Diesel Generation in Demand Response: California's Nagging DR Question

Introduction
This project is the continuation of work that was begun at the request of Bruce Kaneshiro of the California Public Utilities Commission’s Demand Response Section of the Energy Division during the summer of 2010. The primary goal was to assess the role of diesel backup generation in demand-response (DR) programs. There was suspicion among CPUC staff that demand response may have been indirectly incenting participants to rely upon backup diesel generation (BUG’s) outside of the threshold amount allotted by the State of California. Even until now, the true emissions impacts of demand response programs in California --when accounting for the effects of diesel generation-- remain unknown.

As I began to look into the issue it became clear why no one had been able to study it; the issue was multifaceted and nebulous. As complexity began to unfold, I realized that the results of my inquiry would not yield conclusive results during the time frame of a summer internship. The main reason for this was a lack of comprehensive data on when registered diesel engines actually run. However, as a result of our initial inquiry, regulatory processes set in motion by the CPUC and CARB will begin to collect this data and make it available for analysis by the fall of 2011. With the support of the EPA Star Fellowship, I want to apply myself to fully resolving this issue. In doing so, this will be the first comprehensive cost-benefit assessment of demand response that includes the full emissions impact due to diesel generation.


Context
Regulation of Diesel Generators
The main question that this project will try to address is “to what extent do DR participants shift their loads onto diesel generators instead of simply curtailing?” The question focuses on participants in demand response programs who utilize diesel backup generation to respond to events (instead of simply curtailing load) outside of their permitted usage. Understanding the policy framework around this question involves overlapping jurisdictional authorities such as the California EPA, The Air Resources Board, numerous Air Quality Management Districts and the Public Utilities Commission. Some context on the specific roles of each agency in regards to this issue is crucial to an understanding of the ambiguity surrounding it.

The California Air Resources Board (CARB) drafted The Airborne Toxic Control Measure for Stationary Compression Ignition Engines (ATCM) in 2004. Concerns about the toxicity of diesel particulate matter were a prime reason for this regulation. The ATCM specifies the requirements for initial compliance and classifies engines for emergency and prime use. Emergency use engines require less stringent standards because they are only to be used when normal electricity service fails. Prime engines are not used in emergency situations and thus face stricter emissions criteria. The 2004 version of the ATCM established standards for Stationary Compression Ignition Engines operating in demand response programs. Only two DR programs are specifically allowed to use BUG’s: Interruptible Service Contracts (ISC) and the San Diego Gas and Electric Company’s Rolling Reduction Blackout Program (RRBP). Any new customers enrolling in these programs after January 1, 2008 had to meet diesel PM standards of 0.01 g/bhp-hr, the standards set for prime engines. The ATCM mandates than an engine must be used to testing and maintenance a certain amount of hours per year.

Air Quality Management Districts (AQMD’s) are responsible for ensuring that BUG owners remain in compliance with the standards set by the ATCM. Local districts perform inspections of BUG’s every 1-3 years, record hours of usage, and are responsible for enforcement. The enforcement procedures take the form of fines and penalties. Owners of BUG’s are mandated to keep manual logs which indicate each instance of usage of the unit, duration and purpose for operation. All BUG’s are equipped with a tamper-proof meter which tracks cumulative hours of use in the same way that a car odometer tracks miles driven. Monitoring of compliance can vary drastically by district due to logistical necessity; some districts have less than 500 BUG’s while others have more than 9000. BUG owners must file to get a permit with the AQMD of their region. The information from these permits is stored in a local AQMDs database.

The California Public Utilities Commission was the source of this inquiry. In the context of this project, their primary role is to oversee demand response programs developed by utilities. The CPUC can exert authority over utilities, but has no regulatory oversight over BUG owners.

What is the issue?
Although most districts ban the usage of emergency class generators in demand response programs (except for Interruptible Service Contracts and the San Diego Rolling Reduction Blackout Program), the information to determine whether owners actually do this is currently not gathered. AQMD’s normally only record cumulative hours of usage when they do BUG inspections. They typically ensure veracity by inspecting the manual logs and verifying that the number of hours listed matches what is shown on the non-resettable meter of the unit. So, AQMD inspectors, while they know for how long a diesel engine has operated, do not know exactly when the operation occurred. Without knowing the date and time of operation for the engines, it is impossible to determine whether BUG usage coincides with demand response event times.

Another contributing factor may be that it is unclear whether DR customers understand what all applicable regulations are. Some districts have issued compliance bulletins specifying that emergency class generators must be upgraded to prime engine status if they want to participate in a DR Program; other districts (including some of the larger ones) have not reached an official stance on the issue. It is, in any event, likely that the only customers who would find out about whether it is appropriate to use backup generation to respond to a DR event are those who specifically approach the district to ask. It is also likely that most DR customers do not know that this practice is frowned upon by their local AQMD.

Furthermore, vague utility tariff information might contribute to customer propensity to rely inappropriately on backup diesel generation to respond to DR events. Many programs typically thought of as “demand response” do not include specific instructions regarding proper BUG usage. The programs which do make mention of BUG’s in the tariff language invite owners to use their backup generators to respond to the demand response events, but leave it up to them to act within the bounds of their permitted use. The tariff language adds to the general uncertainty surrounding the issue of BUG’s participating in demand response.

How are agencies addressing this issue?
CARB is currently amending the ATCM, the results of which could provide data on when CA BUG’s are used. According to contacts at ARB, The new version will add additional reporting requirements from BUG owners that would close match the list of data provided above. It will also make all new emergency engines adhere to tier 4 emissions standards (0.01 g/bhp-hr), the most stringent category. It proposes to remove the hourly cap on number of hours that engines can run in exchange for the higher emissions controls. The amended version of the ATCM will go to Board for opening comments on October 20th, 2010.

The new ATCM will stipulate that as much of the data be reported electronically as possible. The Board will vote on the measure by early January. The deadline it will extend to districts for the collection of this data will likely be 2-4 months. As such, comprehensive data on the run-time logs of the units could be provided by June 2011.

CPUC will also likely amend utility tariff language in January 2011 with regard to BUG’s. The spirit of these amendments will likely be to mandate that all participants in demand response programs with diesel backup generation provide copies of the run-time logs (detailing when units are run) to the utilities. The driving motivation behind this development will not be to punish BUG owners who participate in DR, but to gather the information necessary to conduct an analysis on the extent of this practice.

These actions in and of themselves, while helpful, may not produce conclusive results. The main reason is due to manpower limitations. In the case of the data being provided by CARB, massive amounts of data analysis must be conducted in order to distill the list of at least 20,700 BUG’s in down to a several hundred DR

Why is this important?
Even though in California stationary sources of pollution are not a majority of overall emissions, the diesel particulate matter (PM) is an ultra-noxious pollutant. Long-term exposure to PM has a well-acknowledged history of correlation with lung cancer and cardiovascular disease. The small particles from diesel exhaust are rough and bind easily to other toxins in an environment, driving up the overall unit risk factor of pollution in an area. Moreover, concentrations of BUG’s are highest in urban areas, so their emissions have a proportionally higher impact per capita. To add further incremental impact, the usage we are concerned with here occurs on DR event days which are typically, the hottest, smoggiest days of the year. In sum, the PM emissions from stationary diesel engines are a dangerous source of pollution that needs to be closely monitored, regulated and understood—no matter how limited we think the impact may be.

Understanding the role of diesel generation among DR participants is an important piece of the overall analysis of the benefits and costs of demand response. Demand response customers earn incentive payments that are funded by ratepayer dollars. The pretext of these subsidies is that demand response promotes grid reliability by encouraging curtailment among program participants. If DR customers are not actually curtailing, but only shifting loads onto their diesel generator, then in effect ratepayers have subsidized the trading of cleaner grid-generation for dirtier diesel. From a reliability perspective, this cost may be outweighed by the other benefits of DR. However, at this point, we do not know enough about how DR customers use diesel generation to tell. No assessments of the value of DR have included the role of diesel generation and looked at the consequent emission impacts of its usage based on real data.

This is an important question. FERC has been aggressively pushing demand response and the dominant philosophy has been to pursue as much capacity as possible. The analysis will allow distinctions between “good” DR and “bad” DR. Additionally, demand response in the Eastern states probably relies much more heavily on diesel generation than California. A study of this nature would provide a benchmark that others could use to adapt to other states and geographic locations.

The issue of diesel generation in DR is also important because it invokes a discussion of the growing role of third-party aggregators such as Enernoc, CPower, Comverge and Energy Connect. DR Aggregators may be relying on backup generation as a selling point to enroll customers. The added sense of security that the potential to rely upon such generation provides is a clear motivating factor for customers to participate in DR programs. DR Aggregator market share is growing rapidly. If reliance on diesel BUG’s is a standard business practice, then this may have substantial impacts on emissions in the future. In addition, DR aggregators market themselves as “green”. If the emissions impacts of aggregator demand response participation are not really known, then this assertion may be misleading. Fully understanding the usage of diesel backup generation in DR is important because it could affect conclusions about the desirability of demand response aggregators.

Finally, California’s utility industry seems destined to venture further into the realm of dynamic electricity pricing in the future. In this world of price spikes and uncertainty for businesses, it is possible that people will increasingly turn to stable diesel backup generators to hedge against price fluctuations and avoid risky curtailment. Looking at how much people are currently relying on diesel backup generation in demand response programs can give us a glimpse into how a future of dynamic pricing might look.

RESPONSE TO DUNCAN CALLAWAY: "Tapping the energy storage potential in electric loads to deliver load following and regulation"

Original:
http://www.gerad.ca/fichiers/activites/act0469/callaway.20090116.direct%20load%20control.pdf

This paper is an important contribution to understanding what demand-side resources like thermostatically controlled loads could accomplish if linked seamlessly into the grid. I realize that this paper is but a first blush analysis of what is possible, given a host of assumptions. However, I am left wondering not whether this strategy is possible or not, but whether it is implementable. No doubt, these sort of questions arise further along in research. But I am still curious as to what the prospects are for such a strategy when we seem to have so much trouble with implementation of demand-side strategies when the goal is many times simpler than remotely controlling the set points on vast amounts of thermostats in real-time according to the generation profile of wind on any given day.

The principle that load shedding can function like storage is very powerful and this paper is inspiring in the way that it demonstrates the outer reaches of where this idea can take us. I wonder if there are not more mundane issues to be worked out first, however. It seems that with the "smart grid" issues that are raised today that these mundane issues are often not as trivial as we initially conceive them to be.

The best example of this that comes to mind is the recent debacle that has arisen over the installation of smart-meters in residential homes. Advanced metering infrastructure not only make good sense from a rational perspective because it allow homeowners to explore the energy consumption of their houses, but is also a fundamental prerequisite to a more dynamic balancing of supply and demand on the grid. They are basically the stepping stone to solutions like those proposed here and a host of others. Yet, even the simple (and free!) installation of this technologies in residential homes has been a disaster so far in California. Citizens in Bakersfield have protested en masse the installation of advanced meters in their homes since PG&E decided to use the city as a test bed. Many have complained that the meters lead to higher bills and view them as simply another tactic of the utilities to nickel and dime them out of more money. I don't see this debacle as being reflective of an overall trend; AMI is by most accounts an eventual certainty. However, I think that it is a good example of the completely different set of obstacles one must face when transitioning from a good and rationally sound idea to actual implementation.
This paper raised several questions to me that may be relevant in moving from the principle of this idea to the practical. The first may be obvious, but I wonder who would actually manage the manipulation of the set points on these thermostatically controlled loads? Would this be something that the CAISO or utilities manage remotely? If it is, does it imply a new set of responsibilities and tasks for the maintenance and upkeep of such a network? CAISO might be a very sophisticated an adept creator of markets and market products, but would they necessarily want an entirely new job to handle?

Even if these loads can be controlled remotely without infringing on occupant comfort levels, would owners of these thermostats still be willing to cede control--even if it is only of the set points--of their house? Americans are very protective of their home environment. I wonder if an idea like this could be twisted by some to be seen as some sort of a privacy invasion. I realize the point made by the paper that the model simulations dictate that " large changes in supply (or demand from other loads) can be followed without compromising the end-use function of the loads subject to control". However, I am not convinced that it would be an easy task to actually get these remotely monitored units installed into people's homes without any hassle. I can just hear the oversimplifying reactionism of the Bakersfield crowd, "NO WAY is the government going to control MY air conditioner!"

Assuming that all the technology exists to realize an operation of this scale, is there sufficient interoperability and technical stability to make this kind of a network actually work? I may be wrong but I suspect that there might be a host of unforeseen installation and interoperability issues, many of which may be impossible to predict. It seems like the only way to know would be to actually do it.

Related to this is the issue of latency, how fast does the response need to be and are technological innovations substantial enough to support it? This might be less of a problem since the paper focuses on providing ancillary services that can have more than an hour of response time. However, the conclusion mentions that "the fidelity of the control signal was assumed to be perfect", and that "the frequency (one per minute)
may be difficult to achieve in practice". That these were mentioned in the end as further avenues of research leads me to believe that this is an issue of concern that will be looked at in the future.

The basic premise of this paper opens the door to using a host of demand-side resources for load following. Technically, it seems that the sky is the limit. However, I cannot read this paper without thinking of the implementation challenges of this strategy. Tackling those, and taking it this lofty proposal from inception to fruition, would be an exciting challenge to undertake. It may be one, however, that would likely involve an entirely different skill set than what it takes to create a model like this.