Distributed Resources: 10.1 Managing Two-Q Demand On The Consumer Side
A host of measures, often lumped together as "distributed resources," can be applied. These include conservation and load management methods, as well as DG and power quality methods. Although diverse in nature and purpose, they all share two characteristics. They try to improve the value the consumer receives from energy in some manner.
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0 ratings0% found this document useful (0 votes)
70 views57 pages
Distributed Resources: 10.1 Managing Two-Q Demand On The Consumer Side
A host of measures, often lumped together as "distributed resources," can be applied. These include conservation and load management methods, as well as DG and power quality methods. Although diverse in nature and purpose, they all share two characteristics. They try to improve the value the consumer receives from energy in some manner.
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 57
10.
1 MANAGING TWO-Q DEMAND ON THE CONSUMER SIDE
A host of measures, often lumped together as "distributed resources," can be applied to re- shape or reduce the demand of electric consumers, as well as to mitigate the impact interruptions and poor voltage have on their usage. These include conservation and load management methods that were traditionally known as Demand Side Management (DSM) as well as Distributed Generation (DG) and a number of power quality methods (PQM). Although diverse in nature and purpose, they all share two characteristics. They try to improve the value the consumer receives from energy in some manner, and they are implemented at or near the actual site of energy consumption. This chapter summarizes these methods and the interactions they have with distribution system planning. Section 10.2 begins with a brief review of demand, energy, and reliability control and management methods and technologies, including both what were traditionally called demand side management and newer power quality improvement systems. Section 10.3 looks at conservation voltage reduction, a utility-implemented mechanism for energy and peak reduction that is complicated to analyze but potentially very useful. Distributed generation (DG) is then covered in section 10.4. Section 10.5 discusses energy storage systems. Planning methods to evaluate the appropriateness of these methods are then covered in section 10.6. This chapter is only a summary of key points. For more detail, readers should see the References. Counting "Nega-Watts" The energy not used and the peak load reduced due to a distributed resource program involving DSM are often referred to as "nega-watts." One problem with DSM in the 20 th century, and continuing today with energy management and demand interruption programs, was accurately counting these nega-watts. Reductions and benefits from many programs in the 1980s and 1990s were sometimes overestimated by mistake (see Chapter 29). Monitoring technologies have improved somewhat since the 1980s and 1990s, but determining how many kilowatts of power usage are avoided, or how much the peak- demand-that-would-have-occurred is really cut, is still often a challenge with some distributed resources. Generally, only verification with DG is straightforward (the power produced by the generator can be metered). 331 10 Distributed Resources Copyright 2004 by Marcel Dekker, Inc. 332 Chapter 10 Planners and utility managers considering energy and peak load reduction measures for their systems today need to carefully consider verification and tracking issues from the beginning of any program. Measures to make certain both that the reductions are real and that they last for the period the utility has planned (often many years) need to be planned carefully. Monitoring and tracking systems are a critical part of any distributed resource program and need to be an integral part of the design of any program and one of the first aspects that should be designed and approved. Who Benefits from Peak Load and Energy Reductions? Who benefits has a good deal to do with who favors energy management and who will take an initiative to employ it. Measures that reduce total energy usage generally have a perceived value only to the consumer, not the utility. They reduce the consumer's energy bill, certainly an advantage in the eyes of most consumers. But conversely they reduce utility revenues, which is certainly not an advantage from its perspective. Thus, usually it is energy reduction measures, as opposed to peak load reduction measures, that are most popular in a purely market-driven energy management environment. Items like insulation, compact fluorescent lights, and other measures that more than pay back their investment with energy savings have been steadily but slowly growing in market acceptance purely on the basis of their benefit to the consumer. By contrast, reductions in peak demand level are valuable mostly to just the utility. It can benefit because it does not need to invest in capacity to meet high, perhaps growing, peak loads nor pay for losses and operating costs driven high by peak demand levels. Residential and small business energy consumers, who pay only for energy, not demand, see no savings from peak load reduction measures and have no incentive to implement it. Large commercial and industrial energy consumers do pay for demand, and therefore are usually directly interested in demand reduction programs. Regardless, since the majority of energy usage in many systems is residential, a utility may find that it has to promote peak reduction programs by using financial incentives and marketing them in a way that mirrors the "old fashioned" utility-sponsored DSM programs of the 1980s and 1990s. An important difference though is that modern utilities pretty much have a free hand to price these incentives and set up their peak load management programs in a market-respond ve way 10.2 ENERGY AND DEMAND MANAGEMENT METHODS Many distributed resource approaches aim to reduce energy and/or the peak demand. Figure 10.1 illustrates the general concept: a decrease in the area under the load curve indicates reduced peak load and load at time of peak (same as peak load in this case). Many of the modern methods employed were referred to as demand side management (DSM) during the 1980s and 1990s. Whether aimed at reducing energy usage or cutting peak demand levels, these methods are often referred to as energy management methods. A Rose by Any Other Name During the 1980s and 1990s many utility regulatory agencies in the United States and other nations required electric utilities to promote energy management methods, referred to as demand side management, wherever it was cost effective. Definitions of what "cost effective" meant and how it was determined varied from state to state, but qualitatively the concept was the same in nearly every regulated venue. The electric utility was considered to be a steward of energy usage in its service territory and was expected to accept as part of its role the responsibility for seeing that its customer used, or at least had the option to use, energy in an efficient and cost effective manner. It was required to offer programs that Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 333 encouraged energy conservation, peak demand shaving, and waste reduction, when and where these programs passed certain evaluation tests of efficacy and economy. For example, an electric utility might be ordered to effect a program to help its customers insulate their homes to reduce energy usage. The utility company would be required to provide free advice as well as building and insulation evaluation services, to make available installation of insulation in homes and small businesses, and to offer financing of these to their consumers. Many states set targets for utilities based on studies of the potential reductions possible from such programs and required that the utilities offer incentives to customers to get them to participate. Thus, many residential energy consumers found themselves offered a reduction in electric rates as an added incentive to insulate their homes or participate in other programs with a similar spirit, programs which would ostensibly save them money even without that additional incentive. An entire industry grew up around these and similar utility DSM programs. Electric utilities found themselves in the business of actively promoting and even implementing measures to reduce the sales of the product they provided. Overall, these DSM programs were not a great success. Often the targets set by state commissions, and claimed to have been attained by utilities, were of doubtful legitimacy, or the program plans were simplistic to the point of guaranteed failure. 1 Over the last two decades of the twentieth century, the term "DSM" gained a particularly bad reputation with utilities, not so much because of a lack of merit in the concepts or technologies employed, but because of the design, evaluation, and implementation that grew out of the regulatory- utility framework that usually gave birth to these programs. But in the author's opinion, the major flaw with the entire DSM industry of the late 20 th century was that it was not market driven: State commissions drove utilities to implement programs. Formulae used by commissions might say that the DSM programs were cost effective, but the homeowners and businessmen clearly did not really think so. If they had, they would have embraced such programs wholeheartedly. This situation was often exacerbated by a particularly obnoxious and arrogant attitude that dominated some DSM programs. Utility DSM departments and regulatory commissions alike were sometimes staffed by fervent "energy conservation fanatics" who believed strongly in their cause and also were certain that consumers did not know what they were doing with respect to energy usage. Basically, consumers then, and today, knew one thing that dominated a lot of their energy usage decisions: electric energy (and gasoline, for that matter) is cheap enough to waste. To a majority of consumers, the matter was not that important, and the hassle associated with managing energy to the standards that the commissions mandated to utilities was just not justified in their eyes by the savings. But none of that should blind modern utility planners to the fact that many of those DSM programs were basically good ideas, measures that improved the quality of the end-use, reducing its cost, and offered improved value. In a de-regulated power industry where consumers have a choice of energy options, some of these programs - those that justify 1 See Chapter 29 on the ways that apparently objective planning studies can be either deliberately or inadvertently corrupted to produce optimistic results. In no venue has the author ever seen so much misrepresentation and mistake as in the evaluation of utility DSM potential and its accomplishment during the 1980s. Errors in "nega-watt" estimates, some deliberate but most due to a combination of optimism and naivete, often were close to 100%. Many of the examples discussed in Chapter 29 are taken from detailed "audits" of such programs done by the author once a utility or commission began to see that results did not match claims. Copyright 2004 by Marcel Dekker, Inc. 334 Chapter 10 Normal load curve. Reduction due to energy efficiency improvement Figure 10.1 Energy reduction methods reduce the total energy usage, usually by lowering energy requirements every hour, as shown here for a daily energy curve. Table 10.1 Basic Types of Energy and Peak Load Reduction Methods Type of Consumer Method R C I Appliance upgrade Distributed generation Building insulation End-use storage Fuel switching Interlocking Lighting efficiency Load control Motor/equip upgrade Renewable energy Site automation UPS and PQ devices CVR X X X X X XX X X X X X X X X X XX X X X X X X X X X X X X X X Reduces Energy Peak X X X X X X X X X X X X X X X X X X X X X Driven by Improves Consumer Improves Market Industry End Use? Hassle? Reliability? X X X X X X X X X X X X X X X X X X No Yes Yes Marginally Maybe No Maybe No A bit No Perhaps Yes No No Can be Maybe Maybe Can be Can be Minor Can be Minor Maybe Yes No No No Yes No Yes Maybe No No No No Yes Maybe Yes No Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 335 themselves to consumers - will see a resurgence as the industry begins to adjust to the opportunities created by de-regulation. The modern equivalent of DSM programs will succeed as market-driven services, when packaged and promoted by savvy energy service companies. Types of Energy Management Methods It is possible to list dozens, even hundreds of different energy management methods, but for the most part, consumer site energy management methods fall into broad categories, as shown in Table 10.1. The table represents a modern, rather than a traditional (1980-1990) DSM perspective. It shows the type of consumer for which each category mostly applies, along with the author's assessment, based on experience, of other salient characteristics. Size of the "X" in a column indicates the relative strength or scale of that column's characteristic with respect to that energy management type. Each of these categories is briefly discussed in the remainder of this section. Appliance upgrades As discussed in Chapters 2 and 3, "appliance" is a category of electrical device that covers all manner of equipment that performs some useful end-use function or purpose for home or business. "Household appliances" such as refrigerators, microwave ovens, washer-dryers, and dishwashers fall into this category, as do "appliances" defined by a broader interpretation such as heat pumps, water heaters, air conditioners, garage door openers, computers and televisions, elevators, and so forth. 2 For purposes of this discussion, lighting is not considered an appliance (it is handled separately, below) but in many discussions of appliances in the broadest sense, it is considered a member of this family. Devices in many appliance categories vary considerably in their energy efficiency. For example, available window air conditioners on sale near the author's home, at the time of this writing, vary from a SEER (seasonable energy efficiency ratio, a measure of efficiency in annual energy usage) of 8 to more than 11, a difference of more than 35%. Appliance upgrade means the replacement of appliances with equipment of similar end-use (i.e., AC for AC unit) but a higher energy efficiency. Energy and peak reductions. An "energy efficient" appliance is one that uses less energy in the course of the year, not necessarily one that uses less energy when running (i.e., has a lower contribution to utility peak load). Energy efficiency standards are based on annual energy use, not device load. For example, SEER replaced EER (energy efficiency ratio, a measure of the electricity an AC unit actually used when operating) as the measure of AC energy efficiency in late 1980s and early 1990s. Regardless, many appliance upgrade programs will reduce both energy and peak demand by roughly equal amounts. For example, some energy efficient window AC units both use less energy during an entire summer season and have a lower load when running. But very often an "energy efficient" appliance will not provide a reduction in peak demand just a reduction in energy usage over a period of a day of more. Planners should never automatically assume they render a reduction in T&D peak loads proportional to their energy reduction. In fact, some "energy efficient" appliances increase peak demand. On- demand water heaters have no storage tank to gradually lose heat as most residential waters 2 By contrast, equipment used in manufacturing, process industries, and similar industrial venues are not appliances because most of it does not cleanly fit the "end-use" definition. Machinery such as rolling mills, welders, and so forth performs a useful function but generally not one that is directly associated with a completed end-use. Copyright 2004 by Marcel Dekker, Inc. 336 Chapter 10 heaters do, and thus are more energy efficient over the course of a year. But they work by using a large heating element buried directly in the pipes leading to hot-water outlets, which uses from 15 to 25 kW, five times the load of a storage-tank residential water heater. Similarly, some types of variable-speed air conditioners and heat pumps actually use more electricity when running during peak periods. They gain their energy efficiency over a year because they can run at variable speeds to efficiently vary output (rather than cycling on and off like normal units, which wastes a small amount of energy) But motor-compressor units that can be operated at variable speed are not as inherently efficient as constant speed units, and thus these units are less efficient (i.e., require more power) when running during hot (peak) periods. Energy efficiency technologies. There are two ways than an appliance's energy efficiency can be increased so that an appliance can obtain added efficiency from one or both manners. First, the electro-mechanical efficiency of the device - the efficiency of its machinery in turning electricity into an end-use product (e.g., cool air, hot water, clean dishes) can be increased. This might be done by employing more efficient motors or pumps, or heat exchangers, or any of dozens of other "tricks" that appliance design engineers can employ to make the device more efficient in turning electricity into the gross end-use product produced by the device. Usually, these types of changes are responsible for any really significant possible improvements that can be made, those representing 20, 30, or 40% improvements in energy, and peak load reductions of noticeable degree. A second form of improvement can be effected by increasing the device's efficiency with respect to its environment or the actual duty cycle it will perform. For example, a dishwasher can be fitted with insulation and a vent to the outside so that heat does not escape into the kitchen around it to raise the temperature of room air that then has to be cooled. A computer can be built to "shut down" all but memory-maintenance activities in a computer when not being used, reducing its energy consumption by 80% when idle. A heat pump can be designed to start its compressors and heat them, several minutes before starting its blowers, marginally improving its performance. Usually, efficient appliances employ measures from both categories to gain their improved performance. However, appliance retrofit programs, which improve the efficiency of existing appliances in a home or business without replacing them, usually include onl> the later category of improvement. The best example is "water heater wrapping," which involves putting a blanket of insulation around the water heater tank and hot water pipes in a home or business. This cuts heat losses of a water heater by a small but noticeable amount (about 5-10%). It is economically justifiable because it is easy to do and inexpensive. Another example is "set back" thermostats, which save energy by scheduling home heating and air conditioners in a more efficient manner. Gradual trend of improvement. A very real appliance upgrade program in the residential sector has been implemented in the U.S. by action of the federal government. The Energy Efficiency Act (1993) set gradually increasing targets for efficiency of household appliances such as air conditioners and refrigerators. Over time, the minimum allowable efficiency of units remitted for sale in the U.S. is mandated to increase. Thus, as older appliances wear out, they will be replaced by more energy efficient devices. Technological advance provides something of the same trend, of a lower rate, through improvements made in appliances by manufacturers. For example, the efficiency of household kitchen refrigerators more than doubled from 1940 to 1990 (i.e., energy required for any amount of cooling fell by half). Since then government regulations have stipulated a slightly higher rate of improvement than this. Utility perspective and programs for appliance upgrades. Generally, a utility can and should count on a certain amount of steady, gradual energy efficiency improvements due to Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 337 appliance upgrades as a factor in its load forecasts. It can promote and encourage efficiency where it makes sense through programs of public awareness, financing (utilities in the author's home state offer financing for major appliances that meet certain standards), and actual incentives (for example a $100 rebate when one turns in an old refrigerator and submits proof of purchase of a new, efficient one, where the need is great). But generally, appliance upgrades happen as programs external to the utility, driven by market forces (owners want efficient appliances because they save them money), technological factors (manufacturers improve efficiency as a competitive advantage), and government mandate. Distributed (backup) generation Backup generation, as opposed to distributed generation, can be viewed as a distinct category of customer-site systems. Contrary to the impression created by many modem DG advocates, DG has been widely used in the industry for decades, but almost exclusively as backup generation - aimed at providing a type of improved reliability. Energy consumers, mostly commercial and industrial users with high costs of sustained interruptions, have created a very large and healthy market for backup generation systems. In the past ten years, a number of utilities have started to offer backup generation as a service. Distributed generation, in all forms, will be discussed in section 10.3. Building shell improvements Improvements in building shell insulation, sealing of doors, windows, and vents, and air handling systems (duct work, etc.) can render a large improvement in the heating and air conditioning energy consumption of a home or office building. Energy requirement reductions of up to 66% are possible in most buildings in the United States, although economically justifiable reduction levels are generally only about a third of that. Building shell energy efficient technology. Adding more and better insulation to older homes, and insulating ductwork exposed to ambient temperatures in attics and crawl spaces, is perhaps the most popular form of energy management in this category. Other popular measures are intense weather-sealing of doors and windows, replacement of windows with double- or triple-paned glass and casements that are heavily insulated and sealed, glazed glass or reflecting film put on windows, and awnings. Like appliance upgrades, a certain amount of this type of improvement occurs as a "natural trend," driven by a combination of market and technological forces. Older homes are often replaced by newer homes built to much higher energy efficiency standards. When combined with upgrades of the cooling and heating units themselves, reduction in energy usage of up to 50% can occur between an old structure and a modern replacement. 3 Improved energy, peak, and user satisfaction. Building shell improvements are one category of energy efficiency improvement that almost always improve consumer satisfaction with end-use performance. Well-insulated homes are famously "less drafty," having fewer unwanted warm and cold corners, and far less temperature variation during the day. This improvement is often a selling point stressed in any program pushing these measures. 3 As an example, in 1990 the author left a charming but old 1,500 square foot home near Pittsburgh for a new 3,300 square foot home in Gary, NC. Built in 1939, the Pittsburgh residence had been upgraded in 1964 and 1987. The 1990 home, of "high energy efficiency standards," used noticeably less energy for cooling each summer, despite having more than double the floor space and having to deal with 20% greater cooling degree days in North Carolina as opposed to Pittsburgh. Copyright 2004 by Marcel Dekker, Inc. 338 Chapter 10 Building shell energy programs almost always makes an improvement in both annual energy usage and peak demand. The impact on energy usage is obvious and well described above, and is usually the motivation for their implementation, at least by consumers. Usually, they reduce coincident peak demand but not non-coincident (individual consumer) peak demand. Since the utility really cares only about the former, most of these programs are at least somewhat a type of peak reduction approach. The peak reductions come about because the building shell improvements reduce the burden on existing heaters, heat pumps, and air conditioners. These units cycle on and off to operate. For example, during peak summer temperature periods, the central air conditioner in a home might operate with a 90% duty cycle (cycling on and off for a total 54 minutes operation out of the peak hour). Insulation and other building shell improvements that reduce energy losses in the home by 10% will reduce the need for cooling during this peak hour. Very likely the duty cycle of this AC unit will fall to about 81%, so that the household's contribution to the utility's coincident peak hourly demand would also drop by about 10%. 4 Customer and utility implementation issues. Building shell energy improvements are not simple to implement nor without a noticeable, often significant, cost. They may also involve a considerable amount of inconvenience (removing and rebuilding the outer walls of a home). In addition, effective economic improvement of a residence's or office building's shell requires considerable attention to detail and some basic engineering analysis and design. Some electric utilities provide services to consumers to promote these types of upgrades; a few finance them, often a holdover program from the regulatory-driven programs of the 1990s. Energy service companies (ESCos, the energy providers in a completely competitive, de-regulated electric market) often make energy efficiency improvements of this type a part of marketing programs aimed at attracting consumers with a "better service, lower overall cost" value package. End-use storage End-use storage includes a number of appliance designs that de-couple, or separate, the time of a consumer's use of the appliance's product from the time of the appliance's use of energy to produce the product. Thus, end-uses that are heavily skewed toward peak times can be satisfied while the usage is pushed to off-peak. A widely, nearly universally applied example is a storage water heater - a "water heater" as far as most homeowners are concerned. Residential water heaters, either electric or gas, usually have a tank that holds from 40 to 80 gallons of hot water, ready for use. This means that their heating elements do not have to, and sometimes do not, operate during the time the homeowner draws water from them. Storage water heaters are widely used not for energy storage reasons, but because the storage of hot water permits instant use of a good deal of hot water, without having to wait for it to be heated. However, they are often retro-fitted with controls that shut down their elements during peak periods, as a peak reduction measure, in company with perhaps larger tanks so they have more product stored for use. End-use storage technologies. Sound and effective storage space heaters and coolers using proven technologies are widely available, at least in some parts of the world. Storage 4 The reduction would not be exactly 10% for a variety of secondary effects and ifs, ands, or buts. For more detail in the behavior of such appliances, see Spatial Electric Load Forecasting, Second Edition, H. Lee Willis, Marcel Dekker, 2002, Chapter 3. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 339 home heaters are widely used in Europe. They consist of an insulated box inside of which is placed an array of ceramic "bricks" of very high energy absorption density which surround resistive heating elements. To reduce peak demand, the elements are run at night, heating the bricks to high temperatures. During the day, they are shut down and a small fan blows air over and through the ceramic brick lattice to produce heated air to keep the home warm. These units are relatively simple and fairly reliable and durable. Similarly, "cool storage" units store "cool" in the form of chilled brine (salt water can be cooled to below freezing while still remaining a liquid) which is then run through radiators and baseboard pipes in a building to cool it during the heat of a summer day. The chiller is run only at night, and the only air conditioning load during the day is the small pump, and perhaps fans, to circulate air. Cool storage units utilize only proven technology, but generally they are fairly large, not home sized, and thus applicable mostly to commercial applications. End-use storage units produce peak load reductions, but usually increase energy use. Both room-size and central thermal storage heating units are widely used in Europe, with residential and small-business market penetration in some regions approaching 50% of all homes using electric heating. Cool storage units have been used in some public schools in Texas. Both are very effective at reducing heating and cooling demand of the affected consumers during utility peak times, by as much as 90%. But these technologies increase both the energy usage and actual demand of the unit, compared to non-storage units. There are several reasons. First, both store heat (or lack of it) and encounter slight losses as that heat escapes during the day. More important, though, thermal storage heaters use resistive heating elements, roughly only half as efficiency of a heat pump. 5 Similarly cool storage units cannot utilize "every cooling trick in the book" as can non-storage AC units. As a result, they use more energy during a 24- hour period, and when they are "recharging" (at night, off-peak) they may use more energy than a non-storage unit would when running at peak during the day. (A thermal storage unit equivalent to an 8 kW heat pump might use 10-15 kW of heating elements.) End-use storage generally appeals only to the utility. It reduces utility peak demand in a very noticeable way. But it increases total energy used by the consumer and thus has no market appeal unless supported by significant rebates or incentives. With very rare exceptions end-use storage is only successful when implemented by the utility. In Europe, the widespread use of thermal storage home heating systems in parts of Germany is due to the utilities there having both encouraged their use and rewarded it through particularly attractive off-peak rates to those users. Fuel switching One option that will reduce electric energy usage and peak demand on nearly any T&D power system is fuel switching - moving major energy uses like space and water heating, cooking, and even cooling from an electric base to natural gas or oil. Few electric utilities consider or encourage this because it cuts deeply into revenues. However, many energy consumers consider and effect these measures, particularly when they face high electric rates. 5 A heat pump cannot be used to drive a compact thermal storage unit, because it does not generate intense heat. Thermal storage units store a lot of heat in a relatively small compact amount of matter (several hundred pounds is all) by using resistive elements to raise its temperature by several hundred degrees Centigrade. By comparison, a heat pump can only generate a differential of perhaps 50 degrees. The units would be too bulky to fit in apartments and small homes, or to retrofit to even large homes. Copyright 2004 by Marcel Dekker, Inc. 340 Chapter 10 During the 1980s and 1990s fuel-switching was an option required for utility study by some state regulatory commissions. It was felt that fuel-switching had the potential to permit utilities to avoid high generation and delivery costs under some circumstances. There was also an opinion among some energy conservation advocates that gas and oil usage for water and space heating was inherently more efficient and should be pursued for that reason. Technology. For applications where gas and oil are viable alternatives to electric power, proven gas appliances are widely available to consumers. Gas heaters, washer-dryers, ranges, and ovens are all "stable" technology and have been available for decades. Gas air conditioners work well, and have been available since the 1970s, but are a constantly evolving and improving technology, with newer units providing more reliability and efficiency than those of even a decade before. Electric energy and peak load reduction. Fuel substitution is an effective way of reducing the use of electric energy and cutting peak load. Shifting of some energy intensive end-uses such as water and space heating and air conditioning to gas can make a big difference in demand and energy needs in an area, if applied to a significant portion of the consumer base there. In some cases, the use of gas rather than electricity results in a net reduction in total energy usage, too. For example, in the case of most residential water heaters, it takes less energy to heat water in a home or business by using gas there than to make electricity with it at a generating plant and move that electric power to the home for use there to heat water. In other cases the situation is not so clear, as with heat pumps in areas where gas delivery and distribution costs, and losses, are relatively high. However, overall, using gas at distributed sites for some of these "heavy" applications is more efficient from an overall energy standpoint than using electricity. For this reason, "fuel switching" enjoyed some popularity with regulators during the 1980s and early 1990s, and was a required option for study for some utilities for that reason. However, the situation is not quite that simple. Generally, emissions and environmental impact are lessened if electric power is burned under the superior control and emissions management of a large central generating plant, utilizing new technology, than at many small, perhaps not as well maintained sites. Then, too, there are the distribution systems to be considered, with a host of issues that often dominate decision-making in a region. For this reason, many of the "fuel switching" studies done in the 1980s and 1990s were simplistic and of little lasting value. Probably the best societal strategy to address the "gas versus electric" question is the one the U.S. has taken: let the market decide. Fuel availability and distribution systems. Availability of distribution systems plays an important part in both consumer and utility perspectives on fuel substitution. Electric distribution networks are nearly ubiquitous, whereas many areas of the nation do not have, and probably will never have, gas delivery systems. The marginal costs of construction and system constraints in different areas vary so much within each category (gas and electric) that it is impossible to generalize which is less expensive to distribute. Consumer perspective. Surveys and market results indicate that consumers make decisions among electric and gas options almost entirely on the basis of cost. But overall, electricity has several advantages with respect to consumer appeal. First, electric distribution systems reach more consumers: there is a significant portion of households in the U.S. and elsewhere in the world that have electric but not gas distribution available. Second, most households could get by without natural gas, but not electricity: gas-powered televisions, stereos, computers, and microwave ovens don't exist, and gas lighting is suitable only for ornamental applications. Finally, a segment of consumers perceives that Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 341 electricity is safer than gas. 6 On the other hand, for a majority of consumers where gas is available it is the less costly energy source for "energy intensive" household applications like water and space heating. Utility perspective. Gas versus electric choices for a utility are either very simple or incredibly complicated. They are simple for an electric utility that views its business as competing in a free marketplace for energy business of consumers. Gas is the competitor, and programs to shift consumers to gas seldom make sense (at least if the utility has been well managed). But utilities that distribute and sell both energy sources face a much more complicated business problem, //"the utility has complete freedom to decide what and how it invests in electric and gas systems, it can decide where to extend or limit gas and electricity distribution, and when and where to promote gas or electric consumption, and how to price one energy source versus the other. It can optimize any of a number of business goals: maximized profit, overall (least) cost, minimized business risk, or maximum consumer choice, etc. Generally, regulatory rules severely limit the utility's choices here and clearly identify what and how it will optimize the use of both systems. Generally, fuel switching is not an approach considered by utilities for tactical reasons Distributed generation. For the standpoint of this discussion, distributed power generators can be viewed as devices that effect a type of fuel switching. They will turn gas into electricity on a site-specific basis, and permit gas to power televisions, lighting, and other purely electric applications. DG will be considered later in this chapter. Interlocking Interlocking is a very simple peak load control mechanism that has an impact on a utility system's load that is incredibly difficult to analyze and predict with accuracy. It involves wiring the thermostats or control mechanisms of two or more appliances at a location in series, so that only one of them can operate at any one rime. The most common application of interlocking is the connection of a home's space conditioner (e.g., heat pump) and water heater in a master-slave arrangement, so that the electric water heater cannot operate whenever the air conditioner/heater is operating. Normally, interlocking is intended to reduce only peak demand while having no impact on energy usage. In practice, if it works well as a peak load reduction mechanism, it will render a very slight, usually insignificant reduction in energy usage, too. Technology. Interlocking involves nothing more than re-wiring of the low-voltage (12- 24 volts DC) control wiring in a home or business. This requires a few dozen feet of 24-volt rated control wire and perhaps replacement of single-pole control relays in the appliances with new double-pole relays. All of this is simple, commodity equipment of very low cost and wide availability. Re-wiring is straightforward. For this reason, it is among the least expensive and simplest peak load reduction mechanisms available. Simple to implement but complicated to model. Interlocking's impact on a utility's load is complicated because its operation interacts with appliance diversity and operations schedules. Accurately modeling and predicting its effects requires detailed assessment of appliance-level load behavior and non-coincident versus coincident load curve shapes, as is discussed in Chapter 3, Section 3.3. 6 Several electric utilities discovered through market surveys and focus groups that they could achieve noticeable market share in electric heating against gas by stressing its "safe and warm" aspect. These marketing programs did not imply gas was unsafe (and neither is the author) but rather simply stressed that electricity was safe, essentially playing on concerns some consumers may have already had. Copyright 2004 by Marcel Dekker, Inc. 342 Chapter 10 NORMAL. Ti me > INTERLOCKED Figure 10.2 At the top, an air conditioner and a water heater each act according to the control of their thermostats. Both use all the energy they want and occasionally their usage overlaps. At the bottom, when interlocked they alternate usage and the "slave" appliance (the water heater), obtains less energy because it is prevented from operating as much as it will if not interlocked. This results in a load reduction during peak times. How it works. Suppose that a particular household has a 5 kW electric water heater interlocked to its 10 kW central electric air conditioner. Further, assume that during the peak summer temperatures, the AC unit will operate at a 95% duty cycle, while the household water heater might want to operate at only a 15% duty cycle. Both of these duty cycles are fairly typical for peak summer hours. 7 Interlocking the two appliances prevents the water heater from operating at more than a 5% duty cycle, that being all that is left to it by the air conditioner. Thus, a reduction in 10% of the water heater's duty cycle is affected, ar>d f he household's energy usage during the peak hour is reduced by Reduction in demand at peak = 10% x 5 kW = .5 kW (10.1) Thus, the household's energy use during this peak hour is reduced by a noticeable amount. Water available for use in its tank might drop in temperature somewhat, however, storage water heaters generally can go for an hour or more without energy at all and still satisfy most household needs, and the one in this example is getting a bit of power during peak, anyway. Energy use is not greatly affected: after the peak period is over, the water heater will "be hungry" and operate at a higher than normal duty cycle until it returns the water stored in its tank to full operating temperature. Figure 10.2 illustrates this. By contrast, during an off-peak summer hour, the AC might operate at an 80% duty cycle. In that case there is more than enough time for both appliances to obtain all the energy they need. Thus, interlocking limits usage only during periods of intense (peak) energy usage, and makes no impact otherwise. 7 Water heaters seldom operate at really high duty cycles. Typically, during periods of peak water heater usage, the average duty cycle for water heaters over a utility system is only about 35%. Copyright 2004 by Marcel Dekker, Inc. 8 Distributed Resources 343 Reduction in weather sensitivity. A recognized capability of interlocking, which was never completely included in utility evaluations of it carried out in the 1980s and 1990s, was its reduction in weather sensitivity. Really extreme high temperatures once-in-ten- year extreme summer weather - can cause air conditioners to reach 100% duty cycle, leading to unexpectedly high peak loads that can overstress electrical equipment. 8 With interlocking, the demand weather sensitivity of a household like that used in the example is cut almost in half. As temperature rises, the AC unit is called upon to remove more heat from the home and its duty cycle increases thereby reducing by a like percentage the duty cycle of the home's interlocked water heater. Thus, the increase in demand seen by the system, as temperature rises above a threshold, is reduced to: (AC connected load - Water Heater connected load) (10.2) (AC connected load) of what it was without interlocking. The threshold is the temperature at which the duty cycles of the two interlocked devices sum to 100%. Reduction in non-coincident demand. Interlocking also reduces the peak non-coincident demand of a household, not only during peak, but all 8760 hours of the year. This makes no impact on household use of energy or convenience of the appliances (homeowners would rarely if ever notice any impact), but it does reduce losses and stress on the utility's service transformers and secondary circuits somewhat. When interlocked, a water heater and an air conditioner/heater can never operate at the same time. As is discussed in Chapter 3, if left to their own, random operation, both devices will cycle on and off as their thermostats dictate, and occasionally their usage will overlap, creating "needle peaks" of the type illustrated in Figure 3.2. Such sharp peak loads usually last only a few minutes, until one or the other of the two shuts off. They occur more frequently on peak days (the appliances operate more often so they have more chance to overlap. But they occur "randomly" during the entire year. Interlocking effectively synchronizes the electric usage of the two devices so that they operate one after the other. Non-coincident demand of the household is cut permanently. But coincident load is not affected except during peak conditions. During off-peak periods the interlocking makes no impact. At those times, even rho-igh the non-coincident load curves of households with interlocked appliances have been altered by "rescheduling" of the "on" periods for one of their major appliances, the household non-coincident curves will "add up" (see Figure 3.4) to the same aggregate coincident load curve as they did when not interlocked. The net effect of this 8760-hour-a-year reduction in non-coincident load are reductions in the peak demands seen by the utility's secondary circuits and service transformers and the losses that occur in the consumer's service drops. Total losses reduction at this level may approach 15%. Voltage regulation to the consumer is also improved, although by only a small margin. The author is aware of no situation where interlocking has been applied because of its impact at this level, but the losses savings and improved voltage regulation are a benefit that can be counted in assessments of its benefit. Poor anticipation of the impact of extreme weather caused several widespread outages for utilities in the U.S. when a one-in-ten summer hit the midwestern and northeastern United States in 1999. Many utility T&D systems had been designed handle typical summer weather, but could not handle extreme weather. Large (25,000+ customers), long (8+ hours), and significant (downtown, City Hall) interruptions resulted. For more, see Chapters 5 and 6 in Willis, 2002. Copyright 2004 by Marcel Dekker, Inc. 344 Chapter 10 Interlocking's history. Interlocking of AC and water heater units was tried by a number of utilities in the southern and southwestern United States, and for electric space heating and water heaters by a small number of rural cooperatives in the northern states, in the 1980s and 1990s. Most of these utilities had relatively high rates of load growth and thus had an incentive to cut peak load. Interlocking developed a checkered reputation, working well in some places and not so well, or not at all, in others. It worked very well in one or two summer peaking systems and well enough to be considered a success in several winter-peaking utilities in the Midwest. However, it failed to provide any peak load reduction at one southwestern utility because planners had estimated duty cycles for AC and water heaters of 95% and 15% respectively, as in the example earlier. In actuality they averaged 90% and 8%. 9 Since their sum was less than 100%, there was no reduction in peak demand. 10 These results were widely disseminated within the industry, in part by manufacturers of load controllers who saw interlocking as a competitor to their products and gave it a poor reputation. Interlocking's potential. Interlocking works well in cases where the utility can verify that the sum of the interlocked devices' duty cycles exceeds 100% at time of coincident peak. It is robust, simple to implement, and effective in these cases. It has the added benefit of reducing stress and losses on the secondary voltage system. It is simple to explain to consumers and to install, makes only a minimal impact on consumer usage, and is quite durable and long-lasting. Only consumers with space conditioning and water heating, or similar loads of that type, are candidates for interlocking. In some utility systems this is a small percentage of the utility's customer base. But in other systems, particularly in rural and suburban areas, a significant portion of the residential customer base falls into this category. Consumers have no incentive to interlock their appliances - it provides no benefit to them. Thus, a utility will have to offer an incentive to gain acceptable and market penetration. In some cases, the most expedient and simple way to gain participation might be for the utility to offer a free replacement water heater, only available as an interlocked device, to anyone who needs a replacement or wishes to add one (new construction, conversions). The utility still gets the energy sales, but sees no or very little peak load increase. Lighting efficiency Approximately one-fifth to one-third of the electrical energy sold by the average electric utility in the United States is used by its customers for lighting applications. Nearly 70 percent of this lighting energy is consumed in the commercial, public, and industrial sectors. Among the numerous types of lighting devices (lamps) available, efficiency varies greatly. However, efficiency is only one of a number of important factors to consumers, 9 Residential AC units are usually sized to operate at 90% duty factor during the hottest normally expected weather in a summer. Water heaters typically have a duty cycle during the hour of peak water heater demand of less than 30% and off-peak duty cycles as low as 10%. But a majority of units in this utility system were located in un-air-conditioned garage spaces, where ambient temperatures around them were typically 125F during the hottest part of the day. As a result, most water heaters operated but a few minutes an hour during summer afternoons. 10 In actuality there was a very small impact, barely measurable, because not all homes were average. In a few, less than 15%, duty cycles did exceed 100% during peak loads, and a slight reduction was effected. However, the action of this program was far below expectation, and the utility abandoned it as "ineffective." In this case, it was. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 345 including initial cost, color of light, lifetime, and operating characteristics (some lamps do not provide light the instant they are turned on, but take several minutes to work up to full output). The efficiencies of the principal types of electric lamps vary considerably. Efficiency is measured by lumens per watt, where a lumen is the measure of light flux (on a per energy basis, one lumen is equal to 0.00413 watt), and varies among types. Energy and peak reduction. Lighting efficiency programs reduce both energy usage and peak load. Lighting efficiency technology. Lighting efficiency improvement programs focused on interior applications, most often on replacing incandescent lighting with fluorescent tubes or compact fluorescent bulbs, which use only one-half to one-third the energy for equivalent amounts of light. A 23-watt compact fluorescent produces equivalent illumination to a 75- watt incandescent bulb. In summer-peaking utilities the bulb produces an additional energy savings. A majority of the energy used by an incandescent bulb is turned directly into heat (the bulb radiates more heat than visible light energy). That heat has to be removed from inside the home by an air conditioner. A compact fluorescent bulb thus reduces air conditioner operation and load, too. From the consumer standpoint, an additional appeal is that compact fluorescent devices have a much longer lifetime than incandescent (lasting an average of six to eight years in residential service). Often interior light fixtures, particularly those with reflectors, are replaced. Better (shinier metal and better shaped to disperse light where needed) reflectors can improve the efficiency of an existing fluorescent light fixture by up to 15%. Exterior lighting efficiency programs, for. exterior building illumination, night-time security, and parking lots and storage yards, often see changes in all aspects of the lighting system as a result of a performance/efficiency upgrade. They may include changes in the number of lighting fixtures, their type and orientation, and their scheduling. Residential lighting upgrade programs usually concentrate on replacing incandescent bulbs with compact fluorescent "bulbs." The degree of improvement in lighting energy efficiency seen in the residential sector is usually greater than in commercial applications. Up to a five-fold increase is possible. In the 1980s, utilities often had to provide considerable incentives to consumers to try compact fluorescent lighting. Wider availability of these devices has made these largely unnecessary. Although the economics work out very well (compact fluorescent "bulbs" more than pay for themselves with reduced energy and replacement costs), consumer acceptance has been slow. Many do not like the starting latency of these devices, a roughly one-second delay between the time the light switch is activated and when the "bulb" begins producing light. There is not much that the bulb's designers can do about that unless the device's energy efficiency is substantially reduced. Compact fluorescent "bulbs" have three other disadvantages with regard to incandescent replacement. Their initial cost is about six times that of incandescent bulbs of equivalent output. Second, compact as they may be, many cannot fit into older overhead light fixtures or table and desk lamps. (Design progress is gradually reducing the size of the smallest compact fluorescent bulbs available, improving this liability.) Finally, fluorescent produces a white light, not yellow, and this represents a change for customers (the white light is sometimes perceived as harsher). Utilities promoted compact fluorescent light bulbs to residential customers during the 1990s, as DSM. Most such programs ended a number of years ago. Usage continues to increase, driven by the consumer savings the devices produce and due to wider availability (stores like Home Depot and Ace Hardware carry them, and they can be found in the light bulb section of some food markets). In addition, newer designs are more compact and include some models with a glass shell so the unit looks very much like a "standard" bulb. Copyright 2004 by Marcel Dekker, Inc. 346 Chapter 10 Commercial and Industrial lighting efficiency often includes replacement of incandescent with compact fluorescent, upgrades of interior fluorescent light trays with improved reflectors, and revised exterior illumination. Finally, many utilities worked with architects and home builders to both design new structures so they used natural illumination well and have efficient lighting throughout (e.g., fixtures that will accept compact fluorescent bulbs). Modern application of lighting efficiency programs. All lighting programs reduce energy sales and thus erode revenues. Thus they do not have a strong appeal to utilities. In particular, programs that reduce night-time illumination load (building security, parking lot and storage yard lighting) have very little appeal. They do not bring a peak load reduction, and cut off-peak sales. But they do appeal to consumers, because they lower cost and potentially provide better illumination. In the residential sector, utilities can depend on trends in increasing use of compact fluorescent lamps to continue. Most utilities have not and probably will not aggressively push residential programs, partly because much residential usage is off-peak and thus does not return a big peak load reduction. Commercial and industrial programs are a different matter. A lot of commercial lighting, particularly interior lighting, is coincident with peak demand. The savings in cost due to a good lighting upgrade can be sufficient to attract businessmen without incentives from the utility. In a de-regulated industry, someone - usually an energy service company can very likely make a profitable business out of offering lighting upgrade survey, design, and implementation services. Load control Load control can be implemented in many ways. But in all cases, it means "utility" control of the operating schedule of appliances like air conditioners and water heaters through some remote means. "Utility" is given in quotations here, because in a de-regulated industry it may very well be an energy service company that implements load control, not the delivery utility. Load control of electric water heaters was a classic DSM program of the 1980s and early 1990s. A utility would offer a customer an incentive, perhaps $5 per month rebate, in return for which it would connect a remote-control relay to the electric water heater at their site. During peak periods, it would open the relay, shutting off the water heater, for a period of one, two, and, in some utility cases, four hours. Many other household and business loads were, and are, candidates for load control, including air conditioners and heat pumps and central electric heaters. Here, the units are not shut down for anything more than a brief period, but this is done frequently. For example, an air conditioner might be shut off for 7.5 minutes during each half hour. This effectively limits its duty cycle to 75%. If during peak times the unit would run at a 90% duty cycle, this effects a reduction in coincident peak load equal to 15% of its connected load. Peak load reductions and rebound. Load control is a peak load reduction measure, which may, during peak days, also reduce energy usage slightly, However, theory and experience showed that it had little effect on energy usage, due to an effect called rebound. As an example, the reader can consider the operation of a water heater that has been denied electric power for two hours after that control is terminated. It will immediately switch on and draws power until it raises the temperature of water stored in its tank back up to its thermostat's setting, operating at a higher duty cycle than it would during the post-peak period, if it had not be "controlled." Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 347 Technology and past applications. Many load control pilot programs were implemented during the "DSM era" and a few grew into stable mainstream programs. Florida Power Corporation (today, Progress Energy, Florida) at one time had several hundred thousand sites under control. Similarly Detroit Edison had enough water heaters under control that it could shave peak load by the equivalent of an entire generating unit. Equipment and control strategies varied greatly. A myriad of different control and communications technologies were developed in the 1980s and 1990s to communicate with load controllers, using several types of power line carrier, phone, and numerous different radio communication methods. Some systems were one-way communication only: the utility sent signals out to the controllers, but got no verification back that they had received the control signal or operated correctly. Other systems, generally more costly, included two-way communication which would verify operation. Utility and consumer appeal Load control has no appeal to energy consumers, at least those not on real-time or demand-sensitive rates. It offers them no advantages. In fact, it can offer a disadvantage: slightly cool "hot" water and slightly warm "cool" air during peak periods. But to a utility, load control can effect significant reductions in peak load in a manner that permits it to "program" the timing, the amount, and the location of the reductions. This has great appeal. However, load control is expensive and complicated. First, it requires incentives. Second, it requires control modules at every controlled site, a communications system to control there, and a central load control dispatch center to coordinate it all. Third, this equipment will need service, increasing maintenance costs. Motor and equipment efficiency upgrades This category of energy efficiency program can best be described as "the industrial equivalent of appliance upgrades." It consists of a variety of programs aimed at improving the efficiency of industrial electric energy use. Among these are the direct analog of residential class appliance upgrades - motor efficiency upgrade programs - in which electric motors are replaced with high-efficiency units that do the same job using less electricity. But in addition there are many other measures that fall into this category of improving energy efficiency of industrial processes, such as programs to replace relatively inefficient boilers, pumps, rolling machines, stamping machines, etc., with equipment that is, ultimately, much less demanding of energy. A considerable portion of the electric energy sold by most utilities is converted into mechanical rotation using electric motors of either induction or synchronous type. Many of these motors are included in air conditioners, heat pumps, chillers, rollers, and other heavy industrial equipment. But beyond those uses, motors are needed for a diverse range of applications, including conveyer belt drives, water well pumps, dryer fans, garage door openers, swimming pool pumps, and hundreds of other applications, in residential, commercial, and particularly industrial classes. Motor sizes, applications, and efficiencies vary tremendously in these classes, but in nearly every commercial and industrial facility, there are numerous motors, of all sizes, and thus a potential for energy savings. Motor efficiency technologies for industrial applications. As was the case in residential appliance upgrades, there are two ways than an industrial facility's energy efficiency can be improved. First, equipment, including motors, but also magnets, heating elements, boilers, and flash processors, can be made more efficient in terms of its use of electricity to produce rotation torque, magnetic attraction, heat, and intense radiation, respectively. Second, the device's efficiency with respect to the mechanical function it performs can be improved. A pump that is turned by a motor can be made 5% more efficient in performing the pumping function. Energy efficiency is improved by 5%. Copyright 2004 by Marcel Dekker, Inc. 348 Chapter 10 The efficiency of an electric motor in converting electric power into rotation power is a function of the type of motor (induction, synchronous, single-phase, three-phase), details in its design and assembly (air gaps, tolerances, etc.) and the choice of materials (better magnetic materials in the stator core, for example) can significantly improve efficiency. As a result, high efficiency motors are quite easy to design and build for any application. They simply cost a good deal more than inefficient motors because they employ more expensive materials and require more parts and closer tolerance assembly. In general, smaller motors of any type are less efficient than larger ones. This is partly due to the fact that small electric motors are built to inexpensive, commodity designs, while larger ones are generally built with some regard to energy efficiency and long lifetime, even if those are not predominant thoughts on the part of the designer. But there are physical reasons why a large motor can be more efficient. In some types of motor design magnetic leakage in the end of the stator winding (which is unavoidable unless the motor is to have infinite length) only doubles even if the motor's size and power is increased by a factor often. Thus, whether built to efficient design or not, motors have a definite efficiency of scale. Figure 10.3 shows the range of efficiencies in percent (HP out versus electrical energy in, with HP rated at 746 watts/HP) for the typical range of small industrial applications (5-50 HP). Replacing a "normal" motor in the smaller size categories with a high efficiency motor can improve conversion efficiency by about 5%. The potential improvement drops to almost negligible amounts for larger sizes. Thus, motor efficiency programs often are best aimed at medium and small motor applications. Tailoring and design are the key. Unlike its residential equivalent appliance upgrades - an industrial energy efficiency improvement program cannot be generalized. A standard type of energy efficient refrigerator will fit in and improve the energy usage of most every household's kitchen, but there is no standard type of "high efficiency motor" and if there were, it would fit in but a small portion of applications. Almost every industrial plant is different for other plants in the details of the exact process it employs, its layout, and the models and types of machines it employs. To achieve energy efficiency, motor and other energy upgrades have to be carefully selected and engineered to fit that particular plant. It is impossible to generalize about motor applications or load patterns. Some motors, such as. those powering conveyer belts, run nearly continuously for long periods and are candidates for replacement with efficient motors. But other motors, like those used to open gates at the plant entrance, etc., run infrequently and only for a short period each time. It would make little economic sense to replace them with efficient motors - the savings could hardly be justified. Gradual trend of improvement. Again, as was the case with residential appliances, industrial equipment and plants in nearly all industries are gradually becoming more energy efficient over time. This is due almost entirely to cost considerations: industrial plant owners and operators are very cost conscious and will seek and continuously drive for cost reductions wherever they can be found. Utilities can count on long term continuation of two trends in industrial energy usage. First, all existing industries will gradually become more energy efficient. Second, new industries will spring up which are more energy intensive than ever, as previously unaffordable, or un-thought-of industries are made profitable by- technological progress. No utility effort is needed to drive these trends. This is "progress" in industry. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 349 95 90 LJ z UJ LJ 80 75 HIGH EFFI CI EMCV 10 MOTOR SIZE - HP 100 Figure 10.3 Standard (bottom) and high-efficiency (top) motor efficiencies. Efficiency generally increases with increasing motor size. Example end-use motor efficiency program. In the 1980s a rural utility in the central United States determined that nearly 40% of its residential customers and about half of commercial and industrial customers had at least one water well pump. These motors operate pumps to pull water out of the ground and push it into a pressurized tank. In this utility's case, the water table was deep in most parts of its service territory, so that the required work for most pumps was considerable. A survey indicated the average water well pump in the system was a .75 HP single-phase motor with an efficiency of 62%, mechanically loaded to about 80% of rated horsepower and operating about 1,000 hours/year, with a duty cycle of 40% during the peak hour. This means that the average motor consumed about 850 kWh during a year, and contributed about .4 kW to system peak. Such motors have a lifetime in service of about 8-12 years. The utility offered a $50 rebate 11 to customers who replaced a failed pump motor with one whose efficiency was 75% or better. (Most fractional horsepower motors are only of about 50-66% efficiency, so 75% is "high efficiency" for a motor of this size range). Such a motor cuts energy usage by 100 kWh per year and reduces contribution to peak by about 50 watts. Further energy savings in this end-use category were probably possible by improving pump mechanisms themselves (many are no doubt inefficient in using the motor's mechanical energy to raise and pressurize water), but this was not pursued. Utility perspective and programs for appliance upgrades. Generally, a utility can and should count on a certain amount of steady, gradual energy efficiency improvements due to improvements in existing industry. Nationwide, and globally, the century long trend of 11 High efficiency motors cost about $70 to $100 per HP more than standard motors. Thus, this incentive covered only about half the cost differential. Copyright 2004 by Marcel Dekker, Inc. 350 Chapter 10 gradual increases in total industry energy use should also be assumed. However, what and how much growth occurs in any particular utility service area is a question whose answer requires careful study. Despite their focus on costs, most industrial companies and plant managers do not want to have to manage their own energy usage. They want to focus on their core business. Substantial opportunity exists for utilities or energy service companies (ESCos) to offer energy improvement and cost reduction services in this sector of utility consumer bases. But while there is tremendous potential in industrial motor and equipment upgrades for large industrial energy consumers, it is not easy to engineer or implement. Good, effective, and economically optimized energy efficiency often requires a partial re-design of each plant, its cooling, waste water, ventilation, process conveyors, and a myriad of other equipment. This can reduce energy usage considerably without having an adverse effect on production. But often a good deal of professional analysis and design work is necessary to do this well. On the plus side, the energy usage of some industrial plants, which costs millions of dollars per year, clearly justifies such effort. Overall, most utilities and ESCos conclude that energy efficiency programs for large industrial energy users are best implemented as a series of customized projects, perhaps marketed under an umbrella program for identification and PR purposes, but individually tailored to each opportunity. A motor efficiency upgrade program offered to a utility's or energy service company's customer base as a whole needs to concentrate on ferreting out and marketing to applications that use smaller motors and where efficiency will be worthwhile. Identification with particular end-uses and consumer segments makes identification of customers and marketing much easier. The water well pump example cited above is an excellent program in this regard. It focused on a wide usage category that customers would find easy to identify ("Do you have a water well?"), the end-use represented a lot of energy and on-peak demand, and it used relatively small sizes of motors, so considerable room existed for improvement. Utilities seeking programs to apply widely across their service area need to look for these same characteristics. A combination of a near-continuous usage, so there is a real savings to be had, and smaller size, for which efficiency improvements provide a larger percentage improvement, offer the best potential for generalized programs. Renewable energy Distributed renewable electric energy generation has an appeal to many consumers, both because they may obtain "free" energy from wind or solar generation at their site, and because most energy consumers understand that renewable generation is more environmentally benign than most central plant technologies. There is about renewable energy an image of living more in harmony with nature, letting it do the work, which also seems to increase its appeal. And, potentially, small renewable power units located at customer sites do have the potential to lower both the energy and peak demand that must be delivered over the T&D system. Technology. There are three types of small, renewable energy power generation sources that represent 99% of all renewable applications. These are solar, wind, and small hydro. These technologies are discussed in section 10.4. Effective, proven solar and wind generators exist in a variety of sizes and types, from very small to megawatt-plus sizes (Willis and Scott, 2000). But none of these units can compete head-to-head with electric utility power on the basis of cost. At present, under ideal ambient conditions, when implemented in sizes that achieve large economy of scale, and when professionally engineered and managed, solar and wind power are only barely able to compete on the fringe of the electric power industry. When implemented in smaller Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 351 sizes and sited at customer locations rather than where they have optimal access to natural energy, solar and wind generation perform less cost effectively. However, this does not stop a significant portion of consumers from trying these technologies anyway. In 1998 the author and colleagues performed a quick survey of rural electric energy consumers in western Kansas and eastern Colorado. Approximately 3% of ranch and farm houses had wind generators, not a large market penetration, but remarkable considering that the local utilities provided no direct support for these programs. However, only 20% of these units were actually being used to generate power. Several had been given up as a lost cause by their owners, maintenance and operation simply being more hassle than the ranchers expected. Wind and solar have disadvantages, but they have some advantages. Solar is firmly entrenched, even dominant, as a source of power for small, isolated, low-energy needs. An example is the emergency telephones located on the side of many rural roads. These use a small solar panel to trickle-charge a battery, that will power the phone when it is needed. Wind generation is used to produce power at isolated facilities and communities where T&D lines do not run and where bringing in fossil fuel for power generation is expensive. Non-dispatchable power. One problem with solar and wind generation is that the users have to take the power when the sun is shining, or the wind is blowing, as the case may be. Solar power is not available at night. Wind generator power fluctuates with wind speed. The power from solar and wind generation is not dispatchable one cannot count on certain amounts of power being there on demand. Providing dispatchable power requires using renewable energy in combination with any of the various storage technologies, including battery, flywheel, and other, that can store the renewable electric power for later use. Non-electric renewable energy applications. Often, electric energy needs can be reduced by using solar or wind power to directly address the end-use function. Two residential applications are drying clothes with wind power (a clothesline) and using solar power to heat water directly. Clothesline drying of washed clothing has certain disadvantages, which is why electric and gas clothes dryers were invented in the first place. However, it works well and uses little electric energy (but more human energy). In some parts of the U.S., outside of cities where grime and pollution are present, it is popular. It is, however, perhaps too archaic an application to use as an example for most utilities. Direct solar water heating is perhaps the most widespread and most effective residential and small business use of renewable energy for a normal electric energy application. The technology is simple, involving only plastic pipes, solar radiators, and a storage tank (needed in electric applications, too). Properly designed systems heat water even on cloudy days and store enough hot water to make it through the longest nights. Small systems can augment, and larger systems completely replace, a heavy household energy usage. In many areas of the U.S., particularly where sunlight is plentiful and electric rates are high, these units make economic sense. Their most serious drawbacks are esthetics (the units are too large to easily hide, and not attractive) and hassle. Maintenance requirements are much more than for a normal electric water heater. Consumer interest. Renewable energy is simply not cost competitive in small, customer- size packages, so it does not have a wide and lasting appeal to most consumers. It also is something of a hassle - units require more maintenance and give more operating problems that most homeowners expect. However, a small percentage of consumers are attracted by the novelty, environmental, and/or individualistic aspects of having their own distributed renewable generation. This segment will probably continue to experiment with renewable generation, but it cannot be expected to grow until the technology makes considerable (50%) improvement in cost effectiveness. Copyright 2004 by Marcel Dekker, Inc. 352 Chapter 10 Utility interest. Few utilities have any incentive to implement renewable energy on a distributed basis. It is not cost competitive. Keeping systems operating will require increases in and new types of maintenance activities. Renewable energy looks best when implemented at the system level (e.g., wind generator parks) with its power used to defer peak and energy needs that would otherwise be satisfied by purchase from the wholesale grid. Site-specific electric-use automation One very workable strategy to reduce the cost of energy use is to use it more intelligently. For households and small businesses, home and business automation systems provide this function, scheduling appliances in a manner that obtains savings due to peak reduction and use of lower-cost off-peak energy. Home and small business automation systems can be regarded as intelligent scheduling and interlocking systems, or microprocessor-controlled demand limiters, or appliance usage schedulers, or, in the case where they communicate with the utility, real-time price control or similar systems. They are all of these. Residential systems are "home versions" of industrial plant control and factory automation systems computerized control systems for the home. Technology. A wide variety of home and small business automation systems are available. Although they differ in details, all have several features. First, they have some way of communicating with major appliances. Many use power line carrier in the household wiring. Others use a very low power radio. A few use infra-red signals, like those used by remote controls for televisions. Second, the automation system has a way of controlling the devices and perhaps of sensing if it has been turned on or wants to turn on. This often usually consists of small "controller" boxes which a homeowner can plug into an electric outlet, into which the appliance is then plugged. Thus, a refrigerator is linked into the home automation system by plugging it into a control box, which is controlled by the system. Heavy appliances, such as heaters, AC units, water heaters, and dishwashers, and some others such as outside lights, etc., have to be hard-wired with controllers, work that often may (should) involve an electrician. But overall, the system needs to be able to control most of the non-interior lighting energy uses in the household or business. The third feature is a central control capability, the "smarts" of the system which monitors usage and controls the appliances. This permits the homeowner to set preferences, schedule usage, and program efficiency goals ("Wash the dishes whenever it is most off-peak during the night, but I will always have hot water at 7 AM in the morning, period."). The homeowner or business owner programs the system with his or her individual preferences and priorities. Set-back thermostat operation, delay of dishwashers and clothes washers to off-peak, and other similar schedule decisions can be made once and then implemented on a continuing basis by the system. Conditional requirements can also be set with regard to time, demand level, and end-use ("Keep the AC set back to 80 degrees and heat to only 50 degrees respectively on weekends.") as desired. The system will routinely schedule those functions as programmed and effect the goals given it. The most effective home and small business automation systems are those that communicate with the utility or energy service company system. Perhaps the soundest approach is real-time pricing, or at least identification of when rates are in various rate tiers. Homeowners can program their systems with a load reduction priority. One homeowner could request that water heating be the device cut first, if load must be limited. Another could identify the swimming pool pump, or air conditioning and Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 353 heating, or the refrigerator and freezers. Regardless, the utility can see substantial response to its price signals and interruption requests from systems like these. Cost of home automation and programmable appliance controllers is currently higher than other load control or interlocking options. Systems cost from $500 to over $1,000 for the master control station plus an additional cost of $50 for each appliance controller. A utility would still have to pay incentives for control of appliances. While this cost is considerably more than other load control options, costs of these systems can be expected to drop in the future. Beyond this, home automation performs numerous other functions besides load control, many of which will attract homeowners exclusive of any load reduction benefit. Peak and energy reduction. Home and business automation systems have the potential to render significant reductions in energy usage and peak demand. However, utility planners need to realize that that is not the major purpose of these systems, as viewed by their owners. The systems are intended to improve value: to lower cost and increase efficiency. Peak and energy reductions come about as part of this process. Modern utility industry perspective. At the present time, home automation systems are in their infancy. Frankly, many do not work well, being somewhat undependable (controllers sometimes don't) and, according to user surveys, too difficult to use. 12 Over time, home and business automation systems will become both more capable and hopefully easier to use and more widely available. They will be built into new homes. Utilities can therefore count on a slow but steady increase in these systems. Automation systems offer utilities and energy service companies a "smart consumer end" to real-time and scheduled-rate pricing systems, something necessary if those systems are going to work well. A company that wants to accelerate these trends and take advantage of these systems and the capabilities they provide should probably begin by effecting some standards within its area of operations. It should identify and perhaps make available to customers a preferred system, compatible with its real-time pricing systems and communications. The systems can be sold on the convenience and lower cost for power that they provide. Incentives generally are necessary only to greatly accelerate the rate of acceptance. Uninterruptible and high-quality power supplies A wide variety of "reliability improvement" and "power quality improvement" equipment is available for installation at an energy consumer's home and business. Many of these devices are appliance specific, intended to be connected to only one device (e.g., a computer). Some are installed on circuits, or groups of circuits, affecting only some loads and some parts of the facility. Others are installed at or near the service entrance and affect the service to the entire site. All share one thing in common. They improve the value the consumer obtains from the use of electricity, and his appliances, by controlling or mitigating unwanted characteristics of the delivered power. These can include voltage sags and surges, interruptions of service, and high harmonics content. Technology and types of devices. Uninterruptible power supplies (UPS) consist of an energy storage device (batteries or, in a few newer systems, flywheels), a charger connected to the utility power system to keep the stored energy resource charged, and a power controller to output power to appliances and equipment as needed. They are 12 A pilot program in the late 1990s, with several hundred homeowners in France resulted in only about 5% of homeowners using the systems on a long-term basis. The rest felt the systems were too complicated to operate. Copyright 2004 by Marcel Dekker, Inc. 354 Chapter 10 available from "commodity" units sold at hardware and home improvement stores for only about $100 to large "site sized" units as big as cargo containers that can serve an entire commercial building. Characteristics that are deliberately varied in the design of various types and qualities of UPS design are energy, power, and power quality. Energy is the total amount of kWh that the unit can store and supply. Power is the rate at which it can supply that energy: a .5 kWh, 250 W UPS can serve up to a 250 watt load (i.e., a home PC) for about two hours. Power quality characteristics - frequency stability, voltage regulation, and lack of harmonics - vary greatly among available UPS. Most low-cost commodity units offer far less in this regard than "utility grade" power quality. Some low cost units output what is basically a chopped square wave, a waveform full of harmonics. Frequency often varies by several Hz either randomly or as a function of load. 13 Voltage may vary by as much as 10% from no to full load. On the other hand, the author's computer doesn't seem to care, working well on such a UPS. A small FM radio, however, buzzes loudly when powered by the same UPS. The higher the power quality the "cleaner" the power that is produced - the higher the cost. Truly "perfect" power quality, voltage that varies very little from an exact sinusoidal waveform regardless of source or load behavior or events, is quite expensive to produce. High quality filters and power quality cleanup in the UPS converter, which provides the output power when needed, can double, triple, or even further increase the cost of the entire UPS over a standard unit. But high quality units are available in a range of types and capabilities, so that potentially a consumer can have whatever he or she wants if they are willing to pay for the quality. However, not all UPS systems provide anything above "mediocre" power quality. Some low-cost UPS appear to have been designed to maximize "store appeal," with the highest possible power and energy levels that can be advertised on the box ("Powers 250 watts for up to 2 hours") and a very low cost. No details on voltage regulation, frequency stability, or harmonics content are given. Waveforms are usually closer to a square wave than a sinusoidal. There is nothing inherently wrong with this, since the units will power most equipment, even computers. And frankly, most consumers would not think to look for such statistics and would not necessarily know what the data meant if they did. But this does emphasize two important points. First, if power quality is important, a buyer needs to be very careful to study the details of specifications, and will need to be very selective when buying a UPS. Second, it would be best to make certain just what power quality is needed: one has to pay a premium for higher quality and it makes economic sense to make certain the capability is needed. Surge and sag protection devices are another popular power quality device with widespread appeal to consumers. The goal of the lowest cost surge/sag protectors is simply to protect an appliance from damage from high or low voltage (transient or otherwise, in most cases). The simplest and least expensive of these devices open a breaker if a surge or sag is detected and expect the owner to close the breaker again to re- activate the device. The appliance (e.g., computer) ceases to operate. These are essentially appliance-specific voltage- rather than current-sensitive breakers. More expensive surge suppressors are available that "shave" incoming surges, allowing the appliance to ride through surges and voltage spikes. 13 A UPS purchased at a local hardware store and used by the author for a desktop PC at his home has an output that varies from 57 to 62 Hz, depending on load. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 355 Similarly, there are devices, including some mid- and high-quality UPS systems, that will shave voltage surges and spikes and fill in voltage sags, maintaining power to the protected appliance regardless. The author's home computer UPS is not one of them. It was advertised as an UPS with a "surge protection circuit." Unfortunately, that is a breaker located on the appliance side of the UPS. A voltage spike is not shaved by this device but instead travels through the UPS to the output side causing the breaker to open, dropping the appliance load. Thus, this UPS, which will keep a computer operating through a service interruption, will drop it instantly if there is a voltage spike. 14 Harmonics mitigation equipment. Harmonics filters and blocks, which eliminate in one way or another off-primary frequency power from circuits, are widely available but not nearly as much a "retail" technology as UPS and surge/spike protection circuits. Harmonics are generated by nearly any non-linear impedance, such as a saturated magnetic core in a transformer or motor, a switched power supply, a light dimmer circuit, or any of numerous other household and industrial processes. But harmonics problems are not nearly as pervasive as interruptions and voltage surges and spikes. 15 Almost all electric energy consumers experience interruptions, and are potentially exposed to sags and surges. By contrast, few will ever experience harmonics to the degree that it will affect the operation of their appliances. Peak and energy impacts. None of these devices make any noticeable impact on peak load and energy usage. Reliability and power quality impacts. The whole purpose of these devices is to improve reliability of operation of consumer appliances. Utility and consumer perspective. Consumers have created a high demand for "commodity" UPS systems and surge/sag protection for computers, big screen televisions, and other home electronics. Stores like Home Depot sell a lot of them. Commercial and industrial businesses also buy a lot of these (often just sending employees out to Home Depot), but they also seek whole-site and more capable systems, particularly for critical loads. A number of companies specialize in offering large, customized UPS systems for entire sites, often in company with backup generation that will start within a few minutes. Traditionally, most electric utilities stayed far away from offering UPS, surge and sag protection, or harmonics abatement equipment, perhaps because to do so would have been to admit that power quality was an issue. However, in the past five years several utilities have begun to sell UPSs, surge/sag protection services, and fee-paid harmonics or power quality improvement services. Utility-Sponsored Load Reduction Programs A 21 st century electric distribution utility may find itself implementing something similar to the DSM programs of the late 20 th century, for several reasons. First, they may be ordered to apply conservation and demand reduction measures where they are cost effective, or at least evaluate them and show that those they are not using are not cost effective. Second, a 14 The author treasures this device, and the box it came in, as a demonstration of "getting exactly what you pay for." The unit's brightly colored box promised to support "up to 250 watts" (it powers a 220 watt computer with no problem), for "up to two hours" (it has kept the computer going for nearly that long), and promised that it "protects computers and similar equipment from voltage spikes in the utility supply." It seems to do this, but doesn't keep the computer energized after the spike. 15 Harmonics, particularly if measured with respect to current, are ubiquitous in most household and business wiring systems. But high harmonic current content seldom causes problems. High harmonic voltage content may (usually will) lead to problems. When harmonics do cause a problem, solutions are not "one size fits all" but need to be carefully tailored to solve the particular problem at the site. Copyright 2004 by Marcel Dekker, Inc. 356 Chapter 10 utility may decide to consider these methods as a way of reducing its costs, without regulatory prompting. Either way, planners will need planning methods that (1) assess energy and demand reductions, (2) can address coincidence and location issues of those reductions, (3) provide reasonable cost estimates of the DSM methods for cost effectiveness evaluation, (4) balance these methods and their benefits and costs against T&D needs and the benefits and costs of traditional system additions. 10.3 CONSERVATION VOLTAGE REDUCTION (CVR) Conservation Voltage Reduction (CVR) involves lowering the load on the distribution system by reducing the voltage on the distribution system, as shown in Figure 10.4. It is a "DSM" program that can be implemented only by the utility. The concept behind CVR is that a reduction in voltage will reduce the load of every appliance used by every customer, and hence reduce peak and energy usage. For pure impedance loads, such as incandescent lighting or resistive water heaters, instantaneous power consumed is proportional to the square of the voltage (Power=V 2 R). Hence a 5% reduction in voltage results in a 10% reduction in its demand. However, some loads such as induction motors draw more current as voltage is reduced -- their load remains constant, or even increases slightly (if losses are considered), as voltage is reduced. The composition of load in a system needs to be taken into account in determining if it is a good candidate for CVR. But overall, on many systems, tests have established beyond any doubt that an instantaneous 3% drop in voltage at the substation instantly reduces the load on a distribution feeder by something approaching 3%, and one of 5% produces about 5%. Practical Matters Associated with Reducing Voltage on the System CVR is implemented on a permanent basis by adjusting transformer turns ratio taps at the substation or adjusting the settings of LTCs or voltage regulators at the substation and out on feeder trunks. Most utilities have little problem reducing voltage by 2-5% for CVR using these means. But making it work in conjunction with meeting all other, and prior, distribution operating goals, particularly that of meeting consumer voltage delivery standards, requires some additional effort. Overall, the impact on the operation and design of the distribution system due to CVR is a lequirement that it operate with less voltage drop overall, and ultimately a need for much better focus on voltage drop and control of voltage during operation of the system. Figure 10.4 shows the most simple, ideal case, one that does not happen very often. There, the feeder was operating with a voltage drop much less than permitted by standard, and CVR could be effected by simply lowering voltage at the substation. Figure 10.5 shows a more common situation. The feeder is close to the voltage drop limit at the end, something that was identified in Chapter 9 as good engineering - use all of the resources available to the planner before spending money on additional capacity or voltage performance. In this case, reducing voltage means that the utility must add voltage control equipment in order to keep voltage above the minimum required by its service quality standard at the end of the feeder and keep it low over the rest of the range of the feeder. Thus, CVR requires more and more carefully placed voltage regulators and switched capacitors. But these changes can be implemented successfully. Whether they are cost-effective depends on the utility's perspective and its energy and peak reduction needs, as will be discussed below. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 357 minimum primary level voltage limit 110 1 Distance - miles Figure 10.4 Conservation voltage reduction lowers voltage on a feeder to the minimum possible. Under the right conditions this will reduce peak load and energy usage. Distance - miles Figure 10.5 Conservation voltage reduction can result in too much voltage drop for customers at the ends of the feeders. Here, customers for the last .5 miles on this lateral branch see voltage below minimum guideline (dotted line) with CVR. Note that the slope of the voltage profile with CVR is steeper than without (total voltage drop to the end is about .45 volt more with CVR). The loads on this feeder are predominantly constant-power loads. As voltage goes down, they compensate somewhat by drawing more current, increasing voltage drop. Copyright 2004 by Marcel Dekker, Inc. 358 Chapter 10 C9 e > 110 minimum primary level voltage limit 0 1 2 Distance - miles Figure 10.6 Two different plans for rather extreme levels of CVR re-engineering of the feeder produce different voltage profiles. Solid line involves reconductoring the feeder for less voltage drop. Dotted line uses the existing conductor and installs a regulator near the feeder midpoint. It applies some CVR across the entire feeder. The reconductoring produces far fewer electric feeder losses than the regulator scheme, but the regulator scheme produces a lower average voltage up and down the feeder. How much can voltage be dropped in a realistic CVR program? The amount of permissible CVR program voltage reduction permissible on utility systems varies a great deal depending on design, load characteristics, etc. However, generally: Rarely can one implement a "theoretical" across the board constant reduction (Figure 10.4). Most often, reductions affect only parts of each feeder, as illustrated by the re-engineered feeder profiles in Figure 10.6. Most feeders can absorb reductions of averaging about 1-1.5% without too much re-engineering. There will be exceptions, feeders where no CVR can be applied without major re-engineering and additions. Reductions of about 3% can be effected through considerable refinement of voltage engineering on feeders and the installation of properly tailored line regulators and switched capacitors. Reductions of up to 5% and in some cases a bit more than 6% can be effected during emergencies, if the utility has previously prepared to handle a 3% reduction without problems, as described immediately above as an emergency peak shaving measure. On many utility systems, for the emergency situation listed above, the voltage delivered to consumers near the end of feeders will fall noticeably below the minimum dictated by the utility's standards for normal voltage service. Such situations are tolerable for brief periods Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 359 because the "emergency conditions" permit application of emergency voltage standards. For example, utilities using ANSI standard C84.1-1989 as their voltage standard can get over three additional volts for emergency CVR. The standard lists 105% to 95% of nominal of utilization voltage as the standard for normal conditions but a range of 105.8% to 91.7% as allowable for certain temporary or infrequent operating conditions, i.e., emergencies. CVR Works Well in Short-Term Tests Without doubt reducing voltage reduces load. This is easy to see with what is called a "notch test": drop the voltage at the substation low side bus by adjusting LTC controls or regulation by a total reduction of 5% while monitoring current. The reduction in power usage is obvious. Averaged over more than 20 utilities the author has worked with on CVR programs, systems generally respond slightly better than linearly: a 5% CVR gives about a 5-6% reduction in load. Raise the voltage back to its original level fifteen minutes later and the load instantly goes up by a like amount. It is hard to argue with this type of "notch test" (short-term drop and rise back in voltage). Every one of these the author has observed has indicated that voltage reduction reduces load by about or a little better than a 1:1 ratio. But this does not mean CVR works this well as an energy reduction measure. What is often not recognized is that such "notch tests" do not mean that that reduction lasts permanently. In fact, CVR interacts with diversity of appliance loads, so that certain appliances "fight" the voltage reduction after a period of time, in a process called "duty cycle rebound." Duty Cycle Rebound: A Complicating Factor Consider an electric water heater rated at 4,000 watts at a nominal 230 volts. What happens when its voltage is reduced by 5% to 207 volts? Its load instantaneously drops by very nearly 10%, to 3,610 watts. However, this water heater is still operating under thermostatic control, and that thermostat still intends to operate the water heater elements so that it maintains the water in its tank at the same temperature as it was prior to the CVR occurring. The water heater thus compensates for its lower heating output by running at a 10% higher duty cycle. For example, where it might normally have operated for fifteen minutes in an hour, a duty cycle of 25% at 4,000 watts, it now stays on for an extra 1.5 minutes. Operating now for 16.5 minutes in the hour (a duty cycle of 27.5%, ten percent more than before), it obtains the same 1 kWh of energy that it would have without CVR: Energy use prior to CVR =15 minutes x 4 kW = 1 kWh Energy use after CVR = 16.5 minutes x 3.61 kW = 1 kWh In the long term daily kWh usage and its contribution to system peak in every hour where it runs less than 100% of the time remains exactly the same as before. Since under normal conditions, even of extreme peak conditions, water heaters never run more than about 60% of the time in any hour, this means that over the course of a year, no net reduction in energy usage by water heaters occurs due to CVR. All draw power at a slightly slower rate, but each draws it for a correspondingly longer period. Also, no peak reduction occurs due to CVR impact on water heaters. Prior to CVR, perhaps there were 100,000 electric water heaters, each of average connected load of 4 kW, operating on the system at 25% duty cycle during the peak hour. That means an average of 25% were activated at any one time during peak hour, so water heater contribution to peak demand was Copyright 2004 by Marcel Dekker, Inc. 360 Chapter 10 25% x 100,000 x 4 kW = 100 MW After CVR has been implemented, the connected demand of each water heater is now only 3.61 kW instead of 4 kW. But duty cycles have rebounded to 27.5% to make up for the slower rate of energy delivery to each water heater. Thus, contribution to peak demand is 27.5% x 100,000 x 3.61 kW = 100 MW Similarly, air conditioners, heaters, electric ovens, stoves, water well pumps pressurizing storage tanks, and most of the other "heavy" loads in most homes and businesses operate on some sort of thermostat of "performance" control, and no long-term reduction in energy usage occurs among these loads. Only lighting, particularly incandescent lighting, and non-controlled impedance and constant current loads do not rebound in this manner. CVR Does Provide Energy Reductions, But of Lower Amounts than "Notch Tests" Indicate But CVR does provide permanent reductions in overall energy usage and in peak demand. They are just a bit lower than might be expected from "notch tests." While a majority of home and business appliances (as measured by demand and energy usage) on most utility systems are of the type that rebound, as explained above, a good portion are not. But roughly 25-30% of electric load is lighting. Another 10% typically consists of inexpensive "regulated" AC-DC power supplies (in low-end electric and consumer equipment) which often looks something like a constant current load. Assuming all lighting responds to CVR like an impedance load would, when voltage is reduced by 5%, the load of this 25-30% portion of demand would drop by 10%, for a reduction amounting to 2.5% to 3% of the total. The constant current load portion would drop by 5%, for a reduction of .5% of the total. Thus, a 5% CVR would obtain a permanent reduction in energy usage and peak loads of about 3-3.5% from a 5% reduction. Of course, not all lighting is constant impedance load, but there a small portion of other non-controlled impedance loads on most utility systems. Overall, the author believes that this rule-of- thumb based model of CVR impact provides the best general estimate of its capability. Long-lasting demand reductions of from 50% to 75% of the amount of voltsge reduction can be obtained. CVR does work, well, as an energy reduction means. It does not work nearly as well as often expected based on short-term tests, which show a reduction of about twice that much, but it can be an effective energy reduction measure. Less rebound during some extreme winter peak conditions Duty cycle rebound cannot occur if the duty cycle of the affected appliances is at or very near 100%. For example, if the duty cycle of the water heaters used in the example above had been at 100% during the peak hour, reduction in demand that hour would have been 10%, not zero. Had it been above 91%, some reduction would have been seen, an amount equal to 100% minus the actual duty cycle. But few thermostatic or pressure-controlled appliances operate at 100% duty cycle, even during peak. Water heaters typically operate, at the time of water heater peak demand, at only between 30% and 15% duty cycle, depending on the characteristics of both usage, and water supply temperature in the utility service area (see Chapter 3). Air conditioners and heat pumps are designed to operate at 80 to 90% during peak conditions. (And in addition, they are mostly motor - i.e., constant power - loads so CVR makes only a small reduction on their demand anyway). Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 361 In order for a major appliance to yield a large reduction during peak, it would have to be an impedance load and be at or near 100% duty cycle. The only appliance in wide usage among households and small businesses that meets this categorization is resistive space heating, either of the direct or radiator and baseboard (hot water flow) type. Also included here are the resistive assist heating elements in many heat pumps designed for intense winter usage. All of these appliances are mostly constant-impedance loads, and can be very near 100% duty cycle during extreme weather winter conditions (worst winter in 20 years, etc.). Thus, during really extreme winter peaks, when heating demand is very high, the load of all types of resistive space heaters is affected by CVR, perhaps greatly. It thus acts as a brake on winter extreme weather peaks. Summer peaks, however, will not react in this way because AC and heat pump loads act mostly as constant-power loads, so even if they are at 100% duty cycle, CVR does not affect them (there is neither a significant load reduction or any rebound). Using CVR as a Peak Shaving Method CVR, implemented as described above (on a permanent basis, lowering voltage to the lowest possible value that meets minimum standards) will reduce peak demand, too, since it reduces demand every hour of the year. However, it can be used as a peak reduction tool only in a way that provides temporary reductions in peak demand that are much greater than this "permanent CVR" amount. To do so, the utility must apply CVR only during peak periods. Consider a summer peaking utility that operates at "normal" voltage levels for most of the year, so that at the beginning of its peak hour, late in the afternoon of a July day, it can drop voltage by 3%. All loads that will rebound, including those like water heaters that will "rebound" by operating at longer duty cycles, instantly drop in load, so the utility sees a system wide reduction of something between 3% and 6%. For the sake of argument here, it will be assumed this reduction is 4.5%, or about three times the long- term impact that a permanent reduction would provide. How long will this reduction last? Tests carried out by the author in the mid 1980s indicated it lasts longer than might be expected. Almost an hour and half passes before there is a noticeable indication the reduction is dropping, and it takes more than two hours before the decrease in loading degrades to the long-term figure of only 1/3 as much. The reason is the duty cycles of water heaters and other appliances that are not operating at the time the reduction is implemented. Consider a water heater that just switched on at the time the voltage reduction is applied. It is at a typical water heater cycle, 25%, so it wants to run for 15 minutes, and will now operate for 16.5 minutes under CVR. Thus it will be about 15 minutes before the "rebound effect" of this water heater's additional duty cycle operation will be seen on the system. In the normal course of events, during the next few minutes, other water heaters, off when the voltage drop took place, cycle back on, exactly as they would have had the voltage stayed normal (voltage reduction makes no difference in the duration of the "off cycle" of any device). They come back on at a time when not quite as many water heaters as one would have expected have shut off, and thus there are a few more water heaters running than there would have been if the voltage had been left the same. All are running at a lower load because of the lower voltage, but over time, due to the slightly longer operating cycle of each, more are operating at any one instant (in fact exactly 10% more) and the total coincident demand due to water heating coincident demand is back to where it would have been had voltage never been reduced. Eventually, 10% more water heaters will be operating at any one time, each at 10% less demand. There will be no reduction. Copyright 2004 by Marcel Dekker, Inc. 362 Chapter 10 Now consider a water heater that switched off just a second before the voltage reduction occurred. Again, operating at 25% duty cycle (and assuming that its cycles take about an hour), it will be 45 minutes before it switches on, at a time it would have even without CVR. Its demand will then by lower than it normally would have been, and it will be a further 15 minutes before its rebound occurs, too. Thus, it will be an hour and fifteen minutes before the system sees its rebound. In actuality, during summer, the duty cycle length of many water heaters is well over an hour, so rebound can take as long as two hours to occur. This correction in the number-of-units-operating-at-any-one-moment occurs only after one complete off-on cycle of all water heaters in the system has taken place, and because water heaters operate infrequently, this can take up to two hours. Thus, gradually during the hour or two after voltage is reduced, duty cycle rebound slowly erodes a portion of the load reduction, at least that of water heaters and similar devices. Other controlled impedance appliances, including heaters and ovens, also react in this way: they only recognize they need to operate longer when they switch on. Thus, rebound takes time to occur, being a function of the duty cycle. After this period of adjustment, the actual peak load reduction will be somewhat less than a simplistic approach suggests. CVR Cost CVR's major cost impact of implementing CVR is at the distribution level, where it can increase cost of feeders in some cases. In other feeder cases, it can be applied at no cost, occasionally with little more effort than the resetting of transformer taps and LDC settings. However, CVR reduces the voltage drop allowed to distribution planners to distribute power, a consideration in systems that are well designed from a load reach standpoint and have no margin in voltage. In some cases, CVR can be implemented only by modifying the feeder system (adding regulators or re-conductoring) to overcome voltage drop problems caused by CVR's effective change in voltage standards. 16 Beyond this, by limiting the voltage drop available, feeders will have less margin to accept new load growth before the (now revised) voltage drop criteria are exceeded - they will need to be reinforced more often and perhaps in a more costly manner. Additionally, the utility needs to consider the impact of CVR on load reach and the loss in value of that. In some cases, the loss in capability of the distribution system is considerable, and the overall cost of future additions and expansion is considerable. This is not to say that CVR is not economical or that it should not be examined as an option, merely to point out that the present and future costs on distribution should be taken into account in developing evaluations of benefit/cost. Utility Perspective Traditionally, most vertically integrated electric utilities preferred to use CVR as a "last ditch" peak shaving method. Using it as an energy reduction method reduces revenues, so they generally do not view it favorably. For this reason, some utilities resisted CVR, and others merely did not respond with the same enthusiasm they would have had if CVR would increase revenues by 3%. Lower voltages also meant less margin for unknowns or voltage regulation problems, and planners, engineers, and operators all had concerns about that. Finally, the appeal of a significant peak reduction capability if used only for 16 Some state utility commissions require CVR implementation at the distribution level but permit utilities to not apply it in cases where they can demonstrate an adverse cost impact. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 363 peak shaving was great. CVR does work, it can be implemented where and when needed and it delivers relatively big peak reductions, if only for brief times. Thus many utilities reserved it for their peak shaving mechanism and did not implement it under normal conditions or as an 8760-hour measure. In fact, most utilities operated, and continue to operate, their systems nearer the high than the low end of the permissible voltage spectrum. This is not due entirely, or perhaps even in the majority, to the fact that this reverses the CVR effects and increases revenues by a small amount. The author has grown convinced in over 25 years of working with utilities that this is mainly a culture attitude. Utility planners, engineers, and operators believe that providing slightly higher than nominal voltage is simply better for their customers. They also know that higher voltage gives them a margin that will help them maintain service during unexpected operating contingencies. Certainly, however, most electric utilities tolerated this situation because it had no adverse, and perhaps a slight positive, impact on revenues. In a de-regulated power industry, many utilities will no doubt continue to have the same perspective about CVR. These companies are essentially still vertically integrated, at least to the extent of being local delivery (T&D and retail sales) companies and still see the revenue reduction as something to be avoided. In particular, utilities under rate freezes have to be very concerned about anything that would erode revenues. They have no mechanism to adjust rates to compensate for a change in the coverage of their fixed costs (see the Rate Impact Measure Test discussion of cost evaluation, later in this chapter). However, utilities that meet the strict definition of wire companies - electric delivery utilities who operate the T&D system but do not actually sell the power transported over them - might take a different view of CVR, depending on the regulatory framework within which they work. Depending on how they are compensated for their delivery services, these utilities may be indifferent to whether the energy delivered over their system is reduced or not. Or they may have a different financial incentive that makes them less reluctant to consider reductions. Finally, many will service at least a part of their income, and some a major portion, by billing for the peak demand they transport over the system (that being what applies against their capital charges for capacity and allocation of a large part of their fixed costs). Therefore, CVR's reduction in peak reduction demand will not appeal to them except as a very "last ditch" emergency measure. 10.4 DISTRIBUTED GENERATION Three Ways to "Burn" Fuel to Provide Electric Power Distributed generation most often uses some form of conventional fossil fuel, like gasoline, diesel or fuel oil, natural gas, propane, methane, or gasified coal, to produce electric power. In every case, regardless of fuel, through very careful design and often intricate timing of events, measured amounts of the fossil fuel are oxidized - purposely combined with oxygen - to produce heat, and perhaps pressure, and, ultimately, electricity. There are three major ways in which oxidizing fossil fuels are used to produce electric power. In two of these three approaches, the fuel is burned and a portion of the heat produced is transformed into mechanical rotation, which spins an electric generator, producing electricity. The first of these approaches, the reciprocating piston engine, uses the heat and pressure from combustion to move a piston inside a cylinder, converts that linear motion to rotation of a crankshaft, and uses that rotation to spin an AC electric generator. Copyright 2004 by Marcel Dekker, Inc. 364 Chapter 10 The heat and pressure created by combustion can also be used to spin a turbine wheel. The fossil fuel is burned very close to the turbine vanes, producing a rapidly moving gas that passes over the turbine wheel's vanes, causing it to rotate. Again, this rotation is used to turn an electric generator. The third way to obtain electric power from fossil fuel is with a fuel cell. Fuel cells do not use combustion -- oxidation is performed by chemical catalyst. Fuel cells are essentially fuel-powered batteries, producing low-voltage DC current via the microscopic catalyst- enhanced oxidation of hydrogen from the fossil fuel, in the presence of electrolytes. This DC power is converted to AC power using electronics. Other, less proven ways to produce electric power from fossil fuel have been developed (i.e., thermophotovoltaic), but these three methods appear to be the only economically viable approaches. From Less than 1 kW to More than 25,000 kW Distributed generators are available in sizes from less than 5 kW to 25,000 kVA (even larger turbine and diesel units are available, but 25,000 kW is the top of the "DG range" being considered in this book). Generally, large reciprocating and combustion turbine generators are designed for heavy, long-term use, and are available in sizes from 1,000 kW on upwards. Large fuel cell systems are available in capacities from 1,000 kW up to 10,000 kVA. These larger size DGs are usually installed at the primary distribution voltage, e.g., on portions of the electric system between 2 and 25 kV phase to ground, and are restricted to applications at large industrial sites, or on the electric utility system itself. Some are applied as base load, used 8,760 hours per year, while others serve as peak-reduction units used only during periods of high power demand. Smaller DG units are available in sizes from 1,500 kW down to as small as 5 kW. These units are intended for very dispersed applications, as generators for individual homes and small businesses or as portable power units for construction sites, etc. Reciprocating piston, fuel cell, and a type of turbine (micro-turbine) are all available in this range. Such "mini" and "micro" generators are almost always installed on utilization voltage level (120/240, 480, or 600 volt) circuits, often on the customer side of the electric utility meter. Applications for these types of units can include providing power for all of the eiectiical demand at a residence or small commercial site, or just providing power for peak shaving. They can also be devoted solely to improving availability of power, including usage in UPS (uninterruptible power supply) and standby or emergency power systems. The Carnot Cycle Most fossil fuel DG units are powered by heat engines. They convert heat into mechanical motion by allowing that heat to move from a place of high temperature to a place of lower temperature in a manner that causes mechanical motion. Fuel cells can also be viewed as electrical heat engines: they create electrical rather than mechanical motion while channeling heat flow from high to low temperature points. The fuel efficiency of a heat engine has an upper bound that is defined by the basic Carnot cycle equation: *high ~ 1 low Upper limit on engine efficiency = (10.3) Thigh where T^is the temperature on the high side of the energy transformation machinery, for example, that inside a turbine engine, and TI OW is the temperature on the low side, for Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 365 example, on the outlet side of the turbine exhaust, all measured in absolute terms (degrees Kelvin). While the Carnot cycle and equation 10.3 above have many implications for DG engineering, the most important implications are: 1. The higher Thigh, the higher the efficiency. It can be raised by using a hotter-burning fuel, by burning the fuel under higher pressures, by burning a leaner fuel mixture (higher air/fuel ratio), or by insulating the combustion areas so the internal areas of the engine run hotter. 2. The cooler TI OW , the higher the efficiency. Heat engines work on a difference in temperature. Regardless of Thigh, the efficiency can be improved if heat can be "dumped" to a lower low temperature. Qualitatively, the relative efficiency of almost all types of DG, and the basic concepts behind various efficiency-improvement approaches, can be understood by applying this simple pair of "Carnot cycle rules." In almost all cases, whether one is talking about diesels, turbines, fuel cells, or even solar power for that matter, units that run at higher temperatures and "dump" their heat to lower temperatures produce more power from the same amount of fuel (or sunlight). Fuel Efficiency and Heat Rate Curves All fossil-powered generating units have an efficiency that varies as a function of output level. The heat rate curve, or efficiency-vs.-output curve, for a generator shows how many BTUs of fuel it requires in order to produce a kWh of electric power, as a function of the amount of power it is producing. Figure 10.7 shows heat rate curves for three DG units, in per unit form, to stress how the efficiency curve shape varies from one unit to another. Note that while curve shapes are quite different, each type - fuel cell, turbine, and reciprocating piston - has some point or range in which it is most efficient, where it requires slightly less fuel per kW than at any other output level. Heat rate curve shapes vary widely among DG units. A DG designer can alter the shape of a unit's heat rate curve by changing various elements in the machine's design, to make the same DG unit have different "optimal" operating points for it? intake, exhaust, fuel injection, generator, and others parts. Piston Engine Driven Distributed Generators Reciprocating piston engines produce rotating mechanical power by the use of pistons - essentially round seals that slide back and forth in cylinders within which a fuel is burned. The burning fuel forces the pistons out. A crankshaft converts that linear motion to rotation and, through its inertia, forces the piston to slide back and forth in the cylinder so the process can be repeated cyclically. Figure 10.8 illustrates the basic operation of a piston engine. The reciprocating piston engine provides the mechanical motion needed to drive a rotating electric generator. Reciprocating piston engines turn at roughly the speed of steam turbines (1,200-6,000 RPM), which is relatively slow compared to some gas turbines, particularly smaller micro-turbines (up to 100,000 RPM). This rather "traditional" speed of rotation for piston engines (similar to that of steam turbines) means that traditional types of generators are most typically used. Four types of generator can be attached to reciprocating Copyright 2004 by Marcel Dekker, Inc. 366 Chapter 10 Example Heat-Rate Curve Shapes 25 50 75 Output: Percent of Maximum Rating 100 Figure 10.7 Heat rate curves for three different DG units, in per unit scale of their maximum output, illustrate the different way in which efficiency varies as a function of output depending on the type and design of a DG unit. engines to produce electric power, synchronous AC, inductive AC, DC with DC/AC conversion, or written-pole synchronous AC generation. By far the most widely used is the synchronous AC generator - the same type used on "real" (large central station) generators. Alternating-current frequency is therefore controlled by controlling engine speed. Piston-engine generators use a gasomis, diesel, natural gas, or propane/methane powered piston engine to spin an electric generator, with the engine's crankshaft and the generator's rotor usually spinning at the same rate on the same or directly coupled shafts. Piston-engine DG is both the most popular DG unit at the present time and the technology that sets the performance/cost benchmark that other types of DG must meet to see any significant market success. Interestingly, it currently accounts for 95+% of all distributed generation on the planet, a figure that has remained stable, or slightly increased, in the past five years. Pistons may be an older technology compared to some other DG types, but they are still a viable, competitive technology. They will likely remain the power source of choice for very small (< 250 kW) electric generators in the foreseeable future. Reciprocating piston engines are a proven, mature, but still improving method to provide power for distributed generation systems. They have potential fuel economies as high as 45%, but their greatest advantages are a low-cost manufacturing base and simple maintenance needs. This last is often an overwhelming advantage: Among all the various types of DG, only a reciprocating piston engine - in particular a diesel engine - is so universally familiar that one can find someone to repair it virtually anywhere on the planet. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 367 Figure 10.8 The basic Otto cycle of most reciprocating piston engines consists of four strokes, each during the movement of the piston through a full back or forth motion that corresponds to one-half revolution of the crankshaft. This particular engine uses flapper valves to control intake and exhaust. Disadvantages of piston-driven generators are a general lack of good waste heat for co-generation applications compared to turbines, exhaust emissions, noise, vibration, and weight. All of these can be mitigated by various means and probably improved greatly in future designs. Most people think of piston engines for DG in terms of the engines that propel automobiles and trucks. Some DG engines are derived from automotive applications. The piston engines in automobiles typically have between four and eight cylinders, in which the pistons are about 2 to 4 inches in diameter, and reciprocate at up to 8,OOO RPM. But the piston engines used to drive larger piston-engine generators (diesel generators up to 50,000 kW have been built) have much bigger pistons, often more than 4 feet in diameter and weighing several hundred pounds each. Engines with such heavy reciprocating masses turn at speeds as low as 50 RPM, far below the RPM rate of smaller engines, partly because it requires too much mechanical strength to reciprocate pistons char heavy at hlgh speed B3ui more important, the exploding gas (fuel-air mixture) inside a cylinder will expand with the same characteristics whether in a small or large combustion chamber. A 5-foot-wide piston with a 5-foot stroke, turning at 150 RPM, moves at the same speed as a piston with a two- inch stroke, moving at 4,500 RPM, and is just as compatible with the burn characteristics of the fossil fuel. Thus, large generators turn at lower speeds, small ones at higher speeds. Low-speed motor-generators operate at 50 to 150 RPM. Medium-speed generators run at up to 800 RPM, and high-speed generators at 1,200 to 1,800 RPM. Note that all these speeds are far below the normal rate of rotation for automobile piston engines. Automotive derived DG units usually run at 1,200-2,400 RPM. Regardless, although there are exceptions, almost all piston-engine DG units use a constant-speed alternating-current generator and run at as close to a constant speed as possible. While proven by decades of use, and not as exotic as some other forms of DG, reciprocating engines have tremendous potential for further future improvement. The impending demise of the internal combustion piston engine has been forecasted by various proponents of newer designs for most of the last one hundred years. Its replacement has successively been predicted to be the turbine, various new designs or cycles involving rotors, impellers, vibrating fluids, or other approaches for heat-to- Copyright 2004 by Marcel Dekker, Inc. 368 Chapter 10 motion conversion, and the fuel cell. Every such prognostication overestimated the pace of development and advantages of the technology predicted to take pistons place, and underestimated the improvement that would continue in piston engines. No doubt, eventually, something will displace the internal combustion piston engine as the most widely used mechanical power source for small and medium applications on this planet. But frustrating as it has been for the proponents of rotating piston (e.g., Wankel) engines, small turbines, fuel cells, and other types of newer concepts, piston- engine technology always seems to stay just far enough in front to justify its continued dominance. And even at the beginning of the 21 century, there is no sign that this replacement will occur any time in the near future. Turbine Driven Distributed Generators A gas turbine generator uses a turbine spun by the expanding high-pressure gases of combustion to rotate an electric generator, as shown in Figure 10.9. Advantages of the gas turbine over all other direct-fueled engines include continuous combustion without reciprocating motion, low vibration, and very high power-to-weight ratio - lightweight turbine generators weigh only a sixth as much as reciprocating piston units of comparable output. Gas turbine driven generation has achieved significant market acceptance in the utility sector, particularly as central station peaking units, and for some larger distributed utility applications in the 10-25 MVA range. Some power producers have committed a significant portion, up to 1008, of their central generating capability to turbines. Gas turbines are also widely used in industrial power applications, in large petrochemical plants, paper mills, and other facilities that have significant power needs. Gas turbines are simple, compact, robust, but not outstandingly efficient devices which can be applied to turn electric generators in distributed generation systems. By varying the design, and using physics in different ways, relatively satisfactory fuel economy and durability can be obtained from turbines over a range of more than four orders of magnitude in size, from about 15 kVA to more than 150,000 kVA. Gas turbine generators fall into three categories of distinctly different design and operating characteristics. The easiest way to distinguish these categories is by size (physical or electrical output) as illustrated in Table 10.2, but utility planners must realize that size is not the only, or even the most sigmficant, difference among turbines in these different categories. Figure 10.9 Turbines use a continuous flow with pressure differences due to intake and output turbine wheels that segment the flow into four distinct regions. The first corresponds to the intake and compression cycles of a piston engine (Figure 10.8), the second to the power cycle, and the third to the exhaust cycle. The turbine produces a high-flow, low pressure exhaust flow (4) that is very suitable for CHP applications. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 369 All three categories of turbine work on the same principle, but use markedly different designs and different types of generators depending on their size. As was the case with piston engines, they vary in design according to size because the flame and gas-expansion characteristics of the gas energy are the same regardless of size. A two-inch turbine in a 20 kW micro-turbine generator must deal with the same expanding gas as a 100-inch turbine wheel in a 150 MW turbine. As a result, the former has to spin at 90,000 RPM to match the larger turbine's blade tip speed at only 1,800 RPM. The largest category is the traditional utility gas turbine generators that range from about 10,000 kVA peak output to more than 150,000 kVA, with a typical unit being about 70,000 kVA. This is the only one of the three categories for which the turbines were designed specifically for electric power applications. Turbines in the two smaller categories are invariably modified versions of turbines originally designed for other applications. In the middle of the size scale are mini gas turbine generators, units in the range of 800 kVA up to about 10,000 kVA. Most of the turbines in this category are based on designs first produced for small aircraft, coastal and military patrol ships, helicopters, or battle tanks. As such, when used for DG, they are not quite as efficient and do not have quite as favorable maintenance and operating costs as the larger turbines designed solely for utility use. On the other hand, many are derived from military designs, which mean they are robust, durable, and capable of operation under very adverse ambient conditions. The smallest category of gas turbine generator, and the one that has gained tremendous cachet as a "revolutionary concept" that will alter the electric industry entirely, is the micro-turbine generator. Units in this category range from less than 20 kVA up to about 750 kVA. The turbines used in these units were mostly designed originally for vehicular application, and many were not originally designed as turbines, but as the turbine in a turbo-charger for a large piston engine. Table 10.2 compares these three categories of turbine generator. More discussion of their design, operation, and similarities and differences can be found in Willis and Scott (2000). Table 10.2 Comparison of Gas turbine Generator Categories Characteristic Available range - kVA About the size of ... Original design based on ... Most typical fuels are . . . Makes about as much noise as Out of service once every . . . Turbine generator usually is ... Turbine spins at about . . . Generator type used is ... Generator turns at about . . . Turbine & generator run at ... Best fuel efficiency is about . . . Can be bought and installed in . Typical cost/kW is ... Micro 20-500 a refrigerator bus, tmck engines nat. gas, diesel a car at 40 mph two years single shaft 70,000 RPM DC with AC conv. 70,000 RPM variable speed 32% a week $950/kW Mini 650-10,000 a large truck aircraft engines nat. gas, diesel 3-4 leaf blowers eight months two shaft 15,000 RPM AC sync. 3,600 RPM constant speed 30% two months $450/kW Utility 12,500-265,000 a building utility needs nat. gas, fuel oil a jet plane year and a half two or three shaft 1,800 RPM AC sync. 1,800 RPM constant speed 37% a year or two $265/kW Copyright 2004 by Marcel Dekker, Inc. 370 Chapter 10 Fuel Cell Powered Distributed Generation Fuel cells take a unique approach to using fossil fuel for producing electricity. Instead of burning fossil fuel to produce motion to drive a generator, they oxidize the hydrogen in a fossil fuel in a chemically controlled (catalyst-driven) process that causes ion migration through an electrolyte, creating a direct electric current and, hence, electric power. If designed and operated appropriately, they are nearly silent, very efficient, and produce virtually no polluting emissions. Fuel cells fall into five categories depending on the chemical basis for their operation. Ranked in ascending order of internal temperature (and thus potential Carnot-cycle efficiency) they are: proton exchange membrane fuel cells (PEMFC), alkaline fuel cells (AFC), phosphoric acid fuel cells (PAFC), molten carbonate fuel cells (MCFC), and solid oxide fuel cells (SOFC). AFCs are unsuitable for electric power application and used almost exclusively only in spacecraft. Alone among fuel cell types they require pure oxygen and cannot not "run" on air. The four of the remaining fuel cells types are potentially viable for at least certain niches in the DG field. In particular, PAFC, MCFC, and SOFC seem suitable for stationary power generation applications. Fuel cells are simple in concept, but very complicated in actual execution. High fuel-to- electrical efficiency comes from high internal temperatures, on par with or higher than inside combustion turbines. This means they must be robust designs made of high- temperature materials. In all of them, the fossil fuel needs to be "reformed" - stripped of its hydrogen, which the fuel cell needs, while the rest of the fuel's contents is discarded (exhaust). Reformers use a type of catalytic converter to do this. Both reformer and fuel cell operate best under high pressure, requiring (high temperature) sealing and pumps for both fuel and air. The power is produced at high, direct current and low voltage and needs to be converted to AC at utilization voltage. All of this results in a machine that is actually quite complicated. Maintenance needs are perhaps no more than those needed by piston-engine DG, but they are very different: One can find a diesel engine mechanic just about anyplace on the planet, but fuel cell service persons are extremely rare. For over three decades, fuel cell technology has been developing and improving, always promising widespread commercial success "just around the corner." Despite considerable work in the last decade, the three barriers to widespread fuel cell application are still high cost, lack of proven durability, and a need for a large, different service infrastructure (the aforementioned "fuel cell mechanics," among other things). Solutions to the second and third will follow automatically upon the first, so it all boils down to whether durable and reliable fuel cells can be produced at a competitive price. This is being addressed by government, industry, and manufacturers, but remains a challenge that no one seems able to meet. For some applications - those having particularly sensitive environments in which noise, vibration, or emissions are a major concern - today's fuel cells are perhaps best for DG applications. But for most distributed applications they are too expensive and, frankly, not yet proven sufficiently for widespread application. Exhaust Heat for Co-Generation and CHP Purposes All fossil-fueled generators produce hot exhaust gases which contain a good deal of remaining energy which can, under some circumstances, be harnessed to provide additional value. Turbines and high-temperature fuel cells (SOFC) are best in this regard, producing hot exhaust gases that are useable for both additional power applications or other purposes. However, piston-engine units are sometimes fitted for this purpose. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 371 Generators that produce additional power from the exhaust gases are called combined cycle units. The exhaust of some turbines is used to boil water to produce steam to power an auxiliary steam turbine in what is called a combined cycle turbine. A Cheng cycle turbine routes the steam back into the gas turbine, where it mingles with the fossil fuel gas. Both types can approach 45% fuel-to-electric efficiency in annual operation. Another option is to use the exhaust heat to boil water for hot-water purposes or for an industrial process. Paper mills often use gas turbine generator exhaust for their pulp boiling. In fact, often paper-mill turbines are designed as a balance between electrical and exhaust water heating applications tailored to the specific plant. DG units are often used to heat water for hot-water applications at smaller industrial plants and in commercial and multi- family residential applications. It is also possible to use absorption cooling equipment to use the waste heat for cooling. Specific piston and turbine "CHP" (cooling-heating-power) units have been designed specifically for office and apartment building applications. A good CHP application - one where the power, hot water and heating, and cooling outputs match the need of the site - can have an overall fuel to end-use efficiency approaching 80%. This makes CHP-based DG so cost-effective that very little can compete against it on the basis of overall cost. Renewable Energy: Wind and Solar Renewable energy resources offer a potential for power production using solar, wind, and a host of other power sources derived from ongoing natural processes of the earth's environment. By their very nature renewable power sources are small, modular, and geographically distributed. For these reasons, and because they connect to the lower voltage parts of the grid, they are often classified as DG. Renewable power systems require no source of fossil or nuclear fuel, as in the case of gas, steam turbine, reciprocating piston, and fuel cell generation units. Some do require burning of a fuel: garbage, animal waste (methane), and bio-mass among them. Often, the motivation for building renewable generation is not to add local peaking support or reliability backup, but to obtain "green" energy production. Regardless an important advantage for remote location applications is that "fuel" is delivered to the site at no charge. For the significant portion of small communities worldwide that use diesel generation for electric generation, the delivery cost of fuel exceeds its actual cost. In addition, environmental impact is deemed low in most cases, and the fact that these power sources are sustainable without using any natural resources is appealing to many persons. Most renewable generators make far less environmental impacts than fossil fuel and nuclear power generation, but are less cost-effective as well. Many are subject to some degree of unpredictability in their energy availability and hence the net power output, and using them to provide dispatchable power depends on combining them with some form of energy storage. Many types of renewable energy also have site requirements that constrain them to locations with the right combination of natural factors. Hydro-generation is by far the most widely used, proven, and cost-effective renewable energy source. Where one has access to continuously moving water, some degree of power generation can be produced. Even small (250 kW) hydro units can be efficient and reliability if well designed, and can produce power that is cost-competitive with power form the grid. Of course, few consumers have access to the requisite river, stream, or waterfall, and there are often legal restrictions on the use of such sources for any purpose, generation included. But for farmers, ranchers, and rural homeowners who do have such sources available to them, small hydro power is an option, particularly if the water source has a rather continuous flow year around and if a small reservoir or pond is Copyright 2004 by Marcel Dekker, Inc. 372 Chapter 10 available to provide "energy storage" in the form of an upstream reservoir. 17 Solar power. Among all possible renewable energy sources, the most flexible and applicable in many respects is solar power. The fuel costs nothing, but is available only during the day. However, unlike wind generation, solar power's "fuel availability" is quite predictable (well-designed solar units will produce power even when the sky is overcast). But daily sunlight cycles remain a major constraint on design and duty cycle, as will be discussed in this section. Photovoltaic generation converts light energy directly into electric power, using any of several types of flat semiconductor diodes built with as large an area as possible and with their p-n junction located very close to the exposed surface. Solar electric cells produce DC electric power at low voltage, typically around 0.5 volt, whenever exposed to sufficient light. A single cell may be less than a square centimeter in size, and produce only a small amount of power, often much less than a watt. Usually many cells are connected in series to provide higher voltage, and in parallel to produce higher current, in what is called a photovoltaic array. As was the case with fuel cells, the DC output of PV solar cells must be converted to AC power by inverter/filter equipment if they are to be connected to the power grid or used to power AC appliances. For very small PV systems, the cost of this conversion equipment can be a significant portion of its cost. PV is a proven technology with more than 200,000,000 PV arrays in operation worldwide - the vast majority of them in calculators and similar small appliances. Solar thermal conversion systems use mirrors to concentrate reflected sunlight to produce intense heat energy which is used to transform water into steam, which in turn drives a steam-turbine generator. Larger (5 MVA and up) systems store great amounts of heat overnight so that they are dispatchable power sources. Smaller solar thermal systems use a Stirling cycle reciprocating engine, and have no storage. Either way, solar thermal units usually employ an AC generator, which means they have electrical output characteristics identical to traditional generation sources. In this respect, they represent a proven and risk-free technology, well understood and for which maintenance and operating experience is considerable. Overall cost is generally two to three times that of fossil fuel generation, but for remote applications where fuel delivery is costly, these simple, robust, and easy-to-repair systems are worth consideration. Wind power generation harnesses the energy in wind to drive electric power generators, using some form of "wind turbine" - essentially an optimized windmill driving an electric generator. Individual wind turbines fall into the DG size range, being between 75 and 3,000 kVA output. However, most commercial applications include groups of these turbines arranged in "wind parks" of between 25 and 200 MVA total, and are treated as "green" central station power plants. Something on the order of 8,000 MW of installed wind turbines were in commercial operation worldwide, with over 1,000 MW of additions being made every year, nearly all of it in wind parks. The advantages of wind generation are no fuel cost, a total lack of exhaust emissions, modularity with a fairly linear power vs. cost relationship for large-scale (wind park) installation, a very robust, simple, proven technology, and potentially a 24-hour per day supply of energy. Disadvantages are somewhat higher cost than for fossil fuel generation, temporal unpredictability of energy production, as well as environmental impacts generally considered to be greater than solar power's. 17 Even a small reservoir, storing a day's worth of water upstream, converts a non-dispatchable "run of river" generator to something that can be relied on to track load well, improving its value as a generation source immensely. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 373 Production cost of the resulting electric power depends greatly on location, local labor costs, and other factors. Most likely, small wind turbines will not be competitive with small fossil-fueled gas turbine generators and other similar distributed generation for sites located on distribution systems of major utilities, but large (3 MVA and up) units will become cost competitive when built in large wind parks in optimal wind areas. That said, the reader should recognize that there is a large market for wind and solar energy in developing countries with rapidly growing loads, because fossil fuel carries additional costs along with it, not often included in direct comparisons of price. Many developing countries simply do not want to import fossil fuel even if affordable wind energy not only defers expenses, but improves their balance of payments. Finally, in developed nations such as the United States, electric services providers are discovering a considerable market for "green power." A good portion of the residential market, perhaps up to 10 percent, will pay a premium for power which has been produced by environmentally benign, renewable power generation methods. Wind turbine power is one viable means of producing this green power in "utility" amounts. Hydro-electric power is a proven and mature technology for electric power production. While most hydro plants have a substantial head (difference between high and low water levels on opposite sides of the water turbine), it is possible to generate electric power in DG amounts with a difference of only a dozen feet and a flow equal to that of many small rivers and large streams. Low-head hydro has been evaluated as feasible and economically competitive at many locations throughout North America, and no doubt large amounts are similarly available worldwide. Hydropower plants, low-head or otherwise, provide good 24-hour-a-day power production. Most low-head installations are "run of river" plants. They do not have a substantial reservoir of water storage behind them to vary output, but instead work with whatever flow the river provides. 18 Thus, their output varies on a seasonal basis and from year to year according to the amount of rainfall upstream. Making a low-head hydro site dispatchable means building a reservoir behind it and making at least minor, and perhaps very major, changes in the water levels and coverage of lakes and rivers. This raises concerns about their environmental impacts on aquatic life and the changes wrought on the natural "cleansing" of the waterways by spring floods. These concerns have limited application of low-head hydro in many areas and in fact have led to initiatives by environmental groups to remove existing small hydro plants in states like Maine. Regardless, low-head hydro is a very proven, robust, and feasible renewable alternative for renewable DG in many areas of the world. 10.5 ELECTRIC ENERGY STORAGE SYSTEMS Energy storage can often augment DG in three ways. First, energy storage can be used for stabilization purposes, permitting the DG to run at a constant, stable output level, even if the load fluctuates greatly and rapidly. Second, proper amounts of storage can provide energy to ride through periods when the DG unit is unavailable, for example, during the nighttime for solar power, or when the DG unit of any type is being maintained or repaired. Third, energy storage can permit a non-dispatchable DG unit to operate as a dispatchable unit by permitting its output at any moment to differ from the power being released to the demand or into the grid. Table 10.3 lists these three purposes of energy storage and the key 18 A large hydro plant may have a substantial water reservoir behind its dam, storing water equivalent to several hundred MW-years of net electrical power. This permits it to be dispatched and the ability to largely disregard short-term shortfalls in river flow. Copyright 2004 by Marcel Dekker, Inc. 374 Chapter 10 characteristics that distinguish them from one another. Energy storage has other uses when applied without DG, either as peak shaving systems for utilities or as UPS backup for customer applications, but those are not emphasized in this chapter. This section begins by examining the application of energy storage to support DG, what it provides and why and when that can be of benefit. It then examines the various storage and operating qualities needed in an energy storage unit in order for it to perform those functions, and compares the various storage methods available. Energy Stabilization Figure 10.10, top, shows the typical load behavior encountered by a DG unit serving a single household load. This non-coincident load curve (see Chapter 3 for a detailed discussion) varies rapidly from over 10 kW to less than 1 kW, many times during the course of a day, almost instantly shifting from high to low state and back again shortly thereafter. The large shifts are caused by major appliances such as space heaters, air conditioners, water heaters, electric dryers, and the like, which switch themselves on and off automatically under their thermostatic control. The short duration of the "needles" is due to the nature of thermostatic control. Seldom do these appliances run for a long period of time, as if on a contiguous basis: temperature is kept within a narrow band by running the unit (e.g., a water heater) for only short periods of time, but frequently, never letting the temperature rise or fall too far. Table 103 The Three Major Applications of Energy Storage with DG Planning Reason Why Energy Storage Is Being Applied in Conjunction with DG System Aspect Energy Stabilization Ride-Through Dispatchability Reason Shave needle peaks in the non-coincident load curve due to large appliances, etc. Benefit Lowers peak DG capacity needed. Improves voltage regulation. Storage Must be enough to "shave" appliance peaks and meet then- short-term needs. Peak Relatively great: all the energy stored must be re- leased in just a few minutes. Method Based on detailed assess- ment of daily load curve, on a minute-to-minute basis. Design Typically high-energy, low- storage design with enough capacity to avoid deep cycle. Provide energy to serve load during periods when DG output is unavailable. Service from PV, etc., can now be maintained during nighttime, etc. Dictated by load during "DG unavailable" times. Usually Vi day's energy. Relatively small, only one- eighth to one-tenth of stored energy. Based on hourly analysis of load needs over a year and DG availability stats. Must achieve size balance between storage size and andDG. Provide energy stored to stabilize DG availability to meet various schedules. DG owner can now bid and sell power contracts for arbitrary schedules. Must be enough to transform the DG schedule into the de- sired sales schedule. Requires more than for ride- though but much less, rela- tively, than for energy stab. Based on hourly analysis of desired schedules, DG avail- ability stats, business cases. Must achieve an overall balance among DG unit size, storage size, and total cost. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 375 Single Household Mid. 6 Noon 6 Mid. Time of Day Coincident Load Curve 10 Mid. 6 Noon 6 Mid. Time of Day 10 Single Household "Filtered" by Storage Mid Noon Time of Day Mid. Figure 10.10 Top, non-coincident load of a single household, showing needle peaks due to appliance activity. Middle - coincident "individual household" load curve, l/100th of 100 such homes, is smoother because random needle peaks of the various customers in the group "cancel out." Bottom, non-coincident curve "filtered" by being put in parallel with a battery that fills in the needle peaks, averaging usage over 30 minute periods. Copyright 2004 by Marcel Dekker, Inc. 376 Chapter 10 Nearly all small electric consumer sites display some such behavior, with needle peaks and rapid shifts in demand level (see Chapter 3). A DG unit that will successfully serve such a load curve must have capacity greater than the high needle peaks. Power quality problems may develop anyway simply because the unit may not be able to react quickly enough to shift its output as rapidly as the demand shifts. Needle peaks occur during all seasons of the year and times of day: they happen more frequently during peak demand periods, but they occur less frequently, but potentially with just as high a load, during off peak periods, too. Equipment serving large groups of homes does not see this type of load curve, as shown in Figure 10.10, middle. The "needles" and "valleys" of individual customers in the large group cancel out, leaving a coincident load curve. The DG unit will see something akin to a coincident load curve as its load, if the load is connected in parallel with energy storage, so that the energy for the needle peaks is drawn from the storage unit, not the DG unit, with the energy "paid back" by the DG during the next valley. This permits the DG unit to run on a smooth, steady schedule, as shown in Figure 10.10, bottom. The energy storage filters, or "smoothes out," the non-coincident load curve. It permits a generator to have a smaller capacity than the needle's magnitude and yet still serve the load with good power quality. The reader who has not gone through this book thoroughly and finds the needle-peak behavior or appliances in Figure 10.10 puzzling can look at Chapter 3's comprehensive discussion of coincident load behavior and why and how it occurs. Ride-Through Capability Solar power cannot provide power when the sun is not shining. Wind turbines cannot provide power when they are becalmed. Even a small hydro may "run dry" in the seasons between rains. No DG unit can provide power when disassembled for maintenance. The DG owner who needs power has two choices during these periods: Use a backup source of power, such as another DG source or the utility grid Provide for one's needs with energy stored when the source was available The second option is often the most appealing, because another DG unit might not be any better (more solar power will not help at night, a fossil unit is unacceptable), and the grid may not be an available or economical option. The energy storage required to address these needs is quite different from that for energy stabilization. To begin with, the total amount of energy needed to be stored is much greater. For example, a system storing 4 kWh (equivalent to what is stored in two heavy-duty "car"- type lead-acid batteries) will stabilize the DG unit serving the individual household load shown at the top of Figure 10.10, so that it sees essentially coincident load behavior. But to make it through a long winter night without power, the DG unit may need to store over 50 kWh. Beyond this, during those periods when it is the sole source of power, the battery will have to be able to meet all the needle peak demands, not just help augment a DG unit. It will need a greater peak capacity, too. Generally, the DG planner will evaluate these energy storage needs using a load duration curve or an hourly simulation analysis, which would examine how often over the year, for how long each time, how much energy would have to be stored in the storage unit for it to do its job. This analysis would also make certain that there is enough DG capacity to "charge" the energy storage unit during off-peak periods. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 377 Dispatchability Energy storage is also needed if a basically non-dispatchable energy source, such as PV, wind, or solar thermal, is to be applied in a dispatchable manner. Viewed purely from the standpoint of the value of the energy, dispatchable energy is worth more that non- dispatchable energy. Dispatchability permits the DG owner to commit in advance to certain production schedules, to provide a promised amount of power at a certain time, regardless of how much power the DG unit is producing at that time. The energy storage needs, as well as planning and analysis concepts for achieving dispatchability, are conceptually similar to those discussed above for ride-through capability. First, there is a load curve, or production schedule that must be met. Then, a sufficient amount of storage capacity, and peak capability, must be arranged to reliably serve this, given the expected production schedule of the DG unit. The differences in planning storage to achieve dispatchability, versus only ride-through capability, are related to more complex and demanding definition of "success" and exacerbated by the planner's knowledge (or lack of it) of the expected production schedule which the dispatchable generation must meet. In planning ride-through, the planner generally has a good idea of the load to be served: one household load. Often detailed data on that load and its behavior over time are available, but even when they are not the target is clearly identified. By contrast, dispatchability means freedom to vary production schedule to meet a variety of schedules by changing output times and amounts. The DG- storage planner must answer the question "how much freedom is needed in being able to control and shape the net output power schedule?" Once that has been determined, a load duration curve or hourly load curve analysis of the extremes of the range desired can be carried out. Second, the planner studying dispatchability has a more complicated "definition of success." Ride-through planning had a simple goal: serve the load with no more than some small probability of failing to meet it. The storage need is dictated by a requirement to serve the load schedule. By contrast, in planning dispatchability the final decision usually becomes an economic balancing act: the more the planner spends on more storage, the more dispatch freedom the resulting system will have, and the more it can earn selling its power. The question is, where is the best payoif? At this point, the planning becomes a business planning situation, beyond the scope of this discussion. Performance Tradeoffs Energy storage units have eight major areas of performance which can, to a certain extent, be traded against one another: Energy density Power density Electrical efficiency Re-charge rate Control system Service lifetime Physical dimensions Cost Each type of storage technology carries with it a different interaction in how compromises must be made among these performance categories. But all have some interrelationship: obtaining more of one means something must be given up in the others. Energy density is a measure of the basic capability of the system: How much energy (kWh) can it store in a specific space - 5 kWh, 50 kWh, or 500 kWh? Power density is the amount of power the unit can give up, per its unit size. A particular storage unit that stores 100 kWh might be able to provide power at 10 kW per hour, meaning it would take ten hours to discharge. Another with the same capacity might be able to provide power at 200 kW, twenty times that rate, fully Copyright 2004 by Marcel Dekker, Inc. 378 Chapter 10 expending its stored energy in half an hour. This is usually measured in kilowatts, or sometimes megajoules. Electrical efficiency is the percent of power pushed into the unit that is available to be withdrawn. A unit with 90% efficiency returns 9 kWh of energy for every 10 kWh banked in its storage. Efficiency of most electrical energy units is a function of both the length of time energy is stored (as a result efficiency is always greater overnight than, say, over a week). Re-charge rate is the rate at which power can be pushed into the unit for storage. A particular 100 kWh storage unit might be able to provide power out at 10 kW, but accept power in at only 7 kW. This means that while it takes the unit only 10 hours to fully discharge at its maximum rate, it takes it a bit over 14 hours to be completely refilled. Control system design determines the degree to which the energy storage system can control the voltage and power quality of the AC power it is providing. Storage systems such as batteries vary their DC output as a function of both the amount of stored energy they have remaining and the rate of power flow they are providing at the moment. But this variable impedance of the battery can be compensated for by proper design of the AC power converter, which can also be designed so it provides harmonic-free, high-quality power. Service lifetime varies greatly among available technologies and depends very much on how the storage units are used. Lead-acid batteries are particularly notorious for having short lifetimes in applications where they are repeatedly charged and discharged completely. Physical dimensions and physical efficiency. The size and weight of a storage system are often important, particularly in automotive or shipboard applications, but also in electric power applications where the issue is whether a sufficient storage capacity will fit at a site. Weight is sometimes a factor in transportation and structural requirements. Cost includes both initial and continuing operating costs. Every storage technology has an initial cost covering the system itself and its installation. Most require periodic inspection, maintenance, and perhaps re-calibration and replacement of key parts (bearings and seals) at periodic intervals. Some storage systems simply need to be replaced every so often. 10.6 DISTRIBUTED RESOURCE COST EVALUATION Traditional Regulatory DSM Evaluation Frameworks Traditionally, DSM was viewed in all cases as an optional resource that would be added if it provided more benefit than its cost. Therefore, regulatory-driven DSM programs were typically evaluated by utilities and commissions alike using a benefit to cost (B/C) ratio analysis. Programs that had a higher benefit than cost were identified as feasible. Among similar DSM options, those with the higher B/C ratio would be the best choice. In selecting a few DSM options from a large "shopping list," the B/C ratios can be used to rank-order the DSM programs. Similarly, consumers, both homeowners and business owners, make decisions about their energy management options based on comparing the benefit and the cost. There are substantial differences in the details of these value systems. Utilities and commissions in the Copyright 2004 by Marcel Dekker, Inc. Total Resource Cost (TRC) Test Distributed Resources 379 "DSM era" - the 1980s and 1990s - usually took a long-term (equipment lifetime, or 30- year investment depreciation) perspective. Most consumers use a very short, "payback period" perspective: "Does it pay for itself in two to three years?" Modern utilities vary, but most plan with a shorter horizon than in the 1980s - often only ten years. 19 The benefits of DSM are the avoided costs of supply (generation, transmission, and distribution capital costs and the costs of energy) and end-use (consumer costs). In a B/C assessment, the avoided costs represent a savings or benefit. Avoided costs are typically expressed in $/kW for capacity-related costs only and in 0/kWh for total costs (capacity and/or energy components). On the other side of the B/C ratio, costs are the expenses associated with a DSM program and may include the costs borne directly by the participants as well as the utility. These costs include such items as the technology (materials and installation), maintenance, and administrative expenses. But details of just what benefits and costs are included varied in the DSM era, and may be important today, to some utilities. There are two major types of cost-effectiveness tests which are utilized to assess DSM options: Rate Impact Measure (RIM) Test The TRC test focuses more on minimizing the cost of electricity services while the RIM test focuses more on minimizing electricity prices to consumers. Each approach represents a different approach. Total resource cost methods label a resource option as cost-effective when the present worth of all benefits over the planning period, no matter who receives them, exceed the present worth of all costs no matter who pays them. The allocation of costs and benefits between the utility and energy consumers - who pays and who gains is not a consideration when applying the TRC test. Neither are questions about whether all, or only a few, consumers win. If overall more win than lose an energy management program is considered be "good." In some regulatory jurisdictions, the TRC test is referred to as the "All Ratepayers Test." The benefits included in the B/C evaluation are all incremental direct savings accruing from the program including consumer savings and the electric utility's avoided costs. Costs include the electric utility's costs related to the program and the total of all consumer costs, if any. Any loss of revenue by the utility is not included. Rate Impact Measure looks at how an energy management program affects the utility rates or the customers total bills (the two are not the same, some states used one perspective, others the other). Functionally, it differs from TRC mostly in that it considers changes in the utility revenue as well as costs. For example, suppose that a utility spends $5,000,000 to implement a program, thereby producing avoided costs for the utility of $6,000,000. Then it has a positive B/C ratio as evaluated by the TRC test. But suppose that the program cuts energy usage of consumers so that the utility loses $7,000,000 in revenue per year. This means that it will eventually have to raise its rates - there are fewer kWh of sales to cover the total of its costs. 19 The reason is uncertainty about future regulation and technology, the future regulatory backdrop is not yet completely defined, utilities are naturally unwilling to make long-term commitments until they know they will be allowed long-term earnings. With respect to technology, utilities learned in the 1980s and 1990s that technological obsolescence is a very real concern with respect to control and automation systems (as in many energy management systems). Quick payback of three to five years - within one or two "technology half-lives" - is the basis for their evaluation. Copyright 2004 by Marcel Dekker, Inc. 380 Chapter 10 The rate impact measure would label such a program as "ineffective" even though it had a B/C ratio greater than one. The basis for this perspective is the program's impact on consumers who didn't or couldn't participate. Suppose the program was "swimming pool pump" control. A homeowner who does not have a swimming pool cannot participate, yet as a result of this program sees his electric rates increase slightly. Modern utilities generally prefer to use a business-case approach for evaluation of distributed resource programs. There, all utility costs, including loss of revenues, are weighed against all utility savings. If B/C ratio is greater than one, and risk from regulatory, technological, and other sectors is considered appropriately small, the program will be approved. Regulators may have concerns about this, particularly if there would be any rate inequities of the type that the RIM test identifies or if the program produces additional costs for consumers not taken into account by the utility. Consumers, who define the market-driven nature of energy management, evaluate energy management on the basis of how it benefits them alone. Their savings versus their costs. This is fine, and something everyone has to accept, but it is interesting that this could have negative rate impacts: if many consumers implemented measures that made sense to them, which reduced their utility bills, it might have enough revenue-erosion impact on the utility that it would require it to raises its rates. Market-driven energy management programs often fail the RIM test. Chapter 6, section 6.4, presents some more detail on TRC and RIM tests. The various methods of conducting cost-effectiveness testing should be considered as different perspectives or various means of assessing a DSM program. Each such measure provides some alternative insight to the feasibility of a program. The final selection of any distributed resource programs depends on who is making the decision, how wide they cast their net on defining costs and savings, and whether they look at only their benefits and costs or those of some of the possible players or those seen by all of society. c o. il X HI (0 Q> Timing of Expense initial Continuing One-time "first" costs that do not depend on usage pattern One-time "first" costs that do depend on usage pattern Periodic, on- going costs that do not depend on usage pattern Periodic, on- going costs that do depend on usage pattern Figure 10.11 Two characterizations, one based on when costs occur and the other on how they vary with usage, result in four categories of costs. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 381 Cost and Benefit Evaluation of Distributed Resources The most important goal of cost evaluation is to include a comprehensive and consistent (comparable) assessment of all costs involved, both for every DR alternative and for any T&D options to which they are being compared. Inclusion of all costs is much more important than in traditional T&D planning studies, because DR options often differ significantly among themselves, and from traditional T&D options, in categories that are ignored or "factored out" of traditional T&D planning studies. One example is O&M costs, which are rarely included in typical T&D studies, because O&M being about the same among all options is usually left out of consideration. However, O&M differs substantially among various DG, DR, and T&D options, and thus must be included in comparisons that cross the DR-T&D boundary. Similarly, traditional T&D options almost never include the "cost of energy" because, regardless of choice, it is usually the same for all options. In DG and DR studies, energy (fuel cost) often varies a great deal, and avoided electric purchases (advantages to consumers, lost revenues to utilities) must be included. Four Cost Categories Costs are characterized in two ways depending on whether they are initial or continuing costs and depending on if they are fixed or variable (Figure 10.11). Initial costs must be dealt with before a DG unit or energy resource can be used. For example, the DG unit itself must be obtained, a site must be prepared for it (foundation, sound abatement walls, etc.), fuel delivery systems installed, electrical connections made, etc. Costs associated with accomplishing all these tasks are the initial costs. Continuing costs are those associated with keeping the unit available and in service. These include taxes, annual inspections or certifications, fuel, repairs, labor costs for service personnel, etc. Usually, these are periodic - monthly, annual, bi-annual, every five years, etc. - expenses that continue as long as the unit is left in service. Fixed costs do not vary as a function of the amount or patterns of usage. The cost of the basic DG unit itself is usually a fixed cost: One must buy or lease the unit regardless of whether it is used every hour of every day or not at all. 20 Likewise, annually required inspections or certifications, as well as annual property taxes, are based on the value of the property and are "fixed," even if they vary year by year, because the unit depreciates in value, etc. 21 Variable costs are those that vary as a function of the amount or type of usage. Fuel is a big component of variable cost: the more power produced, the more fuel used. Other variable costs include certain O&M - those that increase or change depending on the amount of usage. 20 Of course, a lease could be based on usage - in a manner similar to leasing a car on a "per mile" basis - which would make the DG unit cost a variable cost. 21 This is a subtle but important distinction. In some municipalities, local property tax on machinery is based upon assessed value, which decreases with equipment age, and thus the annual taxes due on a DG or other machine decrease each year. Although these costs do change from year to year, they are not variable costs, because they do not alter as a function of usage. Once the user decides to buy and install the DG unit, he has committed to paying all those future taxes, a "fixed" amount regardless of his usage. Copyright 2004 by Marcel Dekker, Inc. 382 Chapter 10 Table 10.4 Example DG Cost Analysis - 1,100 kW Reciprocating Natural-Gas Fired AC Generator Unit Used 3,000 Hours per Year Initial Fixed Costs Cost of DG unit itself Tools and diagnostic equipment Shipping & insurance Site prep., foundation, control house Fuel delivery lines, equipment, meters Backup tank and equipment (10,000 gal. propane storage and propane rollover, supply good for 100 hr. operation) Electric controls, protection (non-grid operation) Design, construction permit, survey, inspection fees, etc. Backup fuel, 10,000 gal at 800 Total "Cost of the Unit" Initial Variable Cost Legal and fees for emissions permit for "more than 1,001 hr./yr." operation TOTAL INITIAL COST Annual Fixed Costs Property taxes, .70 per dollar on $550,000 assessed value Annual mechanical & electrical inspection (req. for warranty) Annual Variable Costs Fuel: natural gas at $3.05 per MBTU 12,000 BTU/ kWh heat rate, for 3,000 hours at full output Maintenance, estimated at .30 per kWh Tear-do wn/rebuild every 10,000 hours (every three years in this case) $72,000 ANNUAL COSTS 3,300,000 kWh/yr. ($431.10/kW) $474,200 $7,800 $8,200 $15,000 $13,500 $50,000 $18,500 $8,400 $8,000 $603,600 $3,200 $606,800 ($551.64/kW) $3,474 $1,370 $120,780 $9,900 $24,000 per year $154,680 (4.68# kWh) Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 383 Table 10.4 shows the costs associated with a hypothetical, but very realistic, V-16 natural gas/propane reciprocating piston driven AC generator capable of producing 1,100 kW output (net) serving up to 1,100 kW of peak load. Although this particular DG technology lacks the cachet of micro-turbines and fuel cells, it is used here to emphasize that pistons should not be "counted out" as an equal, or perhaps even superior, DG technology for application in many cases. The table gives costs for this unit, identified as fixed and variable, as well as initial and continuing, for an application involving an expected 3,000 hours per year of full (1,100 kW) output, in an isolated (stand-alone) application - i.e., not connected to the utility system. This is based on an actual case for a ranch in an isolated area of the pampas in South America which needed power for a series of 75 kW irrigation pumps to be run only about 3,000 hours per year during the crop growing season. The examples in this and subsequent chapters used a gas price of $3.05 per million BTU. This is a lower price (by about 20%) than the annual market price at the time of this writing, but thus represents a slightly optimistic scenario. Assessing Initial and Continuing Costs How costs are to be considered depends on the financial decision-making context of the evaluation (see Chapters 6 and 29 and the earlier part of this section). Table 10.5 shows 20- year operating cost for the DG unit from Table 10.4 when run 3,000 hours per the irrigation schedule expected at the isolated ranch where this unit is being planned to power a series of electric pumps used only during the growing season. Total outlay over 20 years is $3,797,280, production 66,000,000 kWh, or 5.750 per kilowatt hour if undiscounted cost and undiscounted production are used to compute cost/kWh. The cost per Table 10.5 Twenty-Year Evaluation of Table 10.4's DG Unit for 3,000 Hours per Year Operation at Full 1,100 kW Output Study Year Unit &Site 0 $610,274 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 $3,474 MainL &Insp. $11,270 $11,270 $83,270 $11,270 $11,270 $83,270 $11,270 $11,270 $83,270 $11,270 $11,270 $83,270 $11,270 $11,270 $83,270 $11,270 $11,270 $83,270 $11,270 $59,270 Fuel Cost $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 $120,780 TOTAL Evaluated cost = 5.75 cents per kilowatt Annual Cost $742.324 $135,524 $207,524 $135,524 $135,524 $207,524 $135,524 $135,524 $207^24 $135,524 $135,524 $207,524 $135,524 $135,524 $207,524 $135,524 $135,524 $207,524 $135,524 $183.524 $3,797,280 PW Disc. Factor Cost 1.000 0.900 0.810 0.729 0.656 0.590 0.531 0.478 0.430 0.387 0.349 0.314 0.282 0.254 0.229 0.206 0.185 0.167 0.150 0.135 $742,324 $121,972 $168,094 $98,797 $88,917 $122,541 $72,023 $64,821 $89,332 $52,505 $47,254 $65,123 $38,276 $34,448 $47,475 $27,903 $25,113 $34,609 $20,341 $24.791 $1,986,661 NetkWh Production 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3.300.000 Discntd. Production 3,300,000 2,970,000 2,673,000 2,405,700 2,165,130 1,948,617 1,753,755 1,578,380 1,420,542 1,278,488 1,150,639 1,035,575 932,017 838,816 754,934 679,441 611,497 550,347 495,312 445.781 66,000,00028,987,970 hour (undiscounted) or 6.85 cents (discounted) Copyright 2004 by Marcel Dekker, Inc. 384 Chapter 10 Table 10.6 Twenty-Year Evaluation of Buying 3,300,000 kWh/Year at 6.85 Cents, with No Discounted Production, Leading to Evaluated Cost of Only 3.01 Cents Study Unit Year & Site 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 TOTALS SO $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 $0 Evaluated cost = Evaluated cost = Maint. &tasp. $0 $0 $0 $0 $0 $0 $0 so $0 so so so so so so so so so so $0 3,300 MWh Annual X $68.53 Cost $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 $226,220 PW Disc. Net Factor Cost 1 0.9 0.81 0.729 0.656 0.59 0.531 0.478 0.43 0.387 0.349 0.314 0.282 0.254 0.229 0.206 0.185 0.167 0.15 0.135 discounted cost/undiscounted production = $226,220 $203,598 $183,238 $164,914 $148,400 $133,470 $120,123 $108,133 $97,274 $87,547 $78,951 $71,033 $63,794 $57,460 $51,804 $46,601 $41,851 $37,779 $33,933 $30.540 $1,986,661 kWh Production 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3,300,000 3.300.000 66,000,000 $1,986,661/66,000,000 = Discntd Production 3,300,000 2,970,000 2,673,000 2,405,700 2,165,130 1,948,617 1,753,755 1,578,380 1,420,542 1,278,488 1,150,639 1,035,575 932,017 838,816 754,934 679,441 611,497 550,347 495,312 445.781 28,987,971 3.01 cents discounted cost/discounted production = $1,986,661/28,987,971 = 6.85 cents kilowatt, when evaluated with discounted costs and discounted kilowatts, is higher than when evaluated with both undiscounted, indicating that a greater proportion of the costs lies in the short term than does the output. The outp.it is spread evenly over the 20- year period, while the costs tend to cluster near the beginning, particularly in the first year in this case. 22 This is typical in DG studies but not in all DR. Usually the cost analysis of a large capital asset is done in a discounted manner, because some of the costs and much of the production will occur far into the future (see Chapter 5's and 29's discussions of discounting factors used in economic analysis). Thus, 6.85 cents is the more realistic value to interpret from Table 10.5. Always Use Discounted Costs and Energies Table 10.6 shows a bogus analysis of the same DG unit, using a "bias trick" discussed in Chapter 29 (and briefly in Chapter 5) in which undiscounted production over the period is used along with discounted costs for the period to determine the per-kilowatt-hour cost. This results in a computed cost per kilowatt of only 3.01 cents, an impressively low value, one reason, perhaps, why that this "mistake" is sometimes seen in DG and DR studies. 22 Note that the 11.70 value obtained is not the same value one would get if one calculated an annual cost per kilowatt, i.e., undiscounted annual cost over undiscounted annual output, for each year, then discounted those ratios depending on their year's PW factor, and then averaged them to obtain one value. That approach gives the same 8.40 per kilowatt hour as doing no discounting at all. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 385 Many planners understand and accept the discounting of future costs in a long-term planning study, but are uncomfortable with the discounting of future production as done to get the 6.85 cent figure in the example above. Discounting of future costs is an almost universal feature of least-cost regulated utility planning, to the extent that planners experienced with only utility planning may believe it is used in all planning (it isn't, although the time-value of money is always addressed in some manner). Most traditional T&D planning studies do not involve assessment of energy or production or long-term benefit, but instead work to minimize the cost to satisfy an obligation-to-serve criterion. As such, they deal with a discountable quantity only on the cost side of the ledger. Thus, T&D planners have not seen situations that demonstration why it is absolutely necessary to discount future production if future costs are discounted. By contrast, DG and DR studies are nearly always looking at optional or discretionary resources. Neither is needed to meet a minimal obligation to serve. Both must justify themselves on the basis of benefits that outweigh costs. Even when those benefits are measured as "nega-watts" or "mega-watts" rather than dollars, they must be discounted in company with the costs for any benefits versus cost analysis to be valid. The easiest way to see that production kilowatt hours must be discounted like cost is by use of a counterexample, shown in Table 10.6. There, instead of distributed generation power, the 3,300,000 kWh needed by the rancher is purchased at a flat rate of 6.85 cents per kilowatt over the 20 year period. Annual cost is simply .0685250 x 3,300,000 kWh, or $226,220 per year. If evaluated on the basis of discounted cost for the period divided by undiscounted kilowatt hours, the cost/kWh works out to 3.010, an absurdity: buy power for twenty years at 6.850/kWh and the average cost works out to 3.010/kWh. In summary, there are four ways one can compute "cost per kilowatt hour." 1. Undiscounted cost over undiscounted production. This is completely legitimate but neglects "present worth value of money" considerations in planning - the reason that discounting is done. It evaluates cost of the DG case discussed above as 5.750/kWh. That is correct in one sense - over the period total cost divided by total production is 5.750/kWh. However, most of the money has to be spent up front, whereas energy production is constant over the period, a fact missed because discounting is not used at all in this particular case. 2. Discounted cost over discounted production. This provides a legitimate evaluation of what cost/kWh is, taking the time-value-money into account all around. He^e, the evaluated cost is 6.850/kWh, a value that fairly represents the time values of money spent and product gained. Usually, this approach is the "most valid" way to perform such an analysis. 3. Discounted cost over undiscounted production underestimates the cost/unit of any scenario with up-front costs but continuing benefits. This approach should never be used in DG and DR planning - it is always a spurious evaluation. However, planners should look for it because it is often used: such an approach makes its way into many proponent studies of DG and DR, either by ignorance or deliberate intent to confuse. 4. Undiscounted cost over discounted production is equally wrong, but seldom used. Its use would artificially raise the per unit cost, making DR look less attractive. Chapter 29 discusses "bogus" studies, including this and other mistakes often made in planning, as well as various "tricks" that are sometimes deliberately applied to distort the results of planning studies. Copyright 2004 by Marcel Dekker, Inc. 386 Chapter 10 Table 10.7 One-Page Summary of Chapter 10 Distributed resources are customer-side or customer-site mechanisms for either reducing need for or increasing efficiency of energy usage. Distributed resources are often a cost-effective way of solving an energy consumer's needs for either of the Two-Qs (energy or reliability). DR often improves customer reliability and consequently has additional value that should be taken into account. UPS systems are a legitimate DR measure, one that exemplifies DR's reliability impacts. Electric utilities are often viewed as responsible for implementation and administration of DR programs because their monopoly franchise makes them "steward" of energy management practices in their service territory. Distribution generation (DG) and distributed storage (DS) are subsets of DR which are implementable either on the customer or utility side of the meter. DSM will defer transmission additions, but will not defer most distribution projects. Load reductions at the transmission level basically accumulate to look like a reduction in load growth rate. As a result, DSM can delay when transmission projects are needed. At the distribution level, DSM looks different and has a different effect (see below). It will have no impact on the need for projects to extend service to new areas and new customers. DR reduces load density. DSM reduces the load, but not the area that the distribution system must cover, the number of customers that must be connected to the system, or the date when any new customer will move in and demand power. As a result, DR's impact on distribution is best capitalized upon by reducing capacity density of the distribution. Maximizing T&D savings from DR requires re-design of the T&D system. That can require a lot of re-planning, but it usually doubles or triples the distribution-level savings, increasing the overall T&D savings by 50% or more. DR measures that individually do not pass "screening tests" for cost- effectiveness may be cost-effective in an overall plan. If the DSM program is just shy of having enough load reduction to produce a deferral or cancellation, it may be cost-effective to include "marginal" programs to reach the reduction target. DR peak reduction measures like load control and re-scheduling are usually more cost-effective at reducing T&D needs than energy conservation alone. The reason: T&D requirements are linked almost entirely to peak load level. Reducing energy off-peak, which is the additional benefit conservation brings, may cut annual losses costs, but these are a minor aspect of T&D costs. DR and feeder switching interact Properly planned, they can be used to support one another. In particular, feeder re-switching can be planned in conjunction with DSM targeted to augment and multiply the effectiveness of the DSM load reductions. The best areas for DR targeting are not necessarily where T&D costs are highest. DSM should be targeted where the combination of DR cost effectiveness and T&D avoided costs are best. Copyright 2004 by Marcel Dekker, Inc. Distributed Resources 387 10.7 SUMMARY A utility can affect measures on the consumer side of the electric meter that can reduce energy usage, cut peak load, and improve the reliability of service and usability of its product as seen by the consumer. Some of these programs are implemented "by market forces" in spite of what the utility may desire or actions it might take. However, distributed resources have a potential benefit that should not be ignored. Impact on all aspects of the utility's performance, including revenues and consumer satisfaction, must be taken into account. One important point to keep in mind is that a T&D system is the ultimate distributed resource. At the customer end, the T&D system reaches every customer. That is about as distributed as a resource can get. The point is that DR planning should always include the T&D system. Table 10.7 gives a one-page summary of this chapter's key points. BIBLIOGRAPHY J. M. Studebaker, The Electricity Purchasing Handbook, Penwell, Tulsa, 1997. L. J. Vogt and D. A. Conner, Electrical Energy Management, Lexington Books, Toronto, 1977. H. L. Willis, Spatial Electric Load Forecasting Second Edition, Marcel Dekker, New York, 2002. H. L. Willis and G. B. Rackliffe, Introduction to Integrated Resource T&D Planning, ABB Systems Control, Gary, NC, 1994 H. L. Willis and W. G. Scott, Distributed Power Generation - Planning and Evaluation, Marcel Dekker, New York, 2000. Copyright 2004 by Marcel Dekker, Inc.