Sioshansi F.P. Smart Grid: Integrating Renewable, Distributed & Efficient Energy
Подождите немного. Документ загружается.
approximately 8,000 customer accounts onto a new CPP rate
in the fall of 2009, while PG&E moved around 5,000
accounts to PDP in the spring of 2010. A number of those
customers opted out prior to the summer of 2010, leaving
4,100 customers enrolled at SCE (a retention rate of about
50%), and 1,800 customers enrolled at PG&E (a retention rate
of about 36%). About 1,500 customers remained enrolled at
SDG&E in 2010 from the 2,400 that were originally defaulted
(a retention rate of about 60%).
22
22
Several hundred medium-sized customers were transitioned to default
CPP in 2010, adding to the 1,800 large customers in 2008.
Additional changes are planned for CPP at each of the utilities
over the next few years. SDG&E will expand its default CPP
rate to medium-sized, customers whose maximum demand
lies between 20 and 200 kW as installation of smart meters
expands in 2012 and 2013. SCE has proposed to expand
default CPP to customers of less than 200 kW and to offer a
capacity reservation option to CPP participants. Finally,
PG&E plans to expand default PDP to smaller customers in
late 2011 as additional smart metering is rolled out. The total
pool of medium C&I customer accounts exceeds 200,000.
Future load impact evaluations will track the effect of both
expanding enrollment and smaller sized customers on
estimated CPP load impacts.
Price-Response Potential Outside of California
In states other than California, increased availability of smart
meters through smart grid investments will make dynamic
pricing for commercial and industrial (as well as residential)
customers much more feasible than it has been historically.
621
Furthermore, interest in demonstrating benefits from smart
grid investments is likely to provide incentives for utilities
and regulators to explore dynamic pricing.
23
As noted above,
the combined load impacts for CPP and DR programs for
large C&I customers in California amount to more than 1% of
the total system load. Expanding the rates and programs to
smaller but much more numerous medium-sized business
customers could double that amount. In addition,
emergency-based interruptible service programs for large
customers and air conditioner load control programs for
residential and small business customers add another 2–4%
system load-reducing potential.
23
Why invest in smart grids and smart meters, but continue with
not-so-smart pricing?
California differs from other states in a number of ways,
including having relatively fewer large energy-intensive
industrial customers, in part due to a history of relatively high
average electricity prices in recent years; more agricultural
customers; milder weather; and a strong historical focus on
energy efficiency programs. At the same time, it has many of
the same types of office buildings, retail stores, warehouses,
and municipal water utilities that are present in all states.
Some of the more heavily industrialized states might be
expected to generate relatively larger C&I CPP load impacts
as a percent of total load. Finally, a study of potential DR load
impacts sponsored by FERC concluded that current programs
have the potential to reduce total U.S. peak demand by about
4%, while modest expansion of the current “business as
usual” case could expand that impact amount to about 9%.
24
24
622
See FERC [10].
Dynamic Pricing versus DR Programs
Current discussions about the need for and role of
price-responsive demand (i.e., customer load reductions in
response to dynamic retail prices or DR credit payments) in
wholesale markets typically seem to focus unduly on DR
bidding programs arranged by DR service providers and run
through regional ISO/RTOs. This is in contrast to the
seemingly more natural mechanism of dynamic retail pricing
offered by utilities and other retail providers, which reflects
changes in wholesale market prices. Greater penetration of
retail dynamic pricing would obviate much of the need for
DR programs. However, there is a long history of barriers to
adoption of time-based pricing, which include lack of
incentives on the part of utilities, concern by regulators about
potential bill impacts for some customers, as well as metering
costs. These barriers are discussed in more detail in Chapter 3
on the efficiency and ethics of dynamic pricing.
However, DR programs have their own set of issues and
problems. Three of the most important are how to measure
the actual DR load impacts provided, how those load impacts
should be compensated, and whether third-party aggregators
should be allowed to recruit utility customers to participate in
DR programs in non-restructured states. The measurement
issue involves which methodology to use to estimate DR
customers' baseline load, or the counterfactual amount that
they would have used had they not participated in a DR event.
This estimation is necessary because energy not used cannot
be measured directly, but instead must be inferred by
623
comparing observed energy consumption to an estimated
baseline load to determine the difference.
Various baseline methods have been proposed, tested, or used
in different jurisdictions. In a series of baseline analyse
conducted using data for several DR programs in California,
we have shown that the accuracy of baseline loads, and thus
estimated DR load reductions, depends on several factors,
including the baseline method used and the type of customer
(e.g., industrial or commercial, and degree of inherent load
variability), and can often be quite poor.
25
This result raises
questions about both fairness of DR payments and credits and
the uncertainty of DR resources compared to supply
resources.
25
See Braithwait and Armstrong [11].
The DR compensation issue has gained recent prominence
due to a March 2011 FERC ruling that requires ISO/RTOs to
offer DR programs, and requires a uniform DR compensation
structure in which DR customers are to be paid the full
wholesale market price, or locational marginal price (LMP),
without regard for whether the customers are required to pay
for the amount of the load reduction at the generation
component (G) of their retail rate (the so-called LMP minus G
alternative compensation). Critics contend that full-LMP
compensation overpays for DR load reductions, resulting in
an effective subsidy that must be recovered through uplift
charges to all customers, and that the net benefit test required
by FERC's ruling is costly and unnecessary if the subsidy
portion of the compensation is removed. A rehearing of the
FERC ruling has been requested.
26
26
624
In view of these issues, ISO New England has encouraged states in its
footprint to adopt more dynamic pricing and has recommended DR
designs that better mimic dynamic retail pricing. See Chao [12].
Finally, while requiring ISO/RTOs to offer DR programs,
FERC has allowed states to determine whether to allow
third-party DR providers to recruit customers of
state-regulated utilities. Some states have banned such
recruitment, while others are determining whether to do so.
Fully regulated utilities often object to their customers being
recruited by competitive DR providers, on the basis of
potential revenue losses and uncertainty about the loads that
they must schedule in wholesale markets.
Conclusions
The new evidence presented in this chapter on the extent to
which large commercial and industrial customers will
participate in and respond to dynamic price structures such as
critical peak pricing (CPP), as well as in price-based DR
programs, adds to the results of previous studies of residential
customers to confirm that customers on average will respond
when peak prices rise substantially above their normal levels,
or DR credits are offered, on a limited number of critical
days. Two key features of the estimated load impacts—that
they range considerably across industry types (e.g.,
manufacturing, retail stores, and offices) and they tend to be
concentrated in a relatively small fraction of participating
customers—suggest to utilities and policy makers in other
states how load impacts are likely to vary in other settings and
what the potential is for increasing customer price
responsiveness.
As installation of smart metering and defaulting to CPP
continue to reach down to smaller but more numerous
625
business customers in California, load impacts will continue
to expand. Outside of California, the expanding installation of
smart meters and increasing intelligence of the grid should
provide the means and incentives to introduce dynamic
pricing to many more customers. Expansion of dynamic
pricing will also help to reduce the need for DR programs run
by ISO/RTOs, and thus reduce some of the controversial
issues that surround DR programs.
References
[1] Faruqui, A.; Malko, J.R., The residential demand for
electricity by time-of-use: a survey of twelve
experiments with peak load pricing, Energy 8 (1983)
781–795.
[2] Braithwait, S., Customer Response to Electricity Prices:
Information to Support Wholesale Price Forecasting and
Market Analysis. (2001,) EPRI, Palo Alto, CA; 1005945.
[3] Braithwait, S.; O'Sheasy, M., RTP customer demand
response–empirical evidence on how much can you
expect, In: (Editors: Faruqui, A.; Eakin, B.K.) Electricity
Pricing in Transition, pp. 181−190 (2002) Kluwer
Academic Publishers, Boston, MA.
[4] A. Faruqui, S. Sergici, “Household Response to Dynamic
Pricing of Electricity – A Survey of Seventeen Pricing
Experiments,” JI of Regulatory Economics (2010) Vol.
38, No. 2, 193−225.
[5] S.D. Braithwait, D.G. Hansen, J.D. Reaser (Christensen
Associates Energy Consulting), 2009 Load Impact
Evaluation of California Statewide Critical-Peak Pricing
Rates for Non-Residential Customers: Ex Post and Ex
626
Ante Report, Prepared for San Diego Gas & Electric,
April 19, 2010.
[6] S.D. Braithwait, D.G. Hansen, J.D. Reaser (Christensen
Associates Energy Consulting), 2010 Load Impact
Evaluation of California Statewide Demand Bidding
Programs (DBP) for Non-Residential Customers: Ex
Post and Ex Ante Report, Prepared for Southern
California Edison, April 1, 2011.
[7] S.D. Braithwait, D.G. Hansen, D.A. Armstrong
(Christensen Associates Energy Consulting), 2010 Load
Impact Evaluation of California Statewide Aggregator
Demand Response Programs: Ex Post and Ex Ante
Report, Prepared for Pacific Gas & Electric, March 29,
2011.
[8] C. Goldman, N. Hopper, R. Bharvirkar, B. Neenan, R.
Boisvert, P. Cappers, et al., Customer Strategies for
Responding to Day-ahead Market Hourly Electricity
Pricing, Final report to the California Energy
Commission, Lawrence Berkeley National Laboratory
(LBNL-57128), August, 2005.
[9] S.S. George, J. Bode, J. Schellenberg, S. Holmberg
(Freeman, Sullivan & Co.), 2010 Statewide
Non-Residential Critical Peak Pricing Evaluation,
Prepared for San Diego Gas and Electric, April 1, 2011.
[10] FERC Staff Report, A National Assessment of Demand
Response Potential, Prepared by The Brattle Group,
Freeman, Sullivan & Co, and Global Energy Partners,
LLC, June 2009.
[11] S.D. Braithwait, D.A. Armstrong (Christensen
Associates Energy Consulting), 2008 Load Impact
Evaluation of California Statewide Aggregator Demand
627
Response Programs: Volume 2 Baseline Analysis of
AMP DR Program, Prepared for Pacific Gas & Electric,
March 30, 2009.
[12] H.-P. Chao, Price-responsive demand management for a
smart grid world, Electricity J. Vol. 23, Issue 1, January/
February 2010, pp. 7−20.
Appendix
The basic regression model for measuring CPP load impacts
is the following:
In this equation, Q
t
represents the amount of usage in hour t
for a customer enrolled in CPP prior to the last event date; the
b's are estimated parameters; h
i,t
is a dummy variable for hour
i; CPP
t
is an indicator variable for program event days; CDH
t
is cooling degree hours
27
; E is the number of event days that
628
occurred during the program year; MornLoad
t
is a variable
equal to the average of the day's load in hours 1 through 10;
MON
t
is a dummy variable for Monday; FRI
t
is a dummy
variable for Friday; DTYPE
i,t
is a series of dummy variables
for each day of the week; MONTH
i,t
is a series of dummy
variables for each month; Summer
t
is a variable indicating
summer months (defined as mid-June through mid-August),
28
which is interacted with the weather and hourly profile
variables; OTH
t
is a dummy variable indicating an event hour
for a non-CPP demand response program in which the
customer is also enrolled; and e
t
is the error term. The
“morning load” variable was used in lieu of a more formal
autoregressive structure in order to adjust the model to
account for the level of load on a particular day. Because of
the autoregressive nature of the morning load variable, no
further correction for serial correlation was performed in
these models.
27
Cooling degree hours (CDH) was defined as MAX[0, Temperature –
50], where Temperature is the hourly temperature in degrees Fahrenheit.
Customer-specific CDH values are calculated using data from the most
appropriate weather station.
28
This variable was initially designed to reflect the load changes that occur
when schools are out of session. We have found the variables to a useful
part of the base specification, as they do not appear to harm load impact
estimates even in cases in which the customer does not change its usage
level or profile during the summer months.
629
630