Scheduling of Operating Reserves

In Electricity Markets

by

Ahmed Khairi

ECE 666 - Winter 2019

Power Systems Operation Course Project

Electrical and Computer Engineering Department
University of Waterloo
 Table of Contents

1. Introduction ........................................................................................................... 5
2. Classes of operating reserves ................................................................................ 6
3. Operating reserves in electricity markets ............................................................. 9
4. Operating reserves and energy ........................................................................... 11
5. Control action for operating reserve .................................................................. 12
6. Activation ............................................................................................................. 14
7. Shared Activation of Reserve .............................................................................. 17
8. Modeling of Operating Reserve Demand Curves ............................................... 18
Benjamin F. Hobbs’s proposal ............................................................................. 20
Jianxue Wang’s proposal ..................................................................................... 22
Cong Liu’s proposal .............................................................................................. 27
9. Conclusion ............................................................................................................ 33
References .................................................................................................................. 34

2|P age
 List of Figures

Figure 1: Contingency incident on the generator and restoration .................................................. 6

Figure 2: Operating reserve classes ................................................................................................. 6
Figure 3: Load duration curve for 2016 Ontario demand ............................................................. 13
Figure 4: Order of Activation ......................................................................................................... 14
Figure 5: Timeline Condition for Scheduling and Activation ......................................................... 15
Figure 6: Regional transmission organizations of North America ................................................. 17
Figure 7: Flowchart of proposed model by Benjamin ................................................................... 21
Figure 8: Optimal Capacity condition for pay-as-bid mode ........................................................... 24
Figure 9: Flowchart for clearing process of operating reserve ...................................................... 26
Figure 10: Operating reserve demand curve................................................................................. 27
Figure 11: Approximation representation for the step function net disturbance ......................... 30
Figure 12: Social welfare for the RTOLCAP .................................................................................... 31

3|P age
 List of Tables

Table 1: Eligibility of Participants to Offer Reserve ......................................................................... 9

Table 2: Control Action Operating Reserve Offers ........................................................................ 13

4|P age
1. Introduction

Operating reserves is the stand-by power or generating capacity available to be used by the system
operator in short time to be able to supply the necessary load in case of an interruption of a generator or
another disturbance in the power system [1]. It is used for emergency situation to handle the mismatch
between the generation and demand. Such mechanism is important to have a reliable power system.
Contingencies in power system may be one or more of the following incidents [2]:

- A rapid unanticipated increase in demand

- One or more generators get disconnected from the network due to different reasons
- A loss of transmission feeder/s that result of having less transfer capacity to the distribution level
or a limitation on the available amount of power to be transferred

Accordingly, system operators in such situations ensure that there is enough emergency generation
capacity in the form of operating reserve that can be called for in a very short time to re-establish the
balance between generation and demand.

Moreover, the reserve requirements in Canada which is part of the network reliability standards are being
determined by the North American Electricity Reliability Corporation (NERC) and the North Power
Coordinating Council (NPCC). Both organizations set the required amounts of operating reserves,
performance standards and the reserve-sharing program.

5|P age
2. Classes of operating reserves

Operating reserves can be classified into three classes depending on the time required to bring the
required generation capacity into use and the physical behaviour of the utility that deliver it [1-3]:

- 10-minute synchronized (spinning) reserve

- 10-minute non-synchronized (non-spinning) reserve or supplemental reserve
- 30-minute reserve or replacement reserve

Figures 1 and 2 illustrate the contingency incident and classes of operating reserves:

Figure 1: Contingency incident on the generator and restoration

of balance through operating reserves [3]

Figure 2: Operating reserve classes [1]

6|P age
10-minute synchronized (spinning) reserve:

It is used to cover the required supply that is lost due to the largest single interruption or contingency
occurred in the power system. That is, the additional generating capacity which can be achieved by
increasing the power generated from a certain or group of generators to substitute for such loss [1-2]. For
example, an interruption to a single generation which supplies 1000 MW will result in losing the whole
generated capacity (1000 MW), hence the system operators should be able to recover the 1000 MW
through scheduling 1000 MW from the 10-minute operating reserves. As per North American Electricity
Reliability Corporation (NERC) and the North Power Coordinating Council, the maximum amount of 10-
minute operating reserve that shall be synchronized is 100%. This percentage is allowed to be reduced
from utility to utility based on its performance to handle and recover from contingencies incidents [2].

Independent Electricity System Operator (IESO) in Ontario can decrease the amount of the 10-minute
synchronized reserve by 10% in any month provided that it can successfully recover their pre-contingency
supply/demand balance after contingency incident. Moreover, the Independent Electricity System
Operator has a successful record in recovering the pre-contingency supply/demand balance, hence it
schedules the minimum allowable requirement of the 10-minute synchronized reserve which is 25% of
the largest single generation unit. However, in case of failure to re-establish the pre-contingency level
then the 10-minute synchronized reserve must increase by 20% from the minimum allowable percentage
of 25% hence the new minimum 10-minute synchronized reserve following a failure to recover the pre-
contingency will be 45%. It will remain at this level until a successful recovery of a future incident, which
in this case the Independent Electricity System Operator can reduce the level to 25% [2].

10-minute non-synchronized (non-spinning) reserve (supplemental reserve):

It is the additional generation amount that is not connected in currently in the power system but can be
acquired either from the available running generators that can supply this amount by increasing their
generation capacity or acquired by purchasing and importing it from other neighboring utilities [1-2].

The Independent Electricity System Operator usually set the level of the 10-minute non-synchronized
reserve to 75%. Accordingly, adding both the 10-minute synchronized reserve with the 10-minute non-
synchronized reserve will sum up to 100% (25%+75%) [2].

7|P age
It is necessary to highlight that both the synchronized reserve and the non-synchronized reserve should
be able to reach to their required capacities in 10 minutes duration.

30-minute reserve (replacement reserve):

It is the reserved power by the generators and need longer start-up time to reach to their required
capacity. It is used to reduce and relieve the generators who supply the 10-minute synchronized reserve
as well as the 10-minute non-synchronized reserve, hence restoring the system operating reserve [1-2].

The Independent Electricity System Operator has a requirement for the 30-minute reserve that is to be
equal to the greater of [2]:

- Half the capacity of the second largest single contingency condition

- The largest available generator in the network

The 30-minute reserve is not required to be synchronized with the system as per the North American
Electricity Reliability Corporation and the North Power Coordinating Council.

8|P age
3. Operating reserves in electricity markets

There are two schemes of market clearance for purchasing energy and ancillary services (including
operating reserves) [4]:

- Transactions of energy and ancillary services are cleared in aggregated process such as NYISO,
PJM , ISO-NE and CAISO. That is the clearance process are done in the same time for both energy
and ancillary services
- Transactions of energy and ancillary services are cleared separately and sequentially

Each of the three classes of operating reserve has a market that eases the purchase of required reserve
capacity to meet the system demand. In Ontario, there will be scheduling for each of the reserve in the
market associated with their prices every 5 minutes. This comes along the side of normal scheduling of
energy required to meet the demand due to load forecasting not because of contingency incidents [2].

In such market, a process called joint optimization (a dispatch algorithm) is used to schedule the energy
and operating reserve simultaneously. Table 1 shows the market participants who are allowed to provide
operating reserves.

Table 1: Eligibility of Participants to Offer Reserve [2]

Operating reserve Classes Dispatchable Load Dispatchable Generators Imports Exports

10-minute synchronized Yes Yes No No

10-minute non-synchronized Yes Yes Yes Yes

30-minute Yes Yes Yes Yes

As per the North American Electricity Reliability Corporation and the North Power Coordinating Council
imports and exports are not allowed to offer synchronized reserve. Moreover, participants are not
permitted to sell reserve from Ontario to other provinces [2].

There are two requirements to be able to participate in the market:

- Capable to provide the required energy within a specified time duration: 10 minutes for the 10-
minute synchronized reserve, 10 minutes for the 10-minute non-synchronized reserve and 30
minutes for the 30-minute reserve
- Capable to supply the required energy per each class and sustain it up to one hour

9|P age
Furthermore, the participant is required to have a bid or offer in the energy market with an amount more
than or equal to the amount of the his operating reserve offer, hence:

- He can use a price-quantity pair method similar to the way in the energy market however he is
limited up to 5 price-quantity pairs
- He is retracted to have one ramp rate

10 | P a g e
4. Operating reserves and energy

The elasticity of the generators to adjust the output to meet the demand under major contingencies is
what describes the generation side operating reserve services whereas the demand side operating reserve
is described by the flexibility of the demand to change its consumption [5]. When utility participates in
market and schedules for operating reserves, then a joint optimization process is the base to determine
its scheduling. As an example, if a generator offers 400 MW of energy and 400 MW of operating reserve,
then the utility can schedule the entire 800 MW for energy, or for operating reserve, or combination of
both such that each can take up 400 MW. This can be illustrated as:

(𝐸𝑛𝑒𝑟𝑔𝑦 𝑆𝑐ℎ𝑒𝑑𝑢𝑙𝑒 + 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑅𝑒𝑠𝑒𝑟𝑣𝑒) ≤ 𝐺𝑒𝑒𝑛𝑟𝑎𝑡𝑜𝑟 𝐸𝑛𝑒𝑟𝑔𝑦 𝑂𝑓𝑓𝑒𝑟

In the same way, a load can bid for 300 MW of energy and offer operating reserve of 300 MW. In that
way, it can be scheduled for amount less or equal to its bid quantity and receive operating reserve of an
amount less or equal to its scheduled energy. This can be illustrated as:

𝐿𝑜𝑎𝑑 𝐸𝑛𝑒𝑟𝑔𝑦 𝐵𝑖𝑑 ≥ 𝐸𝑛𝑒𝑟𝑔𝑦 𝑆𝑐ℎ𝑒𝑑𝑢𝑙𝑒 ≥ 𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑅𝑒𝑠𝑒𝑟𝑣𝑒 𝑆𝑐ℎ𝑒𝑑𝑢𝑙𝑒

The operating reserve schedule can be known by sending dispatch instructions of the operating reserve
to the selected loads and generators at the beginning of each interval. This schedule provides the market
participants of the amount of operating reserve available for each class and obligated by the utility to be
dispatched once needed.

In the electricity market, the published hour-ahead pre-dispatch schedules provides the market
participants (importers and exporters) of their energy and operating reserve obligations.

11 | P a g e
5. Control action for operating reserve

System operators such as Independent Electricity System Operator in Ontario use several control action
techniques to manage the market shortage in providing necessary supply to match the forecasted demand
and operating reserve requirement such as [6]:

- 3% voltage reduction
- 5% voltage reduction
- Using the 30-minute operating reserve for the 10-minute synchronized and non-synchronized
reserve

These methods have pre-determined price, duration and quantity and are available at all times as input
to a real-time dispatch algorithm.

Adopting this technique is done through voltage reduction in the distribution systems and hence the
consumed energy in the affected area is reduced.

Standard voltage reduction in Ontario is by:

- Reducing the voltage by 3% will result in 1.5% reduction in the total consumed energy (using 3%
for a total load of 30,000 MW will result in reduction of 450 MW to be 29550 MW)
- Reducing the voltage by 5% will result in 2.6% reduction in the total consumed energy (using 5%
for a total load of 30,000 MW will result in reduction of 780 MW to be 29220 MW)

Nevertheless, It is to note that using the voltage reduction technique as a way to compensate using the
operating reserve will result in making this approach unavailable as a way of an emergency control action
to mitigate system reliability issues.

Figure 3 shows an example for Ontario by using the voltage reduction technique with duration and
amount of reduced energy

12 | P a g e
 Figure 3: Load duration curve for 2016 Ontario demand
and the load relief using voltage reduction method [6]

Independent Electricity System Operator has a real-time system for energy and operating reserve
dispatches that utilizes offers from a pair of dummy resources as shown in table 2.

Table 2: Control Action Operating Reserve Offers [2]

10-minute NS
Resource Quantity (MW) 30-minute offer Energy offer
offer

RICHVIEW-230.G_3VR 400 $30 $30.10 $2000

200 $75 $2000

RICHVIEW-230.G_3VR
200 $100 $2000

The control action operating reserve has $2000/MWh is the maximum market clearing price. This
guarantees that the control action operating reserve is activated once finalizing scheduling other reserve.

13 | P a g e
6. Activation

Activation of operating reserve is not based on the offered price for the operating reserve but it depends
on the price of the offered energy corresponding to its resource as illustrated in figure 4.

Figure 4: Order of Activation [2]

The following summary can be obtained from figure 4:

- The order of the activation of operating reserve is Generator A  Generator C  Generator B 

Load X
- The schedules of operating reserve don’t have security constraints since they don’t consider the
grid’s physical limitation
- Even though the activation of operating reserve is done as shown in figure 4, however sometimes
a resource maybe skipped if it is difficult to be activated because of the transmission limitation
- It is not necessary to activate all scheduled operating reserve. For example, a utility may schedule
1000 MW as 10-minute operating reserve, however the activation maybe for 400 MW to handle
the loss of a 400 MW generator

14 | P a g e
It is to emphasize that in the event of activating the scheduled operating reserve, the suppliers must
provide the energy in the allocated time:

- Activation of 10-minute synchronized operating reserve, then the utility must provide the energy
in 10 minutes time
- Activation of 10-minute non-synchronized operating reserve, then the utility must provide the
energy in 10 minutes time
- Activation of 30-minute operating reserve, then the utility must provide the energy in 30 minutes
time

Figure 5 shows an example as illustration of condition # 2 that is mentioned above for a generator
scheduled to supply 50 MW of 10-minute non-synchronized operating reserve [2].

Figure 5: Timeline Condition for Scheduling and Activation [2]

The dispatch was sent at 10:00 with the following instructions:

- Normal energy dispatch amount required = 0 MW

- 10-minute non-synchronized operating reserve schedule required = 50 MW

15 | P a g e
A contingency incident occurred at 10:04, hence the following instructions was sent:

- Activation for operating reserve with a energy target required = 50 MW

- The operating reserve is required to be delivered within 10 minutes

As per the standards followed by the Independent Electricity System Operator, it is required to return to
the pre-contingency condition of supply/demand balance within 15 minutes from the start of the
contingency incident [2]. Unsuccessful attempts to reach to the pre-contingency condition will result in
the followings:

- Real reliability implications

- Causing an increase in the requirement of synchronized operating reserve
- Higher costs to the electricity market

If the participants/suppliers who offered operating reserve can’t provide the required amount in the
allocated time, then:

- The utility may clew back the payments of operating reserve

- The participants /suppliers will undergo compliance investigation and a possibility for penalties
- The participants /suppliers maybe excluded from future participation in the operating reserve
electricity markets

In case of operating reserve activation, then the utility reserve requirement is reduced by the same
amount of the activated reserve. This is essential since if it is not reduced then the market would continue
to schedule the resources to meet the utility operating reserve requirement.

Moreover, the Independent Electricity System Operator is required to recover and return its operating
reserve within 105 minutes immediately after the loss of generation that is greater or equal to 500 MW.
However, the time duration is reduced to 90 minutes for other contingency incidents.

After clearing the incident, the operator will deactivate the operating reserve and increase the
requirement to its normal amount. Usually, the reserve deactivation is done by the Independent
Electricity System Operator in small blocks of 250 MW to meet the ramping constraints of the participants
in the market.

16 | P a g e
7. Shared Activation of Reserve

Shared activation of 10-minute operating reserve is a voluntary program that helps utilities that are
interconnected through transmission system to overcome the insufficient supply occurred because of
sudden generation loss. Members of the North East Power Coordinating Council called as Regional
Transmission Organizations of North America, shown in figure 6, and PJM. Both can participate in the 10-
minute operating reserve program when they experience a supply loss of 500 MW or more except for
Maritimes where their capacity is small and hence they may request a 300 MW or more [2],[7].

Figure 6: Regional transmission organizations of North America [7]

In addition, when a utility call for a shared activation of reserve, then the coordinator of the Shared
Activation of Reserve (New York Independent System Operator) assigns 50% of the energy loss to that
utility or area that suffers from the contingency and the remaining 50% is shared equally among the
available participants. It is to highlight that participants shall inform the coordinator (New York
Independent System Operator) if they cannot supply their shared energy. The following example
illustrates how the shared activation of reserve works:

17 | P a g e
PJM has a loss of generation of 600 MW. Due to transmission limitation. PJM calls the New York
Independent System Operator who is the coordinator to request for shared activation of reserve. Then,
the coordinator will distribute the shares on the available participants at that time as follows:

- PJM = 300 MW (50% of the generation loss)

- New York Independent System Operator = 100 MW
- Independent System Operator – New England = 100 MW
- Independent Electricity System Operator = 100 MW

Once the contingency is cleared, PJM informs the New York Independent System Operator that it is no
longer requiring the shared activation of reserve. Then this information is communicated among the
participants so that deactivation is coordinated.

8. Modeling of Operating Reserve Demand Curves

The market size of operating reserve and other ancillary services is much smaller compared to the energy
market, however the revenue and profit received from such services can reach to the same level of the
energy market. Moreover, the services that don’t change by the operating conditions or few participants
can provide them are traded on the basis of long term contracts [8]. This was the basic principle that
Benjamin F. Hobs proposed his model and it will be discussed later in this report.

It is a fact that the introduction of electricity markets resulted in lowering the electricity prices.
Nevertheless, it triggered a problem called missing money that occurs when the market is not designed
in a proper way that in turn result in a situation that the prices of the energy and ancillary services will not
be sufficient to recover the capital, investment and operational costs of the generation utilities in the
expected period [9-10]. That is the costs for the followings [11]:

- The associated investment costs for creating and generating reserves

- The associated operational costs for maintaining the reserves running without being used
- The associated operational costs for utilizing the reserves

To overcome this issue, different markets schemes were designed such as:

- Forward capacity market (e.g., ISO-NE, MISO, NYISO, PJM)

- Energy-only market

18 | P a g e
In the forward capacity market, energy providers submit their offers assuming a fixed demand or by having
capacity demand curve. Hence at a given time the cleared capacity/energy will meet the demand and as
a result of such auction the cleared price will decrease the risk of under payment to a certain degree.
Moreover, higher capacities in the energy-only market are offered to encourage for higher prices hence
recover their investment and operating costs as well as incentivize for new investment.

As a result, different operating reserve demand curves were proposed to mitigate the missing money
problem. Some of these proposals have been used by system operators in certain markets as discussed in
the following sections.

19 | P a g e
Benjamin F. Hobbs’s proposal:

Benjamin F. Hobs discussed in his paper [12] that having caps for energy prices to mitigate the market
power won’t incentivise the generation companies to invest in building new generation units as there
won’t be the targeted energy revenue to cover its fixed costs. He proposed a dynamic representative
model to reflect the effects of operating reserve margins, the generators benefits and customer costs. A
summary of his proposed model logic (a discrete time simulation with yearly time step) that was evaluated
with the PJM system is as follows and its flow chart is shown in figure 7:

- An auction for a quantity that needs to be available in year y is required to performed at y-4 (4
years ahead). That is the model provides at the beginning inaccurate approximated weather
normalized peak load in year y-4 by using the year y-5’s weather normalized peal load and
considering random economic growth. With the simulation, an accurate peak load is obtained.
After that, the gross margin for the energy and ancillary services (E/AS) is obtained for the
benchmark of combustion turbine (CT) for year y-4. The benchmark for the turbines profit is then
calculated by:
Profit = E/AS gross margin + capacity revenue in that year – fixed annual cost of the CT
It is to note that the load forecast performed in this step is the basis for the demand curve in year
y-4.
- The next step is that the utilities and companies will evaluate their CT profits in year (y-7 to y) if
they want to invest in new generation. The profits for some years are found from (y-7 to y-5) and
in process to be found in (y-4) from above equation. Moreover, the capacity price because of prior
auctions will be known for y-3 to y-1. However, the profit for year y needs to be estimated, as its
auction has not yet performed.
- Next, the risk adjusted forecast profit (RAFPy) for such profits is determined. There are two inputs
for the RAFPy. The first input is the weight sets for profits from years y-7 to y where a weight is
assigned to each profit and more weight can be associated to recent profits. The second input is
the utility function that measure the attitude towards risk. This function will distinguish the less
variable profit stream from the more variable profit stream where the last will be unattractive.
- The following step is to obtain the maximum new capacity that the company is willing to build
based on RAFPy. The assumption is that the higher risk adjusted profit will attract the company to
invest in building new generation capacity. The average load growth followed in PJM market
based on the forecasting model is 1.7%/year.

20 | P a g e
 - The last step is build the supply curve based on the current and future capacities and the bidding
prices for each participant. Then, the supply curve will be joined with the demand curve to yield
the optimal price and committed capacity for year y. After conducting all these steps, then the
process is repeated again for the next year.

Figure 7: Flowchart of proposed model by Benjamin [12]

The proposed model has certain assumptions in order to work smoothly. They are as follow:

- Higher profits attract more investment

- Risk aversion: Tending towards more certain profits over less certain profits
- Expected future profit is based on previous and past profit
- Utilizing the peak capacity of generation

21 | P a g e
Jianxue Wang’s proposal:
Most researches assumes that the required capacity of operating reserve is fixed and inelastic. But due to
different conditions that may arise in real time, the operating reserve may need to be modeled in a
different way that is of elastic nature. Jianxue Wang in his research paper [13] has considered that the
operating reserve electricity market is flexible and needs to be optimized by cost benefit analysis. He
investigated the optimal reserve capacity in uniform price and pay as bid (PAB) auction schemes. The
proposed model was suggested to optimize the reserve capacity and clear the market.

It is essential to maximize the social benefit B or minimize the social cost L through finding the optimal
operating reserve capacity as described in the following equations:

(1)

or

(2)
Where

R: The operating reserve capacity to be optimized

BR(R): The decrease of interruption loss

LR(R): The interruption loss when the there is reserve capacity R

BR(R) and LR(R) can be obtained by the following:

(3)

(4)

Where:

EENS,0 : operating reserve capacity before being purchased

EENS,R : operating reserve capacity after being purchased

IEAR : The Interrupted Energy Assessment Rate in North America or Value of Lost Load (VOLL) in U.K.

The IEAR/VOLL is assumed to be equal for all customers since having capacity shortfall affect the system
as a whole. [] has assumed an approximate value for the IEAR as it is difficult to evaluate the benefit of

22 | P a g e
the operating reserve in an accurate way. Different payment modes can be evaluated, hence when
selecting PAB mode, the cost of purchasing operating reserve can be described in 5:

(5)

Where PRj is the bidding price of resource/unit j

However, in uniform price mode, the above equation will be modified as in 6 with some constraints in 7-
8:

(6)

(7)

(8)

Where:

Rmax,j: maximum reserve capacity of unit j

Rmin,j: minimum reserve capacity of unit j

Trequire: required time of the operating reserve

Ramp,j: ramp speed of unit (in Megawatts per minute)

The second constraint ensures that the unit can provide the needed operating reserve capacity within the
required time that is 10 minutes in the synchronous and non- synchronous reserves.

Jianxue Wang highlighted that the described model above consider the following criteria:

- The operating reserve market is independent from other markets

- The operating reserve is cleared right after the energy market

Further, finding the partial derivative f (2), yields the conditions of optimality of the reserve capacity:

(9)

23 | P a g e
In the purchasing operation, R is not considered as continuous variable, hence the difference equation is
used instead:

(10)

By simplifying above equation, the following is obtained with its inequality conditions:

(11)

(12)

(13)

Where ∆C is the incremental cost for the change in operating reserve capacity ∆R. Moreover, ∆EENS,R is
usually of a negative value due to the increase of operating reserve ∆R. Figure 8 shows the mechanism
under PAP mode for clearing the operating reserve capacity. The bidding prices of the participants are
sorted as PR1<PR2<PR3<…, Moreover, the change of cost under the PAP mode utilizing equation (5) can be
described by (14):

(14)

Figure 8: Optimal Capacity condition for pay-as-bid mode [13]

24 | P a g e
From figure 8, RSj is the accumulative reserve capacity and Cj is the accumulative cost after purchasing R j.
The other curve represents the interruption loss LR(R). the operating reserve capacity at beginning of the
clearing process is 0. In addition, the corresponding EENS,0 and the interruption loss IEAR can be evaluating
the generation system reliability. Hence, the system’s EENS is reduced to EENS,1 by purchasing R1 and the
benefit is therefore is:

(15)

From figure 8, ∆LR1 > ∆CR1 which means that buying R1 is favorable. Furthermore, when buying R2 the
reserve capacity will be RS,2=Rs1 +R2. Also, the EENS,2 will be reviewed to investigate the system reliability
under the condition of reserve capacity of RS,2 ,hence, the benefit of R2 will be:

(16)

In case the benefit is bigger than the cost of R 2 then ∆LR2 > ∆CR2 which means that buying R2 is favorable.
The process is to be continued till ∆LRj ≤ ∆CRJ and hence RSj will provide the optimal operating reserve
capacity. The algorithm proposed by Jianxue Wang in [13] is shown in figure 9.

25 | P a g e
Figure 9: Flowchart for clearing process of operating reserve [13]

26 | P a g e
Cong Liu’s proposal:

In energy-only markets, the lack of co-optimization in real time results sometimes in low energy prices
even for low operating reserve capacity. Due to this fact, the Public Utility Commission of Texas instructed
the Electric Reliability Council of Texas to overcome this issue by integrating the price and reserve into
single model that would present a fair transaction in a real time.

On June 1, 2014, the Electric Reliability Council of Texas introduced the operating reserve demand curves
to enhance the energy prices mechanism, hence mitigating the missing money problem [10]. This scheme
assigns a price on the operating reserve through certain model and add it to the overall energy price. The
operating reserve demand curve estimates the operating reserve price based on a probability that the
reserve falls under the lowest contingency level and the value of lost load. The operating reserve demand
curve is a plot that shows the relationship between the product price (operating reserve price) and the
required demand quantity for a certain time duration as shown in figure 10.

Figure 10: Operating reserve demand curve [10]

Cong Liu in his paper [10] proposed a new way to formulate the operating reserve demand curve. He
started by modeling the dispatch correction problem for each event (w) of set of probabilistic events (Ω)
considering the aggregated model and excluding the transmission constraints as shown in (17-19). The
objective is to minimize the correction cost and the load shedding cost.

27 | P a g e
 𝑃𝑖+∆𝑃𝑖,𝑤
𝑀𝑖𝑛∆𝑃Σ,𝑤 𝑉𝑂𝐿𝐿 ∗ (∆𝐿Σ,w − ∆𝑃Σ,w ) + (∑𝑖∈𝐺 ∫𝑃𝑖 𝑀𝐶𝑖 (𝑥) 𝑑𝑥 ) (17)

𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 ∆𝑃Σ,w ≤ ∆𝐿Σ,w (𝜎𝜔 ) (18)

∆𝑃Σ,w ≤ 𝑟 (𝜋𝜔 ) (19)

Where :

i : Index of generation resources

P : Generation resources base point (dispatch output)
w : Index of probabilistic event
r : Reserve capacity
VOLL : Value of loss load
∆𝑳𝚺,𝐰 : System level net disturbances in event w
∆𝑷𝚺,𝐰 : System level dispatch correction for event w
∆𝑃𝑖,𝑤 : Dispatch correction for resource i and event w
𝑀𝐶𝑖 : Energy offer curve of generating resource i

As shown in 17, the total correction costs of all generators is to be equal to the total aggregated energy
offer curve from the original total generation to the new total generation as described in equation 20.
Therefore, the minimization problem can be changed to a maximization problem as follows:

(20)

Hence,

(21)

(22)

(23)

Where (𝜎𝜔 ) & (𝜋𝜔 ) are the Lagrangian multipliers

28 | P a g e
When applying the Karush-Kuhn-Tucker conditions for the above optimization problem, it will yield the
followings:

(24)

(25)

(26)

In the Electric Reliability Council of Texas, the cap the energy offer is at (9,000 $/MWh) that defines the
VOLL. It is assumed in this proposal that the 𝑉𝑂𝐿𝐿 ≥ 𝑀𝐶Σ (𝑥) for the energy-only market. The operating
reserve demand curve can be formulated after considering all possible events w in Ω as follows:

(27)

Where LOLP is the loss of load probability.

The LOLP(r) in (11) depends on the reserve level as well the probability distribution of the net disturbance
∆𝐿∑ ,𝑤 . The net disturbance is affected by many factors such as the probability of generators’ forced
outages, the probability of demand forecast error and the probability of wind forecast error. In the Electric
Reliability Council of Texas, the probability distribution of the net disturbance is found through analyzing
it statistically from the historical data. The data is divided into 24 groups of 4 seasons and 6 time-of-day
blocks per day. The goal by such data preprocessing is to be able to acquire 24 different normal probability
distribution, hence it will evaluate the loss of load probability for its associated season and time block.

Moreover, it was assumed a step function to approximate the system level net disturbance as shown in
figure 11 where the time duration in the first interval is represented by 𝛿 and the same in the second
interval as 1 − 𝛿.

29 | P a g e
 Figure 11: Approximation representation for the step function net disturbance [10]

Accordingly, the net disturbance is represented by a normal distribution

(28)

There are two forms of reserve capacities that are related to the reaction times (fast, slow) defined in the
Electric Reliability Council of Texas. The first one is the real time online reserve capacity (RTOLCAP) and
the real time offline reserve capacity (RTOFFCAP). The real time online reserve capacity is the supply
capacity margin by generation resources that are participated in the security constrained economic
dispatch that can be started in 10 minutes. In addition, the real time offline reserve capacity is slower than
the first type and it can be synchronized and responds to load disturbance within 30 minutes. The social
welfare for the real time online reserve capacity in [10] in the first interval is determined by integrating
it from 0 to its maximum capacity as shown in figure 12 and calculated by 29 whereas the social welfare
for the real time offline/online reserve capacity in the second interval can be calculated by 30.

30 | P a g e
 Figure 12: Social welfare for the RTOLCAP [10]

(29)

(30)

The last step in this model is to find the price adders noted as RTORPA that is composed of scarcity
adder and price-lift adder
RTORPA = scarcity adder + price-lift adder

31 | P a g e
Where the scarcity adder is the following

(31)

The price lift adder is as follows:

(32)

32 | P a g e
9. Conclusion
Operating reserves is considered an essential service that has its own electricity market to help utilities
suffering to provide enough capacity due to sudden load spike or generators shortfall due to some
incidents to be able to meet the demand or to recover the generator capacity and hence meet supply/load
balance. Scheduling of operating reserves are in three forms (10-minute synchronized (spinning) reserve,
10-minute non-synchronized (non-spinning) reserve and 30-minute reserve. The usage of each of these
reserves undergoes certain conditions and requirements. Due to the need for such services, operating
reserves market has been established where participants can voluntarily enter and hence share the
scheduling and activation of such reserves. Due to the “missing money” problem, participants invest in
operating reserve may not be able to recover their investment and operating costs for this product over
the expected period. Therefore, different models have been proposed in the literature to come up with
operating reserve demand curves that can better forecast the expected needed reserves and hence
mitigate the missing money problem.

33 | P a g e
References
[1] En.wikipedia.org, " Operating reserve," [Online]. Available:
https://en.wikipedia.org/wiki/Operating_reserve. [Accessed 1 Mar 2019].

[2] IESO, "Guide to Operating Reserve," IESO, October 2011. [Online]. Available: http:// www.ieso.ca/-
/media/Files/IESO/Document-Library/training/ORGuide.pdf?la=en. [Accessed 1 March 2019].

[3] Pirbazari, A.M. (2010). Ancillary services definitions, markets and practices in the world. 2010
IEEE/PES Transmission and Distribution Conference and Exposition: Latin America (T&D-LA), 32-36.

[4] Yousefi, A., et al. "Optimum scheduling of spinning reserve by integration of demand response
program." Australian Journal of Electrical & Electronics Engineering, June 2013, p. 209.

[5] F. D. Galiana, F. Bouffard, J. M. Arroyo and J. F. Restrepo, "Scheduling and Pricing of Coupled Energy
and Primary, Secondary, and Tertiary Reserves," in Proceedings of the IEEE, vol. 93, no. 11, pp. 1970-
1983, Nov. 2005.

[6] IESO, "CONTROL ACTION OPERATING RESERVE (CAOR) PHASE 1," IESO, October 2017. [Online].
Available: http://www.ieso.ca/-/media/files/ieso/document-library/public-info-session/2017/caor-
phase-1-20171013-presentation.pdf?la=en. [Accessed 1 March 2019].

[7] En.wikipedia.org, "Regional transmission organization (North America)," [Online]. Available:

https://en.wikipedia.org/wiki/Regional_transmission_organization_(North_America). [Accessed 8
Mar 2019].

[8] P. Wang, H. Zareipour, and W. D. Rosehart, “Characteristics of the prices of operating reserves and
regulation services in competitive electricity markets,” Energy Policy, vol. 39, no. 6, pp. 3210–3221,
Jun 2011.

[9] Hogan, W.W.: ‘Electricity scarcity pricing through operating reserves’, Econ. Energy Environ. Policy,
2013, 2, (2), pp. 65–86

[10] C. Liu, J. Guo, H. Hui and Y. Fu, "Modeling of Operating Reserve Demand Curves and System-Level
Price Adders in Real-Time Energy-Only Market," in IEEE Transactions on Power Systems, vol. 33, no. 5,
pp. 4797-4808, Sept. 2018.

[11] Kilk, K., and M. Valdma. "Determination of optimal operating reserves in power systems." Oil Shale,
vol. 26, no. 3, 2009, p. 220+. Academic OneFile,
http://link.galegroup.com/apps/doc/A216339246/AONE?u=uniwater&sid=AONE&xid=4f0b0fe4.
Accessed 24 Mar. 2019

[12] Hobbs, B.F., Hu, M.C., Inon, J., Bhavaraju, M., Stoft, S., 2007. A dynamic analysis of a demand curve-
based capacity market proposal: the PJM reliability pricing model. IEEE Transactions on Power
Systems 22 (1), 3–11

[13] J. Wang, X. Wang, and Y. Wu, “Operating reserve model in the power market,” IEEE Trans. Power
Syst., vol. 20, no. 1, pp. 223–229, Feb. 2005

34 | P a g e