Demand response has a long history in California. During the California energy crisis, electricity scarcity, real or fabricated, caused many hours in the market to become very expensive. In order to reasonably hedge against those high prices, California’s Public Utility Commission (CPUC) required its utilities to run utility DR programs, where utilities could reduce demand during those high-priced hours. The economics of these programs are very favorable: the avoided market payments far exceeded the payments made to customers.
As these models matured, demand response grew to a sizable contributor of mitigating system peaks. But this presented operational challenges in emergency situations, where having the utility dispatching these resources presented a risk to system stability. Therefore, the California ISO (CAISO) requested that demand response contributing to resource
adequacy be integrated with market operations. This new product did not create the market entrance of new DR as expected, in part because without a capacity payment, the economics for deploying these new DR resources outside of utility programs did not pen out. This, coupled with overly generous compensation within the utility programs, prompted exploring a centralized procurement model where DR could provide offers, and the utility could decide how much to procure. These resources would count toward the utilities’
resource adequacy requirement, and would be required to bid into CAISO’s wholesale dispatch markets.
And while CAISO’s current set of products and procurement mechanisms support some of these aggregators, there is a notion that aggregated wholesale products are likely not the right way to deploy these types of DR at scale. More aggressive deployment of TOU rates or retail models that incentivize customers to respond to price signals, will better align customer action with real-time system needs.
These models are summarized at a high-level in the following figure (figure 22).
We will discuss each of these major evolutions in California’s DR market, discussing some of the notable business models therein, and briefly discuss where California may be heading to further create a role for DR in providing flexibility and renewable integration.
76
Figure 22: Overview of general DR frameworks discussed in this case
4.3 Services DR can provide
Before diving into the details of each phase in DR’s evolution in California, it is helpful to orient around what major grid services DR has the capability to provide, and in particular, which services CAISO wants to capture through its DR product refinements:
1. Peak energy demand reduction 2. Reducing system ramp requirements 3. Firming intermittent renewable resources 4. Relieving network congestion costs
5. Reducing resource adequacy requirements
Below we give a brief description of what those services are, how DR can provide them and in what markets, and finally what types of DR can likely participate.
Part I USA
77 1. Peak energy demand shaving remains the “bread and butter” of demand response,just as moments of scarcity are the bulwark of generator’s earnings in a year. This payment also aligns with where the most value is to the grid69, but as that changes, so too will the market prices for other services, which may attract DR elsewhere. This service requires substantial demand reductions, and has historically involved many large industrial/commercial users, but most types of DR can provide this service. Given the scale of reductions, participant fatigue (when participants become less likely to respond after many events in quick succession) is substantial, and most DR programs require participants to only reduce their load a set number of hours every season.
2. Flattening increasingly steep system ramps: California’s system requires large
amounts of ramping capacity in the late afternoon, when demand reductions coincide with solar generation ramping down. Demand response could contribute to ramp-up, ramp-down services during ramp periods to lessen their severity. DR could also load-shift demand from peak to off peak, reducing the need for system ramps, but that service cannot currently be captured in California’s existing markets. By doing so, DR would also help decommit generators scheduled on to meet ramps later in the day, saving unit commitment costs. Finally, during fast ramp periods the probability of tripping a generator is very high, and demand response, if dispatched instead of generator ramping, also reduces this risk. This kind of service would require automated or controllable loads (likely heating, ventilation, and air-conditioning (HVAC), hot-water heaters, and EV/stationary battery storage), and would need to have low latency in response time.
3. Firming intermittent energy resources: Variable renewable resources diverge from their forecasts and create regulation and balancing challenges. DR can serve as a firming resource in balancing markets to eliminate these divergences. This is currently being done in some microgrid demonstration projects with solar, DR, and batteries. Since this service would require DR to act on faster timescales, it would need to be automated loads, and the latency would need to be low. Hot water heaters, certain types of HVAC, and LED lighting aggregated at scale could help play these functions, as well as EV charging and behind the meter batteries.
4. Relieving network congestion stress: When transmission and distribution assets become congested, the cheapest sources of power cannot be utilized, and demand reduction is a more useful tool to manage congestion rather than altering power flows.
This would look similar to peak load reduction, since that congested area would be paid at the locational marginal price , but revenues from financial transmission rights (FTR) could also be accrued by DR aggregators. Furthermore, DR could help avoid new transmission and distribution infrastructure investment, and part of that money could go to support localized DR deployment efforts70. Most types of DR could provide this service, although the requirements for reliable delivery must be higher when binding congestion is the price of under delivery.
5. Reducing resource adequacy requirements: Every utility is required to demonstrate they have procured sufficient energy and capacity contracts to reliably cover peak
69 Peak shaving potentials from DR is significant, according to analysis, if half of California’s Investor-Owned Utilities (IOU) customers switched to their electricity company’s existing time-of-use (TOU) rates, the reduced peak demand is enough to save the need for about 30 “peaker” 100 megawatt power plants. This same transition to TOU rates could save electric utilities and customers nearly $500 million annually, a nearly 20%
system-wide cost reduction
70 https://www.edf.org/sites/default/files/demand-response-california.pdf
78
demand in the coming years. Dispatchable demand response plays a key role in
achieving this resource adequacy. Therefore, the same way other energy resources and generators are paid to ensure their availability for future years (in California, typically long-term bilateral contracts), DR should get paid similarly. Most types of DR would be eligible provided they could be reliably available during the expected hours of system peak (between 4pm and 7pm for California).
Figure 23: Load profile comparison under real-time pricing.