Page 3 of 5
To adequately account for the nuances of power plant cycling damage mentioned above, we utilized a commercially available production cost model, Cycling Advisor, that includes power plant cycling damage as a constraint. The model also includes ramping cost curves that allowed us to model cycling damage because of varying ramp rates. The methodology used to develop the model and cost inputs have been used in several previous studies  .
Slow and fast DR
Pilot programs in California have demonstrated some success in using automated demand response to provide wholesale ancillary services . The Participating Load Pilots (PLP) authorized by the California Public Utilities Commission (CPUC), demonstrated the viability of fast DR to provide ancillary services . The study used California ISO operating rules on certain Pacific Gas & Electric (PG&E) loads to conduct these pilots, and PG&E correctly recognizes the need for a large scale deployment of these programs to have further insights into performance of this technology.
However, PG&E concluded that automated fast DR as a technology was capable of participating in wholesale ancillary services market.
It should also be noted that other electric markets in U.S. have market products that allow some form of DR to participate in energy markets, including the ancillary services markets.
Literature research done in the past has provided a summary of DR programs that can participate in various markets including – NYISO, PJM, ISO-NE, CAISO, ERCOT and MISO .
This paper will provide a conceptual view of the viability of large scale deployment of DR to provide both energy benefits to utilities, as well as reduced cycling related wear and tear costs on conventional fossil generation.
We performed realistic simulations of a typical electric utility portfolio to illustrate the concept of using DR to reduce cycling-related costs on conventional generation due to wind integration. The simulations, based on hourly load variations, reflect a large (approximately 4,600 MW), utility system whose generation is dominated by large fossil-fueled generation (see Figure 2). The evaluation simulated operations through a full calendar year (CY 2008), representing high natural gas and energy market prices. The first simulation was used to establish a baseline: results reflect operation of the generation fleet without the availability of DR. Results from that simulation were compared to another that included DR. That comparison yielded an estimate of the potential cost reduction due to reduced cycling. Both these simulations were run for two different sets of cycling cost inputs. One set assumed 100 percent of our best estimate of cycling costs as inputs. The second set of inputs included only 50 percent of these costs — essentially representing an “error” omitting half the cycling costs. Since the production cost model minimizes total production cost by including cycling costs, this sensitivity allows us to measure the value of DR when using different levels of cycling cost in dispatch decisions.
Readers should note that simulations provided specific outputs, to value energy benefits of DR in terms of fuel cost due to baseload operation and costs due to low load operation of conventional fossil generators and cycling damage.
However, note that simulations do not account for short duration (< 1 hour) power fluctuations (due to Automatic Generation Control, AGC) and related damage. We believe that fast DR can provide even greater benefits if effects from those short duration power fluctuations on the generation system (cost) are included in the evaluation.