eBook Utility of the Future
eBook Utility of the Future
THE FUTURE
An MIT Energy Initiative response
to an industry in transition
Principal Investigators
IGNACIO PÉREZ-ARRIAGA CHRISTOPHER KNITTEL
Professor, Electrical Engineering, Institute for Research George P. Shultz Professor of Applied Economics,
in Technology, Comillas Pontifical University Sloan School of Management, MIT
Visiting Professor, MIT Energy Initiative Director, Center for Energy and Environmental
Policy Research, MIT
Project Directors
RAANAN MILLER RICHARD TABORS
Executive Director, Utility of the Future Study, Visiting Scholar, MIT Energy Initiative
MIT Energy Initiative
Research Team
ASHWINI BHARATKUMAR MAX LUKE
PhD Student, Institute for Data, Systems, SM, Technology and Policy Program (’16), MIT
and Society, MIT
RAANAN MILLER
MICHAEL BIRK Executive Director, Utility of the Future Study,
SM, Technology and Policy Program (’16), MIT MIT Energy Initiative
JESSE JENKINS
PhD Candidate, Institute for Data,
Systems and Society, MIT
Faculty Committee
ROBERT ARMSTRONG WILLIAM HOGAN
Director, MIT Energy Initiative Raymond Plank Professor of Global Energy Policy, John F.
Kennedy School of Government, Harvard University
CARLOS BATLLE
Research Scholar, MIT Energy Initiative STEVEN LEEB
Professor, Institute for Research in Technology, Professor, Electrical Engineering & Computer Science and
Comillas Pontifical University Mechanical Engineering, MIT
RICHARD SCHMALENSEE
Howard W. Johnson Professor of Economics and
Management, Emeritus
Dean, Emeritus, Sloan School of Management, MIT
265 Part 3: Insights on the Economics of Distributed Energy Resources and the
Competition between Centralized and Distributed Resources
265 Understanding the Value of Distributed Energy Resources
307 Part 4: A Policy and Regulatory Toolkit for the Future Power System
307 A Toolkit for Regulators and Policy Makers
325 Appendix A
325 A Description of the Computational Models Used in the Utility of the Future Study
339 Appendix B
339 A Review of Business Models for Distributed Energy Resources
Full report and related working papers are available at: energy.mit.edu/uof
Foreword
and Acknowledgments
An important evolution in the provision and consumption of electricity services is now under
way, driven to a significant degree by a confluence of factors affecting the distribution side of the
power system. A range of more distributed technologies — including flexible demand, distributed
generation, energy storage, and advanced power electronics and control devices — is creating
new options for the provision and consumption of electricity services. In many cases, these novel
resources are enabled by increasingly affordable and ubiquitous information and communication
technologies and by the growing digitalization of power systems. In light of these developments,
the MIT Energy Initiative’s Utility of the Future study examines how the provision and consumption
of electricity services is likely to evolve over the next 10 to 15 years in different parts of the world
and under diverse regulatory regimes, with a focus on the United States and Europe.
The Utility of the Future study is the first of a new series This study does not attempt to predict the future. We
of reports that is being produced by the MIT Energy follow the dictum of poet and author Antoine de Saint
Initiative (MITEI) to serve as balanced, fact-based, and Éxupéry: “As for the future, your task is not to foresee,
analysis-driven guides to key topic areas in energy for but to enable it.” We identify key barriers and skewed
a wide range of decision makers in government and incentives that presently impede the efficient evolution
industry. This study specifically aims to serve as a guide of the power sector and offer a framework for regulatory
for policy makers, regulators, utilities, existing and startup and market reform, based on a comprehensive system
energy companies, and other power-sector stakeholders of efficient economic signals, that will enable an efficient
to better understand the factors that are currently driving outcome, regardless of how technologies or policy
change in power systems worldwide. The report distills objectives develop in the future.
results and findings from more than two years of primary
research, a review of the state of the art, and quantitative
modeling and analysis.
These technologies are being deployed amidst several for distribution utilities that reward cost savings,
broad drivers of change in power systems, including performance improvements, and long-term innovation;
growth in the use of variable renewable energy sources reevaluation of the power sector’s structure to minimize
such as wind and solar energy; efforts to decarbonize conflicts of interest; and recommendations for the
the energy system as part of global climate change improvement of wholesale electricity markets. This study
mitigation efforts; and the increasing interconnectedness also offers a set of insights about the roles of distributed
of electricity grids and other critical infrastructure, energy resources, the value of the services these
such as communications, transportation, and natural resources deliver, and the factors most likely to determine
gas networks. the portfolio of cost-effective resources, both centralized
and distributed, in different power systems. We consider
The MIT Energy Initiative’s Utility of the Future study
a diverse set of contexts and regulatory regimes, but
presents a framework for proactive regulatory, policy, and
focus mainly on North America and Europe.
market reforms designed to enable the efficient evolution
of power systems over the next decade and beyond. This study does not try to forecast the future or predict
The goal is to facilitate the integration of all resources, which technologies will prevail. Instead, it identifies
be they distributed or centralized, that contribute to the unnecessary barriers and distortionary incentives that
efficient provision of electricity services and other public presently impede the efficient evolution of the power
objectives. This framework includes a comprehensive sector and provides a framework that will enable an
and efficient system of market-determined prices and efficient outcome regardless of how technologies or
regulated charges for electricity services that reflect, policy objectives develop in the future. In addition,
as accurately as possible, the marginal or incremental we recognize that regulatory and policy reform often
cost of providing these services; improved incentives proceeds incrementally and that each jurisdiction faces
EXECUTIVE SUMMARY IX
• Flat, volumetric tariffs are no longer adequate for • Outcome-based performance incentives can reward
today’s power systems and are already responsible utilities for improvements in quality of service, such as
for inefficient investment, consumption, and enhanced resiliency, reduced distribution losses, and
operational decisions. improved interconnection times.
• Peak-coincident capacity charges that reflect users’ • Incentives for longer-term innovation are needed
contributions to incremental network costs incurred to to accelerate investment in applied R&D and
meet peak demand and injection, as well as scarcity- demonstration projects and learning about the
coincident generating capacity charges, can unlock capabilities of novel technologies and practices that
flexible demand and distributed resources and enable may have higher risk or longer-term payback periods.
significant cost savings.
The structure of the electricity industry should
• Granularity matters. The value or cost of electricity
services can vary significantly at different times and at
be carefully reevaluated to minimize potential
different locations in electricity networks. Progressively conflicts of interest.
improving the temporal and locational granularity • Network providers, system operators, and market
of prices and charges for these services can deliver platforms constitute the critical functions that sit at the
increased social welfare. However, these benefits must center of all transactions in electricity markets. Properly
be balanced against the costs, complexity, and potential assigning responsibilities for these core functions is
equity concerns of implementation. thus critical to an efficient, well-functioning electricity
sector. It is also critical to establish a level playing field
• Care must be taken to minimize distortions from
for the competitive provision of electricity services
charges that are designed to collect taxes, recover the
by traditional generators, network providers, and
costs of public policies (such as efficiency programs,
distributed energy resources.
heating assistance, subsidies for renewable energy,
cross-subsidies between different categories of • As experience with restructuring in the bulk power
customers, etc.), and recover residual network costs system has demonstrated, structural reform that
(i.e., those network costs that are not recovered via establishes financial independence between
cost-reflective charges). distribution system operation and planning functions
and competitive market activities would be preferable
• Policy makers and regulators must be wary of the
from the perspective of economic efficiency and would
possibility of societally inefficient “grid defection” if
facilitate more light-handed regulation.
residual network costs and policy charges become too
high. This may suggest an upper limit on the portion • If financial independence is not established, several
of these costs that can be collected in electricity tariffs additional measures are critical to prevent conflicts of
rather than through broader taxes or other means. interest and abuses of market power. These include
stricter regulatory oversight of distribution network
The regulation of distribution utilities must be planning and operation, legal unbundling and functional
improved to enable the development of more restrictions on information exchange and coordination
efficient distribution utility business models. between distribution system operators and competitive
subsidiaries, and transparent mechanisms for the
• Forward-looking, multi-year revenue trajectories with
provision of distribution system services (such as public
profit-sharing mechanisms can reward distribution
tenders or auctions).
utilities for cost-saving investments and operations,
aligning utilities’ business incentives with the continual • Maintaining a data hub or data exchange may
pursuit of novel solutions. constitute a fourth critical function. Such a hub or
exchange would serve several purposes: securely
• Several “state of the art” regulatory tools, including
storing metered data on customer usage, telemetry
an incentive-compatible menu of contracts, an
data on network operation and constraints, and other
engineering-based reference network model, and
relevant information; allowing non-discriminatory
automatic adjustment factors to account for forecast
access to this data to registered market participants;
errors, can better equip regulators for an evolving and
and providing end customers with timely and useful
uncertain electricity landscape.
access to data on their own usage of electricity
services. Responsibility for this function should also be
carefully assigned, with priority given to data security
and customer privacy considerations.
EXECUTIVE SUMMARY XI
PART 1: UNDERSTANDING ELECTRICITY SERVICES AND
HOW DISTRIBUTED ENERGY RESOURCES AFFECT THE
DESIGN AND OPERATION OF POWER SYSTEMS
01
A Power Sector in Transition
1.1 The Facts of Today wind and solar energy; the decarbonization of the energy
system as part of global climate change mitigation efforts;
Electric power systems in the United States, Europe, and the increased interconnectedness of electricity with
and several other parts of the world are experiencing an other critical infrastructure — such as communications
unprecedented set of changes driven by the intersection and transportation — which enhances the importance of
of several key trends: the increasing decentralization of electricity in modern economies. These changes all give
power systems, epitomized by the growing penetration rise to a central question: How will the electricity services
of distributed generation (and more recently, energy that are today primarily provided in a centralized, top-
storage) and more active and price-responsive energy down manner be provided in the future?
communications technologies (ICTs) that enable energy system. Later chapters review the efficacy and cost-
to be produced, transmitted, and consumed more effectiveness of existing and proposed implementations
intelligently and efficiently by agents of any size; the of many of these policies, regulations, and market designs
growth of variable renewable energy sources such as and propose efficient upgrades to electricity regulation
and markets where needed.
1 As distributed resources are adopted, many consumers are becoming what
some authors call “prosumers” — that is, they produce energy at some times and
consume it at others. Rather than distinguishing among producers, consumers, and
prosumers, this report will refer simply to network users or agents.
services that is located in the distribution system. of DERs in the power system, although
still modest in most countries; and (3)
the proliferation of ICTs, which makes it
possible for DERs and flexible demand
1.1.1 The power system is becoming
to participate in the functioning of the power system. The
more distributed
issues related to the integration of renewables at large
Power systems around the world are becoming less scale at the level of the bulk power system have been
centralized as the resource mix integrates distributed extensively analyzed elsewhere, although much remains
energy resources (DERs)2 and new options for providing to be done (for example, Chapter 7 discusses market
and consuming electricity services emerge in the design changes needed to integrate increasing levels of
distribution system. In most power systems, DERs remain intermittent renewables).
minor players in the provision of electricity services;
DERs have, in many cases, been conflated with renewable
nonetheless, smart energy consumption and DER
energy resources. Figure 1.1 illustrates the overlap and the
deployment are generally on the rise.
differences between distributed resources and renewable
There is an abundant and rapidly growing body of literature resources. Many renewable resources can, of course, be
on the actual or potential transformation that is taking or deployed in both a distributed or centralized form. This
may take place in the distribution network, and the term study focuses on the potential role of DERs in an evolving
“distributed energy resources” is profusely employed, power system. We consider DERs that are not renewable,
frequently with different meanings. In this study, a distributed such as micro cogeneration or small, gas-fired, backup
energy resource or DER is defined as any resource capable of turbines. Despite our focus on DERs, we also consider
providing electricity services that is located in the distribution generation that is not distributed, such as a large wind
system. DERs include demand response, generation, energy farm connected in high voltage.
storage, and energy control devices, if they are located and
Figure 1.1: Illustrative Taxonomy of Distributed and
function at the distribution level. DERs can be understood Renewable Energy Resources
even without a precise definition of electricity services, which
is provided in Chapter 2. DER
Demand Response
MORE DECENTRALIZED
Some DERs, including electric vehicles, air conditioners, Electric Vehicles
Source: Liebreich 2016. Reprinted with permission from Bloomberg New Energy Finance (BNEF); figure from a presentation given at BNEF
Summit: New York, April 5, 2016.
This global shift to renewable resources is not without that same year11 (Schmalensee 2016). Indeed, California’s
unintended consequences and growing pains. Power market operator, facing unpredicted net demand12 ramps
prices in Germany have reached as low as negative associated with the penetration of solar PV, is introducing
320 euros per megawatt-hour (MWh) and frequently new market products — such as the Flexible Ramping
remain below zero for hours at a time. The share 10
product — to encourage more flexible resources to enter
prices of E.ON, RWE, and EnBW — three of Germany’s the market (California ISO 2016). Challenges remain in
largest utilities — have collapsed by 45 percent to 66 integrating greater shares of renewable energy, but the
percent over the five years preceding the writing of this trend is clear: Renewable energy resources are no longer a
report, even as the DAX index of German stocks grew niche resource in many global power systems, but rather
steadily. Collectively, Western European utilities have one of the largest sources of new generating capacity.
lost hundreds of billions of dollars of market value in
the past decade. Further, in 2014, German distribution 1.1.4 Power systems are decarbonizing
and transmission system operators curtailed nearly 1.6
The growth of renewable energy is occurring in
TWh of renewable electricity — a 200 percent increase
parallel with (and in part due to) a mounting focus on
over 2013, and in 2015, curtailment rose a further 69
decarbonizing electric power systems. Climate change
percent to 2.7 TWh. Many US states are experiencing
presents an urgent global challenge, and in December
similar challenges with negative electricity prices and
2015, 195 nations came together to negotiate the Paris
curtailment; certain renewable generators in the Electric
Agreement, which commits the world’s nations to
Reliability Council of Texas faced negative prices in
limit global average temperature increases to less than
nearly 18 percent of generating hours in 2011, and many
2 °C above preindustrial levels. Almost every nation in
California Independent System Operator units faced
negative prices for nearly 6 percent of generating hours 11 The California Independent System Operator (CAISO) does not make reliability-
based renewable energy curtailments publicly available. What data are available
10 This phenomenon will be discussed in more detail in Chapter 7. Negative prices can indicate that forced curtailments are increasing in CAISO.
emerge for a number of reasons, including production subsidies and operational 12 Net demand here refers to electricity demand minus any intermittent
constraints for certain power plants. renewable generation.
utility company RWE claims that “conventional power these changes aren’t necessarily new. Referring again
generation, quite frankly, as a business unit, is fighting to Fred Schweppe’s “Power Systems 2000” essay, we
for its economic survival.” It is unsurprising then, that see that even in the late 1970s, some scholars believed
Eurelectric, the trade group for European electric utilities, that demand-side resources could play a central role
predicts a 6 billion-euro decline in wholesale generation in balancing supply and demand and operating power
value between 2013 and 2020, and a 10 billion euro systems in the future. The potential for the emergence
increase in downstream market opportunities — including of distributed generation (at that time, primarily CHP)
DERs, services, and power flow optimization in Europe was one of the primary motivators of the 1970 Public
alone (Eurelectric 2013). These trends have driven the Utilities Regulatory Policies Act in the United States,
restructuring of many major incumbent utilities, spurred and some analysts at the time heralded the imminent
the launch of many new ventures, and led to mergers, arrival of a more distributed and renewable future (Lovins
acquisitions, and initial public offerings. 1976) — a future that only four decades later may now
be a realistic possibility. A degree of healthy scepticism
Some envision that the changes seen today will about pronouncements of sweeping change would thus
continue unchecked. They envision a future in which not be unfounded.
centralized resources, and perhaps even transmission
and major portions of distribution, will become relics At the same time, three converging drivers are
of the past — fossils of the fossil fuel age. In this future, accelerating the rate of deployment of distributed and
consumers meet the majority of their energy needs by renewable resources today.
producing on site, transacting for any remaining energy
needs with their nearest neighbors; many defect from the
grid entirely; the power system is completely upended;
and the roles of energy and network suppliers, power
system operators, and regulators are wholly redefined.
-20%
-40%
-60%
-80%
-100%
2008 2009 2010 2011 2012 2013 2014 2015
First, technological innovation has driven dramatic cost for both electric vehicle batteries and stationary energy
declines in a number of technologies. The cost of wind storage, have reached gigawatt-scale markets, driving
and solar PV — the two leading non-hydro renewable approximately 14 percent annual declines in battery costs
energy technologies globally — have decreased by 40 between 2007 and 2014 (Nykvist and Nilsson 2015). One
percent and 60 percent respectively just between 2008 major US automaker projects that lithium-ion battery
and 2014 (see Figure 1.4). Many solar PV developers and cell costs will drop below $100 per kilowatt-hour by
industry analysts expect the installed cost of utility-scale 2022 — an order of magnitude less costly than 2010 costs
solar PV to fall below $1 per watt before the end of this (Wesoff 2016).
decade (Wesoff 2015), and experts foresee a further 24
These recent innovations do not necessarily mean
percent to 30 percent reduction in wind energy costs by
that these technologies are or will be ubiquitously
2030 (Wiser et al. 2016). Even more impressive is the
cost-competitive today or in the near future. Many
cost reduction in light-emitting diodes (LEDs), which
comparative levelized cost of energy13 analyses exist
have plummeted roughly 90 percent since 2008. These
and frequently highlight that, despite recent innovations,
cost declines are enabling the widespread adoption of
DER technologies still require subsidies or other support
LEDs, a solution that may dramatically reduce the 20
to compete with incumbent resources in many markets
percent of electricity consumed in lighting. LEDs are but
(see Figure 1.5 for an example). We discuss the relative
one example of the improvements in the technologies
economics of various DERs and centralized resources in
that enable homes, businesses, and industries to
Chapter 2 and Chapter 8.
consume energy more flexibly and efficiently. Energy
storage technologies are also progressing at a rapid rate.
Lithium-ion batteries, now the common technical basis
13 Levelized cost of energy is a metric that divides the total capital and operational
costs of a given energy resource by the cumulative energy output of that resource
over its lifetime. This metric is useful for simple comparisons of relative technology
costs, but it provides an incomplete picture when comparing technologies with
very different utilization patterns in a power system — comparing dispatchable
technologies with intermittent renewables, for instance.
$350
$291
$300
$250
$201
$200
$136
$150
$111
$108
$93
$103
$103
$88
$85
$100
$66
$64
$69
$55
$50
$-‐
l
l
e
)
al
S)
e
S
re
c le
ar
tric
ss
nd
a
oa
ycl
PV
CC
ycl
(CC
erm
erm
ma
c le
ho
Cy
Wi
al
C
c
d
C
ic
(
d
C
w/
ele
ffs
ge
Nu
B io
ed
oth
Th
lt a
ine
on
ine
—O
dro
c le
ora
bin
ced
lar
vo
Ge
2
mb
mb
Cy
Hy
en
om
nd
St
oto
So
van
Co
ed
Co
nv
e &
Wi
C
Ph
Ad
Co
bin
as
on
s
tur
Ga
lar
l
G
2
om
ap
fica
ced
So
na
s
C
n
C
2o
asi
van
Ga
rbo
en
d
G
Ad
ced
nv
Ca
ate
Co
van
ith
egr
C w
Ad
Int
IGC
Second, policies — particularly those related to the as feed-in tariffs, auctions, or utility purchase obligations
deployment of renewable energy technologies and (e.g., renewable portfolio standards), as well as indirect
decarbonizing the power sector — have created favorable support schemes such as priority in the dispatch,
investment environments for many of these emerging omission from certain market-bidding procedures, or
technologies, building a positive feedback loop with waivers of balancing costs. In other cases, regulations
technological innovation that has, to date, resulted may fail to capture the increasingly complex realities of
in continued cost declines. Nearly every developed today’s power sector, providing unintended incentives
economy, and many developing economies, have created for DER adoption. Most notably, the recovery of fixed
policy mechanisms to encourage the deployment of network costs through flat, volumetric tariffs, together
renewable generation and increasingly storage as well. with the practice of net metering14 of demand and behind-
In 2014, subsidies for the deployment of renewable the-meter generation, results in skewed incentives for
energy technologies amounted to $112 billion globally network users with embedded generation.
(the consumption of fossil fuels is also subsidized — to
Consumer choice and preference is a third and final driver
the tune of $490 billion globally) (IEA 2015). In 2013 in
of change. DERs may bring an unprecedented level of
the United States, renewable energy resources received
choice to agents that were formerly passive consumers
51 percent of all energy-specific federal subsidies, and
wind and solar collectively accounted for 64 percent of
14 “Volumetric tariffs” refers to the practice of pricing electricity services on a dollar
federal electricity production subsidies (US DOE 2015a). per kilowatt-hour basis. “Net metering” refers to the netting of energy supplied
These policies materialize in specific regulations or locally with energy supplied via the bulk power system. Net metering is typically
performed over a specific time period; that is, all of the energy produced locally
market rules, including direct support mechanisms such over, say, a one-month period is netted with all energy supplied via the power
system in that month.
Figure 1.6: Suite of Models Deployed in MIT Utility of the Future Study
Transmission
Electricity Sector Areas
Centralized Generation
GenX
Distribution ROM+
D-Sim RNM
Network User/DERs DR DRE DER-CAM
Anaya, Karim L., and M.G. Pollitt. 2015. “Options for Electricity-ISAC and SANS Industrial Control Systems.
Allocating and Releasing Distribution System Capacity: 2016. “Analysis of the Cyber Attack on the Ukrainian
Deciding between Interruptible Connections and Firm DG Power Grid.” Electricity-Information Sharing and Analysis
Connections.” Applied Energy 144. Center. ics.sans.org/media/E-ISAC_SANS_Ukraine_
North American Electric Grid: New Policy Approaches to EPA. 2015. 40 CFR Part 60: Carbon Pollution Emission
Address the Evolving Threat.” February 2014. Guidelines for Existing Stationary Sources: Electric Utility
Generating Units. Washington, DC: US Environmental
BNEF. 2016. New Energy Outlook 2016: Long-Term
Protection Agency.
Projections of the Global Energy Sector. London, United
Kingdom: Bloomberg New Energy Finance. Eurelectric. 2013. Utilities: Powerhouses of Innovation.
Brussels, Belgium.
BP. 2016. BP Statistical Review of World Energy 2016.
London, United Kingdom: BP. European Commission. 2014. “2030 Climate & Energy
Framework.” Climate Action. ec.europa.eu/clima/policies/
California ISO. 2016. California Independent System
Operator Corporation Docket No. ER16- ___-000 Tariff strategies/2030/documentation_en.htm.
Amendment to Implement Flexible Ramping Product. European Commission. 2015. Energy Union Package: A
Framework Strategy for a Resilient Energy Union with a
Campbell, R.J. 2015. Cybersecurity Issues for the Bulk Power
Forward-Looking Climate Change Policy. Brussels, Belgium.
System. Washington DC: Congressional Research Service.
R43989; 7-5700. June 10, 2015. www.fas.org/sgp/crs/ European Commission. 2016. “Smart Metering
misc/R43989.pdf. Deployment in the European Union.” ses.jrc.ec.europa.eu/
smart-metering-deployment-european-union.
CSI. 2016. “California Solar Statistics.” Accessed January
9, 2016. California Solar Initiative. Frankfurt School-UNEP Centre, and BNEF. 2016. Global
Trends in Renewable Energy Investment. Frankfurt, Germany.
DDPP. 2015. “Pathways to Deep Decarbonization.” Deep
Decarbonization Pathways Project. Fraunhofer ISE. 2016. Recent Facts about Photovoltaics in
Germany. Freiburg, Germany.
EIA. 2016a. “Wind Adds the Most Electric Generation
Capacity in 2015, Followed by Natural Gas and Global Wind Energy Council. 2016. Global Wind Report:
Solar.” Today in Energy. Washington, DC: US Energy Annual Market Update 2015. Brussels, Belgium.
Information Administration. www.eia.gov/tools/faqs/faq.
IEA. 2015. World Energy Outlook 2015. Paris, France:
cfm?id=108&t=3.
International Energy Agency.
Munsell, Mike. 2016. “US Energy Storage Market Grew The Brookings Institution. 2011. Assessing the Role
243% in 2015, Largest Year on Record.” Greentech Media. of Distributed Power Systems in the U.S. Power Sector.
March 3, 2016. www.greentechmedia.com/articles/read/ Washington, DC.
us-energy-storage-market-grew-243-in-2015-largest-
Trabish, Herman K. 2016. “17% of Hawaiian Electric
year-on-record.
Customers Now Have Rooftop Solar.” Utility Dive.
National Academies of Sciences, Engineering, and February 1, 2016.
Medicine. 2016. The Power of Change: Innovation for
Trabish, Herman K. 2016. “How Storage Can Help Solve
Development and Deployment of Increasingly Clean Electric
the Distributed Energy ‘Death Spiral’: A New Navigant
Power Technologies. Washington, DC: The National
Paper Lays out Three Examples of How Storage Can
Academies Press.
Facilitate the Transition to a More Distributed Grid.” Utility
North Carolina Clean Energy Technology Center. 2016. 50 Dive. June 21, 2016.
States of Solar: Q2 2016 Quarterly Report. Raleigh, NC.
US DOE. 2016. Revolution… Now: The Future Arrives for Five
Nykvist, Björn, and Måns Nilsson. 2015. “Rapidly Falling Clean Energy Technologies—2016 Update. Washington, DC:
Costs of Battery Packs for Electric Vehicles.” Nature US Department of Energy. September 2016.
Climate Change 5: 329-332.
Wesoff, Eric. 2016. “How Soon Can Tesla Get Battery Cell
Costs Below $100 per Kilowatt-Hour?” Greentech Media.
March 15, 2016. www.greentechmedia.com/articles/
read/How-Soon-Can-Tesla-Get-Battery-Cell-Cost-Below-
100-per-Kilowatt-Hour.
02
New Options for the Provision and
Consumption of Electricity Services
This report is not about the future of today’s utility. 2.1 Power System Services
Rather, it is about how the services provided by electric
utilities today will be provided in the future — by the Power systems are rapidly evolving to integrate
utility of the future. Since enabling the efficient provision participation by many agents playing diverse roles;
and consumption of such services is the major focus furthermore, these agents may adopt different roles at
of this study, we begin here by defining what we mean different times (e.g., consumer at one hour, producer at
by electricity services. Understanding these services is another). In the context of the power system, an agent’s
critical to realizing why the new means of providing such role is characterized by the electricity services it provides
services — mainly distributed generation, energy storage, to and obtains from other agents.
and demand response — can be particularly impactful. Electricity services are activities performed in the context
Furthermore, understanding and communicating the value of a power system that have economic value for some
of the electricity services that a given agent consumes agents, regardless of whether this value is monetized.
is a precursor to allowing that agent to be more price- Electricity services are distinguished from other more
responsive and efficient. generic services in that they create economic value by
enabling the consumption of electrical energy, lowering
the costs associated with consuming electrical energy,1
or both. Distributed energy resources (DERs) can provide
electricity services, as can any other energy resource.
1 Note that these costs include both private costs internalized in market transactions
and public costs or externalities not monetized in market transactions.
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 19
Electricity services are activities performed Within this framework, the objective of all planning and
in the context of a power system that have operations is to maximize social welfare — that is, to
maximize the sum of all of the consumer surplus created
economic value for some agents, regardless
by electricity consumption, minus the sum of the costs
of whether this value is monetized. Electricity
of all of the actions required to supply the consumed
services create economic value by enabling electricity.2 These costs emerge from producing,
the consumption of electrical energy, lowering consuming, storing, transmitting, or transforming other
the costs associated with consuming electrical forms of energy into electricity (and vice versa), and
energy, or both. they include both private costs internalized in market
transactions as well as public costs or externalities not
monetized in transactions. This optimization is subject to
Electricity services vary widely in their nature: They multiple constraints that emerge from the aforementioned
may be physical, financial, or information-based, and physical laws, policies, regulations, and contracts.
they may vary in format, with different durations, Viewed through the framework of optimization, a binding
levels of commitment, cost allocation methods, and or active constraint (whether imposed by physical laws,
economic implications. regulations, policies, or any other source) creates a
This section logically defines electricity services. We do shadow price that reflects the marginal value of either
not attempt to define the most efficient, fundamental, relaxing (i.e., changing the value of) the constraint
or mutually exclusive and collectively exhaustive set of or finding alternative ways to meet it.3 That is, this
services. Rather, we provide a definition that is robust to shadow price reflects the increase in social welfare if the
different regulatory, market, and technology paradigms constraint is relaxed at the margin. An active constraint
that can be used to understand how DERs — or any other therefore creates the opportunity of taking action to relax
energy resource — may create value in power systems. or relieve it, increasing the welfare created by the use of
electricity in the power system; this action is an electricity
2.1.1 Power systems operated and planned service, by our definition.
system and the reason for its existence. by electrical energy could
alternatively be provided by
other forms of energy, such
as thermal energy.
2 See Chapter 2 of Regulation of the Power Sector (Perez-Arriaga et al. 2013) for this
mathematical formulation and its implications.
3 In the optimization literature, this price is referred to variously as a “shadow price,”
“dual variable,” or “Lagrange multiplier” of the constraint.
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 21
In theory, in the absence of uncertainty, the only The goal of the imposition of these constraints should
service that needs to exist is the provision of power be to coordinate operational decisions in a manner that
(conventionally decomposed into active and reactive) represents the optimal trade-off between the total system
at a location and a point in time. The prices that are costs for planning and operation and the costs associated
determined in real time could, in theory, coordinate with losing energy supply (i.e., the costs of blackouts due
the consumption and production of power at all points to any number of reasons, from lack of operating reserves
to respect the physical power balance and network to network failures, etc.). In practice, constraints may
constraints and motivate actors to invest sufficiently simply reflect the effect of risk aversion or incentives on
(with the important exclusion of networks, which are not the regulator or the system operator.
fully compensated by spot prices ). 7
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 23
may approximate the transfer capacity of a line by implemented to internalize externalities or achieve other
estimating its thermal state without either measuring the policy objectives. Policy makers may create constraints
conditions of the line directly or accounting for all of the such as mandates for maximum levels of carbon emitted
environmental and system factors that might influence its or minimum levels of renewable energy resources or
true thermal state (and therefore transfer capacity). This DERs used in a power system. For example, maximum
transfer capacity constraint is a simplifying constraint carbon emissions constraints create a price for emitting
made to ensure the safe operation of the power system. carbon, creating a service for carbon emissions reduction.
Typically, these constraints are created to internalize
Reliability constraints simplify complex economic and
externalities that are not currently accounted for in power
technical calculations to develop measures of failure and
systems operation, planning,
and markets or to further other
policy objectives. In the carbon
Another common class of services is created by constraints emissions example, the constraint
implemented to internalize externalities or achieve other enables carbon emitters and/
policy objectives. or consumers to internalize the
“cost” that carbon emissions
impose on society. In cases like
this, it would be economically
maximum System Average Interruption Duration
inefficient for regulators to impose constraints if the cost
Indices (SAIDI) or System Average Interruption
of imposing the constraint exceeded the value of the
Frequency Indices (SAIFI). In North America, these
externality or public good.10
constraints are placed on the transmission system by
the North American Electric Reliability Corporation, and
2.1.5 Services without constraints
at the distribution system by state regulators and other
more local bodies. Thus far we’ve highlighted the various ways that
constraints can create electricity services and how the
Network owners may also include some sort of “network
activation of these constraints creates opportunities
capacity margin” — that is, they may impose a constraint
to create value by providing a service. However, our
on the minimum margin between network transfer
definition of electricity services is not restricted to the
capacity and the peak expected power flows through
creation of services by constraints. Any action that
that network infrastructure. This may take the form of
creates value by enabling the consumption of electrical
a network upgrade heuristic or planning margin. The
energy, lowering the cost of consuming electrical
possibility of this constraint becoming binding creates an
energy, or both, is an electricity service. For example,
opportunity to provide a service by either building more
conservation voltage reduction (CVR) — a technique
network capacity or securing commitments for future
that reduces energy consumption by reducing the
actions that can reduce peak power flows (e.g., through
voltage of the consumed energy — is a service that can,
the installation and operation of DERs, contracting with
in certain cases, increase social welfare by lowering the
flexible demand, etc.).
total cost of energy delivery. However, the economic
value of this service does not emerge directly out of an
2.1.4 Policy and other constraints activated constraint; rather, it is measured by the cost
The first two categories of services emerged from of implementing the CVR solution and the value of the
constraints imposed by the physics of power systems and
10 This is a simple characterization of the use of constraints and market mechanisms
those imposed to simplify, coordinate, and ensure the
to internalize externalities. Externalities can also be internalized with price
safe operation and planning of these systems. Another instruments such as taxes (e.g., a carbon tax for internalizing the cost of carbon
emissions) or incentives (e.g., clean energy credits for internalizing the value of
common class of services is created by constraints clean energy technologies). These pricing mechanisms often implicitly represent a
target for the taxed or incentivized activity or resource.
between the intrinsic economic value of a given service power plants or pumped-hydroelectric energy storage),
and the way it is monetized and compensated in practice. DERs are characterized by relatively small capacities (a
In Chapter 4 of this study, we discuss how services few kilowatts to a few megawatts), and are connected to
should be monetized to lead to efficient outcomes, and lower voltage electricity distribution grids (as opposed to
we address the various intricacies associated with this transmission and high-voltage distribution systems).
monetization. The goal of the comprehensive system of DERs can be divided into two distinct classes: DERs
cost-reflective prices and charges introduced in installed specifically to provide electricity services (e.g.,
Chapter 4 is to ensure that the marginal cost to an agent energy storage devices, solar photovoltaic systems with
of consuming a service or the marginal value for the agent smart inverters, power electronics, or distributed fossil
of providing that service is equal to the marginal increase generation);12 and resources that exist primarily for
or reduction of the total social welfare created by that reasons other than to provide electricity services but that
action. That is, we attempt to align the cost or benefit
of consumption or production for an agent (the private 11 See Ackermann et al. (2001) and Pepermans et al. (2005) for often-cited academic
definitions. See DNV GL (2014) for a more updated survey.
value for that agent) with the cost or benefit accrued to
12 Of course, not every solar PV system or distributed generation system is installed
the entire power system (the system value). Distortions for the sole or even primary purpose of providing electricity services. Some PV
systems, for example, may be installed simply because the systems look “cool” or
are created when the private value created by an action is because their owners want to feel independent from the grid. Nonetheless, for most
investors in solar PV or energy storage, the savings or earnings associated with the
less than or greater than the system value of that action.
sale of electricity services are an important motivator.
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 25
can be harnessed for this purpose (e.g., flexible demand consumption of a service, not the provision of a service.
and electric vehicles). Cost-reflective prices and charges We refer to demand that serves as an economically
for electricity services — which will be discussed further efficient consumer of electrical energy as “price-
in Chapter 4 — can create the conditions to incentivize responsive demand.”
DERs of the first class to be installed only when DERs will
This stems from the fact that demand resources cannot
add value. Furthermore, these cost-reflective prices and
provide energy — they can only consume energy. Of
charges enable existing demand and DERs to create value
course, agents can procure the option to consume energy
by exposing the owner to the potential benefit of engaging
(either through a retail supply contract or by participating
in the market.
in forward markets) and sell that option (Borlick 2010;
We note that demand is both a consumer of electricity Hogan et al. 2012). Alternatively, an agent with flexible
services and a potential supplier of services, and we demand may be able to make a forward commitment
thus consider demand a potential resource. However, to provide a future response, where the commitment
it is important here to distinguish between demand itself has value beyond the final energy consumption (or
providing a service and the more efficient consumption lack thereof). A flexible demand agent may commit in
of electricity services. The act of deciding to consume advance to stand by and be ready to adjust consumption
or not consume electrical energy services based on the in order to deliver an additional service, such as operating
price of such services does not itself provide an electricity reserves, firm capacity, or network capacity margin.
service. Responding to high electricity prices by curtailing This commitment itself delivers value distinct from the
consumption is simply an expression of the degree to actual final consumption of energy, may incur additional
which a given agent values the consumption of electrical standby or commitment costs, and may require payment
energy (i.e., their willingness to pay). This flexible demand of additional compensation beyond avoiding paying for
can be an important source of improved efficiency in energy that was not consumed. We consider these cases
power systems but constitutes the economically efficient to constitute demand as a service provider and refer to
these cases collectively as “demand response.”
However, it is important to remember that DERs are not the only new energy option. Many new
technologies and activities such as network topology control or advanced transformers are capable
of providing electricity services, although they do not fit cleanly into existing definitions of DERs.
Furthermore, as we highlight throughout the report, the efficient consumption of electricity services
may be among the most cost-effective means of improving the success of the power sector in providing
welfare. Thus, we encourage readers to keep sight of the broader focus on decentralizing and expanding
options for providing and consuming electricity services.
Source: Lazard (2015). Levelized cost of storage calculations assume 10-year project life, 1 megawatt (MW), 4 megawatt-hour (MWh)
battery, 350 cycles per year at 100 percent depth of discharge, and $50 per MWh cost of charging. Diesel generator costs are provided
for comparison.
It is important also to note that the technical and First, there are cost and performance differences between
economic characteristics of DERs are as varied as the technologies within a given class of DERs. For example,
DERs themselves, and there is significant variation in cost in the electrochemical battery class, lithium-ion batteries
and performance among even very similar DERs. This is and lead acid batteries have very different performance
critical, as it highlights the futility of attempting to define characteristics. Similarly, within the class of solar PV
a single value for all DERs. Chapter 8 explores this topic
13
technologies, thin film and crystalline silicon (c-Si)
in detail and provides quantitative evidence. systems have different characteristics. Figure 2.1 presents
an example of the significant variation in cost of various
energy storage technologies deployed at commercial and
industrial facilities.14
13 This is not a purely academic concept. The proliferation of DERs has spurred a
multitude of studies, utility programs, and tariff structures that attempt to define
the value of DERs. For example, “value of solar” tariffs — that is, a tariff that
attempts to value all of the benefits (and costs, if applicable) of distributed solar
and applies only to owners of distributed solar systems — has emerged as a popular
regulatory tactic. Spain has implemented a fee, targeted at generation behind the
meter from DERs with capacities above 10 kilowatts, to prevent non-recovery of
regulated network and policy costs. Consolidated Edison — a distribution utility in
New York State — and other utilities have offered bill rebates for customers who
purchase smart thermostats and enable the utility to control them during certain 14 The levelized cost of storage (LCOS) is similar to the levelized cost of energy in
periods. All of these efforts are an attempt to define either the value or cost of that it attempts to create a metric for the comparison of costs across technologies
a specific DER category (e.g., flexible demand or solar PV). In this same vein, a and does not attempt to calculate the value of the technology or the application.
recent draft manual issued by the US National Association of Regulatory Utility LCOS is calculated by dividing the total energy provided by the storage system
Commissioners (NARUC) reviews and evaluates methods to “compensate” DERs (with appropriate charging and discharging losses) by the total cost of building and
by focusing on the value of each technology (NARUC, 2016). operating the system over the system’s lifetime.
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 27
Figure 2.2: AC Energy Produced by 10-Kilowatt c-Si PV Systems in Phoenix, Arizona, Deploying Different
Mounting Systems
3,000
2,500
2,000
kWh
1,500
1,000
500
Second, there are cost and performance differences dramatically different energy outcomes depending on
between technologies of a given type. For example, a orientation and whether or not they employ tracking
lithium-ion battery with two hours of storage capacity systems. In this example, PV systems deploying dual-axis
may cost much less than a lithium-ion battery with tracking systems produced up to 33 percent more energy
four hours of storage (where lithium-ion batteries is in a given month than PV systems without tracking.
the technology type). A c-Si PV module with 15 percent
Finally, there are cost and performance differences
efficiency may likewise cost significantly less than a c-Si
within identical technologies at a given scale. For
module with 20 percent module efficiency (where c-Si
example, the levelized cost of a c-Si PV system installed
PV is the technology type). Figure 2.2 also illustrates
at the 5-kilowatt scale will be significantly higher than
the performance variation within the c-Si PV technology
the levelized cost of a c-Si system installed at the
type. Two identically sized (10 kilowatt) and identically
100-megawatt scale. Figure 2.3 shows the variation in
located (Phoenix, Arizona) solar PV systems can produce
2015 levelized costs of energy (LCOEs) for solar PV at
different scales on Long Island in New York State.
$0.25
$0.20
$0.15
$0.15
$0.06
$0.03
$0.10
$0.00
30 MW 1.5 MW 1 MW 5 kW
30 MW LCOE Premium on 30 MW LCOE
This section introduced some key DER characteristics with the location of provision.16 This emerges from the
with the purpose of highlighting the futility of assigning a physical characteristics of the networks that connect all
single value to DERs and the importance of establishing a power system agents, including losses, capacity limits
system of cost-reflective prices and charges for electricity of network components, and voltage limits at network
services. The following section explores the factors that nodes. DERs, enabled by their distributed nature, can
uniquely differentiate DERs, shedding light on their role provide services in areas of the power system where
in power systems and further underscoring the dramatic these services are most valuable. Understanding the
variation in their value.15 locational nature of DERs allows us to understand the
value of investing in a resource or operating it in one
location versus another.
2.3 What Do DERs Do
Table 2.1 classifies various commonly cited DER values as
Differently? locational or non-locational. In addition, it distinguishes
between values derived directly from the power system
2.3.1 DERs provide locational services
and other values not directly associated with the
The first and primary differentiating factor for DERs is that provision of electricity services.
their distributed nature enables them to provide services
either more effectively, cheaply, or simply in locations
inaccessible to more centralized resources. This capability
is important, because the value of some services changes
15 Assumes capital costs of $1,500 per kilowatt of alternating current (AC) power at
30 megawatts (MW) AC, $2,000 per kilowatt (kW) AC at 1.5 MW, $2,600 per
kW AC at 1 MW, and $4,100 per kW at 5 kW. All systems produce 1,458 kilowatt-
hours (kWh) per kW, or a roughly 16.6 percent capacity factor. This capacity factor
is consistent with a 1-kW AC system at a fixed 40.8-degree tilt and 180-degree 16 The value of many services also changes with time, but this is not important to
azimuth in Long Island, New York (latitude: 40.78, longitude: -73.1). understanding the distinguishing features of DERs.
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 29
Table 2.1: Classification of DER Values17181920
LOCATIONAL NON-LOCATIONAL
Energy
Network Capacity Margin
Firm Generation Capacity18
Network Constraint Mitigation
POWER SYSTEM VALUES Operating Reserves19
Power Quality17
Price Hedging
Reliability and Resiliency
Black-start and System Restoration
Land Use
Emissions Mitigation
OTHER VALUES Employment
Energy Security
Premium Values20
To make this conversation more concrete, Figure 2.4 The value of locational services can be persistently
shows the prices of energy throughout the interconnect different — not just different in one hour as presented
managed by PJM, a regional transmission organization, in Figure 2.4. Figure 2.5 shows a histogram of the daily
in an example hour (July 19, 2013, at 4:05 p.m. ET). As average marginal cost of energy at all PJM aggregate
shown, energy was clearly more valuable near New York, nodes in 2015. While most nodes have a daily average
New York, than near Cincinnati, Ohio, on this day and locational marginal price (LMP) of roughly $22.5 per
at this time. The same concept applies to energy (and megawatt hour (MWh) to $30 per MWh, some nodes
other locational services) within a distribution network. have daily average LMPs above $105 per MWh.
If the value of a service is higher as one goes “down”
into lower voltages of the distribution system, it may be
more beneficial to have DERs provide this service than
centralized resources.
Source: PJM
Figure 2.5: Daily Average Marginal Cost of Energy at All PJM Aggregate Nodes in 2015
Of course, not all services change value based on their of normal or contingency events. Since, except at very
location within the service network. For example, the short time scales, frequency is consistent across an
value of mitigating carbon dioxide (CO2) emissions is entire synchronized interconnected system, the value of
the same wherever the CO2 is emitted in the world, controlling that frequency does not change based on the
since CO2 mixes uniformly in the atmosphere over location of frequency regulation within the system.
time and each ton of CO2 contributes equally to global
DERs can create value by providing locational services
warming. Similarly, operating reserves are deployed to
where centralized resources cannot or when the
contain frequency deviations that emerge as a result
locational value of a service outweighs any added costs
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 31
associated with providing a service in a distributed with retailers, and creating coordination challenges.
fashion (we explore this concept in Chapter 8). A large Aggregators are enabling more price signals to be sent
combined-cycle natural gas turbine may not be able to agents, allowing new agents to become active in
to alleviate the strain on an overloaded transformer the operation and planning of the power system, and
serving a growing neighborhood in a densely populated engaging agents in novel ways. Figure 2.6 portrays the
city, but demand response in the neighborhood or potential role of aggregators as intermediaries that
commercial-scale energy storage could. However, not all enable more granular price signals to be sent between
values are locational, and the distributed implementation different power system actors (e.g., the distribution
of a resource that can also be deployed at scale (e.g., system operator or the transmission system operator)
distributed solar or storage versus centralized solar or and the agents they serve. Aggregation should not be
storage) should only be considered preferentially when accepted as valuable without critical assessment. For
the value of services provided justifies any additional example, aggregation across locations with different
costs due to not fully exhausting economies of unit scale prices could present significant operational challenges.
(discussed in Chapter 8). This study examines what value aggregators create,
which leads to a better understanding of which agents
2.3.2 DERs can be aggregated should be allowed to perform aggregator functions (see
Chapter 6). This requires exploring the different ways
Another key distinguishing feature of DERs is that their
that aggregators can create value (or disvalue) in power
small scale creates new opportunities for aggregators
systems, building on the classification of values highlighted
in power systems. This study adopts a slightly modified
in Section 2.3. This discussion will attempt to lend clarity
version of the definition of an aggregator promulgated by
to pertinent questions related to the role of aggregators
Ikäheimo, Evens, and Kärkkäinen (2010). In the context
such as: Should the power system accommodate many
of this study, “an aggregator is a company that acts as
aggregators or only one centralized aggregator? Who can
an intermediary between electricity end-users and DER
or should be an aggregator (transmission and distribution
owners and the power system participants who wish to
system operators, retailers, third parties, etc.)? What
serve these end-users or exploit the services provided
market design elements may need to be adapted or
by these DERs.” Retailers are therefore a special class of
adopted to accommodate DER aggregators? How should
aggregators that (historically) served only the function
new third-party aggregators be coordinated with existing
of aggregating small electricity consumers — residences
power system aggregators (e.g., retailers)?
and small commercial entities — and procuring power
on their behalf.
2.3.3 DERs give rise to novel In Chapter 6, we review the structure of nearly 150
21 Note that a retailer can perform the function of an aggregator, in which case price
signals would go directly from system operators to the retailer.
22 This study adopts the definition of a business model put forth by Osterwalder
and Pigneur (2010). According to Osterwalder and Pigneur, a business model
is comprised of: a value proposition, key partners, customer segments, a cost
structure, revenue streams, key activities, key resources, customer relationships,
and channels.
CHAPTER 2: New Options for the Provision and Consumption of Electricity Services 33
References NARUC. 2016. Draft NARUC Manual on Distributed Energy
Resources Compensation. Washington, DC: National
ACER. 2013. “Capacity Remuneration Mechanisms and Association of Regulatory Utility Commissioners.
the Internal Market for Electricity.” Ljubljana, Slovenia:
Agency for the Cooperation of Energy Regulators. NREL. PV Watts Calculator. National Renewable Energy
Laboratory. pvwatts.nrel.gov/.
Ackermann, T., G. Andersson, and L. Söder. 2001.
“Distributed Generation: A Definition.” Electric Power Newell, S., K. Spees, J. Pfeifenberger, R. Mudge, M. DeLucia,
Systems Research 57: 195–204. doi:10.1016/S0378- and R. Carlton. 2012. ERCOT Investment Incentives and
Bohn, R.E., M.C. Caramanis, and F.C. Schweppe. 1984. O’Neill, R.P., P.M. Sotkiewicz, B.F. Hobbs, M.H. Rothkopf,
“Optimal Pricing in Electrical Networks over Space and and W.R. Stewart. 2005. “Efficient Market-Clearing Prices
Time.” The RAND Journal of Economics 15: 360–376. in Markets with Nonconvexities.” European Journal of
Operational Research 164: 269–285. doi:10.1016/
Borlick, R.L. 2010. “Pricing Negawatts.” Public Utilities j.ejor.2003.12.011.
Fortnightly 14–19.
Osterwalder, A., and Y. Pigneur. 2010. Business Model
DNV GL. 2014. “A Review of Distributed Energy Generation: A Handbook for Visionaries, Game Changers, and
Resources.” Arlington, VA: DNV GL. Challengers. John Wiley & Sons.
Ela, E., M.R. Milligan, A. Bloom, A. Botterud, A. Townsend, Pepermans, G., J. Driesen, D. Haeseldonckx, R. Belmans,
and T. Levin. 2014. Evolution of Wholesale Electricity Market and W. D’haeseleer. 2005. “Distributed Generation:
Design with Increasing Levels of Renewable Generation. Definition, Benefits, and Issues.” Energy Policy 33:
Golden, CO: National Renewable Energy Laboratory. 787–798. doi:10.1016/j.enpol.2003.10.004.
ENTSO-E. 2013. “Network Code on Load — Frequency Pérez-Arriaga, I. 2011. “Managing Large Scale Penetration
Control and Reserves.” European Network of Transmission of Intermittent Renewables.” MIT Energy Initiative
System Operators. Symposium. Cambridge, MA: Massachusetts Institute
of Technology 1–43.
Gribik, P.R., W.W. Hogan, and S.L. Pope. 2007. Market-
Clearing Electricity Prices and Energy Uplift. Cambridge, MA: Pérez-Arriaga, I.J. 2013. Regulation of the Power Sector,
Harvard Electricity Policy Group. Power Systems. London, United Kingdom: Springer.
doi:10.1007/978-1-4471-5034-3.
Hogan, W., R.D. Tabors, R.D., and M. Caramanis. 2012.
“FERC Order 745 Amici Curae on Demand Pérez-Arriaga, I.J., F. Rubio-Odériz, J.F.P. Gutiérrez,
Response Payments.” J.A. Ogando, and J. Marín. 1995. “Marginal Pricing of
Transmission Services: An Analysis of Cost Recovery.”
Hogan, W.W. 2015. “Electricity Market Design:
IEEE Transactions on Power Systems 10: 65–72.
Energy and Capacity Markets and Resource Adequacy.”
Capacity Markets: Gauging Their Real Impact on Resource
Development and Reliability. Cambridge, MA:
EUCI Conference.
03
Envisioning a Future with
Distributed Energy Resources
As early as 1978, in an article titled “Power Systems power systems could bring to the provision of electricity
2000,” the late MIT Professor Fred Schweppe anticipated services. Some of these changes have already begun. This
that price-driven demand response by residential and chapter consists of four parts. The first part presents a
commercial customers, when enabled by efficient price collection of perspectives about the future of the electric
signals and communication technologies, could play a power sector that depart from traditional assumptions
large role in future electricity systems. Nearly forty years about the sector’s course going forward. The changes
later, the electric power sector is beginning to implement discussed in this section may or may not end up having a
what Schweppe foresaw. But while Schweppe only substantial impact on the future power sector landscape,
considered demand response, this study contemplates but they are conceivable. Each of the perspectives
a wide range of new technologies that will shape the we discuss is associated with high penetration and
electicity industry going forward, including distributed participation of distributed energy resources, and each
generation, distributed storage, and electric vehicles, challenges traditional views about the role of supply and
as well as many diverse and sophisticated forms of demand in electricity markets.
demand response, metering, energy control devices, and
The second part of this chapter presents a collection of
communications and computational capabilities, both
case studies that illustrate the capabilities of distributed
centralized and distributed. Schweppe’s article offered
energy resources (DERs) when they are able to play an
a perspective that was different from the established
active role in power system operations. The case studies
perspective of the time. He envisioned the power system
show how DERs can change the power sector;1 they also
as a meeting point of supply and demand, where supply
illustrate how power systems can function when DERs
and demand play equally important roles and together
are present in large quantities and point to the new
produce a “homeostatic” equilibrium. The purpose of
this chapter is to introduce the reader to the changes
1 This change, to the extent that it may happen, could take diverse forms in different
that a substantial presence of distributed resources in power systems, depending on idiosyncratic factors such as the pre-existing
industrial base or existing technology-specific support schemes.
resources be realized.
Today, customers who adopt DERs are primarily 74 percent of total electricity use and nearly half of
motivated by the opportunity to reduce their energy this consumption occurs in the commercial sector
bills. This opportunity exists as a result of policies and (EIA 2015). Energy use by HVAC systems accounts
regulations. An increasing number of DER technologies, for the largest fraction of total commercial electricity
including solar photovoltaics (PV), batteries, home consumption and is often the most variable category.
and-power (CHP) systems, fuel cells, and others, are and structures can provide energy storage when
now cost-effective in certain applications. This presents strategically heated or cooled. Methods such as model
many opportunities and challenges for providing the right predictive control (MPC) can be used to determine
mix of policy incentives. In addition, the emergence of optimal cooling strategies that exploit thermal storage
distributed technologies has the potential to enhance the to minimize energy costs in addition to potentially
provision or consumption of electricity services. providing other services, such as operating reserves.
Increasing knowledge and interest in acquiring enabling
This section uses case studies to illustrate how customers technologies and software, combined with HVAC
may opt to use different DERs depending on the financial systems that can facilitate more flexibility in building
benefits they provide — both by avoiding having to operation, may prompt commercial building owners or
procure services from the system and by generating operators to participate in ancillary services markets.
revenues in return for providing services to the system.
The first case study involves reducing energy costs by To explore this potential, we simulate a medium-sized
managing building heating and cooling loads in response office building, located in Boston, Massachusetts,
to hourly electricity prices. In this case study, buildings that participates in ancillary services markets while
can also provide ancillary services to the system in optimizing hourly energy consumption to reduce total
competition with traditional providers, such as centralized costs.5 In this simulation, the building is subject to
generators. The second case study builds on the fact that either wholesale locational marginal prices (LMPs), as
gas and electricity are increasingly becoming substitute established by the independent system operator for
fuels for a variety of end-user services such as heating New England (ISO-NE), or flat retail electricity rates.6
and cooling. It illustrates how various factors, including At the same time, the building can generate revenues
retail gas and electricity prices, climate conditions, and from ISO-NE’s day-ahead markets for frequency
load profiles, determine the competitiveness of electricity regulation and spinning reserves. Table 3.1 displays
and gas DER technologies. The third and final case results from two price scenarios for a typical July day:
Note: The table shows monthly weekday scaled estimates for optimal energy cost, ancillary services revenue, and net operating cost for
two July price scenarios in which a medium-sized office pays either the LMP or average retail rate for electricity.
Scenario A features higher ancillary services prices Ancillary services revenue by itself may not be a
throughout the day, relative to Scenario B. In addition, sufficient economic motivator for owners of small
in each scenario, the medium-sized office building is office buildings to invest in certain HVAC systems
subject to either hourly LMPs or a flat retail rate of and associated technology and control systems.
14.5 cents per kilowatt-hour (¢/kWh) for its energy However, facilities with metering and required building
automation systems (BASs) might
see additional benefits in other
demand response or energy cost
Building loads, such as loads for heating and cooling, minimization programs.
can be managed to reduce electricity bills by reacting to
hourly energy prices and providing operating reserves. As this example illustrates, the
effect of tariff design is an important
determinant of the quantity of
ancillary services provided by
office buildings. Under flat volumetric retail rates for
purchases (reflecting average prices for July 2015). The
electricity, medium-sized office buildings provide a
office building is able to recover 122–141 percent of its
lower quantity of ancillary services than under dynamic
HVAC energy costs when it pays the LMP, resulting in
energy market prices (LMPs). This is because under a
a negative net operating cost. If subject to the average
flat volumetric retail rate, the energy consumed by the
retail rate for electricity, the office building saves 13–26
building is valued at the retail tariff while the regulating
percent of electricity costs in both scenarios. In this
energy provided to the wholesale market is valued at
case, the flat rate prevents smarter energy consumption
the lower wholesale LMP.
that could enable the building to optimize its usage
in response to the changing value/cost of electricity
throughout the day.
7 For this analysis we use projected costs from various US agencies. Projected total
costs for heat pumps are based on the US Energy Information Administration (EIA)
report, Updated Buildings Sector Appliance and Equipment Costs and Efficiencies,
April 2015 (available at: www.eia.gov/analysis/studies/buildings/equipcosts/
pdf/full.pdf). Projected costs for residential combined heat and power engines
are based on ARPA-E’s “Generators for Small Electrical and Thermal Systems -
GENSETS Program Overview” (available at: arpa-e.energy.gov/sites/default/files/
documents/files/GENSETS_ProgramOverview.pdf). Finally, projected costs for
residential CHP fuel cell systems come from the US Department of Energy’s “Fuel
Cell Technologies Office Multi-Year Research, Development, and Demonstration
Plan - 3.4 Fuel Cells,” Updated November 2014 (available at energy.gov/sites/prod/
files/2014/12/f19/fcto_myrdd_fuel_cells.pdf).
8 Further description of the methodology and results can be found in Dueñas, P. and
K. Tapia-Ahumada, “Interplay of Gas and Electricity Systems at Distribution Level,”
Utility of the Future Memo Paper at: energy.mit.edu/uof.
Note: Annual savings and expected internal rates of return are shown for two electricity-to-gas price ratio scenarios (we assume high
and low price ratios of 3.6 and 1.9, respectively) and are calculated with respect to a reference case based on a high-efficiency gas-fired
condensing boiler. Calculations consider technology cost projections,10 not current costs. ICE — internal combustion engine; SE — Stirling
engine; PEMFC — polymer electrolyte membrane fuel cell; SOFC — solid oxide fuel cell; and AA HP — air-to-air heat pump.
9 Refer, for example, to Hawkes, A., E. Entchev, and P. Tzscheutschler. 2014. “Impact
of Support Mechanisms on Microgeneration Performance in OECD Countries.”
On behalf of International Energy Agency — Energy in Buildings and Communities
Programme (IEA- EBC Annex 54) published by Technische Universität München,
Germany, October 2014. In this reference, the technology cost of an internal
combustion engine (ICE) of 1.2kWe is around $8,000, the cost of a Stirling
engine (SE) of 1.0kWe is approximately $6,500, the cost of a polymer electrolyte
membrane fuel cell (PEMFC) of 0.75kWe is over $11,500, and the cost of a solid
oxide fuel cell (SOFC) of 0.70kWe is above $15,000. 11 Further description of the methodology and results can be found in Huntington, S.
10 Assumed technology costs are as follows: IC-$4,500/kWe; SE-$4,500/kWe; “Case study: Battery Storage vs Flexible Demand.” Utility of the Future Memo Paper
PEMFC-$2,500/kWe; SOFC-$2,500/kWe; AA H-$4,000/kWe. at: energy.mit.edu/uof.
Figure 3.4: The Impact of Flexible Demand and Battery Cost on Battery Profitability in Texas
Note: Levels of network investment are shown for connecting different penetrations of low-carbon technologies to the British distribution
system by 2050. Source: EA Technology (2012)
The UK analysis illustrates how passive grid operation This necessarily implies coupling network planning and
and a fit-and-forget approach to network planning lead operation so that network constraints are managed not
to unnecessarily large costs, particularly in systems only at the planning and connection stages but also
with high DER penetration. In such systems, costly grid during real-time operation. Active network management
reinforcements will be triggered by situations that occur in turn relies on the extensive utilization of ICTs to enable
relatively infrequently (e.g., a few times per year) such as the advanced monitoring and control capabilities of grid
very high injections or withdrawals from solar DERs. In assets and network users. One of the greatest challenges
these cases, active management or curtailment over short for regulating distribution companies in the future will
periods of time can often reduce the costs associated with be managing their close interaction with DERs that they
such peaks. Moreover, the need for grid reinforcements do not directly control, but that assist in the operation
can result in long lead times for connecting new network of the distribution grid. The flexibility associated with
users. To avoid such inefficiencies and to facilitate an DERs may become essential for day-to-day network
efficient transition to a distributed system, distribution operation. Distribution companies will have to become
companies will need to adopt innovative ways to manage true “system operators” in the sense that they acquire
their networks. network services, such as forward commitments of DER
13 These parameters are set forth in the German Renewable Energy Sources Act
or EEG.
As shown in Figure 3.6, in some cases the total economic impact of meeting a significant (greater than
20 percent) fraction of load with distributed PV generation can be large — in the most extreme case,
even doubling the cost of the network.
This raises important questions. How should these costs be allocated? If the network users that
generate the costs are required to pay for them, would they still decide to invest in PV generators?
Are there “non-wires” solutions that can lower total system costs?
14 For more details about the methodology used, we suggest reading Chapter 7 of MIT’s Future of Solar Energy study.
Despite this cost-mitigating effect, procuring distributed storage is not necessarily the most cost-
effective approach to manage the network impacts of distributed solar, for two reasons. First, it is not
clear whether storage is cheaper than conventional network solutions (at current battery prices, storage
is almost certainly not cheaper than conventional network solutions in the majority of cases). Second,
there could be other, more cost-effective solutions such as PV curtailment or demand response. As
we propose in this study, a cost-reflective system of prices and charges would reveal the least-cost
solutions to meeting network challenges.
Figure 3.6: Impact of Distributed Solar PV on Network Cost with Different Levels of
Energy Storage
Note: The chart shows the cost-mitigating effect of energy storage (represented by a storage factor, SF). Source: MIT (2015)
PV inverters are a class of technologies that provide voltage support by changing the reactive
power injection and absorption levels depending on the active power output of the solar PV system.
When coupled with PV inverter control strategies, line losses associated with solar PV generation
could be significantly reduced, while line transfer capacities and the loadability of networks (that is,
the maximum load that can be supplied at peak hours) could be significantly increased.15
Figure 3.7: A 34-Node IEEE Test System (Feeder) with Five PV Generators
29
12 28
Grid
27
11
1 10
VR1 24 26
2 3 4 6 7 8 9 13 14 25 30 31 32
23
5 33
20
21 22
VR2 34
19
15 16 17 18
Note: The installed PV capacity at each node containing a generator is the peak demand at that same node.
Figure 3.7 shows a 34-node IEEE test system or feeder with five PV generators at different nodes. The
maximum transfer capability of each branch or line at the maximum feeder load without PV is limited by
voltage constraints in the feeder.
15 For more details about the methodology used, please refer to Abdelmotteleb I., T. Gómez, and J.P. Chaves. 2016. “Benefits of PV Inverter Volt-Var Control on Distribution Network
Operation.” IIT-Comillas Working Paper. www.iit.comillas.edu/html20/publicacion/mostrar_publicacion_working_paper.php.en?id=284.
Figure 3.8: Variation in the Transfer Capabilities of Feeder Branches with PV Integration
80
PV PV + Inverter Control
70
60
Loadability Increase (%)
50
40
30
20
10
0
0 25 50 75 100
PV Output with Respect to PV Installed Capacity at Peak Hours (%)
16%
PV PV + Inverter Control
14%
12%
10%
Losses (%)
8%
6%
4%
2%
0%
0 25 50 75 100
Note: In Figure 3.9a (top) network loadability is measured as the percent demand increase that can be supplied at peak hours before the feeder
reaches its transfer capacity. In Figure 3.9b (bottom) network losses are shown as a percent of demand, and at different levels of PV generation
with and without PV volt-var inverter control strategies.
In sum, the integration of distributed generation (PV in this example) with volt-var inverter control strategies,
in distribution networks with transfer capabilities that are mainly limited by voltage constraints (such as
networks in rural areas), can improve system efficiency by reducing energy line losses and increasing network
loadability, thereby potentially deferring the need for network reinforcements.
The best-known effect of high levels of VRE penetration In power systems where most of the variability of VRE
is the so-called merit order effect: Zero variable cost generation will be absorbed by thermal generation,
generation displaces generators with higher variable these thermal units will be forced to rapidly change their
costs, thereby immediately reducing wholesale output and to start up and shut down more frequently.
market prices. Because VRE technologies are only Consequently, costs associated with the cycling of
intermittently available, the reduction in prices occurs thermal plants will increase and, for large penetrations
only in those periods when solar or wind generation is of VRE technologies, may offset any cost reductions
available, although it is likely to lead to lower average associated with the merit-order effect. Because of these
wholesale prices. thermal cycling costs, overall energy costs can increase.
Figure 3.11 illustrates this dynamic using a simulation
of the impact of solar generation in ERCOT, the power
system that serves most of the state of Texas.17
Note: The figure shows total short-term production cost and average short-term production cost for thermal generators only associated
with increasing solar penetrations in ERCOT, a thermal-dominated system. (MIT 2015)
In sum, VRE technologies have a twofold effect on spot unpredictability of VRE generation, can lead to electricity
prices. Prices can increase due to the cost of thermal price divergence between sequential markets, such as
generation cycling in certain periods and prices can day-ahead and intraday or real-time markets. A review of
decrease when VRE production displaces more expensive wholesale market rules is needed to address these issues;
generation in other periods. In periods when high VRE we return to this topic in Chapter 7.
production displaces costlier production, prices can fall
Active participation of DERs in wholesale markets
to almost zero, reflecting the fact that there is excess
electricity supply. Prices may even become negative when Current electricity wholesale markets present numerous
inflexible thermal generators are unable to come offline barriers to the participation of DERs. In most cases, these
barriers simply result from technology evolving at a faster
pace than electricity market rules and
regulations. The participation of DERs may
Increasing penetration of variable renewable energy be hindered by a lack of clear rules or by
sources, such as solar PV and wind, will increase the rules that were designed for large traditional
need for flexible demand and generation to adapt to resources and have not been updated.
their intermittent output. Pioneering experiences in DER integration
have illuminated the most urgent reforms
required to enable DER participation in
wholesale electricity markets. This section
(because it is technically infeasible or uneconomic), draws on some of these early experiences to highlight the
and renewable generation is benefitting from priority most relevant barriers that exist today. It also bears noting
dispatch rules and/or production subsidies. This dynamic that reforms to allow for DER participation are only the
can have the overall effect of increasing price volatility first step in a necessary redesign of electricity markets.
in spot electricity markets, which, combined with the Such reforms must be followed by efforts to ensure the
Definition of market products The challenges associated with DER participation are
In order for DERs (aggregated or not) to compete very different across market segments. For example,
efficiently with conventional resources they must be able in capacity markets, the key challenge is assessing the
to participate in all relevant markets, from long-term contribution to firm capacity by resources with very
capacity markets to real-time and balancing markets. different technical characteristics. Traditionally, many
capacity markets only allowed
conventional generators to join.
More recently, demand response
Current market rules must be examined to enable the has been allowed to take part in
participation of DERs and to level the playing field between a growing number of capacity
centralized and distributed resources. markets due to its ability to reduce
energy loads and alleviate stress
on the grid during emergency
conditions.18 However, assessing
In markets that allow DER participation, DER involvement the actual capacity value of VRE and storage technologies
may still be hindered by market products that were still constitutes a barrier to tapping their potential
designed with the capabilities of conventional resources contribution toward the overall capacity of the system.
in mind and that therefore fail to fully reflect the needs
of the power system or properly take into account the
18 Demand response (DR) programs have been the primary way in which DERs — not
capabilities and limitations of DERs. The remuneration
connected at the bulk power level — have had an opportunity to participate in US
received for delivering these ill-designed (or outdated) wholesale markets.
While some DERs were previously able to enter the market in limited ways, CAISO is the first grid
operator in the United States to spell out a process for grouping various DERs to reach the threshold for
market participation.
According to CAISO’s President and CEO, “It’s critical that the ISO collaborate with distribution system
operators, regulators and market participants to harness these valuable distributed resources.”*
* www.marketwired.com/press-release/california-iso-leads-historic-push-for-distributed-energy-resources-2132153.htm
While SEDC has acknowledged progress by some EU member states in implementing network
codes that are favorable to demand response, it gave other member states a negative grade based
on five remaining regulatory barriers to the deployment of DR resources: (1) demand response is
not accepted as a resource; (2) baseline load estimates are inadequate and/or non-standardized,
making it impossible to measure the amount of load reduction attributable to demand response; (3)
technology-biased program requirements are present; (4) aggregation services are not fully enabled;
and (5) standardized processes between balancing responsible parties (BRPs) and demand response
aggregators are lacking.
Figure 3.13: Distribution LMPs Using Marginal Loss Values Similar to the Spanish Distribution System
19 For more details on the modeling used to produce this figure, see Appendix A.
locational marginal prices for electricity at different voltage priorities — must be established.
• Detect (identify the occurrence of a cybersecurity event, and enable timely responses);
• Recover (restore the capabilities or critical infrastructure services that were impaired through a
cybersecurity event).
Solar Roof
CLOUD
Solar Irradiation
Maintenance
Market Hourly Prices
TSO Signals
DSO Signals
Building Contract
Fan Coils External Data
DHW
Thermostat
Accurate Heating
Storage
+
HOT CYCLE
METHANE CCHP
COLD CYCLE
Distribution
Power Grid
Data generated at each building would converge (in voltage limit violations, imminent transformer capacity
the cloud) with information about the power system, limits, etc.). The distribution system operator may also
such as the spot price of wholesale energy, the prices of send emergency signals that limit the power that can
operating reserves, information about requests for bulk be injected or withdrawn by any user for a prescribed
power system support (e.g., to meet expected ramps or time interval. However, the use of electricity prices that
maintain appropriate frequency levels), information about embed information about the availability of generation
the activation of commitments for demand reductions, or about active network constraints should minimize
and information pertaining to feeder-level distribution (though probably not eliminate) the need for these types
system markets and operating conditions (e.g., imminent of emergency signals.
Interdisciplinary Systems Laboratory. August 2016. web. Insurance Implications of a Cyber Attack on the US Power
NIST. 2014a. Framework for Improving Critical Scottmadden Management Consultants. 2015. Energy
Infrastructure Cybersecurity Version 1.0. National Industry Cybersecurity Report. July 2015.
Institute of Standards and Technology. February 12, 2014.
SEDC. 2015. Mapping Demand Response in Europe Today.
www.nist.gov/cyberframework/upload/cybersecurity-
September 30, 2015. www.smartenergydemand.eu/
framework-021214.pdf.
wp-content/uploads/2015/09/Mapping-Demand-
NIST. 2014b. Interagency Report on Guidelines for Smart Response-in-Europe-Today-2015.pdf.
Grid Cybersecurity. NISTIR 7628: Volumes 1-3. National
Shelar, D. and S. Amin. 2016. “Security Assessment
Institute of Standards and Technology. September 2014.
of Electricity Distribution Networks under DER Node
Nourian, A. and S. Madnick. 2015. “A Systems Compromises.” IEEE Transactions on Control of
Theoretic Approach to the Security Threats in Network Systems.
Cyber Physical Systems: Applied to Stuxnet.” IEEE
Smith, E., et al. 2016. “Going Beyond Cybersecurity
Transactions on Dependable and Secure Computing. MIT
Compliance: What Power and Utility Companies Really
Sloan School Cybersecurity Interdisciplinary Systems
Need to Consider.” IEEE Power and Energy Magazine 14(5):
Laboratory. December 2015. web.mit.edu/smadnick/
48-56. September/October 2016.
www/wp/2015-07.pdf.
Stockton, P. 2016. “Superstorm Sandy: Implications
NRC. 2012. “Terrorism and the Electric Power Delivery
for Designing a Post-Cyber Attack Power Restoration
System.” In Mitigating the Impact of Attacks on the Power
System.” The Johns Hopkins University Applied Physics
System. National Research Council. Washington, DC: The
Laboratory. NSAD-R-15-075.
National Academies Press. doi:10.17226/12050. www.
nap.edu/catalog/12050/terrorism-and-the-electric- Taft J.D. and A. Becker-Dippman. 2015. Grid Architecture.
power-delivery-system. Richland, WA: Pacific Northwest National Laboratory.
January 2015. www.pnnl.gov/main/publications/external/
NYDPS. 2014. “Reforming the Energy Vision. NYS
technical_reports/PNNL-24044.pdf.
Department of Public Service Staff Report and Proposal.”
State of New York Department of Public Service. CASE Ruester S., I. Pérez-Arriaga, S. Schwenen, C. Batlle, and
14-M-0101. Proceeding on Motion of the Commission in J.M. Glachant. 2013. From Distribution Networks to Smart
Regard to Reforming the Energy Vision. April 24, 2014. Distribution Systems: Rethinking the Regulation of European
Electricity DSOs. THINK Project, Topic 12. Florence, Italy:
Ofgem. 2013. “Strategy decisions for the RIIO-ED1
European University Institute.
electricity distribution price control. Business plans and
proportionate treatment.” As Supplementary annex to Ton D.T. and M.A. Smith. 2012. “The U.S. Department of
RIIO-ED1 overview paper. Office of Gas and Electricity Energy’s Microgrid Initiative.” The Electricity Journal 25(8):
Markets. 26b/13. March 4, 2013. 84-94. October 2012.
04
A Comprehensive and Efficient
System of Prices and Regulated
Charges for Electricity Services
4.1 The Importance of Getting result in more efficient response of demand connected
at all voltage levels and also in enhanced efficiency of
Prices and Charges Right operations and investment in the bulk power system. The
Distributed technologies allow the agents in a power growing integration of DERs increases the importance of
system to meet their own energy needs and to deliver well-designed economic signals and the ramifications of
electricity services to the system. Distributed energy poorly designed signals.
resources (DERs) bring new options for service provision
DERs are a new class of potential competitors for the
and enable demand for electricity services to become
provision of electricity services traditionally delivered
increasingly price-responsive. Well-designed prices and
by centralized generators and network assets. Economic
charges become ever more important in this environment,
signals that accurately reflect the costs and values of
as agents are increasingly capable of adapting their
utilizing varying resources for service provision will enable
behaviors to power system conditions at specific
the creation of a level playing field on which centralized
locations, during particular times, and in relation to what
and distributed energy resources can compete and
specific services are being consumed or provided.
complement one another in an efficient resource portfolio
Acknowledging the importance of designing economic (Pérez-Arriaga et al. 2013). These economic signals serve
signals that reflect power system costs and serve as to coordinate planning and operational decisions related
efficient signals for distributed decision-makers is not to all resources and make it possible to achieve efficient
new (Bonbright 1961; Schweppe 1978; Schweppe et outcomes — particularly if the resources are numerous
al. 1988). Even in the absence of DERs, there are clear and distributed. Ideally, these signals will elicit efficient
benefits to using efficient economic signals, since they responses from all resources, at all times, no matter
where they are located in the network.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 75
Centralized and distributed resources are sited in This study advocates the need for a
different locations and have different sizes and temporal comprehensive system of prices (for those
patterns for both production and consumption. The
services provided in markets) and regulated
only way that these two categories of resources can
charges (for remuneration of the network
jointly and efficiently operate, expand, compete, and
collaborate, is in the presence of a comprehensive
activities and any policy costs included in
system of economic signals — market-determined prices electricity rates); such a system can act as
and regulated charges1 — with adequate granularity to the nervous system of the power sector,
capture important variations in the value of the specific coordinating the actions of many disparate
services across time and location. This study advocates providers and consumers of electricity services
the need for a comprehensive system of prices (for those
by communicating prices and charges that
services provided in markets) and regulated charges (for
reflect time- and location-specific conditions.
remuneration of network activities and any policy costs
included in electricity rates); such a system can act as
the nervous system of the power sector, coordinating the
To date, most power systems have employed simplified
actions of many disparate providers and consumers of
prices and charges to allocate power system costs to
electricity services by communicating prices and charges
most electricity customers; these are proving inadequate
that reflect time- and location-specific conditions.
in the face of increasing penetration of DERs and
Ideally, these prices and regulated charges will reflect opportunities for flexible demand (see Section 4.3).
all the operating conditions and investment needs of the Such simple methods exhibit limited (if any) temporal or
system, and the markets for electricity services will allow spatial differentiation and result in tariffs that typically
the participation of all system users. For example, to fully bundle costs of all of the services that customers receive
realize the possible benefits of distributed storage or in a nontransparent manner. As a result, some electricity
price-responsive demand,2 energy prices that sufficiently users are making inefficient investments in DERs and
reflect temporal and spatial variations in the costs of are over- or undercompensated for the services that
meeting electricity demand or providing electricity they provide to the power system. At the same time,
generation must be communicated to distribution system many more opportunities to deliver value are being
users, aggregators (including retailers), and/or automated left untapped because of inadequate compensation.
energy management systems. These economic signals The result is a power system that costs more than is
must also encourage the placement and operation of necessary — a loss in overall efficiency — and is unable
DERs when and where they can prove more cost-effective to reveal and appropriately compensate the value that
than centralized providers of services. DERs and price-responsive demand can provide, as will be
shown in the following sections.
1 Tax incentives and other policy incentives can be part of the ensemble of economic
signals that the power system agents are subject to, beyond the basic system of
prices and charges.
2 In this report, “price-responsive demand” means the efficient spontaneous
response of demand to the existing price. See Chapter 2 for precise definitions of
“price-responsive demand” and “demand response.”
Those looking for greater detail on how electricity prices Furthermore, prices and charges for electricity services
and charges can be computed with an increased level should be nondiscriminatory, and thus agnostic to the
of granularity should also read Section 4.4, which first particular activities for which the network is used, or
presents an efficiency-driven approach to the pricing “technology neutral.” Accordingly, any cost-reflective
of electricity services and the design of network and component of prices and regulated charges should
policy charges, then considers how practical challenges be based exclusively on the individual injections and
to the implementation of such an “efficient ideal” withdrawals at the network connection point, regardless
can be addressed via implementation of proxies with of the specific technology producing those injections or
manageable complexity. withdrawals.4 Such an approach differs from a “value of
resource” approach that provides incentives targeted
Finally, those who want to be convinced of the increases
toward specific resources, and is instead similar to
in efficiency gained by avoiding common errors in tariff
a “value of service” approach in that the costs and
design and progressively improving the granularity in time
benefits — or overall value to the power system — of any
and location conveyed by prices and charges should read
resource can be fully captured via prices and charges for
Section 4.5, which provides examples of the benefits3 of
services such as those described in Chapter 2 (NARUC
employing prices and charges that more and more closely
2016). In addition, cost-reflective prices and regulated
approximate the efficient ones.
charges should be symmetrical, with a marginal injection
4 From the point of view of the power system, it does not make any difference
whether a change in the power injected or withdrawn at a given connection point
and moment in time has been caused by turning off an appliance, discharging a
battery, or injecting more power from a photovoltaic panel or micro cogeneration
unit. The impact on the overall system does not depend on the technology involved,
so prices and charges shouldn’t either. Moreover, it is a hopeless task to try to
intrude behind the meter and apply different rates depending on the nature of the
device used. There may be reasons to favor or disfavor particular technologies
besides power system efficiency, but efficiency will be sacrificed if this is done.
The allocation of residual costs is a different topic (addressed later in this chapter)
3 This study also comments on the costs and complexity associated with increasing where the most efficient cost allocation criterion may distinguish between types of
granularity in time and location, but to a smaller extent. consumers, for instance according to their price elasticity.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 77
Figure 4.1: Smart Meter Deployment
4.1a: Smart Meters Installed in the United States as 4.1b: Expected Level of Smart Meter Deployment in the
of 2015 European Union in 2020
Source: JRC Smart Electricity Systems and Interoperability (2016) Source: Institute for Electric Innovation (2016)
at a given time and place compensated at the same To establish a level playing field for all resources,
rate that is charged for a marginal withdrawal at the cost-reflective electricity prices and regulated
same time and place.5 Designing prices and charges in
charges should be based only on what is
a way that is technology-neutral and captures relevant
metered at the point of connection to the power
differences in the costs and benefits offered by injections
and withdrawals effectively obviates the need for defining
system — that is, the injections and withdrawals
a DER-specific customer class and subclasses. Any of electric power at a given time and place,
subsidies or penalties — or other kinds of explicit support rather than the specific devices behind the
or burden on specific technologies or resources — can be meter. In addition, cost-reflective prices and
legitimate and justified in many cases (for instance, to regulated charges should be symmetrical, with
internalize some externalities), but they should be applied
injection at a given time and place compensated
in addition to and separately from the economic signals
at the same rate as that charged for withdrawal
that will be developed in this chapter.
at the same time and place.
other energy-related services, such as operating reserves or firm charges also enable the
economic sustainability
capacity; (3) charges for network-related services; and (4) charges
of regulated services
to recover policy costs.
via recovery of
regulated costs
reserves), as well as other regulated taxes and items that (such as distribution
regulatory authorities decide to include in the electricity network costs and policy costs). As will be shown later
tariff. Among the diverse criteria advocated by experts in this chapter, while prices and charges that provide
and embodied in most tariff designs, two principles are economic signals by reflecting marginal or incremental
dominant: allocative efficiency and sufficiency to recover costs contribute to recovery of regulated network
the regulated costs. Only the second one is routinely met
6 costs,7 such prices and charges alone are unlikely to
in practice. be sufficient for full cost recovery. Regulated costs not
recovered via cost-reflective prices and charges — the
A comprehensive system of prices and charges that “residual costs” — should be recovered in a minimally
provides economic signals and enables cost recovery distortive manner. In addition, electricity bills have often
consists of the following core elements: (1) a price for been a convenient tool used by politicians to allocate
electric energy (active and reactive, in principle); costs derived from energy policy objectives such as
(2) prices or charges for other energy-related services, energy efficiency, climate change mitigation, or income
such as operating reserves or firm capacity; (3) charges for redistribution. These taxes and a large component
network-related services; and (4) charges to recover policy of policy-related costs represent a significant part of
costs (such as taxes and costs incurred supporting energy electricity bills in some jurisdiction (Figure 4.2). Any such
efficiency or renewable targets). costs not affected by changes in electricity consumption
Following the first objective, efficient economic signals add to the amount of “residual costs,” and must also be
should try to capture and reflect the marginal or recovered in minimally distortive manner.
incremental costs of the production and utilization of If not carefully designed, charges to recover the residual
electricity services. Such signals serve as the key tools costs may distort the economic signals received by
with which to coordinate all the planning and operational power system users, leading to decisions that increase
decisions made by the diverse range of power sector the overall cost of power systems and/or shift policy and
agents to achieve efficient outcomes. For services provided network costs onto other users. For example, inefficient
competitively, the corresponding markets generally provide installation and use of behind-the-meter generation
the required prices. For other services, regulated charges may be incentivized if reducing volumetric electricity
must be designed to send efficient signals reflecting each consumption allows users to reduce their payment
user’s marginal or incremental contribution to the regulated of policy costs or regulated network charges that are
costs (such as network capacity). allocated via volumetric rates but are not a direct function
of volumetric consumption (see Boxes 4.2 and 4.3 for
more details). This study proposes methods to allocate
residual costs and in a manner that minimizes economic
6 Besides allocative efficiency and economic sustainability, other commonly
distortions.
mentioned criteria are: transparency in making public the procedure that is
followed to compute prices and charges; as much simplicity as possible, to facilitate
understanding and acceptance, without compromising other more important 7 For example, the differences among locational marginal prices (LMPs) of energy
criteria; consistency with the adopted regulatory framework, specifically the level in the nodes of the network, due to the effect of losses and congestion, contribute
of liberalization and unbundling of the different activities; and stability to minimize to network cost recovery because they generate revenues when charging for
deviation from the status quo as a basic regulatory principle meant to reduce consumption and paying generation at their individual LMPs.
the risk perceived by the agents in the power sector and to enhance the social
acceptability of the electricity tariffs.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 79
Figure 4.2: Breakdown of Residential Electricity Bills in Different Jurisdictions in 2014–2015
Care must be taken to minimize distortions On the basis of fundamental principles of economics and
from charges that are designed to collect power systems, this study proposes a framework for the
design of prices and charges according to what can be
taxes, recover the costs of public policies
labeled an “efficient ideal” approach (i.e., an approach
(e.g., efficiency programs, subsidies for
based on achieving economically efficient outcomes).
renewable energy, cross-subsidies between The implementation of such an approach under a range
different categories of consumers, etc.), and of specific technical, regulatory, policy, and social
recover residual network costs (i.e., those contexts may present numerous challenges; therefore,
network costs that are not recovered via cost- alternative approximate or quasi-ideal schemes will also
reflective charges). be considered.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 81
communication technologies available for metering, data charges. Retail liberalization brings another layer of
management, and communication. Progress has also been complexity, since this enables end consumers to choose
made in understanding the complex economics of the their suppliers. This forces the regulated electricity tariff
power sector. to be reduced to the “access tariff,” i.e., the regulated
component of the tariff that is meant to recover the
Determining the economic signals for each agent when
regulated cost components: transmission and distribution
all the activities involved in supplying electricity are
network costs and policy costs to which all consumers
performed by a single vertically integrated company and
should contribute. Costs attributed to the competitive
the rest of the agents are pure consumers, mandated
components of electricity supply — electricity production
to purchase the electricity from that same company, is
and the commercial activity of retailing — are then freely
a relatively straightforward task. The company receives
established by the independent retailers and incorporated
remuneration based on the efficient cost of electricity
in the “integral tariffs” (i.e., those that include all
supply as estimated by the regulator. The end consumers
components) that retailers offer to the customers. In most
cover this cost via regulated electricity tariffs. The
jurisdictions there is an “integral default tariff” designed
prevalent structure of these tariffs is very simple, with
by the regulator as a default option for customers who
no time or spatial differentiation, since it is designed
do not want to shop around for a better option. Again, in
for price-inelastic consumers with similar and stable
most jurisdictions, both the default and the competitive
consumption patterns. However, it is possible to maintain
integral tariffs are very simple, with low granularity levels.
this traditional regulatory framework while exposing
customers to more differentiated (granular) economic This study confronts the ultimate layer of complexity,
signals that would allow them to respond by modifying which occurs when any agent connected to the power
their demand patterns and installing storage or generation grid at any voltage level cannot be simply classified as
behind the meter, as any other customer could do in a either a consumer or a generator, and each agent has
more market-based regulatory environment. Improved the capability of providing — by itself or in aggregation
economic signals could also enable the provision of with other agents — all kinds of electricity services to
electricity services by DERs to the incumbent company. any of the other agents in the power system (these may
Large customers of traditionally regulated companies include transmission or distribution system operators,
have long been subject to much more granular signals producers, consumers, or any combination thereof). In
in the price of electric energy, charges for peak demand such circumstances, which may come to characterize the
(either for demand coincident with the system-wide power sector in the next decade or beyond, the need for
peak or for individual peak loads), deviations with more sophisticated economic signals becomes apparent.
respect to pre-established consumption patterns, and Regulation must anticipate this scenario (which is already
interruptibility contracts related to firm capacity services. incipiently present in some power systems) with a sound
In some jurisdictions more advanced tariff schemes have approach to the design of prices and charges.
been implemented for small and medium customers
as well: time-of-use (TOU) pricing, critical peak pricing 4.3.1 A cursory review of current
(CPP) and, very rarely, real-time pricing for the energy electricity tariff practices9
component of the tariff.
While complex wholesale and “ancillary services”
In power sectors that have been restructured and markets have evolved in many jurisdictions that capture
liberalized at the wholesale level, the activities of the marginal price of electricity services in varying
wholesale generation, transmission, system operation, degrees of detail, the electricity tariffs that are seen by
and distribution are unbundled. Therefore, the prices and end consumers and DER owners have been uniformly less
charges for the services provided among these activities complex to date. This is particularly true for residential
and to end consumers have to be established either by
specific markets that reveal the prices or by regulated 9 The profound changes in tariff design that the presence of DERs requires — and
that are made possible by the availability of advanced meters — suggest it is not
worthwhile to spend much time dwelling on the analysis of present electricity
tariffs.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 83
pricing practices for large C&I customers provide valuable more granular prices and charges, accounting for the
guidelines for constructing a system of retail-level actual system operating conditions and applying prices
economic signals that can enable efficient utilization of and charges to the individual injections and withdrawals
DERs in electricity service provision. Indeed, large C&I at any given time and connection point. Some power
customers receive more granular prices precisely because systems, such as that in use in Spain, have started to
they have historically had more options for responding to apply wholesale day-ahead hourly energy prices to
prices, which increases the costs of ill-designed tariffs. residential customers in the integral default tariff.
The same rationale is now becoming true for smaller
customers as well, as opportunities for DER adoption and
price-responsive or flexible demand proliferate.
NEM is widely used in the United States and also in some European countries. Figure 4.3 shows the net-
metering policies that were in place in 41 states of the United States as of July 2016. In seven European
Union countries, electricity customers can net their demand with electricity produced from solar panels
any time during that same year (European Commission 2015).
10 NEM is an approach used for the compensation of distributed generation and other DERs. NEM charges system users for net energy consumed (that is, energy consumed minus
energy generated during the time period) during each netting period, typically one or two months (note that, the longer the period, the more distortional the netting is).
DC
41 States + DC,
AS, USVI, & PR have
mandatory net
metering rules
Statewide distributed generation compensation rules other than net metering (4 states + 1 territory)
Source: www.dsireusa.org
Under NEM, payment for any costs recovered via volumetric charges in the electricity tariff — including
energy, network, and policy costs — is reduced or entirely avoided by distributed generation (DG) users,
leading to cross-subsidization of DG network users by customers without DG, since only energy costs
should be netted by DG production. NEM plus volumetric tariffs asymmetrically value behind-the-meter
generation above electricity at the same location that does not come from behind the meter. Also note
that DG production typically peaks at a different time than demand, and therefore customers with DG
typically use the network more than customers without DG. In addition, injections of DG power may stress
the network and lead to increased distribution network costs (MIT 2015). The full recovery of network and
policy costs requires an increase in volumetric rates, further encouraging the deployment of DG.
How can this criticism of NEM be compatible with our basic principle that “cost-reflective prices and
charges should be based only on what is metered at the connection point,” i.e., the net of the production
and consumption of all the devices behind the meter? The problem with NEM, as it is used in many
power systems today, is twofold. On the one hand the netting is done over a long period of time, as
required by the standard meters still in use today in most power systems, so that the injection of power
into the grid from a solar panel at noon on a sunny day when the energy price is low is netted with the
consumption at peak demand time on a cold evening when the energy price can be much higher. On the
other hand, when the volumetric component of the electricity tariff includes network and policy costs,
these costs are avoided when the metered injection is netted with the metered consumption.
Netting the internal generation and demand of all the devices behind the connection point to the
network during a short time interval (one hour or less) is what we propose. However, the actual
implementation of NEM — over long periods of time and accompanied by volumetric tariffs that
include network and policy costs — introduces serious distortions in the tariff system.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 85
4.4 Design of Prices and manner by which the more advanced economic signals
typically available at the wholesale level can filter down
Charges: From First Principles to all voltage levels in distribution. As explained in
to Implementation Chapter 3, in practice one can expect that — at least
for some time — intermediary agents or aggregators
As discussed at the beginning of this chapter, the main
will manage the response of residential and small I&C
objectives of a comprehensive system of prices and
customers. Such aggregators can receive the signals of
charges are to promote efficient outcomes (e.g., to
prices and charges for all the relevant electricity services
minimize the cost of desired electricity services) and
corresponding to all their customers and make use of
to recover regulated network and policy costs. In this
diverse estimates or forecasts of demand, weather, plant
section, the fundamental principles of power system
availability, state of charge of storage, network flows, and
economics and engineering are applied to provide the
other information in order to decide how to participate in
framework for an “efficient ideal” approach to designing
the several markets for services. This participation may
prices and charges — that is, an approach based on
include making use of the responsive demand and DERs
achieving the most economically efficient outcome, while
of its customers, sending to each one simple signals to
guaranteeing the recovery of the regulated costs.
obtain efficient responses at all times, and maintaining
As indicated in the introduction, reading this section prespecified levels of comfort or meeting other customer
can be omitted by those who are just interested in the needs. A likely scenario is that competing aggregators
principles (Section 4.2) and the recommendations will offer retail customers lower flat prices in return for
(Section 4.6) as well as by those with a good knowledge being able to manage and automate a defined degree of
of the design of prices and charges in a market flexibility in the timing of DER demand and operation.
environment, although such readers may wish to at
The following sections will cover the economic signals
least skim the sections on firm capacity (Section 4.4.3),
related to key electricity services (e.g., those services
allocation of network charges on the basis of responsibility
responsible for the largest share of power system costs),
in new network investments (within Section 4.4.4), and
making use of the breakdown of prices and charges that
the treatment of residual regulated costs (both residual
was presented in Section 4.2. First, we will discuss the
network and policy costs) (first part of Section 4.4.5).
prices for electric energy, operating reserves, and firm
In practice, the level of granularity in space, time, and the capacity. Next, we will discuss methods for determining
disaggregation of services reflected in these economic forward-looking cost reflective network charges and policy
signals will depend on several factors, including: trade- costs charges and finally how to allocate the residual
offs among the efficiency gains associated with increased regulated costs (both network and policy residual costs).
granularity; the availability of communication and
computational technologies and implementation costs 4.4.1 The price of electric energy
necessary to increase granularity; and the acceptance
The economically ideal price for electric energy given
of different prices and charges by various agents.
current power system conditions is embodied in the
Therefore, alternative simplified approaches must also
“nodal” or locational marginal price of electric energy
be considered. Where possible, we present a continuum
at each point of connection and at each moment in
of options and underscore the key trade-offs relevant to
time, calculated on the basis of the costs of supply and
evaluating alternative designs for prices and charges (see
the demand in response to these prices (Schweppe
Section 4.6 on implementation).
et al. 1988). Locational marginal pricing is the ideal
All economic signals at all voltage levels are relevant that power systems should try to achieve, while taking
and will be considered. However, our focus will be on the into consideration the implementation costs and the
prices and charges that apply to electricity customers incremental benefits of progressively increasing the
and DERs connected in the distribution network and the temporal and spatial granularity of energy prices.
Accurately computing locational marginal prices (LMPs) network, or zones of the transmission system where the
at the connection point of each agent anywhere in the nodal LMPs typically have similar values at all times).
grid is a long way from current practices in tariff design While simplifications in any of these three dimensions
(see Section 4.3). Yet, any sensible simplifications are are unavoidable in practice, this section will illustrate
possible, and our objective in this chapter is to understand that some level of granularity is necessary to capture
the implications of each option and to make a choice that significant differences in the marginal cost or value of
brings more benefits than costs in each particular context. energy.
There are three major possibilities for simplifying the ideal 4.4.1.1 The temporal dimension
reference LMP: (1) to make use only of the active (also
First, the marginal cost of electrical energy varies with
known as “real”) power component of the LMP, ignoring
time, and it can do so significantly. These variations arise
reactive power in energy prices11; (2) to use average
from changes in load patterns and generation costs.
values of LMP over a period of time, which could range
The changes in the cost of energy over the course of a
from as little as five minutes to as much as one or more
day or other time period are incompletely reflected in or
years; and (3) to use average values of the LMP over a
approximated by existing rate structures like flat annual
geographical portion of the network, which could typically
or multiannual rates, time-of-use (TOU) rates, and critical
range from the individual connection point of each
peak pricing (CPP) tariffs.12 As an illustrative example,
agent (i.e., no simplification) to the entire considered
Figure 4.4 shows one week of zonal settlement prices
power system, with intermediate options (e.g., applying
in the Austin Energy load zone of the Electric Reliability
average LMP values at the head of the low-voltage
Council of Texas (ERCOT) system. The settlement price
feeder, distribution zones with homogeneous network
characteristics, individual nodes of the transmission
12 Under TOU rates, customers are charged different static, predetermined rates in
each of multiple time blocks (e.g., different days and hours of each day). Once set,
11 The instantaneous power (product of the voltage and the current at any given time) the TOU prices are not updated or allowed to fluctuate for some fixed period. TOU
can be split mathematically into a component that has a positive average value (so rates are often coupled with a peak/off-peak demand charge. Under CPP or variable
called “active power”) and another one (“reactive power”) that withdraws and then peak pricing, customers are charged according to whether or not the retailer or
delivers power in equal amounts, so that the average value is zero. This reactive distribution utility has identified a particular hour or set of hours as “critical,” with
power component goes back and forth between the supplier and the receiver, only loads ranging from low, to standard, to high, to critical (for a particularly high
heating the wires and other network components or contributing to the violation of load). Customers receive day-ahead or few-hours-ahead notice of the anticipated
some network constraint, with no useful purpose other than being an inseparable occurrence of a critical period during which they will be charged a CPP rate. The
part of the actual complete power. CPP rate is set in advance (i.e., it does not vary in real time).
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 87
Figure 4.5: Relative Frequency (in Time) of ERCOT Wholesale LMPs Over One Year (2015)
is a load-weighted average of the LMPs at all nodes energy prices spike (sometimes an indication of a
within the load zone and also an average of the prices network constraint, signaling the need for network
computed every five minutes into 15-minute intervals investments), significant cost savings could be achieved
(it thus already includes some degree of simplification via relatively small reductions in energy use. These cost-
along both temporal and spatial dimensions). Figure 4.4 saving opportunities are not realized, however, with flat
also shows a TOU rate offered by Austin Energy, as well energy prices or most TOU rates. On the other hand,
as an illustrative CPP that is consistent with the LMPs CPP schemes can reflect most of the cost variation and
in that the designation of critical event hours coincides capture most of the available response to such price
reasonably well with the actual occurrence of peak load variations, if the identified “critical” peaks coincide with
hours. Although the TOU and CPP rates are derived from the actual periods of generation or network capacity
predicted LMP values (through estimates of peak load scarcity. Figure 4.5, which depicts the relative frequency
hours, contingencies, and other events that will affect of the value of the hourly LMPs in the Austin Energy load
LMPs), the economic signals received by the customer zone of the ERCOT system, indicates that, while instances
from TOU and CPP rates can differ significantly from the of price spikes can drive LMPs to very high values, such
actual short-term marginal prices. spikes are rare and thus idiosyncratic. Therefore, spikes
cannot easily be estimated a priori or approximated by
The mismatch between temporal simplifications for
TOU rates. This is also an indication that even hourly
energy prices, such as flat or TOU rates and the true
energy prices reflecting day-ahead market clearing
marginal cost of energy, is particularly apparent at times
prices — e.g., computed 12 to 36 hours in advance — may
of either generation scarcity or network congestion.
fail to reflect the actual operating conditions of the power
During scarcity events or other periods when wholesale
system.13 This issue is treated in detail in Chapter 7.
13 Most so-called “real-time” pricing programs do not reflect the real-time settlement
prices, but rather convey day-ahead clearing prices to end-users.
4.4.1.2 The spatial dimension due to losses may be in the realm of 5 percent to 10
percent, depending on the size of the power system and
The marginal cost of electrical energy also differs by
the specific operating conditions. However, as Figure 4.6
location within a network. These differences arise
shows for the PJM power system, vast differences in LMP
from the presence of losses within transmission and
(due to both losses and congestions) may exist at any
distribution lines, and from the occurrence of active
given time. In this case, prices in the eastern portion of
network constraints that limit power flow in sections of
the PJM system are an order of magnitude higher than
the network.14
in the western portion, illustrating the importance of
At any given time, the differences in LMPs among the spatial granularity in energy prices, particularly at times of
nodes of a transmission network can be very significant, binding network constraints.
particularly when there are active network constraints.
At the transmission level, the differences in LMP values
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 89
Figure 4.7: Frequency Distribution of 2015 PJM Average Nodal LMP Over One Year
The situation depicted in Figure 4.6 is not an unusual one. is computed for each one of the bidding zones shown in
Figure 4.7 shows the distribution of the annual average LMP Figure 4.8. These zonal prices are used instead of LMPs
value at each locational node of the PJM system in 2015. and are cleared via coordination among different power
This distribution provides clear evidence of the need for a exchanges (as distinct from system operators), neglecting
locational characterization of the energy prices in the PJM network constraints within each bidding zone. After
system. Similar differences in the marginal price of energy prices are cleared, internal transmission constraints are
at different locations are observed in other power systems. subsequently solved through diverse and uncoordinated
market mechanisms. This approach improves the
Other power systems have adopted spatially simplified
tractability of price computation by limiting consideration
approaches to LMPs, such as zonal prices (with each zone
of network constraints to those arising from a very
comprising many transmission nodes that are assumed to
simplified representation of cross-border flows, which has
share the same energy price) or the use of a single, uniform
facilitated the integration of multiple national markets over
energy price throughout an entire transmission region.
a large geographical area into a common trading platform
In the European Union, a uniform zonal price for energy
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 91
distribution LMPs could not be useful. Rather, advanced and low distribution voltages respectively, although
meters, inexpensive ICT solutions and DERs that could at particular moments, when demand levels are high,
use distribution LMPs effectively have become available the effect of marginal distribution losses on LMPs can
only recently. Therefore, the question is: How important be as high as 40 percent for low-voltage levels. Other
is it to account for spatial variations in distribution? Let’s references also report marginal loss factors that indicate
examine losses and congestion separately. relevant differences in the value of energy consumed
within distribution networks that is not captured in
Using actual loss data from the Spanish power system
LMPs calculated for wholesale nodes (see Box 4.2).
in an aggregated fashion, it was possible to estimate
These numbers change with different penetration levels
the “marginal loss effect” for each voltage level, i.e.,
of distributed generation in the network. In any case,
the factors that, multiplied by the wholesale market
these variations create a different — and typically higher,
price, allow us to estimate the average LMPs at the
for moderate penetrations — local economic value for
high-, medium- and low-voltage levels of the distribution
distributed generation than for generation connected at
network at a given moment in time and also as an average
the transmission level.
over a year or other time period. Over the course of
the year, LMPs in distribution average 11 percent to 17
percent higher than the wholesale LMPs for medium
Figure 4.9: Distribution Locational Marginal Prices when Aggregated Distribution Losses are
Considered in the Spanish Power System
4.9a: Yearly average prices for the current 4.9b: Marginal prices for 9 a.m. during a
Spanish system day in April when the Spanish system had
a penetration level of solar photovoltaics
connected at low voltage (under 1 kV) providing
35 percent of total generation capacity
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 93
Figure 4.10: Layout of Urban Distribution Network Used in Price-Computation Simulations
Next, with a more detailed model of distribution system Figure 4.11a illustrates the spatial distribution of LMPs in
operations, we can characterize the behavior of LMPs in the network shown in Figure 4.10 in the absence of any
distribution networks under a diverse range of conditions. network congestion; that is, with differences in LMPs
Figure 4.10 shows the layout of an urban distribution arising solely from network losses. Figure 4.11b provides a
network (designed according to a street map of a portion close-up view of the low-voltage (LV) laterals connected at
of Austin, TX) that we use to illustrate the computation of a single medium-voltage (MV)/LV transformer, illustrating
distribution LMPs. 16
the price variation across nodes along MV and LV lines.
As expected, the maximum differences in nodal prices
along the MV and LV feeders relative to the transformer
occur at the ends of the LV laterals — i.e., farthest from
the transformer. In Figure 4.11b, the increase in nodal price
as we move farther along the LV lines (away from the
transformer) reflects the marginal cost of distribution line
losses. At the relatively light loading of the network used
in this example, the range of spatial differences in prices
along radial LV lines reaches just over 5 percent.
16 Note that all connections between medium-voltage and low-voltage lines occur
at MV/LV transformers (pink squares). What may appear, in the figure, to be
connections between MV and LV lines without a transformer are in fact shared
rights-of-way by MV and LV lines that, when plotted in the diagram, give an
appearance of line discontinuity.
4.11a: Distribution-level active power nodal prices (in $/MWh) across the entire
considered network during a single hour
4.11b: Distribution-level active power nodal prices along a small fraction of the network connected at a single
MV/LV transformer
Topology of zoomed-in network section Active power nodal prices of zoomed-in network section
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 95
Figure 4.12: Spatial Variation in Distribution-Level Active Power LMPs Caused by Network Congestion
in the Network of Figure 4.10
Most LMPs are in the range of $35 to $40 per megawatt-hour (MWh), while LMPs at congested nodes rise to the cost of nonserved
energy of $300/MWh since load must be curtailed to abide by constraints on transformer capacity.
Next, the same network is examined in the presence of Under such hypothetical future conditions, prices
congestion. Currently, congestion is generally uncommon would serve a critical role in communicating scarcity
in distribution, as the only practical solution for a violated of network capacity to the network users. Figure 4.12
physical network constraint — once transformer tap depicts the spatial variation of active power LMPs for the
changers, voltage regulators, and network reconfiguration same distribution network as in 4.10 and 4.11, but in the
have been employed — is load curtailment on the presence of congestion (reflected by the load on multiple
concerned feeder. Network capacity planning margins are transformers exceeding the rated transformer capacities).
therefore typically generous, and with low load growth, The price at congested nodes (those downstream of a
congestion in existing networks may be sufficiently congestion) rises to the cost of nonserved energy (CNSE),
infrequent that prices will not capture congestion-related or cost of voluntary demand curtailment, which in this
signals in many distribution networks. In the future, example has been set to $300 per megawatt-hour (MWh).
however, if price-responsive DERs become commonplace
and/or demand becomes more price-responsive,
networks may be more closely adapted to the actual
needs of network users,17 resulting in more frequent
congestion, caused by such factors as binding network
constraints that limit the response the demand and DERs
to prices and charges.
17 The level of adaptation of the network to current system conditions also depends
on the existing mandatory technical procedures to be used when planning and
designing upgrades. These procedures may have to be adapted to permit utilities to
manage distribution networks with binding constraints.
18 In the future, many of the devices that have inverters, such as solar panels and
batteries, could provide voltage control and reactive power compensation in
distribution systems (see Kroposki 2016 and Moghe et al. 2016). These devices 19 Voltages at all nodes are expressed “per unit,” or normalized to a base voltage
could obtain some compensation when committing their capabilities to the system quantity, which is defined to be the voltage at the transmission/distribution
and mitigating price volatility. substation (the pink star in Figure 4.10) in this system.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 97
Figure 4.13: Spatial Variation in Distribution-Level Reactive Power LMPs in the Network of Figure 4.10, with 40 Percent
PV Penetration (Measured as Installed Capacity as a Percent of Peak Load)
4.13b: Spatial variation of distribution level reactive power LMPs (dollars per Mvarh)
4.13c and 4.13d illustrate the concurrence of the highest per-unit nodal voltage and the most negative reactive power nodal price.
since it is at those nodes that reactive power absorption can set of networks is needed to evaluate this assertion.
mitigate the voltage rise caused by greater PV output. This
information can be used in several ways, such as: (1) the 4.4.1.5 The choice of granularity level
reactive power nodal price can be sent to network users with While the cost of providing energy to users throughout
the expectation that they will dynamically react to it and the power system can vary significantly by time period
alleviate the voltage constraint; (2) the sizeable magnitude and location, as shown in Section 4.3, most existing rate
of reactive power LMPs can be used to flag an area of the structures seen by end-users reflect few, if any, of these
distribution network for the installation of a capacitor or temporal and spatial variations in cost, and they are
inductor bank by the distribution utility; (3) reactive power exclusively based on active power (i.e. excluding reactive
nodal prices may help identify candidate nodes at which power).20 The question that remains is: how much additional
to run an auction to solve a voltage or reactive power benefit can be gained by more granular prices and charges?
compensation problem through a long-term contract; or (4) Obviously, the answer will depend upon the trade-off
reactive power LMPs may signal the need for an adjustment between the costs and the benefits of greater granularity.
of static settings of inverters in the relevant area of the
Figure 4.14 illustrates several possibilities for enhancing
distribution network (if the grid-code allows the utility or
the temporal and spatial granularity of the energy price,
distributed resource to do so).
where an implementation choice would consist of some
In summary, reactive power LMPs may be useful for combination of an option along the time dimension
flagging opportunities to install compensation devices, and another along the spatial dimension, or variations
change equipment settings, or run long-term auctions for thereof. Options along the time dimension reflect different
compensation services. The utilization of reactive power degrees of price or rate variation, or differences in the
prices as market signals applied to nodal reactive power degree to which prices or rates reflect real-time operating
injection and consumption in day-ahead and real-time conditions. These range from rates that do not change
local markets in which market agents bid prices and with time over the course of a day to real-time prices
quantities, may lead to volatile results and small gains
from implementation. Further research in a representative 20 In the European Union, customers with power factors (the ratio of active power to
apparent power) below a prescribed limit are penalized, restricting the flexibility of
customers to absorb or inject reactive power.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 99
Figure 4.14: Choices in the Adoption of Temporal and Spatial Granularity in Energy Pricing
(for some time interval such as 15 minutes or one hour) distribution-level granularity by adding loss factors to
for end-users. Starting from a flat, uniform tariff for an the transmission LMPs to account for distribution losses.
entire year or more, the first improvement (see Section Finally, distribution LMPs can be computed, perhaps for
4.3) is a TOU rate that assigns rates to blocks of time varying aggregations of distribution nodes (such as only
based on expected energy prices. Over the course throughout the HV network, at each HV/MV substation,
of a day, a user may see roughly three or four rates or at each MV/LV transformer), or, as nodal prices for
corresponding to different times-of-use. A step toward every distribution node. Figure 4.14 also indicates that
greater granularity is the use of CPP, which enables the making use of the reactive component of the energy
utility to alert customers of “critical” events, such as an price might be seriously considered when using LMPs
hour of particularly high load, during which the utility more deeply throughout the distribution network, where
can charge a significant CPP adder. The next option is voltage-related problems may become more frequent,
the use of day-ahead hourly wholesale prices (with or harder to solve, and can be addressed via reactive power
without locational differentiation) or, finally, the use of pricing. Evaluating whether or not prices for reactive
closer-to-real-time spot prices (which may be hourly or power need to be extended to distribution nodes requires
sub-hourly), as discussed in Chapter 7. Along the spatial 21
consideration of their role and significance in both
dimension, starting again with a tariff with no locational signaling efficient operation and investment, and (to a
component, the first step is to include zonal wholesale lesser extent) contributing to network cost recovery.
LMPs, moving next to nodal wholesale LMPs, then into
Figure 4.15 illustrates the spread of energy prices during
a single hour (active power prices) at varying levels of
21 Note that temporal granularity is understood here as updating closer to real time, spatial granularity for an urban distribution network
regardless of whether the length of the considered time interval is hourly, half
hourly, every five minutes, etc. (although shorter time intervals are typically used that is not experiencing congestion (yielding nodal price
closer to real time). However, this is not always the case. For example, intra-day
and balancing markets in many European markets have the same granularity
variations that result solely from losses). The network is
(hourly). Also note that using a market as a reference does not necessarily imply comprised of a transmission substation (pink star), HV/
using the same time intervals of that market when doing settlements. For instance,
the real-time market price in the United States has been used to make hourly, daily, MV substations (pink diamonds indicated by arrows in
or monthly settlements.
4.15b), and MV/LV transformers (pink squares), along Since most users (93 percent) in the modeled network are
with HV (132 kV), MV (20 kV), and LV (400 V) lines. LV users, most nodes exhibit the LV marginal price, shown
in green. The remaining 7 percent of users (MV users)
4.15a illustrates the case in which the same wholesale
face a slightly lower MV marginal price, shown in blue.
LMP is conveyed to all distribution system nodes. 4.15b
Finally, 4.15d shows the active power LMPs at all MV and
shows zonal LMPs computed at each HV/MV substation
LV distribution nodes.22 The spread of prices increases
(indicated by the black arrows). The marginal cost of HV
with each successive step toward full nodal pricing,
distribution network losses (which are quite low in this
capturing and revealing more to network users about their
example) is added to the wholesale LMP to produce a
impacts on marginal losses.
marginal price at each HV/MV substation that accounts
for HV distribution losses. 4.15c adds to the wholesale The benefits of greater granularity will be discussed in
LMP a voltage-level-specific loss factor that estimates the Section 4.5, once we review how prices and charges for
marginal losses at each voltage level. That is, an average the most relevant electricity services are computed. The
marginal loss factor is added to the LMP for all LV users to primary costs of improved granularity are computation
account for their impacts on marginal losses throughout and implementation costs (see Box 4.3). Implementing
the network, and, similarly, an average marginal loss the computation and use of distribution LMPs, if deemed
factor is added to the LMP for all MV users to account for
22 Reactive power marginal prices can be computed for all distribution nodes as well,
their impacts on marginal losses throughout the network. but they are not shown in the illustration here.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 101
convenient, may require significant time since ICTs must regulation, operating reserves, synchronized reserves,
be deployed, and robust algorithms must be verified in and black-start capabilities) accounted for approximately
sufficiently representative networks. Implementation 0.9 percent of the total wholesale cost (Warner-Freeman
requirements can be much lower for simplified, or less- 2016). These values can change over time due to changes
granular, approaches. For example, the pass-through in fuel costs, market rules, etc. In 2008, the breakdown
either of wholesale prices or of LMPs computed at the of the PJM wholesale cost of electricity was: 84percent
transmission-distribution interface node to network users energy, 10percent capacity, 4percent transmission, and
downstream (preferably with the addition of some loss 2percent reserves. In 2014, the ISO New England ancillary
factors approximated for large distribution zones) would services market (which includes operating reserves) was
have low implementation costs beyond the deployment of roughly 2.6 percent of the size of the energy market, and,
advanced meters, while yielding sizeable efficiency gains in 2015, the ancillary services market was 3.7 percent of
relative to the status quo. the size of the energy market (ISO New England 2015).
expected to slowly move from energy to capacity markets, 25 These economic signals may be received by users’ smart energy boxes or by
aggregators that manage users’ appliances automatically. What matters is that
at least in this specific system. economic signals with adequate spatial and temporal granularity are computed and
available.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 103
caps or simplified energy prices), it is efficient for some discussion that follows focuses on the distribution network,
additional signal(s) to convey users’ impact on generation although much of what will be presented also applies
capacity costs. For example, in the presence of a capacity to transmission.
market, a peak-coincident generation capacity charge can
From a regulatory standpoint, distribution and transmission
restore an efficient short-run incentive and determine
networks are customarily treated as regulated monopolies,
the allocation of marginal capacity costs among users.
with the corresponding regulatory authority establishing
This charge would reflect the coincidence of an end-
for each year the appropriate remuneration. There are two
user’s consumption (or network withdrawals) with the
main priorities when designing network-related economic
aggregate peak in the zone of the power system for which
signals: economic sustainability and allocative efficiency.
firm-generation capacity requirements are defined (this is
Economic sustainability means the totality of the funds
analogous to the manner in which users’ peak coincidence
collected via network charges must equal the regulated
contributes to network capacity requirements as described
network costs, since this is the amount that the regulator
in Section 4.4.4). DERs that inject power into the grid
has established must be recovered. Allocative efficiency
during aggregate peak periods should be credited with a
means efficiency in the allocation of the network charges.
symmetrical payment or bill credit reflecting the reduction
Here the objective is to maximize the utility of the network
in firm capacity requirements.
users (i.e., global welfare), eliciting their efficient response
In practice, peak-coincident capacity charges create and the associated impact on the power system while
significant and volatile short-run signals. Thus, if end-users minimizing economic distortions.
are able to participate in a capacity market, either directly
The ideal solution would be to achieve both objectives with
or via aggregators, agents with good forecasts of their
a single shot: By applying perfectly efficient LMP signals
likely availability during scarcity periods and preferences
to network users, the network would acquire power at the
for more certain revenues can commit themselves to
injection nodes paying the appropriate LMPs for the energy,
contribute to firm capacity requirements by participating
then transport the power and sell it (minus any incurred
in a capacity market a priori. By bidding firm capacity — a
losses) at the withdrawal nodes, receiving for the energy
forward commitment to either inject capacity or reduce
delivered the LMP of each node. As LMPs are generally
withdrawals during aggregate peak periods — users can be
higher at consuming nodes than at generating nodes, this
hedged against short-run scarcity signals for the amount of
system generates “network rents” or revenues. In an ideal
firm capacity bid into and cleared in the capacity market.
world, LMPs would thus both send economically efficient
Any consumption above the quantity cleared in the capacity
signals to all network users and generate sufficient revenues
market would remain exposed to the full short-run signal,
to recover regulated network costs.
including the peak-coincident generation capacity charge.
In the real world, however, the revenues recovered by
4.4.4 Network charges this method over the life of the network are only a small
fraction of the cost of well-adapted network investments
In Section 4.4.2 we have discussed locational marginal
(see Boxes 4.4 and 4.5). In the absence of economies of
prices (LMPs), which determine the price for electrical
scale in network investment, of noneconomically justifiable
energy at any node of the network and at any instant of
engineering design requirements, and of planning errors
time, incorporating all the short-run impacts of generation,
leading to excessive and irreversible investments, it has
demand, and network effects (losses and network
been proved that LMPs would completely recover the
congestion). LMPs precisely determine the price of the
fundamental energy service. Our next task is to allocate
and recover the regulated network costs efficiently. The
the distribution level and have only been adopted at the 3. The “residual network costs,” i.e., the fraction of network
transmission level in certain jurisdictions. In this context, costs that is not allocated on the basis of the cost-
the potential contribution of network rents from LMPs to reflective method noted above,26 should be allocated in
the least distortive possible manner.
cost recovery is purely hypothetical.
4. Although this may not be considered a priority at
We are therefore back at square one in the search for an
the moment, the allocation of the costs of the distribution
efficient network cost allocation method, but we have and transmission networks must gradually take into
learned something: LMPs, the perfect short-term marginal account the increasingly blurred boundary between these
energy prices that incorporate all network effects, are networks, so that ultimately all network users participate
not adequate long-term signals and do not secure full in all network costs according to the same principles.
recovery of regulated network costs. With only LMPs, the 5. Ramsey pricing27 advocates the use of the inverse
network users lack information about the implications of elasticity rule to optimize the efficiency (i.e., maximize
their network utilization patterns on the needs for future welfare) of the allocation of the costs of monopolies, or of
network investments. some prescribed amount of taxes, under some conditions.
It happens today that the availability and decreasing
Before embarking on the search for the most efficient cost of distributed generation and storage result in the
cost allocation approach, it is important to review the possibility of grid defection (i.e., radical disconnect from
the grid). (For the substantial amount of costs that can
requirements with which such an approach must comply
be avoided by defection, see Figure 4.2.) Grid defection
in any realistic situation.
is an extreme form of elasticity to price that has to
1. Since LMPs are perfect short-term energy prices, any be taken into account from an efficiency perspective
additional economic signal that is associated with when deciding the regulated costs to be included in the
the actual pattern of network utilization should be electricity tariff and in the design of the tariff itself (see
well-justified. Any charge not justified by efficiency Section 4.4.5 for a more detailed discussion).
objectives (e.g., designed to incentivize efficient short-
6. It is a common regulatory practice all over the world
and long-run behavior) should be applied in the least
that all consumers supplied by the same distribution
distortive manner, i.e., avoiding interference with LMPs
company, connected at the same voltage level, and
(or their implementation by proxy) and with any other
under the same type of contract (e.g., flat rate, critical
well-justified economic signal.
peak pricing, etc.), are subject to exactly the same
tariff — regardless of where they are connected. This
implies substantial subsidization of customers in high-
cost portions of the network (such as rural networks)
by those in lower-cost portions of the network (such
as dense urban areas). This cross-subsidization is
generally accepted without complaint, either because
people ignore that this socialization or averaging of
26 Or whatever fraction of the total cost could be recovered via congestion and lost
“rents” associated with distribution LMPs if implemented.
27 See Section 4.4.5.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 105
the distribution network costs is taking place or 4.4.4.1 How to make use of distribution location
because such cross-subsidization is desired to address marginal prices if available
distributional concerns.28 If cross-subsidization is
desirable, the advanced methods of tariff design that We have already said that so far no power system uses
are proposed in this study should make compatible LMPs at the distribution level and that, even if LMPs
the twin goals of sending efficient economic signals were computed and used as energy prices, the potential
and maintaining the “equal-tariff-for-all” principle, at contribution of LMPs to network cost recovery is likely to
least approximately. We discuss solutions to these
be modest. Why then worry about distribution LMPs in
challenges in Section 4.6.
this section on network charges?
7. Even if the average value of the tariff, or the cost of
electricity for the household, remains more or less the First, distribution-level LMPs, or different approximations
same once the advanced tariff designs that this study of them, may start being used in some systems, since they
proposes are implemented, the volatility over time of are the reference or gold standard for efficient short-run
network charges, and of economic signals in general, signals. If this is the case, particularly in tandem with price-
could be a matter of concern, as many customers
responsive DERs and demand, network capacity may come
may feel that they are unqualified to respond and take
closer to mirroring actual utilization patterns and more
advantage of the opportunities that this information
may bring them. We also discuss this challenge in significant differences may emerge between nodal prices
Section 4.6. in distribution networks, increasing the “network rents”
obtained via LMPs. Although such revenues would likely
8. DERs can provide network services, deferring
network-related investments that would otherwise continue to be insufficient to recover most of the network
be necessary. In order for DERs to have an economic costs, the collected amounts could be substantial and
incentive to provide these services, they need the their utilization should be supervised. Depending on the
correct economic signals with sufficient granularity in specific separation of activities in each jurisdiction, these
space and time. funds would be collected by the distribution company,
9. Finally, it has to be remembered that it is not only the the incumbent retailer, or even independent retailers or
quantity to be paid, but also the format (lump annual aggregators. These funds could then be used to partly
sum, $/kW, $/kWh charge, or other format) that offset the network costs that will have to be recovered with
matters in tariff design. We have already explained
network tariffs.
in this chapter (Box 4.1) how charging network or
policy costs with a volumetric tariff leads to multiple
economic distortions that are exacerbated by net
metering. Format is critical in adopting a charge that
sends an efficient signal without distortion.
28 Socialization usually has limits. For instance, rural customers usually have lower
quality service standards. Therefore, although they are subsidized, they do not
receive exactly the same service as urban customers.
29 Demand growth was simulated in three urban networks and three rural networks in which no network expansion occurred in response to demand growth.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 107
While the above results demonstrate that it would be unlikely for LMPs to provide full distribution
network cost recovery under typical network design and operating conditions, multiple factors
contribute to the exact magnitude of cost recovery from nodal prices. The prevalence of congestion at
various locations throughout the network and the frequency of congestion at any given location over
the period of cost recovery are instrumental in determining the amount of recovery. In an “overadapted”
network built with significantly greater capacity than peak network use, congestion will occur less
frequently; thus, nodal prices will less frequently rise to the cost of nonserved energy. As a result, nodal
prices will yield less network cost recovery than in a more tightly built (or “well-adapted”) network that
experiences more frequent congestion.
Much of the variation in cost recovery trends seen among the networks depicted in Figures 4.16a
and 4.16b can be explained by such differences in network adaptedness. For example, the network
called Rural 1, which is somewhat overadapted to base demand and experiences relatively infrequent
congestion over the full range of examined load-growth scenarios, exhibits steady cost recovery
throughout the range of load growth. In contrast, Rural 2 is overadapted until load grows by 30 percent
over base demand, at which point congestion increases significantly, and a sharp rise in cost recovery is
apparent. Figure 4.16c illustrates the relationship between the magnitude and frequency of congestion
(as measured by the amount of load curtailment required to abide by network constraints) and the
amount of recovery of the network cost annuity for the networks shown in Figures 4.16a and 4.16b.
This chart shows recovery of the network cost annuity from distribution LMPs as a function of load curtailment for each of the six networks
in Figures 4.16a and 4.16b, assuming cost of nonserved energy is $300 per megawatt-hour.
This chart reflects the impact of the cost of nonserved energy on the percent recovery of the network cost annuity with distribution LMPs
(base demand, no load growth, in urban network similar to that of Austin, TX).
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 109
4.4.4.2 Assigning responsibility for new the corresponding economic signals should be conveyed to
network investments the network users. The first challenge is to define what can
be understood as a “peak” during which peak-coincident
We want to relate the patterns of network user behavior
network charges are applied. Do charges apply only in the
and the network costs30, and convey the corresponding
one single moment of greatest demand? Should the peak
economic signal to the network users. This signal, if it
periods be specified in advance, or should the charges
reflects actual system conditions, will interfere with
be applied ex post to the actual peak periods, whenever
the LMP that ideally would be employed. Does this
they occur? Should we define the peak at the level of the
make economic sense, given that the LMPs, either at
entire power system, the HV/MV substation, or the MV/
the transmission or distribution levels, are the efficient
LV substation? As with efficient charges for energy, efforts
short-term signals? We maintain that adding an economic
to define efficient signals for network utilization must
signal on top of LMPs (if employed) — for the agents
determine the appropriate degree of both temporal and
connected to a node where the withdrawal or injection of
spatial granularity.
power at a given moment in time indicate the need for new
investment — makes economic sense. The action of the To answer these questions satisfactorily, the proposed
agents has two separate economic consequences: First, approach must be able to send signals that reflect any
to modify the cost of operation (this is captured by the actual stressful conditions in the system, capturing
LMP) and, second, to indicate the need for new network patterns of contribution to the peaks; therefore it must
investments (which the LMP alone, due to economies of use an ensemble of the highest peaks and not just
scale in network investment, enhanced by engineering the highest single peak. The results obtained could be
planning rules, discreteness and irreversibility of network averaged over a certain area (to be specified for each
investments, fails to capture, in practice by a large amount, case) to avoid having very different charges for customers
as the collected experience until now seems to indicate). whose locations are very close. What follows is a
discussion of a possible implementation approach that
The literature on network cost allocation at the
meets all of these requirements.
transmission level is enormously abundant (in contrast
with that for distribution), since the design of distribution It can be safely assumed that distribution networks are
network charges was considered straightforward under planned and designed to accommodate peak capacity
the top-down paradigm. We can therefore borrow ideas requirements under a range of plausible operating
from transmission to be used at the distribution level. scenarios and anticipated system peaks (accounting
The criteria for cost allocation should be to quantify the for variations in season, weather patterns, load growth,
impact that the actual behavior of network users has and other factors). Making use of a sufficient number
on creating the need for new network investments (by of distinct system peaks (in withdrawals or demand
connecting to the network, installing new electricity-using and in injection or generation) throughout the year to
appliances or DERs, or augmenting the ones that they allocate the capacity charges ensures that variability in
have). Box 4.5 summarizes several approaches that are network use patterns is accounted for and minimizes
inspired by this principle. randomness in the computation of network charges.
Thus, the distribution utility and/or regulator can allocate
4.4.4.3 The format of the peak-coincident network
the annuitized capacity-related network costs (for peak
capacity charge
demand and peak generation) to be collected over
A peak-coincident network charge, however it is calculated, multiple periods (months may be reasonable, or portions
can be an efficient method for guiding long-term utilization of months) within a year, based on the probability and
of electricity networks and recovering a portion of magnitude of anticipated peak conditions. For instance,
regulated network costs. A key question, however, is how assume that the 3031 highest peaks in demand of the
30 In particular, the reinforcements that will be needed in the future or have been
recently made as a consequence of the behavior of each specific network user. 31 Thirty, or a similar number, reduces the randomness in the application of the
charges and captures the most significant peak conditions in the power system.
A second approach, from Abdelmotteleb, Gómez, Chaves, and Reneses (2016), is based on the marginal
participation method, which consists of first selecting those critical network components with a scarce
margin of installed capacity over utilization. Then, the network users that contribute to the flows in the
critical network components are identified so they can be charged for future reinforcements.
A third approach, from Pérez-Arriaga and Bharatkumar (2014), starts by determining the weight in the
distribution network cost (total or by voltage level, area, or feeder) of the two major cost drivers for
incremental network investment: meeting the expected peak demand and meeting the expected peak
reverse flow due to distributed generation. Assuming that the costs of future network investments
and the weights attributable to peak demand and peak distributed generation remain consistent with
estimates derived from a planning model for the existing network, it is possible to evaluate per-unit
charges for the contribution of coincidental peak demand and coincidental peak generation.
A fourth approach, from quite a different perspective, is presented in detail in Chapter 6.3.3, whereby
regular auctions of options on network capacity would be held to elicit information on network users’
willingness to pay for future network capacity investments as well as DER alternatives. The results of this
auction would then determine network users’ responsibility for network cost recovery.
In all these approaches, the strength of the economic signals sent to network users through their network
charges should be modulated to account for the capacity margin of the network under consideration
(e.g., at the aggregate level or, by voltage level, zone, or feeder) over the existing maximum power
flows. The objective is clear: Do not incentivize unnecessary changes to network utilization because
that distorts the economic signal and reduces efficiency. For instance, if a distribution feeder has ample
capacity to accommodate existing peak network-use as well as significant growth in network utilization,
the economic signals conveying users’ contributions to projected reinforcements should be weak or
nonexistent. Under such circumstances, it would be inefficient to encourage an even lower utilization
of the network assets. In contrast, if a network is being utilized at or near its full operational capacity,
or is projected to be so soon, then the economic signals conveying the marginal or incremental cost of
projected reinforcements should be stronger. Determining the most effective format for the relationship
between the strength of the charge and the degree of network utilization is a key aspect of the design, and
it will also depend on the response of distributed agents in a diversity of contexts.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 111
previous year occurred during five hours in January, nine network users consume or inject power during periods
in February, three in March, six in July, five in August, of peak demand or production. For example, injection
and two in December. This effectively allocates the peak from distributed generation (DG) during periods of peak
demand-driven costs to each month according roughly system consumption may help reduce the aggregate
to the probability of a system peak occurring within that network withdrawal peak, reducing the need for network
period. Dividing the total recoverable peak demand-related investments. Injections that occur during the aggregate
costs across months according to anticipated system peaks withdrawal peaks can thus receive a negative network
provides network users with economic incentives to guide capacity charge — in the form of a payment or credit on
their network utilization behaviors and shift network use a user’s bill — because of the value of DG in serving load
away from potential peak periods. Within each month to and reducing power flow and system congestion. Likewise,
which some fraction of total peak demand-related costs power injection during periods of system peak injection
have been allocated, each network user’s share of the total can lead to a positive capacity charge, while withdrawals
capacity costs can be computed according to that user’s during the peak injection period should be compensated
ex post share of the actual peak or collection of actual symmetrically. This symmetry in peak coincident network
highest peaks each month — i.e., according to the ratio of charges/payments mirrors the symmetry in LMPs and is
the user’s kilowatts of consumption to the system kilowatts critical to ensure all resources are placed on a level playing
of consumption. An entirely parallel procedure must be field and compensated or charged in a nondiscriminatory
followed for peak generation-related costs. and nondistortionary manner.
Note that the total sum that must be collected from users 4.4.4.4 What to do with the residual
for network capacity investments in a given month is network costs?
established a priori and does not depend on the actual
It is possible that only a modest fraction of the total
behavior of the network users during that period. However,
revenue requirement for a distribution network can be
the allocation of the monthly charge among network
recovered with the “peak-coincident network capacity
users does depend on each user’s contribution to the
charges” and via network rents from LMPs, if they are
actual peaks that month, creating efficient incentives for
employed. Therefore, we shall define “residual network
network utilization in the short run. The magnitude of total
cost” as the fraction of the network cost that still remains
recoverable capacity costs allocated to each month is
unrecovered after the application of LMPs (if used)
based on the anticipated peaks used in distribution network
and the peak-coincident network capacity charges.
planning because the anticipated peaks (rather than
These three components of the network cost have been
actual peaks) guide network investments. With a record
represented in Figure 4.17.
of the historical system peaks plus regulatory validation
for the projected peaks used in the course of determining Once the efficient short-term signals (LMPs, if used) and
distribution utility remuneration, it is likely long-term signals (peak-coincident network capacity
that the anticipated system peaks will match the actual charges) have been applied, there is no need to send
system peaks reasonably well, producing an accurate additional economic signals to influence the behavior or
allocation of total capacity-related network costs to several investments of network users. Indeed, any other charges
forecasted peak periods. Operations and maintenance based on the relationship between the pattern of user
costs that vary according to capacity requirements (i.e., on injections and withdrawals would result in economic
a per-kilowatt basis) can be added to each user’s share of distortions that would increase the cost of the power
capacity-related investment costs on a per-kilowatt basis.32 system. Other economic reasoning is necessary to find the
criterion for allocating these residual network costs in a
These “peak-coincident network capacity charges” can
manner that minimizes economic distortions. This criterion
be either positive or negative, depending upon whether
will be common for the residual network costs and the
policy costs, and it will be presented in the next section.
32 We illustrate the application of this format for network charges in Boxes 4.6 and 4.10.
4.4.4.5 The end of the “top-down paradigm” much smaller. The mindset about who should pay for
distribution network costs will thus have to be changed
Our proposed approach to the design of network charges
accordingly, paving the way for methods more like what is
follows the classic distribution/transmission separation:
now used for the transmission network.
(1) the agents connected at the transmission level
(this includes the agents connected in the distribution Under these hypothetical conditions, a separation of the
network, conveniently aggregated at the transmission/HV networks into “meshed” and “radial” would make more
distribution transformation nodes) cover the transmission sense than the present division between “transmission”
network costs; and (2) the agents connected to the and “distribution,” which is mostly artificial and follows
distribution network pay the distribution network costs. different criteria in the diverse power systems.33 Any
This approach makes perfect sense for current power sound method for determining transmission charges
systems in which the distribution network has been would be applied to the meshed part of the network, and
designed to bring power to end consumers, who are the the resulting charges for nodes connecting the meshed
obvious major beneficiaries of the distribution network. If and radial networks would be passed to the agents in
a feeder does not have enough capacity, the consumers the radial feeders.34 Similarly, the cost of each radial
connected to that feeder have to be disconnected, feeder would be allocated to the agents connected to it,
resulting in a heavy loss of individual utility (therefore including the agent at the head of the feeder.35
these consumers would benefit much from a grid
reinforcement), whereas the large generators, connected
to the transmission network, only reduce their sales a 33 Note that in most distribution networks, the high-voltage part has a meshed
structure, as in transmission. The medium-voltage network is also meshed, but it is
small amount. operated in a radial configuration, with some breakers open, but with the possibility
of reconfiguration, if needed. The low-voltage network is radial, typically with a
This situation may change in the future if distribution tree-like structure.
34 These charges should be based on who is responsible for network investments and
networks are populated by a multiplicity of DERs, would adopt the format of peak-coincident capacity charges. In other words, the
including abundant distributed generation and storage. radial feeder would be seen, from the meshed network perspective, as a generator
(power injection) or a demand (power withdrawal) at the corresponding meshed/
Under these conditions, the response to a shortage of radial connection node.
capacity at the distribution level will be to activate the 35 The peak-coincident capacity charges that are transferred from the radial to the
meshed part of the network at each node will have to be allocated to the generators
local resources, and the loss of utility for the agents and demands connected to this meshed network following some criterion of
responsibility in the injections or withdrawals at the meshed/radial connection
connected in the congested distribution area will be
nodes.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 113
4.4.5 Charges to recover policy costs and categories there is no clear criterion for assigning costs
in every specific tariff, but we shall highlight the potential increase (or decrease) the marginal cost of compliance
consequences of incorporating a large percentage of with such policies, as explained in Batlle (2011) and
these costs into electricity tariffs. further developed in Batlle et al. (2016).37 For example,
in the case of a 20% renewable electricity obligation,
In the discussion that follows, we include both residual increasing total electricity demand by 10 kWh would
network and policy costs together since both share two key require an increase of 2 kWh of electricity supplied
features. First, both are costs that we assume regulators by qualifying renewable electricity sources. If these
have decided to recover via electricity tariffs (later in this resources receive public policy support in the form
section we will discuss whether some such costs could of direct subsidies or tradable renewable certificates
be recovered by other means). Second, for each of these (also known as renewable energy credits), then the
36 The European Commission has so far considered that directly charging electricity 37 Note that if the policy objective is established in terms of a fixed quantity
consumers for subsidies to renewables is compatible with the European rules requirement completely decoupled from the evolution of consumption, then there
on state aids and the environmental-polluter-pays principle (Ecofys 2014). We would not be room for a cost-reflective policy cost signal. Changes in electricity
suggest other possibilities later in this section. consumption would not affect the cost of compliance with this fixed quantity
requirement. In such cases, all costs would have to be allocated as residual costs in
a minimally-distortive manner, as discussed below.
Figure 4.18: Simplified Power System Configuration Illustrating the Impact of the Design of
Network Charges
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 115
The effect of three network tariff designs is discussed: (1) an “efficient ideal” approach, consisting of a
forward-looking peak-coincident charge computed as the incremental costs (Y) divided by the expected
load growth (as proposed earlier in this section), plus a fixed charge per customer computed based on
a Ramsey-like approach (see Section 4.5) for collecting the remaining residual network costs (Z); (2)
a volumetric charge ($/kWh) to recover the entire network cost; and (3) a fixed charge to recover the
entire network cost. The forward-looking peak-coincident charge would be applied to the total load at
critical hours at which a further increase in load would require the network reinforcements. The residual
network cost (Z) would be equal to the total network cost (X) minus the amount recovered (W) with
the peak-coincident charge during the critical hours, that is Z=X-W (all of these values are annuities).
The customers are assumed to optimally react to each tariff design by minimizing their total cost of
electricity supply from the grid plus the cost of any deployed responses to the received economic
signals. When presented with the efficient ideal approach, the customers would reduce their load during
critical hours (through any of the available options for responding) if the cost of these alternatives is
lower than the bulk energy price plus the forward-looking peak-coincident charge. In this case, the
network reinforcements would be avoided if the expected load growth is sufficiently reduced such that
it can be supplied with the existing network capacity. Otherwise, if the cost of the response alternatives
is higher than the cost of the energy supplied from the grid, including network charges, the customers
would buy all the energy from the grid and pay the associated costs. As a result, the efficient ideal
approach successfully guides the agents to find the most efficient solution for the system. In contrast, if
total network costs (X) are allocated based on a volumetric charge, the customers may find it beneficial
to invest, for instance, in a backup generator to partially self-supply electricity and reduce their
volumetric charges, with no specific incentive to focus on reducing load or network use during peak
demand hours. In this case, the system cost in general would increase, since the tariff may incentivize
the deployment of a less efficient generator. Finally, with only a fixed network charge, the customers
would decide to fully supply their load from the grid, as they face no marginal incentive to reduce
consumption during peak hours. As a result, a fixed charge to recover total network costs could lead to
more expensive network reinforcements in future periods.
Efficient network cost recovery may be comprised of three components: surplus from locational
marginal prices (if any), a peak-coincident capacity charge, and a fixed charge to recover
residual costs.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 117
wealth. One viable proxy for wealth, which perhaps has The presence of a substantial amount of policy
the advantage of being perceived as equitable in the and residual network costs in the electricity
design of electricity rates and is also indirectly related to
tariff has several negative consequences and
electricity consumption, is the customer’s property tax
invites a serious reflection on this aspect of
or the property size,42 and either of these proxies could
be used to implement a Ramsey pricing approach. Other
tariff design.
similar metrics might also be acceptable.
4.4.5.3 Costs to be included in the tariffs As indicated before, this report will not make
recommendations on which policy costs should or should
There are several negative consequences of including
not be included in the electricity tariff, since this is very
residual costs that are not directly affected by changes
jurisdiction-dependent. However, since the drawbacks of
in electricity consumption in electricity tariffs. First,
including these costs in the tariff have been indicated, it
there is a very clear risk of efficiency loss, if these costs
seems pertinent to reflect on this topic.
are allocated in the tariffs in such a manner that they
distort cost-reflective prices and network charges. Every cost item within the residual network and
Second, if the adopted tariff design is such that the policy costs should be subject to scrutiny. In many
policy and residual network costs end up increasing the jurisdictions, the subsidies to promote low-carbon
volumetric component of the price of electricity for the technologies — renewable generation, in particular — are a
end customers, this will have the effect of discouraging significant component of the policy costs in the electricity
the use of electricity with respect to competing tariff. Climate change is the major justification for
alternatives and slowing down the transition to a more incurring these costs. Certainly, electricity production is
electrified future, where electricity appears as the an important contributor to greenhouse emissions, both
dominant energy vector, generated from carbon-free in the United States and in the European Union, but other
resources and supplying also the needs for mobility sectors of the economy are as well. Since electric power
and space conditioning of buildings. Finally, even if our appears to be easier to decarbonize than other sectors,
recommendation on how to include the residual network a strong effort is in progress to decarbonize electricity
and policy costs in the tariff is followed, we have to be faster and more drastically. It therefore seems justifiable
aware of the potential implications that loading the tariff to share the burden of electricity decarbonization — i.e.,
with policy and residual network costs could have on grid the corresponding policy costs — with customers of
defection (see Box 4.7 and the following discussion), other sectors that are spared this extra effort (see Batlle
perhaps incentivizing defections that result in a reduction 2011 for a detailed discussion). In the United States a
of social welfare that would not have happened if residual substantial fraction of the cost of renewable support
network and policy costs were lower. schemes is born by federal, state, or local taxes.
42 There is a causal relationship between the amount of energy consumed and the
need for more renewable production or more energy efficiency measures, since
the targets for these measures are frequently expressed as a percentage of total
energy consumption. Unless this relationship is clearly established, we recommend
avoiding recovering these costs with a charge in the tariff that is proportional to the
energy consumed, as this would artificially modify the short-term price of energy.
Rather, we recommend the use of a proxy, such as a property tax, to allocate
costs. Using this type of proxy would ensure that allocation is related to energy
consumption without modifying short-term signals for electricity services.
4.4.5.4 Implications on grid defection grid defection, i.e., one that reduces social welfare. Grid
defection has implications for the remaining network users,
The present trend toward widespread availability and
which have to shoulder whatever costs are not paid by the
decreasing cost of distributed generation and storage
defectors. Thus, tariff design should try to ensure that no
results in the possibility of grid defection (i.e., radical
excessive or ill-assigned costs in the tariff make separating
disconnect from the grid),44 from a physical (for
from the grid attractive, particularly if better-designed
customers that meet some conditions, such as being
economic signals could prevent defection.
able to install some embedded generation within the
household or business premises) and economic (given A measure that can be explored — although we are
the substantial costs to be avoided by defection) aware of the potential legal obstacles — is to establish
standpoint. Grid defection is an extreme form of elasticity that those who defect from the grid should continue
to price, and it has to be taken into account — from an paying residual network charges (e.g., the cost of any
efficiency perspective — when deciding the regulated network assets left stranded by defections) or perhaps
costs to be included in the electricity tariff and the design receive some credit (if, by disconnecting from the grid,
of the tariff itself. they contribute to the deferral of network investments),
and maybe pay some other policy costs (guaranteeing
The implications of tariff design on grid defection have
that all customers contribute to recover an appropriate
to be carefully examined. The economic viability of grid
share of the policy costs underwriting established
defection depends critically on the volume of costs that
societal objectives) via an exit charge, thus ensuring
can be avoided by disconnecting from the grid, which in
that other network users do not have to pay for these
principle include all the costs of supplying the energy and
costs, increasing their tariffs. Some challenging design
network services plus the policy costs. Figure 4.19 shows
decisions are necessary to implement exit charges: How
the costs and the savings of grid defection from the
much of the estimated cost of any stranded investment
should the defector pay? Is this just one lump payment
and, if not, for how long should the defected customers
have to pay this charge? Would these consumers
continue receiving electricity bills from the utility?46
43 We are not proposing to give up the network-related short-term economic signals
(LMPs, in detailed or approximated form) or the long-term ones (peak capacity
network charges), which are necessary for efficient operation and investment,
but to consider whether the majority of the network costs (the residual network
charges) should be included in the electricity tariff or not. In the same way that
public schools are paid for by all taxpayers, regardless of whether or not they have
children, and public libraries are supported by those who never use them, and most
roads are paid by general taxes regardless of how much each taxpayer uses them,
the existence of an electric network reaching to every building can be considered a
necessary basic infrastructure in any developed society, and thus paid for via taxes
and not specific electricity rates. Access to connection to the electric network is 45 A customer’s decision to defect may be driven by other motivations besides strict
an added value for real estate, even if its owners decide not to connect to the grid monetary savings, such as a strong preference for renewable technologies or
and to self-supply electricity for their own needs. Note also that there is already a the satisfaction derived from being energy self-sufficient. These motivations are
high level of averaging or cross-subsidization in the allocation of the distribution beyond the scope of the present discussion.
network costs in the tariffs of every power system, whereby all customers 46 An additional difficulty that arises when addressing these implementation issues
connected at the same voltage level are subject to the same tariff, regardless of the is the difference in the economic lives of the assets involved on both the electric
actual distribution costs in their location. grid side (lines, transformers, large generators) and the customer side (PV panels,
44 The refusal of a new agent to connect to the network can be seen as another form batteries, heat pumps, ICT investments, etc.) — and the expected time horizon for
of grid defection, with different implications. investment recovery.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 119
Figure 4.19: Costs and Benefits of Grid Defection from Regulators and policy-makers must carefully
the Customer Viewpoint
monitor for conditions that could lead to a
serious threat of inefficient grid defection.
If these arise, they must reconsider the
costs that are included in the tariff as well
as other measures to prevent substantial
cross-subsidization among consumers and
a potential massive grid defection with
unforeseen consequences.
Some studies, however, claim that customers may find it economically attractive to disconnect from
the grid in the near future in some power systems in developed countries (Rocky Mountain Institute,
Homer Energy, and Cohnreznick Think Energy 2014; CSIRO 2013). Most of the published studies on grid
defection forecast future retail prices (based on current cost components) and compare those prices
with the estimated average cost of an isolated system. These studies usually underestimate the costs
of the typical, low-reliability, off-grid systems. EPRI (2016) estimates the net present costs of off-grid
supply for a US single-family household (Figure 4.20) for different estimated levels of reliability (as
a percentage of load served). The results show that the isolated system is more than 10 times more
expensive than the grid-connected one for the same reliability level, and still about five times more
expensive if a very low reliability level (80 percent) is considered acceptable for the off-grid system.
Net present cost and levelized cost are shown per kilowatt-hour for off-grid systems with solar photovoltaics and storage, with and without
demand management, and with a diesel generator for different reliability levels.
The costs included in a tariff are critical to determining the economic viability of grid defection. For
instance, the Australian Energy Market Commission recognizes the need for adjusting network tariffs
as “there is a high risk of grid defection” (AEMC 2016). The situation might evolve into what has been
termed the death spiral (Costello and Hemphill 2014; Felder and Athawale 2014; Graffy and Kihm
2014), whereby the incentives for grid defection increase with the number of defecting customers
because residual network costs and policy costs are borne by fewer remaining customers.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 121
Box 4.8: Cybersecurity-Related Costs
The computation of appropriate prices between DERs and transmission system operators (TSOs) or
distribution system operators (DSOs) requires consideration of cybersecurity. TSOs/DSOs will have
the responsibility for monitoring nonutility DER suppliers and assuring malware or cyber attacks are
not introduced into the TSO/DSO control or pricing systems. Cyber risks can come from DER control
systems, demand response systems, dispatchable energy storage systems, smart meters, and other
DERs that communicate with TSO/DSO control systems. As DERs participate in the price formation by
making buying and selling bids in different markets, digital connections to utility pricing and accounting
systems will need to be sufficiently protected and monitored. Fixed charges can be used to recover the
cost of meeting protection standards (NIST 2014).47
47 For example, US North American Electric Reliability Corporation critical infrastructure protection standards, European Commission standards, state Public Utility Commission
standards, or National Electric Sector Cybersecurity Organization Resource suggested requirements.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 123
Box 4.9: End-user Response to Prices: The Effect of Locational and
Spatial Granularity
Using a detailed model of a distribution network, which includes a variety of building types as network
users (see the description of the D-SIM model in Appendix A), we computed energy prices with
different degrees of spatial and temporal granularity and let each network user make optimal (i.e., cost-
minimizing) investment and operational decisions. Our results, shown in Figure 4.21a, suggest that the
type and location of DERs likely to emerge in a distribution network are very sensitive to the structure of
the energy price. Furthermore, locational prices (distribution-level LMPs) can effectively indicate which
locations will have the highest network benefits, potentially relieving network constraints at a cost that
is less than traditional approaches (e.g., reinforcing wires, upgrading transformers, etc.).
4.21a shows that the most attractive DER investments change under different energy price structures. Each panel in the figure corresponds
to the same distribution network. Flat prices, for example, lead to investments in solar photovoltaics (PV) with no discernable geographical
pattern. In contrast, LMPs calculated down to the meter lead to investments in heating, ventilation, and air conditioning (HVAC) controls
that are clustered around the area of congestion.
An important observation is that the net present value of a particular DER investment in response
to granular LMPs diminishes quickly as other similar investments are made in the same area of the
distribution network. As we show in Figure 4.21b for this particular case, the expected savings for a
single network user from the installation of a price-responsive heating, ventilation, and air conditioning
(HVAC) system are considerably smaller if we assume that other neighbors will do the same.
Simultaneously, network users who did not invest would still benefit from the exposure to lower LMPs.
As explored further in Chapter 8, the ability of DERs to address network constraints as an alternative
to traditional “wires” solutions is a key driver of the economic value of DERs. Figure 4.22 compares the
costs of investing in DERs to the costs of network infrastructure investments (in the network modeled
above) under a variety of economic signals. As shown in Figure 4.22, the lowest-total-cost solution
occurs when the best 10 DER investments, in terms of net present value, are installed under the LMP
pricing scenario. Looking back at Figure 4.21a, those investments are HVAC controls, all located in the
area of congestion. The traditional utility solution — replacing the transformer — actually results in
greater operation benefits in the form of reduced nonserved energy48 and losses, but carries a higher
upfront investment cost, making the total net cost higher than the best DER solutions. DER investments
that emerge under flat and substation-level LMPs result in some reduction in losses, but because they
are located outside of the congested region (see Figure 4.21a), they do not reduce nonserved energy,
and thus produce higher net costs.
This figure shows the net costs of different investment decisions related to the alleviation of congestion in the distribution network. In the
baseline situation, a certain amount of nonserved energy (NSE) is needed to respect network constraints. Each alternative scenario has
a certain upfront cost and is to some extent effective in reducing the losses and NSE with respect to the baseline scenario. The red bars
correspond to the net cost-benefit balance with respect to the baseline.
48 Nonserved energy is assumed to come from the continuous forced disconnection of load at a high price. As explained in Appendix A, this approximation is necessary to compute
LMPs as the dual variables of the nodal power balance constraints. On the building side, the cost of temperature deviations was made very high, making them effectively hard
constraints.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 125
Increased granularity in energy prices is a sound way to identify DER investments that can help alleviate
network problems — a function carried out in a similar fashion by LMPs in the transmission network.
However, given the strong coupling between the expected profitability of investment decisions made
by multiple agents, along with the magnitude, irreversibility, and long asset life of such investments,
it would be very difficult to prevent inefficient substitution and free-riding. Instead, the necessary
coordination could take the form of long-term contracts, cleared through public auctions, between the
distribution company and providers of alternatives to network reinforcements, including DERs, perhaps
via aggregators. See Chapter 8 for more examples of the system-wide value that DERs can provide
when efficiently located and operated.
Increased granularity in energy prices is a sound way to identify DER investments that can
help alleviate network problems. However, since DER investments may experience rapidly
diminishing marginal returns, influenced by surrounding DER investments, network users may
need to coordinate among one another and with the distribution company to prevent inefficient
substitution and free-riding.
We model two cases, one reflecting a high-load-growth scenario with higher peak-coincident generation
capacity and network capacity charges and a low-load-growth scenario where these charges are more
modest.51 We simulate a residential building in Westchester County, New York,52 with electric cooling
using the Demand Response and Distributed Resources Economics Model (see Appendix A). The
building’s air conditioning (AC) demand is controlled by a home energy management system or “smart
49 Historical 2015 hourly wholesale energy prices for NYISO Zone I are used.
50 We assume average losses in a distribution network are 5 percent and use a quadratic function to estimate marginal losses in each hour as a function of historical 2015 NYISO
Zone I aggregate load (e.g., Westchester County, New York).
51 Generation capacity costs are assumed to be $70 per kilowatt-year (kW-yr) in the low-growth case, reflecting 2013–2015 average NYISO capacity clearing prices, and $115/kW-yr
in the high-growth case, reflecting NYISO load and capacity forecasts for 2016–2026 (NYISO 2016). Distribution network capacity costs range from $25/kW-yr in the low-growth
case to $120/kW-yr in the high-growth case, reflecting two values within the range of transmission and distribution deferral benefits estimated in several regulatory avoided cost
filings (e.g., Neme and Sedano 2012). Costs are converted to peak-coincident charges by allocating annual costs proportionately to the top 100 hours of aggregate demand at the
NYISO zonal level for generation capacity charges and top 100 hours of average residential load profile for the distribution network charge.
52 The household consumes 6,600 kWh annually and is equipped with a 4 kW central air conditioning system.
The results show that the efficient ideal approach provides the lowest customer bills, the greatest
reduction in the user’s contribution to marginal system costs, and full recovery by the utility of the costs
to serve the customer. Simplified alternative tariff designs show reductions in customer bills compared
to traditional flat rates but result in either under- or over-recovery of utility costs and thus (implied)
transfer payments. These results are summarized in Figure 4.23 for the high-demand-growth scenario,
in which a larger share of total network costs are considered marginal (rather than sunk or residual), and
generation capacity costs are higher, reflecting expected capacity investments to accommodate peak
demand growth. As Figure 4.23 illustrates, under the TOU and CPP tariffs, the customer’s bill falls short
of fully recovering the marginal power system costs incurred to serve that customer, including marginal
energy costs and marginal or incremental generation and network capacity costs, as well as the share
of total residual costs allocated to the customer. In contrast, under the flat tariff (Flat) and flat tariff
with noncoincident customer demand charge (Flat + NCP), the customer’s bill exceeds the marginal
costs incurred to serve the customer and the residual costs allocated to that customer. Only under the
efficient ideal approach does the bill accurately match the customer’s contribution to marginal system
costs and share of residual costs.
Figure 4.23: Summary of Customer Bills and Utility Cost Recovery by Tariff Schedule —
High-Demand-Growth Scenario
53 We assume a 0.5°C of temperature deviation with no penalty, reflecting the fact that HVAC equipment naturally fluctuates around a target temperature. Deviations from 0.5°C
-1.5°C are penalized at $0.1 per hour, while penalties increase to $0.5 per hour for 1.5°C -2.5°C deviations and rise steeply for each additional degree beyond, reflecting the fact
that residential consumers have a demonstrated price elasticity for air conditioning and have shown a willingness to tolerate both precooling and higher temperatures during event
hours (EIA 2014; Herter and Okuneva 2013).
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 127
Furthermore, customer bills under the efficient ideal approach are 22 percent lower than bills under
the flat rate for the high-growth scenario and 6 percent lower for the low-load-growth scenario (where
peak prices are not as high). Changes in the user’s consumption patterns also reduce marginal system
costs (including energy costs and marginal or incremental generation and network capacity costs) by
33 percent under the cost-reflective tariff as compared to flat rates for the high-growth scenario, and
by 17 percent under the low-growth scenario. This typical household could save as much as $400 per
year on electricity bills solely by employing flexible AC strategies in response to improved tariffs,54 with
system costs incurred to serve this customer falling by an equal magnitude. Greater savings could be
achieved by employing other smart appliances, price-responsive demand curtailment, cost-effective
energy efficiency measures designed to reduce consumption during high-price periods, and other cost-
saving methods. In aggregate, more efficient, cost-reflective tariffs could unlock billions of dollars in
savings at a state or national scale by reducing the need for investment in new network or generating
capacity, and by reducing consumption during periods with the highest marginal energy costs.
Five cases, together featuring different ways to allocate costs, are compared, and the results presented
in Figure 4.24.
First, the reference case allocates regulated costs as fixed charges (euros per customer). Therefore,
flexible consumers respond only to hourly locational marginal prices (LMPs). In this reference case,
there are no peak-coincident network charges, the total annual network costs amount to 9.5 billion
euros, and policy costs amount to 15.7 billion euros. Operational costs in this reference case are lowest;
however, future incremental network costs will be highest due to the fact that agents do not experience
a peak-coincident network charge (and therefore do not change behavior when they are operating in a
way that is costly to the network).
The second case represents the proposed efficient ideal approach (EIA), where the application
of the peak-coincident network charge corresponding to the estimated costs of required network
The intermediate case allocates the total network costs to a peak-coincident charge (instead of
adopting the cost-reflective approach of the second case), and recovers policy costs with a volumetric
charge. Energy prices are computed as hourly LMPs. In this case, operational costs are higher than in the
previous two cases due to the distortionary impact of the volumetric charge and the excessive response
of on-site thermal generation, without achieving further network cost reduction.
In the fourth case, all regulated costs are allocated on a volumetric basis (i.e., on a euro/kWh of energy
basis), and exports to the system are remunerated at the retail price. Again, energy prices are computed
as hourly LMPs.
Finally, the fifth case employs a flat average energy price for the whole year, together with volumetric
charges to recover network and policy costs.
Figure 4.24: System Operational Costs and Flexible Resources Dispatch for Selected Cases
Hourly price differences incentivize electric vehicles (EVs) to charge when prices are low and inject
energy to the network when prices are high, whereas flat energy prices do not incentivize EVs to inject
energy since flat prices eliminate the possibility of arbitrage between prices. In addition, in the flat-
price and volumetric cases, backup generation is dispatched at almost its maximum capacity, since
on-site generation reduces payments of the residual network costs and policy costs embedded in the
volumetric tariffs. In contrast, backup generation is not dispatched in the reference case because of its
high variable costs.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 129
Flexible customers can net demand with local generation, reducing the energy flow from the rest of
the system, as well as energy losses and LMPs for the whole system — thus decreasing system costs
and customer payments. Reductions in LMPs (and customer payments) are higher at the low-voltage
nodes where flexible consumers are located, since the effects of their behavior permeate through more
of the network. As indicated, under volumetric charges, reducing electricity imports from the grid
allows flexible consumers to reduce their payments of regulated charges. The volumetric costs avoided
through customers’ modifications of their demand profiles must be collected from inflexible customers
to ensure full recovery of regulated costs.
Well-designed prices and charges can incentivize investments in and operation of DERs in ways
that yield lower costs for customers and the system as a whole. Ill-designed prices and charges
can incentivize outcomes that yield lower costs for some customers but may not necessarily yield
lower costs system-wide.
4.6 Implementation Issues 3. Demand and DERs can respond to all of these prices,
creating value for themselves and increasing the
The material presented in this chapter has illustrated overall efficiency of the power system. Demand and
how prices and charges for different electricity services DERs can also sign contracts (individually or via
aggregators) with system operators to commit to
vary in time and by location and has demonstrated how
specific response patterns in relation to the provision
economically rational agents respond to these economic
of each of these services or any other services that
signals in ways that impact the power system conditions system operators may require.
that initially produced these very signals. More specifically:
4. These three levels of differentiation (“granularity”) in
1. There are significant temporal and locational time, location, and the services that are considered,
differences in prices for electrical energy — i.e., active must be reflected in the design of efficient economic
and reactive power — at the wholesale level that signals intended for the agents connected to the
depend upon the operating conditions and designs distribution network. Our simulations of the responses
of specific power systems. But significant variations of agents to an ensemble of prices and charges featuring
in energy prices can occur within the distribution different granularity levels have shown significant
network as well, at all voltage levels. These locational variation in both operational and investment outcomes.
energy prices internalize the costs of energy This can translate into substantial gains in global
production as well as the impact of network losses and efficiency for the power system and individual agents
operating constraints. when granularity is increased.
The fact that significant temporal and locational 3. Wholesale energy prices can be extended to all
voltage levels of the distribution network by means
differences exist in the “efficient ideal” real-time
of loss factors with different possible levels of
prices and charges of the most significant services detail or accuracy. Loss factors should change in
does not necessarily imply that the maximum level of time depending on the level of power flows in every
granularity must be adopted to achieve a satisfactory direction, and they may be considered uniform per
level of efficiency. What follows is a discussion of ad hoc voltage level or differentiated by distribution zones
simplifications that may reduce implementation costs or even feeders. At peak times (of demand or of
distributed generation) loss factors may depart from
significantly without a significant efficiency loss. In some
unity significantly, creating a difference in value
cases, the simplifications may be required because of the between centralized and distributed generation that
difficulty of modifying well-established or, on the contrary, could decide the financial viability of some DERs (at
very recently adopted regulation. Other simplifications least it would impact their efficient response) and
should be avoided at all costs. In most cases, the therefore must not be ignored
following sequence of improvements toward “getting the 4. Depending upon the characteristics of the power
prices and charges right” will capture low-hanging fruit system, peak capacity signals may become critically
first at reasonable implementation costs: important to reduce the need for future investments in
generation and networks. Once there is an advanced
1. The first recommendation is to fix a flaw in tariff meter available that can record the hourly or subhourly
design that has important consequences. Policy costs injection or withdrawal of power at every connection
and residual network costs should be removed from point, there is no conceptual difficulty in applying
the volumetric ($/kWh) component of the tariff and coincidental peak capacity charges for firm capacity
charged differently, preferably as an annual lump sum and for responsibility in network investment. This
(conveniently distributed in monthly installments), with notably enhances the level of comprehensiveness
the magnitude of each customer’s charge dependent of the system of prices and charges without adding
upon some proxy metric of lack of price elasticity or significant implementation effort or cost, although the
some measure of wealth, such as the property tax or difficulty of regulatory design and acceptance should
the size of a system user’s dwelling, or just via ordinary not be underestimated.
taxes.55 The seriousness of the threat of grid defection
must be carefully considered in determining which costs 5. It has been demonstrated in US electricity markets
are included in the electricity tariff. that transmission LMPs can be computed precisely
and frequently, and that sometimes the differences
2. In most power systems, energy prices vary between LMPs can be significant between nodes of a
significantly over time each day (except in those power system at a given moment in time. These prices
systems with large amounts of hydro storage). In have excellent spatial granularity, but are presently
typical power systems with significant proportions of only applied to generation, while broad average zonal
thermal generation, high penetration of intermittent prices are applied to the rest of the agents. This policy
renewables with zero variable costs increases this has to be reconsidered and the loss of efficiency
time variability. Hourly or subhourly energy prices evaluated, in particular with regard to the participation
can be easily communicated to end customers via of DERs in the wholesale markets and the impact on
price-responsive demand.
55 If there is any indication of the actual or potential “contribution to the peaks” of
each customer (e.g., in some countries each customer has to contract a certain
amount of peak power, which cannot be exceeded at any time without triggering 6. The current situation is very different in the EU
the disconnection of supply), then some fraction of the residual network costs wholesale market, where only very coarse zonal
could be allocated in proportion to this amount of contracted power, as another
proxy to the inverse of price elasticity and a crude incentive for peak reduction in
prices are used, and most of the granular spatial
the absence of better signals. However, this can incentivize the use of batteries or differentiation is lost. Chapter 7.2.1.2 discusses in
distributed generation to reduce the charges for policy and residual network costs.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 131
detail the EU market strategy, where the level of larger transformers, or more voltage regulators — may
representation of the transmission network has been now be addressed by aggregations of DERs committed
simplified in favor of an easier integration of the to supply services via competitive auctions.57 It is critical
national markets into a single EU trading platform
that those commitments provide the same level of
while maintaining national control of system operation.
certainty as network assets do, in order to ensure the
7. An interesting question is whether the lack of detailed same level of reliability. The next three chapters identify
locational prices at transmission level — as is the
the inefficient barriers that may exist at the distribution
case in the European Union, and also up to a point
and bulk power levels to the efficient participation of
in the United States for demand and DERs — implies
that we have to give up the locational signals at the DERs in the provision of electricity services, and they offer
distribution level. Certainly, averaging LMPs over a suggestions for the removal of these barriers.
large zone sacrifices that locational value. But we
have to realize that, despite any averaging of LMPs
at the transmission level, we may want to signal
that DERs have a high locational value in one feeder
Granularity matters. The prices and
and none in another (see Chapter 8 for a detailed regulated charges for electricity services vary
treatment of locational values). Therefore, despite the significantly at different times and in different
inconsistency, it can make economic sense to apply
detailed locational signals at the distribution level,
locations in electricity networks. Progressively
even if it is not done in the transmission network.56 improving the temporal and locational
8. The next step is to further differentiate by location granularity of prices and charges can deliver
at the distribution level. Presently, active network increased social welfare; however, these
constraints at the distribution level are infrequent, benefits must be balanced against the costs,
since, without adequate demand and DER response, a
network constraint violation can only be fixed by load complexity, and potential equity concerns of
or generation curtailment. A full alternating current implementation.
representation of power flow, with active and reactive
components, along with both active and reactive
prices, will be needed to capture the impact of network
constraints, which in distribution are frequently related
to voltage problems.
58 For the sake of simplicity, in this discussion we ignore the fact that in power
systems with liberalized retailing this statement is only valid for network charges,
not necessarily for retail prices, as each customer may contract the energy supply
from a different retailer.
59 Indeed, this may not always be the case, as volumetric consumption is imperfectly
correlated with wealth (Wood et al. 2016).
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 133
References Costello, K.W., and R.C. Hemphill. 2014. “Electric Utilities’
‘Death Spiral’: Hyperbole or Reality?” The Electricity
Abdelmotteleb, I., T. Gómez, and J. Reneses. 2016. Journal 27 (10): 7–26. doi:10.1016/j.tej.2014.09.011.
“Distribution Network Cost Allocation Using a Locational
and Temporal Cost Reflective Methodology.” Power CNMC. 2015. “Report in the Provisional Settlement
Systems Computation Conference (PSCC), 2016, 1–7. Power 1/2015 of the Power Sector: Analysis of Results and
Systems Computation Conference. ieeexplore.ieee.org/ Monthly Monitoring of the Annual Estimate of Revenues
Loss-factor-and-regional-boundaries/-/media/ files/files/ec-fraunhofer-isi-ecofys-2014-design-features-
BEAA8BCF75814061B12C1427D21F41A1.ashx. of-support-schemes.pdf.
Ausgrid. 2011. “Distribution Loss Factor Calculation EIA. 2014. “Price Elasticities for Energy Use in Buildings of
Batlle, C. 2011. “A Method for Allocating Renewable and Off-Grid Power for Residential Consumers.”
Energy Source Subsidies among Final Energy Consumers.” Electric Power Research Institute. www.epri.
policy charges and electricity bills for a distributed future: Renewable Energy Self-Consumption.” ec.europa.
Biggar, D.R. and M. Hesamzadeh. 2014. The Economics and Market Overview.” ec.europa.eu/eurostat/
Bonbright, J. 1961. Principles of Public Utility Rates. New Felder, F.A. and R. Athawale. 2014. “The Life and Death
York, NY: Columbia University Press. of the Utility Death Spiral.” The Electricity Journal 27 (6):
9–16.
Caramanis, M., E. Ntakou, W.W. Hogan, A. Chakrabortty,
and J. Schoene. 2016. “Co-Optimization of Power FERC. 2016. “PJM at a Glance.” Federal Energy Regulatory
ISO New England. 2015. “The Importance of a Molzahn, D.K., J.T. Holzer, B.C. Lesieutre, and C.L.
Performance-Based Capacity Market to Ensure Reliability DeMarco. 2013. “Implementation of a Large-Scale
as the Grid Adapts to a Renewable Energy Future.” Optimal Power Flow Solver Based on Semidefinite
Revised discussion paper. October 2015. Programming.” IEEE Transactions on Power Systems 28 (4):
3987–98. doi:10.1109/TPWRS.2013.2258044.
Joskow, P. and C. Wolfram. 2012. “Dynamic Pricing
of Electricity.” American Economic Review: Papers and NARUC. 2016.“ Distributed Energy Resources Rate
Proceedings, 102 (3): 381-385. Design and Compensation.” National Association of
Regulatory Utility Commissioners. http://www.naruc.org/
JRC Smart Electricity Systems and Interoperability. 2016.
rate-design/.
“Smart Metering Deployment in the European Union.”
October 31, 2016. ses.jrc.ec.europa.eu/smart-metering- Neme, C. and R. Sedano. 2012. “US Experience with
deployment-european-union. Efficiency as a Transmission and Distribution System
Resource.” Montpelier, VT: Regulatory Assistance Project.
Kroposki, B. 2016. “Can Solar Save the Grid?”
February 2012.
IEEE Spectrum 53 (11): 42–47. doi:10.1109/
MSPEC.2016.7607026. NIST. 2014. “Interagency Report on Guidelines for Smart
Grid Cybersecurity.” NISTIR 7628: 1-3. National Institute
Lewis, W.A. 1941. “The Two-Part Tariff.” Economica, New
of Standards and Technology. September 2014.
Series 8 (31): 249–70.
CHAPTER 4: A Comprehensive and Efficient System of Prices and Regulated Charges for Electricity Services 135
Ntakou, E., and M. Caramanis. 2014. “Price Discovery in Rubio-Oderiz, F.J. 1999. “Methodology of Transmission
Dynamic Power Markets with Low-Voltage Distribution- Network Cost Allocation under a Competitive Regulatory
Network Participants.” 2014 IEEE PES T&D Conference Framework.” Doctoral dissertation (in Spanish). Madrid,
and Exposition, April 2014. IEEE, 1–5. doi:10.1109/ Spain: Institute for Research in Technology, Universidad
TDC.2014.6863212. Pontificia Comillas de Madrid.
NYISO. 2016. “2016 Load and Capacity Data ‘Gold Book.’” Rubio-Oderiz, F.J., and I.J. Perez-Arriaga. 2000. “Marginal
Rensselaer, NY: New York Independent System Operator. Pricing of Transmission Services: A Comparative Analysis
April 2016. of Network Cost Allocation Methods.” IEEE Transactions on
Power Systems 15 (1): 448–54.
Ofgem. 2014. “Bidding Zones Literature Review.” London,
United Kingdom: Office of Gas and Electricity Markets. Schneider, I., and C.R. Sunstein. 2016. “Behavioral
www.ofgem.gov.uk/sites/default/files/docs/2014/10/ Considerations for Effective Time-Varying Electricity
fta_bidding_zone_configuration_literature_review_1.pdf. Prices.” SSRN Scholarly Paper ID 2841049. Rochester,
NY: Social Science Research Network. papers.ssrn.com/
Ofgem. 2015. “Wholesale Energy Markets in 2015.”
abstract=2841049.
London, United Kingdom: Office of Gas and Electricity
Markets. September 9, 2015. Schweppe, F.C. 1978. “Power Systems ‘2000’: Hierarchical
Control Strategies.” IEEE Spectrum 15 (7): 42–47.
OMIE. 2016. Price Report 2015. Operador del Mercado
doi:10.1109/MSPEC.1978.6367813.
Ibérico de Energia (Spanish market operator).
Schweppe, F.C., M.C. Caramanis, R.D. Tabors, and
Pérez-Arriaga, I., and Y. Smeers, 2003. “Guidelines on
R.E. Bohn. 1988. Spot Pricing of Electricity. Dordrecht,
Tariff Setting.” In Transport Pricing of Electricity Networks,
Netherlands, and Norwell, MA: Kluwer Academic
Francois Leveque, ed. Dordrecht, Netherlands, and
Publishers.
Norwell, MA: Kluwer Academic Publishers. ISBN:
1402074557. August 2003. Warner-Freeman, J. 2016. “PJM Markets Report.”
PJM. www.pjm.com/~/media/committees-groups/
Pérez-Arriaga, I., S. Ruester, S. Schwenen, C. Batlle, and
committees/mc/20161024-webinar/20161024-item-
J.-M. Glachant. 2013. From Distribution Networks to Smart
08a-markets-report.ashx.
Distribution Systems: Rethinking the Regulation of European
Electricity DSOs. THINK Project, Topic 12. Florence, Italy: Wood, L., R. Hemphill, J. Howat, R. Cavanagh, S.
European University Institute. Borenstein, J. Deason, and L. Schwartz. 2016. “Recovery of
Utility Fixed Costs: Utility, Consumer, Environmental and
Pérez-Arriaga, I., and A. Bharatkumar. 2014. “A
Economist Perspectives.” Berkeley, CA: Lawrence Berkeley
Framework for Redesigning Distribution Network Use of
National Laboratory. www.publications.lbl.gov/islandora/
System Charges Under High Penetration of Distributed
object/ir%3A1005742.
Energy Resources: New Principles for New Problems.”
Cambridge, MA: MIT Center for Environmental and
Energy Policy Research, Massachusetts Institute of
Technology. ceepr.mit.edu/publications/reprints/614.
05
The Future of the Regulated
Network Utility Business Model
2 The authors note that establishing the allowed revenues of the utility, known
variously as the rate-making or remuneration process, is only one of the regulator’s
1 In this chapter, we begin from the current situation and assume that regulated core responsibilities. A number of important regulatory challenges fall outside
distribution utilities are responsible for network planning, construction, the scope of this paper, which focuses only on improving the core rate-making or
maintenance, and operation. Going forward, a thorough reconsideration of remuneration process. For example, regulators will likely need to establish a suite
electricity industry structure and assignment of core responsibilities for market of performance incentives as well as the core incentives established by the process
platform operation, system operation and planning, and network provision herein (Cossent 2013). In addition, after the total volume of allowed revenues is
is needed. This may result in different structural arrangements, and these established, rates or tariffs for network users must be designed to ensure allowed
considerations are discussed thoroughly in Chapter 6. If responsibility for network cost recovery, establish efficient signals for network use, and address other
planning and operation, as well as network construction and maintenance, are regulatory concerns, such as equity. For more on this topic, see Chapter 4 (also
ultimately divided among multiple entities (e.g., some form of independent Pérez-Arriaga and Bharatkumar 2014).
distribution system operator), that may entail changes to the proposals herein, 3 The level of information asymmetry varies across jurisdictions. Regulators try to
although in general, the regulatory measures recommended in this chapter apply mitigate the information asymmetry by requesting ample technical and economic
equally well to any profit-motivated firm responsible for optimizing the design, information from the companies; this is the case in the United Kingdom, Spain, and
operation, and maintenance of distribution systems. other jurisdictions.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 137
uncertainty about future technological change and 2013). Advanced metering, dynamic market-based
demand for network services (Cossent 2013; Ofgem prices, time-varying rates, and energy management
2013a). To reduce costs for consumers, regulators are systems have the potential to make electricity loads more
simultaneously concerned with two forms of efficiency: responsive to economic and operational signals than ever
productive efficiency, which has to do with firms finding before (Conchado and Linares 2012; Hurley, Peterson,
the least costly way to produce a given output or service and Whited 2013; Schisler, Sick, and Brief 2008). Efficient
(in both the short and long term),4 and allocative price signals, new information and control systems, and/
efficiency, which is concerned with ensuring consumers or novel market actors are necessary to manage and
pay prices that reflect incurred production costs, coordinate each of these DERs and their associated
minimizing the economic “rents” extracted by producers. services (see also Chapters 4 and 6).
In practice, regulators must carefully balance inherent
These emerging technologies constitute a set of
trade-offs in the design of regulatory mechanisms,
important new users of distribution systems, potential
incentivizing productive efficiency by rewarding utilities
new competitors for the delivery of electricity services
for pursuing cost savings on the one hand and maximizing
to end users (Kind 2013), and possible suppliers of
allocative efficiency by minimizing rents collected from
services to distribution companies seeking to harness
ratepayers on the other hand (Joskow 2014).5 Finally,
DER capabilities to avoid network investments or improve
regulators must achieve this balance of efficiency
system performance (Poudineh and Jamasb 2014;
incentives all while preserving incentives for quality of
Trebolle et al. 2010).
service and loss reduction (Gómez 2013; Cossent 2013).
Distribution utilities may need to make substantial
On top of these persistent challenges, the proliferation
investments to accommodate increased penetration
of distributed energy resources (DERs) and information
of DERs. In many jurisdictions, these new investments
and communications technologies (ICTs) is now actively
will coincide with significant expenditures necessary
transforming the delivery of electricity services and
to install and manage advanced meters and modernize
the use and management of distribution systems in
aging distribution systems to take advantage of new ICT
many jurisdictions. Distributed generation and storage
capabilities and active system management techniques
introduce bidirectional power flows and, at significant
(Cossent et al. 2011; Eurelectric 2013; MA DPU 2014;
penetration levels, entail profound changes to the
NYDPS 2015a). At the same time, the pace of change and
real-time operation and the needs for investment in
impact of distributed energy resources on distribution
distribution systems (Cossent, Gómez, and Frías 2009;
systems is highly uncertain. For example, while solar
Cossent et al. 2011; Denholm et al. 2013; Olmos et al.
photovoltaic (PV) systems generated less than 1 percent
2013; Pudjianto et al. 2014; Strbac et al. 2012; Vergara
of Germany’s electricity in 2008, solar supplied 7.5
et al. 2014). Widespread electric vehicle adoption could
percent of German electricity consumption and roughly
likewise necessitate new network investments and may
20 percent of installed capacity in 2015 (Wirth 2016).
enable new “vehicle to grid” services (Fernández et al.
While rapid solar adoption in Germany was principally
2011; Gómez et al. 2011; Momber, Gómez, and Söder
of DERs exacerbates fundamental regulatory the cost-saving opportunities made available by DERs,
smart grid technologies, and active system management
challenges and strains conventional approaches
techniques. Utilities only profit from any realized savings
to the remuneration of distribution utilities.
until the next “rate case,” when regulators reset rates to
Unless proactive reforms are made to update align with the cost of providing service. Utilities are thus
the regulation of distribution utilities, outdated encouraged to focus primarily on short-term cost savings,
network regulation may become a key barrier to sacrificing the opportunities that could be unlocked if
the efficient evolution of power systems. utilities were incentivized to invest with a longer-term
view. In addition, this approach requires regulatory review
of expenditures associated with thousands of individual
distribution system assets, which has always posed an
5.1.1 New uses of distribution networks expensive challenge for regulatory commissions with
and drivers of network costs limited staff and resources (Gómez 2013). The changing
First, new demands on networks, new alternatives nature of cost drivers and the emergence of novel
to traditional network investments, and increased cost-saving opportunities will further aggravate this
competition for end-user service delivery will pressure challenge, making it difficult for regulators to identify and
distribution utilities to focus on delivering improved disallow all but the most obviously imprudent or wasteful
outputs at a competitive cost. investments, further weakening incentives for firms to
manage productive efficiency.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 139
Finally, cost of service regulation is typically backward- rapidly evolve, benchmarking based on past utility
looking in nature, focused on reviewing the prudency performance or cost will no longer provide an accurate
of incurred investments after the fact. For utilities in an estimate of the forward-looking efficient frontier. At
actively evolving marketplace, this introduces substantial the same time, different patterns of DER adoption in
regulatory risk that can impede utility efforts to innovate different regions can introduce much more heterogeneity
and take advantage of new technologies and capabilities. between distribution network costs, further challenging
In reviewing the prudence of utility investments, regulators statistical benchmarking approaches (Cossent 2013).
typically rely on the incremental development of For example, the availability of solar, wind, biomass/
established best practices with the implicit assumption biogas, and combined heat and power resources differs
that the past is an appropriate guide for the future. substantially from location to location and is likely to
Traditional regulation thus frequently requires utilities to lead to the divergent evolution of distribution networks
justify novel investments and departures from established in different regions. Without improved, forward-looking
practices by proving that such changes will result in a net tools to assist in accurately estimating efficient network
reduction in utility costs (Malkin and Centolella 2013). If costs, regulators may set ex ante revenues that are poorly
a utility adopts a novel technology that fails to perform aligned with realized costs, leading either to excess utility
as expected, regulators may disallow cost recovery. As profits and reduced allocative efficiency (if revenues are
a result, utilities are often slow to adopt innovative too generous) or to increased risk that firms will not be
technologies and practices and may instead go through able to finance necessary investments adequately (if
a protracted cycle of internal testing and performance revenues are too low).
validation, small-scale pilot projects, data collection and
In addition, as the sector evolves, distribution utilities are
assessment, and presenting results to regulators before
likely to develop more intimate and immediate knowledge
finally, after many years, adopting improved technologies
about new cost drivers and opportunities than the regulator,
or practices system-wide. Cost of service regulation can
heightening information asymmetries and creating new
thus present a major barrier to the evolution of distribution
opportunities for strategic behavior. Tools to overcome the
utilities in light of both changing customer needs and new
regulator’s information disadvantages will thus become
DERs and smart grid technology capabilities.
even more critical for effective incentive regulation.
Regulatory frameworks that employ a forward-looking,
In summary, regulators must be equipped with forward-
multi-year revenue trajectory or “revenue cap” (such as
looking tools to identify the impacts of new DER-related
RPI-X and similar frameworks) are also challenged by
network uses on distribution costs and overcome
the evolving nature of the electricity marketplace. The
information asymmetries. In addition, regulators need
emergence of new cost drivers and changing customer
remuneration mechanisms that both incentivize utilities
needs make it increasingly difficult to establish an
to accommodate cost-effective DERs and to take
effective forward-looking, or ex ante, revenue (or
advantage of new DERs or smart grid opportunities
price) cap. Regulators often employ statistical frontier
to improve cost and performance outputs. As always,
benchmarking and yardstick approaches to assist them
regulators must also ensure firms can raise necessary
in establishing ex ante estimates of efficient network
debt and equity to finance needed investments and, to
costs (Jamasb and Pollitt 2003; ACC 2012a, 2012b). Yet
the extent feasible, limit firms’ rents.
as network uses, cost drivers, and utility best practices
CHAPTER 5: The Future of the Regulated Network Utility Business Model 141
updated to fully exploit new opportunities to effectively 5.1.4 Need for increased levels
balance increasingly important trade-offs between these of innovation
two expenditure categories.
To keep pace with a rapidly evolving landscape, electricity
Under traditional cost of service regulation, utilities earn distribution utilities will also need to focus more on
a regulated return only on capital investments. Allowed harnessing long-term innovation—by participating in
returns are calculated based on the utility’s “rate base” applied research and development efforts, investing in
or regulated asset base, which includes the cumulative, demonstration projects, engaging in the technological
non-depreciated share of capitalized expenditures. learning that emerges from those projects, and
Under cost of service regulation, utilities can thus be disseminating knowledge and best practices between
discouraged from reducing CAPEX, as this may impact network utilities. Uncertainty about how networks will
their rate base and allowed returns. At the same time, the evolve implies that the technological solutions that will
intrinsically poor incentives for cost saving under cost of lead to the greatest levels of productive efficiency in
service approaches make it unlikely that firms will fully the medium to long term (i.e., in periods that extend
exploit the most efficient trade-offs between capital and beyond the regulatory period) are also uncertain. The
operational expenditures. technologies and systems that will be most efficient for
facilitating active network management in distribution
While multi-year revenue trajectories or revenue caps
networks with high penetrations of DERs are simply
will reward firms for efficiently reducing total costs, this
not known today. Therefore, there is a need for greater
form of regulation can also distort incentives between
investment in demonstration projects and accelerated
savings achieved via reductions in CAPEX versus OPEX.
knowledge-sharing or “spillovers” among utilities. This
While incentive regulation will reward the utility equally
will involve undertaking experimental projects where the
for saving a dollar of CAPEX or a dollar of OPEX, if only
potential cost savings are inherently uncertain, may only
CAPEX is capitalized into the utility’s regulated asset base,
be realized in the medium to long term, if at all, and may
then that dollar in reduced CAPEX will also involve a
not fully accrue to the utility incurring the costs due to
reduction in the regulated asset base and thus a reduction
spillover. However, despite the need for increased levels of
in the allowed return on equity and a corresponding
long-term innovation, spending on research, development,
decline in net profit for shareholders. This decline in net
and demonstration (RD&D) by network utilities has been
profit will offset some portion of the efficiency-related
declining (Meeus and Saguan 2011). Today’s regulatory
income, distorting trade-offs between OPEX and CAPEX
frameworks, including both cost of service and incentive
and potentially encouraging over-investment (Ofgem
regulation, do not adequately incentivize these types of
2009, 2013b).
risky projects and the technological learning that emerges
Regardless of which regulatory approach is used, from them.
regulators therefore need mechanisms to equalize
Under cost of service regulation, utilities are only
incentives for CAPEX and OPEX savings and to ensure
incentivized to engage in innovation to the degree that
utilities fully exploit these opportunities.
short-term cost savings can be retained by the utility
until the next regulatory review or rate case (Bailey
1974; Malkin and Centollela 2013). Since this period
CHAPTER 5: The Future of the Regulated Network Utility Business Model 143
Cybersecurity investments suffer from a type of market gather more detailed information about network users
failure in which the benefit of secure operations is shared and as the number of Internet-connected devices in
among electricity providers and network users, but each distribution networks rises, privacy concerns may be
actor’s private incentive to invest in adequate cybersecurity exacerbated. Regulators may therefore need to adopt new
preparedness is too low to justify the cost. Network utilities, approaches to privacy regulation. For example, EU privacy
many of which face slow or negative electricity demand regulations approved in April 2016 give citizens control
growth and relatively uncertain financial futures, tend to of their personal data and create a uniformly high level of
make investments that increase financial returns rather data protection (European Union 2016b).
than spend money on cybersecurity, which has limited
direct financial benefit. Network utilities will therefore 5.1.6 New regulatory tools for a more
require additional regulatory incentives to invest in distributed power system
adequate levels of cyber protection.
To address the combination of challenges described above,
Current cybersecurity regulations vary across regulators can employ a novel combination of established
transmission and distribution systems, across regions, best practices or state-of-the-art regulatory tools. The
and across utility-owned and independently-owned remainder of this chapter describes key regulatory
DERs. In the United States, the North American priorities and possible mechanisms for accomplishing
Electric Reliability Corporation (NERC) has developed these objectives. At the broadest level, these regulatory
cybersecurity regulations at the bulk power and priorities can be divided into two categories. In Section
transmission levels (NERC 2016). In Europe, the first 5.2, we present a set of priorities for improving traditional
European legislation on cybersecurity, the Network and approaches to establishing the core remuneration of
Information Security (NIS) Directive, entered into force distribution utilities, accounting for new cost drivers,
in July 2016 (European Union 2016a). Regulations at the new demands and uses of the distribution network, and
distribution-system level (including DERs and customer increased uncertainty while establishing incentives for
connection points) are lacking, therefore a baseline set cost-saving efficiencies. Section 5.3 then discusses
of cybersecurity standards for all distribution networks additional measures necessary to incentivize utilities
is needed. Continuous improvements in best practices to improve specific outputs that are not captured by
will also be required to meet evolving cyber threats. core remuneration methods, including quality of service,
Achieving good cybersecurity in the future demands losses, asset health, cybersecurity preparedness, and
good governance as well as regulation because rule- long-term innovation.
making is not sufficiently nimble to keep up with evolving
threats. Instilling good governance is difficult. Industry
organizations such as the Institute of Nuclear Power 5.2 New Methods for Establishing
Operations have been reasonably successful for the the Core Remuneration of
nuclear industry, but there is currently no equivalent for
Distribution Utilities
the cyber and physical security of electricity networks.
The emergence of DERs, as well as of ICT-enabled
In addition to strengthening cybersecurity standards
technologies and systems, has led to increased
and regulations, privacy regulations also may need to be
uncertainty about the evolution of network needs, cost
updated to address issues raised by the influx of DERs
drivers, and opportunities. As a result, the fundamental
and Internet-connected devices. As network utilities
challenges of economic regulation—namely information
asymmetries between the regulator and the utility—are
CHAPTER 5: The Future of the Regulated Network Utility Business Model 145
Forward-looking, multi-year revenue trajectories with profit- In the most extreme form, a
sharing mechanisms can reward distribution utilities for pure “revenue cap” can be
11 Indeed, in an extreme case where allowed utility revenues precisely track costs, the
only way for a utility to increase profitability is to increase investments whenever
allowed returns on capital invested exceed the utility’s cost of capital.
12 These regulatory regimes are sometimes classified as “incentive regulation.” We
avoid using that term here because all forms of regulation incentivize some
behaviors, and performance-based output incentives can be incorporated into
a cost of service regulatory regime to incentivize key outputs as well. Under the
umbrella of “incentive regulation,” we distinguish between (1) “performance-based”
mechanisms, which focus on incentivizing key outputs or improvements in quality-
of-service metrics (e.g., reducing the number of minutes per year of non-served
energy), and (2) multi-year revenue trajectory or revenue cap mechanisms,
which specifically promote production cost reductions over a multi-year period
(e.g., three to five years or more). This section is focused on the latter, discussing
mechanisms for balancing incentives for productive and allocative efficiency
and managing the challenges involved in this task, while Section 5.3.1 focuses
on performance-based or output-based incentive mechanisms. Establishing
appropriate “incentives” for desired outcomes is a central focus throughout both
sections.
13 Note that once the regulatory contract is set and the revenue trajectory established, 14 Under Ofgem’s regulations, the incentive-compatible menu of contracts is known
the utility will maximize profit by reacting efficiently to the ex post realization of as the “information quality incentive” (IQI); it debuted for capital expenditures
demand for network services and the availability of new technologies or practices in the 2004 distribution price control review process (DPCR4; Ofgem 2004;
that lower costs below its initial expectations. Incentive compatibility thus ensures Crouch 2006). The mechanism is now an integral part of Ofgem’s RIIO regulatory
utilities make their best projections available ex ante, while a multi-year revenue framework (RIIO stands for “revenue set to deliver strong incentives, innovation,
trajectory with earnings sharing incentives ensures utilities are rewarded for and outputs” [Ofgem 2010c, 2013b]) wherein the IQI is used to establish a utility’s
reducing costs as much as possible ex post. profit-sharing and total allowed expenditures.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 147
Box 5.1: Establishing an Incentive-Compatible Menu of Contracts
A menu of contracts is an important tool for incentivizing utilities to accurately reveal their expectations
of efficient future expenditures. Employing a menu of contracts in the remuneration process can thus help
reduce information asymmetry between regulators and utilities and mitigate incentives for utilities to engage
in strategic behavior (such as inflating expectations of future costs to increase allowed revenues or returns).
A menu of contracts works by establishing the strength of a profit-sharing factor that rewards cost-saving
efficiency efforts such that the sharing factor reflects the ratio between an estimate of efficient network
costs over the regulatory period submitted by the utility as compared to the regulator’s own estimate of
efficient expenditures. Using the method introduced in Cossent and Gómez (2013) and described further in
Jenkins and Pérez-Arriaga (2017), the regulator only needs to select four discretionary regulatory parameters
to create a continuous, incentive-compatible menu of contracts:
1. The weight placed on the regulator’s estimate of efficient network expenditures relative to the utility’s
estimate, ω. This weight should depend on how reliable the regulator believes his or her estimate of future
expenditures is likely to be relative to the accuracy of the firm’s estimate. A higher value places more
weight on the regulator’s estimate, while a lower value places more weight on the firm’s estimate.
2. The reference value for the profit-sharing factor (the portion of cost savings/increases to which the utility
is exposed, also known as the efficiency incentive rate), SFref, which corresponds to the case in which
the utility’s estimate of future expenditures aligns with the regulator’s estimate. A profit-sharing factor
of 1.0 corresponds to a pure revenue cap contract while a value of 0.0 corresponds to a cost of service
contract. The regulator can thus tune the sharing factor to establish the strength of efficiency incentives
faced by utilities in order to manage trade-offs between incentives for efficiency and rent extraction. This
parameter also plays an important role in managing the effects of forecast errors. Regulators should thus
take into account the degree of uncertainty about future network costs when establishing this factor. In
general, a higher profit-sharing factor (i.e., the firm is exposed to most of the risks and rewards of cost
savings) performs better under lower levels of uncertainty, while a lower profit-sharing factor (which
shares most risks and rewards with ratepayers) performs better under greater uncertainty (Schmalensee
1989; see also Ofgem 2010a, pp. 84-87, for further discussion of regulatory considerations in establishing
the sharing factor or incentive rate).
3. When the method is applied to several utilities simultaneously, the value of this parameter, SFroc , is set
to control the spread in efficiency incentives faced by different utilities during the regulatory period. The
parameter determines the change in the profit-sharing factor as the ratio between the changes in the
utility’s estimate and the regulator’s estimate. A larger value results in a wider range of profit-sharing
factors offered, while a smaller factor results in a tighter range.
4. The reference value for an additional income payment, AIref, is used to ensure the utility will always earn the
greatest profit when the firm accurately reveals its expectations of future costs—that is, it ensures “incentive
compatibility” of the menu of contracts. This reference value specifies the additional income (on top of
the allowed cost of capital) that applies when the utility’s estimate of future costs aligns exactly with the
regulator’s estimate. The selected value can be used to tune expected profit margins for the utility.
An example menu of contracts is shown in Table 5.1. The first row of the table describes the ratio between the
regulator’s estimate of efficient expenditures and the estimate submitted by the utility. Given this ratio, the
regulator sets the utility’s allowed ex ante expenditure baseline (Row 2) as well as the profit-sharing factor (Row
3) and the additional income (Row 4) awarded to ensure incentive compatibility. The first four rows of Table
5.1 thus fully define the menu of contracts, while the remaining rows illustrate the efficiency incentive earnings
(or penalties) associated with the level of actual network expenditures the utility manages to achieve ex post.
Shaded cells in these rows correspond to cases in which the utility’s ex ante expenditure forecast matches actual
expenditures, demonstrating the incentive-compatible nature of this matrix. For any realized value of network
costs (the horizontal row in the bottom half of Table 5.1), the utility will earn the greatest revenues when realized
costs match its ex ante forecast. Efficiency incentives are also preserved, as lowering realized costs below the
utility’s forecast (i.e., moving up in a vertical column) will increase the utility’s final revenues (and vice versa).
Note that while this table shows discrete values in each column, the formulas actually generate a continuous
menu of contracts for any ratio between regulator and utility expenditure estimates.
15 The menu of contracts in this table uses the following discretionary parameters: ω = 0.66; SFref = 0.7; SFroc = -0.01; AIref = 1.0.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 149
5.2.1.2 Equalizing incentives for operational and regulated asset base (from which depreciation and cost of
capital expenditures capital revenue allowances are calculated). The remainder
of TOTEX is designated as “fast money,” which is treated
In addition to incentivizing utilities to pursue cost-saving
as an annual expense. Critically, the regulator fixes these
efficiencies, regulatory mechanisms should take care
shares at the start of the regulatory period based on an
to avoid distorting a utility’s incentives to invest in
estimate of the efficient split between CAPEX and OPEX
capital assets rather than operational expenditures. As
in total expenditures. As such, the share of CAPEX and
discussed in Section 5.1.3, utilities face increased trade-
OPEX in actual utility expenditures is free to depart from
offs between traditional capital investments in network
this expected share without impacting the utility’s return
assets and novel operational and network management
on equity. Under this approach, both OPEX and CAPEX
strategies that harness DERs. To encourage utilities
savings face the same efficiency incentives—that is, a
to find the most efficient combination of capital and
dollar of OPEX savings and a dollar in CAPEX savings
operational expenditures, financial incentives related to
produce the same improvement in utility earnings—
CAPEX and OPEX need to be equalized.
freeing the utility to exploit the expanding frontier of
Incentives are typically skewed by conventional cost-saving trade-offs between both types of expenditure.16
regulatory approaches, which add approved capital
Alternative measures have been enacted by the New York
expenditures directly to a utility’s regulated asset base or
Department of Public Service (NYDPS 2015a, 2016b)
rate base, while operational expenditures are expensed
consistent with its cost of service–based regulatory
annually. Even if utilities are properly incentivized to
framework and US accounting practices. New York
pursue cost savings via a profit-sharing incentive, under
establishes allowed revenues or “base rates” based on
this financial framework, a dollar in reduced CAPEX will
a “forward test year” or a forward-looking estimate of
also involve a reduction in the utility’s regulated asset
expected utility costs. This estimate is based in part on
base and thus a reduction in the allowed return on equity
a capital investment plan submitted by the utility and
and a corresponding decline in net profit for the utility’s
reviewed by the regulator in each rate case. Once base
shareholders. This decline in net profit will offset some
rates are set for a given year, a utility could conceivably
portion of any efficiency-related income awarded by
increase earnings by withholding funds from capital
the regulator, distorting trade-offs between OPEX and
projects included in the base rates. While this provides
CAPEX and potentially encouraging overinvestment
an efficiency incentive, New York regulators were
in conventional network assets (Ofgem 2009, 2013c;
historically concerned that this incentive could lead to
Jenkins and Pérez-Arriaga 2014; NYDPS 2016b).
underinvestment and degradation of network quality,
Ofgem has developed a mechanism for equalizing these or to rewards for utilities that inflate their estimates
incentives that is known as the total expenditure or of future expenditures during rate cases. Therefore, a
“TOTEX-based” approach (Ofgem 2009, 2013c). Under a “clawback” provision was established wherein regulators
TOTEX-based approach, the regulator establishes a fixed automatically reduced a utility’s allowed revenues if
portion of total utility expenditures (TOTEX), referred to capital expenditures fell below approved levels, returning
as “slow money,” that will be capitalized into the utility’s
17 The New York Public Service Commission recognizes that the clawback reform
discussed here is “not a complete solution to issues around capital and operating
expenses” and that a TOTEX-based approach that eliminates the distinction
between capital and operational expenditures is “a more comprehensive way to
address the potential capital bias” (NYDPS 2016b, pp. 100-101). In addition, the
NYDPS restricts waivers of the clawback mechanism specifically to cases “directly
linked to a demonstration of the DER alternative that replaced the capital project.”
This implies increased regulatory risk for utilities, as they must demonstrate, on a
case-by-case basis, the particular measures undertaken to harness DERs in lieu of
network investments. A multi-year revenue trajectory with profit-sharing incentives
and a TOTEX approach to capital bias would thus provide greater regulatory
certainty, a reduced regulatory burden, and enhanced efficiency incentives.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 151
5.2.2.1 Forward-looking benchmarks for efficient that utilities know more about their expected costs and
utility expenditures opportunities than regulators—and the likelihood that the
future will be different from the past, meaning that prior
As DERs proliferate and network users become more
best practices fail to offer a good guide for future costs.
diverse and active, the demands placed on electricity
networks will evolve. Likewise, the emergence of DERs, One approach to reduce information asymmetry between
active system management, and ICT capabilities in regulators and utilities has already been described:
distribution systems will change the set of tools available the incentive-compatible menu of contracts. With
to utilities, shifting the “efficient frontier” for utility best this approach, utilities profit most when their actual
practices. Methods for establishing the allowed revenues expenditures match their reported forecasts; utilities are
for distribution utilities should therefore become thus rewarded for submitting their most accurate and
forward-looking, based not on the best practices and truthful forward-looking estimates of efficient network
customer preferences of the past, but rather on estimates costs to regulators during each regulatory review
of what an efficient utility can accomplish and what period or rate case. This method has been successfully
diverse network users are likely to demand in tomorrow’s implemented by the Ofgem since 2004 (where it is
electricity landscape. known as the Information Quality Incentive) and offers
a powerful tool to reduce
information asymmetry
(Ofgem 2004, 2009, 2013b).
The future will not look like the past, so the past no longer
provides a useful benchmark for efficient future expenditures. Regulators can also use other
Forward-looking benchmarking tools are needed. ways to incentivize utilities to
submit high-quality business
plans. For example, Ofgem
expedites the regulatory review
Diverse regulatory bodies have pioneered a number of process for the utilities that submit the highest-quality
forward-looking approaches to establish allowed revenues. business plans during each distribution price control
Fourteen US state regulatory commissions have used review process (Ofgem 2010a, 2013d).18 Utilities fast-
“forward test years” to establish utility cost of service, tracked during the first RIIO electricity distribution price
including commissions in California, New York, Wisconsin, control review process, which was completed in 2015,
and Illinois, and the experience has been overwhelmingly finished the bulk of the regulatory process a year ahead
positive (Costello 2013, 2014). Regulators in Europe and of other utilities (Lowry and Woolf 2016). This significant
elsewhere commonly employ multi-year revenue or price reduction in regulatory costs and improved regulatory
cap regulation that requires the establishment of forward- certainty offers a more intangible but still important
looking ex ante estimates of efficient utility costs as well, incentive for utilities to submit high-quality business
and they have developed a variety of tools to produce plans and reduce information asymmetry.
higher quality forecasts and benchmarks.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 153
Box 5.2: How Does a Reference Network Model Work?
An RNM typically takes as input the location and power injection/withdrawal profile of all network users
as well as a catalog containing technical and cost information about available equipment, probability of
component failure, and the cost and time burden of maintenance. Given these inputs, the RNM constructs
a network to serve all users while minimizing total costs (including capital expenditures, operational
expenditures, and a specified penalty for thermal network losses) and meeting three specified quality of
service constraints: (1) maximum system average interruption duration index (SAIDI); (2) maximum system
average interruption frequency index (SAIFI); and (3) maximum and minimum acceptable voltage range at
every node.
For regulatory benchmarking purposes, it is important to take into account the established layout of the
utility’s network and sunk investments in network components. The RNM should thus be run in a “brownfield”
or network expansion mode, taking as inputs the existing network layout and location of the utility’s network
users and specifying the layout of network reinforcements and extensions necessary to serve projected
changes in network use over the regulatory period. Regulators would thus have to require utilities to report
information on their existing networks in a standard format including the location, voltage level, contracted
capacity, and injection/withdrawal profile of all existing network connections (loads and DERs); the
layout, impedance, and capacity of electrical lines and protection devices; and the capacity and location of
transmission interconnection substations, high-/medium-voltage substations, and transformers. Explicitly
taking into account the heterogeneous nature of distribution networks is thus another key advantage of the
RNM over statistical benchmarking techniques.
The regulator must also maintain a detailed library of the standard network components used by the
RNM, including cost and performance characteristics of cables, overhead lines, distribution transformers,
substation components, and protection devices. This catalog should adequately characterize the real
investment alternatives the utility may face. The library should be updated regularly to reflect the current
cost of standard components and expanded to include any new components entering common use, such
as ICT equipment and advanced power electronics. To minimize opportunities for the strategic inflation of
reported component costs, the regulator or suitable government agency21 should determine costs for library
components by benchmarking efficient unit costs across multiple utilities.
To ensure the RNM is suitable for use in regulatory proceedings, it must be accurate enough to simulate
established industry best practices. Over time, as novel, emerging techniques such as active system
management, coordinated dispatch of DERs, or other measures become standard practice, the model
should be updated to include these practices. However, it is appropriate for the RNM used in benchmarking
to reflect current best practices at the beginning of the regulatory process. New methods adopted during
the regulatory period that successfully reduce total system costs (or improve performance) will then be
rewarded appropriately, and the model can be updated periodically to reflect any practices that have become
widespread. In this manner, this proposed method works well despite the evolving nature of distribution
network management, provided a suitable RNM is available that captures the best practices of an efficient
network utility at the beginning of the regulatory period.
21 In many US states, there may be a very limited number of distribution utilities from which any individual regulator could draw data. A federal agency such as the US Department of
Energy or a designated independent entity may assist regulators in developing RNM frameworks and standard inputs for use by regulators.
5.2.2.2 Tuning the multi-year revenue trajectory Regulators employing multi-year revenue trajectories
and profit-sharing incentives to distribute the risk have two tools to manage the impact of such benchmark
of benchmark errors errors (beyond improving the quality of the benchmark
The regulatory tools described above can help overcome itself using the tools described above).
information asymmetry and improve the accuracy of First, the length of the regulatory review process directly
regulatory benchmarks and forward-looking revenue affects its exposure to uncertainty and benchmark error
trajectories. Yet in today’s uncertain era, benchmark (Ofgem 2010a, 2013a). The longer the regulatory period,
errors can still occur when regulators fail to anticipate the greater the opportunity for the efficient frontier
the emergence of new cost-saving technologies or to evolve and move out of step with the established
practices within the regulatory period that shift the benchmark. Conversely, the shorter the period, the
efficient frontier. To some degree, the regulator’s more frequently regulators can update their estimates
forward-looking benchmark of efficient expenditures of efficient utility costs to reflect evolving best practices
can lag behind the best practices likely to emerge over (although at the cost of reduced incentives for productive
the next regulatory period. For example, if the regulatory efficiency). Second, the strength of any profit-sharing
benchmark is based on best practices at the beginning incentive also helps manage exposure to uncertainty
of the regulatory process, then the adoption of any new about future costs and demand. In particular, under
methods that successfully reduce total system costs will lower levels of uncertainty, a higher profit-sharing factor
be rewarded, and regulators can update their estimates (meaning the utility is exposed to most of the risks
of best practices accordingly. However, if the regulator’s and rewards of cost savings) performs better, while a
benchmark becomes too outdated, utilities may earn lower profit-sharing factor (which shares most risks and
substantial rents, distorting allocative efficiency. rewards with ratepayers) performs better under higher
CHAPTER 5: The Future of the Regulated Network Utility Business Model 155
levels of uncertainty (Schmalensee 1989). In both cases, • Revenue trigger provisions that specify a rule to
however, mitigating exposure to benchmark error and increase or decrease revenues by a specified amount
thus improving allocative efficiency comes at the cost of if certain pre-specified trigger events occur (e.g.,
substantial departures from forecasted network uses).
reduced incentives for pursuing cost-saving measures and
improving a utility’s productive efficiency. The twin goals • Reopener thresholds that specify clear conditions
of allocative efficiency and productive efficiency should under which expenditure and revenue determinations
will be revisited and revised (e.g., to account for major
therefore be carefully balanced. Regulators can tune both
new legislative requirements, changes in the tax code,
the length of the regulatory period and the strength of the
substantial departures from forecasted network uses,
profit-sharing factor to achieve the desired balance. or other major unanticipated cost drivers).
5.2.2.3 Automatic adjustment mechanisms to • Reviews of output requirements halfway through the
resolve forecast errors regulatory period. For example, Ofgem’s RIIO multi-year
revenue trajectory period lasts eight years but includes
In addition to benchmark errors, there will always be a mid-period review at four years to capture changes in
uncertainty regarding the evolution of network uses, output requirements.
cost of capital, and network component costs over the • Automatic adjustment factors or forward-looking
regulatory period. This uncertainty can lead to “forecast volume drivers—i.e., predetermined formulas to adjust
errors,” where costs rise or fall unexpectedly due to revenues if key cost drivers (e.g., number of customer
new network uses (e.g., growth in solar PV penetration connections, load growth, DER penetration) depart
or electric vehicle adoption) or other cost drivers (e.g., from ex ante forecasts.
changes in cost of capital or commodity prices) that fall An RNM can be employed not only to assist in initial
outside the utility’s direct influence. If forecast errors are benchmarking of efficient utility expenditures, but also
unmitigated, the economic sustainability of regulated to calculate ex ante automatic adjustment factors to
utilities could be jeopardized if allowed revenues are set correct the estimate of efficient network expenditures
too low; alternatively, utilities could profit from significant to account for any deviations from the forecast for load
economic rents if the revenue trajectory is too generous. growth and DER adoption. Jenkins and Pérez-Arriaga
For example, the rapid and unanticipated growth of (2017) describe in detail how an RNM can be used to
electric vehicle adoption could drive new demand for determine the appropriate formulas—in short, regulators
network upgrades that a multi-year revenue trajectory can employ an RNM to estimate network costs across
fails to anticipate. Alternatively, anticipated load growth a range of uncertainty scenarios designed to capture
could fail to materialize due to economic recession. The likely load patterns, DG penetration, electric vehicle
general principle should be that utility earnings are tied adoption, or other important and uncertain cost drivers.
to the exercise of managerial effort and quality of service, The regulator then runs the RNM in a brownfield mode
not to factors outside the utility’s direct influence. to calculate efficient network expansion costs under
Fortunately, regulators have developed a range of each of these scenarios. Finally, the regulator determines
mechanisms designed to manage impacts when the the relationship between deviations in cost driver
actual evolution of network uses and cost inputs values and efficient network costs by performing a
diverges from forecasts. For a discussion of uncertainty multivariate linear regression on the resulting estimated
mechanisms available for ex ante or forward-looking total network expenditures for each scenario. The
regulatory approaches, see Ofgem (2010a, 2013a). coefficients of this regression function, which we call
Options for managing uncertainty include: “delta factors,” prescribe simple formulas to adjust the
estimated TOTEX baseline ex post based on the realized
• Indexing provisions to adjust allowed expenditures
evolution of network uses or other key cost drivers. Delta
based on changes in economy-wide price indexes (e.g.,
inflation of input costs, fuel price changes, or changes factors reduce the risk that the revenue determination
in average cost of debt). will need to be reopened during the regulatory period,
CHAPTER 5: The Future of the Regulated Network Utility Business Model 157
5.3.1 Outcome-based incentives In this section, we discuss and make recommendations
regarding regulatory strategies for the design of
Short-term economic efficiency is not the only goal of
outcome-based performance incentives. We focus on
regulation. Other important outcomes include improving
four performance areas—commercial quality, continuity
the quality of electrical service, reducing energy losses,
of electrical supply, voltage quality (which together
proactively stewarding network assets, ensuring
comprise quality of service), and energy loss reduction—
workplace and public safety, and others. These outcomes
drawing from best practices in Europe and the United
are not incentivized by core
remuneration frameworks since
achieving them imposes costs on
Outcome-based performance incentives can reward utilities
distribution companies, and such
for improvements in quality of service, such as enhanced
frameworks focus on reducing
resiliency, reduced distribution losses, and improved
costs. Therefore, many regulatory
authorities have created
interconnection times.
additional incentives to promote
these critical outcomes.
Ideally, such mechanisms should aim to replicate the States before delving into specific performance areas.
outcomes of efficient markets, where consumers are However, we first present some general considerations
willing to pay more for products or services that deliver for the design and implementation of outcome-based
greater value. Value for money is the overarching concern, incentives, which will be useful for the discussion.
not simply lowest cost. Outcome-based incentives are
thus a critical complement to the cost-saving efficiency
incentives discussed above.
Performance
Metric
Today
Establish performance
target & time length Penalty
Target
Establish reward-penalty
(“bonus-malus”) scheme Reward
Time
Source: Aggarwal and Harvey (2013)
CHAPTER 5: The Future of the Regulated Network Utility Business Model 159
Box 5.3: Establishing an Economically Optimal Performance Target
Careful analysis is essential to setting any performance target for an outcome-based incentive. From
a social perspective (i.e., a perspective that includes the interests of both customers and companies),
the optimal level of performance is the level at which the marginal benefit of additional performance
equals the marginal cost of supplying it. The benefit—or willingness to pay (WTP)—of an increment
of performance improvement is usually difficult to quantify, so in practice WTP is often approximated
by its inverse: the cost of—or willingness to avoid (WTA)—incurring an increment of performance
deterioration. From this perspective, the optimal level of performance is that which minimizes the
total cost function—that is, the sum of the company’s cost function and the customers’ cost function
(Fumagalli et al. 2007; Gómez 2013; Rivier and Gómez 2000). This concept is illustrated in Figure 5.3.
Total Costs
Costs
Customer Costs
Company Costs
Quality
In practice, of course, determining these values is not straightforward. Establishing customers’ WTA
is an inexact science that usually depends upon customer surveys and can vary widely within a given
distribution network. Likewise, while estimates of a company’s cost for delivering various levels
of performance can be more accurate, these estimates too are usually rough (at best) due to the
complexity of distribution networks.
Regulators should also bear in mind that optimal levels of quality will change and may be very different
for different types of customers; therefore, performance target levels should evolve over time and
may need to become more granular. Over time, the perceived importance of a given performance
goal for network users may vary. Likewise, the cost to the distribution utility of implementing a given
performance goal may vary over time as technology costs change and as economies of scale are
realized. Consider the example of reducing connection times for network users with solar PV. If prices
Finally, regulators must be wary of the fact that oftentimes achieving a specific performance outcome
may increase a company’s OPEX and/or CAPEX efficiency. Therefore, in jurisdictions that include cost
efficiency incentives in the core regulatory framework (such as the efficiency incentives described in
Section 5.2.1), regulators must ensure that efficiency incentives and outcome-based incentives are
tightly coupled—that is, that the expected costs associated with the performance outcome are built into
the cost-efficiency trajectory to which the utility is subject. This becomes increasingly important as the
share of earnings that a network utility can achieve via outcome-based incentives becomes larger.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 161
Box 5.4: Establishing Output-based Incentive Formulas in a Reward–
Penalty Scheme
The inclusion of a financial reward–penalty scheme in a performance incentive allows the regulated
company some degree of discretion: Given the performance target and the associated financial
incentives, the company is expected to employ its superior knowledge of costs to deliver an efficient
level of performance (Sappington 2005). Financial reward–penalty schemes have delivered positive
results where they have been applied (Fumagalli et al. 2007). However, they are complex to design and
therefore should be approached with appropriate deliberation and analytical rigor.
The incentive formula of a financial reward–penalty scheme can take numerous forms, including linear,
quadratic, and step function. As much as possible, the shape and the slope of the formula should reflect
customers’ willingness to pay for increases in the performance level of a given outcome. Practical
considerations, such as ease of implementation, may also come into play in the choice of an incentive
formula. Figure 5.5 presents examples of possible functional forms that reward–penalty schemes may
take, as well as some of their potential benefits and drawbacks.
• A linear function is a popular functional form since it is simple to understand and to administer.
Figure 5.5a displays a linear function with a deadband and caps. The deadband helps reduce the
risk of the distributor achieving a reward/incurring a penalty due to a change in performance
that occurs as a result of factors outside the distributor’s control. A potential drawback of a
linear function with a deadband is that it might incentivize the distributor to perform at the most
negative performance level within the deadband (Point Z). A linear function need not have a
deadband or caps. For example, incentive formulas for TOTEX efficiency are typically linear with
no deadband or caps.
• A sigmoidal function (Figure 5.5b) may be used as an alternative to a linear function with a
deadband, since the slope of the sigmoidal function (i.e., the rate of change of the reward/
penalty) is relatively low at small deviations from the performance target. The parameters of
the function can also be tweaked so that the maximum reward and penalty occur at the same
performance level as a similar linear function with a deadband. Moreover, the fact that the
slope of the sigmoidal functions increases/decreases as the deviation from the target increases/
decreases provides an additional incentive to strive for high performance levels and avoid low
performance levels. However, a sigmoidal function may be more difficult to administer than a
linear function.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 163
5.3.1.2 Quality-of-service incentives reporting requirements and moving (if necessary) to
the establishment of a performance target and (again,
Service quality in the electricity distribution and retail
if necessary) to the establishment of a reward–penalty
sectors spans many technical and non-technical aspects.23
scheme. Introducing regulatory elements in succession
The aspects that are normally regulated can be grouped
provides time to monitor and fine-tune each of the
into three areas, one of which is non-technical and
elements, which are often quite complex.
two of which are technical. First, commercial quality
includes ensuring adequate performance on a number of Table 5.2 provides a list of some of the most frequently
service quality measures, such as the provision of new regulated commercial quality services in the European
electrical connections, meter reading, billing, and the Union. These include services that are provided before the
handling of customer requests and complaints. Second, supply of electricity begins and those that are provided
continuity of supply is related to events during which during an active contract. Services that are provided
the voltage at a customer connection drops to zero.24 during an active contract can be further divided into
Finally, voltage quality is concerned with minimizing those that are provided regularly and those that are
deviations from desired levels of voltage characteristics. provided occasionally. Services that are provided before
The design of regulation for each of these three services supply begins include connection, meter installation, and
will be significantly affected by the emergence of DERs. responses to requests for information. The metric typically
Moreover, the transition to actively managed networks used to regulate these services is the waiting time for the
will mean that DERs may increasingly participate in the provision of a service following a service request.
provision of some of these services. Each of these three
Services that are regularly provided during an active
performance dimensions, and best practices for their
contract include billing, meter reading, and customer
regulation, are briefly discussed below.
service. The performance of these services is usually
Commercial quality regulation addresses the non- assessed with metrics that measure regularity and
technical aspects of quality-of-service regulation and is accuracy such as the number of incorrect bills, the
concerned with maintaining minimum levels of quality frequency of meter readings, customer satisfaction with
for services that relate to the interaction between respect to the precision of information provided by call
service providers (the distribution utility or retailer) and center staff, etc. Services that are occasionally provided
customers. Commercial quality regulation employs all include responding to a customer’s request for technical
three of the regulatory elements described in Section assistance (e.g., to address problems with the meter or
5.3.1.1: the establishment of performance metrics the supplied voltage), responding to information requests,
(and reporting requirements), the establishment of a etc. Metrics used to regulate these services typically
minimum performance target, and (rarely) reward and include the amount of time it takes for the service
penalty schemes. However, it is recommended that provider to respond to requests and results of customer
when introducing a new commercial quality incentive, satisfaction surveys.
regulators do not adopt all three elements at once but
rather introduce them incrementally, beginning with
23 For a richer discussion of service quality regulation, see Fumagalli et al. (2007).
24 According to the European Norm EN 50160, a supply interruption is a condition in
which the voltage at the supply terminals is below 1 percent of the declared voltage.
The declared voltage is normally the nominal voltage of the system (i.e., the voltage
by which the system is designated or identified), unless a different voltage is
applied by agreement between the supplier and the customer (CENELEC 1999).
REGULAR OCCASIONAL
Providing (supply and meter) Accuracy of estimated bills Responding to failures of a distributer’s fuse
Increasing use of DERs will usher in a variety of new do not recommend the adoption of all three regulatory
considerations for commercial quality regulation. As elements at the outset. Rather, we recommend the
electricity customers increasingly become electricity gradual implementation of each regulatory element in
service providers, their ability to quickly and efficiently succession to allow ample time to fine-tune each element.
gain access to electricity services markets will become The introduction of sound continuity of supply regulation
more important. Additional issues such as the availability can take one to two years’ work (three to four years if
of and ease of access to information about electricity reliable data are unavailable) on the part of both the
services markets will also grow in importance. Network regulator and the distribution companies.
users will increasingly look to distribution utilities or
Continuity of supply regulation is primarily concerned
retailers to provide platform services that enable them
with three performance categories: the number of
to enter electricity services markets. Commercial quality
outages experienced by customers in a given time
regulation will need to keep pace with these evolving
period, the duration of outages (which includes both the
needs and priorities, and if it is successful in doing so will
average duration of outages as well as the total duration
contribute to innovation and to the transition to actively
of outages in a given period), and the intensity of outages
managed networks.
(i.e., total amount of load that was curtailed). Moreover,
Continuity of supply regulation seeks to minimize continuity of supply regulation is concerned both with
interruptions of electrical supply—that is, events during “long” unplanned interruption events and with “short”
which the voltage at a customer connection drops to events. Unplanned interruptions are classified as “long”
zero.25 In practice, continuity of supply regulation uses when they last more than three minutes, and as “short”
all three of the elements described in Section 5.3.1.1: the when they last for up to three minutes (CENELEC 1999).
establishment of performance metrics (and reporting
With respect to long unplanned interruptions, three of the
requirements), the establishment of a minimum
most frequently used system-level statistical indicators
performance target, and reward and penalty schemes.
are: average number of interruptions per customer per year
However, as with commercial quality regulation, we
(also known as the system average interruption frequency
index, SAIFI); average interruption duration per customer
25 For a richer discussion of continuity of supply regulation, see Cossent (2013).
CHAPTER 5: The Future of the Regulated Network Utility Business Model 165
Figure 5.6: Shape of the Incentive Formula for Continuity of Supply Regulation in Different Countries
per year (also known as system average interruption In practice, there is significant diversity in the incentive
duration index, SAIDI); and energy-not-supplied. functions chosen by the different countries that have
Regulation to minimize short interruptions (less than or implemented reward–penalty schemes for continuity of
equal to three minutes26) has been less common than supply. Figure 5.6 shows the shape of incentive functions
regulation to minimize long interruptions. Nonetheless, for a variety of European countries. Figure 5.6a shows
in recent years this metric has received more attention countries that use linear functions (one with no cap and
from customers (particularly industrial customers, since two with caps), while Figure 5.6b shows countries that
they are the most sensitive to short interruptions) and utilize caps as well as deadbands.
regulators. As with long interruptions, when data are
Continuity of supply regulation has been widely adopted
available for the number of customers affected by short
both in the United States and in the European Union and
interruptions, system-level indicators of performance can
has proven extremely successful in many jurisdictions
be calculated as the average number of short interruptions
in significantly reducing outages (both in number and
per customer per year (also known as the momentary
frequency). For example, as shown in Figure 5.7, in Italy
average interruption frequency index, MAIFI).
over a span of three regulatory periods (2000–2011), the
After a target has been set, the regulator must determine number of customer minutes lost was reduced from more
the functional form that relates level of quality to financial than 180 in 1999 to less than 60 in 2011 at an average
reward or penalty. To the degree possible, the functional cost to customers of 2.5 euros per customer per year.
form should reflect customers’ WTP for increases in Figure 5.7 distinguishes between customer minutes lost
quality as reflected, for example, in customer surveys. that are “within utility control” and customer minutes lost
as a result of events that are outside the control of the
distribution operator (“outside of utility control”). Italian
26 “Micro-cuts,” losses of supply that last only fractions of a second, are generally
regulated under voltage quality standards and regulations and not under continuity regulators consider outage events that are under the
of supply regulation. Micro-cuts can be caused by a fault in a piece of equipment,
changes in network configuration, and other reasons; they can be a serious problem control of the distribution utility to be those that originate
for some industrial applications.
160
140
Customer Minutes Lost (SAIDI)
120
100
80
60
40
20
0
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011
Year
Source: Lo Schiavo (2016)
in the distribution network (low or medium voltage) and Moreover, regulators will have to carefully consider how
those that are not caused by a force majeure event (i.e., an to update continuity of supply regulation in light of the
event that stresses the distribution network beyond its increasing penetrations of technologies into distribution
designed resilience limits). Outage events that are beyond networks. For an increasing number of network users,
the control of the distribution utility include those that the cost of a service interruption is not only the cost of
originate in the transmission system, those that trigger a lost energy consumption but also the cost of foregone
system-wide response, and those that are caused by force revenues from distributed generation production
majeure events. (Cossent 2013).
Increasing penetrations of DERs and ICT-enabled smart Also, the increasing use of electric vehicles will link
grid technologies will provide new opportunities and mobility to electricity supply, which may significantly
challenges for continuity of supply regulation that, if increase customers’ WTP for continuity of supply.
acted upon, will contribute to innovation and to the Finally, continued deployment of ICT-enabled smart grid
transition to actively managed networks. For example, the technologies and systems will very likely reduce the costs
presence of DERs may contribute to system reliability associated with providing continuity of supply. All of
through islanded operation in which a portion or portions these changes mean that regulators that have established
of the electricity network may function even if there outcome-based performance incentives for continuity
are electrical outages in other areas of the network of supply will have to regularly reevaluate the target
(Cossent 2013; McDermott and Dugan 2003). In this levels, incentive structures, and reward–penalty amounts
case, well-designed output-based performance incentives associated with these incentives.
for distribution utilities—combined with regulations that
Finally, voltage quality regulation is concerned with a
equalize incentives for OPEX and CAPEX—could lead to
subset of factors that describe deviations of voltage from
markets for distribution-level reliability services in which
its ideal waveform. Such deviations can lead to damage or
DERs play an active role.
to the malfunctioning of customers’ electrical equipment.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 167
To date, the economic regulation of voltage quality has Despite the complexities associated with voltage quality
been relatively rare; voltage has typically been regulated regulation, increased AMI deployment will create new
using technical standards. Nonetheless, increasing opportunities for this area of regulation. The widespread
penetrations of advanced metering infrastructure (AMI), deployment of distributed power electronics on
and thus the increasing ability to gather distribution distribution feeders will enable new volt and var (volt-
feeder-level data, is creating new opportunities for ampere reactive) optimization (VVO) capabilities that
voltage quality regulation. At the same time, interest in were previously not available. Specifically, these new
voltage quality regulation has increased in recent years capabilities will make it possible to regulate voltage in real
due to a proliferation of electronic devices that are time and within a much narrower band than is currently
extremely sensitive to voltage quality. Voltage quality required under most existing technical standards. New
also lends itself to all three of the regulatory elements VVO capabilities can contribute to significant reductions
discussed at the beginning of this section: performance in technical losses, improvements in energy efficiency,
metrics (and reporting requirements), minimum and load reduction, for example via conservation
performance targets, and reward and penalty schemes. voltage reduction (CVR), whereby voltage levels are
kept at or very close to the minimum safe voltage level.
The types of data required for voltage quality regulation
It is plausible that output-based incentives for CVR
are more technical and more difficult to obtain than
performance could begin to emerge in jurisdictions with
data for the other two types of service quality regulation.
the requisite technical capabilities.
Voltage supplied to customers has a number of
characters—including frequency, magnitude, waveform, 5.3.1.3 Minimization of energy losses
and symmetry of three-phase voltage—each of which
Electrical losses are defined as “the difference between
exhibits variations during the normal operation of an
the amount of electricity entering the transmission
electrical system (CENELEC 1999). Deviations of these
system and the aggregated consumption registered
characteristics from their nominal values are called
at end-user meter points” (ERGEG 2008) and are
voltage disturbances.
comprised of technical or physical losses (i.e., those
Due to these complexities and the large costs associated that occur as a result of heat and noise as electricity
with voltage measurement at every connection point passes through network components) and non-technical
in a network, the regulation of voltage quality has or commercial losses (i.e., electricity consumption that
typically been achieved via technical standards and is not metered such as consumption at distribution
bilateral arrangements between affected customers substations, energy theft, non-metered consumption, and
and distribution utilities. The use of outcome-based metering errors). This section only considers technical
performance incentives for the regulation of voltage or physical losses, which are the most common type of
quality has been relatively rare. The few cases that losses in most developed countries.
we are aware of are in Latin America. In Argentina,
Technical electricity losses in distribution networks
for example, economic regulation for power quality
increase costs for customers and may increase CO2
indicators was introduced in the 1990s following the
emissions (and emissions of other pollutants from power
privatization of distribution companies in Greater Buenos
plants).27 In a vertically integrated setting (in which the
Aires. Other Latin American countries introduced similar
utility is the owner/operator of power plants, transmission,
schemes following the privatization of publicly-owned
and distribution networks), the utility has an incentive
national utilities. Moreover, several jurisdictions have
to reduce technical losses to a level at which the cost of
implemented outcome-based performance standards
further loss reduction is equal to the marginal benefit of
that rely on “non-technical” metrics of voltage quality
that reduction (where benefit is defined as the upstream
to enforce voltage quality regulation. These include,
for example, customer waiting time before receiving a
27 Of course, the degree to which technical losses increase CO2 emissions depends on
response following a voltage-related request. the carbon intensity of the electricity generation fleet.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 169
On the other hand, DG and DERs may also present more efficient networks. It is also very likely that the
opportunities for energy loss reduction. For instance, use of outcome-based incentives for the improvement
DG could contribute to energy loss reduction if it were of these services has contributed to the creation of new
sited close to consumption points. Demand-response knowledge, technologies, and processes that will play
technologies could also contribute to energy loss an important role in the transition to actively managed
reductions, although the significance of this contribution electricity networks.
may depend on the price elasticity of the electricity
These facts have led some authorities to move toward
demand of customers that are participating in the
regulatory frameworks in which outcome-based
demand-response program (e.g., Shaw et al. 2009 and
performance incentives play a major role. Perhaps the
Venkatesan et al. 2012). The use of ICT-enabled smart
best example is in the United Kingdom, where in 2011
grid technologies may also contribute to technical
the UK regulator (Ofgem) launched a new regulatory
loss reductions, since these technologies will assist
framework—titled “Revenue = Incentives + Innovation +
network utilities in real-time network monitoring and
Outputs” (RIIO) —in which outcome-based performance
automation and in the dispatch of DERs that contribute
incentives play a central role in determining network
to loss reduction. However, it is likely that a significant
utility profits.
share of the loss reductions that are enabled by smart
grid technologies could be achieved as a byproduct of Figure 5.8 shows the relative strength of incentives
technology deployment that occurs via quality of supply associated with the core remuneration framework—and
regulation (Cossent 2013). With respect to electric with additional outcome-based performance incentives—
vehicles (EVs), there is ample evidence to suggest that for distribution utilities in the United Kingdom under
high penetrations of EVs are very likely to contribute to RIIO. The vertical axis is the rate of return on regulated
increased losses (Clement-Nyns et al. 2010; Fernández et equity (ROR), and the horizontal axis shows different UK
al. 2011; and Peças Lopes et al. 2011, 2009a, 2009b). distribution companies. As Figure 5.8 shows, the allowed
ROR of network utilities is linked to core efficiency
Due to the challenges associated with regulating energy
incentives (the “RPI formula effect,” “ex ante reward/
loss reductions, and the fact that these challenges
penalty,” and the “TOTEX efficiency incentive”) as well as
will only be exacerbated by increasing penetrations of
to a variety of important outcomes: improving network
DERs, regulators must tread extremely carefully with
reliability (“interruptions incentive scheme,” “guaranteed
respect to the design of performance-based incentives.
standards for reliability,” “guaranteed standards for severe
Incorporating the effect of DERs on technical losses into
weather”), improving asset health (“health index”)30,
the methodologies used to determine the reference value
improving customer service (“broad measure of customer
of losses will prove critical in the design of any incentive
service”), improving connection times and network user
mechanism aimed at loss reduction.
5.3.1.4 New frontiers for outcome-based 30 Performance incentives may be put in place to ensure that the health of
distribution network assets remains sufficiently high and the risk of network
performance regulation asset failure remains sufficiently low. The United Kingdom is a good example of
a jurisdiction that has implemented this type of incentive. In the first regulatory
The preceding paragraphs have discussed some of the period under the United Kingdom’s new RIIO framework (RIIO-ED1), the UK
regulator introduced an asset “risk index” (RI) that is a composite of an asset
best practices with respect to outcome-based regulation “health index” (HI) and an asset “criticality index” (CI). The RI is assessed on an
in four critical performance areas: commercial quality, increasing scale of 1 to 5 (Ofgem 2013e). UK distribution utilities are required
to demonstrate how their allowed investments will contribute to network asset
continuity of supply, voltage quality (which together risk mitigation, and a specific “risk score” is assigned to each target a utility
must meet. Under this scheme, if a distribution utility achieves its risk mitigation
comprise quality of service), and energy loss reduction. performance target, it will not receive a penalty or reward at the end of RIIO-ED1.
In each of these areas, the implementation of outcome- On the other hand, if a distribution utility fails to meet its risk mitigation target and
is not reasonably justified in underperforming, it will be subject to a penalty of 2.5
based performance incentives has helped steer network percent of the avoided costs that are associated with under-delivery. Likewise, if a
utility exceeds its target, it will be rewarded 2.5 percent of the incremental costs
utilities in directions that are beneficial for network
associated with over-delivery. Notably, in the United Kingdom, ensuring adequately
customers and that have led to safer, more reliable, and low levels of asset failure risk is considered secondary to the primary objectives of
safety (public and workplace safety) and network reliability (Ofgem 2013e).
engagement (“guaranteed standards for connections,” During the ex ante establishment of future costs, Ofgem
“time to connect and incentive on connections carefully considers the impact that achieving outcomes
engagement”), and reducing electricity losses (“losses might have on these costs, and network utilities are
discretionary reward”). unable to claim certain cost efficiency gains if they do not
also successfully meet performance targets.
Figure 5.8 reveals that success or failure in meeting
minimum targets for certain outcomes can lead to an The United Kingdom is not the only jurisdiction that
increase or decrease (respectively) of 2 to 3 percentage envisions outcome-based performance incentives
points in the network utility’s allowed rate of return on playing an important role in the transition to actively
regulated equity, comparable to the possible gains and managed electricity networks. In 2014, New York
losses associated with efficiency incentives that are part launched Reforming the Energy Vision (REV), an
of the core remuneration framework (Ofgem 2014). ambitious policy and regulatory initiative aimed at
decarbonizing the state’s energy sector. A major goal of
As suggested in Box 5.3, regulators must recognize
the REV initiative is the transformation of the electric
that achieving a specific performance outcome often
power system from a passively managed one to an
increases a company’s OPEX and/or CAPEX efficiency;
actively managed system in which DER generators
they therefore must link outcome performance incentives
participate in the operation and planning of networks.
(which fall outside the core remuneration framework)
Efforts are under way to completely rethink the regulation
to cost efficiency incentives in the core remuneration
of distribution utilities to align revenue streams with the
framework. The United Kingdom, which has moved
goals of active system management.
toward a more outcome-based approach to regulation,
has done a good job of coupling outcome-based One of the mechanisms by which New York’s regulatory
performance incentives and cost efficiency incentives. authority seeks to align the incentives of network utilities
CHAPTER 5: The Future of the Regulated Network Utility Business Model 171
with the goals of the REV initiative is by developing a knowledge among network utilities.31 Since it is unclear
set of outcome-based performance incentives (which how networks will evolve, it is also uncertain what
it has termed “earnings adjustment mechanisms” or technological solutions will lead to the greatest levels
EAMs) (NYDPS 2016). In particular, regulatory staff of productive efficiency in the medium to long term
have proposed outcome-based incentives that would (i.e., lengths of time that are typically longer than the
apply to five performance areas seen as critical to regulatory period). The technologies and systems that
achieving the objectives of the REV initiative: peak load will be most efficient for facilitating active network
reduction, energy efficiency, customer engagement management in distribution networks with high
(i.e., the education of, engagement with, and provision penetrations of DERs are simply not known with precision
of data to customers), affordability (i.e., the promotion today. Therefore, there is a need for greater investment
of low-income customer participation in DERs and a in demonstration projects and applied research and
reduction in the number of terminations and arrearages), development (NAS 2016).
and interconnection (i.e., improvements in the speed
One of the most straightforward regulatory approaches
and affordability of distributed generation units).
for achieving increased levels of innovation is to
The regulatory mechanisms that would incentivize
offer explicit financial incentives, outside of the core
performance improvements in these areas are explicitly
remuneration framework, for network utilities to spearhead
temporary and would be updated and reassessed
demonstration projects. The most common way to do this
regularly (NYDPS 2016).
is via so-called “input-based” financial incentives, whereby
Notably, some of the performance areas that the regulatory demonstration projects are capitalized and included in
commission has decided to focus on (namely peak load the regulated asset base (in a cost of service regulatory
reduction and energy efficiency) are largely beyond the context). It is also possible to incentivize demonstration
control of the distribution utility, since events and actions projects via output-based financial incentives such as
external to the utility will affect
how these performance areas
change. It is partly for this Incentives for longer-term innovation are needed to accelerate
reason that regulators decided investment in applied R&D and demonstration projects that can
to link these performance
provide lessons about the capabilities of novel technologies that
incentives to financial rewards
may be risky or may have long-term payback periods.
but not to financial penalties.
Distribution utilities will be
remunerated if performance
improves but will not be penalized if performance degrades. those discussed in Section 5.3.1. However, this is not
This is a significant departure from the design of outcome- recommended, since the outcomes of such projects are
based performance incentives in other jurisdictions, in inherently uncertain. Nonetheless, as was discussed in
which incentives are linked to outcomes that can be fully or Section 5.3.1, outcome-based performance incentives will
mostly attributed to distribution utility performance. inevitably incentivize some additional near-term innovation.
This portion of the report presents case studies of three
5.3.2 Explicit incentives for jurisdictions that have had success promoting innovation
long-term innovation in electricity networks via input-based regulatory
mechanisms. The case studies illuminate best practices
Finally, increased uncertainty about the evolution of
from Europe and the United States and may provide
network needs, cost drivers, and opportunities will
guidance to regulatory authorities considering the adoption
intensify the need for long-term innovation, including
of incentives for innovation.
expanded investment in demonstration projects that
produce technological learning and dissemination of 31 Here we refer to activities focused on supporting potential solutions that may
not be commercially ready within one regulatory period, or that are extremely
uncertain with respect to performance or returns. The objective is for utilities to
become consistent adopters and integrators of novel solutions.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 173
Figure 5.9: Installed Capacity of Intermittent Renewable Energy Sources in Italy, 2006–2015
30 Solar PV
Wind
25
Installed Capacity (GW)
20
15
10
0
2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Year
As a result of these challenges, regulators in Italy have The results of this research motivated the Italian
implemented a number of new regulations, including regulatory authority to solicit competitive offers for
32 Enel is the largest distribution utility in Italy, serving 86 percent of the country’s
total electricity demand in 2011.
REVERSE POWER FLOW > 1% OF TIME REVERSE POWER FLOW > 5% OF TIME
1100
1000
900
Number of Primary Substations
800
700
600
500
400
300
200
100
0
2010 2011 2012 2013 2014 2015 2010 2011 2012 2013 2014 2015
Year
Source: Lo Schiavo (2016)
innovative demonstration projects with the primary operating conditions (voltage, current, frequency), as a
objective of reducing RPT and thereby improving the result of the pilot project.
DG hosting capacity of distribution networks (AEEG
Eight pilot projects were deployed under this regulatory
2010).33 A committee of experts conducted a selection
framework, and as a result, several functionalities
process on behalf of the regulatory authority, and
pertaining to active network management have been
selected demonstration projects were capitalized and
tested. In particular, the pilot projects have shed
benefited from an additional 2 percent weighted average
light on the interactions between distribution system
cost of capital (WACC) on top of the standard rate of
operators (DSOs) and transmission system operators
return for a period of 12 years.34 The committee assessed
(TSOs), smart voltage control, active power modulation,
the different proposals using a variety of parameters,
anti-islanding functionalities, fast MV fault isolation,
including qualitative indicators and technical scores, the
and electricity storage. The Italian regulatory authority
cost of the project, and an indicator known as Psmart.
is currently focused on using lessons from the pilot
Psmart is defined as the increase in DG production that
projects to develop outcome-based incentives related
can be connected to the grid without undermining safe
to the transition to active system management. A
33 Reverse power flows can contribute to overvoltage in electricity networks and detailed analysis of various active management
therefore can pose significant operational, safety, and security risks. The DG functionalities was undertaken in 2015, and it was
hosting capacity of a distribution network is constrained by these risks. Therefore
innovations that can reduce reverse power flows will increase the DG hosting proposed that two functionalities—TSO–DSO data
capacity of distribution networks.
exchange and smart voltage control in MV networks—
34 To participate in the selection process, demonstration projects had to meet three
main requirements: (1) distribution networks in which the projects were deployed should be considered for outcome-based regulation
had to show an RPT of at least 1 percent annually; (2) projects had to focus on
development and deployment (i.e., not basic research); and (3) projects had to
(AEEG 2015a, 2015b).
meet standard protocols for any communication applications involving network
users.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 175
CASE STUDY 3: NEW YORK projects under way.35 Examples include CenHub Marketplace,
a project in which a distribution utility and a technology
In the United States, New York State has been at the
company have teamed up to build an online portal for
vanguard of developing regulations to promote long-
energy products and services that provides customers with
term innovation. Under the REV initiative, the goal
personalized recommendations and offers enhanced data
is for electricity distribution companies to become
analytics; Clean Virtual Power Plant, a project in which a
“distribution system platform” (DSP) providers, meaning
utility is partnering with DER providers that bundle solar
the majority of their revenues will be generated from
with storage to aggregate electrical supply and thus serve as
providing market or platform services rather than from
a virtual power plant to provide grid services on clear days;
expenditures on capital equipment and from sales
Flexible Interconnect Capacity Solution, a project in which
volumes (NYDPS 2016). The intent is to change the
a utility is partnering with a technology company to offer
regulated utility business such that distribution utilities
a new, less costly, and faster way for customers and third
facilitate DER participation in electricity markets
parties to connect large DG projects to the grid by providing
and such that network utilities are agnostic between
an “infrastructure as a service” alternative to traditional
traditional network upgrades and DER-based “non-
interconnection; and others.
wires” solutions.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 177
5.3.3.2 Expanding cybersecurity regulation in the 5.3.3.3 Longer term approaches for building
United States and Europe cybersecurity resilience
To improve cybersecurity in the United States, state To address the long-term cybersecurity challenges faced by
regulatory commissions could work together to establish network utilities, we recommend a multifaceted approach.
standards applicable to distribution utilities and modeled
First, cybersecurity regulations or standards should
on the existing NERC Critical Infrastructure Protection
consider network and market operations that are
standards. This would remove the uncertainty surrounding
performed in the cloud in order to make data interfaces
what cybersecurity standards will be applied to DERs.
consistent, clear, and secure. Performing network
Using an approach similar to NERC, authorities could
and market operations in the cloud with consolidated
announce regulations that apply to distribution utilities
information would require enhanced security, robust
(and large distributed generators) along with a date
monitoring, and artificial-intelligence and machine-
that the regulations would come into effect. State public
learning technologies to monitor out-of-norm activities.
utility commissions would be responsible for new critical
Individual DER operators and aggregators would provide
infrastructure protection guidelines applicable to distribution
secure, authenticated data to a private or hybrid public/
utilities. These cybersecurity standards would apply to
private cloud data repository and network utilities with
DERs owned by both utilities and non-utilities. Harmonizing
appropriate cybersecurity protections, and market
regulatory standards at the transmission, distribution, and
operators would assess the veracity of the information.
end-user levels would contribute to economies of scale and
In this way, operations would be consolidated; instead
reduce the cost of cybersecurity preparedness.
of relying on separate systems with different protocols
In Europe, the European Commission encourages member and control software, this plan would ensure all data
states to make the most of NIS coordination mechanisms. and controls pass through major portals. If executed
Building on those, the Commission will propose how well, this type of configuration could lead to more
to enhance cross-border cooperation in the case of a secure and efficient operations and would enable DER
major cyber-incident. Given the speed with which the providers to interact with and utilize utility data. However,
cybersecurity landscape is evolving, the Commission the integration of DER communications with network
will also evaluate ENISA, which will possibly lead to operations could create risks of wider outages in the
the adoption of a new mandate. Common coordinated case of attack; therefore, it is recommended that DERs
standards are likely to be more effective at reducing risks be required to have adequate built-in security measures
for involved member states than piecemeal guidelines that before connecting to the grid and to cloud systems that
vary by state. control network and market operations.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 179
References Bailey, E.E. 1974. “Innovation and Regulation.” Journal of
Public Economics 3: 285–295.
ACCC. 2012a. “Benchmarking Opex and Capex in Energy
Networks.” Australian Competition and Consumer Bauknecht, D. 2011. “Incentive Regulation and Network
Commission and Australian Energy Regulator working Innovations.” EUI working paper RSCAS 2011/02.
ACCC. 2012b. “Regulatory Practices in Other Countries: Beesley, M.E. and S.C. Littlechild. 1989. “The Regulation of
Benchmarking Opex and Capex in Energy Networks.” Privatized Monopolies in the United Kingdom.” The RAND
Australian Competition and Consumer Commission and Journal of Economics 20(3): 454. doi:10.2307/2555582.
Australian Energy Regulator. May 2012. Bolton, R. and T. Foxon. 2011. “Governing Infrastructure
AEEG. 2009. Monitoraggio dello sviluppo degli impianti Networks for a Low Carbon Economy: Co-evolution
di generazione distribuita in Italia per l’anno 2006 ed of Technologies and Institutions in UK Electricity
analisi dei possibili effetti della generazione distribuita Distribution Networks.” Competition and Regulation in
sul sistema elettrico nazionale. Autorita per l’Energia Network Industries 12: 2–27.
Elettrica e il Gas. Regulatory decision ARG/elt 25/09 Bourreau, M. and P. Dogan. 2001. “Regulation and
(only in Italian). Innovation in the Telecommunications Industry.”
AEEG. 2010. Procedura e criteri di selezione degli Telecommunications Policy 25(3): 167–184.
investimenti ammessi al trattamento incentivante di cui CENELEC. 1999. “Voltage Characteristics of Electricity
al comma 11.4 lettera d) dell’Allegato A alla deliberazione Supplied by Public Distribution Systems.” European Standard
dell’Autorit a per l’energia elettrica e il gas 29 dicembre EN 50160.
2007, n. 348/07. Autorita per l’Energia Elettrica e il Gas.
Regulatory decision ARG (only in Italian). Clemenz, G. 1991. “Optimal Price-Cap Regulation.” The
Journal of Industrial Economics 39(4): 391–408.
AEEG. 2015a. Criteri per la definizione delle tariffe per
l’erogazione dei servizi di trasmissione, distribuzione e Conchado, A. and P. Linares. 2012. “The Economic Impact
misura dell’energia elettrica nel quinto periodo regolatorio. of Demand-Response Programs on Power Systems: A
Autorita per l’Energia Elettrica e il Gas. Regulatory Survey of the State of the Art.” In A. Sorokin et al., eds.,
consultation 544/2015/R/eel (only in Italian). Handbook of Networks in Power Systems. Berlin, Germany:
Springer. dx.doi.org/10.1007/978-3-642-23193-3_11.
AEEG. 2015b. Smart distribution system: promozione
selettiva degli investimenti nei sistemi innovativi di Cossent, R. 2013. “Economic Regulation of Distribution
distribuzione di energia elettrica. Autorita per l’Energia System Operators and Its Adaptation to the Penetration
Elettrica e il Gas. Regulatory consultation 255/2015/R/ of Distributed Energy Resources and Smart Grid
CHAPTER 5: The Future of the Regulated Network Utility Business Model 181
Jamasb, T., P. Nillesen, and M. Pollitt. 2004. “Strategic Kahn, A.E., T.J. Tardiff, and D.L. Weisman. 1999. “The
Behaviour under Regulatory Benchmarking.” Energy Telecommunications Act at Three Years: An Economic
Economics 26(5): 825–843. dx.doi.org/10.1016/j. Evaluation of Its Implementation by the Federal
eneco.2004.04.029. Communications Commission.” Information Economics
and Policy.
Jamasb, T. and M. Pollitt. 2003. “International
Benchmarking and Regulation: An Application to Kind, P. 2013. “Disruptive Challenges: Financial
European Electricity Distribution Utilities.” Energy Policy Implications and Strategic Responses to a Changing
31(15):1609–1622. doi:10.1016/ S0301-4215(02)00226-4. Retail Electric Business.” Washington, DC: Edison Electric
Institute. Retrieved from www.eei.org/ourissues/finance/
Jamasb, T. and M. Pollitt. 2008. “Liberalisation and
Documents/ disruptivechallenges.pdf.
R&D in Network Industries: The Case of the Electricity
Industry.” Research Policy 37: 995–1008. doi:10.1016/j. Laffont, J-J. and J. Tirole. 1993. A Theory of Incentives in
respol.2008.04.010. Procurement and Regulation. Cambridge, MA: MIT Press.
Jenkins, J.D. 2014. “Economic Regulation of Electricity Littlechild, S.C. 2006. “Beyond Regulation.” CWPE 0616.
Distribution Utilities Under High Penetration of EPRG 0516. www.eprg.group.cam.ac.uk/wp-content/
Distributed Energy Resources: Applying an Incentive uploads/2014/01/eprg0516.pdf.
Compatible Menu of Contracts, Reference Network
Lo Schiavo, L. 2016. “Incentive Regulation of Smart
Model and Uncertainty Mechanisms.” Master’s thesis.
Distribution Systems in Italy: from Input-based
Cambridge, MA: Massachusetts Institute of Technology.
Demonstration to Output-based Deployment.”
June 2014.
Presentation given to the Utility of the Future Study team
Jenkins, J.D. and I.J. Pérez-Arriaga. 2014. “The at MIT on May 6, 2016.
Remuneration Challenge: New Solutions for the
Lo Schiavo, L., M. Delfanti, E. Fumagalli, and V. Olivieri.
Regulation of Electricity Distribution Utilities Under
2013. “Changing the Regulation for Regulating the
High Penetrations of Distributed Energy Resources
Change: Innovation-driven Regulatory Developments
and Smart Grid Technologies.” MIT Center for Energy
for Smart Grids, Smart Metering, and e-Mobility
and Environmental Policy Research, working paper WP
in Italy.” Energy Policy 57: 506–517. doi:10.1016/j.
2014-005. Cambridge, MA: Massachusetts Institute of
enpol.2013.02.022.
Technology. September 2014.
Lockwood, M. 2016. “Creating Protective Space for
Jenkins, J.D., and I.J. Pérez-Arriaga. 2017. “Improved
Innovation in Electricity Distribution Networks in Great
Regulatory Approaches for the Remuneration of
Britain: The Politics of Institutional Change.” Environmental
Electricity Distribution Utilities with High Penetrations of
Innovation and Societal Transitions 18: 111–127. doi:10.1016/j.
Distributed Energy Resources.” The Energy Journal 38(3):
eist.2015.05.007.
89-117.
Lowry, M.N. and T. Woolf. 2016. “Performance-based
Joskow, P.L. 1989. “Regulatory Failure, Regulatory Reform,
Regulation in a High Distributed Energy Resource Future.”
and Structural Change in the Electrical Power Industry.”
Lawrence Berkeley National Laboratory, Future Electric
Brookings Papers on Economic Activity.
Regulation Series Report No 3. LBNL-1004130. Berkeley, CA.
Joskow, P.L. 2014. “Incentive Regulation in Theory and January 2016.
Practice: Electricity Distribution and Transmission
MA DPU. 2014. Order 12-76-B: Investigation by the
Networks.” In N.L. Rose, ed., Economic Regulation and Its
Department of Public Utilities on Its Own Motion
Reform: What Have We Learned. Chicago, IL: University of
into Modernization of the Electric Grid. Boston, MA:
Chicago Press.
Massachusetts Department of Public Utilities. June 12,
2014.
Müller, C. 2012. “Advancing Regulation with Respect to Ofgem. 2010a. “Handbook for Implementing the RIIO
Dynamic Efficient Network Investments: Insights from the Model.” London, United Kingdom: Office of Gas and
United Kingdom and Italy.” Competition and Regulation in Electricity Markets.
Network Industries 13: 256–272.
Ofgem. 2010b. “Low Carbon Networks Fund Governance
NAS. 2016. The Power of Change: Innovation for Document v.1.” London, United Kingdom: Office of Gas
Development and Deployment of Increasingly Clean Electric and Electricity Markets.
Power Technologies. Washington, DC: National Academy
Ofgem. 2010c. “RIIO: A New Way to Regulate Energy
of Sciences.
Networks.” London, United Kingdom: Office of Gas and
NERC. 2016. “Critical Infrastructure Protection Standards.” Electricity Markets.
Washington, DC: North American Electric Reliability
Ofgem. 2013a. “Strategy Decision for the RIIO-ED1
Corporation.
Electricity Distribution Price Control: Uncertainty
NSF. 2011. NSF Industrial Research and Development Mechanisms.” London, United Kingdom: Office of Gas
Information System (Historical Data 1953-2007). and Electricity Markets.
National Science Foundation. www.nsf.gov/statistics/iris/.
Ofgem. 2013b. “Strategy Decision for the RIIO-ED1
Accessed August 25, 2016.
Electricity distribution Price Control: Outputs, Incentives
NYDPS. 2015a. Order Adopting Regulatory Policy and Innovation.” London, United Kingdom: Office of Gas
Framework and Implementation Plan. Case 14-M-0101: and Electricity Markets.
Reforming the Energy Vision. Albany, NY: State of New
Ofgem. 2013c. “Strategy Decision for the RIIO-ED1
York Public Service Commission, Department of Public
Electricity Distribution Price Control: Financial Issues.”
Service. February 2015.
London, United Kingdom: Office of Gas and Electricity
NYDPS. 2015b. Staff proposal: Distributed System Markets.
Implementation Plan Guidance. Albany, NY: State of New
Ofgem. 2013d. “Strategy Decision for the RIIO-ED1
York Public Service Commission, Department of Public
Electricity Distribution Price Control: Business Plans and
Service.
Proportionate Treatment.” London, United Kingdom:
Office of Gas and Electricity Markets.
CHAPTER 5: The Future of the Regulated Network Utility Business Model 183
Ofgem. 2013e. “Strategy Decision for the RIIO-ED1 Schisler, K., T. Sick, and K. Brief. 2008. “The Role of
Electricity Distribution Price Control: Reliability and Demand Response in Ancillary Service Markets.” IEEE
Safety.” London, United Kingdom: Office of Gas and Transmission and Distribution Conference and Exposition,
Electricity Markets. 2008: 1–3. dx.doi.org/10.1109/tdc.2008.4517087.
Ofgem. 2014. “Final Determinations for the Slow-Track Schmalensee, R. 1989. “Good Regulatory Regimes.” The
Electricity Distribution Companies.” London, United RAND Journal of Economics 20(3): 417–436. dx.doi.
Kingdom: Office of Gas and Electricity Markets. org/10.2307/2555580.
Olmos, L., R. Cossent, T. Gómez, C. Mateo, J. de Joode, Strbac, G., M. Aunedi, D. Pudjianto, P. Djapic, F. Teng, A.
M. Scheepers, F. Nieuwenhout, J. Poot, M. Bongaerts, D. Sturt, D. Jackravut, R. Sansom, V. Yufit, and N. Brandon.
Trebolle, B. Doersam, S. Bofinger, U. Cali, and N. Gerhardt. 2012. Strategic Assessment of the Role and Value of Energy
2009. “Case Studies of System Costs of Distribution Storage Systems in the UK Low Carbon Energy Future.
Areas.” IMPROGRES No. WP 4, Deliverable 5. Petten, London, United Kingdom: Carbon Trust. Retrieved from
Netherlands: Improvement of the Social Optimal www.carbontrust.com/media/129310/energy-storage-
Outcome of Market Integration of DG/RES in European systems-role-value-strategic-assessment.pdf.
Electricity Markets Project.
Trebolle, D., T. Gómez, R. Cossent, and P. Frías (2010).
Pérez-Arriaga, I. J. and A. Bharatkumar. 2014. “A “Distribution Planning with Reliability Options for
Framework for Redesigning Distribution Network Use of Distributed Generation.” Electric Power Systems Research
System Charges Under High Penetration of Distributed 80(2): 222–229. dx.doi.org/10.1016/j.epsr.2009.09.004.
Energy Resources: New Principles for New Problems.”
Vergara, C., I. Perez-Arriaga, C. Mateo, and P. Frías.
CEEPR Working Paper No. 2014-006. Cambridge MA:
2014. “Estimating the Aggregate Impact of Distributed
MIT Center for Energy and Environmental Policy Research.
Photovoltaic Generation over Distribution Networks.”
Poudineh, R. and T. Jamasb. 2014. “Distributed IIT-Comillas Working Paper No. IIT-14-034A. Madrid,
Generation, Storage, Demand Response and Energy Spain: Institute for Research in Technology (IIT). Comillas
Efficiency as Alternatives to Grid Capacity Enhancement.” University.
Energy Policy 67: 222–231. dx.doi.org/10.1016/j.
Weisman, D. 2005. “Price Regulation and Quality.”
enpol.2013.11.073.
Information Economics and Policy 17: 165–174.
Pudjianto, D., M. Aunedi, P. Djapic, and G. Strbac.
Wirth, H. 2016. “Recent Facts about Photovoltaics in
2014. “Whole-Systems Assessment of the Value of
Germany.” Frieberg, Germany: Fraunhofer Institute for
Energy Storage in Low-Carbon Electricity Systems.”
Solar Energy Systems. April, 22 2016.
IEEE Transactions on Smart Grid 5(2): 1098–1109. dx.doi.
org/10.1109/TSG.2013.2282039. Woodman, B. and P. Baker. 2008. “Regulatory
Frameworks for Decentralised Energy.” Energy Policy 36:
Rivier, J., T. Gómez. 2000. “A Conceptual Framework for
4527–4531.
Power Quality Regulation.” Proceedings of the IEEE Ninth
International Conference on Harmonics and Quality of Power:
469–474.
06
Restructuring Revisited:
Electricity Industry Structure
in a More Distributed Future
6.1 Restructuring Revisited Until rather recently, the debate about restructuring has
been mainly framed within the boundaries of the bulk
After nearly a century of exclusive monopoly provision of power system (i.e., wholesale generators and high-voltage
electricity services, a wave of restructuring began in the transmission networks). Today, the growth and future
1980s and spread across much of the world in the 1990s potential of a variety of distributed energy resources—
and 2000s. With the apparent exhaustion of economies including solar photovoltaics (PV), other forms of distributed
of unit scale in power generation due to the emergence generation, and electrical and thermal energy storage
of smaller-scale electricity generation technologies, such devices—and more price-responsive and flexible electricity
as gas-fired plants, wind turbines, and combined heat demand has sparked a new wave of debate (CPUC 2014;
and power generation, policy makers began to reconsider CEER 2015a; Corneli and Kihm 2015; de Martini and Kristov
the natural monopoly characteristics of bulk power 2015; European Commission 2009, 2010; Pérez-Arriaga, et
generation, and in some cases, even retail supply (Joskow al. 2013; NYDPS 2014, 2015a, 2015b).
and Schmalensee 1983). This era saw considerable debate
over the proper structure of the electricity sector and the
appropriate roles and responsibilities, in particular, of
electricity network utilities, system operators, and market
operators (FERC 1996, 1999; European Commission 2003;
Millán 2006).
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 185
Market platforms, network providers, and examined next. The third section of this chapter analyzes
system operators perform three critical how the aggregation of DERs may create value, presently
and in a foreseeable future. The chapter concludes with a
functions that sit at the center of all
mostly descriptive presentation of existing and proposed
transactions in electricity markets. Properly
non-traditional business models for harnessing DERs,
assigning responsibilities for these core some of which may play a significant role in providing
functions is thus critical to an efficient, electricity services in the future.
well-functioning electricity sector and
for establishing a level playing field for
competitive provision of electricity services
6.2 Electricity Industry
by both traditional generators and network Organization in a More
providers and new businesses harnessing DERs. Distributed Future
6.2.1 Three crucial electricity
While there are many close parallels to restructuring industry functions
debates of the past, contemporary challenges require In electricity markets, the trading and provision of
extending discussions about electricity industry structure services occurs at three levels. First, producers and
“all the way to the bottom”—to encompass not only consumers of electricity (or their representatives) buy
distribution network owners and operators, but also end and sell energy from one another, often taking advantage
consumers, aggregators (including retailers), and new of market platforms, such as power exchanges or
competitive business models that harness distributed centralized markets run by system operators. Second, a
energy resources (DERs). The first round of electricity physical transmission and distribution network must be
restructuring focused on establishing competitive markets built and maintained to deliver energy from generators
for large-scale power generators and hinged on the role to consumers. Third and finally, system operators need
of transmission utilities and bulk power system operators to plan network development, procure certain technical
and their relation to competitive market segments. This services from market agents, and coordinate the dispatch
new round of debate revolves around the proper role of of these market agents and network assets to reliably and
electricity distribution utilities, the potential need for a efficiently operate electricity systems.
distribution system operator (DSO), the establishment of
market structures or platforms that enable the provision Market platforms, network providers, and system
of multiple electricity services by DERs of all kinds, and operators thus perform three critical functions that sit
the emerging role of DER aggregators and other new DER at the center of all transactions in electricity markets.
business models. This makes properly assigning responsibilities for these
core functions critical to an efficient, well-functioning
This chapter considers the question of electricity industry electricity sector and critical to establishing a level
structure from four different perspectives. First and most playing field for competitive provision of electricity
fundamentally, the chapter offers different views on services by traditional generators and network providers,
how to best allocate the core responsibilities of market and by new businesses that harness DERs. Considering
platform, system operation, network ownership, and the appropriate industry structure and assignment of
data management at the distribution and retail level responsibilities in a more distributed future thus requires
of the power system in order to enable an efficient familiarity with these three core functions, together with
and competitive sector. The new and more complex an understanding of which agents and actors are best
relationship between “bulk” system operators at the equipped to carry out each function.
transmission level and distribution system operators is
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 187
(common in the United States).3 In the second option, market platforms to serve them, can thus benefit from
system operators create new markets to re-dispatch more than 30 years of experience with restructuring in
and balance supply and demand when market platform bulk generation and transmission networks.
transactions would lead to infeasible power flows or the
real-time physical position of generators or loads differ 6.2.2.1 Market platforms alone are insufficient
from market schedules (common in European electricity First, as the restructuring of wholesale electricity markets
markets).4 In practice, the line between “security- unfolded, it quickly became apparent that establishing a
related” markets and “regular” markets is somewhat competitive market platform (or “power pool”) alone was
fluid and may shift over time.5 In addition, uncertainty insufficient to ensure competitive electricity generation
about load growth over time; long lead times to build and supply. In practice, both the system operator and
new generators; lack of trust in the stability of policy network provider functions can significantly affect the
or regulatory conditions; risk aversion on the part of ability of market agents to buy or sell electricity services.
generators, regulators, and consumers; and lack of full The entities responsible for system operation and
market participation by demand may give rise to a market network provision are in a position to exercise “vertical
for “firm capacity” contracts (e.g., capacity remuneration foreclosure”6 by planning, building, or operating the
mechanisms), another security-related market for which system in a manner that negatively impacts the ability
the system operator may be responsible. of certain upstream suppliers to access downstream
customers or vice versa. As the US Federal Energy
Finally, electricity network assets have traditionally
Regulatory Commission (FERC) observed:
been operated “statically,” with network components
sized to accommodate peak demands (a “fit and forget” “ When utilities control monopoly transmission facilities
strategy) and lines only occasionally switched in or out and also have power marketing interests, they have poor
of service. More “active” network management practices incentives to provide equal quality transmission service
are emerging, however, and could yield considerable to their power marketing competitors. …The inherent
improvements in cost and performance, including characteristics of monopolists make it inevitable that
dynamically switching lines (Hedman et al. 2010, 2011). they will act in their own self-interest to the detriment of
System operators may therefore also be responsible for others by refusing transmission and/or providing inferior
network topology control. transmission to competitors in the bulk power markets to
favor their own generation, and it is our duty to eradicate
6.2.2 Lessons from restructuring bulk unduly discriminatory practices.” (FERC 1999)
power systems
The European Commission (EC) has concurred, noting:
Although debates over industry structure at the bulk power
“ Without effective separation of networks from activities
system level have not fully ended (and perhaps never will),
of generation and supply (effective unbundling), there is
decades of experience in restructuring have highlighted a
an inherent risk of discrimination not only in the operation
number of necessary structural and regulatory practices
of the network but also in the incentives for vertically
that underpin healthy electricity markets and operations.
integrated undertakings to invest adequately in their
Today’s debates, over the proper role of distribution
networks.”7 (European Commission 2009)
system operators, owners, and users and over proposed
3 For discussion and definition of the roles of regional transmission organizations 6 Vertical foreclosure, a term from the industrial organization literature, encompasses
(RTOs) or independent transmission system operators (ISOs) in the United States, a variety of practices whereby a vertically-integrated firm exercises its position
see FERC (1999). in the market to provide discriminatory preference to affiliated upstream or
4 For discussion and definition of the roles of transmission system operators (TSOs) downstream firms, either by excluding or negatively impacting the ability of
in Europe, see Slot et al. (2015). downstream buyers to access upstream suppliers that compete with the firm or
5 The definition of security-related markets usually reflects some technical by excluding or negatively impacting the ability of upstream suppliers to access
constraints on the speed at which generators or demand can respond to security- downstream customers that do business with the firm (Tirole 1988).
related market prices or control signals and practical concerns about transaction 7 In this context, vertical integration refers to transmission or distribution network
costs, all of which can change over time. utilities that are vertically integrated with generation and/or retail supply functions.
practice, both the system operator and network provider integrated utilities to use their
transmission assets to favor their own
functions can have significant impacts on the ability of
generation, but instead attempts to
market agents to buy or sell electricity services.
reduce the ability of utilities to act on
those incentives.” (FERC 1999)
Network planning and expansion must be performed in
In light of this view, FERC (1999)
a transparent and impartial manner under appropriate
concluded that a “better structured market where
regulatory oversight to ensure that network expansion
operational control and responsibility for the transmission
and interconnection processes do not privilege certain
system is structurally separated from the merchant
agents over others. Less immediately obvious but perhaps
generation function of owners of transmission” would
equally critical, actions by system operators to resolve
best facilitate competition and eliminate incentives for
security-related concerns (e.g., re-dispatch of loads
vertical foreclosure in bulk power system activities. The
and generators) or procure security-related system
EC reached a similar decision in 2009, finding that the
services (e.g., reserves) can also advantage some market
“rules on legal and functional unbundling of [transmission
participants and penalize others.
system operators]” it established in 2003 “have not
led to effective unbundling.” This led the EC to call for
6.2.2.2 Independence of system operation and
“structural separation” that would “remove the incentive
network planning are critical
for vertically integrated undertakings to discriminate
Given these concerns, regulators in the United States, against competitors as regards access to the network,
Europe, and elsewhere have established requirements as regards access to commercially relevant information
for “functional and legal unbundling” of transmission and as regards investments in the network” (European
network and system operation functions from competitive Commission 2009, 2010).
market activities (FERC 1996; European Commission
2003). Functional and legal unbundling typically entailed 6.2.3 New challenges in
open access to transmission systems by all generators,
distribution systems
establishment of transparent and non-discriminatory
transmission tariffs (including purchases of ancillary While many lessons can be learned from prominent
services), creation of a functionally separate legal entity models for industry structure at the level of the bulk
responsible for transmission business activities, and other power system, these models cannot be simply “copied
mandates to prevent transmission network utilities from and pasted” into the distribution system context. A
preferentially advantaging certain generators or load- number of differences are important when considering
serving entities (such as subsidiaries of the network utility’s how to translate these lessons to the realities of
owner) or foreclosing market access by certain agents. distribution system structure and operation. This section
explores some of the salient differences between bulk and
However, after initial restructuring efforts generated distribution systems and key challenges for restructuring
several years of experience, both FERC and the EC in the distribution context.
ultimately concluded that open access tariffs and
functional and legal unbundling were generally insufficient
to establish effective incentives and a well-functioning,
competitive wholesale marketplace. As FERC has argued:
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 189
6.2.3.1 Distribution networks are 6.2.3.2 Markets for distribution system services
orders of magnitude more complex than are local and may be illiquid
transmission networks
Markets for distribution system services may also be far
While transmission network owners may manage a few less liquid than markets for bulk power system operation
hundred to tens of thousands of lines, substations, and and wholesale markets. Bulk power system operators
generators, many distribution networks incorporate maintain frequency across large geographic areas.
hundreds of thousands or millions of network components While network constraints can segment these markets,
and users. The order-of-magnitude greater complexity of reserves, balancing, capacity, and other ancillary services
distribution networks challenges both planning and active are generally fungible commodities that can be procured
system operation. For example, while transmission asset from a wide range of generators or loads across a system
investments tend to be large, discrete investments that operator’s territory. In contrast, distribution system
facilitate long-term planning, competitive procurement operators are concerned primarily with local voltage
processes, and transparent consideration of “non-wires” constraints and local network reliability and resiliency
alternatives, distribution investments are smaller, more issues. Only a small number of service providers with
numerous, and often must be completed in a shorter assets in very specific locations may be able to provide
time period. This greater complexity also increases these services, particularly in the operational time scale.
Markets for DSO services may therefore
prove illiquid in practice and may require
As experience with restructuring in the bulk power further regulation and oversight to ensure
system has demonstrated, ensuring the independence that participants do not exercise market
of market platforms, network providers, and system power. Alternatively, DSO services could
operators at the distribution level is important to a be procured via longer-term contractual
arrangements, which, by allowing new
competitive industry landscape.
entrants to bid, would increase competition
and mitigate the exercise of market power
during periods of system stress. Here
the regulatory burden associated with ensuring that again, lessons can be learned from bulk power systems,
vertically integrated utilities do not exercise vertical where the critical and highly location-specific nature
foreclosure and consistently provide non-discriminatory of black-start capabilities and “must-run reliability
access to distribution networks and markets. Light- resources” necessary for local voltage stability led to the
handed regulation of distribution utilities may therefore development of regulated, long-term contracts for these
require structural changes that mitigate incentives illiquid, location-specific services. In addition, market
for discriminatory behavior and/or changes to utility monitors may be required to identify competitiveness
remuneration that establish incentives for utilities to issues and recommend corrective actions on an ongoing
efficiently serve DERs and consistently exhaust cost- basis.
saving and performance-improving opportunities across
both capital and operational expenditures.8 Finally,
the complexity of distribution networks makes it more
difficult to decouple network planning from operation.
This increases economies of scope between network
ownership and system operation, as compared to
transmission systems.
8 See Chapter 5 and Jenkins and Pérez-Arriaga (2017) for more on economic
regulation of distribution network utilities.
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 191
Table 6.1: Salient Differences Between Bulk and Distribution Systems
BULK SYSTEM Hundreds to thousands of Focused primarily on Fungible Can supply services
LEVEL lines and substations. A few regulating frequency. Security- commodities to wholesale markets
large, discrete investments related services (balancing, markets with and TSO/ISO security-
needed annually, making reserves) can be procured many participants related markets
network expansion as fungible commodities and prominent (ancillary services).
decisions easily supervised across wide geographic areas use of locational
and contestable. Meshed (exceptions include black marginal prices
network structure. start and voltage regulation). (at nodal or zonal
Operations can be relatively level).
easily decoupled from
maintenance decisions.
OPTIMAL
DISTRIBUTION DISTRIBUTION SYSTEM RETAIL DERs AND
REGULATION
NETWORK OPERATOR (DSO) TARIFFS LOADS
REVENUE ($)
DISTRIBUTION Thousands to millions Focused primarily on End consumers Can supply services
SYSTEM LEVEL of lines, substations, regulating voltage and (and most to wholesale markets
transformers, and other restoring outages. Additional DERs) pay or are and TSO/ISO security-
assets. Many smaller complexity due to need to compensated related markets, as well
investments needed on a maintain phase balance, based on as to DSO security-
continual basis, making topology changes, volatility regulated or related markets
network expansion in voltage, power quality competitive and directly to end
decisions more difficult concerns, and in some cases retail tariffs that consumers.
to supervise or contest. less complete electronic currently lack
Meshed and radial network information and visibility of cost-reflective
structure. network assets. Location- locational
specific and potentially illiquid and temporal
markets for security-related granularity.
services (voltage control,
congestion management,
reliability). May exhibit
significant economies of scope
with network maintenance.
roles and responsibilities of distribution are sufficient (CAISO 2015), the DNO/SO would have
the capability to compute and communicate locational
system operators
marginal prices within the distribution system (in close
Just as in transmission and bulk generation, market coordination with the local ISO/TSO) that capture the
platforms, system operation, and network provision are marginal cost of local voltage or thermal constraints,
all critical functions that facilitate competition and distribution losses, and transformer wear and tear, and
efficient operation of distribution networks and DERs the capability to communicate the location- and time-
in lower voltages. Here we present three potential specific value of DER-provided services (Caramanis
models for assigning these core functions to different et al. 2016). In addition, cost-reflective network tariffs
actors, focusing in particular on the different roles and can convey the long-term cost of network expansion
responsibilities of DSOs. in different locations and ensure adequate recovery of
network costs. Prices and charges in distribution systems
6.2.4.1 Distribution network owner and operator
and the benefits and trade-offs associated with greater
The first option parallels the TSOs that have emerged degrees of spatial granularity in energy and network
to manage the bulk power system in Europe and other charges are discussed further in Chapter 4.
jurisdictions. It combines the functions of distribution
To facilitate equal competition between traditional
network provider, distribution system operator, and
network investments and contracts or payments for
market platform for distribution services within a single
DER services, the DNO/SO should be regulated in
utility. This distribution network owner/system operator
a manner that (1) rewards utilities for reducing total
(DNO/SO) would be responsible for creating and
expenditures and improving network quality of service
maintaining the physical distribution network, operating
and (2) equalizes incentives between operational and
the distribution system (including actively managing
capital expenditures, as discussed in detail in Chapter 5.
distribution network assets and coordinating the dispatch
Alternative or complementary measures include
of DERs to provide distribution network services), and
conducting transparent “distribution resource plans”
establishing and operating any markets or procurement
(similar to the integrated resource plans undertaken by
processes for DSO services (e.g., voltage control,
traditionally regulated, vertically integrated utilities),
congestion relief, enhanced resiliency).11
mandating full consideration of “non-wires” alternatives
As distribution system services are highly location- to all network investments through regulation (CPUC
specific and may be illiquid, the DNO/SO could lead 2014; NYDPS 2015a, 2015b), and undertaking fully
competitive and transparent procurement processes to transparent procurement processes or organized
establish longer-term contracts for system services.12 auctions for DER services. While these measures
This would reveal the price of alternatives to traditional improve incentives for DSOs to actively engage DERs
network investments and allow the utility to effectively and loads to reduce network expenditures, they do not
unlock the potential of DERs to reduce total system align the fundamental business incentives of the utility
costs or improve system performance. Eventually, if (as performance-based regulation does, for instance).
In addition, effective oversight of so many individual
distribution network assets and investments can involve
11 The markets established by the DNO/SO would be specifically for services used
within the distribution system, as distinct from existing markets for ancillary significant regulatory costs.
services used in bulk power systems, which are already well established by bulk
system operators. Note that it is possible to separate responsibility for market
platform operation from the system operation, planning, and maintenance
As with TSOs, this version of the DSO would have to
functions of the DNO/SO (we discuss a variation of this model below). Indeed, be sufficiently independent of competitive activities,
dividing the market platform role from the system operator and network provider
roles closely parallels the bulk power system structure common in Europe, where both upstream and downstream (e.g., centralized and
market platforms are managed by power exchanges and system operation and
network provision is managed by TSOs.
decentralized generation upstream, and retailing/
12 See Section 6.3.3 for more on how this procurement process could be structured. aggregating and DER ownership downstream), to ensure
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 193
impartial planning and operation of distribution systems context would entail financial independence from any
and impartial procurement of services from DERs or companies active in wholesale energy or ancillary
their aggregators. Indeed, since the 3rd Energy Package service markets and from retailing or DER ownership or
(European Commission 2009), European regulation aggregation within the service territory of the DNO/SO.14
has recognized the importance of independence in As an alternative to structural unbundling, independence
distribution system operation and established a minimum can be approximated by the establishment of legal
level of separation between distribution utilities and unbundling and effective restrictions on exchange of
retailing and generation activities. As the Council of information and coordination between the DSO and
European Energy Regulators explains: any other subsidiary or sister companies engaged in
competitive activities within the DSO’s service territory.
“To create a level playing field in retail energy markets, all
However, legal unbundling entails a greater regulatory
competitive actors need to compete on the same terms.
burden to be effective15 and, as experience in the
DSOs should facilitate this by acting as neutral market
restructuring of bulk power systems has demonstrated,
facilitators. This requires a sufficient level of unbundling
legal and functional unbundling measures are second-
between suppliers and associated DSOs. As energy
best alternatives to structural independence. This is
networks are regulated monopolies, DSOs have exclusive
likely to be especially true in more complex distribution
access to all customers within their geographic network
networks, as discussed above.
area. Without sufficient unbundling, this has the potential
of disturbing competition in the market.” (CEER 2016) This DNO/SO model offers advantages in terms of
maintaining potentially significant economies of scope
The New York Public Service Commission (PSC)
between network ownership, planning, and operation,
has voiced similar concern that “unrestricted utility
since both network provider and system operation
participation in DER markets presents a risk of
functions are the domain of the same utility. In addition,
undermining markets more than a potential for
if properly regulated and separated from competitive
accelerating market growth” (NYDPS 2015a). The
activities, as discussed above, the DNO/SO becomes an
PSC’s 2015 order establishing utilities as “distribution
impartial facilitator of much greater competition between
system platform” companies (similar to the DNO/SO
conventional and distributed energy resources and a new
model outlined here) thus restricts these companies
customer for DER services wherever they provide cost-
from owning distributed generation or storage assets
effective alternatives to traditional network investments.
except in limited circumstances.13 “Markets will thrive
The primary challenge to implementing this approach is
best where there is both the perception and the reality
the level of industry restructuring and regulatory reform
of a level playing field,” the PSC has argued, “and that is
that may be necessary in some jurisdictions.
best accomplished by restricting the ability of utilities to
participate” in DER markets (NYDPS 2015a).
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 195
wires companies operating within a given IDSO territory where they would compete with the utility’s business
may have advantages over other competitors, such as units. A vertically integrated DSO would therefore have to
DER providers or aggregators that could offer alternatives be subject to close regulation to minimize opportunities
to network investments. This information asymmetry may for vertical foreclosure, including requirements to procure
act as a barrier to the efficient integration of non-wires network services through transparent and open auctions
alternatives to traditional network investments. that allow DERs to compete on a level playing field. This
option is similar to the way that vertically integrated,
6.2.4.3 Closely regulated, vertically traditionally regulated utilities may be required by
integrated utility regulators to fairly and transparently consider signing
The final option we consider for restructuring at the contracts with independent power providers in lieu of
distribution level is to incorporate all of the critical building and owning their own generation. In addition,
functions, including ownership of the distribution the vertically integrated DSO would be required to
network, distribution system operation, and any markets offer transparent, cost-reflective, open-access tariffs
for DSO services, into a vertically integrated, closely for connecting DERs to its networks and to adjudicate
regulated utility.16 This utility could also be responsible for interconnection requests in a timely manner, paralleling
retailing, generation, and/or transmission ownership and the open-access tariffs and interconnection processes
operation. The challenge in this model is to appropriately required of transmission system owners. Finally, the
regulate the utility such that its incentives are aligned regulator would also need to closely oversee the utility
with the efficient integration of DERs and such that any to ensure that system planning accounts for DERs as
potential for the utility to employ its central platform role another integral customer of the utility’s services and
to foreclose opportunities for DERs to access markets as a provider of alternatives to utility-owned generation
(i.e., vertical foreclosure) is minimized. and network investments. This could be facilitated via
long-term integrated system planning processes, similar
The most effective way to align the utility’s incentives
to the integrated resource planning processes required of
would be to manage the utility under a regulation that
regulated utilities at the bulk power system level (CPUC
equalizes incentives for savings between operational
2014; Hawaii PUC 2014; Wilson and Biewald 2013).
and capital expenditures (similar to the distribution
owner/operator model discussed above), as discussed The greater complexity of distribution networks
in Chapter 5. If implemented effectively, the utility’s significantly increases the regulatory burden of effectively
distribution business unit could be made financially overseeing each of these critical functions at the
impartial between wires and non-wires options in network distribution level. The challenges associated with ensuring
planning and operation and could be incentivized to non-discriminatory behavior and facilitating open
pursue the most cost-effective strategies for building, competition in bulk power systems ultimately led most
maintaining, and operating the distribution system. regulators to pursue reforms that ensured the structural
independence of transmission system operators. Similarly,
However, given the utility’s vertical integration into
the challenges and costs associated with sufficiently
generation and/or retailing activities, it will be impossible
regulating a vertically integrated DSO should not
to render the utility fully impartial to DER adoption and
be underestimated.
integration, as DERs have the potential to compete
directly with the utility’s retailing, generation, or
transmission businesses. This raises the potential for
the utility to exercise vertical foreclosure by managing
the distribution system in ways that disadvantage DERs
16 Note that this model is not compatible with current law in the European Union
(European Commission 2009, 2010).
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 197
Table 6.2: Benefits and Challenges of Different Industry Structures
CLOSELY-REGULATED,
DISTRIBUTION NETWORK
INDEPENDENT DISTRIBUTION VERTICALLY-
OWNER/SYSTEM OPERATOR
SYSTEM OPERATOR (IDSO) INTEGRATED
(DNO/SO)
DISTRIBUTION UTILITY
BENEFITS • Economies of scope from • Independent system operator acts • Captures economies of
combining distribution market as neutral facilitator of markets and scope between all three
platform, system operation, and distribution system operation. distribution system
network provider functions. • Does not require owners of functions as well as bulk
• Structural unbundling minimizes distribution network assets to system functions.
incentives for discriminatory be unbundled from competitive
behavior and ensures DNO/ affiliates.
SO acts as neutral platform for
competitive market activities.
• As a second-best alternative,
functional independence can be
approximated via legal unbundling
and sufficient “Chinese walls”
between the network company and
competitive affiliates.
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 199
second best option. If the DSO is independent of other networks and bulk system operators was foundational
competitive agents, the DSO can act as a neutral manager for competitive wholesale markets, independence of
of the data hub. To the degree DSOs are integrated with distribution system operation and planning is essential to
competitive market segments, the importance of the providing a truly neutral trading platform for competition
data hub responsibility argues for further enhancement between centralized and decentralized resources, as well
of functional independence or the establishment of an as between traditional network investments and solutions
independent data hub manager (Smart DCC 2015). As to distribution operation challenges that harness DERs
the New York PSC contends, “asymmetry regarding (e.g., “non-wires” solutions).
system information if continued will result in a barrier to
new market entry by third parties and ultimately impede
innovation and customer choice” (NYDPS 2015a). As experience with restructuring in the bulk
Decisions about the governance of the data hub are thus system has demonstrated, the best solution,
strongly related to the other structural choices discussed
from a competitive market perspective, is
above. Furthermore, the possibility cannot be ruled out
structural reform that establishes financial
that future ICT developments might render a centralized
data hub unecessary, while making compatible the independence between the distribution
selective preservation of privacy, non-discriminatory system operator (DSO) and any affiliates
access to data of commercial value, and DSO access to in competitive markets, including adjacent
data needed to perform security functions. wholesale generation and ancillary services
markets and competitive retail supply and DER
6.2.6 Restructuring “all the way to
markets within the DSO’s service territory.
the bottom”
To sum up, distribution networks and system operations
are now at the heart of modern electricity markets. Just as Experience in bulk power systems and well-
restructuring transmission network utilities was essential established regulatory principles lead to a set of clear
to create a level playing field for competition at the bulk recommendations for restructuring at the distribution
power system level, regulators and policy makers today level, although regulators must also contend with the
must carefully reconsider the roles and responsibilities implications of existing industry structure and the
of distribution utilities. The time has come to address costs of any proposed transition in each context. For
incentives and create structures for the efficient provision example, the best solution from a competitive market
of electricity services by a diverse range of conventional perspective is structural reform that establishes financial
and distributed energy resources and network assets. independence between the DSO and any affiliates in
In particular, assignment of responsibility for three competitive markets, including adjacent wholesale
core functions—market platform, system operator, and generation and ancillary services markets and competitive
network provider—must be carefully considered. In retail supply and DER markets within the DSO’s service
addition, a fourth function, the data hub or data exchange, territory. Between the two primary alternatives to
may become increasingly relevant to unlock competitive achieve structural independence—namely, a combined
markets at the distribution system and retail level. distribution network owner/system operator (DNO/
SO) with financial independence from competitive
One of the most challenging and contentious aspects
affiliates and an independent distribution system operator
of this new era of restructuring will be taking the steps
(IDSO)—only the former has so far proven its practical
needed to establish sufficient independence between
viability at the distribution level. Furthermore, the DNO/
competitive market activities and critical distribution
SO construct captures the significant economies of
market platform, system operation, and network planning
scope between system operation and physical network
processes. Just as the independence of transmission
every regulatory jurisdiction and power sector context. Different solutions for addressing coordination challenges
As a second-best alternative, various forms of legal between the bulk power and distribution system
and functional independence can be established. These operators are currently being studied by researchers21 and
structures will need to be complemented by transparent considered by policy makers and various stakeholders.22
mechanisms (e.g., auctions or markets) for selecting This section does not aim to resolve these ongoing
services where DERs and centralized network services discussions—instead, it highlights the main coordination
might compete to ensure that no conflicts of interest are challenges for transmission and distribution system
exercised. It must also be noted that efforts to establish operations, regardless of how responsibilities for the three
legal and functional unbundling ultimately proved (or four) core functions are assigned.
insufficient at the bulk system level in many jurisdictions
(FERC 1999, European Commission 2009). Measures that
fall short of making DSOs financially independent may If DERs and flexible demand actively
prove more problematic as competitive markets for DERs
participate in energy- and security-related
and for DER-enabled services develop further. Facilitating
markets and if bidirectional power flows
a level playing field between DERs and conventional
approaches to generation and network services is likely become common, the conventional distinction
to remain challenging if distribution network utilities are between the “bulk power system” and the
vertically integrated into competitive market segments—in “distribution system” will become increasingly
that case, significant regulatory oversight will be required. blurry. This makes coordination between both
system operators increasingly necessary for
the efficient and reliable functioning of the
power system.
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 201
6.3.1 Reconsidering boundaries facilitates a single topological interface between system
between transmission and distribution operation problems at the transition from meshed to
radial networks. In contrast, separating meshed networks
system operation
into independent operational problems is technically
First, it must be noted that boundaries between challenging—as current efforts to coordinate operational
transmission and distribution systems and the respective decisions and power flows across jurisdictional
domains of bulk power system operators (TSOs or boundaries between interconnected ISOs or TSOs
ISOs) and distribution system operators (DSOs) are demonstrate. Finally, of the two candidates for operating
somewhat arbitrary. The voltage ratings of transmission meshed networks and associated markets, bulk power
and high-voltage “sub-transmission” distribution lines system operators (ISOs or TSOs) have much greater
often overlap. In addition, the “high-voltage” portions of experience in this area than DSOs.
distribution networks are typically meshed networks, just
like transmission systems. “Medium-voltage” distribution Technical considerations, of course, must be taken
networks (also called “primary feeders”)23 are typically up alongside jurisdictional, regulatory, and economic
also constructed in a meshed fashion (particularly in considerations as well. Depending on the jurisdiction,
urban and semi-urban networks), although, to simplify changing responsibility for system operation (e.g., to
power flows and reduce the severity of short-circuits, encompass meshed “distribution” networks in bulk power
they are regularly operated with some circuits “open,” system operations) may require changes in law, regulation,
which renders them radial in practice.24 The physical or asset ownership. In other jurisdictions, such changes
configuration of these networks is important, because may result from a natural evolution of steadily expanding
electrons care little for jurisdictional boundaries and and integrating energy markets and power systems. This
respect only Kirchhoff’s Laws. Power flows across the study therefore takes no final position on the ideal structure
meshed portion of networks that are considered part for any particular jurisdiction, but rather urges a thoughtful
of the “distribution system” thus have complex and reconsideration of traditional arrangements as part of a
interactive impacts on power flows, congestion, and broader acknowledgement of the end of the conventional
system operation decisions in “transmission networks” “top-down” paradigm (see Chapter 3).
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 203
runs an optimization for the whole system (PJM 2015). At temporally linked markets, would play an important
the same time, a subset of DERs managed by curtailment role in coordinating decisions where conflicts arise
service providers aggregate different resources to provide between various services. This would help ensure that
economic and emergency demand response to PJM for DERs commit their capabilities to the most valuable
bulk power system operations (PJM 2014). Although combination of services while respecting their various
distribution utilities are informed when DERs register technical constraints.
with PJM, distribution utilities are not currently informed
when these resources are actually scheduled to provide 6.3.2.4 Real-time dispatch and activation
system services. Moreover, the aggregator can at any of services
time change the location of the specific DERs that are In real time, different services need to be activated, final
providing the required response, so long as deliverability positions need to be determined, and active resources
to the system operator is assured. Because this may need to be dispatched. DER owners or third-party
cause constraints in distribution networks in the future, aggregators would take on responsibility for activation
such disconnects in communication should be avoided based on dispatch directives, price signals, and penalties
wherever possible. for non-fulfillment of previous commitments.
6.3.2.3 Energy prices at the interface between 6.3.2.5 Emergency conditions and
transmission and distribution system operators restoration arrangements
Conceptually, it is possible to compute and communicate Information on changes in energy schedules and available
prices at the TSO and DSO levels that are consistent with resources may need to be sent to both operators during
one another without resorting to a massive, centralized emergency situations, when system security is in danger
optimization process. For example, once the TSO and fast action is needed. In such cases, the TSO may
computes energy prices for the meshed network, these need support from the DSO to reduce or curtail loads
prices could be sent to the DSO to be incorporated in or generation connected to the distribution network.
the computation of prices for the distribution system In addition, the DSO may have local issues (e.g., line
or when clearing local markets in radial networks (see faults) that may be relevant to communicate to the TSO
Chapter 4 for further details). This decomposition is best and may require support from resources connected at
facilitated by simultaneously computing prices for the the TSO level. Advanced network codes may need to
entire meshed portion of the network, before propagating incorporate new actions and procedures in emergency
down to radial network branches using a centralized or situations and establish communication protocols
distributed solution method. Iterative solution methods between operators, such as Europe’s Network Code on
may be required to reach convergence between DSO Emergency and Restoration (CEER 2015a; ACER 2015).
and TSO prices.26 If locational marginal prices (LMPs) Restoration arrangements between TSOs and DSOs
are not computed in the distribution system, a second- will enhance the resilience of the system and provide
best solution may be implemented, such as wholesale more cost-effective solutions compared to the current
prices with an estimated distribution loss factor, local situation where DSOs do not intervene (CEER 2016).
market mechanisms, and/or contracts. In any case, a Data exchange between TSOs and DSOs will require
comprehensive system of prices for energy, ancillary the standardization of formats and protocols for each
services, and network services required by both TSOs timeframe so as to reduce the exchange time and
and DSOs, together with a well-coordinated series of increase data usage (CEDEC et al. 2016).
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 205
to network expansion. DERs commit to a firm put option pay a fixed cost ($/kW-yr) for the capacity options at the
which network utilities can exercise at periods of network clearing price and would in turn have their consumption
congestion, up to the contracted firm capacity quantity. hedged against congestion and peak-coincident network
Appropriate penalties for non-performance would be capacity charges for the length of the option contract.
assessed. Note that DERs co-located with loads capable Users who did not clear would not pay any fixed cost for
of reducing net withdrawals during periods of network network capacity and would be fully exposed to short-run
congestion can likewise reveal and monetize their value prices for all consumption. DER owners who cleared in
by allowing network users to reduce their requirements the supply side of the auction would sign put options with
for network capacity and lower their options bids performance penalties and receive a fixed payment ($/
accordingly, reducing the costs these users will pay for kW-yr) at the clearing price. Finally, the network utility
options contracts.29 would invest in any remaining network capacity expansion
needed to accommodate cleared demand for network
Network users could submit their own bids, or
capacity after accounting for the aggregate ability of
alternatively, sign up for aggregators who submitted
cleared DERs to reduce network peaks. Box 6.1 provides a
bids on their behalf. Aggregators better able to reflect
real-world example of a DER auction that was recently run
or predict user preferences would be able to offer lower
by Con Edison to defer a costly network upgrade in a high-
costs to their customers, as they would face lower fixed
growth area of its service territory. While this auction was
costs of options and lower exposure to LMPs. Finally,
not an options auction such as the one described above, it
any users who did not submit bids, either directly or via
nonetheless shares many of the same characteristics.
aggregators, would have a conservative estimate of their
demand for firm network capacity added to the left-hand
side of the bid stack at a very high willingness to pay. In
other words, a very conservative estimate would be used
for any users that did not submit bids, with these users
then obligated to pay fixed costs reflecting this estimate.
This is in effect what utilities do to today absent better
information about network users’ willingness to pay for
network capacity. Users with lower willingness to pay or
more accurate information about their preferences have
a strong incentive to reveal that preference by submitting
more accurate bids.
29 Note that this means behind-the-meter DERs that are not providing net injection at
times of system congestion but rather reduce a user’s net withdrawals would bid on
the “demand side” of the auction, rather than the supply side, while those that are
able to inject at times of system peak bid on the supply side.
Clearing price
Cleared firm DER supply put options Cleared firm network capacity additions Clearing quantity
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 207
Box 6.1: Brooklyn Queens Demand Management Program Demand
Response Auction
Facing growing peak load in parts of Brooklyn and Queens, New York, Consolidated Edison, Inc. (Con
Edison) has launched the Brooklyn Queens Demand Management (BQDM) Program. The BQDM
Program will procure 52 MW of peak load reduction by summer 2018, with 41 MW of the total load
reduction provided by customer-sited solutions such as demand response (DR), energy efficiency,
energy storage, fuel cells, solar PV, and combined head and power (NYDPS 2014; Con Edison 2016a).
The remaining 11 MW are expected to come from utility-sited solutions such as voltage optimization.
Con Edison will use these resources to provide load relief during critical hours on peak summer days,
and the deployment of these resources are meant to defer the construction of a $1.2 billion substation
upgrade by five years or more.
As part of the BQDM Program, Con Edison ran auctions to competitively procure distributed resources
that would deliver load relief between the hours of 4:00 p.m. and 12:00 a.m. (in two separate four-hour
blocks) during the summers of 2017 and 2018. The auctions for 2017 and 2018 delivery were held
on July 27 and July 28, 2016 and resulted in Con Edison accepting offers for 22 MW of peak
hour demand management services from 10 providers including Stem, EnerNOC Inc., Innoventive
Power, Direct Energy, Power Efficiency, Demand Energy Networks, Energy Spectrum, and Tarsier
(Bloomberg 2016). The BQDM auction clearing prices ranged from $215/kW-year to $988/kW-year
(Bloomberg 2016).
Figure 6.3: Illustration of Brooklyn Queens Demand Management Program Auction Mechanics
DR Resource Blocks
are bid in at
decreasing prices
Price ($/kW/year)
DR Resource
Bid Blocks
Winning DR
Resource Blocks
Con Edison intends to call a BQDM demand reduction event when it expects that the following day will
be a “peak day” in the BQDM Program area. Con Edison defines “peak day” differently for resources
providing load relief between 4:00 p.m. and 8:00 p.m., and those providing relief between 8:00 p.m.
and 12:00 a.m. For the former (4:00-8:00 p.m.), DR resources are called upon when the anticipated
peak electrical load in the BQDM area the following day is no less than 97 percent of the summer peak
forecast. For the latter (8:00 p.m. to 12:00 a.m.), DR resources are called upon when the anticipated
peak electrical load in the BQDM area the following day is no less than 93 percent of the summer peak
forecast. Con Edison expects that there will be an average of three to six calls each year for each product.
30 Each descending clock auction began with a 15-minute timer and any bids placed in the final two minutes of the timer added an additional two minutes to the timer. The auctions
ended when the timer ran out or when two hours had passed (whichever occurred first).
6.4 The Value and Role of the evolving power system. In Europe’s liberalized retail
markets, debate is centered around the functioning of
Aggregation retail markets, the ability of retailers32 to deliver desired
levels of consumer engagement and value-added
6.4.1 The emergence of aggregators services, and the value or cost of superimposing third-
As DERs proliferate and opportunities for active or flexible party aggregators over these retailers (Eurelectric 2015;
demand grow, “aggregators” of these resources have the European Commission 2015; Smart Grid Task Force
potential to help unlock the value of distributed resources 2015). On the other hand, independent DER aggregators
and bring them into energy markets at scale. Aggregation are already highly active in US markets and stakeholders
is defined here as the act of grouping distinct agents in a are attempting to design market rules to ensure these
power system (i.e., consumers, producers, prosumers, or aggregators flourish when they provide true value
any mix thereof) to act as a single entity when engaging creation as opposed to regulatory arbitrage (CAISO 2015;
in power system markets (whether wholesale or retail) or NYDPS 2014).
selling services to the system operator(s).31
Clarifying these debates requires an understanding of
The emergence of a new breed of DER aggregators has the mechanisms by which aggregation creates value. In
sparked debate over the value and role of aggregators in many cases, aggregators are performing roles today that
may not deliver value to power systems but rather reflect
31 This definition builds upon and narrows the definition provided by Ikäheimo, Evens,
and Kärkkäinen (2010): “an aggregator is a company who acts as an intermediary 32 Retailers or retail electricity providers (REPs) are aggregators in the sense that they
between electricity end-users and DER owners and the power system participants who aggregate a number of dispersed consumers (and, at some times, producers) and
wish to serve these end-users or exploit the services provided by these DERs.” Ikäheimo act as a liaison between these agents and wholesale markets. REPs also comply
et al.’s definition encompasses both aggregation in the context considered in with power system regulations, perform hedging functions, and undertake other
this chapter, as well as the potential role of platform markets. We recognize the activities on consumers’ behalf. Some REPs, such as MP2 Energy in the United
existence of other definitions of aggregators; in practice, the definition of an States, are performing roles traditionally attributed to third-party aggregators,
aggregator can be restricted or expanded depending on regulations that define the such as brokering demand response for capacity and ancillary services market
roles and activities aggregators can perform. participation (MP2 Energy 2016).
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 209
opportunities to arbitrage inadequate regulation. In other 6.4.2 The present and future value
cases, aggregation delivers real value, but this value may of aggregation
become less significant in the future, as technological
To clarify how and for whom aggregation creates value,
change reduces the costs of information provision,
we discuss two types of economic value: private value and
coordination, or transactions. Other activities may deliver
system value. Aggregation has system value if it increases
enduring value. This section discusses the potential role
the social welfare of the power system as a whole. Private
of aggregation in power systems and the ways in which
value is an increase in the economic welfare of a single
aggregation can deliver value to both private stakeholders
agent or subset of agents. Private value creation may
and the power system as a whole. This discussion has
or may not align with system value creation. As we will
implications for pertinent questions, such as: Should
demonstrate, aggregations with private value may create
the power system accommodate many aggregators or
economic value for certain agents at the expense of
only one centralized aggregator? Who can or should be
system-wide economic efficiency. Aggregation may also
an aggregator (transmission and distribution system
simply lead to a rent transfer between market actors. In
operators, retailers, third parties, etc.)? What market
other cases, aggregation helps align private value with
design elements may need to be adapted or adopted to
opportunities to improve the efficiency of power systems
accommodate DERs? What is the best feasible level of
and deliver system value.
unbundling between aggregators and other traditional
utilities or retailers? We can distinguish three broad categories of aggregation
(Figure 6.4). First, aggregations with “fundamental” value
do not depend on the specific regulations, level of
Fundamental Aggregation
SYSTEM VALUE
• Economies of scale
• Economies of scope
• Risk management
• Competition and innovation
Transitory Aggregation
• Engaging power system agents
• Management of complexity
• Deployment of automation technologies
• Closing information gaps
Opportunistic Aggregation • Coordinating agents for system operations
• Ill-designed end-user tariffs
• Inadequate regulations for procurement
PRIVATE VALUE
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 211
contracts for differences, for example). Aggregators may that incentivizes efficient behavior. However, current
therefore deliver value by managing uncertainty—acting as power system technology is not capable of calculating
intermediaries between small consumers/producers and and communicating the multiplicity of price signals that
volatile markets to provide hedging solutions to market are relevant for different agents at theoretically optimal
players. For example, in retail markets, retail electricity levels of temporal and spatial granularity. An aggregator
providers (i.e., demand aggregators) offer stabilized may be able to intervene to close gaps in information
prices to their consumers. The value of this type of risk- on system operation between the system operator and
hedging service is discussed in Littlechild (2000). various agents. Furthermore, one aggregator can gain
from economies of scale by processing information
6.4.2.2 Transitory value of aggregation from multiple agents, whereas costs would otherwise
Aggregators may create value as the power system be multiplied by the number of agents processing this
transitions from current regulations and technologies to information independently.
a more advanced or idealized future. Temporary value
It may be difficult to motivate agents to take action if
is not inherent to aggregation, but may be unlocked by
the sums of money that can be gained are small and if
aggregators. Opportunities for agents in the distribution
the required action is complex. However, when taken
system to increase system efficiency by engaging with
in sum, small actions by numerous agents can create
the bulk power system are increasing as ICTs enable
significant value for the system. For example, Opower
loads to become more price-responsive and as DERs are
achieves peak energy reductions by communicating when
increasingly deployed. However, market complexities,
and how energy should be saved. Opower estimates
that the average customer in Nevada
may save $33 per year (the estimated
Aggregators deliver fundamental system value where national average is $28 per customer
they harness economies of scale or scope or manage per year). This equates to just over $2 in
risk for participating agents. savings per month. Without intervention,
a customer may not be motivated to
take action that would produce such
a small sum of savings on his or her
information gaps, lack of engagement, and other biases
own. However, Opower estimates that its service has
may prevent the value in these distributed assets from
the potential to avoid over 140 megawatts of generating
being unlocked. Aggregators may create system value by
capacity in Nevada and over 4.5 gigawatts nationally
managing or eliminating these factors.
(Opower 2016).
An agent may be capable of providing a service (or set
The problem of small financial incentives is exacerbated
of services) to the system but may lack the information
by the problem of system complexity. Navigating the
required to do so effectively. For example, consumers
power system’s various rules and codes may be very
often lack information in a number of areas: when
difficult for an agent that has not historically engaged in
system peaks occur, what the prices are for various
power system operations. An aggregator may be able to
services they consume, what technologies are available
navigate these complexities on an agent’s behalf. Many
to help them control consumption, what the prices of
companies, such as EnerNOC and Comverge, handle
these technologies are, etc. Additionally, consumers
complex registration and bidding processes on behalf of
may lack the ability to forecast this information into the
the agents they serve, enabling the system to benefit from
future, which is critical for hedging risk or scheduling
the services these agents provide and enabling the agents
bids and consumption. Finally, power system operations
to benefit from previously untapped revenue streams.
and planning require the provision of many energy
The value these companies provide stems from engaging
services—these services need to be priced in a manner
an otherwise unengaged customer, but aggregators may,
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 213
A to provide reserves at the current clearing price R but by capitalizing on economies of scale associated with
not for resource B to do so. However, in a system with a the ICT infrastructure needed to manage EV charging.
penalty (P) for the non-delivery of committed reserves, Additionally, EV aggregators may harness economies
the aggregator will dispatch the uneconomical unit B of scope to provide multiple services to EV owners at
when unit A is not available so long as (R + P) > MCB. lower cost, including charge management, energy supply
Thus, it will be economically rational for the aggregator purchases, and other vehicle-related services, such as
to dispatch unit B, even if another resource, C, managed regular maintenance, navigation, or other services. Finally,
by a different market participant, is available with a cost EV aggregators may provide value by managing supply
MCC < MCB. Allowing aggregators to net penalties in this cost volatility and risk for EV owners.
manner can lead to aggregation that improves the private
At the same time, wherever prices and charges do not
value of DERs but does not improve (and may indeed
fully reflect time- and location-varying marginal prices
for all electricity services, EV aggregators
may provide transitional value in the near
Where aggregation creates only private opportunistic term by aggregating EV fleets and creating
value, regulations, policies, or market rules should appropriate incentives to motivate efficient
generally be modified to align private value with charging. In the future, smart charging
systems embedded in EVs may perform
system value.
this function automatically without the
need for intermediary aggregators. In
addition, if transaction costs dictate
minimum sizes for participation in wholesale energy
harm) overall system efficiency. Furthermore, allowing
markets, capacity markets, or balancing or ancillary
aggregators to net penalties can act as a barrier to entry
service markets, aggregators will be critical to bring
for smaller market agents. This is one of several possible
potentially cost-effective EV flexibility into these markets.
examples of opportunistic aggregation.36
Finally, EV aggregators may provide value by overcoming
Where aggregation creates only private opportunistic informational barriers in cases where aggregators are
value, regulations, policies, or market rules should better able to predict EV charging patterns and manage
generally be modified to align private value with associated uncertainty than system operators.
system value. 37
EV charging may also give rise to opportunistic
aggregation. For example, many European markets apply
6.4.3 An example of aggregation:
dual imbalance penalties, wherein individual agents are
Charging electric vehicle fleets penalized for departing from their scheduled consumption
As an example of all three forms of aggregation— or production in either direction, even when only one of
fundamental, transitory, and opportunistic—consider those directions contributes to overall system imbalances
the case of electric vehicle (EV) charging. Under a (and thus reserve or balancing market dispatch costs).
fully efficient system of prices and regulated charges In such markets, EV aggregators can significantly reduce
for electricity services (see Chapter 4), individual EV imbalance penalties by aggregating many individual
owners could charge their vehicles in a manner that EV charging locations and bidding a single aggregate
maximizes both private welfare and system efficiency. In charging schedule. This aggregation of imbalance
this case, EV aggregators may deliver fundamental value positions is privately valuable, but does not improve
overall system efficiency. A single imbalance price,
36 For further discussion of opportunistic aggregation, see Burger et al. (2016).
37 Exceptions include cases where the opportunistic private value created by regulation
is explicitly acknowledged and desired as a form of subsidy, or where the cost of
regulatory reform exceeds benefits from eliminating inefficiencies and transfers.
38 That is, if the system overall was “short” and had insufficient supply, agents that
were consuming more than their scheduled quantity or producers who were
producing less would be penalized at the single imbalance price, since both
would be contributing to the overall system imbalance and thus to the costs of
dispatching reserve or balance market resources. In this scenario, agents that were
39 Note that despite the name, this category also includes “behind-the-meter” energy
consuming less or generating more than scheduled would actually be reducing the
storage companies.
overall system imbalance and lowering costs, and thus should not be charged an
imbalance penalty (as they would be under dual imbalance price systems). 40 Note that operating reserves take different definitions depending on the system.
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 215
Figure 6.5: Electricity Services Summary
ELECTRICAL &
SERVICE(S) DEMAND RESPONSE & EMS THERMAL STORAGE SOLAR PV
Energy 37
Non-Electricity Services 6 17 6
Firm Capacity 9
While operating reserve services are rather lucrative, the Spees, Newell, and Pfeifenberger 2013). As one might
markets for these services are very small, typically making expect, the only business models in our dataset that
up less than 4 percent of total energy costs (Denholm et provide energy in a distributed fashion are those that
al. 2013; Rebours et al. 2007; Masiello, Roberts, and Sloan utilize solar PV. Where a business model is described as
2014). Intense competition for the provision of operating providing “operating reserves,” the business can provide
reserve services could prove challenging for demand the range of primary, secondary, or tertiary reserves;
response and energy storage technologies. On the other other business models may provide only a single type of
hand, payments for firm capacity can be quite significant reserves (notably, secondary reserves).
(on the order of hundreds of dollars per megawatt-day)
depending on desired reserve margins (tens of gigawatts
in certain systems) (Crampton, Ockenfels, and Soft 2013;
DER Provider 16 6
Residential 5 6
Industrial &
Load-Serving Entity
5
Load-Serving Entity
& Regulated Utility 4
Figure 6.6 summarizes the customers targeted by the also refers to utilities that are vertically integrated such
business models in our sample. Businesses that act as that they also function as the load-serving entity.
intermediaries between (and therefore service providers
Figure 6.7 shows that most business models in our
to) two agents are represented with a double-sided
dataset are targeting end users directly. DR and EMS
arrow (<->) connecting the two customer segments.
companies primarily target larger commercial and
Commercial, institutional, or municipal customers are
industrial customers, while solar PV and storage
represented by the abbreviation “C/I/M.” The customer
companies target more diverse customer segments. A
segment “DER Provider” indicates that the company is
smaller number of companies sell services directly to
selling its products to businesses that then integrate
regulated utilities, independent system operators (ISOs),
one or more DERs at customer sites. The “Regulated
transmission system operators (TSOs), or regional
Utility” customer segment refers to network companies—
transmission organizations (RTOs).
distribution, transmission, or both. This customer segment
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 217
Figure 6.7: Summary of Revenue Streams
ELECTRICAL &
REVENUE STREAM(S) DEMAND RESPONSE & EMS THERMAL STORAGE SOLAR PV
Subscription Fees
& Brokerage Fees 37
Asset Sale 14 6
Figure 6.7 summarizes the revenue streams leveraged Using the same data as the data presented in this section,
by the business models in our sample. Many companies Appendix B explores each of the three DER technology
leverage multiple revenue streams. The structure of categories in much greater detail. For each of the three
the revenue streams often depends on the customer DER categories, we defined a relatively small set of
segments that are targeted. For example, many solar business model “archetypes;” the archetypes, in turn,
PV integrators offer consumers the option to directly encompass many individual business models. These
purchase their PV system, lease the system, or take archetypes circumscribe the key value creation and
out a loan for the system. Notably, the vast majority of capture components of the most common DER business
DR companies leverage subscription fees or brokerage models that exist today.
fees (or a combination thereof) for their services.
Such brokerage fees are typically structured as shared 6.5.3 Discussion and conclusions
savings arrangements or as fees on payments earned in
Our analysis here and in Appendix B leads to four
markets. Given the nascence of the storage market and
observations about the DER business landscape that we
the difficulties in predicting end-user savings, storage
believe have important implications for electric utilities,
business models have so far relied heavily on asset sales
DER entrepreneurs, policymakers, and regulators.
or financing.
Figure 6.8: Taxonomy of Demand Response and Energy Management System Business Models
EMS
PROVIDERS
Non-Electricity
Services
Secondary
Frequency Control
MARKET-BASED
CAPACITY & RESERVE DR
Firm Capacity
& Secondary UTILITY-BASED
Service(s)
Firm Capacity
Customer Segment(s)
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 219
Surveying a large number of DER business models The influence of policy and regulation
suggests that these myriad businesses can be clustered on DER business models indicates
into a relatively small set of well-defined DER business that many DER businesses today
depend to some degree on capturing
model archetypes.
opportunistic value created by current
regulatory
6.5.3.2 Policy and regulation are key drivers of the or policy regimes. As discussed in
structure of DER businesses Section 6.4, this type of value capture accrues to private
stakeholders but not to the electricity system as a
Our second observation is that current DER business
whole, and thus typically results in economic transfers.
model archetypes appear to be driven more by regulatory
If regulation and policy improve over time, opportunistic
and policy factors than by technological factors. The
value capture should be minimized and DER businesses
influence of policy and regulation on business structure
will evolve to derive a greater share of their revenues from
is arguably most evident with the solar PV business
the creation of transitory or fundamental value, which
model archetypes, but it applies to all of the technology
delivers real efficiency improvements to electric power
categories we considered. As Figure 6.9 illustrates,
systems.
three solar PV archetypes have emerged (as well as
several solar-plus-storage archetypes). The largest
cluster consists of distributed PV finance and integration
companies, which primarily leverage net metering or The influence of policy and regulation on DER
feed-in tariff policies and creative financial structures business models indicates that many DER
to maximize the value of available tax credits or other businesses today depend to some degree
incentives and overcome capital barriers to end-user on capturing opportunistic value created
adoption. Utility-scale solar PV finance and integration by current regulatory or policy regimes. If
companies build megawatt-scale solar installations and
regulation and policy improve over time,
have largely emerged in response to state renewable
opportunistic value capture should be
portfolio standards in the United States and feed-in tariff
policies in Europe. Finally, community solar providers minimized and DER businesses will evolve to
have emerged to capitalize on economies of unit scale derive a greater share of their revenues from
or to enable consumers located in areas unsuitable for the creation of transitory or fundamental
their own solar PV production to procure solar from value, which delivers real efficiency
community solar projects. The community solar market improvements to electric power systems.
is still relatively small (tens to hundreds of megawatts in
the United States) and geographically restricted to policy-
friendly environments. Without key policy and regulatory
elements, we would not have seen the widespread
emergence of these three solar PV archetypes. Similarly,
the majority of business models within the “market-based
capacity and reserve DR” archetype depend on capacity
remuneration mechanisms that were established by
regulators and legislatures.
TECHNOLOGY
Non-Electricity
MANUFACTURERS
Services
Capacity &
Operating Reserves
SOLAR-PLUS-STORAGE
END-USER OPTIMIZATION
Energy & Firm
Capacity
UTILITY-SCALE PV
FINANCIERS & INTEGRATORS
Energy COMMUNITY
SOLAR PROVIDERS
DISTRIBUTED PV
FINANCIERS & INEGRATORS
Residential C/I/M C/I/M & Industrial & Residential Residential & C/I/M &
& Industrial & Load-Serving <-> C/I/M & Industrial Industrial
C/I/M Load-Serving Entity DER Provider <-> <->
Entity ISO/TSO/RTO ISO/TSO/RTO
Residential Residential C/I/M Industrial Load-Serving Residential Residential & DER
& & Industrial Entity & & C/I/M C/I/M & Provider
C/I/M ISO/TSO/RTO <-> ElectricVehicle
& Industrial ISO/TSO/RTO <->
ISO/TSO/RTO
Customer Segment(s)
6.5.3.3 DER business models are not static and change as new regulations and policies are introduced,
evolve over time old ones expire, and existing ones change. Thus, the third
key observation to emerge from our analysis is that the
Indeed, our analysis does not suggest that the DER
business landscape for DERs 10, 15, or 20 years from now
business model landscape will continue to be defined and
will likely look very different than the landscape of today.
driven by regulation and policy indefinitely. Continued
cost declines and technological innovation may well
6.5.3.4 DER business models compete to provide a
lead to markets for DERs that are less defined by policy
limited range of electricity services
and regulatory conditions and more by the delivery of
fundamental or transitory value. Our analysis also does A fourth policy-relevant observation centers on the
not suggest that the business model archetypes we diverse range of DER business models that currently
have identified are here to stay. On the contrary, the fact compete for market share in providing a limited set
that these archetypes are sensitive to regulation and of commodity electricity services (Figure 6.5). DER
policy environments suggests that they are very likely to businesses compete within archetypes (e.g., many DR
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 221
companies compete to provide capacity and ancillary
services), but as DER business models and technologies
become increasingly mature we can expect increased
competition across archetypes. For example, DR and
energy storage technologies will likely compete to provide
reserves and capacity-based services. We might also
expect distributed PV businesses to face competition
from utility-scale PV businesses. Because these
business models generally provide commodity services,
differentiation beyond price may be difficult to realize.
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 223
Eurelectric. 2013b. Power Distribution in Europe: Facts and Hawaii PUC. 2014. The Commission’s Inclinations on the
Figures. Brussels, Belgium: Eurelectric. February 2013. Future of Hawaii’s Electric Utilities: Aligning the Utility
Business Model with Customer Interests and Public
Eurelectric. 2015. Designing Fair and Equitable Market
Policy Goals. Honolulu, HI: State of Hawaii Public Utility
Rules for Demand Response Aggregation. Brussels, Belgium:
Commission. April 2014.
Eurelectric. March 2015.
Hedman, W., M.C. Ferris, R.P. O’Neill, E.B. Fisher, and
European Commission. 2003. Directive 2003/54/EC
S.S. Oren. 2010. “Co-optimization of Generation Unit
Concerning Common Rules for the Internal Market in
Commitment and Transmission Switching with N-1
Electricity and Repealing Directive 96/92/EC. Brussels,
Reliability.” IEEE Transactions on Power Systems 25(2):
Belgium: European Commission. June 2003.
1052-1063. May 2010.
European Commission. 2009. Directive 2009/72/EC
Hedman, W., S.S. Oren, and R.P. O’Neill. 2011. “Optimal
Concerning Common Rules for the Internal Market in
Transmission Switching: Economic Efficiency and Market
Electricity and Repealing Directive 2003/54/EC. Brussels,
Implications.” Journal of Regulatory Economics 40(2): 111-
Belgium: European Commission. July 2009.
140. October 2011.
European Commission. 2010. “Interpretative Note on
Hogan, W.W. 1997. “Transmission rights, through the
Directive 2009/72/EC Concerning Common Rules for
looking glass: a false dichotomy between active and
the Internal Market in Electricity and Directive 2009/73/
passive trading.” The Electricity Journal 10(8): 86-91.
EC Concerning Common Rules for the Internal Market in
Natural Gas: the Unbundling Regime.” Commission Staff i3Connect. 2016. Cleantech Group. i3connect.com.
Working Paper. Brussels, Belgium: European Commission. Accessed March 1, 2016.
January 2010.
Ikäheimo, J., C. Evens, and S. Kärkkäinen. 2010. DER
European Commission. 2015. “Delivering a New Deal for Aggregator Business: the Finnish Case.
Energy Consumers.” Brussels, Belgium.
Joskow, P.L. and R. Schmalensee. 1983. Markets for Power:
FERC. 1996. Order No. 888, Final Rule. Promoting An Analysis of Electric Utility Deregulation. Cambridge, MA:
Wholesale Competition Through Open Access Non- MIT Press. November 1983.
discriminatory Transmission Services by Public Utilities
Jenkins, J.D. and I.J. Pérez-Arriaga. 2017. “Improved
(Docket No. RM95-8-000). Washington DC: US Federal
Regulatory Approaches for the Remuneration of
Energy Regulatory Commission. April 24, 1996.
Electricity Distribution Utilities with High Penetrations of
FERC. 1999. Order No. 2000, Final Rule. Regulation Distributed Energy Resources. The Energy Journal 38(3).
Transmission Organizations (Docket No. RM99-20- doi:10.5547/01956574.38.3.jjen.
000). Washington DC: US Federal Energy Regulatory
Kranning, M., E. Chu, J. Lavaei, and S. Boyd. 2013.
Commission. December 20, 1999.
“Dynamic Network Energy Management via Proximal
Friedrichsen, N. 2015. “Governing Smart Grids: The Case Message Passing.” Foundations and Trends in Optimization
for an Independent System Operator.” European Journal of 1(2): 70-122.
Law and Economics 39(3): 553-572. June 2015. First online
Littlechild, S.C. 2000. “Why We Need Electricity
July 2012.
Retailers: A Reply to Joskow on Wholesale Spot Price
Glover, J.D., M.S. Sarma, and T.J. Overbye. 2011. “Power Pass-through.” Cambridge, UK.
Distribution.” In Power System Analysis and Design, Chaper
14. 5th Edition. Cengage Learning. January 2011.
Masiello, R.D., B. Roberts, and T. Sloan. 2014. “Business Parsons, J., C. Colbert, J. Larrieu, T. Martin, and M.E.
Models for Deploying and Operating Energy Storage Mastrangelo. 2015. “Financial Arbitrage and Efficient
and Risk Mitigation Aspects.” Proceedings of the IEEE Dispatch in Wholesale Electricity Markets.” Center for
102:1052–64. doi:10.1109/JPROC.2014.2326810. Energy and Environmental Policy Research, Working
Paper 2015-002. Cambridge, MA: Massachusetts
Mallet, P., P. Granström, P. Hallberg, G. Lorenz, and
Institute of Technology. February 2015.
P. Mandatova. 2014. “Power to the People. European
Perspective on the Future of Electric Distribution.” Pérez-Arriaga, I. 2011. “Managing Large Scale Penetration
IEEE Power and Energy Magazine: 51-64. doi: 10.1109/ of Intermittent Renewables.” MIT Energy Initiative
MPE.2013.2294512. Symposium. Cambridge, MA: Massachusetts Institute of
Technology.
Mansur, E.T. and M.W. White. 2012. “Market Organization
and Efficiency in Electricity Markets.” Working paper. Pérez-Arriaga, I., S. Ruester, S. Scwenen, C. Batlle, and
January 13, 2012. J-M Glachant. 2013. From Distribution Networks to Smart
Distribution Systems: Rethinking the Regulation of European
Millán, J. 2006. Entre el Estado y el Mercado. Tres décadas
Electricity DSOs. THINK Project, Topic 12. Florence, Italy:
de reformas en el sector eléctrico de América Latina (in
European University Institute.
Spanish). Banco Interamericano de Desarrollo, 2006.
ISBN: 1597820288. PJM. 2014. Retail Electricity Consumer Opportunities for
Demand Response in PJM’s Wholesale Markets.
MP2 Energy. 2016. First-Class MPACT Energy
Management. www.mp2energy.com/demand-response. PJM. 2015. Demand Response Fact Sheet. June 29, 2015.
Accessed March 1, 2016.
Rebours, Y.G., D.S. Kirschen, M. Trotignon, S. Rossignol.
NYDPS. 2014. Case 14-M-0101 Proceeding on Motion 2007. “A Survey of Frequency and Voltage Control
of the Commission in Regard to Reforming the Energy Ancillary Services - Part II: Economic Features.” IEEE
Vision: DPS Staff Report and Proposal. Albany, NY: State Transactions on Power Systems 22:358–66. doi:10.1109/
of New York Public Service Commission, Department of TPWRS.2006.888965.
Public Service.
Ruiz, P.A., E.A. Goldis, A.M. Rudkevich, M.C. Caramanis,
NYDPS. 2015a. Order Adopting Regulatory Policy C.R. Philbrick, and J.M. Foster. 2016. “Security-constrained
Framework and Implementation Plan. Case 14-M-0101: transmission topology Control MILP Formulation Using
Reforming the Energy Vision. Albany, NY: State of New Sensitivity Factors.” IEEE Transactions on Power Systems,
York Public Service Commission, Department of Public doi: 10.1109/TPWRS.2016.2577689.
Service. February 2015.
Slot, T., H. Dijk, R. Haffner, and A. van der Walle. 2015.
NYDPS. 2015b. “Staff White Paper on Ratemaking and Options for Future European Electricity System Operators.
Utility Business Models.” Albany, NY: State of New York Brussels, Belgium: European Commission. December
Public Service Commission, Department of Public Service. 2015.
July 2015.
Smart, D.C.C. 2015. Building a Smart Metering Network
for Great Britain. London: Smart Data Communications
Company, Ltd. October 2015.
CHAPTER 6: Restructuring Revisited: Electricity Industry Structure in a More Distributed Future 225
Smart Grid Task Force. 2015. Regulatory Recommendations
for the Deployment of Flexibility. Report of Expert Group 3
to the European Commission. Brussels, Belgium.
07
The Re-evolution of Short- and
Long-Term Electricity Market Design
7.1 Toward an Electricity and support the chain of decisions necessary to match
electricity production and consumption in the most
Market for All economically efficient manner. This supply–demand
One of the key objectives for power system restructuring balance, however, is not the result of interactions between
was the establishment of wholesale markets that allowed market agents alone. Regulators, system operators,
competition in the generation of electricity. Wholesale and market designers have historically intervened to
markets replaced vertically integrated utilities with an enhance investment signals via capacity mechanisms
open-access model, whereby various generating companies and/or technology-specific incentives (e.g., so-called
could compete to meet a share of electricity demand. clean energy technology support mechanisms). Such
interventions have been used, for example, to ensure
This transformation required replacing centralized
the availability of sufficient resources to cope with
decisions with competitive processes, and creating
contingencies that could endanger the reliable operation
multiple markets with different time scales. Today’s
of the power system.
wholesale electricity markets comprise a series of
sequential trading opportunities — from long-term Given the complexity of electric power systems,
markets that are cleared years in advance, to very electricity wholesale market design remains a topic of
short-term balancing markets (namely in the European heated discussion, even after decades of experience.
context) and so-called real-time markets in the United Design decisions for markets and investment support
States — with the ultimate goal of meeting electricity mechanisms are further complicated by a host of
demand in an efficient, secure, and environmentally new challenges, including adapting to technological
acceptable way. These wholesale market mechanisms developments with respect to distributed energy
allow market agents to manage their risk exposure resources (DERs) and information and communications
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 227
technologies (ICTs). DERs present specific challenges 7.2 Leveling the Playing Field:
to market design due to their small size (compared
with centralized resources) and because they are often
Short-term Markets for an
affected by local distribution network constraints that are Efficient Integration of the New
not typically considered in wholesale markets. Moreover,
Diversity of Resources
some DERs — such as electric vehicles, batteries, and
demand response — represent a new class of technologies. The short-term market mechanisms examined in this
Each has its own set of operating constraints and, in most section are the so-called energy markets: day-ahead and
cases, little experience with market integration. Most intraday markets. Prices in these markets are frequently
DERs have not played a significant role in electricity used as a reference for many other transactions, as
markets previously, but now have the potential to expand well as for the reserves and imbalance markets, which
very quickly — which means that electricity market design operate closer to real time and are thus closely linked to
needs to be ready ahead of time. operational security conditions. For reference, we rely on
the US and European electricity market models, which
The goal of this chapter is to address the challenges
represent two different but increasingly converging
that currently inhibit efficient integration of all types of
design archetypes. The most prominent difference
resources — centralized and distributed, conventional
between these two models is found in the division of
and emerging — into electricity markets. Achieving this
responsibilities between the power exchange and the
requires a series of new or updated market mechanisms
system operator. However, it is not possible to attribute
and platforms to generate price signals that will drive
all differences to this foundational decision. The next
efficient operational and investment decisions. An
sections describe several specific market design elements
in-depth review of existing subsidy mechanisms and
while highlighting differences and similarities.
a transition to more market-compatible schemes for
better integrating clean energy technologies are also
7.2.1 Energy (day-ahead and intraday)
needed. The reimagining of support mechanisms is
market mechanisms
all the more critical now that some renewable and
distributed technologies are reaching competitive Energy (day-ahead and intraday) mechanisms are a
levels — comparable to those of more conventional sequence of short-term market mechanisms that operate
generating technologies — in some contexts. up to one day prior to the actual delivery of electricity.
They are designed to allow market agents and the system
This chapter is divided into two parts. The first
operator to match supply and demand in a secure and
(Section 7.2) focuses on the design challenges associated
efficient manner. This section focuses on organizing
with short-term market mechanisms (i.e., markets that
markets that are compatible with system operation
range from day-ahead to real time) and analyzes options
requirements while also producing more accurate
for efficiently integrating market trading and system
market signals. It then introduces the concept of market
operation. This section also examines reserve and
“granularity” to describe the temporal and geographical
balancing markets, which are intimately related to short–
resolution achieved by market products and price signals.
term markets. The second part (Section 7.3) reviews
The section concludes with a discussion of the elements
needed changes to long-term capacity mechanisms and
of auction design. For each of these topics we focus on
clean energy support mechanisms. Specifically, the focus
changes that are necessary to achieve an efficient co-
of this section is on redesigning these mechanisms to
integration of both conventional generation technologies
enhance their compatibility with existing markets and
and DERs.
minimize associated market distortions.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 229
being the “real-time market,”4 which provides dispatch of these designs allow generators to make adjustments
instructions a few minutes before delivery in five-minute to their energy schedules, which helps to improve the
intervals. These dispatch instructions are partly based on efficiency and accuracy of the initial day-ahead market.
bids (updated if allowed), but also on the ISO’s forecast of
Continuous markets allow market agents to submit
load and renewable energy production.
bids and offers to modify schedules at any time before
In practice, this implies that — in the United last gate closure. Each bid and each offer consists of
States — different resources are effectively subject a quantity of energy with a price (a “price–quantity
to different last gate closures, on grounds that some pair”). Continuous markets are cleared on a first-come,
participants, such as consumers and renewable first-served basis — in essence, they arrange a bilateral
producers, are more efficiently integrated through transaction whenever a buy order (bid) is matched with
ISO forecasting rather than through active market a sell order (offer). Conversely, auction-based intraday
participation. While this may be the case, especially when market sessions resemble day-ahead auctions, albeit
the engagement of consumers and the maturity level of with several market sessions throughout the day. The
renewable technologies are low, ample evidence suggests auctions are centrally cleared with marginal prices.
that renewable generators and demand response While continuous markets provide greater flexibility
providers can significantly improve their forecasts and to market agents, as transactions can be made at
overall performance when faced with cost-reflective any time, intraday auctions have several advantages
price signals.5 If it can be shown that there is value in from the perspective of efficient price formation.
transferring forecasting/scheduling responsibility from Auction-based intraday markets improve liquidity,6 in
some market participants to the ISO, the ISO should part by concentrating all transactions that reflect the
perform this task as a mere intermediary, passing all accumulated events that occurred since the previous
costs incurred for forecasting and balancing of deviations session. Thus, the prices that result from these auctions
on to market participants. This way, market agents could provide a more reliable reference price compared to
determine which arrangement is truly more efficient prices in an illiquid continuous market, which might not
and would have the choice to transfer their forecasting reflect system marginal cost at a given time. Liquidity
responsibility to the ISO — or not. is of particular relevance in markets that are dominated
by a small number of large companies, as is the case in
Efficient intraday prices and market designs
many restructured power systems. In such markets, low
Market designs affect price signals produced in the liquidity is particularly challenging for new and small
intraday timeframe, (i.e., between the day-ahead and real- market entrants, such as DERs and small aggregators.
time markets) and result in different balances between Small market participants would further benefit from
electricity trading and system operation. intraday auctions, since they usually lack continuous
In Europe, where intraday markets play a significant role, 24/7 trading desks. Moreover, intraday auctions have the
current discussions are focused on how to implement advantage of implicitly pricing transmission capacity. In
these markets most efficiently. Two market designs have a continuous market, by contrast, transmission capacity
generally prevailed: continuous intraday markets and is allocated at zero cost (in other words, the bid–offer
intraday auctions that occur at prescribed times. Both spread is zero) on a first-come, first-served basis.
4 The “real-time market” involves a number of processes that go beyond what it 6 Market liquidity can be defined as a market’s ability to allow assets to be bought
is commonly understood as an exchange. In most cases, US real-time markets and sold at stable prices. Liquidity can be quantified in different ways. For example
provide dispatch instructions before real time, but prices are computed in separate Hageman and Weber (2015) develop a trading-volume-based analysis of liquidity
ex post processes. Also, as described by Helman et al. (2008): “In the real-time in European intraday markets that also highlights an important consideration: To
market, the system operators manage congestion on a minute-to-minute basis in properly assess liquidity in different power systems, trading volumes must be
part through auction prices and in part through non-market operating decisions.” compared with a benchmark expected volume, which does not depend on the market
5 For example, wind forecast errors have gradually decreased due to the exposure of design but on the characteristics of each system. For example, in a system with high
these generators to imbalance prices (Herrero et al. 2016). renewable energy production, the expected trading volume in intraday markets is
also high due to forecast errors. If this system has higher trading volumes than a
mostly thermal system, the difference cannot be attributed to market design.
7 As discussed in Neuhoff et al. (2016), the addition of an intraday auction to the 8 The California ISO features intraday markets, although with a limited scope. The
German continuous intraday market increased liquidity and led to greater market hour-ahead scheduling process is binding for imports, and the fifteen-minute
depth (revealing market participants’ capacity/flexibility) and reduced price volatility. market is binding for some renewable resources.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 231
Box 7.1: Enhancing Intraday Price Signals in US ISO Markets —
Results of a Modeling Exercise
(Herrero et al. 2016)
ISOs could benefit from receiving forecast updates directly from producers, since producers can better
account for local conditions, but only to the extent that producers are appropriately rewarded for the
value of accurate forecasting. Although receiving forecast corrections as soon as possible has clear
economic value, under the two-settlement system used in ISO markets, this value is not fully disclosed
and costs are not properly allocated. To improve the efficiency of the market, each intraday commitment
process should be accompanied by its own intraday settlement. This leads to the proposed multi-
settlement system, which permits allocation of intraday costs according to cost causality principles and
creates efficient signals for market agents to improve forecast accuracy.
In our modeling exercise, we compare outcomes for two renewable energy units that deviate from their
forecasts at the same time to illustrate the incentives produced by intraday settlements versus the
two-settlement system (Figure 7.1). Unit 1 corrects its forecast later (Figure 7.1c), while Unit 2 performs
a better forecast and is able to make a correction sooner (Figure 7.1b). The higher cost of the later
correction is captured by intraday prices. However, if the two-settlement system is used, these two
deviations are lumped together and both units pay the real-time price for the full difference between the
day-ahead forecast and the real-time deviation. In this way, the two-settlement system loses the more
precise intraday signal.
Figure 7.1: Sequential Dispatches as PV Output Is Corrected and Corresponding Intraday Prices
PV unit 1
PV unit 2
Energy Price ($/MWh)
(a) Day-ahead dispatch (b) Intraday 1 dispatch (c) Intraday 2/final dispatch
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 233
Box 7.2: One Effect of Low Temporal Granularity
Among the most visible effects of breaking electricity production into hourly blocks is the deviation
of system frequency at the transition from one hour to the next. This leads to an inefficient use of
operating reserves that could be avoided with more continuous representations of demand — for
example, via more granular dispatch intervals.9
9 Another alternative to deal with this inefficiency would be to convert constant hourly schedules into continuous piecewise linear schedules.
“The Federal Energy Regulatory Commission is revising its regulations to address certain practices
that fail to compensate resources at prices that reflect the value of the service resources provide to the
system, thereby distorting price signals, and in certain instances, creating a disincentive for resources to
respond to dispatch signals. We require that each regional transmission organization and independent
system operator align settlement and dispatch intervals by: (1) settling energy transactions in its real-
time markets at the same time interval it dispatches energy; (2) settling operating reserves transactions
in its real-time markets at the same time interval it prices operating reserves; and (3) settling intertie
transactions in the same time interval it schedules intertie transactions.” (FERC 2016)
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 235
DERs increase the need for nodal pricing in wholesale markets, integration within the same
while simultaneously making it necessary to expose DERs to country (the case in the United
price signals with the same temporal and spatial resolution. States), is challenging because
of the need to coordinate
different states with different
geographical consumer discrimination, and institutional institutions (including different
and governance issues. Different analyses have shown ISOs) and regulations. But the challenge is an order of
how the gains from implementing nodal pricing — in magnitude greater when it comes to integrating markets
the context of congested systems — clearly outweigh across different countries, as in the case of the European
implementation costs.11 At the same time, zonal pricing is internal energy market.
usually deemed appropriate in strongly meshed systems
From wholesale to retail granularity
with easy-to-forecast power flows because it simplifies
zones identification. The arguments for zonal pricing As mentioned above, a nodal market could be
often rest on the need to create larger bidding zones compatible with average (zonal) pricing for low-
so as to reinforce liquidity and competition, and induce voltage consumers. Specifically, nodal prices are often
investment signals that are more stable for generators averaged, both in the United States and Europe, to
and less discriminatory toward demand resources. compute retail tariffs with lower (effectively, zonal)
However, many of the points often made in favor of zonal resolution. However, price averaging at the distribution
pricing are, to say the least, highly debatable. For example, level will be increasingly inefficient (as we have already
in nodal markets with long-term hedging, market liquidity pointed out, this is also true for temporal averaging), as
is not necessarily affected — financial transmission rights low-voltage consumers start responding to price signals,
(FTRs) and hubs12 play a key role here. On the other both in the short and the long term.
hand, nodal prices can be compatible with whatever Note that consumers can be exposed to market prices
geographical resolution is used in end-consumer prices through tariffs, or through direct market participation. If
(in other words, retail customers do not necessarily need consumers become active market participants, the logic
to face nodal prices in nodal markets). used to clear markets would have to account for the
The fact is that large shares of dispersed variable division of the market into two granularities (nodal prices
generation are already causing congestion problems to for generation and averaged zonal prices for consumers).
increase, for reasons that are sometimes localized and This would create additional implementation challenges
sometimes structural. Growing congestion between, for and inefficiencies. Clearing a market where participants
example, northern and southern Germany makes power face different prices gives rise to arbitrage opportunities
flows less predictable and makes it more difficult to and is computationally complex.13 Computational
define effective zones (Figure 7.3). Also, increased DER complexity is also a source of inefficiency to the extent
participation will improve liquidity in electricity markets, that it limits the introduction of other necessary market
but requires ever more localized investment signals. design elements (the next section, for example, discusses
the need for more complex bidding formats). Ultimately,
We recommend a shift toward providing more detailed as the emergence of DERs erodes the traditional
and robust spatial resolution at the wholesale level, separation between generation and demand, and between
and in this respect nodal prices represent the ideal transmission and distribution, exposing all market
framework if workable. Regarding workability, clearly, the
most important hurdles for implementing nodal prices 13 The Italian power exchange is an example of a market where market participants
face different prices. Generation receives zonal prices in the day-ahead market,
stem from institutional and governance issues. Market while for political reasons the same weighted average price (known as “PUN” for
Prezzo Unico Nazionale or “single national price” in Italian) is used for all demand
(including large consumers actively bidding in the market). The market clearing
algorithm explicitly considers this policy, which results in a very complex problem
11 See, for instance, Neuhoff and Boyd (2011).
given that the algorithm must search for prices that ensure supply–demand
12 In US nodal markets, trading hubs that use an average price across a set of nodes equilibrium, while also making the PUN equal to the weighted average of the zonal
have provided adequate liquidity. This is a suitable trade-off to avoid liquidity prices and satisfying transmission constraints, in an iterative fashion, with no
problems as a result of moving toward a higher level of spatial granularity. mathematical guarantee that a solution exists.
First, zonal pricing should reflect transmission constraints between countries but, in many cases,
these constraints are not actually associated with lines between countries — rather, they reflect
congestion within a country. This occurs because TSOs sometimes limit international flows to avoid
domestic congestion.
Second, nodal pricing simulations illustrate that congestion and price patterns would vary considerably
between different scenarios for wind penetration. Therefore, approaches aimed at defining price zones
within countries are not suitable to address internal congestion, as the zones would either have to vary
depending on system conditions (which would be impractical for contracting purposes) or be small
(and thus be essentially equivalent to nodal pricing).
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 237
participants to price signals with the same temporal The European approach is based on the concept of a
and spatial resolution will become unavoidable (at least simple auction, where market agents indicate their
with respect to the transmission network, although the willingness to buy or sell electricity using price–quantity
extension of such price signals to the distribution level pairs. The EU market, nonetheless, has quickly evolved
would increase efficiency too, as discussed in Chapter 4). toward increasingly complex bidding formats, which
do not specifically capture the characteristics of any
7.2.1.3 Auction design particular type of unit but are instead formulated in a
Fitting bidding formats to new resources general way with the goal of providing a level playing
field to all market participants. The key information
While electricity is often defined as a commodity (in the
provided in the bid is a price to buy or sell electricity, but
sense that one MWh of electricity is indistinguishable
the bid can also include implicit information about the
from another), experience has shown that electricity
conditions under which an agent is willing to buy or sell
markets are more complex than markets for any other
(e.g., agents can present block bids and other complex
commodity. In addition to the constraints that are
conditions). These rather limited bidding formats can be
imposed by the network, electricity producers face
handled reasonably well by large generators that can use
various operational constraints (e.g., start-up, minimum
their ability to manage large portfolios to compensate for
power output, ramping constraints) that have a significant
a lack of complexity, while at the same time benefiting
impact on the cost and thus the operation of generators.
from disclosing the minimum amount of information
New resources (including demand-side resources) will
about their operating cost structures. However, limiting
introduce their own operational constraints, which will
the amount of information contained in bids particularly
also need to be addressed — further complicating matters.
hinders the participation of small producers, makes it
These constraints are usually most relevant in short-term
more difficult to monitor market power, and does not
markets (i.e., day-ahead, intraday, and real-time markets)
improve the efficiency of the resulting dispatch.
and therefore must be considered in the design of
electricity auctions. A hypothetical example is useful for comparing the US
and EU approaches. This example involves a thermal
The goal of a market exchange is to facilitate trading to
power plant that is bidding in a day-ahead auction.
help participants manage their risks and efficiently match
Designing its bid is straightforward in an ISO market.
available supply with demand. In this respect, a key
The generator must submit a multi-part bid containing
element for electricity markets is the design of bidding
relevant information on its operating constraints and
formats that allow market agents to include information
opportunity costs, and the auction will reveal the most
about their costs and operating constraints. US and EU
efficient use of all contributing resources. Bidding in a
markets follow two different approaches to solve this
European market, the same unit could submit a block
problem. In the United States, costs (e.g., variable and
order that made its electricity production conditional on
start up) and operation constraints are represented
being committed for a given number of hours and at a
in bidding formats in a straightforward way, using
given average price (incorporating start-up cost). It could
essentially the same unit commitment and dispatch
also use a profile block order to express that its order
optimization models that were used prior to restructuring.
is only valid if a minimum ratio of the bid is executed
ISOs require generators to submit multi-part offers that
(representing its minimum power output constraint).
represent detailed operational (and opportunity) costs
Incorporating operating costs and constraints in this
(e.g., the California ISO uses warm, intermediate, and
bidding format requires making assumptions about
cold start-up offers), and also the technical constraints
what the most likely operating regime will be on the
that apply to different generating units (e.g., offers in the
following day, based on prior knowledge and the ability
Midcontinent ISO include generators’ minimum run times
to predict future market results. In the European Union,
and down times).
this predictability has decreased due to uncertainty
Differences in bidding formats between the US and EU 15 Assuming that resources had convex cost structures, otherwise simple bids cannot
adequately represent operating costs.
markets can be attributed to a fundamental difference in 16 More precisely, when the market-clearing problem is non-convex.
the models that are used for market clearing and pricing 17 We continue to use the term “side payment” throughout this section — while avoiding
the use of “uplift” — because, in most jurisdictions, the term “uplift” is broader and
also includes other system charges such as different operating reserve charges.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 239
Box 7.5: Bidding a Storage Resource in Electricity Markets
(Usera et al. 2016)
Significant limitations to optimally bidding an energy storage system (ESS) exist both in the US market
(with some exceptions) and in the EU market. In many cases, bidding formats require the participant
to anticipate specifically in which periods the ESS should be producing electricity and in which periods
it should be consuming electricity, and to place separate bids as a generator and as a consumer at
different times of the day according to these predictions.
However, it is possible to devise more suitable bidding formats. The California ISO already provides a
bidding format designed for ESS participants (also called non-generating resources or NGRs), which
can account for capacity limits, ramp rate constraints, maximum and minimum energy constraints,
and state-of-charge constraints subject to charge/discharge efficiency rates. This is an extension of
the multi-part bid approach to new resources, and its main challenge is the added computational
complexity of market models.
In Europe, the best bidding format available would be a combination of linked and exclusive block
orders (which are not available in all power exchanges). Using linked block orders, the ESS participant
can connect its production and consumption bids to ensure, for example, that the storage system is
not called to discharge before it has been charged. This still requires anticipating optimal periods for
consumption and production, but by using exclusive block orders the agent can submit multiple, linked
block-order pairs and have only the most profitable pair accepted (Figure 7.4). This combination of
bidding formats still does not ensure the most efficient use of the resource but it is a considerable
improvement over other, simpler bidding formats.
Figure 7.4: Linked and Exclusive Block Order Usage for a Storage Asset
Purchase order
Exclusive block order
Selling order
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 hours
This example makes use of a long-term capacity expansion model wherein generation investment
decisions are based on market remuneration. For a realistic, thermally-based electricity system with a
high penetration of renewable energy sources, the model determines the resulting generation mix in the
case of traditional marginal cost pricing (i.e., a non-linear or discriminatory pricing scheme) and in the
case of a simplified version of so-called convex-hull pricing (i.e., a linear or non-discriminatory pricing
rule). Figure 7.5 shows the generation mixes that result from applying the two different pricing rules.
0 5 10 15 20 25 30 35 40
Installed Capacity (GW)
Three important conclusions can be drawn from this simulation. First, the way in which non-convexities
are reflected in the uniform price can have a significant impact on the prices that drive market agents’
investment decisions. Second, the linear pricing rule appears to promote a more efficient energy mix (see
details in the referenced paper). And third, contrary to what a superficial analysis might suggest, a linear
pricing rule does not necessarily produce higher energy prices than a non-linear pricing rule; in fact, a
linear pricing rule can lower the average energy price since it attracts generation technologies with lower
variable costs. However, these results also suggest that the gains resulting from different investment
decisions under different pricing rules are small and easily swamped by other considerations.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 241
because some bids are rejected as needed to ensure that Therefore, allocation rules must be updated so that side
all accepted bids are profitable at the uniform price (in payments are not borne exclusively by either the demand
other words, an offer will never be accepted if doing so or generation part of the market, and the computation of
requires a side payment). these allocation decisions must be incorporated in the
market-clearing problem.
The advantages of the US ISO approach are clear: The
market clearing mechanism is such that it makes the EU markets, on the other hand, avoid side-payment
most efficient use of all resources. However, the use problems at the expense of short-term dispatch efficiency,
a trade-off that should be carefully
reconsidered. An additional problem
in EU markets is computational
The demand side of electricity markets should not bear complexity. The EU market-clearing
the full cost of side payments. EU and US markets should model, which includes uniform pricing
improve the balance between optimal dispatching and and revenue sufficiency constraints,
uniform pricing. is intrinsically more difficult to solve
than an optimal unit commitment
and dispatch model. These two
problems are aggravated by the
of side payments distorts price signals and could lead increasing complexity of bidding formats and the entry
to inefficient investment and operating decisions. This of DERs. Thus it is necessary to find new solutions that
problem has been acknowledged for a long time, and bring EU markets closer to optimal dispatching. Some
ISOs have implemented slight variations to the traditional such alternatives are at an early stage of discussion (PCR
marginal cost pricing to minimize its impact (FERC 2014). 2016), but European legislation is ambiguous on the
Box 7.6 summarizes a modeling exercise that assessed possibility of introducing discriminatory side payments
the extent to which long-term investment incentives can (Madani et al. 2016).
be affected by differences in pricing rules.
7.2.2 Reserves and Balancing Markets
Innovations in US pricing rules will improve price signals
and reduce the need for side payments, but those side To maintain reliable operation, power systems must have
payments that remain will be increasingly challenging the ability to respond to sudden changes in generation or
to allocate (O’Neill et al. 2016). Currently in the United consumption in order to constantly balance both. Once
States, side payments are charged to demand, which is the day-ahead market is cleared, reliability is achieved
the least distortionary approach given that demand is through the use of different types of reserves that can
relatively price inelastic. This choice also comes from be competitively procured by the system operator. This
the perception of side payments as generation costs, section reviews the changes that should be taken into
since they originate from complex generation offers. consideration when redesigning the reserve products
However, as soon as demand resources are allowed to that are traded, as well as the market mechanisms for
use complex bidding formats (e.g., to reflect some of acquiring these resources.
the usual constraints that could also apply to demand
7.2.2.1 Reserves markets
resources, such as a minimum consumption level
that would prevent a demand bid from being partially Redefining reserve products
accepted, or a minimum run time), some demand bids The reserve products currently purchased by system
would equally require side payments. Demand resource operators are tailored to the characteristics of
providers can also experience revenue insufficiency if conventional generating technologies. The transformation
their bids are accepted in the market at a certain price of the power system changes reserves requirements, not
and they are later charged additional side payments.
Figure 7.6: Regulation Signals for Two Secondary Reserve Products in the PJM System
Note: Regulation signals are shown for Reg A (blue) and Reg D (green). Source: PJM (2013)
The figure shows the main difference between the two products: The Reg A signal includes a smoothing
(low-pass) filter to better fit ramping requirements to the capabilities of traditional resources, while
Reg D uses a faster regulation signal (responding to higher frequency perturbations). The Reg D
product was designed to take advantage of small storage (battery) resources, which have virtually no
ramping constraints but must contend with a different constraint: namely limited energy availability. To
overcome this barrier, the control system for Reg D is implemented in such a way that the net energy
required is zero over a short period of time (95 percent of the time the controller converges in less than
15 minutes). This ensures that storage resources will not be required to sustain a response for a long
time and will be able to maintain a relatively stable state of charge. Importantly, Reg D is not exclusively
available for batteries — many hydro resources are also qualified to provide this service.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 243
only because some new technologies are more variable Some established simplifications in reserve product
than conventional technologies, but also because these definitions have become barriers to DER participation
new resources may offer a new source of flexibility. For and should be reconsidered. For instance, most power
example, distributed solar PV inverters could be used systems have combined products for upward and
to support primary frequency control, but they will do downward reserves (especially for secondary/frequency
so only to the extent that this service is appropriately restoration reserves) that could easily be replaced by
rewarded. To fully exploit this potential flexibility, it is two separate products. Separately procuring upward and
important to remove inefficient barriers in current product downward reserves not only results in a more efficient
definitions and to create innovative reserve products that use of resources (by not over-procuring one type of
value different capabilities (see example in Box 7.7). reserve only because the other is needed), it also allows
for the participation of resources that are best suited to
However, there is a delicate balance between defining
provide only one of these services. For example, wind
reserve products that are better suited to a variety of
or solar generation can be readily curtailed to provide
downward reserves, while various
forms of controllable demand
Innovative reserve products should be implemented to fully can easily be curtailed to provide
using DERs for reserve provision should be removed. The requirement for minimum bid
sizes is another barrier that can
be removed. While the cost of
metering and control equipment
new technologies, and creating new products for each will continue to provide economies of scale for large
individual technology. The objective of new reserve resources, markets should not impose artificial barriers
products should always be to provide value to the power to entry for small resources. If minimum bid sizes are
system; new products should not be exclusively tailored unnecessary for the operation of the market, they should
to the capabilities of a particular technology. If new not be enforced. At the same time, DER aggregation
products are adapted to facilitate the use of a particular should be facilitated to help DERs realize economies of
technology, as has been done in the past for conventional scale, incur lower transaction costs, mitigate risk, and
technologies, these adaptations should also contribute contribute to a more efficient system.
value to the electricity system. Product adaptations to
Efficient pricing of reserves
support a particular technology (e.g., for policy purposes),
on the other hand, are not advised, as they would inhibit Implementing a reserve market is necessary but not
effective competition. sufficient to efficiently price reserves. The definition of
reserve requirements and reserve settlement mechanisms
Aggregating resources can help maintain the balance significantly affects the remuneration of reserve services
between developing refined reserve products and and is an important determinant of an efficient pricing
developing too many reserve products. An aggregator can system, for reasons we discuss in this section.
combine the responses of different devices to meet the
requirements of the system operator, thereby exploiting
new sources of flexibility without overly complicating
reserve product definitions.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 245
markets, the quantity of reserves required is fixed, capacity is more uncertain than that of a conventional
regardless of the cost of those reserves. However, these generator) but it can provide balancing energy in real
reserve requirements should reflect the changing value of time. In cases like this, joint procurement of capacity
reserves to the system, and should surely be periodically and energy can be a barrier for the participation of some
reexamined (see further discussion in Box 7.8). Assessing DERs. This barrier could be overcome by organizing large
this value is difficult, and any attempt to do so entails DER aggregations, which can provide more firm reserve
significant intervention into markets. Therefore, the long- commitments. This approach, however, still would not
term benefits of flexible reserve requirements should be guarantee the most efficient use of energy from remaining
weighed against the costs of such intervention. resources. For this reason, separating the procurement of
reserve capacity and energy is recommended. If the two
Operating reserves require the provision of both system
products are separated, capacity can be procured ahead
capacity (MW) and the use of balancing energy (MWh).
of time as a kind of “insurance” to provide the certainty
In many cases, capacity and balancing energy are
and reliability needed for system operation — thereby
jointly procured for various services. In such cases, only
still providing incentives for DER aggregation — while the
resources that are selling capacity can provide real-time
decision about which resources will provide energy in real
balancing energy (typically via a pro-rata allocation). This
time can be postponed for maximum efficiency.
is detrimental for technologies that cannot guarantee
availability when reserves are procured (often, a day or
more ahead) but that could be efficiently deployed in real
time. As an example, a fleet of electric vehicles may not
be able to commit much capacity in advance (its available
Figure 7.7 illustrates the impacts of implementing an ORDC through a simple case study. Using a
realistic electricity system, the study shows that as the reliability target increases (thereby shifting the
ORDC outward), the long-run installed capacity of peaking units increases. The implementation of the
ORDC also serves as a risk hedge for these scarcity-oriented units, as it increases the frequency with
which these units are remunerated in the short term.
Figure 7.7: Illustrative ORDCs and Optimal Capacity Mix in a Sample Case
10,000
Optimal capacity mix [GW]
80
Operating reserve value
8,000
60 Peaker
[$/MW/h]
6,000
40
4,000 Midload
2,000 1 2 3 20
Baseload
0
0 2,000 4,000 6,000 8,000 10,000 12,000
0
Operating reserve level [MW] 1 2 3
A simplified version of the ORDC was implemented in ERCOT, the system that serves most of the state
of Texas, in 2014. Additionally, most ISOs implement some form of scarcity pricing, and discussions are
currently underway to redefine flexible reserve requirements.19
19 We recommend the implementation of an ORDC, as it improves the market signal for agents in the short and long terms. However, as we discuss in Section 7.3.1.1, implementing
the ORDC by itself is unlikely to solve reliability problems (when detected as such), as the key reason behind the alleged need to implement capacity mechanisms in today’s
markets happens to be the large regulatory and planning uncertainty that investors face.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 247
7.2.2.2 Imbalance settlements Maintaining special treatment for particular technologies
has negative consequences for overall system costs. It
This section analyzes three market design elements that
dilutes price signals that could incentivize investment in
are related to imbalance settlements: (1) the possibility to
technologies with load-following capabilities. Moreover,
exempt some resources from balancing responsibilities;
it inhibits the ability of renewable energy resources and
(2) the aggregation of resources for imbalance settlement
DERs to update their production/consumption schedules
purposes; and (3) the imbalance settlement pricing
in close-to-real time (see the previous discussion on
scheme. The design of each of these elements has
intraday markets).
implications for the efficiency of the others.
Imbalance pricing
Balancing responsibility and exemptions
In addition to determining balance responsibilities, the
Balancing responsibility refers to the extent to which
prices or charges that are applied to imbalances must
market agents are financially responsible for their
also be determined. Using the European terminology,
deviations (from their physically declared programs) in
there are two major approaches to imbalance pricing:
real-time operation. Balancing responsibility is enforced
so that resources have incentives to help maintain the • Single imbalance pricing. A single reserve energy price
physical balance of the electricity system and, secondarily, at a given location (zone/node) and time is used to
to maintain the financial balance of the electricity system. settle deviations from the market program. Negative
deviations pay for positive deviations and the result is a
In an effort to support renewable energy technologies, zero-sum mechanism for the system operator.
many countries permit renewable producers partial or full
• Dual imbalance pricing. Different prices are used to settle
exemption from balancing responsibility (in other words, positive and negative imbalances. In this case, positive
the costs associated with renewable energy imbalances deviations are not paid the full value of the injected
are socialized). In other cases, renewable energy energy, and negative deviations pay for more than their
resources are “balance responsible” but are subject to cost. This results in a non-zero-sum mechanism.
different, more favorable charges than other resources. In theory, single imbalance pricing provides optimal
The impact of these exceptions is relatively small with low incentives, since it means that a balancing responsible
penetrations of renewable resources. However, as these party (1) only profits if it deviates from its market program
technologies become widespread, differential treatment in favor of system requirements (whether the deviation is
leads to market inefficiency and should be eliminated so up or down) and (2) pays exactly the marginal cost of its
that all resources face similar market risks. In systems deviation when it deviates against system requirements.
where renewable energy resources are exposed to these In practice, however, system operators may prefer to
signals, the ability to forecast their future production avoid an incentive-based system and opt for a more
has significantly improved (see, for example, Herrero et command-and-control type of mechanism.20 In this
al. [2016]). The same reasoning applies to DERs, which, case, dual imbalance pricing may be used to penalize all
at the low-voltage level, usually do not have the same deviations from the market program to create incentives
balancing responsibilities as centralized resources. for agents to each keep a balanced program. In other
words, dual imbalance pricing
promotes more conservative
Renewable energy technologies and DERs should be exposed behavior and decreases the risk
to the same balancing requirements as other technologies. of suffering high imbalances, but
does so at the cost of distorted
market signals (thus giving rise to
associated inefficiencies).
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 249
Box 7.9: Wind Contribution to Security of Supply in Systems with Large
Storage Capabilities
In most thermal systems, scarcity conditions, though infrequent, can arise from a deficit of installed
capacity that translates into scant reserves during a few dispersed hours of very cold or very hot weather.
This is the situation that prevails, for example, in the eastern United States, in areas such as New York
or New England, where capacity mechanisms have been instituted. Resulting market prices reflect this
potential for scarcity conditions, as shown in the series of real-time prices in the New York ISO during the
polar vertex event of 2013–2014. The left graph in Figure 7.8 shows prices during this period of extreme
cold weather that reflect the scarcity of installed capacity, not of energy. In such cases, intermittency
results in a low capacity credit for wind or solar photovoltaic (see also footnote 23 later in this section).
In hydro-dominated systems with large storage capacity, as is the case in Colombia (with more than 60
percent hydro), rationing takes place due to lack of available energy, not capacity. This is also clearly
reflected in day-ahead market prices, as shown in the right graph of Figure 7.8. Day-ahead market prices
are very flat, since the regulation capability that comes with hydro resources gives much flexibility to the
unit commitment process. Thus, in this context, the intermittency of wind and solar is not an issue.
Figure 7.8: Spot Market Prices for Electricity in the NYISO and Colombian Systems
2500 900
800
2000 700
600
US$ /MWh
1500
$/KWh
500
400
1000
300
500 200
100
0 0
Jan 01 Jan 02 Jan 03 Jan 04 Jan 05 Jan 06 Jan 07 Jan 08 Jan Feb Mar Apr May Jun
2014 2016
NYISO RTM prices, first week of January 2014 Colombian exchange prices, first semester 2016
In some systems it can also be the case that the pattern of wind production is largely complementary
with that of key generation resources. Again, this is the case in Colombia, as illustrated in Figure 7.9,
which was provided by Colombia’s Mining and Energy Planning Unit (UPME 2015) in its detailed
planning reports.
200 6.5
150 6
50 5
Wind
0 0
2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 251
In capacity-constrained systems, as in Great Britain, response should be integrated in the wholesale market as
Germany, Italy, or (in the United States) PJM or New much as possible, with all associated consequences.25
England, scarcity conditions arise because there is not
enough installed capacity available to meet load at
every given moment during extreme weather events. In “Non-conventional” technologies (renewables,
these systems, the intermittency of renewable production storage, demand response, etc.), whether
is an issue.23
distributed or not, should be integrated in
7.3.1.2 Interaction of capacity mechanisms with capacity mechanisms on an equal footing with
other support mechanisms other technologies. If coherently designed,
The debate about whether or not to isolate developing this integration is fully compatible with other
technologies from market signals, and to what extent, has support mechanisms for these technologies.
been intense. However, the discussion has been limited to
the interaction of these new technologies with short-term
markets. The discussion now needs to be extended to Capacity mechanisms not only provide an investment
include increasingly widespread capacity mechanisms. incentive, they are also intended to provide market
First, capacity markets should allow “non-conventional” incentives for making committed resources available
energy resources of all types (renewable generation, during scarcity conditions, with the objective of
demand response, storage, etc., whether distributed or not) optimizing the efficiency of the system. If the design
to participate in any kind of capacity mechanism. At the of the capacity mechanism is robust and the reliability
same time, to the best of their capabilities, these resources product is technology-neutral, all technologies that
should be subject to the same conditions as any other participate in system reliability are acknowledged and
technology that participates in the capacity mechanism. exposed to an economic signal that prompts them
As discussed above, these conditions should be defined to to be available when the system most needs them.
ensure that resources contribute to system reliability.24 Non-conventional technologies should not be favored,
either by being exempted from penalties or by receiving
Nevertheless, regulators in many countries have so far
over-generous recognition for their firm energy/capacity
decided that renewable and demand response resources,
contributions.
which already benefit from other kinds of incentives (such
as grants, feed-in tariffs, targeted programs, etc.), are not Mechanisms to support renewable technologies are
eligible for remuneration through the capacity mechanism. not incompatible with this approach. As discussed in
This has been the case, for example, in Great Britain’s Section 7.3.2.2, the ideal renewable support scheme is
Capacity Market (DECC 2014). The regulatory practice one in which revenues obtained by renewable plants, from
of excluding DERs from these markets misses the key all markets in which they participate, including energy
point we are making here: that these technologies need and capacity markets, are deducted from the subsidy they
to be exposed to all market signals in order to maximize require to support their deployment.
the benefits they provide. Thus, renewables and demand
7.3.1.3 Capacity mechanism design elements and
distributed energy resources
23 In 2017/2018, the PJM capacity market (known as “RPM” for Reliability Pricing
Model) cleared 803 MW of wind power plants and 116 MW of solar resources
Applying the same rules to any technology that bids
in its base residual auction (PJM 2014). On average, wind and solar units were
assigned a 13 percent and 38 percent capacity factor, respectively, during the in an auction does not necessarily guarantee that all
qualification phase prior to the auction.
technologies actually compete on a level playing field.
24 This is already taking place in Brazil, as analyzed in Mastropietro et al. (2014),
where generation capacity mechanisms are a component of the income from
renewables. Renewable technologies started being cleared in specific renewable
auctions, but they ended up being selected also in conventional long-term auctions 25 As highlighted by Finon and Roques (2012), “a more comprehensive approach
for new electricity supply. As discussed in the next subsection, however, this based on a strong public governance has been proposed to deal with both
convergence is not yet complete and some rules still need to be harmonized. objectives of decarbonisation and system reliability through the same policy
instrument which would be a market-wide capacity forward auctioning.”
however it is designed, will be better suited to some can opt for multi-year contracts (up to
fifteen years for projects above a capital
technologies than to others.
expenditure threshold of £255/kW),
only one-year contracts are available for
The numerous design elements that need to be defined demand response (since it is deemed to
when implementing a capacity mechanism heavily be less capital intensive and can change more rapidly — for
condition which technologies will be better adapted example, by moving to another location). From the
to the mechanism. These design elements include, perspective of investors, the longer the duration of the
among many others, physical guarantees, financial contract, the better. Although it makes sense to try to
commitments, contract duration, lead-time, penalties for adapt contract duration to the different capital structures
non-compliance, etc. Different resources call for different of different technologies, we suggest that regulators should
contract conditions. For instance, a seven-year contract seek to avoid extremely lengthy contracts as much as
might work for a gas turbine, but may not be suitable for a reasonable. For example, due to the effect of discounting,
large hydro project. future income has less and less impact on decision-making
at the time of the auction, depending on how far out this
A combination of reliability products
income occurs from the present. Therefore, contract
One well-known option for solving the problem that durations of thirty years are seldom justified with the high
capacity mechanisms inevitably favor some technologies discount rates used by the generation sector in South
over others is to define different products in the America, for instance.
mechanism.26 Obviously this differentiation comes at a
Recently, the most controversial set of issues relates to
cost: The regulator needs to define a methodology to
the specific design of the firm capacity product (and
compare apples and oranges, since each technology
particularly to the obligations embedded in the product).
actually offers different products that are hard to
Many examples can be cited here. Walton (2016) points
compare. (For example, what is better suited for a certain
out that units selling the so-called Capacity Performance
system? A 10-year contract for 1000 MW of gas turbines
product in PJM are expected to be online year-round,
or a 20-year contract for 1000 MW of large hydro?) As a
meaning that solar generators or demand response
result, the methodology needed to clear the auction ends
programs focused on summer cooling loads would
up being rather complex, and efficiency becomes a matter
struggle to bid this product.27
of debate. Although there is no perfect solution for this
problem, some good practices should be considered. In many cases, the obvious solution to these matters is
aggregation, which can either be performed by the system
With the advent of new technologies that can contribute
operator or be simply left to market agents. In the first
to system reliability (e.g., biomass, geothermal, wind,
case, the regulator can define tailor-made products — for
solar, demand response, storage, etc.), existing challenges
example, defining seasonal products (such as summer
related to defining capacity products are exacerbated. As a
and winter) and then defining the most efficient way to
result, multiple controversies arise. Regarding the duration
clear a combined auction (some agents might bid for
of reliability contracts, for example, in Great Britain’s
only one of the products while others might bid for both).
capacity market, different contract lengths are available to
Alternatively, the regulator can design a product that
26 When the first capacity mechanisms were implemented back in the 1990s, implies a maximum number of hours of response for each
particularly in thermal systems (such as those in the eastern United States),
this problem of technology bias was initially not perceived as such. Regulators
provider and then manage different providers in the best
determined that system expansion was going to be led by just one or at most two way possible when scarcity is detected.
types of generation technologies with similar financial and cost structures (i.e.,
gas- and coal-fired plants). On the contrary, Latin American systems faced, from
the start, different alternatives: namely hydro and thermal plants. As a result, most
capacity mechanism designs included reliability products tailored to different 27 Other mechanisms, such as the one implemented in the New York ISO, hold winter
generation technologies, such as ten- to thirty-year contract terms and three- to and summer auctions separately.
seven-year project lead times (Batlle et al. 2015).
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 253
The market price as the efficient indicator of reliability issues. On the other hand, intraday and balancing
critical periods markets are also subject to price fluctuations caused by
In any case, our recommendation is to take the market sudden events (such as the outage of a nuclear plant or,
price as the indicator of a critical period in which firm in systems with high renewable penetration, a decline in
capacity is needed (in other words, use market price intermittent generation due to bad forecasting). These
to identify scarcity or rationing conditions). The spot, situations can provoke temporary generation scarcity
balancing, or real-time market price should be used as the even if total load is far from the peak and do so over a
critical period indicator and scarcity conditions should be time horizon that extends beyond the period covered by
defined as characterizing the period of time during which ancillary services. In this case, the system faces firmness
the reference market price exceeds a predetermined and flexibility issues. However, the selection of resources
strike price. should also take into account the signal that the capacity
mechanism is providing to the generation
mix. While all units are more or less
technically capable of producing if notified
A proper and transparent definition of the critical
one day ahead, certain technologies
period indicator that takes as a reference the price of (including baseload technologies, such
a sufficiently liquid short-term market provides good as coal power plants) would not be able
incentives for enhancing flexibility. to take part in the balancing market
because ramp constraints prevent them
from responding on such short notice.
Therefore, a capacity mechanism that uses
As discussed in Batlle et al. (2015), this approach should
the balancing market as the reference market provides
be increasingly valid in a future scenario that features
agents with a signal that discourages the installation of
increased elasticity of demand. In fact, as long as the
new baseload units. This may not be the objective or it
amount of completely inelastic demand in the market
may be precisely the objective, if the aim is to enhance the
(i.e., the demand that bids at the price cap) declines, it
installation of so-called flexible units. An alternative would
will become more and more difficult to define, in the long
be to define different products in the mechanism, but using
term, the demand that “must” be served. Consequently, it
a combination of reliability products creates problems
will also become more and more difficult to identify near-
that have been discussed above, since it unequivocally
rationing conditions using only a comparison between
implies that the regulator determines how much capacity
peak demand and generation available. Rather, any time
is to be installed of each sort of technology or group
that prices rise above the marginal cost of the most
of technologies. Again, the best approach is system-
costly generators can be considered a period of capacity
dependent and will need to account for different trade-offs.
limitation, in which resources committed in capacity
markets should be available. How to avoid mixing objectives
This critical period indicator obviously assumes the It is common in regulatory debates to find arguments
presence of a liquid reference short-term market in the against certain rules for capacity mechanisms that set
system. The selection of the reference market also affects barriers to the promotion of certain technologies. In
the kind of scarcity conditions that are covered, and that fact, rules should be defined in such a way as to provide
the regulator wants to have covered, by the capacity market agents with incentives to plan, and later operate,
mechanism. On the one hand, day-ahead markets are their investments to maximize system reliability — not to
only capable of capturing emergency situations related to promote one technology or another. Thus, the capacity
the combination of high loads (as on days of peak winter mechanism needs to require availability just when the
demand) and reduced supply (due to fuel constraints or system needs it, which is why the market price is the best
a dry year that limits hydro production) — that is, pure indicator of scarcity.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 255
heavily penalized), but this demand will not be required costs (wind and solar) have an incentive to keep running
to pay the corresponding tariff item. The second option even when prices are negative in order to access support
involves forcing the whole system demand to be covered payments. Also, reserve costs can increase unnecessarily
by the capacity mechanism, while at the same time if renewables are not exposed to balancing and regulation
allowing consumers to sell demand-response products market signals. As discussed in Section 7.2, this is because
in the capacity market. With this approach, demand the absence of market signals gives these generators no
response competes with other reliability providers in incentive to improve their production forecasts, which
the supply curve of the capacity mechanism (while in turn results in the need to commit larger quantities of
with the previous approach demand response is simply reserves. Finally, ill-designed subsidies for clean energy
removed from the demand curve) and is subject to the technologies can lead to long-term distortions in the
same contract provisions, including penalties for under composition of the generation fleet (see Box 7.10).
delivery. In this case, it is essential to properly define how
Promoting the integration of clean energy technologies
to calculate the customer baseline (Chao 2011), in order
in markets, while at the same time trying to reduce risks
to assess the performance of the demand resource during
for investors in these technologies to just the level that
scarcity conditions.
they can properly manage, calls for careful consideration
of design elements in support mechanisms for renewable
7.3.2 Support schemes for clean
technologies, as we discuss next.
energy technologies29
The key to a proper analysis of renewable technology
The interference of support mechanisms with market support schemes: Focusing on design elements instead
signals today of labels
From early efforts to implement support mechanisms, the Because labels for traditional renewable support policies
debate about whether or not to isolate renewables from (e.g., feed-in tariffs, feed-in premiums, auctions, etc.)
market price risk has been intense. In most systems, it are rarely applied in a standard or uniform manner and
are commonly subject to
misinterpretation, this study
The market penetration and maturity of renewable focuses on the design elements
technologies have grown to the point that these technologies of these policies instead
can now be reasonably expected to manage market risks like of resorting to traditional
other, more conventional generators. classifications.30 We begin by
highlighting three high-level
decisions in the design of any
type of support mechanism:
was assumed that the inefficiencies derived from reduced
wholesale market exposure were outweighed by the • Desired degree of technical and operational integration.
There are typically specific technical requirements for
benefits of reducing investment risk for these emerging
the grid connection of renewable technologies, such
technologies. As renewable energy technologies mature as effect on reactive control (IEA-RETD 2015). The
and reach higher levels of market penetration, however, most reasonable approach is to ensure that, to the
the system impacts of insulating renewables from market extent technically and economically feasible, renewable
signals are becoming more difficult to ignore, as they generation complies with the same grid standards as
lead to both short- and long-term distortions. First, some conventional generation.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 257
Most production-based incentive schemes provide 7.3.2.2 Capacity-based mechanisms
renewable generators with incentives to produce even
Capacity-based subsidies are intended to support
when market prices are negative. This can turn out to
investments on a per-installed-MW basis. In principle,
be highly inefficient for the system. Negative prices
capacity-based subsidies avoid market distortions by
have been a major concern in systems where high
decoupling payment and performance. These subsidies
levels of renewable generation are being supported with
can be implemented following different criteria (for
production-based mechanisms (Götz et al. 2014).
example, the per-installed-MW fixed subsidy can be
Alternatives exist to avoid this inefficiency even if the proportional to cost), depending on the generation
choice is to use production-based schemes. For example, technology and scale being targeted. The remainder of
if the production incentive is based on an increasing this section focuses on how to design a capacity-based
function of spot prices, the problem of negative prices mechanism for a technology that is subject to wholesale
disappears.32 (In that case, however, the pure marginal market signals.
price would still not be the guiding signal for short-
In cases where the supported technology perceives
term operation.) Other alternatives for minimizing the
market signals (and there is therefore an associated
inefficiency that results from negative prices include
market remuneration), these capacity- or investment-
capping total support payments in those periods with
based subsidies are intended to cover (on a
negative prices or directly banning negative bidding by
per-installed-MW basis) the difference between a plant’s
generation that receives regulatory support.
upfront investment cost and the present value of any
Production-based schemes and market compatibility expected market revenues after subtracting operating
costs, such that the project is likely to be profitable and
Production-based schemes always suffer from the
thus attractive to investors.
fundamental tension between creating revenue
certainty (which at this stage should not be seen as The cornerstone of these subsidization mechanisms is
such a particular problem, since the long-term risk an estimate of how much a project will earn or need to
position of investments in renewable projects does not pay by operating in various segments of the market (e.g.,
potentially including markets for
capacity and other services). In
Paying for production alters market price signals, leading to practice, this estimate can be
changes in generators’ operating behavior. This has a negative based on an ex ante forecast of
impact on the economic efficiency of the system. future market revenues for a given
renewable power plant over the
course of its life, or conversely,
compensation can be calculated
essentially differ from that of investments in conventional ex post, based on market results. The problem with ex
generation) and incentives for high-quality projects on post compensation based on a plant’s actual market
one hand, and limiting distortions on the other. performance is that compensation becomes unavoidably
“production-based” — and therefore causes distortions.
Production-based schemes create incentives to roughly
An alternative that avoids this problem is to define the ex
maximize total energy generation. However, as illustrated
post payment using a benchmark plant. The benchmark
in Box 7.10, projects with the highest capacity factors do
plant is a reference plant (synthetic or real) that would
not necessarily produce the highest-value energy.
ideally reflect an efficient and well-managed installation.
Market rents would be calculated for this reference plant
and additional support would be estimated based on the
32 For example, a percentage of the market price would result in a penalty for amount determined to be needed to make that plant break
producing during periods of negative prices.
Figure 7.10: Day-Ahead Market Prices As Well As Market and Premium Income for Solar
Production for Different Panel Orientations
100
DAM prices
80
Tilted South
700
Tilted West
60
$/MWh
500
Watts
40
300
20
100
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
even. This would provide greater investment certainty be profitable with the support provided, then clearly
without at the same time distorting short-term market additional rules will be necessary to achieve an efficient
signals. Furthermore, using real plants creates yardstick result. This can be tackled via minimum performance
competitive pressure for renewable power plants to requirements (e.g., minimum operating hours) or
outperform their competitors: Since all plants receive the minimum standards for component efficiency or system
same support payment, those plants that perform better design (although setting such standards is challenging in
(because they are more efficient or have better forecasting light of rapidly evolving technology).
abilities or a better sales strategy, etc.) will
earn higher returns overall.
Avoiding the problem of paying for nothing Capacity-based subsidies avoid market distortions
If a low-cost, low-performance facility by decoupling payment and performance. Using
(which typically is not the intended target reference plants can mitigate the risk of payments
of the support mechanism) turns out to flowing to low-quality projects.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 259
Auction-based mechanisms are needed to bid the Auctions, if workable, should be used in the procurement
percentage of the reference plant compensation process to secure the most competitive projects.33 Plants
Note that the use of a reference plant is compatible with would bid a percentage of the reference support payment
different mechanisms to determine final remuneration, for a number of years. The conditions of this support
including price- or quantity-based mechanisms (such should remain fixed over the duration of the contract
as auctions). For example, once the reference plant is to provide investors with greater revenue security. The
established, an auction could be used for price discovery, use of regular auctions will result in updated reference
where bids could consist of declaring the percentage of investment costs that reflect changes in technology
breakeven compensation needed by the reference plant cost and performance, market conditions, and access to
(for example, 80 percent or 110 percent). resource-rich locations for new projects (the best sites
should gradually fill up).
Reference plants can be defined using different criteria,
such as location and size, and they can change over time
to reflect technology changes.
The solution proposed allows for a prompt
7.3.2.3 Conclusions and recommendations on convergence with capacity mechanisms
the design of support mechanisms for clean and creates a natural competitive pressure
energy technologies
to optimize investment costs, siting,
If the policy objective is to support a particular renewable and production.
technology, the approach we advocate would function
as follows: Any new plant brought into operation would
first be assigned to a particular reference facility that The capacity-based support scheme outlined above has
corresponds to the same technology type (wind, solar, several notable features. First, it is highly compatible
etc.) and similar project size. with competitive electricity markets compared with a
A reference support payment methodology is calculated production-based scheme because it severs the link
for each reference facility based on the difference between production and payment: Only market signals
between the reference facility’s market revenues and the are left to dictate operating decisions. This compatibility
annuity of its investment cost, plus an adder to ensure has to do not only with the short-term energy market
a reasonable rate of return. If possible, reference plants but also with capacity mechanisms. Indeed, support for
should be based on actual installed plants (e.g., the clean energy technologies should already be starting to
median 10 percent by market revenue or investment cost converge with these mechanisms as the competitiveness
to better reflect actual bidding behavior). of renewable technologies is approaching (and in some
cases surpassing) that of conventional generation
technologies (Mastropietro et al. 2014). Second, the
approach we propose guards against the traditional risks
Efficiency considerations argue in favor of
of capacity-based schemes by combining performance
auctioned capacity-based support payments
requirements with natural incentives: Generators can
that make use of reference facilities to increase their profits by beating the reference facility (e.g.,
estimate market revenues through superior bidding strategies, better forecasting/
fewer imbalances, superior location for wind/solar
projects, etc.). This creates a natural competitive pressure
to optimize production and siting decisions. Third, the use
At the same time, auctioned capacity-based support payments should be the preferred alternative to
support renewables (and other immature technologies). One of the reasons to favor this approach is
that it would allow for prompt convergence with capacity mechanisms.
In systems that have chosen to implement capacity mechanisms and pursue a renewable penetration
objective, there is, at this stage, a straightforward way to make these two mechanisms converge: Add a
constraint in the capacity auction in such a way that a minimum amount of renewable investment would
be guaranteed.
A reliability product should be defined, aimed at acknowledging the actual contribution of each
technology to the future reliability of the system (for example, such a product would reward actual
production at times when the market price reveals that the system has reached a near rationing event).
All technologies in place should be exposed to the same commitment. If, for some reason, there are
higher-level policy objectives to assure a minimum penetration of one technology in particular, a
descending-clock auction could be easily designed to clear this amount. In that case, two different
prices would arise: one for the promoted technology (higher, if the constraint is activated) and another
price for the rest.
This solution presents numerous advantages: Once the auction is cleared, it exposes all technologies
to the same incentives. More importantly, it guarantees that once the learning curve of the promoted
technology improves sufficiently to allow the technology to fully compete with the others, this fact is
naturally revealed by the auction, as the constraint ends up being inactive.
of auctions introduces competition into the procurement support payments too often lead to market distortions
process, ensuring that investment costs are not inflated in and inefficient outcomes. The support schemes of the
an attempt to secure larger support payments. future will necessarily have to be more sophisticated
to spur investment without obscuring the benefits of
Properly designing all these sorts of schemes is fairly
competitive markets.
complex, especially compared with simpler production-
based mechanisms that have dominated support
schemes in many parts of the world for decades. But
this highlights an important point: Delivering subsidies
effectively is challenging. Simple and straightforward
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 261
References DECC. 2013. Electricity Market Reform: Capacity
Market — Detailed Design Proposals. Cm 8637. Department
Artelys. 2015. France-Germany Study: Energy Transition and of Energy & Climate Change. June 2013. www.gov.uk.
Capacity Mechanisms. Report for the Union Française de
l’Electricité and the German Association of Energy and DECC. 2014. Implementing Electricity Market Reform
(ERM) — Finalised Policy Positions for Implementation of
Water Industries. ufe-electricite.fr/.
EMR. Department of Energy & Climate Change. June 2014.
Batlle, C., C. Vázquez, and J. Barquín. 2007. “The www.gov.uk.
Balancing Market Maker: A Tool to Enhance Liquidity in
EFET. 2013. Design Principles for Capacity Mechanisms.
the French Balancing Mechanism.” The Electricity Journal
European Federation of Energy Traders. February 2013.
20 (6): 76-86. July 2007.
www.efet.org.
Batlle, C., P. Mastropietro, P. Rodilla, and J.I. Pérez Arriaga.
EIA. 2014. “Solar Photovoltaic Output Depends on
2015. “The System Adequacy Problem: Lessons Learned
Orientation, Tilt, and Tracking.” Today in Energy. U.S.
from the American Continent.” In Capacity Mechanisms in
Energy Information Administration. November 19, 2014.
the EU Energy Market: Law, Policy, and Economics: 119-139.
www.eia.gov.
Oxford, England: Oxford University Press.
European Commission. 2013. European Commission
Barroso, L. A. and C. Batlle. 2011. “Review of Support
Guidance for the Design of Renewables Support Schemes.
Schemes for Renewable Energy Sources in South
Commission Staff Working Document 439 final. Brussels.
America.” IAEE Third Quarter 2011: 27-31. International
November 2013. ec.europa.eu/.
Association for Energy Economics.
European Commission. 2014. “Guidelines on State Aid for
Bird, L., M. Milligan, and D. Lew. 2013. Integrating Variable
Environmental Protection and Energy 2014-2020.” Official
Renewable Energy: Challenges and Solutions. National
Journal of the European Union (2014/C 200/01). June
Renewable Energy Laboratory. NREL/TP-6A20-60451.
2014. eur-lex.europa.eu/.
September 2013. www.nrel.gov.
European Commission. 2016. Interim Report of the
Boot, P. and B. van Bree. 2010. A Zero Carbon European
Sector Inquiry on Capacity Mechanisms. Report from the
Power System in 2050: Proposal for a Policy Package.
Commission. Commission Staff Working Document 119
ECN-E — 10-041. Petten, Netherlands: Energy Research
final. Brussels, Belgium. April 13, 2016. ec.europa.eu/.
Centre of the Netherlands. April 2010.
FERC. 2013. Centralized Capacity Market Design Elements.
Butler, L. and K. Neuhoff. 2008. “Comparison of Feed-in
Commission Staff Report AD13 7 000. US Federal Energy
Tariff, Quota and Auction Mechanisms to Support Wind
Regulatory Commission. August 23, 2013. www.ferc.gov.
Power Development.” Renewable Energy (33): 1854–1867.
FERC. 2014. Operator-Initiated Commitments in RTO and
Cochran, J., L. Bird, J. Heeter, and D. Arent. 2014.
ISO Markets. Staff Analysis in Docket No. AD14-14-000.
Integrating Variable Renewable Energy in Electric Power
US Federal Energy Regulatory Commission. December
Markets: Best Practices from International Experience.
2014. www.ferc.gov.
National Renewable Energy Laboratory. NREL/TP-
6A00-53732. April 2012. www.nrel.gov. FERC. 2016. Settlement Intervals and Shortage Pricing in
Markets Operated by Regional Transmission Organizations
Couture, T.D., D. Jacobs, W. Rickerson, and V. Healey.
and Independent System Operators. Order No. 825, 155
2015. The Next Generation of Renewable Energy Policies:
FERC 61,276. US Federal Energy Regulatory Commission.
How Rapid Change Is Breaking Down Conventional Policy
June 16, 2016. www.ferc.gov.
Categories. National Renewable Energy Laboratory. NREL/
TP-7A40-63149. February 2015. www.nrel.gov.
CHAPTER 7: The Re-evolution of Short- and Long-Term Electricity Market Design 263
Neuhoff, K., J. Barquin, J.W. Bialek, R. Boyd, C.J. Dent, F. Usera I., P. Rodilla, and C. Batlle. 2016. “Regulatory
Echavarren, T. Grau, C. von Hirschhausen, B.F. Hobbs, F. Framework for Energy Storage Systems.” MIT Energy
Kunz, C. Nabe, G. Papaefthymiou, C. Weber, and H. Weigt. Initiative Utility of the Future Working Paper.
2013. “Renewable Electric Energy Integration: Quantifying
Walton, R. 2016. “New PJM Capacity Rules Challenge
the Value of Design of Markets for International
Demand Response Aggregators.” Utility Dive Newsletter.
Transmission Capacity.” Energy Economics 40: 760–772.
June 8, 2016. www.utilitydive.com.
Neuhoff, K., N. Ritter, A. Salah-Abou-El-Enien, and
Weissbach, T. and E. Welfonder. 2009. “High Frequency
P. Vassilopoulos. 2016. “Intraday Markets for Power:
Deviations within the European Power System: Origins
Discretizing the Continuous Trading?” DIW Discussion
and Proposals for Improvement.” In 2009 IEEE/PES Power
Paper 1544. doi:10.2139/ssrn.2723902.
Systems Conference and Exposition.
O’Neill, R., A. Castillo, B. Eldridge, R.B. Hytowitz.
2016. “Dual Pricing Algorithm in ISO Markets.” IEEE
Transactions on Power Systems PP(99): 1–1. doi:10.1109/
TPWRS.2016.2614891.
08
Understanding the Value of
Distributed Energy Resources
8.1 Introduction This framework is designed to create a level playing
field for all resources, centralized and distributed, to
Preceding chapters of this study have outlined a contribute to the cost-effective provision of electricity
framework for proactive reform designed to enable the services. It should also ensure that resources are properly
efficient evolution of the power system over the next compensated and charged for the services they provide
decade and beyond. The goal of this framework is to and consume, revealing when, where, and how these
facilitate the cost-effective integration of both distributed resources deliver value or drive costs in electric power
and centralized resources. The framework includes four systems. With this framework in place, this chapter
elements: (1) a comprehensive system of prices and focuses on understanding the value of distributed energy
regulated charges for electricity services that sends resources (DERs) and providing insights about the factors
sufficiently granular signals about the temporal and that are most likely to determine the portfolio of cost-
locational value of resources (Chapter 4); (2) improved effective resources, both centralized and distributed, in
regulation for distribution utilities that incentivizes different power systems.
integration of cost-effective alternatives to conventional
network investments, rewards utilities for cost savings Section 8.2 revisits the taxonomy of electricity services
and performance improvements, and encourages introduced in Chapter 2 and focuses on the ways in which
ongoing innovation (Chapter 5); (3) reconsideration of the value of some of these services changes depending
the structure of the electricity sector, with a focus on the on where these services are provided or consumed within
role of distribution utilities and on minimizing potential the power system. This focus on “locational value” is
conflicts of interest (Chapter 6); and (4) upgrades to key to understanding when, where, and how DERs can
wholesale electricity and ancillary service markets to provide potentially significant additional value relative to
remove unnecessary barriers to the participation of larger-scale centralized resources. Section 8.3 discusses
distributed resources (Chapter 7). the potential for unlocking contributions from existing
resources, including flexible and price-responsive
LOCATIONAL NONLOCATIONAL
• Energy
• Network capacity margin • Firm generation capacity
POWER SYSTEM VALUES • Power quality • Operating reserves1
• Reliability and resiliency • Price hedging
• Black-start
• Land value/impacts
• CO2 emissions mitigation
OTHER VALUES • Employment
• Energy security
• Premium values2
The value of other services is nonlocational — that is, the DERs can provide a range of services, including services
value of the service does not change based on where it with locational and nonlocational value. Due to their
is delivered in the power system. For example, operating distributed nature, DERs can be sited and operated12
reserves are deployed to contain frequency deviations in areas of the power system where the services they
that emerge as a result of imbalances between supply and provide are most valuable. Understanding the specific
demand due to forecast errors or unexpected failures of
1 The value of firm capacity and operating reserves may differ by location when frequent
power plants or transmission lines. Except at very short network constraints segment electricity networks and prevent delivery of capacity or
reserves to constrained locations. This leads to the designation of zonal requirements
time scales, frequency is consistent across an entire
for capacity and reserves in some jurisdictions. However, even in such cases, the value
synchronized interconnected system. As a result, the value for these services are uniform within these zones and thus remain constant across
wide geographic areas. As such, these services can be considered non-locational at
of frequency regulation is typically consistent across that least within broad segments of a given interconnected transmission system.
entire system. Similarly, the value of mitigating carbon 2 “Premium value” is a catch-all term that refers to the value that DER owners may
derive from factors that are inherent to DERs but that are not directly associated with
dioxide emissions depends on the social cost of carbon the electricity services that a given DER provides. For example, some consumers may
(EPA 2016) and does not change if the resource driving derive value from independently producing their own power or aligning their electricity
production with personal values (such as environmental attributes). In such cases,
emissions reductions is distributed or centralized. these “premium values” are private—that is, they accrue only to the individual(s) who
own the DER(s). Accordingly, these premium values should not be internalized by
public policies or electricity tariff design.
50.4%
8.4% 7.3%
2.3% 2.9%
0.1% 0.1% 0.9% 0.4% 0.4% 0.5%
<1 1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-80 81-90 91-100 >100
USD per MWh
5 Losses reported in Figure 8.2 include both technical losses due to resistance and
nontechnical losses from theft or other unmetered consumption. In some countries,
nontechnical losses account for a significant share of total losses—hence the larger
values in this figure may reflect significant nontechnical losses. Locational energy
value is affected only by marginal technical losses. Nontechnical losses should thus
be disregarded when considering the locational value of energy.
Total resistive losses rise quadratically with power flow as at times and locations of heavy network loading and high
described by Equation (1): marginal losses can thus capture higher locational value.
Note however that while DERs are typically assumed
(1)
to reduce losses, they can, in some cases, contribute
to losses in the distribution network by creating
Where Ploss is total resistive losses, I is electrical current, R reverse power flows across lower distribution voltages
and V are resistance and voltage for the network branch, (Schmalensee et al. 2015; Goop et al. 2016).
and P is instantaneous power. Thus marginal losses as a
function of power are the derivative of Equation (1) with
respect to power and rise linearly as power increases:
(2)
Losses also contribute to the locational value low-voltage networks increases. This example presents
of energy, although they generally have a results for a case study that employs load and solar PV
profiles consistent with a distribution network in the
smaller effect than network constraints.
Electric Reliability Council of Texas (ERCOT) system.
However, DERs that generate power (or reduce
Initially, the production-weighted average marginal losses
net consumption) at times and locations of avoided by each MWh of solar PV output ranged from
heavy network loading and high marginal approximately 6 percent to 19 percent in distribution
losses can capture higher locational value. networks with average annual losses ranging from
3 percent to 9 percent. In other words, by reducing
marginal distribution losses, each MWh produced by
In addition, because marginal losses decrease as the the first few solar PV systems located in the low-voltage
current or power flow through lines declines, the marginal portion of a distribution network in ERCOT is worth on
value of loss reduction falls as more DERs produce power average 6–19 percent more than the average MWh of
or reduce consumption during periods of high network solar PV produced at transmission voltage. However,
loading.6As an example, Figure 8.4 illustrates the decline as solar PV production increases in a given hour, the
in the locational value of distributed solar PV systems marginal losses in that hour steadily fall, reducing the
due to distribution loss reductions as the share of PV in marginal value of each additional unit of solar output.
6 As distribution losses are only rarely metered and reported, we estimate distribution
losses for Figure 8.3 by representing the distribution network as a single resistive
device. Using aggregate zonal load profiles from the independent system
operator for New York (NYISO) and the relationship between power and losses
in Equation (1), we establish the equivalent R/V2 coefficient that corresponds
to a given annual average technical distribution losses. In each case, we assume
transformer core losses represent a constant 0.5 percent of average load.
7 As with Figure 8.3, we estimate distribution losses for Figure 8.4 by representing
the distribution network as a single resistive device. Using aggregate zonal
load profiles from ERCOT and the relationship between power and losses in
Equation (1), we establish the equivalent R/V2 coefficient that corresponds to
a given annual figure for average technical distribution losses. In each case, we
assume transformer core losses represent a constant 0.5 percent of average load.
Solar PV profiles are from the National Renewable Energy Laboratory’s PVWatts
Calculator. We assume any hours with negative net load correspond to reverse
power flows and increase resistive losses. For more on the impact of distributed PV
systems on distribution network losses, see Schmalensee et al. (2015), Chapter 7.
In this case, the power system is segmented into five distinct voltage levels, with a simplified network
representation (see Appendix A for details). This case study resembles a Spain-like test system in
the 2020–2025 timeframe.9 Figure 8.5 shows the existing installed generation mix at each voltage
level in this system. All conventional generation (including large hydro) is connected at high-voltage
transmission and existing renewable sources are connected across a range of voltage levels. Least-
cost unit commitment and real-time dispatch of all resources are modeled using the ROM model
(see Appendix A for details), and LMPs for each voltage level are calculated using a direct current
power flow with quadratic losses. Effective resistances are estimated for each voltage level to closely
approximate reported annual losses in the real Spanish system, and load profiles are created for each
voltage level consistent with aggregate profiles for Spain.10
In each case, the penetration of either distributed or centralized solar PV is steadily increased, and the
effect on average energy losses and the market revenues of solar PV systems is calculated (assuming
all resources are compensated at LMPs, including prices at the distribution level). Smart charging of
EVs or energy storage devices may increase demand at low-price hours when the solar generation is
available, increasing marginal prices received by solar owners and reducing the prevalence of solar
generation curtailment (see Schmalensee et al. 2015). To illustrate this potential impact, we also model
cases assuming a fleet of 1.5 million EVs representing around 2 percent of total electricity demand and
connected at low voltage levels. In addition, once EVs are charged to fulfill mobility requirements (based
on EV parameters and mobility data from Banez-Chicharro et al. [2014]), we model EVs as capable of
providing energy back to the system in order to minimize total system costs.11 The total power capacity
of the EV fleet is equivalent to 3 percent of installed electricity-generating capacity in this system,
although not all EVs can dispatch simultaneously due to mobility requirement constraints.
8 Note that as EV manufacturers typically do not support discharging from EVs to the grid at present, these effects could also be generalized to reflect the impact of other forms of
energy storage at similar penetration levels.
9 The future generation scenario has been obtained from the Renewable Deployment Plan (Plan de Energía Renovables), approved by the Spanish government at the end of 2011
(IDAE 2011).
10 The Spanish regulator reports aggregated flows, losses, generation, and demand for an electric equivalent network for the whole country. Data are available at CNMC (2014).
11 The EV fleet dispatches to minimize total system costs, subject to constraints on battery state-of-charge and requirements that ensure individual vehicle mobility needs can be
met. A round-trip charge–discharge efficiency of 90 percent is considered, but battery degradation is not accounted for in this example. This example may therefore overestimate
the degree of discharge flexibility for EV fleets given current battery-life characteristics. If future technological improvement increases battery life, greater EV battery discharge
may be cost-effective, consistent with this case study.
As Figure 8.6 illustrates, increasing penetration of distributed solar PV at lower voltage levels can help
reduce total transmission and distribution losses by reducing power flows across transmission and
distribution lines. In contrast, if centralized solar PV is installed at high-voltage transmission nodes,
the effect on losses is negligible, as centralized PV displaces other generation at transmission voltage
level, with little effect on power flows. At high solar PV penetration levels (higher than 19 percent in
our case study), losses start to increase due mainly to reverse flows in distribution networks (in the
case of distributed PV) or replacement of distributed generation (in the case of centralized PV). Figure
8.6 also illustrates the same diminishing marginal effect of solar on losses discussed in Section 8.2.1.1.
Furthermore, EV charging increases demand during periods of high solar output, attenuating the
increase in losses due to reverse flows at higher penetration levels of distributed solar PV.
Figure 8.7 depicts average market revenues per MWh produced by both distributed and centralized PV
at increasing PV penetration levels, as well as the impact of smart EV charging on solar revenues.12 At
low PV penetration levels, distributed solar earns greater market revenue than centralized solar due to
the increase in locational value associated with avoided energy losses (i.e., higher LMPs at low-voltage
nodes). However, as PV penetration increases, marginal losses decline during the periods when solar
PV generates power, reducing the locational value of distributed PV. At higher PV penetration levels
(around 10 percent in Figure 8.7), the locational value of distributed solar starts to become negative
(i.e., lower LMPs at low-voltage nodes than at high-voltage nodes) and consequently market revenues
of centralized solar become greater than those of distributed solar. Smart EV charging, by shifting stored
energy, increases consumption at PV production hours and thus results in higher prices at those hours,
thereby increasing PV market revenues for both distributed and centralized solar.
12 Once again, we assume that compensation for all resources reflects LMPs at each voltage level, including distribution. In practice, actual compensation may differ, depending on
applicable market rules and tariff designs.
Figure 8.7: Annual Market Revenue per MWh for Solar PV Systems at Increasing Penetration
(Spain-like Test Case Example)
then DERs can substitute partially or fully for Matpower (Zimmerman et al. 2011) to calculate optimal power flows subject
to alternating current power flow and network constraints. Curtailable loads (or
conventional investments in transmission equivalently, distributed generators) are simulated at each load point, and the
objective of the model is to minimize the total net load reduction necessary to
and/or distribution infrastructure. render power flows feasible under increasing levels of aggregate network load.
Increments in aggregate network load are assigned in the following way: First, the
network is divided into zones corresponding to each feeder (i.e., any aggregation
of consuming nodes for which there is only a single connection point to a MV-LV
substation). Then, for each zone/feeder, specific node or bus values are assigned
in proportion to the peak load values of those specific buses in the pre-existing
network. This is best illustrated with a simple example. Suppose that a specific
network zone contains two consumers, A and B. In the pre-existing network,
13 In practice, transmission and distribution network operators apply engineering consumers A and B had peak loads of 60 kilowatts (kW) and 40 kW, respectively,
safety margins to all network components to ensure that technical limits are and therefore a total peak load of 100 kW. Now suppose that this zone has been
not exceeded. If DERs can be reliably dispatched or counted on to respond to assigned a new peak load value that is equal to 1.1x the pre-existing peak load value,
pricing signals, operators may relax these margins and rely on DER behavior to or 100 kW x 1.1 = 110 kW. In this case, since customer A’s peak load in the pre-
avoid violating network constraints. This would allow network components to existing network accounted for 60 percent of that zone’s total peak load, customer
be operated more closely to true technical limits and would allow the network to A’s new peak load is 0.6 x 110 kW = 66 kW, whereas customer B’s peak load is 0.4 x
increase its capacity margin without a physical upgrade of network components. 110 kW = 44 kW.
See Section 2.2.1.3 for further discussion.
16 If DERs or load curtailment occurs at suboptimal locations, greater total reductions
14 Note that DERs must be capable of providing these services with sufficient in aggregate peak load are necessary than depicted in Figure 8.8. The figure thus
reliability that, in aggregate, operation of DERs can substitute for network presents an optimal case reflecting a lower bound on the required change in
investments without degrading quality of service. This may require performance aggregate peak net withdrawals to accommodate a given increment of peak load
contracts or other measures to guarantee sufficient performance and quality. growth.
Figure 8.9: Optimal Location and Magnitude of Net Withdrawal Reductions Necessary to Accommodate Aggregate
Peak Demand Increases Without Network Upgrades in a Representative European Urban Network
17 Areas of the network with aging assets due for replacement could also afford
opportunities for network capacity deferral value.
0.9
0.8
0.7
0.6
0.5
0 50 100 150 200 250 300 350 400 450 500
Hours
At the same time, as load grows, network constraints highest peak demand), 36 hours exceed a load factor of
will become binding more frequently, requiring DER 0.8 and 157 hours exceed 0.7, and so on. In other words,
dispatch in a greater number of hours of the year as well. accommodating each marginal increment of load growth
This dynamic can be illustrated by examining the load without network upgrades requires both more MW
duration curve for a given portion of the network. For and more hours of reductions in peak withdrawals. This
example, Figure 8.10 depicts the 500 hours of highest steadily reduces the marginal value of employing DERs
load for an average low-voltage aggregate load profile for network capacity deferral.18
in Spain. As this figure illustrates, the number of hours
that exceed a given threshold load factor (defined as the
hourly load as a per unit share of the annual peak load)
steadily increases as the threshold falls. So while only
9 hours exceed a load factor of 0.9 (or 90 percent of
To address this challenge, UK Power Networks partnered with Smarter Grid Solutions to deploy an
Active Network Management (ANM) system in an area of the network spanning 70 square kilometers
(Smarter Grid Solutions n.d.). Through this program, UK Power Networks offers generators “managed”
connections—that is, connections that allow the utility to engage in real-time control of DG in response
to prevailing grid conditions. As part of the connection offer, UK Power Networks provides applicants
with an estimate of how often their generation will be curtailed to manage network constraints. These
estimates help developers with long-term planning and typically show a very low anticipated reduction
in annual energy export volumes. In addition to managing DG output, the ANM system coordinates a
range of smart devices such as Automatic Voltage Control relays, Dynamic Line Rating relays, and a
19 Also known as a phase angle regulating transformer or phase-shifting transformer, quadrature boosters are specialized transformers that control power flow along lines
downstream of the transformer by shifting the phase angle at the head of one of the lines. Quadrature boosters provide a means to relieve network constraints by redirecting
power flows from heavily loaded lines to more favorable paths.
Figure 8.11: Histogram of 2014 System Average Interruption Duration Index Numbers for US Utilities
Table 8.2: Estimated average reliability value for different customer classes in the United States and average
cumulative annual outage durations21
LARGE COMMERCIAL
RESIDENTIAL SMALL COMMERCIAL
CUMULATIVE ANNUAL & INDUSTRIAL
(PER CUSTOMER (PER CUSTOMER
OUTAGE DURATION (PER CUSTOMER
PER YEAR) PER YEAR)
PER YEAR)
Finally, deployment of DERs is commonly justified with both cases, we assume the solar systems are price takers
appeal to local employment or job creation benefits. and do not affect the marginal value of each service. If
These employment benefits may be considered locational solar PV penetration increases, the marginal value of solar
as well, since the jobs associated with deploying DERs PV services may decline (as discussed in Section 8.2.1).
(i.e., construction, installation, etc.) would be created In addition, quantitative results are case-specific and
within the local region where the DERs are installed. are meant to be illustrative rather than generalizable to
Analyzing the employment impacts of distributed other power system contexts. We calculate the locational
difficult to incorporate all appropriate short- and long- • A distributed PV system at a specific location in the
term equilibrium impacts on overall employment. For power system has an incremental locational energy
example, jobs benefits associated with DERs are often value equal to the difference between the LMP at the
simply a welfare transfer or loss, as the opportunity PV system’s location and the price at the location
costs of paying more for DERs translate to employment of the marginal generator in the NYISO system. We
losses in other sectors of the economy that now see therefore calculate the production-weighted average
lower expenditures (NYSERDA 2011). A full and accurate locational energy value for the two zones using
accounting would consider the general equilibrium historical 2015 hourly day-ahead NYISO market
implications of energy investment decisions on overall prices and rooftop solar PV production profiles for
employment and welfare. In other words, just because Islip (Long Island) and Monticello (Mohawk Valley),
a particular technology is more labor-intensive to install New York. This yields an incremental value of $24.0/
doesn’t mean it is necessarily a net job creator in the MWh for Long Island and $2.3/MWh for Mohawk
economy as a whole (Tsuchida et al. 2015). Thus, the Valley, relative to the energy value at the system-
distributional benefits or costs of DER deployment for a wide marginal generator.
particular locality are difficult to judge.
22 NYISO only publicly reports zonal (as opposed to nodal) electricity prices.
locational values for CVR of $11.1/MWh in Long 24 NYISO defines “Summer Peak Hours” as the hours beginning 14, 15, 16, and 17
during the three-month period from June 1 through August 31, inclusive. “Winter
Island and $1.7/MWh in Mohawk Valley. Peak Hours” are defined as the hours beginning 16, 17, 18, and 19 during the three-
month period from December 1 through the last day of the immediately following
• Similarly, employing the highest and most typical February. The capacity value of solar PV is defined as the average capacity factor
during these peak hours. See NYISO 2016, “Installed Capacity Manual,” Version
distribution network investment deferral values 6.32, New York Independent System Operator, February 2016, Section 4.5.1.
from Cohen, Kauzzman, and Callaway (2016), we 25 If PV systems are paired with energy storage and configured to safely island during
outages, their reliability value could be higher. In that case, however, much of the
assume $60/kW-yr of investment deferral value for reliability value could be appropriately attributed to the storage system, which
could charge from grid-connected power and provide backup supply without
the Long Island case and $0/kW-yr for the Mohawk necessarily being co-located with the PV system. The incremental reliability value
Valley case. In other words, we assume that PV of the solar system in this case would be equivalent to the extended duration of
self-sufficient supply due to the ability to recharge the storage system during
output in the Long Island case is coincident with islanded operation.
$100
$80
$60
11.1
$40
5.6
24.0
$20
$0
Locational Locational Conservation Network Generation Reliability Total locational
energy value: energy value: voltage reduction investment capacity premium value
transmission distribution deferral
losses
Figure 8.13: Locational Value of Distributed Solar PV — Mohawk Valley, New York (Average-Value Example)
$10
Average Value per MWh Produced
$6
3.1
$4
2.3
$2
$0
Locational Locational Conservation Network Generation Reliability Total locational
energy value: energy value: voltage reduction investment capacity premium value
transmission distribution losses deferral
vehicle fleets
Electric vehicle (EV) adoption is expected to steadily
increase over the next decade. Projections vary, but
one source estimates that EV market share may reach
approximately 20 percent of new vehicle sales worldwide
by 2030 (BNEF 2016). Such estimates suggest that
EV charging could account for a significant increase in
electricity demand. Enabling smart charging of EVs is
therefore key to reducing the impact of this new class of
demand on power system costs.
In particular, different EV charging strategies can have Source: Mateo, Frías, and Miralles (2016)
a substantial effect on the cost of distribution networks
as well as on the marginal cost of energy and generating Figure 8.16: Illustration of EV Charging Strategies on Top
capacity. Mateo, Frías, and Miralles (2016) use the of Energy Consumed by Medium- (MV) and Low-Voltage
Reference Network Model (RNM) to analyze the effect (LV) Consumers
of charging strategies on five large-scale distribution
networks (see Appendix A for a description of the
RNM). They consider three scenarios for increasing EV
penetration in Spain based on policy scenarios developed
by Hassett and Bower (2011) (Figure 8.15).33 The authors
consider four categories of vehicle size: quatricycle (L7e),
passenger vehicle (M1), goods-carrying vehicles with
a maximum laden mass of 3,500 kilograms (N1), and
goods-carrying vehicles with a maximum laden mass
of 12,000 kilograms (N2). The authors also distinguish
between three specific types of EV technology: plug-in
battery electric vehicles (PBEVs), which have no energy
source besides the battery; plug-in hybrid electric Source: Mateo, Frías, and Miralles (2016)
vehicles (PHEVs), which use a combustion engine after EVs are modeled under three different charging strategies
their batteries are depleted; and plug-in extended-range (smart, valley, and peak). Figure 8.16 illustrates EV
electric vehicles (PEREVs), which use a combustion charging demand in each hour as the difference between
engine to provide electrical power and overcome range the dots and bars, where the bars reflect other non-EV
limitations. Additional assumptions about specific electricity demand in each hour. The “smart charging”
connection times, energy consumption for each type of strategy is responsive to price signals or dispatch signals
electric vehicle, and distance traveled by each type of that reflect local distribution network constraints and
vehicle are based on Hassett and Bower (2011) and Ball, is designed to minimize impact on distribution network
Keers, and Alexander (2010).34 costs. The “valley charging” strategy charges vehicles on
33 EVs are assigned to each of the five representative test networks in proportion to
the aggregate load in each network.
34 See Mateo, Frías, and Miralles (2016) for a detailed description of technical
parameters for different types and sizes of EVs.
The value of utilizing smart inverters is, again, dependent output than load. In contrast, because PV output is
on where in the network distributed resources are sufficiently matched to load elsewhere in the network, the
located. For example, Figure 8.19 illustrates the locational presence of smart inverters yields no value for reactive
marginal price of reactive power for one hour in an urban power absorption at these locations.
distribution network with PV capacity equal to 10 percent
of peak load. Reactive power absorption is particularly
valuable in one region of the network with greater PV
8.4 Economies of Scale and possible cost declines through 2025.37 As Figure 8.20
illustrates, smaller-scale solar systems have higher annual
Distributed Opportunity Costs costs than larger systems due to economies of unit scale in
Many technologies suitable for distributed deployment, installation, system component costs, and maintenance.38
including solar PV, electrochemical energy storage, and Failure to exhaust economies of unit scale for solar PV
fuel cells, can be deployed across a range of scales. systems therefore results in incremental unit costs relative
Each of these resources also exhibits varying degrees to larger-scale systems.
of economies of unit scale, or declining costs per unit of
capacity as the size of the system increases.
Resources that can be deployed at multiple
For example, solar PV systems can be deployed at the
kilowatt (kW) scale on residential rooftops, at a scale of
scales, such as solar PV and energy storage,
several hundred kW to several MW on commercial or exhibit economies of unit scale—that is,
industrial rooftops or in ground-mounted arrays, or at declining costs per unit of capacity as the
a scale of tens to hundreds of MW in so-called “utility- size of the system increases. By failing to
scale” solar farms. Figure 8.20 depicts the estimated exhaust economies of unit scale, smaller-scale
annual cost of ownership for PV systems of varying scales
deployment of these resources therefore
in the United States, including annuitized capital costs35
results in incremental unit costs relative to
and fixed operations and maintenance (O&M) costs.
Estimates are given for costs in 201536 and for a range of
larger-scale systems.
37 Projected cost reductions from 2015 to 2025 assume a 25, 50, and 66 percent
decline in installed cost and a 0, 25, and 50 percent decline in annual O&M costs
35 Per Lazard (2015a), the capital annuity assumes a 7.68 percent after-tax weighted across our high, medium, and low cost estimates, respectively. For comparison,
average cost of capital (8 percent cost of debt, 12 percent cost of equity, 40 percent NREL (2016) projects cost reductions through 2025 ranging from 4 to 65 percent
total business tax rate, 40 percent equity share) and 25-year asset life. for installed costs and 0 to 53 percent across solar systems of various sizes. Note
that economies of unit scale depend significantly on installation and balance-
36 The cost shown for 10–100 MW systems in 2015 is from Lazard (2015a). The costs of-system costs; in some countries (e.g., Germany, Australia), installed costs for
shown for 1–2 MW and 1–10 kW systems are based on average economies of unit smaller-scale systems can be lower than in the United States.
scale across estimates from Lazard (2015a) and NREL (2016).
38 Note that interconnection costs or transmission extensions needed to access
locations suitable for utility-scale solar projects can result in greater variation
in the cost of larger-scale solar projects. The figures shown here assume typical
interconnection costs and assume no transmission expansion.
The rate at which economies of scale are exhausted varies cases, after factoring in both costs and benefits, smaller
significantly from technology to technology. Traditional is not better. Indeed, when incremental unit costs exceed
nuclear, hydroelectric, natural gas, and coal-fired power locational value, society incurs “distributed opportunity
plants are typically installed at scales of several hundred costs” when small-scale DERs are deployed in lieu of more
MW to more than a thousand MW, thereby maximizing cost-effective resources.
cost reductions due to economies of unit scale. In
To illustrate this trade-off, consider our case study of
contrast, economies of scale for solar PV units appear
locational value for distributed solar PV systems in
to be exhausted at the scale of tens to hundreds of MW.
New York (presented in Section 8.2). The high-value
Electrochemical energy storage systems tend to exhibit
case in Long Island, New York, produced nearly $85 per
even more rapidly diminishing economies of unit scale, as
MWh in locational value, while the average-value case
depicted in Figure 8.21.39
in the Mohawk Valley produced roughly $8 per MWh in
In any case, economies of scale matter, even for distributed locational value. However, solar PV systems deployed in
resources. For resources that can be deployed at multiple a distributed fashion to capture this additional locational
scales, incremental unit costs must be considered value also incur incremental unit costs. Figure 8.22
alongside possible increases in locational value associated compares the locational value for both the Mohawk
with smaller-scale, distributed installation of these Valley and Long Island cases to incremental unit costs for
resources. Trade-offs between locational value on the distributed PV systems at the kW and MW scales. The
one hand and incremental unit costs due to economies figure includes both estimated costs for 2015 and a range
of unit scale on the other hand can help identify the ideal of possible cost declines through 2025. To facilitate this
locations and applications for these resources. In many comparison and account for annual energy production,
we present incremental unit costs as the difference
39 Per Lazard (2015a), the capital annuity assumes 7.68 percent after-tax weighted
average cost of capital (8 percent cost of debt, 12 percent cost of equity, 40 percent
total business tax rate, 40 percent equity share) and 10-year asset life. Estimates
for 2015 cost, economies of unit scale, and 2025 cost declines are based on a
range of estimates in Lazard (2015b) and on personal correspondence with GTM
Research. Projected cost reductions for 2025 assume 50, 66, and 75 percent
declines in installed cost and 0, 25, and 50 percent declines in annual O&M costs
across high, medium, and low cost estimates, respectively.
in levelized cost per MWh40 between distributed In contrast, the much greater locational value in the high-
PV systems in Long Island and Mohawk Valley, New value case we constructed for Long Island outweighs the
York, and a 100 MW, utility-scale, fixed-tilt solar PV incremental unit costs for MW-scale distributed solar
system with an equivalent solar insolation quality.41 As systems at estimated 2015 and projected 2025 costs. At
Figure 8.21 illustrates, incremental costs for both kW- estimated 2015 costs and at the high end of projected
and MW-scale distributed solar PV systems exceed the 2025 costs, smaller, kW-scale rooftop systems would still
relatively modest locational benefits in the typical value entail a distributed opportunity cost in the Long Island
case we constructed for Mohawk Valley. This holds for case, as incremental unit costs still exceed locational
both 2015 estimated costs and all 2025 cost projections. value. However, incremental unit cost for rooftop-scale
In short, locational benefits are not sufficient in the PV systems falls below the locational value for the
illustrative Mohawk Valley case to justify deploying solar medium- and low-range 2025 cost projections. In the
PV at a distributed scale. Larger, utility-scale solar would high-value Long Island case, distributed solar PV is thus
deliver greater net benefits in this case, while small-scale economically justified, although the greatest net benefit
PV deployment incurs distributed opportunity costs. would accrue to the largest-scale PV system capable of
capturing the locational value in that part of the network.
In other words, even at locations where distributed
deployment is economically justified, exhausting
economies of unit scale to the greatest extent possible in
40 Note that levelized costs are inadequate to compare technologies with very
different production patterns or locations (e.g., comparing solar PV systems to
that setting yields the greatest economic benefits.
combined cycle gas plants or nuclear generators) because the value of these
resources can differ substantially. In this case, where we are comparing solar PV
systems with very similar production profiles and location, and explicitly factoring
in differences in value, levelized cost is an appropriate metric.
41 Solar PV production profiles from NREL PVWatts Calculator assuming fixed-tilt
systems at all scales. Annual production for Long Island location is 1,458 MWh per
MW installed and 1,376 MWh for Mohawk Valley. See note 40 above for cost and
discount rate assumptions. Levelized cost of electricity is calculated using a formula
from NREL (2016). Note that utility-scale solar systems with one-axis tracking can
exhibit even lower levelized costs than the fixed-tilt system used for comparison here.
locational value and incremental unit costs due is likely to be inefficient in many locations. At specific sites,
such as in portions of the network that are experiencing
to economies of unit scale can help identify
frequent congestions or in areas that are confronting
the ideal locations and applications for these
rapid growth in electricity demand, locational value can
resources. In many cases, after factoring be significant and can give distributed solar or storage
in both costs and benefits, small-scale technologies an economic advantage over larger-scale
deployment of DERs can incur “distributed deployment of these resources. On the other hand, in
opportunity costs.” Smaller is not always better. well-developed networks that experience congestions and
service interruptions only infrequently, locational value
can be modest, as the Mohawk Valley case illustrates. In
This example illustrates important trade-offs between such locations, economies of unit scale generally outweigh
locational value and economies of unit scale. While the locational value given current and projected costs for solar
values and distributed opportunity costs presented here PV and electrochemical energy storage systems. In these
are case-specific, the basic heuristic can be broadly settings, innovation in the underlying technologies and/
applied to develop insight about both the value and cost or learning-by-doing to reduce system installation costs
of distributed deployment of solar PV, energy storage, and may be needed to flatten economies of unit scale before
other resources that can be deployed at multiple scales. widespread distributed deployment of these resources
becomes economically efficient.
To fully evaluate these complex trade-offs and interactions in a long-term context, new models must
be employed that capture the salient dynamics. GenX, a comprehensive electricity resource capacity
expansion model, is described more extensively in Appendix A. It incorporates state-of-the-art
formulations that capture and optimize trade-offs between economies of unit scale and locational value
for a variety of DERs and conventional resources. GenX employs a simplified zonal representation of
electricity networks with electricity resources available for deployment at one or more transmission
and distribution voltage levels, each exhibiting different unit costs, with power flows and constraints
between zones (Figure 8.23). Using mathematical formulas that capture in sufficient detail the results
of extensive offline modeling of distribution network power flows, the GenX model accounts for the
impact of DERs and electricity demand on distribution network losses and reinforcement costs as well
as transmission-level losses and constraints or reinforcements. The model also captures operational
constraints on generators, operating reserve requirements, and opportunities for flexible demand or
price-responsive demand curtailment.
HV HV HV
MV MV MV
LV LV LV
We steadily increase the cost of transmission network reinforcements to explore trade-offs between
distributed storage with greater locational value and MW-scale storage with lower unit costs. (Figure 8.24).
As transmission network costs increase, the optimal location for energy storage shifts from transmission
voltage at the 25 MW scale to medium-voltage distribution at the 100 kW scale, while expansion of
transmission capacity also declines. The 100 kW system exhibits the lowest unit cost for a system that
can be sited where it relieves the transmission constraint and defers network investments. As such, the
model never selects the more costly 5 kW storage option for inclusion in the least-cost resource portfolio.
Distributed solar PV (at 5–10 kW and 100 kW–1 MW scales) is also an option, but it is not selected by
the model in this case, because PV production is non-coincident with peak demand in this test system
and thus cannot contribute to transmission network deferral. Furthermore, the optimal duration of energy
storage in this case is two hours, with a small amount of four-hour storage appearing only when annuitized
transmission reinforcement costs exceed $180,000 per MW-year.43 Shorter duration storage thus appears
sufficient to reduce peak demand during periods of binding transmission constraints, thereby avoiding
the need for reinforcements—at least given this specific load profile. In addition, as total storage capacity
increases modestly across the cases, solar PV capacity increases, reflecting the complementary nature of
these resources. Likewise, the capacity of natural gas open-cycle and combined-cycle units declines slightly
as storage and PV capacity increases, indicating that solar and storage are at least partial substitutes for
gas-fired resources in this context.
These results are meant to be purely illustrative and to demonstrate the kind of system-wide analysis
that is important to develop insights about the role of DERs in the power systems of the future and to
explore when and how DERs can add value to power systems.
42 Hourly demand profiles are based on data from the Spanish regulatory commission (CNMC 2014) and scaled to a peak demand of 100 GW. For this test system, we assume the
following existing capacity mix: 5,000 MW of open cycle gas turbines, 20,000 MW each of combined cycle gas turbines, pulverized coal, and nuclear reactors; 30,000 MW of
onshore wind; and 10,000 MW of solar PV. Energy storage and solar PV costs are the mid-range 2025 costs reported in Figure 8.20 and Figure 8.21. Fuel price assumptions are
mid-range assumptions for Europe in 2025 from IEA (2015), and capital and operating cost estimates for other resources are from NREL (2016).
43 In this case, 100 kW-scale lithium-ion storage systems with a 2:1 energy-to-power ratio have installed system costs of $724/kW ($362/kWh), while 100 kW-scale systems with a
4:1 energy-to-power ratio have installed costs of $1,305/kW ($326/kWh). With a lower installed cost per kW of rated power capacity, the model thus favors the shorter-duration
storage except in cases when larger energy capacity is significantly more valuable.
Commission and Australian Energy Regulator Working Energy Resources.” IEEE Power Energy Magazine 6: 30-40.
Ball, E.B.R., N. Keers, and M. Alexander. 2010. EIA. 2016. Electric power sales, revenue, and energy
Mobile Energy Resources in Grids of Electricity. MERGE efficiency Form EIA-861 detailed data files. Washington,
Project Deliverable D2.1. Janaury 4, 2010. www. DC: US Energy Information Administration. www.eia.gov/
documents/20140203_154622_76425_Deliverable_2.1_ EIA. 2014. Price Elasticities for Energy Use in Buildings of the
Modelling_Electric_Storage_devices_for_Electric_ United States. Washington, DC: US Energy Information
Vehicles.pdf. Administration. October 22, 2014.
Banez-Chicharro, F., J.M. Latorre, and A. Ramos. EIA. 2015. Annual Energy Outlook 2015. Washington, DC:
2014. “Smart Charging Profiles for Electric Vehicles.” US Energy Information Administration. April 2015.
Computational Management Science 11 (1–2): 87–110.
doi:10.1007/s10287-013-0180-8. EPA. 2016. The Social Cost of Carbon. Washington, DC:
US Environmental Protection Agency. September 29,
Blum, D. and L.K. Norford. 2014. “Dynamic Simulation 2016. www.epa.gov/climatechange/social-cost-carbon.
and Analysis of Ancillary Service Demand Response Accessed October 28, 2016.
Strategies for Variable Air Volume HVAC Systems.”
HVAC&R Research 20(8): 908-921. November 2014. EnerData. 2015. Energy Efficiency Indicators: Rate of
Electricity T&D Losses, 2014. www.wec-indicators.
Campbell, R.J. 2012. Weather-Related Power Outages and enerdata.eu/world-rate-of-electricity-T-D-losses.html.
Electric System Resiliency. Washington, D.C.: Congressional Accessed October 17, 2016.
Research Service. doi:R42696.
EPRI. 2015. Alternatives to the 15% Rule: Modeling and
CNMC. 2014. “Methodology for Computing of Hosting Capacity Analysis of 16 Feeders. Palo Alto, CA:
Transmission and Distribution Charges” (in Spanish). Electric Power Research Institute.
Comisión Nacional de los Mercados y la Competencia.
www.cnmc.es/Portals/0/Ficheros/Energia/Circulares/ Georgiopoulos, S. n.d. Interruptible Connections:
09
A Toolkit for Regulators and
Policy Makers
Throughout this study we offer recommendations for practitioners. Chapters 4 through 8 contain detailed
proactive regulatory, policy, and market reform designed analyses, modeling results, and theoretical underpinnings
to enable the efficient evolution of the power system for each of the recommendations made herein; where
over the next decade and beyond. The purpose of our more detail is desired on the recommendations or on
recommendations is to enable all energy resources, questions of practical implementation, readers should
whether distributed or centralized, to participate in the refer to previous chapters in this report.
efficient provision of electricity services while achieving
The task facing those responsible for the reliable and
other public policy objectives. The overarching framework
cost-effective planning, regulation, and operation of
for these recommendations has two interrelated pillars:
future power systems is daunting; the sheer length and
Establish a comprehensive system of prices and regulated
weight of this study seem to confirm that. However,
charges that applies to all network users, and remove
many of the recommendations outlined in this chapter
inefficient barriers to the integration and competition
and throughout the report are implementable and can
of distributed resources and centralized resources alike.
deliver significant benefits today. Throughout the report
Those barriers include inadequate remuneration schemes
we pay careful attention to the practical implementation
for distribution utilities, power sector structures that may
challenges of our recommendations and, where possible,
impede fair competition, and wholesale electricity market
we highlight examples of their real-world implementation
design flaws.
and outline steps that bridge theory and practice. These
This chapter collects our core policy and regulatory discussions are intended to enable proactive policy and
recommendations for establishing efficient prices regulatory reform of the power sector; proactive reform
and charges and for removing inefficient barriers. It is the only defense against being caught flat-footed
is intended to be used as a quick reference for policy by challenges that seem minor today, but could seem
makers, market designers, regulators, and other insurmountable tomorrow.
Recommendation 4: Find the efficient level The first measure to be adopted is not about enhancing
of granularity of prices and charges for each granularity, but about reducing the distortion caused
power system given its characteristics and by charging policy costs (e.g., for efficiency programs,
regulatory context. subsidies for renewable energy, etc.), taxes, and residual
network costs (i.e., network costs that are not recovered
The prices and charges for electricity services may
via cost-reflective charges) within the volumetric,
vary significantly with space and time when computed
dollar-per-kilowatt-hour ($/kWh) component of the
according to the most precise cost-reflective methods.
tariff, as is today the practice in most power systems.
However, this does not necessarily imply that regulators
Any policy costs, taxes, and residual network costs
must adopt prices and charges with the maximum level
that are not directly affected by changes in electricity
of granularity in order to achieve a satisfactory level
consumption or injection should be removed from the
of efficiency. Implementing more granular prices and
volumetric ($/kWh) component of the tariff and charged
charges involves costs. Thus, temporal and locational
in a manner that minimizes distortions of cost-reflective
granularity should be increased only as long as the
prices and charges for electricity services.
incremental efficiency gains exceed the incremental
implementation costs. Complexity and equity concerns
related to implementation (to be discussed later) are
relevant factors that should also be considered.
network users. Rebates to equalize average to make trade-offs between capital expenditures (CAPEX)
and operational expenditures (OPEX), including contracts
charges and mechanisms to hedge month-
with or payments to DERs that are capable of contributing
to-month bill volatility are some of the
to more efficient network operation. Dealing with these
instruments that can be used to properly new uncertainties and challenges in an efficient manner
address these concerns. will require utilities to become more innovative; however,
today’s regulations lack incentives for innovative behavior.
Finally, as the power system becomes more digitalized
and as DERs proliferate, distribution systems will face new
9.2 Recommendations for cyber threats that current regulations are not adequately
prepared to manage.
Removing Barriers
Powerful regulatory tools exist to manage these changes.
Efficient pricing alone is not sufficient to ensure the
These regulatory tools can be divided into two categories:
cost-effective development of the power sector over
(1) improved approaches to distribution remuneration
the coming decade and beyond. Establishing sound
that can account for new cost drivers, new demands
economic pricing is a critical step in enabling all resources
and uses of the distribution network, and increased
to compete efficiently to provide the electricity services
uncertainty; and (2) additional incentive mechanisms
required in the power system. However, many inefficient
for achieving specific outcomes that are not captured
barriers exist today that prevent DERs from participating
by improved distribution remuneration, including
on an equal footing with centralized resources and
performance and quality of service improvements
vice-versa. These barriers can be grouped into three
and long-term innovation. Our key regulatory
categories: (1) inadequate regulation of distribution
recommendations are summarized here; a detailed
networks for the presence of DERs; (2) power sector
discussion can be found in Chapter 5.
structures that lead to conflicts of interest between the
network and the commercial components of distribution Recommendation 11: Implement cost efficiency
companies; and (3) biased or inadequate wholesale incentives for distribution utilities.
market rules.6
Our first recommendation is to more closely align the
business incentives of the distribution company with the
remuneration structure, starting with a forward-looking
revenue trajectory. Regulators can better align utility
incentives for cost-saving investments and operations and
ensure that the benefits of improved utility performance
are shared between utility shareholders and ratepayers
by adopting state-of-the-art regulatory mechanisms. The
distribution system services are critical to prevent conflicts service providers and aggregators.
Closer to real time, all resources should participate in Market clearing and pricing mechanisms will become
balancing and real-time markets for the management of more complex as refinements in bidding formats
their energy imbalances. Most system operators allow are introduced. An outstanding debate is whether
large market players to self-manage their imbalances discriminatory side payments (uplifts) should be used
(netting positive and negative deviations of several power in addition to uniform market prices, and if so, how
plants within the same portfolio) and settle only their to allocate their cost. Different approaches create
net deviations in the balancing market. Liquidity and significantly different price signals, which affect both
transparency in balancing markets would improve if this the short- and long-term decisions of increasingly
kind of aggregation were avoided. responsive market agents. To avoid discrimination against
demand resources, the allocation of side payments
Recommendation 23: Adapt auction rules to (when they cannot be avoided) should be incorporated
incorporate new operational constraints and in the market-clearing procedure in a way that ensures
new resources. revenue sufficiency for all market agents (whether on the
In addition to the constraints that are imposed by the generation side or the demand side).
network, electricity generators are subject to internal
physical constraints (e.g., start-up,
minimum power output, ramping
constraints) that have a significant Market operators should adapt the definition of services
impact on their operation and cost. and the bidding formats, with an adequate spatial and
New resources (including demand- temporal granularity, to incorporate new operational
constraints and new resources.
tools used in this study consider the technical and scheduling of resources from an end-user or system
economic characteristics of electricity systems, as well perspective. In these types of models, investment
as the interactions among various parts of the system. decisions are fixed, and the objective is to determine
These tools aim to quantify the economic trade-offs how resources would be utilized (e.g., when generation
between providing electricity services from centralized units would be scheduled and dispatched, when
versus distributed energy resources and to evaluate the storage facilities would be charged or discharged, etc.)
impact of sectoral trends and changes (technology cost to minimize end-user costs and/or to minimize total
declines, changing regulatory environments, new energy system costs. Other models focus on the investment
and environmental policies, etc.) on the efficient mix of and planning decisions that would minimize costs in the
these resources. medium to long term. These types of models tend to use
more simplified representations of the operation of the
APPENDIX A: A Description of the Computational Models Used in the Utility of the Future Study 325
system. In general, a single model cannot represent in full Electricity network modeling is also a core aspect of
detail the complexity of both operation and investment the Utility of the Future study as it allows us to quantify
decisions for both networks and energy resources (this the mutual impacts of different elements connected
limitation of electricity system models is sometimes at various parts of the network, either at the level of
referred to as the “short blanket” problem1). transmission or of distribution voltage. The network
characteristics condition the methodologies used to
Modeling the technical capabilities of DERs in detail
model network effects. For instance, transmission
is challenging due to the complexity of the underlying
networks usually have a tractable number of nodes that
physical processes as well as to limitations in
can be individually included in a unit commitment or a
computational power. Examples of the former include the
security-constrained economic dispatch model. Since
thermodynamic performance of thermal storage systems
the focus of this study is on distribution, we only use
and the degradation characteristics of electrochemical
simplified representations of the transmission network in
storage technologies. These processes are complex
using the Reliability and Operational Model (ROM) and
but can have significant impacts on DER costs, on the
the Optimal Electricity Generation Expansion (GenX)
services that DERs are capable of providing, and on
model, which facilitates inclusion of distribution network
the value that these resources can capture in markets.
features at an abstracted but salient level of detail.
Furthermore, single DER technologies are often capable of
providing multiple electricity services, either one at a time Distribution networks have a complex structure with
or simultaneously with certain restrictions. Therefore, millions of connection points and assets. They have a
detailed and sufficiently accurate representations of DER meshed topology in high voltage, and a radial topology in
technology characteristics are essential for determining medium and low voltages. The Reference Network Model
what services might be provided by these technologies (RNM) allowed us to design large-scale networks that
and at what costs. The most relevant parameters that resembled what an efficient distribution company would
affect technology performance can be obtained from build. We were then able to use those networks to run
detailed, technology-specific models and integrated power flow and optimal power flow simulations. RNM
into less detailed models that represent a variety of DER also allowed us to do distribution network expansion
technologies. These less-detailed models can then be planning studies under different load growth and DER
used to estimate how different DERs might compete with penetration scenarios.
one another for the provision of electricity services, a task
By combining the capabilities of RNM, DR DRE, and
that would be computationally difficult with more detailed
Matpower, the Distributed Energy Systems Simulator
representations. The Utility of the Future study utilizes the
(D-Sim) model is able to calculate the economically
Distributed Energy Resources Consumer Adoption Model
rational DER investment and operation decisions made
(DER-CAM)2 and Demand Response and Distributed
by several thousand network users coordinated by price
Resources Economics (DR DRE) model to analyze the
signals. D-Sim can also calculate the impact of these
DER technical capabilities and end-user responses to
decisions on the operation of the distribution network.
price signals.
1 If you pull a short blanket up to cover your head, you cannot also cover your feet—
and vice versa.
2 DER-CAM was developed by the Lawrence Berkeley National Laboratory. This
model has been used in the Utility of the Future study through a collaboration
agreement. Further description of this model can be found in building-microgrid.lbl.
gov/projects/der-cam.
APPENDIX A: A Description of the Computational Models Used in the Utility of the Future Study 327
Figure A.2: Schematic Representation of Demand Response and Distributed Resources Economics Model
3 NREL’s PVWatts Calculator estimates the energy production and cost of energy of
grid-connected photovoltaic (PV) energy systems. It is available at pvwatts.nrel.gov.
APPENDIX A: A Description of the Computational Models Used in the Utility of the Future Study 329
Figure A.3: Distribution System Used in the Distributed Energy Systems Simulator
This example of a distribution system used in the A.3.5 Experiments run in D-Sim for the
Distributed Energy Systems Simulator includes the Utility of the Future study
distribution network itself and models of price-responsive
• Competitiveness of DER investments under different
network users connected to each node. The network users
price signals (Chapter 4.5)
can react to expected future price signals by changing their
short-term behavior and by adopting DER technologies. • Sensitivity of the economic performance of a given DER
investment given other DER investments and network
reinforcements (Chapter 4.4)
A.3.3 Main D-Sim inputs
• Calculation of locational marginal pricing at distribution
• A map image and aggregate statistics for the distribution
level and network cost recovery (Chapter 4.4)
area: Load density, distribution of network user types, and
electrical equipment characteristics are included
• Definition of network user types: Responsive users are A.4 Reference Network Model
represented through a combination of manageable load
and fixed, active, power-consumption profiles for Large-Scale Network
Distribution Planning
A.3.4 Main D-Sim outputs
• Hourly data: for nodal prices (locational marginal pricing A.4.1 Description
at distribution level), nodal voltages, congestions, and
The Reference Network Model (RNM) is a very
losses per line
large-scale planning tool developed by Instituto de
• Yearly data: for energy payments by network users, Investigación Tecnológica—Comillas (Mateo et al. 2011;
network charges by user, total remuneration of the Gómez et al. 2013) that plans the electrical distribution
distribution network owner, as well as identification of
network starting from the geographic coordinates and
efficient network reinforcements
profiles of network users. It designs networks comprising
• Optimal DER investments: from a private point of view several voltage levels from low to high and includes
(either centrally coordinated or not), assuming that the
plans for both substations and feeders. The planning
users are price-takers
A.4.2 Methodology
There are two versions of the RNM: Greenfield and
Brownfield. The Greenfield RNM builds the network from
scratch, starting with network users and moving up to
transmission substations. The Brownfield RNM expands
Figure A.4: Example of Rural Network Built by the Reference Network Model
APPENDIX A: A Description of the Computational Models Used in the Utility of the Future Study 331
A.4.3 Main RNM inputs A.4.5 Experiments run in RNM for the
• Customers, DERs, and supply HV substations: locational Utility of the Future study
data (rectangular coordinates), power demand, installed • Impact of specific DERs (e.g., PV, storage, demand
capacity, and daily load or generation profiles response, and electric vehicles) and a combination of
• Library of standard installations: substations, DERs on distribution network costs and investment
transformers, electrical lines, switching and (Chapter 8.3.2)
protection equipment • Design of network charges considering different networks
• Technical and economic planning parameters: and categories of network users (Chapter 4.4.4)
continuity of supply targets, demand increase rate, loss • Implications of DER penetration on the remuneration of
factors, simultaneity coefficients distribution system operators (Chapter 5.2.2)
• Geographic data and constraints: orography, lakes,
nature reserves
A.5 Reliability and Operational
A.4.4 Main RNM outputs Model with Distribution Network
• Graphical results: feeder layout, substation locations, Representation
customers, and DER connections
Medium
voltage
Figure A.5 shows different possible ways to transform periods, which occur monthly or even less frequently.
voltage levels, reflecting the complexity of real Therefore, different samples are needed to compute the
distribution networks. For instance, urban networks EEN parameters.
usually have direct transformation from high-voltage
transmission levels to medium-voltage distribution levels A.5.4 Computation of
(as shown at the left-hand side of Figure A.5), whereas equivalent resistances
in rural areas transformation typically happens at several
The EEN represented in Figure A.5 has a radial
distribution voltage levels (as represented in the right-
configuration. Since parallel flows are nonexistent, and a
hand side of Figure A.5). Each branch of the network
DC power flow approach is being used, it is not necessary
represented in Figure A.5 can aggregate many thousands
to calculate an equivalent network reactance. In the DC
of networks with similar topology.
power flow formulation, reactive power flows and voltage
drops are also neglected. A quadratic approximation for
A.5.3 Main EEN inputs
energy losses is considered for the calculation of the
An EEN can be computed based on available measured equivalent network. Some of the ROM model results are
data from distribution companies. The following data are operation costs, locational marginal prices, and reliability
required at each voltage level: generation, aggregated indexes. This model has been employed in research
demand, energy flows, and network energy losses. These projects funded by the European Commission, such as
data may only be available for specific time periods. MERGE (2011) and TWENTIES (2013).
For example, measures are usually taken during tariff
APPENDIX A: A Description of the Computational Models Used in the Utility of the Future Study 333
Some of the main characteristics of ROM are: A.5.5 Experiments run in ROM for the
• a day-ahead unit scheduling optimization followed by a Utility of the Future study
sequential hourly simulation in a one-year period;
• Effect of energy losses on marginal prices for each
• Monte Carlo simulation of many yearly scenarios to voltage level (Chapter 3.2.3 and Chapter 4.4.1.3)
deal with the stochasticity of demand and intermittent
• Effect on locational marginal pricing of a mix of DERs
renewable generation; and
connected at LV—e.g., PV plus storage and backup
• detailed operation constraints regarding unit generation (Chapter 8.2.1.1)
commitment and dispatch.
• Effect of prices and charges on the dispatch of DERs (e.g.,
In addition, when management of hydro resources and electric vehicles, solar PV, demand response, and backup
generation) connected at low voltages (Chapter 4.5)
seasonal pumped storage exceeds the analysis time
frame, the hydro operation can be computed by another
higher–level model and taken as an input in ROM.
HV HV HV
MV MV MV
LV LV LV
APPENDIX A: A Description of the Computational Models Used in the Utility of the Future Study 335
further relaxes the problem and allows commitment and solar PV. Other possible central station resources
of fractions of a plant. Both options introduce modest include geothermal, biomass, solar thermal, pumped
abstraction errors but significantly improve computational hydro storage, and possible thermal storage for solar
performance, enabling greater detail in other features, thermal or nuclear units. Eligible DERs include: solar
such as network complexity. Since on/off decisions for PV, electrochemical storage, thermal storage, flexible
individual DERs and even OCGTs are fast and occur in small demands, and batteries.
increments, representing them as continuous decisions is
Power flows between zones and voltage levels are
also a minor abstraction.
modeled as simple transport flows. Maximum power
Capacity investment and operational decisions are flows across these interfaces capture key network
indexed across each node or zone in the system, enabling constraints. Future versions could include nodal network
the model to select the optimal location of capacity representation in the meshed portion of the system with
investments and operations in each location. Thus DC power flow constraints, although this dramatically
the model balances the different economies of scale increases the dimensionality of the problem.
at different voltage levels on the one hand, with the
Losses are a function of power flows between voltage
differential impacts or benefits of location at different
levels or zones, implemented as a piecewise linear
zones or voltage levels on the other hand—a key
approximation of quadratic resistive losses. Losses due
advantage over other models.
to power flows within zones are related to injections
Eligible central station generation resources include: and withdrawals by means of several hundred power
combined cycle gas turbines, open cycle gas flow simulations in realistic networks, as we show in
combustion turbines, pulverized coal, nuclear, wind, Figure A.8 for a semi-urban network.
APPENDIX A: A Description of the Computational Models Used in the Utility of the Future Study 337
References Silverstein, A. 2011. “Transmission 101.” Presented at
the NCEP Transmission Technologies Workshop. www.
De Sisternes, F. and M.D. Webster. 2013. “Optimal naruc.org/grants/Documents/Silverstein%20NCEP%20
Selection of Sample Weeks for Approximating the Net Load T-101%200420111.pdf.
in Generation Planning Problems.” ESD Working Paper
Series 2013-03. Cambridge, MA: Massachusetts Institute Tabors, R., A. Rudkevich, and R. Hornby. 2014. “Rate
Mateo, C., T. Gomez, A. Sánchez-Miralles, J. Peco, and A. Reneses. 2014. “Economic Benefits of Active Demand
Candela. 2011. “A Reference Network Model for Large- for Stakeholders.” Project: ADVANCED/WP6 – T6.3/
Scale Distribution Planning with Automatic Street Map D6.3. Funded by European Commission within “FP7 -
190–97. doi:10.1109/TPWRS.2010.2052077. Van den Bergh, K., E. Delarue, and W. D’haeseleer. 2014.
MERGE Project. 2011. “Functional Specification for Tools “DC Power Flow in Unit Commitment Models.” TME
to Assess Steady State and Dynamic Behavior Impacts, Working Paper-Energy and Environment. www.mech.
MIT. 2015. The Future of Solar Energy. Cambridge, MA: and W.L. Kling. 2006. “Usefulness of DC Power Flow
Massachusetts Institute of Technology. mitei.mit.edu/ for Active Power Flow Analysis with Flow Controlling
adaptation, incumbent utilities1 risk falling into a a novel empirical review and analysis of the business
“death spiral” that threatens their financial viability models for three of the most widely deployed DERs: solar
(PWC 2015; Kind 2014). photovoltaics, electricity and thermal storage, and demand
response. We define the key “value capture” and “value
1990 or
4 9% 7% 3 6% 6% 6 12% 9%
Earlier
1991-1995 1 2% 3% 1 2% 2% 2 4% 5%
5 Note that despite the name, this category also includes “behind-the-meter” energy
storage companies, as discussed in Section.
Africa 0 0% 0% 0 0% 1% 1 2% 1%
Asia
0 0% 4% 4 8% 8% 4 8% 9%
Pacific
Central
and South 0 0% 0% 0 0% 0% 0 0% 0%
America
Europe
10 22% 22% 12 25% 24% 19 38% 38%
and Israel
Middle
0 0% 0% 0 0% 0% 0 0% 1%
East
North
36 78% 74% 32 67% 67% 26 52% 52%
America
EMS
PROVIDERS
Non-Electricity
Services
Secondary
Frequency Control
MARKET-BASED
CAPACITY & RESERVE DR
Firm Capacity
& Secondary UTILITY-BASED
Service(s)
Firm Capacity
Customer Segment(s)
Re
, Fe ven
c ity aint MARKET-BASED CAPACITY e u
pa tr S ub s on e St
e nt: Ca Cons AND RESERVE DR sc U rea
m rm , r ipt tilit m
P ay r Fi rves ion y R s:
o Fe eve
r v ice nts f Rese ided es nu
e e g ov es
S ym tin Pr ,
a
P e tio ra n
Information Streams:
Op itiga Information Streams:
M Energy Usage, Availability,
Bids, Dispatch Signals
Control Signals
LOAD RESOURCE:
ISO/TSO/RTO RESIDENTIAL, C/I/M, OR INDUSTRIAL
Service Provision:
Load Modulation in
Response to Control Signals
Monetary Flow
Electricity Service Flow
Information Flow
Businesses within this archetype most commonly target (Dehghanpour and Afsharnia 2015). Secondary reserves
large commercial, institutional, or municipal (C/I/M) and are favored among DR providers, as secondary reserves
industrial customer segments. This is due to a number are dispatched less frequently and the required response
of factors, including market rules such as minimum time is typically lower than it is for primary reserves.
bid-size requirements, transaction costs, and customer
These businesses typically make a profit by taking a
acquisition costs (Behrangrad 2015). These businesses
portion of the revenues generated from the sales of these
often provide an EMS (or similar product) to the targeted
services—i.e., by brokering market revenues (brokerage
customer to optimize the customer’s energy consumption
fees)—and/or by charging for the use of the energy
(and production in the case of customers with on-site
management software that enables the demand control
distributed generation); the EMS also enables customers
(subscription fees). Note that the business model’s
to participate in DR programs offered by an independent
revenue is a brokerage fee, rather than a commodity sale;
system operator (ISO), with market interaction facilitated
the business distributes the revenues associated with
by the DR business. In certain cases, the businesses do
commodity sales to the DR resources under contract.
not provide an EMS-like product, and loads are simply
controlled through alternative measures (e.g., phone Table B.3 includes several examples of business models
calls instructing customers to manually respond). that fall under this archetype. EnerNOC and REstore
These business models most commonly leverage are typical examples of businesses selling to C/I/M
customer loads such as lighting; heating, ventilation, and industrial customers. Ohmconnect is an interesting
and air conditioning (HVAC) units (chillers and fans); exception in that it explicitly targets residential customers
refrigeration units; other variable-frequency drive units; through a user-friendly app. By sending signals to
idiosyncratic industrial process loads; and customer-sited homeowners directly or communicating with home
generation such as backup diesel or gas units, fuel cells, area network – connected devices, Ohmconnect enables
or batteries (Palensky and Dietrich 2011; Hansen et al. residential consumers to participate in power system
2014). Among the various types of operating reserves, DR markets. Encycle targets mostly small and midsize
is most commonly deployed for secondary reserves (i.e., C/I/M facilities, which are commonly overlooked among
contingency reserves); however, the provision of primary DR businesses.
reserve services is becoming increasingly common
B.4.2 Utility-based capacity and reserve In these cases, distribution utilities seek an explicit
s, Re
rvice UTILITY-BASED CAPACITY Fe ven
Se es ue
: ity AND RESERVE DR on St
ms tric Ut rea
t rea Elec ilit m
y R s:
S r s ev
ue fo ee en
ven nts on F ues
e
Re ym ript i
Pa bsc Information Streams: Information Streams:
Su Service Requirements, Energy Usage, Availability,
Dispatch Signals, Availability Control Signals
LOAD RESOURCE:
REGULATED UTILITY RESIDENTIAL, C/I/M, OR INDUSTRIAL
Service Provision:
Load Modulation in
Response to Control Signals
Monetary Flow
Electricity Service Flow
Information Flow
brokerage fees (i.e., keeping a share of the revenue earned An exception to the above model is the “behavioral”
from the sale of the DR resource to the utility). Many of model. Businesses that use this model tend not to provide
the businesses operating in market environments also explicit control or dispatch signals, but rather provide
operate in these regulated environments; examples include “nudges” and targeted incentives to create a response
EnerNOC and Comverge. (Allcott 2011; Gillingham and Palmer 2014). These
businesses sell their services directly to the regulated
In market-driven environments, transaction and customer
utility and do not engage with the consumer outside of
acquisition costs have driven DR business models to
the context of the behavioral program. The revenue model
target larger C/I/M and industrial customers, yet the
is typically based upon subscription fees and shared
regulated environment allows for greater participation of
savings (brokerage fees) charged to the utility. Examples
residential loads. However, the technical requirements
of business models of this type include Opower and
of coordinating very fast responses from residential
Tendril. Table B.4 includes several examples of business
loads has limited the majority of the business models in
models within this archetype.
the archetype to providing only capacity and secondary
reserves (Mathieu et al. 2012). The most common loads
that are used by residential DR companies are HVAC
units. Examples of business models of this type include
Comverge and EcoFactor. Nest, a residential smart
thermostat provider, offers a similar service through its
Rush Hour Rewards program.
Information Streams:
Energy Usage, Control Signals CUSTOMER:
EMS PROVIDER RESIDENTIAL, C/I/M,
Revenue Stream:
OR INDUSTRIAL
Subscription, Brokerage (Shared Savings, Etc.)
Service Provision:
Minimization of Electricity Service Costs
Monetary Flow
Non-Electricity Service
Information Flow
B.5 Electricity and Thermal Energy storage technologies are diverse in their uses
(Ruester et al. 2012; Schoff 2015). However, despite
Storage Business Model such diversity, the related business models can be
Archetypes clustered into three major archetypes. Figure B.5 shows
the business model landscape for electrical and thermal
Electricity and thermal storage technologies6 are
storage. Note that business models that deploy energy
often lauded as critical components of a clean energy
storage technologies in conjunction with solar PV are
future. It is therefore not surprising that energy storage
discussed in the solar PV section that follows.
deployments have been rapidly increasing (DOE 2015).
Pumped hydro energy storage and molten salt thermal
storage account for the vast majority of installed energy
storage capacity to date, but these technologies are poorly
suited to distributed applications (DOE 2015). Lead-acid
technologies make up the bulk of distributed energy
storage installations globally, although lithium-ion (Li-ion)
and other advanced technologies are gaining traction
(Agnew and Dargusch 2015; Schmalensee et al. 2015).
6 Note that thermal storage technologies in this case exclude the storage of thermal
energy in the heated or cooled space of a building, as this is categorized as demand
response.
REVENUE STREAM(S)
Non-Electricity
Services Asset Sale
Asset Sale & Lending/
Renting/Leasing
Primary Frequency Lending/Renting/
Control & Voltage Control Leasing & Brokerage Fees
Lending/Renting/
Service(s)
Firm Capacity 13
1
Customer Segment(s)
From Figure B.5 we see three major (non-manufacturer) B.5.1 Energy storage for network services
archetypes. The primary defining feature of each
Numerous studies have highlighted the value of
archetype is its level of integration within power system
energy storage technologies for network and system
operations. A number of business models are focused
applications, including various network capacity and
on the provision of ICT-based optimization and control
ancillary service benefits (Denholm et al. 2013; EPRI 2013;
services for energy storage technologies. Some of these
Akhil et al. 2013). Certain states have begun legislatively
businesses actively deploy projects and provide energy
requiring utilities to procure storage assets; for example,
services to power system operators, while others simply
California’s Assembly Bill 2514 requires the state’s largest
provide optimization and control products to other
utilities to procure 1.3 gigawatts of storage by 2020
businesses or end users. Business models that fall into
(California State Assembly 2010). A cluster of business
the latter category are classified, along with technology
models that we call “energy storage for network services”
manufactures, as providing non-electricity services.
has emerged to meet this market. Figure B.6 shows
the general structure of this business model archetype;
dotted lines are used between the financing function
and the electrical and thermal storage (ETS) resource
management/deployment function to indicate that this
function could be performed internally or by partners that
are external to the business.
Monetary Flow
Electricity Service Flow
Information Flow
Table B.6: Network Service Business Models for Electricity and Thermal Storage
The majority of the businesses in this archetype are either B.5.2 End-user optimization for
technology-agnostic project developers, project developers energy storage
with outsourced manufacturing but proprietary technology,
End-user systems involve installing storage assets
or technology developers with downstream integrated
“behind the meter” at customer sites. These behind-
project development arms. Businesses in this archetype
the-meter systems have historically been deployed at
tend to either serve vertically integrated regulated utilities
customer facilities to manage peak demand charges and
or system operators, or they install batteries at industrial
arbitrage between low and high energy price hours under
sites with the intent of providing network services. These
time-of-use or real-time pricing tariffs (actions that are
businesses tend to earn revenues either from the sale
collectively referred to as “bill management”) (Neubauer
or financing of the storage assets or from the sale of
and Simpson 2015; Masiello and Roberts 2014). These
electricity services (typically firm capacity and operating
customers may be eligible to participate in bulk system
reserves valued at market prices).
markets but may desire not to. Alternatively, these
customers may be in regions that do not allow distributed
assets to participate in markets. The general structure for
this group of companies, a business model that we call
“energy storage for end-user-optimization,” is depicted
in Figure B.7.
ELECTRICITY
SERVICE PROVIDER:
REGULATED UTILITY,
LOAD-SERVING ENTITY
InformationStreams: Electricity
Energy Usage, Service Payments:
Electricity Services Payments for Energy
Price Signals Networks, etc.
Monetary Flow
Non-Electricity Service
Information Flow
To date, the primary motive for the deployment of Business models within this archetype typically earn
residential energy storage systems has been to increase revenues either through a form of shared or guaranteed
the profitability of solar PV systems through increasing savings arrangements7—e.g., brokerage fees—or through
“self-consumption” (i.e., minimizing the export of the sale and financing of storage assets (St. John 2016).
energy produced on site) (Masiello and Roberts 2014; Shared or guaranteed savings arrangements are less
Hoppmann et al. 2014; Johann and Madlener 2014). For common at the residential level. Industry analysts expect
commercial and industrial customers, the primary motive that the financing options that contributed to the rapid
has been the avoidance of demand-based consumption rise of distributed solar (discussed below) will have the
charges (i.e., charges per kilowatt of peak demand) same effect on energy storage (Wesoff 2015).
(Neubauer and Simpson 2015). As technology costs have
fallen, providing backup power to residential customers
and critical commercial and industrial loads has also
emerged as a driver (Nair and Garimella 2010).
Table B.7: Energy Storage for End-User Optimization Business Model Examples
Sonnen (Europe)
In response to these emerging market-based Finally, there exists a small subset of companies that are
opportunities, businesses are attempting to deploy attempting to provide operating reserves to system operators
storage technologies behind the customer meter that through the use of thermal storage in bricks and water
simultaneously attempt to lower costs to the end heaters. Given the large potential for the use of residential
user and participate in power system markets. These water heaters, these businesses have targeted not only
businesses most commonly attempt to connect C/I/M commercial and industrial loads but also residential loads.
and industrial customers with ISO markets, providing These companies tend to operate on a brokerage-fee basis
firm capacity, operating reserves, and mitigating network similar to their larger storage counterparts.
constraints. These businesses tend to earn revenue on the
sales of the storage assets and/or through brokerage fees
on market-based revenues (e.g., fees levied for managing
market interaction on behalf of the battery host). Certain
Figure B.8: Business Model Structure for Generic End-User and System Co-optimization
Monetary Flow
Electricity Service Flow
Information Flow
TECHNOLOGY
Non-Electricity
MANUFACTURERS
Services
Capacity &
Operating Reserves
SOLAR-PLUS-STORAGE
END-USER OPTIMIZATION
Energy & Firm
Capacity
UTILITY-SCALE PV
FINANCIERS & INTEGRATORS
Energy COMMUNITY
SOLAR PROVIDERS
DISTRIBUTED PV
FINANCIERS & INEGRATORS
Residential C/I/M C/I/M & Industrial & Residential Residential & C/I/M &
& Industrial & Load-Serving <-> C/I/M & Industrial Industrial
C/I/M Load-Serving Entity DER Provider <-> <->
Entity ISO/TSO/RTO ISO/TSO/RTO
Residential Residential C/I/M Industrial Load-Serving Residential Residential & DER
& & Industrial Entity & & C/I/M C/I/M & Provider
C/I/M ISO/TSO/RTO <-> ElectricVehicle
& Industrial ISO/TSO/RTO <->
ISO/TSO/RTO
Customer Segment(s)
Monetary Flow
Electricity Service Flow
Information Flow
Table B.9: Solar-Plus-Storage for End-User and System Co-optimization Business Model Examples
A number of business models have emerged that attempt B.6.1.2 Solar-plus-storage end-user optimization
to bring “firm” solar PV resources to market by pairing
As noted above, solar-plus-storage systems have been
solar PV and storage technologies. The aggregations of PV
most commonly deployed at customer sites to increase
and storage (and, in some cases, other technologies, such
self-consumption (most often in the face of explicit
as demand response and distributed generators) are often
incentives to do so), provide backup power, and minimize
termed “virtual power plants,” or “VPPs.” Revenue streams
electricity demand charges. Figure B.11 presents a general
are structured around the sale and financing of the assets
structure of this business model archetype, which we
and fees for brokering market interactions on behalf of the
have called “solar-plus-storage end-user optimization.”
system hosts. In certain cases, businesses will own the
projects and earn revenues on the sales of energy (most
often under long-term power purchase agreements),
operating reserve, and capacity services (i.e., commodity
sales revenues). Table B.9 presents some examples of
businesses currently operating within this archetype.
ELECTRICITY
SERVICE PROVIDER:
REGULATED UTILITY,
LOAD-SERVING ENTITY
InformationStreams: Electricity
Energy Usage, Service Payments:
Electricity Service Payments for Energy
Price Signals Networks, etc.
Monetary Flow
Non-Electricity Service
Information Flow
Many US states have explicit subsidies for energy storage B.6.2 Solar photovoltaics business
technologies. In recent years, governments, including the model archetypes
German government, have begun offering subsidies for
PV system integrators fall into three key archetypes, each
solar PV paired with energy storage (KfW 2013). These
with significant internal nuances, which are explored
businesses operate very similarly to their pure-play
below. All grid-connected PV systems operate with a
storage or solar counterparts. They tend to sell products
DC/AC inverter. These inverters have the capability to
directly to residential and C/I/M customers and structure
modulate their power factor, providing or consuming
revenue streams around the sales and financing of solar
active power (during producing hours) and reactive power
and storage assets. Examples of companies within this
(at all hours). Modulating the power factor of distributed
business model archetype are presented in Section B.6.2.
PV and storage systems has been shown to be effective at
maintaining distribution voltage in certain cases (Turitsyn
9 The power factor is defined as the ratio between the active power (kW) and the
apparent power (kVA).
10 Niche financing options such as “property assessed clean energy” are emerging,
but the dominant methods remain direct and third-party ownership.
Re
As ven
s u
SOLAR PV FINANCE S ale et S e St
AND INSTALLATION s, ale rea
Sp /F m
ac ina s:
FINANCING RESOURCE: eR n
en cing
Lending/Renting/Leasing t/
Le , En
SOLAR PV RESOURCE as erg
eP y
ay
me
Information Streams: nts
Solar PV System Production
Information Streams:
Energy Usage
Monetary Flow
Electricity Service Flow
Information Flow
The exact structure of the project deployment financing A significant amount of variation exists in the role and the
may dramatically change the economics of projects. degree of vertical and horizontal integration of the solar
However, the financing structure does not change PV integrator. For example, some small installers partner
the basic components of the business model (i.e., the with larger financiers; others outsource the installation
business is still earning revenues through a lease-like of the systems. Still others perform the installation and
payment). For a detailed description of deployment financing functions themselves. SolarCity has announced
financing methods for solar PV, see Lutton (2013) plans to further vertically integrate into manufacturing.
and Speer (2012). Furthermore, there are a number of Certain businesses have undergone horizontal integration.
methods for selling bundles of solar PV projects after For example, a number of load-serving entities (LSEs)
the projects have been installed; these options fall into have begun to provide solar PV. In these cases, the LSE
the categories of securitization and bonding. These earns revenues on the sale or financing of the asset as
methods can significantly lower the cost of capital well as on subscription fees that are typically levied on
for the businesses installing the assets (Motyka et al. retail customers. Some companies that have traditionally
2015). However, these post-installation bundling and focused on home security services, such as Vivint Solar
sales methods do not dramatically change the electricity and Alarm.com, have also horizontally integrated into
services aspects of the business models discussed. That solar PV installation.
is, the business must still identify/locate a customer,
Finally, a significant number of business models
install a system, and deliver an electricity service that
have emerged that provide support services, such
generates a revenue stream sufficient to justify the cost
as generating sales leads, performing pure financing
of capital.
functions, or performing system maintenance. These
services are categorized as non-electricity services.
Figure B.13: Generic Utility-Scale Solar PV Finance and Installation Business Model Structure
UTILITY-SCALE PV
Information Streams:
FINANCE AND INSTALLATION
Production Forecasts, Production
REGULATED UTILITY, PV RESOURCE:
ISO/TSO/RTO Revenue Streams: Mono/Poly Crystalline Silicon, Thin Film, etc.
Asset Purchase/Lease, Commodity Purchase FINANCING RESOURCE:
Bridge Financing, Lending/Renting/Leasing
Service Provision:
Energy
Monetary Flow
Electricity Service Flow
Information Flow
B.6.2.2 Utility-scale PV finance and installation to industrial or regulated utility customers, while the
renewable energy credits (RECs) associated with the
It is difficult to demarcate the scale of installation that
energy are sold to another party (again, most often the
should be considered “utility scale,” as many “utility-
load-serving entity). In certain cases, these businesses
scale” plants are connected at distribution voltages
sell large projects to commercial or industrial customers
(SMA Solar Technology 2012); for example, as of March
and then sell credits associated with the energy
2013, 95 percent of solar PV capacity in Germany was
produced to utilities (Wiser et al. 2010). These business
connected to low- and medium-voltage networks, despite
models require financing structures that are often quite
the significant number of multi-megawatt-scale plants
different from those deployed at smaller scale plants.
(von Appen et al., 2013). Nonetheless, many businesses
Utility-scale PV business models tend to be focused
focus exclusively on developing large, multi-megawatt-
on the establishment of PPAs. Figure B.13 presents a
scale PV plants. The primary driver behind these types of
general structure of such “utility-scale PV finance and
installations in the United States has been the fulfillment
installation” business models.
of “renewable portfolio standards” that require utilities
or other agents (commonly the load-serving entity) There are many supporting roles and distinctions within
to procure a certain quantity of solar PV (or credits the category of utility-scale PV providers. Certain
associated with solar) by a certain date (Steward and businesses are focused entirely on procuring the rights
Doris 2014). Therefore, large-scale solar businesses to land, ensuring that PPAs are signed, and sourcing
commonly sell to multiple parties: The energy is sold into contractors to perform construction. These “developers”
wholesale markets or directly under long-term power will often eschew any ownership of the project after
purchase agreements (PPAs, i.e., commodity sales) construction. Other engineering, procurement, and
construction companies (EPCs) focus on exactly what in unsuitable areas to procure solar PV. Community solar
that name suggests—engineering, procuring supplies involves installing large solar PV plants located away from
for, and constructing the projects. EPCs tend not to take the customer site. Customers can purchase the rights to
ownership stakes in the projects in which they participate. a portion of the output of the solar plant, or can purchase
an equity stake or share in revenues from a portion of
Business models within this archetype utilize
the plant outright (Coughlin et al. 2010). The business
securitization methods similar to those deployed by
earns revenues by charging the customer for access to
the distributed PV business models discussed above
the PV system outputs (brokerage fees). The community
(Motyka 2015). As with the distributed business models,
solar provider will typically sell the plant’s output under
the method of securitization, bonding, or direct sale of
a long-term PPA and distribute the associated revenues
the assets after installation functions primarily to lower
to the project’s shareholders (Coughlin et al. 2010). The
the cost of capital for the installation and does not
community solar provider approach has been particularly
significantly change the manner in which the electricity
popular among regulated utilities that see it as a way to
services are provided. Table B.12 presents examples of
leverage their strengths and provide a value-added solar
companies within this archetype.
service (Siegrist et al. 2013). The community solar market
B.6.2.3 Community solar providers is still relatively small (tens to hundreds of megawatts
in the United States) and geographically restricted to
Many residences and commercial buildings are not
policy-friendly environments, but it is expected to grow
proper sites for distributed PV installations because of
over the next decade (Munsell 2015). An interesting and
shading, building ownership challenges, and other factors.
related model is that of solar “crowd funding” startup,
“Community solar providers” have emerged to capitalize
Mosaic. Mosaic allows individuals (e.g., homeowners
on economies of unit scale or to enable consumers located
or business owners as opposed to banks) to offer funds
COMMUNITY
SOLAR BUSINESS
ed
uc Re
rod FINANCING RESOURCE:
Br ven
n t: gy P Lending/Renting/Leasing
e r n As oker ue S
ym ne tio se ag tre
e Pa for E : d uc SOLAR PV RESOURCE
t S e F am
c s s ro s: ale ee
rvi nt am , P on s/ s, s:
Se yme S tre asts isi Fin
P a n e c o v an
tio For Pr cin
r ma tion r v ice g
o
Inf oduc Se ergy
Pr En
Monetary Flow
Electricity Service Flow
Information Flow
of Its Operations Throughout the State of Arizona.” Local PV Program Response to Financial Innovation
Vol. E-01933A-1. Tuscon, AZ: Arizona Corporation and Disparate Federal Tax Treatment in the Residential
Systems.” Nature Climate Change 5: 315–8. doi:10.1038/ Energy, et al. 2015. “The Economics of Load Defection:
Systems 28: 3874–84. doi:10.1109/TPWRS.2013.2259269. Matayoshi, et al. 2014. “Solar Power & Energy Storage
Policy Factors vs . Improving Economics.” Morgan Stanley
Allcott, H. 2011. “Social Norms and Energy Conservation.” Blue Paper.
Journal of Public Economics 95: 1082–95. doi:10.1016/j.
jpubeco.2011.03.003. California ISO. 2015a. “ISO Board Approves Gateway to
the Distributed Energy Future.” Folsom, CA: California
Amusa, M. 2013. “How the Biggest U.S. Utilities Are Independent System Operator.
Planning for the Future.” Utility Dive. www.utilitydive.com/
news/how-the-biggest-us-utilities-are-planning-for-the- California ISO. 2015b. “Expanded Metering and Telemetry
Cappers P., C. Goldman, and D. Kathan. 2010. “Demand EEI. 2013. “Key Facts about the Electric Power Industry.”
Response in U.S. Electricity Markets: Empirical Evidence.” Washington, DC: Edison Electric Institute.
Energy 35: 1526–35. doi:10.1016/j.energy.2009.06.029.
EIA. 2015. Electric Power Sales, Revenue, and Energy
Coughlin J., J. Grove, L. Irvine, F.J. Janet, S.J. Phillips, L. Efficiency Form EIA-861 Detailed Data Files. US Energy
Moynihan, et al. 2010. “A Guide to Community Solar: Information Administration.
Utility, Private, and Non-Profit Development.” Prepared by
EPRI. 2013. “Cost-Effectiveness of Energy Storage in
Bonneville Environmental Foundation and Northwest SEED.
California: Application of the EPRI Energy Storage
Golden, CO: National Renewable Energy Laboratory.
Valuation Tool to Inform the California Public Utility
CSIRO. 2013. “Change and Choice—The Future Grid Commission Proceeding R. 10-12-007.” Palo Alto, CA:
Forum’s Analysis of Australia’s Potential Electricity Electric Power Research Institute. doi:3002001162.
Pathways to 2050.” Newcastle, Australia: The
EPRI. 2014. “The Integrated Grid: Realizing the Full Value
Commonwealth Scientific and Industrial Research
of Central and Distributed Energy Resources.” Palo Alto,
Organisation.
CA: Electric Power Research Institute.
Davidson, C., T.L. James, R. Margolis, R. Fu, and D.
Farivar, M., C.R. Clarke, S.H. Low, and K.M. Chandy.
Feldman. 2014. “U.S. Residential Photovoltaic (PV)
2011. “Inverter VAR Control for Distribution Systems
System Prices, Q4 2013 Benchmarks: Cash Purchase,
with Renewables.” 2011 IEEE International Conference
Fair Market Value, and Prepaid Lease Transaction Prices.”
on Smart Grid Communications 457–62. doi:10.1109/
Golden, CO: National Renewable Energy Laboratory.
SmartGridComm.2011.6102366.
Dehghanpour, K. and S. Afsharnia. 2015. “Electrical
Gillingham, K. and K. Palmer. 2014. “Bridging the Energy
Demand Side Contribution to Frequency Control in Power
Efficiency Gap: Policy Insights from Economic Theory and
Systems: A Review on Technical Aspects.” Renewable and
Empirical Evidence.” Review of Environmental Economics
Sustainable Energy Reviews: 41: 1267–76. doi:10.1016/j.
and Policy Journal 8: 18–38. doi:10.1093/reep/ret021.
rser.2014.09.015.
Green Charge Networks. 2014. “The City of Lancaster
Denholm P., J. Jorgenson, T. Jenkin, D. Palchak, B. Kirby,
Aims to Become the First Net-Zero City in the U.S. with
and M.O. Malley. 2013. “The Value of Energy Storage
Energy Storage System from Green Charge Networks.”
for Grid Applications.” Golden, CO: National Renewable
Lancaster, CA: Museum of Art and History.
Energy Laboratory.
Gruber, T.R. 1993. “A Translation Approach to Portable
Drury, E., P. Denholm, and R. Margolis. 2013. “Sensitivity
Ontology Specifications.” Knowledge Acquisition 5:
of Rooftop PV Projections in the SunShot Vision Study to
199–220.
Market Assumptions.” Golden, CO: National Renewable
Energy Laboratory.
Richter, M. 2012. “Utilities’ Business Models Shariatzadeh, F., P. Mandal and A.K. Srivastava. 2015.
for Renewable Energy: A Review.” Renewable “Demand Response for Sustainable Energy Systems:
Sustainable Energy Reviews 16: 2483–93. doi:10.1016/j. A Review, Application, and Implementation Strategy.”
rser.2012.01.072. Renewable Sustainable Energy Reviews 45: 343–50.
doi:10.1016/j.rser.2015.01.062.
Richter, M. 2013. “Business Model Innovation for
Sustainable Energy: German Utilities and Renewable Sharma A., K. Melkonyan, F. Xie, W. Wilkinson, J. Berg,
Energy.” Energy Policy 62: 1226–37. doi:10.1016/j. J. Jin, et al. 2015. “Top Solar Power Industry Trends for
enpol.2013.05.038. 2015.” IHS.
Rocky Mountain Institute. 2013. “New Business Models Siegrist, C.R., B. Barth, B. Campbell, B. Krishnamoorthy,
for the Distribution Edge: The Transition from Value Chain and M. Taylor. 2013. “Utility Community Solar Handbook:
to Value Constellation.” Boulder, CO. Understanding and Supporting Utility Program
Development.” Washington, DC: Office of Energy
Ruester S., X. He, J. Vasconcelos, and J.-M. Glachant.
Efficiency and Renewable Energy, Department of Energy.
2012. “Electricity Storage: How to Facilitate Its
Deployment and Operation in the EU.” Florence, Italy: SMA Solar Technology. 2012. “Technology Compendium
European University Institute. 3.4: PV Grid Integration.” Niestetal, Germany.
Schmalensee, R., V. Bulović, R. Armstrong, C. Batlle, P. Smart Grid Task Force—EG3. 2015. “Regulatory
Brown, J. Deutch, et al. 2015. The Future of Solar Energy. Recommendations for the Deployment of Flexibility.”
Cambridge, MA: Massachusetts Institute of Technology. Brussels, Belgium: Expert Group 3.
Schoettl, J.-M. and L. Lehmann-Ortega. 2011. Speer, B. 2012. “Residential Solar Photovoltaics:
“Photovoltaic Business Models: Threat or Opportunity for Comparison of Financing Benefits, Innovations, and
Utilities?” Handbook of Research on Energy Entrepreneurship, Options.” Golden, CO: National Renewable Energy
R. Wustenhagen and R. Wuebker, eds. Cheltenham, Laboratory.
United Kingdom: Edward Elgar Publishing. 145–71.
St. John, J. 2016. “The Next Wave of Energy Storage
Schoff, R. 2015. “Energy Storage: Technology Applications Financing: Green Charge Closes $50M Non-Recourse
and Grid Integration.” MIT Energy Initiatiative Associate Debt Funding.” Greentech Media.
Member Symposium: Storage, Renewables, and the
Steward, D. and E. Doris. 2014. “The Effect of State Policy
Evolution of the Grid. Cambridge, MA: Massachusetts
Suites on the Development of Solar Markets.” Golden, CO:
Institute of Technology. 1–16.
National Renewable Energy Laboratory.
Seel, J., G.L. Barbose, and R.H. Wiser. 2014. “An Analysis
Teece, D. 2010. “Business Models, Business Strategy and
of Residential PV System Price Differences between the
Innovation.” Long Range Planning 43: 172–94. doi:10.1016/j.
United States and Germany.” Energy Policy 69: 216–26.
lrp.2009.07.003.
doi:10.1016/j.enpol.2014.02.022.
energy.mit.edu