Assessment of Demand Response Market Potential and Benefits … of Demand... · July 2015...

Assessment of Demand Response Market Potential and Benefits in Shanghai

Prepared by

Environmental Change Institute & Oxford Institute for Energy Studies University of Oxford

Prepared for

Natural Resources Defense Council

Supporting Partners

China Prosperity Strategic Programme Fund of the Foreign and Commonwealth Office The Energy Foundation

July 2015

eci

© Wilson Hui © Christian Schnettelker © Andrew Imanaka

July 2015

Assessment of Demand Response Market Potential and Benefits in Shanghai

Authors

The Environmental Change Institute, University of Oxford

Mr Yingqi Liu Dr Nick Eyre Dr Sarah Darby

The Oxford Institute of Energy Studies, University of Oxford

Mr Malcolm Keay Dr David Robinson Dr Xin Li

Prepared for

Natural Resources Defense Council

Supporting Partners

China Prosperity Strategic Programme Fund of the Foreign and Commonwealth Office The Energy Foundation

Assessment of Demand Response Potential and Benefits in Shanghai Acknowledgements 1

ACKNOWLEDGEMENTS

This report was produced by the collaboration of team members from the Environmental Change Institute (ECI) and the Oxford Institute for Energy Studies (OIES), University of Oxford. ECI and OIES worked together to synthesise international experience of using, supporting and valuing DR, while ECI led on the market potential assessment and OIES had the leading role on valuation analysis, with significant input from each other.

We would like to express our gratitude to Ms Hyoungmi Kim and her colleagues at the Demand-Side Management Team of NRDC China Programme, for their guidance, support and feedback through all stages of the project. We are very grateful for the valuable support from the Shanghai DR Implementation Team: Shanghai Economic and Informatisation Commission (EIC), Load Control Centre of State Grid Shanghai Municipal Electric Power Company, NRDC, Shanghai Twenty-First Century Energy Conservation Technology Co., Ltd, Shanghai Electrical Apparatus Research Institute (Group) Co., Ltd, Beijing Nari Smartchip Microelectronics Technology Co., Ltd, Tongji University, Honeywell and LY Enerlytics. We would like also to extend our gratitude to these experts who have provided highly valuable and insightful comments and guidance: Mr Pierre Bull from NRDC, Mr Steve McCarty from Energy Solutions, Dr Frederich Kahrl from E3, Ms Elizabeth Reid from Olivine, Professor Yanmin Han from Tongji University and Dr Shaodi Zhang from Shanghai Electrical Apparatus Research Institute. Finally yet importantly, we would like to thank NRDC Science Centre, China Prosperity Strategic Programme Fund (SPF) of the British Embassy in Beijing, and the Energy Foundation for their sponsorship of this study.

The contents of this paper are the sole responsibility of the authors. They do not necessarily represent the views of the Environmental Change Institute, the Oxford Institute for Energy Studies or any of their Members.

2 Assessment of Demand Response Potential and Benefits in Shanghai Contents

CONTENTS

Acknowledgements ................................................................................................................................................. 1 Contents .................................................................................................................................................................. 2 List of Figures .......................................................................................................................................................... 3 List of Tables ........................................................................................................................................................... 4 Executive Summary ................................................................................................................................................. 5 1. Introduction ................................................................................................................................................... 19 2. International Experience ............................................................................................................................... 20 2.1 Background .......................................................................................................................................... 20

2.1.1 Rationale for supply side approaches ......................................................................................... 20 2.1.2 Electricity systems are changing ................................................................................................. 20 2.1.3 Historical background of DR development in the OECD countries ............................................. 22

2.2 Characterising the DR resources ......................................................................................................... 23 2.2.1 Uses of DR in the electricity system ............................................................................................ 23 2.2.2 Studies of DR potential in the US and the UK ............................................................................. 29 2.2.3 Scale and sources of current DR resources ............................................................................... 31

2.3 Value of DR resources ......................................................................................................................... 37 2.3.1 Methodology for assessing the benefits of DR resources ........................................................... 37 2.3.2 The valuation of DR in the UK ..................................................................................................... 38 2.3.3 The valuation of DR in the US ..................................................................................................... 41

2.4 Enabling frameworks for developing the DR market ............................................................................ 42 2.4.1 Business models of DR programmes .......................................................................................... 42 2.4.2 Regulatory incentives for industry stakeholders to promote DR ................................................. 46 2.4.3 Roles of aggregation ................................................................................................................... 50 2.4.4 Customer technical capability, incentive and education .............................................................. 52 2.4.5 Reliability of DR resources .......................................................................................................... 54 2.4.6 Enabling technologies ................................................................................................................. 56

2.5 Concluding remarks and implications for China ................................................................................... 56 2.5.1 International experience .............................................................................................................. 56 2.5.2 Implications for China .................................................................................................................. 57

3. DR Market Potential and Valuation in Shanghai ........................................................................................... 60 3.1 Introduction .......................................................................................................................................... 60 3.2 Framework and Methodology .............................................................................................................. 60

3.2.1 Assessment of DR potential ........................................................................................................ 60 3.2.2 Assessment of DR benefits ......................................................................................................... 66

3.3 Approach for Assessing DR Potential and Values in Shanghai ........................................................... 68 3.3.1 Assessment of DR potential ........................................................................................................ 68 3.3.2 Assessment of DR benefits ......................................................................................................... 80

3.4 Results of Assessment ......................................................................................................................... 84 3.4.1 Estimate of DR potential ............................................................................................................. 84 3.4.2 Estimate of DR benefits .............................................................................................................. 85

3.5 Recommendations for further research ............................................................................................... 89 References ............................................................................................................................................................ 91 Appendix ................................................................................................................................................................ 93

Assessment of Demand Response Potential and Benefits in Shanghai List of Figures 3

LIST OF FIGURES

Figure 1 Reported Potential Peak Demand Reduction of DR programmes in the FERC surveys ........................ 32 Figure 2 DR programmes and their reported potential peak demand reduction in the FERC surveys ................. 32 Figure 3 DR resources in wholesale markets of RTOs/ISOs and their reported potential peak demand reduction ............................................................................................................................................................................... 33 Figure 4 DR programmes in RTOs/ISOs wholesale markets and their potential peak demand reduction ............ 34 Figure 5 Emergency DR resources in PJM by load-reducing method and sector ................................................. 35 Figure 6 Short-Term Operating Reserve (STOR) resources by fuel type in recent seasons ................................ 36 Figure 7 Potential cost savings of DR programme from domestic sector and small and medium enterprises by 2030 ....................................................................................................................................................................... 39 Figure 8 Peak generation capacity investment to meet demand from the domestic sector and SMEs in the UK . 40 Figure 9 CO2 emission savings from demand response (by 5% reduction) up to 2025 ........................................ 41 Figure 10 Costs and benefits of DR in a California electric utility .......................................................................... 42 Figure 11 Regulatory mechanisms for recovering DSM costs in the US ............................................................... 48 Figure 12 Generator Equivalent Availability Factor (EAF) in PJM RFM from 2007 to 2014 .................................. 55 Figure 13 Key steps in analysing the DR market potential .................................................................................... 61 Figure 14 Total electricity consumption by sector in Shanghai for recent years ................................................... 68 Figure 15 Historical growth and estimate of highest system peak demand in Shanghai ...................................... 69 Figure 16 Forecast for peak demand in future years (2020, 2025 and 2030) ....................................................... 69 Figure 17 Aggregate hourly load of electricity demand in Shanghai on a typical summer weekday (6 August 2014) ............................................................................................................................................................................... 70 Figure 18 Steps for calculating generation capacity cost ...................................................................................... 81 Figure 19 Assessment of DR market potential in Shanghai .................................................................................. 84 Figure 20 Avoided capacity cost between 2020 and 2030 with 7% discount factor .............................................. 86 Figure 21 Avoided capacity cost between 2020 and 2030 with 0% discount factor .............................................. 86 Figure 22 Avoided capacity cost between 2020 and 2030 with 10% discount factor ............................................ 87 Figure 23 Avoided energy cost .............................................................................................................................. 87 Figure 24 Avoided CO2 cost .................................................................................................................................. 88 Figure 25 Avoided T&D cost .................................................................................................................................. 88 Figure 26 Range of all other avoided costs (energy, CO2 and T&D) ..................................................................... 88 Figure 27 Total avoided costs between 2020 and 2030 ........................................................................................ 89

4 Assessment of Demand Response Potential and Benefits in Shanghai List of Tables

LIST OF TABLES

Table 1 Participation of DR resources in capacity mechanisms ............................................................................ 24 Table 2 Participation of DR resources in ancillary services ................................................................................... 27 Table 3 Examples of commercial and industrial DR trials under the Low-Carbon Network Fund ......................... 29 Table 4 Amount of capacity by fuel type in the first T-4 auction of GB Capacity Market ....................................... 37 Table 5 Potential benefits of DR to different activities of the electricity system ..................................................... 37 Table 6 Benchmark price for different type of services ......................................................................................... 38 Table 7 Avoided generation capacity and network costs ...................................................................................... 40 Table 8 Description of key components in the estimation of avoided costs following CUPC methodology .......... 41 Table 9 Business models of DR under different electricity industry structures ...................................................... 43 Table 10 Examples of techniques in estimating customer participation rate ......................................................... 64 Table 11 Examples of techniques in estimating average load impact ................................................................... 65 Table 12 Estimates of annual electricity use and number of customers by sub-sector in 2020, 2025 and 2030 .. 73 Table 13 Distribution of sampled customers with load profile information ............................................................. 74 Table 14 Estimate and assumption of Peak Load Factors (PLFs) by sub-sector .................................................. 75 Table 15 Estimated total peak demand and average per-customer peak demand by sub-sector for 2020, 2025 and 2030 ................................................................................................................................................................ 76 Table 16 Assumptions for customer participation rate in future milestone years (2020, 2025 and 2030) ............. 77 Table 17 Assumptions for average per-customer load impact* ............................................................................. 79 Table 18 installed capacity and investment of selected gas-fired plant or unit ...................................................... 82 Table 19 Cost parameters and tax rates of CT and CCGT plants ......................................................................... 82 Table 20 Estimated capacity cost (RMB/kW-year) ................................................................................................ 83 Table 21 Parameters used in the estimation of other avoided costs ..................................................................... 83 Table 22 Technical characteristics of plant used to estimate other avoided costs ................................................ 83

Assessment of Demand Response Potential and Benefits in Shanghai Executive Summary 5

EXECUTIVE SUMMARY

The electricity systems in China are changing. The need for large-scale integration of intermittent renewables and penetration of electric vehicles, as implied by the low-carbon energy transition, poses challenges to system operation and network management. Moreover, the ever-increasing electricity consumption and peak demand, and the emerging policy objectives of reducing environmental impacts and improving economic efficiency, underline the urgency to change the way electricity systems are operated and planned in China. There is a growing interest, particularly from high-level policy makers, to explore the potential for demand side resources to contribute to the supply-demand balance and to meet the key objectives of carbon emission reduction, environmental protection and economic efficiency.

Demand response (DR) is increasingly seen as an important demand-side resource in achieving these objectives. By definition, DR refers to programmes to reduce electricity use, especially at times of system peak, by exposing customers to pricing signals that reflect system marginal cost of generation or give them financial incentives in return for their commitment to reduce demand when asked to do so. As part of the orderly electricity use programmes, utilities in China have some experience running administrative load management programmes, including load shifting and avoidance projects, as well as restricted use and demand curtailment in the event of severe power shortage. However, without the market-based mechanisms typically employed in DR, administrative load management often lacks the ability and flexibility in promoting the willingness of customers to participate, and may not be ‘fit-for-purpose’ for the long-term development of demand-side resources.

The development of DR has increasingly become more important in China. In 2014, Shanghai became the first pilot city in China, as designated by the National Development and Reform Commission (NDRC), to trial municipality-level DR programmes. Focusing initially on commercial and industrial (C&I) customers, the pilot aims to explore market-based mechanisms to support the procurement of DR resources. While the Shanghai DR pilot is under way, the NDRC has issued a notice to four cities in China (i.e. Beijing, Suzhou, Tangshan and Foshan, referred to as demand-side management (DSM) cities), requesting them to plan and undertake DR pilots. Since the experience of implementing DR is limited in China, especially with models based on market mechanisms, there is immense value in learning from leading international experiences in countries (e.g. the UK and the US) where DR has been successfully used to address electricity system needs.

This study contributes to the DR pilot in Shanghai and similar efforts in other parts of China, by synthesising international experience in using and supporting DR resources, and by conducting a very preliminary assessment of the market potential and values of DR Shanghai. To share international experience, this report reviews the background and current situation of using DR resources, the existing efforts in the UK and the US to estimate the potential and value of DR, and the regulatory and policy enablers for promoting efficient DR. Given the difference in terms of regulatory and market conditions of electricity systems between China and OECD countries, this report also highlights a number of implications and caveats for interpreting evidence from OECD countries. To estimate the potential and the benefit of DR, the study first outlines the general framework and methodology for the assessment, before continuing to use international evidence and local data in Shanghai to gauge the DR potential and benefit there. This exercise provides an example of how to conduct such assessments and offers a number of recommendations for future research in the same area.

International Experience

§ Changes in electricity systems create needs and opportunities for demand-side resources like DR

Since electricity is difficult and expensive to store, supply and demand need to be matched at all times. Traditionally, the burden of maintaining the supply-demand balance largely falls on the supply side, with the demand side being treated essentially as passive. This is mainly due to the characteristics of electricity supply that are perceived to be most significant, such as controllability and reliability, relative ease in coordinating and aggregating supply-side, and the economics of supply-side flexibility.

Emerging changes in the electricity system have however sparked interest in using DR and other demand-side resources as complements or even alternatives to supply-side solutions. These changes include the increasing

6 Assessment of Demand Response Potential and Benefits in Shanghai Executive Summary

penetration of intermittent renewables and distributed generation as well as smart meters and appliances, market liberalisation and reliance on pricing signals for system balancing, declining costs of smart technologies for monitoring and managing demand, and government policy objectives of energy decarbonisation and environmental protection.

In the US, the history of DR can be traced back to the 1970s when regulated utilities offered DR programmes to deal with rising summer peak demand. After the electric industry restructuring, a number of issues such as the power crisis in 2000-2001 and sharp electricity price spikes led policymakers and regulators to see the value of DR for the efficient functioning of wholesale markets. In Europe, while policy has traditionally focused on energy efficiency, new trends such as growing peak demand, ambitious targets for renewables, lower costs of smart metering and other enabling technologies, and the potential of an ‘active’ demand-side have heightened the importance of DR in EU energy policy.

§ DR resources can be used to deliver different system services

Supply-demand balancing occurs at different timescales (e.g. yearly, daily or hourly). In light of this, a variety of DR products have been designed to meet different system needs. The nature and importance of these DR products depends very much on the specific nature of the electricity system concerned (characteristics such as peak demand period and drivers of system peak demand, needs for different system services, etc.). A further distinction can be made between DR resources for helping overall system balancing or resource adequacy and those for addressing geographically specific network constraints.

Resource (capacity) adequacy – the biggest potential savings to the system come from long-term reduction in capacity requirements, particularly for generation. In the US and Europe, many unbundled systems1 have introduced, or are setting up, capacity markets to give incentives for the construction or retention of the generating capacity needed to fulfil resource adequacy obligations (e.g. load serving entities in the US are required to contract for capacity that is 10-20% above forecast peak demand), and to balance the system at a time of growing importance of intermittent renewable energy sources, notably wind and solar PV. In a number of cases, these markets allow the participation of DR resources. Such schemes are generally based around ensuring a level of generating capacity or DR to ensure that expected demand could be met, via guarantees from the bidders to provide generation supply or demand side resources. Examples include forward capacity markets of ISO-NE and PJM2, the Great Britain (GB) Capacity Market and other capacity mechanisms in NYISO and MISO3.

Economic response – Economic or price-responsive DR refers to a straightforward situation of customers responding to electricity prices by reducing (or increasing) consumption, of their own volition and in their own direct interests rather than because they have undertaken to provide a particular service for a payment. In periods when system prices are higher, normally at times of system peak demand, consumers are expected to reduce their demand to save money; they may shift some of that demand to periods when prices are lower. Economic or price-responsive DR can include retail time-based programmes offered by utilities (e.g. time-of-use tariffs, critical peak pricing). In some unbundled markets in the US (e.g. PJM, ISO-NE and NYISO), DR resources can also bid directly into wholesale energy markets, and reduce the power drawn from the grid during hours of high wholesale prices. With the penetration of intermittent renewables (e.g. wind and solar), DR resources may be used to increase demand in the event of high generation output and low wholesale prices, especially coupled with storage.

Ancillary service – since the electricity system often experiences short-term changes in the supply-demand balance, ancillary services are necessary to ensure system reliability. In the US, for regions with an unbundled electricity sector, independent system operators (ISOs) or regional transmission operators (RTOs) offer ancillary service markets to help manage the operation and balancing of transmission systems; for other regions with a bundled structure, it is the vertically integrated utility companies or balancing authorities that provide or buy such services. Eligible DR resources can be offered into these ancillary service markets (e.g. reserve services in ERCOT4 and PJM). In the UK, National Grid, the system operator for Great Britain, maintains the Balancing Services (e.g. for Reserve Services and Frequency Response) to support the supply-demand balance of the GB

1 The term “unbundled” refers here to the separation of transmission (and sometimes distribution) network assets from generation, allowing for competition 2 PJM is currently re-examining its approach to demand response in the light of a Court decision last year (http://www.pjm.com/~/media/documents/reports/20141007-pjm-whitepaper-on-the-evolution-of-demand-response-in-the-pjm-wholesale-market.ashx). 3 Regional transmission organisations (RTOs) or independent system operators (ISOs) in the US: Independent System Operator New England (ISO-NE); PJM Interconnection (PJM); New York Independent System Operator (NYISO); Midcontinent Independent System Operator (MISO). 4 Electric Reliability Council of Texas.


Transmission System; DR resources have been successfully participating in these markets, with most of them found in the Short-Term Operating Reserve.

Network charges and regulation – the electricity sector in the UK is unbundled, with transmission and distribution networks separated legally from each other and from other segments (e.g. competitive generation and retail markets). Regulated network charges are levied on energy suppliers and are then passed on to consumers to recover the cost of maintaining and upgrading transmission and distribution networks. Currently there are a number of opportunities in the network charges to enable the participation of DR resources from large customers (e.g. TRIAD avoidance5, distribution use of network charges, and bi-lateral agreements with network operators).

§ DR resources can make meaningful contributions to system operation and planning, and programmes in different countries have utilised DR resources from various sectors and end-uses

Assessments have estimated significant DR potentials in different countries. For example, the National Assessment of Demand Response Potential in 2009, which was commissioned by the US Federal Energy Regulatory Commission (FERC), estimated that under the most optimistic scenario6, 188GW of peak demand reduction could be achievable by 2019 in the US, representing a 20% of reduction in the projected peak demand without DR in that year7.

In accordance with the US Energy Policy Act of 2005, FERC conducts biennial national surveys of progress in advanced metering and DR development in the US. The reported potential peak demand reduction of DR programmes increased from 29.7GW in the 2006 survey to more than 66GW in the 2012 survey (Figure ES 1). For the NERC regions in the US8, this also marks an increase in the ratio between total reported potential peak demand reduction of DR programmes and total non-coincident summer peak load9, from 3.9% in 2005 to 8.5% in 201110. During the period of 2006-2012, significant increase is seen in the reported potential peak demand reduction of wholesale and C&I DR programmes, each of which takes up around 40% of the total potential peak demand reduction of DR programmes in the 2012 survey. Residential DR programmes grew by 40% in their reported potential peak demand reduction over the same period.

Dispatchable resource adequacy DR like curtailable programmes11 and direct load control (DLC) programmes12 contribute markedly more reported potential peak demand reduction, while some price-based programmes grew rapidly within the period between the 2006 and 2012 surveys (Figure ES 2). In the 2012 survey, curtailable programmes and DLC represent nearly 70% of the total reported potential peak demand reduction. Most of the C&I DR resources are enrolled in curtailable programmes and price-based DR programmes (mainly time-of-use tariffs). The bulk of wholesale DR resources in the US have concentrated on curtailable programmes and demand bidding & buy-back programmes, while they also participate in ancillary markets such as markets for spinning and non-spinning reserves. In comparison, the residential DR resources mainly participate in DLC and price-based DR programmes (mainly time-of-use tariffs), while DLC programmes also suit small C&I customers.

5 Large commercial and industrial customers can reduce their electricity demand for the three half-hourly periods with highest system peak demand in a year, so as to reduce their payment for transmission network use of system charges (TNUOS). 6 The most optimistic scenario assumes universal penetration of smart metering, with time-based tariffs being offered as the default option and with the use of enabling technologies. 7 Based on the NERC assessment, 2008 Long Term Reliability Assessment, for national (non-coincident) summer peak demand, which considers energy efficiency but not DR. 8 Only the US portion of North American Electric Reliability Corporation regions are covered (i.e. Hawaii and Alaska are excluded). 9 Sum of summer peak demand of NERC regions in the US. It is on the non-coincident basis since the peak demand of NERC regions occurs at different hours in the summer months. 10 Calendar year 2005 is the reporting period of the 2006 FERC survey and calendar year 2011 is the reporting period of the 2012 FERC survey 11 Curtailable programmes refer to dispatchable DR programmes (e.g. interruptible load, emergency DR and load as a capacity resource) that require customers to reduce their electricity demand when instructed by utilities or system operator. Interruptible load refers to load subject to curtailment or interruption under tariffs or contracts that provide a rate discount or bill credit for agreeing to reduce load during system contingencies. Emergency DR Programmes give customers a financial incentive to commit to reduce load if pre-defined emergency conditions are triggered. Load as a capacity resource refers to DR that commits to make pre-specified load reductions when system contingencies materialise. 12 In DLC programmes, customers allow utilities or system operator to remotely control their electricity demand in return for DR payment.


Figure ES 1 Reported potential peak demand reduction of DR programmes in the FERC surveys

Note: % of Total Noncoincident Summer Peak Demand in NERC Regions represents the ratio between reported potential peak reduction of DR programmes in the US portion of NERC regions (i.e. excluding Alaska and Hawaii) and total noncoincident summer peak load in the US portion of NERC regions. Source: Based on EIA (2013); FERC (2006, 2008, 2011a, 2012)

Figure ES 2 DR programmes and their potential peak demand reduction in the FERC surveys

Source: Adapted from FERC (2006, 2008, 2011a, 2012)

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

8.0%

9.0%

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

2006 2008 2010 2012

Ratio of D

R P

otential Peak R

eduction and Noncoincident S

umm

er P

eak Dem

and in NE

RC

Regions

Rep

orte

d P

oten

tial P

eak

Dem

and

Red

uctio

n (M

W)

C&I Residential Wholesale Other % of Total Noncoincident Summer Peak Demand in NERC Regions

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012

Curtailable Programme DLC Price-based DR Economic Response in Wholesale Market

Resource Adequacy Economic Ancillary Services Others

MW

C&I Residential Wholesale Other


The size of existing DR resources in the wholesale markets differs between RTOs/ISOs (Figure ES 3). Compared with other regional system operators, PJM and MISO have markedly higher reported peak demand reduction of DR resources13. The reported peak demand reduction of DR constitutes 5-10% of the total peak demand of CAISO, ISO-NE, MISO, NYISO and PJM, while their contribution is lower in ERCOT (2-3%) and SPP14 (3-4%).

DR resources can come from a variety of end-uses. For example, in the US, on-site generation, HVAC and manufacturing are the predominant means of load-reduction (70-90%) for the Emergency DR in PJM forward capacity market, while there is some notable participation of lighting and refrigeration as well. As for the source of DR, around 50% of the peak reduction potential comes from industrial or manufacturing sectors, followed by 20% from some commercial and public customers and 15% from residential end-uses.

As for Great Britain, which is a winter-peaking system, The Demand Side Response in the Non-Domestic Sector in 2012 estimated 1.2-4.5GW (or 3-8% of system peak demand) of DR potential15 in non-residential buildings. Moreover, the GB Electricity Demand project indicated 18GW (or 34% of system peak demand) of loads could be regarded as technically suitable for DR in 2012. This potential would rise up to 25-32GW (or 37-55% of projected system peak demand) in 2025.

However, the current DR participation in markets is modest in the UK. Ward et al. (2012) estimated a size of 1-1.5GW of the DR participation in 2012, with the ‘true’ load reduction (as opposed to demand reduction enabled by back-up generation) amounting to around 400-600MW. Even considering the new DR capacities as procured in the DSBR16 and the GB Capacity Market, the DR participation is at most 2GW, which makes up only around 3% of the maximum winter peak demand of around 58GW. In the Short-Term Operating Reserve of National Grid, which has taken up most of the existing DR resources so far, demand-side resources have contributed 1.4-2GW

13 This may partly reflects the larger number of customers (and thus size of loads) they serve. But advocacy and regulatory efforts to capture DR resources and utilise their system benefits are greater in scale may have contributed to the more significant role of DR in these markets. 14 Southwest Power Pool (US). 15 Amount of end-use demand that is regarded as technically suitable and flexible to be reduced during the system peak period, without considering the economic or other factors likely to influence customers’ participation. 16 Demand Side Balancing Reserve offers payment to large electricity users to reduce demand or use embedded generation between 4pm and 8pm of winter weekdays following the instruction from National Grid. National Grid launched it in 2014, alongside the Supplemental Balancing Reserve (SBR), to address the medium-term anticipated decline in capacity margin during winter periods for the next few years before the first delivery year of the GB Capacity Market in 2018.

Figure ES 3 DR resources in wholesale markets of RTOs/ISOs and their reported potential peak demand reduction

Source: Based on data in FERC (2011b, 2014)

0%

5%

10%

15%

20%

25%

30%

35%

40%

0

2000

4000

6000

8000

10000

12000

14000

16000

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

CAISO ERCOT ISO-NE MISO NY-ISO PJM SPP

Percentage of Peak Dem

and R

epor

ted

Pote

ntia

l Pea

k D

eman

d R

educ

tion

(MW

)

Potential Peak Reduction of Wholesale DR Programmes Percentage of Peak Demand


in recent seasons, representing around 50% of the total STOR resources. However, most of these demand-side resources are in fact supply solutions (e.g. embedded generation), while ‘true’ load reduction has only contributed 110-237MW of capacity, which makes up 4-7% of the resources in recent STOR seasons. Moreover, for the moment, there is also limited DR participation in other balancing services of National Grid and the GB Capacity Market.

§ DR can bring significant benefits to the electricity system

The benefits of DR are of many kinds, including avoided capacity cost, avoided energy cost, avoided network cost, avoided environmental externalities, participant bill savings and other benefits like system security (Table ES1). As the ‘avoided costs’ are not directly observable, assumptions have to be made about what costs would have been incurred in the absence of DR, which nonetheless involves uncertainty. It must be borne in mind that the value of DR resources can vary significantly from one power system to another, depending on the configuration of the power system, on the methodology used in the evaluation of DR, and especially on the basis for building capacity. Where generation and networks are built to meet peak demand, demand may be ‘peakier’ than is optimum if prices do not fully reflect the cost of supply at peak times. In this case, demand response may effectively substitute for the inadequacy of price signals and may avoid the construction of unnecessary capacity.

A number of international studies have estimated the benefits of DR. For example, for the UK, Element Energy and RedPoint (2012) estimated the potential benefits of DR from households and small & medium enterprises at £500m per year in the high demand scenario in 2030. Ofgem (2010) provides another estimation of the potential to avoid generation and network capacity costs in the UK through DR. If 10% (or 4.6-5.7GW) of the peak demand could be shifted, the Ofgem study estimated the annual avoided capital cost savings and the annual network investment savings at £265-536m and £28m respectively. For the US, the Brattle Group used a simulation-based approach to estimate the impacts of DR (power curtailment by 3% at each Mid-Atlantic zone’s peak load) on the PJM energy market in 2005. It concluded that the benefits of DR to the entire PJM system were between $65 and $203 million per year, depending on market conditions. Moreover, regulatory bodies and other stakeholders have used the Avoided Cost Calculator developed by Energy and Environmental Economics to assess avoided cost savings.

Table ES 1 Potential benefits of DR to different activities of the electricity system

Operation Expansion Market1

Generation

• Reduce energy generation in peak times: reduce cost of energy and possibly emissions2

• Facilitate balance of supply and demand (especially important with intermittent generation)

• Reduce operating reserves requirements for increase short-term reliability of supply

• Avoid investment in peaking units • Reduce capacity reserves

requirements or increase long-term reliability of supply

• Allow more penetration of intermittent renewable sources3

• Reduce risk of imbalances

• Limit market power

• Reduce price volatility

Demand

• Consumers more aware of cost and consumption and even environmental impacts

• Give consumers options to maximize their utility

• Opportunity to reduce electricity bills or receive payments

• Take investment decisions with greater awareness of consumption and cost

• Increase demand elasticity

Transmission and distribution

• Relieve congestion • Management contingencies, avoiding

outages • Reduce overall losses • Facilitate technical operation4

• Defer investment in network reinforcement or increase long-term network reliability

Retailing1

• Reduce risk of imbalance


• New products, more consumer choice

Note: 1Only applicable in liberalized systems; 2 Depends on the electricity mix; 3It can be considered a benefit in system where renewable generation is encouraged; 4 Keep frequency and voltage levels, balance active and reactive power, control power factor, etc..

Source: Conchado and Linares (2012), Table 3


§ Technical and socioeconomic enablers are important for the long-term development of DR market

A number of regulatory and market enablers for the development of DR markets are summarized here. To create the market for DR, utilities and system operators should face incentives for using DR as a resource and have confidence in its reliability, while customers need to have sufficient benefits and support for their long-term participation. While this summary draws mainly on examples where wholesale markets exist and final prices are determined through competition in retail markets, many of the considerations are also relevant to other electricity system structures (e.g. vertically integrated utility). However, market regulators and policymakers should take into account the specific conditions in the system of interest in designing enabling regulatory and policy environment.

Business models of DR programmes – there should be opportunities and mechanisms for DR to provide system services. Depending on the ownership structure and characteristics of the electricity sector, different business models can be used to procure DR resources, including retail price-based DR programmes, incentive-based dispatchable programmes managed by system operators, and participation of consumers in wholesale electricity markets (Table ES 2).

Table ES 2 Business models of DR under different electricity industry structures

Market Structure Sponsoring Entities

DR Programme Offerings1 Non-dispatchable Programmes Dispatchable Programmes Types Implementation Types Procurement

Vertical Integration

Vertically integrated utility

Demand-based tariff TOU and CPP

Administrative tariff DLC and curtailable programmes

Bilateral contracts Administrative offerings Dedicated auctions

Single-Buyer Model

‘Single-buying entity’ (or system operator)




Wholesale Competition

System operator Distribution utilities

Demand-based tariff TOU, CPP and RTP2

Administrative tariff Response to/integration with wholesale energy market

DLC and curtailable programmes Wholesale market participation

Bilateral contracts Administrative offerings Dedicated auctions Wholesale markets (‘multiple resource auctions’)

Full Competition

System operator Distribution network operators Retail suppliers

Demand-based tariff TOU, CPP and RTP Time-based network charges




1Generally speaking, non-dispatchable DR refers to price-based DR, while dispatchable DR typically means incentive-based DR. However, the degree of ‘dispatchability’ may differ amongst resource types, depending on factors such as the stringency of the non-performance penalty. 2 In principle, real-time pricing may also be offered under vertical integration or single-buyer model but this is not common.

Enabling regulatory incentives for stakeholders like utilities to promote DR – regardless of the market structure, one vital prerequisite for the long-term development of DR is that key stakeholders recognise the value of DR (e.g. economic, reliability and environmental benefits) and face appropriate incentives to capture these benefits. Financing (at least from private capital) requires that the value stream from DR is reasonably predictable and can be captured by the actors making the investment. For regulated utilities, regulatory incentives (e.g. cost-effectiveness in resource procurement) should exist to encourage them to capture the financial and operational benefits of DR. Moreover, for the long-term viability of DR programmes, regulated utilities should be able to recover programme costs and lost revenues that can be justified. Regulators may even allow additional incentives for these utilities based on their performance in promoting DR. For wholesale markets and restructured electricity industries, predicting and capturing the value from DR may be more problematic as there may be a number of value streams, some of which (e.g. network benefits) may not naturally come directly to investor. In that case, even cost effective DR may not happen because no single actor can capture its value. In short, if there isn’t a market in the relevant value, it will be difficult to know what the real value is, and certainly difficult for any private actor outside the utility to capture it. This points to the importance of a stable, transparent market structure that allows DR to be valued and the benefits to be captured. Regulatory efforts should remove barriers for DR resources to participate in these markets, ensure that pricing signals and market rules provide the stakeholder incentives for engaging in DR, and support the opportunities for DR to earn appropriate returns.


Aggregation can play enabling roles for DR resource procurement – aggregation can build on innovative business models, delivering ‘value-adding services’ to system operators, utilities and customers. Three main aggregation business models include curtailment service providers, support of direct DR procurement of system operator/utilities, and provision of other energy-related services. Benefits of aggregation are multiple, including tapping small DR resources, customer engagement and market development, and performance risk management and providing scope for innovation. More importantly, aggregators should have the ‘natural’ incentive for maximising DR procurement, whereas utilities may earn less revenue when consumers engage in DR. To support aggregators, regulatory authorities and policymakers should promote their access to customers and DR opportunities, create market for aggregated DR products and ensure the financial viability of the business model when the service helps to lower system costs.

Customer technical capability, incentive and education are important – for customers to engage with and participate in DR programmes in the long term, they need the technical capability to deliver DR, appropriate financial benefits to participate in DR programmes, and some awareness of why DR is important and how best to participate. Customers should have access to resources for identifying DR opportunities and enabling technologies. This includes access to aggregators or other sources of technical advice, two-way metering and communication, and energy management and feedback systems to monitor the delivery of DR. Programmes may consider offering subsidised or free technical support and incentives for enabling technologies. Depending on the needs of systems and the familiarity of customers with DR, the programmes on offer should also be diverse to fit with different needs and characteristics of electricity use. Moreover, customers should earn adequate benefits to justify the costs incurred in participation (e.g. direct cost in enabling technology and transactional costs), while the programme payment needs to be in line with the value of DR to the system. In light of this, DR programmes need to give customers confidence in the potential financial returns, and ensure that customer can see the value of DR (e.g. the impact of DR payment on the consumers’ electricity bill). Finally, sustained customer education and outreach are essential for the long-term development of DR programmes. This has largely to do with the different needs customers may have at different stages of the ‘customer journey’ towards understanding how to be a smarter consumer. When customers are not familiar with DR, education and engagement are important to address concerns they may have about the risks of participation. Once customers gain some experience and have a better understanding of the benefits of participation, engagement and outreach can provide opportunities for encouraging them to develop DR capabilities even further.

Reliability of DR resources – electricity utilities and system planning authorities need to have confidence in the reliability of DR resources, with specific requirements depending on the service characteristics (e.g. frequency response or as a capacity resource). This is essential for treating DR as an alternative to supply-side solutions. While specific programme designs (e.g. penalty for under-performance in delivering DR commitment) can give incentives for reliability, technologies that promote controllability of end-uses also have an enabling role. Moreover, it is equally important for programme sponsors to undertake regular evaluation, measurement and verification (EM&V) to strengthen the evidence for the reliability of DR resources. This is especially valuable for non-dispatchable DR programmes, since they do not typically involve incentives for reliability.

Enabling technologies – Advanced Metering Infrastructure (AMI) is one essential technology for market-wide DR programmes, especially the price-based (non-dispatchable) programmes. Other technologies (e.g. smart appliances, energy management systems or process control systems for C&I customers) are also important enablers for the DR programmes, in particular those that require quick response and automation. Moreover, the granular data about real-time demand for various end uses may also be useful in increasing the ‘visibility’ of the impacts of DR actions, and in identifying the potential to provide different DR programmes. Finally, the ownership of on-site generation could strengthen the motivation and ability of customers to provide DR, although on-site generation is really an alternative supply-side resource and could well involve emissions and energy costs that are as great as, or greater than, those avoided by the system.

§ Difference in system operation creates barriers for DR in China, but ongoing electricity market reform provides opportunities

While the drivers for promoting DR and its potential benefits are relevant to China, the experience from OECD countries may not be directly transferable, essentially because the Chinese power system has developed very differently from systems in OECD countries. This difference highlights the need to design DR programmes to reflect the specific regulatory context in China as well as to change the practices and institutional framework governing electricity system operation. It also implies several practical considerations in estimating the potential and value of DR resources.


Administrative demand planning is not compatible with market-based DR. While originating from the 1970s when electricity supply was short, the practice of annual demand planning still exists today. There are a number of limitations to demand planning, such as the economic losses of administrative measures, inadequate incentives for individual customers to change their electricity demand voluntarily, and limited scope for developing DR products to meet different system needs (e.g. beyond system emergency, such as ancillary services to promote the integration of intermittent renewables). The transition from administrative demand planning to market-based DR requires not only regulatory changes, but also appropriate customer engagement and programme design. Moreover, administrative demand planning and its effect of demand suppression can undermine the value of market-based DR.

A rigid institutional and governance framework for resource planning and dispatch can create constraints for DR. As discussed further in Kahrl and Wang (2014), annual generator output plans and unit commitment plans, together with grid operating plans, form the basis for generation dispatch planning17. Annual generator plans typically require the maintenance of certain hours of operation for generation capacities. Together with the fixed schedule for interregional and interprovincial power exchange18, this dispatch model risks constraining the scope for using DR, to the extent that the deployment of DR can have significant impacts on the operating hours of generation plants. Moreover, the multi-level hierarchy of dispatch model may also introduce complexity in the potential sharing of DR resources (and the costs and benefits) amongst provinces.

Inadequate drive for economic efficiency in resource planning and dispatch may disadvantage DR. Unlike the UK and the US, the incentive for economic efficiency is largely absent in the electricity system of China. For example, the lack of optimised economic dispatch (e.g. day-ahead and operating reserves) across all generation types (e.g. coal, gas, hydro and renewables), or even the ad hoc approach to dispatch in some cases, means that less economically efficient units may be running at the expense of more efficient ones. The implication for DR is two-fold: first, the potential benefits of cost-effective DR may not be fully realised in the existing model of system dispatch and operation; secondly, there is little incentive for considering DR as an alternative resource in the electricity system planning.

Lack of pricing signal in electricity system operation makes it difficult to assess DR value. It is often possible to value DR resources in the UK and some US states by referring to what they (or an alternative resource) can earn in competitive wholesale markets for capacity, energy and ancillary services, financial transmission rights and CO2 emission permits. For systems with competitive market mechanisms, DR resources may compete directly against supply-side and other demand-side resources, or be procured by a competitive mechanism to determine the price for specific DR services. For systems without competitive market mechanisms, there is usually some economic information that allows the administrative price for DR to reflect an estimate of avoided costs. By contrast, the electricity system in China is still mostly subject to central planning, with prices, operating hours of power plants and peak demand pre-defined. Since the competitive market mechanism does not exist to determine prices, DR forms part of the planning process rather than being driven by price signals. The inadequate pricing signal makes it difficult to assess the full economic benefits of DR. Moreover, as customers do not face marginal prices that reflect system operation conditions and economic costs, they may see less benefit than otherwise from participating in DR, and have limited incentives.

However, a number of recent regulatory provisions in China should be able to support DR development in the medium- or long-term. In the Opinions on Further Deepening the Power System Reform as issued in 2015, the State Council has not only heightened the importance of DR and other demand-side solutions in ensuring the supply-demand balance, but also highlighted its objectives for electricity pricing reform and introducing market-based mechanisms. Further development in these areas will contribute to the strengthening of pricing signals in system operation, and the flexibility of resource planning and dispatch. Moreover, there is also an intention to cut back on the practice of administrative demand planning in an orderly manner, and to allow voluntary interruptible contracts between customers and utilities. In principle, these efforts should create a more enabling environment for DR.

A ‘phased’ approach for developing the DR market looks appropriate, given the specific regulatory characteristics of the electricity system in China. At the early stage, DR programmes may consider simple designs (e.g. curtailable programmes for resource adequacy to targeted customer groups), even on a pilot basis,

17 Annual generator output plans are drawn by provincial-level planning agencies to guarantee hours of operation for generators. For provinces using energy efficiency dispatch, unit commitment plan is made based on the dispatch order table mandating the order for dispatching generation. Grid operating plan incorporates transmission system security considerations and constraints. 18 Interregional and interprovincial power exchange schedule is fixed before provincial dispatch organisation makes the dispatch plan.


to promote customer interest and understanding of this market-based approach, which is different from the administrative demand planning. Once customers gain more knowledge of participating in DR programmes, opportunities may arise for introducing more DR programmes to fit with the diverse customer characteristics and different system needs. From the viewpoint of the ‘customer journey’, this approach has value in enhancing customer engagement and learning, and sustained customer education and assistance should support it. Utilities and aggregators have very important roles to play in this. Moreover, it is important to conduct regular programme evaluation to develop the evidence of DR reliability, to share ‘best practices’ and to understand where improvement is needed. Meanwhile, on the regulatory level, it is advisable to leverage the opportunities of ongoing electricity market reform to ‘familiarise’ customers with ‘market characteristics’ of electricity prices. Examples can include the introduction of time-of-use tariffs or critical peak pricing to increase the potential benefits for delivering DR during high-priced periods. Regulators and utilities should also improve the accounting system and system service definitions to enable accurate assessment of the value of DR to system operation. This should complement the regulatory incentives for more cost-effective resource procurement and the engagement of utilities.

DR Market Potential and Valuation in Shanghai

Estimating the potential and benefits of DR resources in specific markets is an important part of efforts to promote the utilisation of such non-supply-side resources to provide system services in China. The analysis can help utilities, system operators, government and other key stakeholders in identifying the scale and source of DR resources that can be expected, and their potential values to the electricity system as well as the targets and strategies for DR development.

§ Framework and Methodology for Assessing DR Market Potential

Market potential refers here to the level of potential to reduce demand at times of system peak19, after taking into account practical considerations (e.g. programme design, incentive, customer engagement, characteristics of electricity use, regulatory and market conditions) that can affect the participation of customers in providing DR resources and/or the level of response in reducing electricity demand. International studies for DR market potential are heterogeneous in their analytical approach. However, they share some common features in the overall framework for conducting the potential analysis, one of which is the ‘bottom-up’ approach in assessing DR potential. There are strong reasons to support the use of such ‘bottom-up’ approaches. For example, the potential of various customers to reduce demand and thus deliver DR largely depends on their characteristics of electricity use, which are likely to vary across different customer segments. The existence of specific end-uses and the ownership of enabling technologies can influence the amount of DR individual customers can deliver. Moreover, customer segments may have different capabilities and willingness to participate in DR programmes, given their specific participation requirements, incentives and other programme characteristics. Figure ES 4 shows the framework and steps for the ‘bottom-up’ approach in estimating the DR market potential.

Figure ES 4 Key steps in analysing the DR market potential

§ Framework and Methodology for Assessing DR Benefits

The benefits flowing from DR are calculated by reference to the costs which the DR programmes enable the utility to avoid (known as avoided cost) – i.e. the costs that it would have incurred in meeting the extra demand which would have existed in the absence of the DR programmes. Different types of avoided costs are often considered.

19 It is important to note that DR can be used to address system issues not necessarily occurring during system peak periods (e.g. when large generation capacity unexpectedly becomes unavailable, or the need to integrate intermittent renewable energy).

Determine scope of analysis

Segment customers

Develop average per-

customer load profile

Estimate participation

rate

Estimate average load

impact DR market potential


Avoided costs of new generation capacity – this is derived from the reduction in generation capacity that would have been needed to satisfy peak demand (including a planning reserve margin20 – PRM – and taking account of the avoided incremental T&D network losses) without DR. The estimated DR potential (usually in MW) can be used as a reference point to calculate the relevant capital costs of power plant that would provide equivalent generation capacity. We adopt the generation cost model developed by Energy and Environmental Economics (E3) to calculate the avoided generation capacity cost. Figure ES 5 describes the steps for calculating generation capacity costs.

Figure ES 5 Steps for calculating the avoided generation capacity cost

Avoided energy costs – this can be calculated in various ways, for instance, by using historic and estimated future load profiles along with a forecast of the average value of wholesale energy (or of wholesale market prices where these exist). In a region without a wholesale energy market, avoided energy costs can be determined by comparing the energy costs under two circumstances: one with DR and one without DR. However, it is difficult to estimate avoided energy costs because there is no true counterfactual.

Avoided costs of ancillary services – these are equally open to the difficulty of defining the counterfactual. In some countries, there are markets for these services; the prices in these markets may provide a reasonable estimate of their value to the system.

Avoided transmission and distribution costs – in practice, avoided transmission costs are difficult to estimate because they depend on the time period, the specific location and the overall configuration of system. The starting point is usually to identify future potential network congestion and consider the capacity needed to relieve that congestion in the absence of DR.

Potential costs of implementing DR – it must be borne in mind that there are also potential expenses when implementing DR. For instance, the capital expenditure involved in the installation of DR equipment. The balance between costs and benefits will depend on the nature of the system concerned. In our analysis, due to the absence of information, we do not estimate the potential costs of implementing DR.

§ Estimated DR Market Potential in Shanghai

This study focuses on the DR market potential of direct load control (DLC) for air conditioning (AC) programmes and curtailable programmes up to 2030, based on three scenarios designed to reflect the different levels of participation rate and average per-customer DR load impact. Figure ES 6 shows the assessment results of DR market potential in future milestone years (2020, 2025 and 2030). Under the ‘top-performing’ scenario, the analysis shows that the market potential of DR resources could reach 2.5GW in 2030, representing 4% of the forecast peak demand in that year. The ‘moderate’ scenario assessed the market potential at 790MW in total or 1% of the forecast peak demand in 2030. As for the more conservative ‘basic’ scenario, the DR market potential was estimated at a low end of 214MW or 0.3% of the forecast peak demand in 2030.

20 We have added a reserve margin to the calculation of capacity avoided but that in future analyses the question would require more detailed consideration in the context of the planning and operation of the Shanghai system.

Identify reference power plant and relevant technical

specifications Bottom-up estimate of

investment costs Estimate fixed operation and

maintenance costs Estimate avoided cost of

generation capacity


Figure ES 6 Assessment of DR market potential in Shanghai

C&I curtailable programmes contribute a predominant share in the estimated DR market potential. In the assessment, around 64-73% of the DR market potential is coming from curtailable programmes in the C&I sectors in Shanghai. Industrial curtailable programmes, in particular, may contribute 43-59% of the total DR market potential as estimated for individual milestone years under different scenarios. Under the ‘top-performing’ scenario, out of the 2.5GW of total DR market potential in 2030, curtailable programmes for commercial and industrial customers may respectively deliver 0.5GW and 1.1GW of peak demand reduction.

Residential DLC for AC programmes can make a significant share of the contribution. While DLC programmes for AC for small- and medium-sized C&I customers may make only a minimal contribution, residential DLC for AC programmes constitutes 23-33% of the estimated DR market potential for milestone years under different scenarios. Under the ‘top-performing’ scenario, residential DLC for AC programmes may be reducing peak demand by 0.8GW in in 2030.

§ Estimated Value of Avoided Costs from DR in Shanghai

The total avoided costs of DR in Shanghai include both the avoided generation capacity costs associated with the hypothetical gas-fired plant and the other avoided costs (including avoided energy costs, avoided CO2 emission costs, and avoided T&D costs) associated with the reference plant. This study assumes that the estimated DR market potential is achieved by load shedding only, and not by load shifting or on-site generation21. Figure ES 7 shows the estimated annual total avoided costs between 2020 and 2030. It shows that total avoided costs could reach 811.2 million RMB in 2030, when future cash flows are discounted at 7%. Avoided generation capacity costs contribute most to the total avoided costs (their share ranges between 68.3% and 79.7%), and are followed by avoided energy costs (between 17.8% and 28.1%). Avoided T&D costs and avoided CO2 costs together account for between 2.5 % and 3.8% of the total avoided costs.

21 This is mainly due to the limited evidence as uncovered by this study to show the relationship between load shedding, load shifting and use of on-site generation in contributing to the peak reduction on the programme-level.

1 1 2 4 4

14

21

1 3

6 5 15 22

15

45

100

31

66 102

110 236 364

330

707

1,092

8 24 55 29 89

199

80

242

536

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

3.5%

4.0%

4.5%

-

500

1,000

1,500

2,000

2,500

3,000

2020 2025 2030 2020 2025 2030 2020 2025 2030

Basic Moderate Top Performance

MW

Residential DLC Industrial DLC Commercial DLC

Industrial Curtailable Commercial Curtailable % of Forecast Peak Demand


Figure ES 7 Total avoided costs between 2020 and 2030

However, there are a number of caveats. On the one hand, the assessment only considers dispatchable DR (i.e. incentive-based programmes). Furthermore, due to inadequate information, we have not attempted to estimate the avoided costs of network expansion. For these reasons, our results may under-estimate DR market potential and value. On the other hand, there are reasons why our analysis may overstate the value of DR. First, since the valuation assumes that all market potential is achieved from load shedding only, the avoided costs22 may be lower if part of the potential DR is actually achieved from load shifting or on-site generation respectively. Second, the practice of administrative demand planning has suppressed demand. This may well reduce the ‘real’ economic benefits of DR since capacity may be less than the optimal level to meet peak demand. Third given the lack of detailed projections of peak demand and electricity use (e.g. for 2020-2030), this study made very simple estimates, which may not accurately reflect the future trend of peak demand and electricity use. Finally, we have not estimated the costs of introducing DR.

§ Areas for future research

Finally, this study also identifies a number of areas where further research will be valuable:

Strengthen load profile research. One key challenge of this study is the lack of typical load profiles for different customer segments. Future research will benefit from more rigorous analysis of customer load profiles, especially covering a large sample and maintaining long metering duration. This will help researchers in identifying key patterns in electricity use, and thus refining the customer segmentation. Since AMI has a very high penetration in Shanghai, there are great opportunities to leverage it to improve the understanding of customer load profiles. The load profiles of various end-uses (e.g. lighting, AC, plug loads and refrigerator) will also facilitate the analysis of customer potentials to deliver DR.

Develop a more robust evidence base for participation rate and DR load impact. Many factors may influence the acceptance and capability of customers in delivering DR, including the flexibility potential of business activities and their electricity use, the availability of enabling technologies and their functionality, the cost-benefit case for participating in DR programmes and other solutions (e.g. on-site generation or fuel switching). As the DR pilot develops, more locally specific evidence or data will emerge to show how these factors may differ amongst various customer segments. This will inform future research in considering the market potential for different customer segments. To achieve this goal, comprehensive and regular evaluation of DR programmes is necessary.

22 From a system perspective

0

100

200

300

400

500

600

700

800

900

low high low high low high low high low high low high low high low high low high

2020 2025 2030 2020 2025 2030 2020 2025 2030

Basic Moderate Top performing

Mill

ion

RM

B p

er y

ear

Avoided capacity costs Avoided energy costs Avoided CO2 costs Avoided T&D costs


Analyse empirically the results of the 2015 summer DR pilot programme in Shanghai. The data from this pilot programme could be very useful in identifying the impact of DR on different customer categories and specific end uses. This information could be studied in detail to guide the design of DR programmes in Shanghai and other cities or regions.

Undertake analysis on the relative potential of different DR strategies. There is value from future research to consider the strategies customers are likely to follow to deliver peak demand reduction (e.g. load shedding, load shifting or use of on-site generation) in Shanghai. Such information will support the valuation analysis in terms of making informed assumptions about avoided energy and CO2 costs and system-level avoided capacity costs. This involves better understanding of the characteristics of business activities and their electricity use, the role of enabling technologies in supporting different DR strategies, and the cost-benefit considerations for customers.

Detailed assessment at the end-use level and for other DR types. Due to data limitations, this study could not undertake an assessment of the potential for different end-use categories (e.g. air conditioning, industrial process or refrigeration) or other DR types (e.g. non-dispatchable price-based DR). Future research will benefit from the availability of end-use-level granular load profiles for different customer segments, and more evidence showing the price elasticity of customers regarding their electricity use.

Further define DR products. With the role of DR in system operation and planning becoming more important, it is useful to assess the potential and value of individual DR products, thus offering a more detailed gauge of the DR potential and value for system planning and operation. However, this may add to the need for evidence or data to indicate how different customers are likely to participate and deliver response for various DR products. In other words, this requires a more robust understanding of customers as noted above.

Develop long-term peak demand forecast. The largest benefits of DR are most likely to be the avoided costs in generation and network capacity over the long term. DR programmes typically have a ‘ramping-up’ period before certain levels of participation rate or load impact can be achieved. For these reasons, it is worthwhile to consider the DR potential and valuation over the medium or long term. This requires extending the timeframe for peak demand forecast to a longer term, by taking into account the likely changes that may be expected to materialise in the electricity system and influence peak demand.

Develop a customer engagement strategy. As indicated above, sustained customer education and outreach are essential for the long-term development of DR programmes, and there needs to be provision for them in any comprehensive DR plan. A review of international experience in customer engagement would be a useful initial contribution to this.

Develop better information on system costs in order to better estimate system benefits. More accurate measures of avoided cost require more information on which generation plant will be affected by DR and on the specific incremental cost of those plants. In the longer term, in a market-based system or one where prices reflect marginal costs and dispatch is based on marginal cost, the system’s avoided costs could be calculated by forecasting the hour by hour avoided system electricity costs. Furthermore, it is important to determine whether and to what extent DR would contribute to the avoidance of network capacity expansion.

Develop estimate of DR programme cost. Assessing the cost of DR programmes is essential for understanding the true value they can bring to the system operation and planning. It allows us to estimate the net value of DR programmes and make informed decisions as to how programmes should be designed (e.g. how to reduce programme cost) or whether particular programmes should be launched at all (e.g. cost-effectiveness of DR).

Consider adding other environmental externalities. The current estimates include CO2 as potentially important avoided cost. Future research might include other avoided environmental externalities, including PM10 emissions

Assessment of Demand Response Potential and Benefits in Shanghai Introduction 19

1. INTRODUCTION

The electricity system in China, like that in many other countries, is changing. The low-carbon energy transition requires large-scale integration of intermittent renewables and penetration of electric vehicles, which poses major challenges to system operation and network management. With electricity consumption and peak demand ever increasing, the growing environmental concerns and the policy interest in economic efficiency underline the urgency in changing the way the electricity system is operated and planned in China. While regulatory provisions and power industry policies tend to focus on the supply side, there is an emerging trend of exploring the potential of demand-side in contributing to the supply-demand balance and to the key objectives of carbon emissions mitigation, environmental protection and economic efficiency.

Utilities in China have some experience running administrative load management programmes. As part of orderly electricity use programmes, utilities engage in load shifting and avoidance projects, while restricted use and demand curtailment are also sometimes used to address severe power shortage situations. Moreover, there are four pilot cities in China (i.e. Beijing, Suzhou, Tangshan and Foshan) trialling demand-side management (DSM) programmes. However, such administrative load management programmes typically focus on permanent peak demand reduction or lack flexibility in influencing customers’ willingness to change their electricity usage patterns. Against this backdrop, there is high-level policy interest in promoting demand response (DR), as an alternative, cost-effective solution to address issues in the electricity system.

DR refers primarily to programmes inducing the reduction of electricity use during peak periods, by exposing customers to pricing signals that reflect the system marginal cost of generation or giving them financial incentives in return for their commitment to reduce peak demand when asked to do so23. DR has been successfully utilised to provide various system services in the US, Canada, the UK and continental Europe, with some of these international DR markets being more developed. The National Development and Reform Commission has designated Shanghai as the jurisdiction in China to undertake the first municipality-level DR pilot, with the key objective of exploring the potential for market-based mechanisms to support DR in the long term. In 2014, Shanghai undertook a DR pilot aiming to reduce system peak demand in the early afternoons (e.g. 1p.m.-3p.m.) of weekdays in summer months.

This study contributes to the understanding of the potential for DR in Shanghai by synthesising international experience in using and supporting DR resources, and conducting a preliminary assessment of the market potential and values of DR. Sharing international experience (Section 2), this report reviews the background and current situation of using DR resources, existing efforts in the UK and the US to estimate the potential and value of DR, and regulatory and policy enablers for promoting DR. Given the difference in terms of regulatory and market conditions of electricity systems between China and OECD countries, the report also highlights a number of implications and caveats for interpreting evidence from OECD countries. For estimating the potential and value of DR (Section 3), the study first synthesises the general framework and methodology for the assessment, before drawing on international evidence and local data in Shanghai to gauge DR potential and benefits. This exercise provides an example of how to conduct such assessments and offers a number of recommendations for future research.

23 DR can also be used to provide system flexibility in off-peak periods, for instance to cope with supply disruptions or with intermittent renewable generation. DR includes not only demand reduction, but also increasing demand.

20 Assessment of Demand Response Potential and Benefits in Shanghai International Experience

Assessment of Demand Response Potential and Benefits in Shanghai International Experience

2. INTERNATIONAL EXPERIENCE

2.1 Background

Interest in electricity demand response (DR) is growing worldwide following changes in technology, economics and policy. The underlying issue, central to the operation of electricity systems, is that electricity is difficult and expensive to store, especially on consumers’ premises. Supply and demand generally need to be matched at all times to ensure the stability and safety of the system. Traditionally, the main burden of this matching has fallen on the supply side, while demand has been treated as essentially passive. In practice, what has normally happened is that the level of demand over the relevant timescale (e.g. the next hour, the next day or years into the future) is forecast and the electricity system operator then takes steps to ensure there is enough supply capacity to meet that demand. In a centralised system, most of the tasks are undertaken by a central operator; in a market system, price signals provide the incentives for private companies to build and operate supply capacity but the same broad principle has traditionally applied – the primary task is to meet demand and it is the supply side that is supposed to provide the flexibility to ensure that there is enough generation to meet it.

2.1.1 Rationale for supply side approaches

This supply-focused system has worked well, and has been based on what have traditionally been seen as basic characteristics of electricity supply:

§ Controllability and reliability. The electricity systems have consisted of a relatively small number of large-scale generators (whose operation can be directly controlled by the system operator) supplying millions of customers (whose usage cannot be directly controlled or, in most cases, even monitored in real-time). Electricity system operators have therefore found it much easier to influence supply than demand.

§ Coordination and aggregation. Even if the demand side is directly controllable, the transaction costs and uncertainties associated with coordinating and aggregating a large number of customers have been seen as a major obstacle to the use of demand side resources, especially when instant response is needed.

§ Economics of flexibility. Technically, there is little to choose between demand and supply in terms of flexibility. Indeed, reducing demand is often a simple matter of turning off a piece of equipment24, while generation sources often take time to ramp up to the required level of production. In economic terms, however, there has in the past been much more flexibility on the supply side; start-up and ramp-up costs have been a relatively small part of the total, and generating plants have been built with a view to forming part of an interconnected system of supply and operating accordingly. The demand side is different – electricity is a complementary good and is usually only a small part of a household’s or business’s total costs. Electricity is price inelastic – its value to the consumer is normally much higher than the price being paid, so demand is not very dependent on price. As a result, the so-called Value of Lost Load (VOLL – the amount a consumer would have to be paid to forego a unit of electricity consumption) is normally seen as very high by comparison with the cost of electricity. For instance, when the UK Electricity Pool was introduced in the 1990s, it included an element representing VOLL, which, at £2/kWh, was roughly 100 times the then wholesale electricity price; a more recent estimate puts the value at £17 (LE 2014).

2.1.2 Electricity systems are changing

While there are good reasons for the traditional orientation towards providing flexibility on the supply side, electricity systems are changing and there is a growing interest in considering the demand side as a system resource, just as much as the supply side. This shift reflects some powerful underlying trends:

§ Technological changes are taking place on both supply and demand sides. Intermittent renewable sources, like wind, are increasingly penetrating electricity systems, making the supply side less controllable, while distributed sources such as solar photovoltaic are becoming more widespread, changing traditional patterns

24 However, customers may have to reschedule some of the business or industrial activities, which could involve significant operational changes.

Assessment of Demand Response Potential and Benefits in Shanghai International Experience 21

of supply – it can no longer be taken for granted that the supply side is more controllable and requires less aggregation than the demand-side. Major technical developments are also under way on the demand-side of electricity, including the introduction of smart meters and smart appliances. The growth of localised generation sources can also be seen as a potential demand-side resource, contributing to greater flexibility and controllability on the demand-side.

§ Furthermore, as markets liberalise, operators are getting more used to the idea that price signals, rather than central control, can be used to help balance the system. In many liberalised markets, decisions on plant construction are now a matter for private investors – in some, plant owners self-dispatch rather than being dispatched by the system operator (though the operator still retains responsibility for short term balancing). Confidence in well-designed markets as a way of providing security has grown. Against this background, system operators are also becoming more open to the idea of using the demand-side as a resource, despite the lesser degree of direct controllability.

§ The economics are also changing. The costs of coordination and aggregation and of remote monitoring and control of use on customers’ premises are coming down with the new smart technologies. In addition, the traditional assumptions about VOLL are becoming out of date. Any particular figure, like the £17/kWh cited above, conceals a huge range of different values for particular customers in particular situations. In the past, a single figure was considered necessary because it was not practical to identify separate consumer groups – i.e. it was not possible to cut off individual customers while supplying others in the vicinity. But changes in technology mean that it is now often feasible to identify customers with low VOLLs (or for those customers to identify themselves and make their own decisions about when it is worth their while reducing demand). Meanwhile, the economics of the supply side are also changing. The growth of intermittent generation is requiring flexible generators to adjust their generation output more frequently and their hours of operation are decreasing. As a result, start-up and ramp-up costs are increasing as a proportion of total revenue. The idea that the economics of flexibility favour the supply side is becoming at any rate less clear-cut, if not already out of date.

§ Government policy is also changing in many countries, in particular with an increased emphasis on decarbonisation and reducing environmental impacts. Governments are encouraging the building of renewables and other low carbon plants, most of which are intermittent and non-dispatchable (like wind and solar) or inflexible in economic terms (like nuclear and plants with carbon capture and storage). In short, government policies are helping build in the trend of increasing supply side inflexibility and diversity. Concerns about climate change are also leading to more emphasis on valuing reductions in carbon emissions in particular and lower resource use generally: the power station which remains un-built and the tonne of carbon not emitted are increasingly seen as desirable goals in themselves.

In addition to these changes, rapid technological and economic developments are also occurring in relation to storage. As noted at the beginning of this paper, it is the fact that electricity is difficult to store which underlies much of the interest in DR. If economically viable storage was available, it would have a fundamental effect on system operation. The technologies are progressing rapidly; while not yet fully economic, they may in future have a game-changing impact. Broadly speaking, storage can be considered in three main categories:

§ Centralised utility storage. This is the main traditional form of storage. In many systems, the main form is pumped storage, which uses water as a storage medium. But the viability of this storage type depends on the availability of suitable sites. In general, this form of storage can be considered an alternative to DR.

§ Consumer storage. Technically, when dedicated storage is located on consumer premises, from the point of view of the rest of the system, it essentially forms part of that consumer’s DR potential. Interest has been growing in storage, although often the concern has not been to provide DR but security of supply. For instance, big financial institutions like Goldman Sachs (which have very high VOLLs) often invest in battery storage to ensure uninterrupted supply. Such storage could be a complement to DR - consumers with their own storage find it much easier to reduce effective demand on request.

§ Transport storage. Many countries are promoting electric transport for environmental reasons. If it develops significantly, this would have big implications for electricity systems. While it would add a significant new demand, it could also lead to a scale increase in the consumer storage capacity. It is too early to say how this will affect system operation and the viability of DR. However, major changes in electricity pricing at wholesale and retail level are likely to be needed in order to optimise the integration of this new resource and these changes are also likely to reinforce interest in DR.



Other types of storage can also be relevant; for instance, it is generally easier to store heat than to store electricity. This can open up scope for indirect storage of electricity – for instance, via off-peak electricity combined with storage heating, or by the use of heat pumps with large-scale storage of hot or cold water to provide efficient temperature control. Another area of current interest arises from the very rapid fall in the cost of solar photovoltaic panels – when used in combination with electricity storage (usually battery), solar photovoltaic can enable consumers to be largely independent of the electricity system.

2.1.3 Historical background of DR development in the OECD countries

The history of demand response programmes in the U.S. can be traced back to the 1970s before the restructuring of its electric industry- Regulated utilities would offer time-based tariffs (e.g. time-of-use pricing) and incentive-based programmes (e.g. direct load control, interruptible/curtailable programmes), usually to address rising summer peak demand, largely driven by the greater use of air conditioning. With the electric industry restructuring in the 1990s, competition was introduced in wholesale markets and the system was ‘unbundled’, with monopoly networks separated from competitive generation and (in some cases) retail businesses. Regional system operators – namely regional transmission organisations (RTOs) or independent system operators (ISOs) – were established in some regions to manage the wholesale markets25. RTOs/ISOs are also responsible for the operation of transmission systems, generation coordination and long-term regional planning. Currently there are seven regional system operators26 in the US, which manage around 60% of the electricity flow (EIA, 2011). For other regions (e.g. the Pacific Northwest, Southeast, Southwest and Inter-mountain West), however, the vertically-integrated (or ‘bundled’) utility model is maintained whereby utilities and/or balancing authorities in these areas operate demand response programmes (Hurley et al., 2013).

As the industry restructuring shifted the responsibility for maintaining power system reliability from regulated utilities to RTOs/ISOs, there was less incentive for utilities in those regions to operate demand-side management, which includes demand response. As a consequence, the total spending for demand-side management in the U.S. declined from $2.7bn in 1993 to $1.3bn in 2003 (Hurley et al., 2013). However, subsequent issues after the restructuring (e.g. the Western States Power Crisis of 2000-2001, sharp electricity price spikes) led policy makers to see the value of demand response in ‘the efficient functioning of wholesale electric markets’ (Cappers et al., 2010). As further discussed below, regulatory support and a number of modifications to existing market rules have been provided by Federal Energy Regulatory Commission (FERC) and state regulators to support the participation of demand response in wholesale markets. Providers of demand response resources (e.g. utilities and curtailment service providers) can participate in the energy, ancillary and capacity markets, or otherwise, offer a diversity of cost-effective services for the operation of wholesale markets and/or electric networks.

DR programmes can be largely divided into dispatchable programmes (i.e. can be asked by system operators or utilities to deliver load reduction as needed, in response to system reliability events or high wholesale prices) and non-dispatchable programmes (e.g. with time-varying pricing, where customer response to pricing signals may not be certain). Section 2.4.1 further discusses the types of DR programme offerings. Hurley et al. (2013) classified them depending on the extent of integration with wholesale markets and the characteristics of DR resources:

§ Full Integration with Wholesale Markets. The DR resource can participate and set the clearing price in the energy (e.g. day-ahead or real-time markets), ancillary (e.g. frequency regulation, spinning and non-spinning reserves) and/or capacity markets. A brief description of how DR resources can take part in these wholesale markets can be found in Table 1-3. Utility DR programmes can participate in the appropriate wholesale markets, while curtailment service providers (CSPs), which aggregate the DR resources, and wholesale customers can direct offers into wholesale markets as well.

§ DR Programmes Reactive to Wholesale Markets. Some DR programmes are limited in their abilities in setting the market-clearing price, but they can be dispatched in response to market signals. An example is the day-ahead load response programme of ISO-NE, where the DR resources whose bid price is lower than the day-ahead clearing price are dispatched in the real-time energy market. In other words, DR participation only impacts the price in the real-time energy market but not that in the day-ahead energy market. Given the limited influence of real-time energy markets on consumer costs, the day-ahead load response programme can be seen as one that essentially reacts to market signals.

25 Wholesale markets may include energy markets, ancillary markets and, in some cases, capacity markets. 26 Including ISO New England (ISO-NE), New York ISO (NYISO), PJM, Midwest ISO (MISO), Southwest Power Pool (SPP), California ISO (CAISO) and Electric Reliability Council of Texas (ERCOT)


§ Non-Market DR Programmes. In cases where wholesale markets do not exist in parts of the US (e.g. the Bonneville Power Administration in the Pacific Northwest) or where DR does not participate in the wholesale markets (e.g. California), local utilities or balancing authorities can dispatch DR programmes as cost-effective resources for their system needs.

In Europe, until recently, the energy policy of the EU, and in most of its Member States, has focused more on energy efficiency and its potential in addressing environmental and energy security concerns than DR. Coupled with other factors including the absence of DR-enabling infrastructure (e.g. smart metering infrastructure) and market liberalisation, the development of DR initiatives was ‘largely neglected at policy level’ (Torriti et al., 2010). However, a number of emerging trends such as growing peak demand, ambitious targets for intermittent renewable energy, declining costs in smart metering and other enabling technologies, and the potential of ‘active’ demand-side have heightened the importance of DR in EU energy policies. Article 15.8 of the Energy Efficiency Directive (EED) in 2012 requires Member States to ‘ensure national regulatory authorities encourage’ DR and other demand-side resources to ‘participate in wholesale and retail markets’, and ‘promote the access and participation of DR resources in the balancing, reserve and other system service markets’. Member States also need to remove any disincentive in T&D tariffs hindering DR participation.

In 2014 the Smart Energy Demand Coalition assessed the progress of EU Member States in complying with the EED, based on criteria such as market access, enabling programme design, measurement and verification (M&V) and financial compensation (SEDC, 2014). According to their analysis, Member States vary in the extent to which their regulatory framework is supporting the development of DR. While the DR markets in Belgium, Great Britain, Switzerland, Finland, France and Ireland have demonstrated a relatively high level of commercial development, regulatory barriers in other Member States limit the growth of DR programmes. However, it is necessary to note that other European countries (e.g. Italy, Spain, and Sweden, in addition to those mentioned in the earlier list) also have some experience of using DR resources from industrial sectors (e.g. interruptible programmes) as reserves, or introducing time-of-use tariffs for commercial and residential customers (Torriti et al., 2010).

2.2 Characterising the DR resources

2.2.1 Uses of DR in the electricity system

Matching of supply and demand occurs across different timescales. In light of this, there are various categories of DR – and different policy and market instruments – that have been designed to meet the needs of supply-demand balancing on different timescales. The nature and importance of these categories depend very much on the specific nature of the system concerned (e.g. system characteristics such as peak demand period and drivers of system peak demand, needs for different system services, etc.). Moreover, a further distinction can be made between DR resources for helping balance the electricity system as a whole or ensure resource adequacy in the system (i.e. from a system operator perspective), and those for addressing geographically specific network constraints (e.g. ensuring distribution network has enough capacity to meet the demand). While balancing the system as a whole and ensuring resource adequacy are likely to be more significant concerns in the long term, particularly because generation capacity costs are typically higher than network assets, network constraints (e.g. distribution network constraint) may emerge early in specific locations as a result of changes in demand (e.g. greater uptake of electric vehicles and integration of distributed generation). As discussed later, using DR to address distribution network issues is an emerging area in network operation and regulation.

This section introduces the various system services where DR resources can make contributions. It uses case studies of the US and the UK on how DR can participate in the operation of electricity systems and markets. It is important to note that these electricity systems have different characteristics (e.g. drivers of peak demand, market structure, regulatory and policy background), which have influenced how various DR resources can contribute to system management and operation. For example, as opposed to many regions in the US, the electricity system in the UK is a winter-peaking system, which has an important bearing on what and how different technologies or end-uses can be used to provide DR services. Therefore, the valuation of DR depends on the system characteristics and is not directly transferable from one system to another.



Resource (capacity) adequacy

The biggest potential savings to the system come from long-term reductions in capacity requirements, particularly for generation. In the US and Europe, many unbundled systems have introduced, or are setting up, capacity markets or mechanisms to give incentives for the construction or retention of the generating capacity needed to fulfil the resource adequacy obligations (e.g. load serving entities in the US are required to contract for capacity that is 10-20% above forecast peak demand), and to balance the system at a time of growing intermittent generation. In a number of cases, they allow the participation of DR resources. Such schemes are generally based around ensuring a level of generating capacity or DR to ensure that expected demand could be met, via guarantees from the bidders to provide the supply or demand side resource at a particular time in the future. Table 1 introduces different capacity markets in the US and the UK, and how DR resources can offer into these markets.

Table 1 Participation of DR resources in capacity mechanisms

Programme Description The US Forward Capacity Markets

In some RTOs/ISOs, forward capacity markets are established to procure sufficient capacities to meet the forecast system peak demand (including a reserve margin) several years in the future. Forward auctions are typically held every year to solicit bids of capacities for future delivery-years and there are usually additional auctions near the delivery year to ‘fine-tune’ the capacity procurement. If capacity resources clear the auction, they receive a stream of payments at the auction price and are committed to delivering the pre-defined capacities during system reliability or emergency conditions. Non-compliance with the commitment will result in penalty. Currently there are two forward capacity markets in the U.S. – the Reliability Pricing Model of PJM and the Forward Capacity Market of ISO-NE. Both of them allow the participation of DR and other demand resources (e.g. energy efficiency, distributed generation): • ISO-NE – two DR products (i.e. Real-Time DR and Real-Time Emergency Generation) need to respond within 30

minutes of receiving the dispatch order from ISO-NE; • PJM – Emergency DR (e.g. Limited Summer DR, Extended Summer DR and Annual DR) needs to respond within

30 minutes of receiving the dispatch order from PJM. Other Capacity Mechanisms

DR resources may also participate in other types of capacity mechanisms: • NYISO – DR can take part in the Installed Capacity Market as a Special Case Resource (which receives monthly

payment and has the commitment to respond) and the voluntary Emergency DR programme (which receives the higher of real-time energy market price or $500/MWh). DR resources are called if operating reserves are short.

• MISO – resource adequacy programmes include the Emergency DR Programme (which pays the DR resource at the higher of its offer price or the Locational Marginal Price for its zone), the Load Modifying Resources Programme (DR resources can be used to meet part of the monthly capacity obligation of distribution utilities and are paid based on the contract with utilities), and the DR Resource Programme (DR resource bids into the energy market).

The UK GB Capacity Market

As part of the Electricity Market Reform in GB, a forward Capacity Market has been established for securing sufficient capacity to meet peak demand. Based on the peak demand forecast, for every delivery year, one four-year-ahead auction (T-4) is held, which is complemented by a year-ahead auction (T-1) to refine the capacity procurement. Subject to a number of criteria (e.g. minimum size of 2MW, pre-qualification requirements), capacity resources cleared in the auctions are obliged to deliver the contracted level of capacity (e.g. extra generation or demand reduction) during system stress events, or face penalties if they fail to deliver the contracted level. The Capacity Market Warning (CMW) is issued in the anticipation of a system stress event and becomes effective in 4 hours’ time. During the period when CMW is effective and system stress event prevails, capacity resources should deliver the capacity obligations or otherwise face penalties. If no CMW has been issued before the system stress event, the start of the latter triggers the former and capacity resources must deliver 4 hours after that. The delivery of capacity obligation is ‘load-following’ (e.g. if only 80% of the total contracted capacity is required, each capacity provider only needs to deliver up to 80% of the contracted capacity obligation). The GB Capacity Market allows the participation of DR resources (including embedded generation and small storage) alongside traditional generation capacity, while the Electricity Demand Reduction pilot is trialling whether electric energy efficiency can also participate in the future. DR resources participating in the GB Capacity Market will have an obligation to deliver capacity (‘capacity obligation’) in return for monthly payment determined through the capacity auctions. DR resources can have concurrent commitment to deliver in the GB Capacity Market and the Balancing Services of National Grid (e.g. STOR, but not DSBR as described later). To avoid penalising DR resources, if the delivery of Balancing Services affects the capacity obligation delivery, baseline and delivery will be adjusted to account for the impact. However, if the resources respond to pricing signals, the capacity obligation will not be adjusted. For the purpose of supporting DR participation, separate ‘transitional arrangements (TAs)’ will be introduced with a first transitional auction in 2015. Under Stage 1 of the TAs, DR resources can participate in the two year-ahead auctions for the delivery years of 2016/17 and 2017/18, either as a ‘load-following’ product or a ‘time-banded’ product. Penalty regime for non-delivery of DR resource is also weaker than that of generation capacity.

Source: DECC (2014c); Hurley et al. (2013)

In the UK, National Grid launched in 2014 the Demand Side Balancing Reserve (DSBR) alongside the Supplemental Balancing Reserve (SBR), to address the medium-term anticipated decline in capacity margin during winter periods for the next few years before the first delivery year of the GB Capacity Market in 2018. DSBR offers payment to large electricity users to reduce demand or use embedded generation between 4pm and 8pm of winter weekdays following the instruction from National Grid. Resources need to respond within 2 hours and sustain for at least 1 hour. DSBR is called as the ‘last-resort’ reserve after all other services are exhausted. DSBR is not intended for customers who engage in TRIAD avoidance (see below) or already have STOR contracts for winter periods. The tendering process procures DSBR resources and the payment is based on a


‘stepped schedule’ (e.g. first 25% of delivered capacity is not paid and second, third and forth 25% is paid at 50%, 150% and 200% of the agreed utilisation rate respectively). Resources may also bid for a set-up fee of £10,000/MW to support them in building the capability for demand reduction.

An alternative approach is to encourage long-term reduction in peak demand, for instance by providing incentives for installing equipment with lower demand requirements, with the aim of reducing long-term capacity needs in general. In January 2015, the UK held an auction to procure electric energy efficiency projects based on their potential impact on system peak demand. In the end, the government offered funding to 22 projects that are required to demonstrate via M&V their peak demand impacts during 4-8pm of winter weekdays27. The cost of peak demand reduction in the auction is £229/kW. Most of the demand reduction involved improved lighting system and the auction winners were mainly companies that aggregated demand from multiple sites. It should be noted that the relationship between energy efficiency and DR can be complicated: while better insulated building that is electrically heated may allow customers to engage in DR services for a longer duration, more efficient electric equipment may reduce the potential of DR capabilities of particular end-uses (e.g. the amount of load that can be shifted). In terms of programme delivery, there may well be value in integrating energy efficiency and DR to strengthen the value proposition to customers, and to capture programme delivery efficiency (e.g. installing low-cost control systems in energy efficiency retrofit projects to allow DR participation). For these reasons, it is worthwhile to consider energy efficiency and DR in a coordinated manner.

Economic response

The previous categories refer to the provision of specific services in return for payment. Economic or price-responsive DR refers to a straightforward situation of customers responding to electricity prices by reducing (or increasing) consumption, of their own volition and in their own direct interests rather than because they have undertaken to provide a particular service for a payment. Clearly, for this to be effective, there need to be appropriate price signals in place, as discussed in Section 2.4. Indeed, if all electricity and network prices reflected costs perfectly, there might be no need for further demand side measures. But in practice, all existing systems fall well short of this theoretical ideal, and other mechanisms described in this paper may have a role to play.

Economic or price-responsive DR can include retail time-based programmes offered by utilities (e.g. time-of-use tariffs, critical peak pricing). As shown in Section 2.2.2, some types of time-based programmes (especially time-of-use tariffs) have seen substantial growth in the US in recent years. In the UK, major energy suppliers have been offering static time-of-use tariffs (e.g. ‘Economy 7’ and ‘Economy 10’) for a long time. Customers with night-time electrical storage heaters or who can shift their electricity use to low-cost non-peak hours in the night are likely to benefit from such tariffs. With the rollout of smart metering to residential and small commercial customers, there are attempts by energy suppliers to trial innovative price-responsive DR programmes, including those that are aimed to address the issues of distribution network operation (e.g. dynamic residential price-responsive DR projects under the Low Carbon Network Fund). Given the uncertainty of customer response to price signals, time-based DR programmes are not usually perceived as ‘firm’ resources that can be dispatched for electric system operation.

In some unbundled markets in the US, DR resources can also bid directly into the wholesale energy market, and reduce the power draw from the grid during hours of high wholesale price. With the penetration of intermittent renewables (e.g. wind and solar), DR resources may be used to increase demand in the event of high generation output and low wholesale price28, especially coupled with storage. Examples include:

§ PJM – the Economic DR Programme allows DR resources to clear in the energy market and reduce load on a voluntary basis when the wholesale price exceeds the PJM net benefit price, which is published monthly to represent the price at which the benefit of a reduced wholesale price from the Economic DR resource is greater than the payment to the Economic DR resource.

§ ISO-NE – in the Price Response Programme, DR resources are allowed, on a voluntary basis, to reduce demand when the day-ahead energy market price exceeds $100/MWh, and are paid at the higher of the real-time energy price or $100/MWh29. The Day-Ahead Load Response Programme is another route whereby the DR resource can bid into the real-time market if its price is lower than the cleared day-ahead

27 Weekdays during November and February 28 There is, however, only limited experience on the pilot-basis with this form of DR in the U.S. 29 The Price Responsive DR does not actually ‘clear’ the real-time energy market, and acts only as a ‘substitute’ to the load to be procured in real-time market. So the cost of Price Response Programme is charged to the load on a pro-rata basis outside the real-time market.



market price. In other words, DR participation has an impact on the real-time price but not the day-ahead price. The DR resources will be paid at the day-ahead price for the amount of load-reduction cleared in the real-time market, and at the real-time price for any additional load-reduction.

§ NYISO – the Day-Ahead DR Programme allows DR resources to offer load curtailment into the day-ahead energy market, in a similar way to generation. If cleared, the DR resources are committed to deliver the cleared amount of load reduction in the specified hours. A minimum bid price applies for the DR resources to ensure they only participate during high-priced hours.

Ancillary services

Since the electricity system often experiences short-term changes in the supply-demand balance, ancillary services are necessary in ensuring the reliability of system. Table 2 provides a brief description of how DR resources can participate in ancillary services. In the US, for regions with an unbundled electricity sector, independent system operators (ISOs) or regional transmission operators (RTOs) offer ancillary markets to help manage the operation and balancing of transmission systems; for other regions with a bundled structure, it is the utility companies or balancing authorities that offer such services. In the UK, National Grid, the system operator for Great Britain, maintains the Balancing Services (e.g. Reserve Services and Frequency Response) to support the supply-demand balance of the GB Transmission System. Ancillary services typically include operating reserves and other short-term ancillary services:

§ Operating reserves refer to reserve or back-up capacities that can be brought into operation quickly, in the case of sudden loss of power in the system (e.g. generation break-down, transmission going out of service) or unforeseen increases in electricity demand. In a number of power systems in the US and Europe, it is common for DR resources to provide operating reserves.

§ Other short-term ancillary services (e.g. frequency response or regulation service), which are needed to ensure the maintenance of acceptable frequency, require dispatching resources within seconds or minutes to address system imbalances over even shorter period of time. Some DR resources are able to provide such short-term ancillary services and many power systems in the US make considerable use of such DR.


Table 2 Participation of DR resources in ancillary services

Programme Description The US Reserve Services (e.g. spinning and non-spinning reserves)

Spinning reserve is synchronised with the grid and able to respond to contingency events within 10 minutes, and non-spinning reserves can respond within 30 minutes to replace the capacities of spinning reserves. The typical response duration is 10-120 minutes. Examples include: • ERCOT – the Load Resources Programme is a spinning reserve where the DR resource needs to reduce

demand within 10 minutes of receiving a dispatch order and return to 95% of the pre-event demand within 3 hours after the event. The DR resources are paid at the day-ahead offers of availability. Another programme is the Emergency Interruptible Load Service (non-spinning reserve), in which ERCOT solicits and contracts with DR resources and requires them to be available for dispatch during all or some of the defined hours for different seasons. Contracted resources must respond within 10 minutes of dispatch order and are paid at contract price.

• PJM – Tier 2 of Synchronised (Spinning) Reserve allows the participation of DR via market-based clearing of the DR resource offer, which needs to respond within 10 minutes of dispatch order.

Balancing Services and Load-Following Services

Balancing services (e.g. frequency regulation) and load-following services are for normal fluctuation of demand and supply. These services are becoming more important as the penetration of intermittent renewables increases. The DR resources suitable for such services may include loads with storage (e.g. thermal storage, air/water storage) that can respond almost immediately to a dispatch order. The participation of DR resources in such services (e.g. regulation market of PJM) is very limited so far, but there are attempts in other RTOs/ISOs (e.g. ISO-NE) and regions (e.g. Bonneville Power Administration) to promote or trial the role of DR resources in these services.

The UK Reserve Services (e.g. Fast Reserve and STOR)

Fast Reserve is the quickest reserve service for National Grid to address sudden or unforeseen supply shortages (e.g. unpredictable short-term demand increase, short-term frequency control). It can respond within 2 minutes of dispatch order via increased generation output or reduction in demand. Fast Reserve is subject to a minimum capacity of 50MW and minimum ramp-up and-down rate of 25MW/minute. To ensure operational flexibility, the resource has a minimum commitment of 2 minutes and the normal maximum commitment of 5 minutes to sustain delivery, while it is also required of the capability to commit for at least 15 minutes. Aggregation of units smaller than 50MW is allowed. Fast Reserve is a contract-based mechanism whereby resources deliver the contracted level of power within the agreed parameters if dispatched by National Grid. Monthly tendering processes determine the distribution of Fast Reserve procurement targets (published month-ahead) amongst qualified resources. Short-Term Operating Reserve (STOR) is another reserve service that requires response (e.g. generation or demand reduction) within a maximum of 240 minutes following the National Grid dispatch, while resources that can respond within 20 minutes are typically procured. The minimum period for delivering the contracted level of power is 2 hours. Optimal STOR is the level of power National Grid manages on a daily basis (which is based on the historic availability profile of resources and the amount of STOR MW procured). Resources that can respond within 20 minutes contribute to the Optimal STOR MW, while other resources may also be procured when the economics of doing so are strong. STOR resources should be able to commit at least 3 times a week and have a minimum power level of 3MW. Aggregation is allowed. STOR is a contract-based mechanism whereby resources deliver the contracted level of power within the pre-agreed parameters if dispatched by National Grid. To reflect the seasonal changes in system demand, a total of 6 seasons are designated for STOR procurement (working and non-working days for 3 seasons). Tendering processes that take place three times a year procure the STOR resources, and resources can choose to tender for one or more STOR seasons (subject to the maximum contract length of 2 years). Under contract to National Grid, STOR resources receive availability payment and utilisation payment (if dispatched). There are three types of STOR services:

• Committed service – resources must be available for all Availability Windows during the contract term; • Flexible service – resources can choose the Availability Windows for which they plan to be available; • Premium flexible service – resources can choose the ‘premium’ Availability Windows of greatest value to

National Grid, and National Grid offers to take 85% of the offered days. Frequency Response (e.g. FFR and FCDM)

Firm Frequency Response (FFR) complements other Frequency Response Services in managing the second-by-second system balance. For Low Frequency service (i.e. demand is higher than supply), resources must deliver full contracted level of power within 10 seconds and sustain for another 20 seconds, or deliver full contracted level of power within 30 seconds and sustain for 30 minutes. For High Frequency service, resources need to deliver full contracted level of power within 10 seconds and sustain indefinitely. The minimum level of power is 10MW and aggregation is allowed. National Grid procures FFR via monthly tendering processes. Resource payment consists of an availability fee, a nomination fee (for being called upon to provide the service) and optional fees (e.g. response energy payment for demand-side resource). Frequency Control by Demand Management (FCDM) is needed when there is a large frequency deviation (e.g. caused by significant loss of generation), whereby demand customers are prepared for their load to be interrupted for 30 minutes. The call frequency of FCDM is 10-30 times per year. Resources need to respond within 2 seconds of dispatch and deliver for at least 30 minutes. Minimum level of power is 3MW. National Grid procures FCDM resources via bi-lateral agreement.

Source: Hurley et al. (2013); National Grid (2013a, b, 2015)

Network charges and regulation

The electricity sector in the UK is unbundled, with transmission and distribution networks separated from each other and other segments (e.g. competitive generation and retail services). Regulated network charges are levied on all competing energy suppliers and then passed on to consumers to recover the cost of maintaining and upgrading transmission and distribution networks. Currently there are a number of opportunities in the network charges to enable the participation of DR resources from large customers:

§ TRIAD Avoidance – Transmission Network Use of System Charges (TNUOS) are levied on generators and energy suppliers to recover the cost of transmission network upgrade and maintenance. The share of TNUOS to be borne by energy suppliers is based on the load they serve during three half-hourly periods of



highest system demand in a year (‘TRIADs’), which are not known beforehand but typically occur in the late afternoon or early evening of winter weekdays. Energy suppliers then pass TNUOS on to their consumers. To avoid high TNUOS, large commercial and industrial customers (e.g. half-hourly metered) can choose to reduce electricity demand or use back-up generation during these periods.

§ Distribution Use of System Charges (DUOS) and bi-lateral agreement with Distribution Network Operators (DNOs) – similar to TNUOS, DUOS is charged to energy suppliers and then passed on to consumers to recover the cost of running distribution networks. For large commercial and industrial customers (e.g. half-hourly metered), DUOS is broken down into a number of components: 1) capacity charge, a fixed daily charge based on the maximum agreed load to customer’s site; 2) unit charges that are based on the actual consumption for three differently-priced time periods of day (e.g. 4pm-7pm of weekdays is the highest-priced period); 3) reactive power charges for customers with reactive power (e.g. air conditioning) higher than the pre-determined level; and 4) a fixed daily charge for maintaining connection. Therefore, half-hourly metered customers can reduce their DUOS by reducing demand in high-priced unit charge periods, reactive power use (e.g. by installing capacitors) or maximum agreed load. Alternatively, DNOs can have bi-lateral agreement with large half-hourly metered customers (e.g. non-firm connection agreement, whereby customers allow their load to be curtailed if the distribution network is constrained, in return for lower network connection fees and ongoing DUOS charges) and non-half-hourly metered customers (e.g. radio teleswitch to control electrical heating) to seek an alternative to the reinforcement of distribution networks and defer investment.

The electrification of transport (e.g. electric vehicles) and heating (e.g. heat pumps), and penetration of distributed generation (e.g. solar PV) are at the heart of low-carbon energy transition strategies in the UK. However, they also pose challenges (e.g. increased peak demand and need for substantial investment in network infrastructure) for the operation and management of networks, especially the distribution networks. This provides the context for assumptions that DR will become more valuable in future electricity systems. To support DNOs in finding innovative solutions to these challenges, Ofgem, the market regulator, has established the Low Carbon Network Fund (LCNF) over 2010-15 to enable DNOs to trial new technical, operational and commercial arrangements (e.g. DR and other demand-side solutions). A number of trials focus on the potential of DR resources in dealing with more ‘local’ network problems (e.g. as alternatives to more traditional options such as temporary generation to the affected networks or disconnecting customers), thus deferring the capital investment or providing an alternative solution to managing the distributed network constraints. This can translate into significant financial savings for DNOs and ultimately consumers as well. Table 3 summarises some of the LCNF trials featuring DR solutions for commercial and industrial customers.

As shown above, there is a wide range of DR instruments designed to meet different needs and requirements of electricity system operation. They should be adapted to the specific characteristics of the system concerned (e.g. needs for different system services, overall policy objectives). In terms of procuring DR, there may be difference in the incentive or dispatch regime (e.g. conditions to trigger dispatch, characteristics of response) between the purposes of short-term system balancing, long-term resource adequacy and addressing network issues (especially distribution network constraints). In other words, there is no ‘one size fits all’ approach to developing and utilising the potential DR resources. There may be synergies or conflicts in the utilisation of DR resources by different parties (e.g. DNOs, system operator, or suppliers), depending on whether the same DR provider can serve the requests of all these parties. The timeframe and frequency for implementing DR events differs between parties (e.g. DNOs need DR resources to address network constraints or substation-level peak demand, and the system operator may need such resources more frequently for reserve purposes). Moreover, the procurement of DR resources by DNOs is location-specific, potentially limiting the ability of DNOs to find alternative resources if certain DR resources are already procured by other parties (e.g. clause of exclusivity in some of the contracts). In light of these considerations, there is need for an enabling commercial and/or operational framework to coordinate the procurement of DR and information sharing amongst different parties (e.g. UKPN, 2014)


Table 3 Examples of commercial and industrial DR trials under the Low-Carbon Network Fund

DNOs Projects

UK Power Network (UKPN)

Low Carbon London – C&I Demand Response UKPN contracted with 37 industrial and commercial customers (21 customers for load reduction in water pumping or HVAC with a total DR capability of 4.2MW; 16 customers for embedded generation with a total DR capability of 14MW). Most of the customers participated via third-party aggregators. Compensation for DR customers includes availability payment (£50/70/100/MW/h, depending on whether the DR is load reduction or embedded generation and whether it serves an existing network constraint) and utilization payment (£200/MWh). DR customers must be available for a dispatch request for a time window associated with the network substation peak: 10am-4pm, 12pm-6pm, 9am-9pm, 7am-7pm, 8am-8pm or 2pm-8pm). DR dispatch request is sent to the contracted customers via automated systems without supply interruption. The DR call is limited to once per day, three times a week and ten times per trial period. For most of the trial participants, a maximum response time of 30 minutes is required, while the DR event duration is fixed at 1 hour for 19 customers and allowed for 1-3 hours for 18 customers. Two trials were completed for summer 2013 (June to August) and winter 2013/14 (December to February). A total of 185 DR events (i.e. one DR request to one customer as one DR event) were called, which provided a total response of 254MWh. The average DR event duration was 1.26 hours, providing an average DR of 1.4MWh. The response rate to DR calls was 90% and most of the DR events were responded to within the 30-minute window. Two-thirds of the DR events achieved a compliance rate (i.e. percentage of event duration when actual DR is equal to or higher than the contracted amount) of 90% or higher.

Northern Powergrid

Customer-Led Network Revolution – C&I Demand-Side Response Trials Northern Powergrid carried out its first DR trial for commercial and industrial customers in 2012, with three sites participating via two aggregators. A total of 13 DR calls were made via telephone to the two aggregators, ten of which were responded to. The payment consisted of availability payment and utilization payment, which were based on the STOR market of National Grid. The availability window was 3pm-7pm of weekdays and response time was 20 minutes. The 2nd trial was completed in spring 2014, for which 14 sites had signed up (one directly with the DNO and 13 via aggregators). A total of 33 DR calls were made and 31 of them were responded to. Two different payment models were used: for ten sites, availability payment (£10/MW/h) and utilization payment (£300/MWh) were paid for the delivered DR that was measured with the benchmarking methodology (i.e. baseline determined as the half-hourly metered load immediately before the DR call); for the other four sites, daily charge payment (£306/MW/day) was used with a ‘floor methodology’ (i.e. maximum demand level during the DR event was agreed with individual sites, and the DR delivered was the difference between the agreed maximum demand level and the agreed average demand over relevant time periods). The availability window was 3pm-7pm of weekdays and response time was 20 minutes. A combined reliability rate (i.e. availability rate multiplied by the utilization rate, which measures the percentage of DR calls that are responded to) was 47%, whereas if sites not available for the whole trials were removed, the combined reliability rate was 83%.

Source: MacLennan (2014); UKPN (2014)

2.2.2 Studies of DR potential in the US and the UK

Many studies have estimated the potential of DR resources in different sectors at varying scales (e.g. national, regional or for a specific system or utility service area). These studies typically use a ‘bottom-up’ approach for estimating DR potential. As noted in FERC (2009), there is strong support for such an approach. For example, the potential of customers to reduce demand and thus deliver DR largely depends on the characteristics of their electricity use during system peak periods, which are likely to vary across different customer segments. Moreover, the existence and cost of certain technologies (e.g. back-up generation, energy control and management system, ownership of air-conditioning or other equipment making notable load impacts) is another factor influencing the amount of DR individual customers can deliver. Section 3 discusses in detail the general framework and methodology for conducting DR potential assessment. Here, some of these studies done at the national level are reviewed to demonstrate the scale of actual DR or its potential in the US and the UK.

§ A National Assessment of Demand Response Potential commissioned by FERC for the US in 2009

This study is the first DR potential analysis nationwide in the US, which looked at the achievable potential for DR programmes (e.g. incentive- and price-based) by 2019. This assessment involved:

• Constructing the average load profiles of customer segments (e.g. residential, small/medium/large C&I customers) during the system peak period;

• Estimating the per-customer load impact30 for different DR types and customer segments – for price-based programmes, the estimation relied on analysis of price elasticity and assumed price differential; for incentive-based programmes, it was based on synthesis of existing programme evidence31;

30 Percentage of load during the peak period that customers can reduce to provide DR 31 For direct load control programmes involving central air conditioning only, an average load reduction (in kW) based on FERC survey or default assumption is used; for other incentive-based programmes, percentage of load (%) that can potentially be reduced (based on FERC survey) is used.



• Estimating the participation rate for different DR types and customer segments – this relied on evidence on the current level of participation (e.g. FERC DR survey) and assumptions regarding pricing policies (e.g. if dynamic pricing is the default choice) and penetration of AMI. The estimated participation rates can be seen as reflecting the economic (i.e. benefits and costs customers face in taking part in a particular programme) and/or non-economic factors that can influence the uptake of particular DR programmes in a given population. While considering the benefits and costs of DR for individual customers is helpful conceptually, the heterogeneity of customer segments makes it very difficult to accurately estimate benefits and costs for individual customers. However, as discussed in Section 2.4.6, the benefits customers get from providing DR services must be greater than the costs (e.g. technology investment, transaction costs) if they are to be motivated.

• Cost-effectiveness analysis was also used to screen DR programmes and determine whether to include them in the assessment32.

According to the assessment, if DR programmes available and planned in 2009 could be expanded with a higher availability of advanced metering and other enabling technologies (‘Expanded Business-as-Usual Scenario’), a demand reduction of 82 GW could be achievable by 2019 against the projected demand baseline without DR33 (9% decrease). Under the most ambitious scenario (‘Full Participation’, i.e. universal roll-out of advanced metering, time-based tariffs as default choice and use of enabling technologies), the potential of DR resources could reach 188 GW by 2019, which represents a 20% reduction in the projected peak demand without DR in that year34.

§ Demand Side Response in the Non-Domestic Sector for Great Britain in 2012

This analysis, led by Element Energy Limited, focused on the ‘technical potential’ of DR in non-residential buildings of Great Britain (i.e. the amount of end-use demand that is regarded as technically suitable and flexible to be reduced during the system peak period, without considering the economic and other factors likely to influence customers’ participation). It started by constructing the load profiles for customer segments but, different from the FERC study above, it characterised the load profiles at end-use level (e.g. lighting, HVAC, water heating) based on metering data and available energy consumption data. Then end-uses regarded as technically suitable for DR services (e.g. HVAC, hot water with storage, refrigeration and lighting) were identified, and assumptions were made on the percentage of their load that can be ‘realistically’ reduced to provide DR services, under different scenarios to consider varying degrees of technical flexibility to reduce the demand from different end-uses. These assumptions were made based on expert opinion and existing literature. In other words, the outcome of this analysis should be seen as the ‘technical DR potential’, based on a subjective assessment of technical suitability and flexibility.

The assessment estimated the ‘technical DR potential’ in the non-residential sector of the UK to be 1.2-4.5GW (or 0.6-1.8GW, if lighting is excluded), with retail, education and commercial offices making up around half of the potentials in each assessment scenario. The ‘technical potential’ of 1.2-4.5GW represents 3-8% of system peak demand in the early evenings of typical winter weekdays.

§ GB Electricity Demand – 2012 and 2025 for Great Britain in 2014

Similar to the Element Energy study, this analysis as part of the GB Electricity Demand project also focused on constructing end-use-level load profiles for different customer segments (e.g. residential, C&I). However, this study only identifies the end-uses regarded as technically suitable to be shifted (e.g. space and water heating, wet and cold appliance, and compressed air) without taking into account the technical flexibility (the degree to which specific end-uses may be shifted) or economic factors (benefits and costs customers face in taking part in particular programmes) and other factors that may influence the ability and willingness of customers to become flexible with specific end-uses.

For the system peak period of early winter weekday evenings, the assessment indicated around 18GW (or 34%) of loads as potentially suitable for DR in 2012, with the residential sector contributing 9-10GW of the potential. For 2025, the size of load regarded as technically suitable for DR was estimated to rise to around 25GW (or 37%

32 The Total Resource Cost (TRC) test was used. If DR programmes require the installation of enabling technologies (e.g. dynamic pricing programmes, direct load control), the benefits (e.g. avoided capacity costs) and costs (e.g. technology adoption costs, programme implementation) are estimated for that programme in each state. If the benefits are greater than costs, that programme is then included in the DR potential study for that state. For other programmes not requiring enabling technologies, they are assumed to be cost-effective and included in potential analysis. 33 Based on the NERC assessment, 2008 Long Term Reliability Assessment, for national (non-coincident) summer peak demand, which considers energy efficiency but not DR. 34 Refer to Footnote 33


of the projected peak demand under the ‘business-as-usual’ scenario in the 2050 Pathways of the Department of Energy and Climate Change) or even 32GW35 (or 55% of the projected peak demand, under the ‘greenest’ scenario that assumes significant changes in the electricity system, including high penetration of electric vehicles and heat pumps).

2.2.3 Scale and sources of current DR resources

Besides the DR potential studies, this report also reviewed DR resources that currently participate in various DR programmes offered by the utility retailers or the system operators via wholesale markets in the US and the UK.

Experience in the US

In accordance with the Energy Policy Act of 2005, FERC conducts biennial national surveys of the progress in advanced metering and demand response development. Findings of the recent surveys are summarised below.

§ Total reported potential peak demand reduction36

The results for four FERC surveys between 2006 and 2012 are available (Figure 1). Residential and commercial and industrial (C&I) DR programmes refer to those offered and operated by retail utilities, while wholesale DR programmes37 are those directly bid in the wholesale markets (e.g. energy, ancillary and capacity markets) of RTOs/ISOs or offered by wholesale power marketing agencies including the Bonneville Power Administration and the Tennessee Valley Authority. The way in which DR resources support the electricity system operation (e.g. participation via utility retail programmes or direct offer into wholesale markets) depends on the regulatory and policy context of specific regions (e.g. existence of wholesale markets, regulation and law allowing participation of DR resources in these markets).

The reported potential peak demand reduction of DR programmes increased from 29.7GW in the 2006 survey to more than 66GW in the 2012 survey. For the NERC regions in the US38, this also marks an increase in the ratio between total reported potential peak demand reduction of DR programmes and total non-coincident summer peak load39, from 3.9% in 2005 to 8.5% in 201140. During the period of 2006-2012, significant increase is seen in the reported potential peak demand reduction of wholesale and C&I DR programmes, each of which takes up around 40% of the total potential peak demand reduction of DR programmes in the 2012 survey. Residential DR programmes grew by 40% in their reported potential peak demand reduction over the same period.

§ Programme trends

Dispatchable resource adequacy DR like curtailable programmes41 and direct load control (DLC) programmes42 contribute markedly more reported potential peak demand reduction, while some price-based programmes grew rapidly within the period between the 2006 and 2012 surveys (Figure 2). In the 2012 survey, curtailable programmes and DLC represent nearly 70% of the total reported potential peak demand reduction. Most of the C&I DR resources are enrolled in curtailable programmes and price-based DR programmes (mainly time-of-use tariffs). The bulk of wholesale DR resources in the US have concentrated on curtailable programmes and demand bidding & buy-back programmes, while they also participate in ancillary markets such as markets for spinning and non-spinning reserves. In comparison, the residential DR resources mainly participate in DLC and price-based DR programmes (mainly time-of-use tariffs), while DLC programmes also suit small C&I customers.

35 Around 7GW of the potential is due to the projected significant load of electric vehicles (EVs), which are considered technically suitable for DR. 36 Reported potential peak demand reduction refers to the capabilities of DR programme participants, as opposed to the actual peak demand reduction achieved by these participants, which were reported by the respondents to the FERC surveys. Since not all the entities surveyed responded, FERC used additional information to estimate the scale of potential peak demand reduction, which, however, is not reflected here. 37 As some retail DR programmes are also enrolled in the wholesale markets, the FERC surveys deducted the reported peak reduction of these programmes from that of wholesale DR programmes to avoid double counting. 38 Only the US portion of North American Electric Reliability Corporation regions are covered (i.e. Hawaii and Alaska are excluded). 39 Sum of summer peak demand of NERC regions in the US. It is on the non-coincident basis since the peak demand of NERC regions occurs at different hours in the summer months. 40 Calendar year 2005 is the reporting period of the 2006 FERC survey and calendar year 2011 is the reporting period of the 2012 FERC survey 41 Curtailable programmes refer to dispatchable DR programmes (e.g. interruptible load, emergency DR and load as a capacity resource) that require customers to reduce their electricity demand when instructed by utilities or system operator. Interruptible load refers to load subject to curtailment or interruption under tariffs or contracts that provide a rate discount or bill credit for agreeing to reduce load during system contingencies. Emergency DR Programmes give customers a financial incentive to commit to reduce load if pre-defined emergency conditions are triggered. Load as a capacity resource refers to DR that commits to make pre-specified load reductions when system contingencies materialise. 42 In DLC programmes, customers allow utilities or system operator to remotely control their electricity demand in return for DR payment.



Figure 1 Reported Potential Peak Demand Reduction of DR programmes in the FERC surveys

Note: % of Total Noncoincident Summer Peak Demand in NERC Regions represents the ratio between reported potential peak reduction of DR programmes in the US portion of NERC regions (i.e. excluding Alaska and Hawaii) and total noncoincident summer peak load in the US portion of NERC regions.

Source: Based on EIA (2013); FERC (2006, 2008, 2011a, 2012)

Figure 2 DR programmes and their reported potential peak demand reduction in the FERC surveys

Source: Adapted from FERC (2006, 2008, 2011a, 2012)

0.0%

1.0%

2.0%

3.0%

4.0%

5.0%

6.0%

7.0%

8.0%

9.0%

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

2006 2008 2010 2012

Ratio of D

R P

otential Peak R

eduction and Noncoincident S

umm

er P

eak Dem

and in NE

RC

Regions

Rep

orte

d P

oten

tial P

eak

Dem

and

Red

uctio

n (M

W)

C&I Residential Wholesale Other % of Total Noncoincident Summer Peak Demand in NERC Regions

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

40,000

2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012 2006 2008 2010 2012

Curtailable Programme DLC Price-based DR Economic Response in Wholesale Market

Resource Adequacy Economic Ancillary Services Others

MW

C&I Residential Wholesale Other


§ DR resources in RTOs/ISOs43

Figure 3 shows the reported potential peak demand reduction of DR programmes in wholesale markets of RTOs/ISOs. Compared with other regional system operators, PJM and MISO have markedly higher reported peak demand reduction of DR resources44. The reported peak demand reduction of DR constitutes 5-10% of the total peak demand of CAISO, ISO-NE, MISO, NYISO and PJM, while their contribution is lower in ERCOT (2-3%) and SPP (3-4%).

Figure 3 DR resources in wholesale markets of RTOs/ISOs and their reported potential peak demand reduction

Source: Based on data in FERC (2011b, 2014)

The 2010 and 2012 FERC surveys also included information on how these DR resources participated in the individual wholesale markets (e.g. energy, ancillary and capacity markets) of RTO/ISO in 2009 and 2011. As can be seen in Figure 4, the distribution pattern of DR resources amongst individual wholesale markets varies between RTOs/ISOs. Curtailable programmes for the resource adequacy purpose (e.g. emergency DR programme and participation in the forward capacity market) take up a predominant share of DR resources in ISO-NE, MISO, NYISO and PJM45. Demand bidding & buy-back programme also constitutes a notable share in PJM (e.g. PJM Economic Demand Response programme). In ERCOT, where there is no established capacity market, the DR resources concentrate on providing spinning reserves, while they also contribute to resource adequacy via the emergency DR programmes. For California, the DR resources are predominantly operated by distribution utilities, and not integrated into the wholesale markets of CAISO. But distribution utilities need to inform CAISO of expected load curtailment for the next day.

43 It is possible that retail utilities may offer programmes to aggregate and bid DR resources into the wholesale markets. The FERC surveys deduct, from the wholesale DR programmes, the expected peak reduction capability of those resources that can be linked with retail DR programmes to avoid double counting. However, Figure 5 and 6 only reflect the expected peak reduction capability of DR programmes in these RTOs/ISOs and have not been adjusted for double counting. 44 This may partly reflects the larger number of customers (and thus size of loads) they serve. But advocacy and regulatory efforts to capture DR resources and utilise their system benefits are greater in scale may have contributed to the more significant role of DR in these markets. 45 For PJM, the changes in the enrolment of DR resources in emergency programmes and load as capacity resources between the 2010 and 2012 surveys are largely due to the change in how PJM categorised its Emergency DR programmes in these two surveys and in programme participation.

0%

5%

10%

15%

20%

25%

30%

35%

40%

0

2000

4000

6000

8000

10000

12000

14000

16000

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012

2009

2010

2011

2012


Percentage of Peak Dem

and R

epor

ted

Pote

ntia

l Pea

k D

eman

d R

educ

tion

(MW

)

Potential Peak Reduction of Wholesale DR Programmes Percentage of Peak Demand



Figure 4 DR programmes in RTOs/ISOs wholesale markets and their potential peak demand reduction

Note: CAISO – California ISO; ERCOT – Electricity Reliability Council of Texas; ISO-NE – ISO New England; MISO – Midwest ISO; NYISO – New York ISO; PJM – PJM Interconnection; and SPP – Southwest Power Pool Source: Adapted from FERC (2011a, 2012)

Given the market size of DR in PJM46, this study also reviews how the DR resources are distributed by sector and demand-reducing methods. Emergency DR and Economic DR resources are the two broad categories of DR participation in PJM. Emergency DR participates in the three-year forward capacity market (i.e. the Reliability Pricing Model), and if cleared in the capacity auction, will commit to reduce the demand to a pre-defined level in the system reliability or emergency events. If the DR resource does not deliver the pre-defined load reduction, it will be subject to penalties. There are three types of Emergency DR (i.e. Limited DR, Extended Summer DR and Annual DR), which are differentiated by the availability commitment and maximum response duration. By contrast, Economic DR resources can offer to participate in the energy market and deliver load reduction on the voluntary basis when the wholesale price exceeds the PJM net benefit price1. Economic DR resources may also provide ancillary services (e.g. the Synchronised Reserves, Day Ahead Scheduling Reserves, and Regulation), subject to the availability of infrastructure and the qualification of PJM. For the recent three delivery years between 2012 and 2015, Emergency DR resources contribute nearly 80% of the total peak reduction potential of DR resources (PJM, 2013, 2014, 2015).

As shown in Figure 5, on-site generation, HVAC and manufacturing are the predominant means of load-reduction (70-90%) for the Emergency DR resources in PJM forward capacity market, while there is some notable participation of lighting and refrigeration as well. Regarding the source of these DR resources, around 50% of the peak reduction potential comes from the industrial/manufacturing sector, which is followed by 20% from some commercial and public customers and 15% from residential end-uses.

46 PJM is currently re-examining its approach to demand response in the light of a Court decision last year (http://www.pjm.com/~/media/documents/reports/20141007-pjm-whitepaper-on-the-evolution-of-demand-response-in-the-pjm-wholesale-market.ashx).

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

2009 2011 2009 2011 2009 2011 2009 2011 2009 2011 2009 2011 2009 2011


Rep

orte

d P

oten

tial P

eak

Red

uctio

n (M

W)

Curtailable Programme Direct Load Control Demand Bidding & Buy-Back Ancillary Service Other


Figure 5 Emergency DR resources in PJM by load-reducing method and sector

Source: Based on data from Monitoring Analytics (2015); PJM (2013, 2014, 2015)

Experience in the UK

The current participation of DR resources in the UK market is modest at best. Based on the discussions with key industrial stakeholders, Ward et al. (2012) estimated 1-1.5GW of DR participation in 2012, with the ‘true’ load reduction (as opposed to demand for grid electricity being reduced through use of back-up generation) amounting to around 400-600MW. Even considering the new DR capacities as procured in the DSBR and the GB Capacity Market, the DR participation is at most 2GW, which makes up only around 3% of the maximum winter peak demand of around 58GW.

§ Balancing Services of National Grid

National Grid has classified the resources in the STOR market into two broad categories: Balancing Mechanism (BM) resources and non-BM ones. BM resources are those directly connected with the GB transmission system or large enough to be registered in the BM, while non-BM resources are generally smaller and connected to the distribution network. For National Grid, non-BM resources are perceived as ‘demand-side resources’, but include embedded generation as well as actual load reduction. Figure 6 shows the fuel types of STOR resources in the recent seasons. While non-BM resources have contributed 1.4-2GW in recent seasons, representing around 50% of the total STOR resources, most of them are supply-side solutions such as embedded generation (e.g. diesel

0

1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000

2012/13 Delivery Year 2013/14 Delivery Year 2014/15 Delivery Year

Reg

iste

red

Peak

Dem

and

Red

uctio

n (M

W)

On-Site Generation HVAC Manufacturing

Refrigeration Lighting Water Heating and Other

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

2012/13 Delivery Year 2013/14 Delivery Year 2014/15 Delivery Year

Perc

enta

ge o

f Reg

iste

red

Peak

Red

uctio

n

Industrial/Manufacturing Residential Office Building Schools Hospitals Others



generators, CCGT and OCGT). True load reduction has only contributed 110-237MW of capacity, which makes up 4-7% of the resources in recent STOR seasons.

There is also some participation of DR resources in other Balancing Services of National Grid. For example, the Firm Frequency Response procures 3-5MW from commercial heating equipment; for the Frequency Control by Demand Management, the potential market size for DR resources is estimated to be over 500MW (Ward et al., 2012). In the Fast Reserve service, tele-switched storage heating contributes around 250MW as well.

For DSBR, National Grid has the target of procuring the maximum de-rated capacity of 330MW for the winter of 2014/15 as a pilot, subject to the balanced consideration of the benefits (e.g. potential economic savings from peak demand reduction) and costs (e.g. incentives given for procuring specific resources) of resources being offered. A total of 319MW cleared in tendering, representing resources from 431 individual sites or metering points with the capability to sustain load reduction for 1-4 hours (National Grid, 2014). For the winter of 2015/16, the maximum target for the de-rated capacity for DSBR and SBR is 1,800MW.

Figure 6 Short-Term Operating Reserve (STOR) resources by fuel type in recent seasons

Source: Data from National Grid

§ GB Capacity Market

The first T-4 capacity auction for the delivery year 2018/19 took place in late 2014, which concluded at a clearing price of £19.4/kW/year. Out of the total capacity of 49,259MW awarded contracts, DR resources contributed only 174MW (0.35%). Table 4 shows the amount of capacity, by fuel type, which entered the auction after pre-qualification and cleared the auction. It should be noted that the auction was dominated by existing generation plants, which were effectively bidding for payments in return for undertaking not to close down before the delivery year. In most cases, that involved very little, if any, extra cost for them and as a result the clearing price in the auction was low. More than 70% of the pre-qualified DR resources exited the auction; however, this was a product of particular circumstances and does not reflect the competitiveness of DR resources as against new generating capacity.

There is also some criticism that the capacity market as currently designed could disadvantage DR resources against generation. For example, in the T-4 auctions, new generation capacities can have up to 15 years of capacity agreement, while DR resources can only get a one-year contract. The Energy and Climate Change Committee proposed that an increase in the contract length for DR resources be considered (ECCC, 2015) and the government has expressed the intention to review this concern as well as others over the summer. Moreover, there are other stakeholder concerns that the ‘one-size-fits-all’ model (e.g. number and duration of dispatch, ramp-up time, possible exclusion of time-banded DR products in the enduring mechanism) could risk deterring the participation of some DR resources (SmartGrid GB, 2013).

0

500

1,000

1,500

2,000

2,500

3,000

3,500

4,000

Season 7.3 19/8/2013 - 23/9/2013

Season 7.5 28/10/2013 - 3/2/2014

Season 8.3 18/8/2014 - 22/9/2014

Season 8.5 27/10/2014 - 2/2/2015

BM Total Non-BM Others Non-BM Load Reduction


Table 4 Amount of capacity by fuel type in the first T-4 auction of GB Capacity Market

Capacity Entering Auction (MW) Capacity Clearing Auction (MW) Capacity Exiting Auction (MW) CCGT 31,106 22,259 8,847

Coal/Biomass 13,731 9,232 4,499

Nuclear 7,876 7,876 0

CHP & Auto-generation 4,777 4,235 542

OCGT and Reciprocating Engines 3,446 2,101 1,345

Storage 2,748 2,699 49

Hydro 682 682 0

DSR 603 174 429

TOTAL 64,969 49,259 15,710

Source: Data from National Grid (2014)

2.3 Value of DR resources

2.3.1 Methodology for assessing the benefits of DR resources

There is general consensus that DR can bring significant benefits to the power system. Conchado and Linares (2012) provide a helpful table summarising the potential benefits of DR related to different activities in the electricity market (see Table 5). As mentioned above, the major benefits of DR lie in the avoidance of system costs that would otherwise be incurred to ensure that supply matches demand. As Table 5 suggests, benefits are of many kinds, including avoided capacity cost, avoided energy cost, avoided network cost, avoided environmental externalities, participant bill savings and other benefits (such as improved system reliability). Some can be expressed in monetary terms, which are considered as cost savings to various market actors. For example, DR provides a new source of flexibility to the system operator. It can reduce the need for new peak generation capacity potentially to the benefit of all consumers, and lower the invoiced costs for DR programme participants. Avoided environmental externalities will be beneficial to society as a whole.

Table 5 Potential benefits of DR to different activities of the electricity system

Operation Expansion Market*

Generation

• Reduce energy generation in peak times: reduce cost of energy and possibly emissions b

• Facilitate balance of supply and demand (especially important with intermittent generation)

• Reduce operating reserves requirements for increase short-term reliability of supply

• Avoid investment in peaking units • Reduce capacity reserves

requirements or increase long-term reliability of supply

• Allow more penetration of intermittent renewable sources c

• Reduce risk of imbalances

• Limit market power • Reduce price

volatility

Demand

• Consumers more aware of cost and consumption and even environmental impacts

• Give consumers options to maximize their utility

• Opportunity to reduce electricity bills or receive payments

• Take investment decisions with greater awareness of consumption and cost

• Increase demand elasticity

Transmission and distribution

• Relieve congestion • Management contingencies, avoiding

outages • Reduce overall losses • Facilitate technical operation a

• Defer investment in network reinforcement or increase long-term network reliability

Retailing*

• Reduce risk of imbalance


• New products, more consumer choice

Note: *Only applicable in liberalized systems; a Keep frequency and voltage levels, balance active and reactive power, control power factor, etc.; b Depends on the electricity mix; c it can be considered a benefit in system where renewable generation is encouraged. Source: Conchado and Linares (2012), Table 3

Therefore it is important to evaluate a DR programme from the perspective of different beneficiaries, especially when trying to encourage participation. Moreover, as the ‘avoided costs’ are not directly observable, assumptions have to be made about what costs would have been incurred in the absence of DR, which nonetheless involves uncertainty. It must be borne in mind that the value of DR resources can vary significantly from one power



system to another, depending on the configuration of the system, on the methodology used in the evaluation of DR, and especially on the basis for building capacity. Where generation and networks are built to meet peak demand, demand may be ‘peakier’ than is optimum if prices do not fully reflect the cost of supply at peak times. In this case, DR may effectively substitute for the inadequacy of price signals and may avoid the construction of unnecessary capacity. Section 3.2.2 discusses in detail the methodology for estimating the benefits of DR.

2.3.2 The valuation of DR in the UK

The GB Electricity Demand project is undertaking research to estimate the benefits of DR and to identify the potential for demand response and demand reduction in the GB market. It looks at the value of DR from the perspective of different market actors, including the system operator, suppliers, distribution networks and aggregators (Ward et al., 2012).

§ As mentioned earlier, the system operator in the UK benefits from DR’s potential to provide ancillary services, such as Frequency Response, Fast Reserve and Short Term Operating Reserves (STOR). The total value of these services amounts to £383 million47 per annum. However, consumers with back-up generation supply most of this. The ‘true’ demand response (i.e. without back-up) accounts for only around one-quarter of the demand-side STOR provision, or 200MW. Suppliers of these services, whether on the demand or supply side, are compensated according to their availability and the use of their services by the system – see Table 5 below.

§ Suppliers perceive DR value in terms of their avoided costs48. However, suppliers tend to prioritise secure supply and believe that the potential savings or DR to individual consumers are very small; there are few policy instruments aimed directly at the provision of demand response by suppliers and there has tended to be limited interest from this section of the market.

§ There has been more interest from distributors (i.e. the companies that own and manage the distribution network but do not sell electricity), encouraged by the regulatory shifts mentioned above. Government policy is likely to lead to the development of new loads (such as electric vehicles and heat pumps) which may present challenges for distribution networks. A relatively small shift in demand patterns could lead to a significant reduction in the cost of connecting these new loads – in very broad terms, a 10% load shift away from the peak could enable investment in distribution networks to be deferred by 20 years. So there is considerable interest in DR from this part of the system49. However, avoided costs in distribution networks vary significantly between different network areas (remote vs. central locations) and types (high voltage vs. low voltage). A possible range for average annual avoided costs has been calculated as between £40 and £60 per kW – again a relatively small proportion of the average household bill.

§ Aggregators offer balancing services to the system operator through DR. The services involve up-front investment costs (e.g. hardware and software) and are usually compensated on the basis of participation level, so there is some risk for the aggregator, but also a strong incentive to provide an efficient service. Possible earnings can be quite substantial, amounting to £2-3k pa for a single-site 100kW customer and £1-2m pa for a major multi-site customer.

The GB Electricity Demand study considers avoided costs using the ‘benchmark price’ approach, which lists a range of benchmark prices for different types of services to indicate the possible avoided costs from DR (Table 6).

Table 6 Benchmark price for different type of services

Services Benchmark price and description

Firm Frequency Response (2011-2012) £50 – 60/kW/pa Split between a tendered fee for availability, holding and utilisation

Fast Reserve (2009 average) £50/kW/pa Availability payment: £44/kW/pa; utilisation payment: £6/kW/pa

Short term operating reserve (2011-2012) Availability payment: £8/MWh; utilisation payment: £225/MWh TRIADs TNUOS demand charges: £10.74/kW/pa to £31.17/kW/pa Distribution Network Avoided reinforcement costs £40-60/kW/pa

Source: Ward et al. (2012)

47 It consists of £193 million for Frequency Response, £92 m for Fast Reserve and £98 m for STOR. 48 Suppliers in the UK buy electricity in wholesale markets and resell it to retail consumers. They do not generate, transmit or distribute electricity. Their costs, including the cost of electricity purchased, represent about 50- 60 % of the average household customer end-bill. 49 Distribution charges represent about 18% of the average household customer end-bill.


Element Energy and RedPoint (2012) examined the potential benefits of DR from domestic sectors and Small and Medium Enterprises (SME) in the UK up to 2030. The potential benefits of DR are calculated as avoided capital and operational costs on generation and distribution. This study considers four hypothetical DR programmes, including Static Time of Use (SToU), Load Control 1 (applied on 30 peak days per year), Load Control 2 (applied on all days in year) and Critical Peak Pricing (CPP) under projections of future demand at low, central, high level and also with high level of heat pumps (High HP). Figure 7 shows that the potential cost savings of DR from the domestic sector and SMEs increase over time (because more flexible loads such as electric vehicle become available over time) and approach £500 million per year in the ‘high demand’ scenario in 2030.

Figure 7 Potential cost savings of DR programme from domestic sector and small and medium enterprises by 2030

Source: Element Energy and RedPoint (2012), page 22

The study assumes that avoided peak generation is provided by OCGT plants. The avoided capacity cost is calculated by comparing the peak generation investment with DR (SToU, LC2 and CPP scenarios) and without DR (BAU scenario). The most significant potential for capacity cost savings happen with the CPP in both the High Demand and High HP demand scenarios in 2030. The annual savings related to avoided generation capacity are estimated at £266 million, based on the reduction of peak generation capacity of 3.2 GW of OCGT plants (see Figure 8)50.

These calculated avoided capacity costs are cited in the Parliamentary Office of Science and Technology note (Number 452, January 2014) as an indication of the DR programme benefits. However, the study only examined the avoided capacity costs from DR in the domestic sector and SMEs; it would presumably be much higher if it had included industrial consumers.

50 The reduction of peak generation capacity (MW) and avoided capacity costs (£m) are calculated as the values in the BAU scenario minus the values in the SToU, LC2 and CPP scenarios.



Figure 8 Peak generation capacity investment to meet demand from the domestic sector and SMEs in the UK

Source: Element Energy and RedPoint (2012), page 29

Ofgem (2010) provides another estimation of the potential to avoid generation and network capacity costs in the UK through DR. The demand profiles of two typical winter days in 2009 and 2010 and one typical autumn day in 2009 are used to compare actual and reduced demand (based on assumed 5% and 10% peak demand shifts). The avoided capacity costs are calculated by multiplying the shifted peak load (in GW) by the capital costs of peaking plant (assumed at £93.73/kW/year for CCGT plant and £57.69/kW/year for OCGT plant). The Ofgem report also addressed the avoided network investment, calculated by assuming that the peak demand reduction would result in the same reduction in network investment (i.e. 5% peak demand reduction would result in a 5% reduction in network investment). The annual planned network investment is assumed at £280m. Table 7 presents the results from this study, which confirms that avoided generation capacity costs are the source of most of the potential savings.

Table 7 Avoided generation capacity and network costs

Shift 10% peak load (Between 4.6 and 5.7 GW)

Shift 5% peak load (Between 2.2 and 2.8 GW)

Annual capacity cost savings £265m to £536m £129m to £261m Annual network investment savings £28m £14m

Source: adapted from Ofgem (2010), Demand Side Response, Table 2-1, available from: < https://www.ofgem.gov.uk/ofgem-publications/57026/dsr-150710.pdf>

Other studies estimate the impacts of DR on CO2 emission savings. For instance, Hesmondhalgh et al. (2014) estimate the impact of demand reduction on CO2 emissions in the UK (Figure 9). They show that the impact on CO2 emissions of a 5% reduction in demand is greater over weekday winter peak hours than at low demand periods such as summer weekends51. However, the impacts of DR on CO2 emissions in summer off-peak hours are greater than during off-peak winter hours, because DR in the summer is sufficient to enable the system to avoid use of coal, which would otherwise be the marginal source of generation. The study also projects future emissions reductions; however, in the case of the United Kingdom, the potential for DR to reduce CO2 emission reduction will decline over time because of progressive decarbonisation of the UK power system.

51 In the UK, electricity demand usually reaches its peak between 4 and 7pm in weekday in winter.


Figure 9 CO2 emission savings from demand response (by 5% reduction) up to 2025

Source: Hesmondhalgh et al. (2014), Figure 18

2.3.3 The valuation of DR in the US

In the US, the Brattle Group used a simulation-based approach to estimate the impacts of demand response (power curtailment by 3% at each Mid-Atlantic zone’s peak load) on the PJM energy market in 2005 (The Brattle Group, 2007). The simulation captures the real time fluctuations in supply, demand and transmission and incorporates the associated price changes. It concludes that the benefits of DR to the entire PJM system were between $65 and $203 million per year, depending on market conditions. The energy market price can be reduced by 5 to 8 percent on average with the 3% curtailments of selected zones’ peak load. In addition, the study provides a rough estimation of the potential benefit for DR programme participants. There are two sources of benefits from DR participation: a) avoided energy costs due to reduced load, which is estimated to be $9-26m per year; and b) avoided capacity needed to meet the peak load, which is estimated to be valued at up to $73m.

Other studies adopted the Avoided Cost Calculator that was developed by Energy and Environmental Economics for the California Public Utilities Commission to estimate the avoided costs of electricity (Energy and Environmental Economics, 2004). Three components are considered including the avoided generation costs, avoided transmission and distribution costs, and avoided environmental externalities. The estimation is based on hourly energy and capacity prices at different locations in order to reflect the variations of energy and capacity costs over time and area. The total avoided costs are calculated as the sum of all hourly-avoided costs in one year. The method was extended to include three more components in 2011 (Energy and Environmental Economics, 2011). For example, avoided generation costs are separated into avoided capacity costs and avoided energy costs. It also includes avoided ancillary services costs and avoided renewable portfolio standard52 in its update. The calculator was originally developed to calculate the avoided electricity costs for energy efficiency programmes. In order to address demand response specifically, the California Public Utilities Commission (2010) drew on the above method and made adjustments based on demand response characteristics. The main components in calculating avoided costs related to DR are given in Table 8.

Table 8 Description of key components in the estimation of avoided costs following CUPC methodology

Avoided costs Description

Avoided Capacity Costs The annualized fixed cost of a new combustion turbine, less the net revenues (gross margins) that the CT could earn operating in the real-time energy and ancillary services markets

Avoided Energy Costs Hourly values of energy in both the day-ahead and real-time markets (the appropriate value stream depends on the DR program characteristics)

Avoided Environmental Costs The value associated with a reduction in greenhouse gas emissions resulting from avoided thermal generation

Line Losses Additional costs resulting from line losses between the point of generation and the point of retail delivery

Source: California Public Utilities Commission (2010) 2010 Demand Response Cost-Effectiveness Protocols, Table 3.

52 Renewable portfolio standard is a regulation that requires electricity supply companies to generate a specific proportion of their electricity from renewable energy sources.



Figure 10 presents a comparison of the benefits and costs of a DR programme in a California electric utility using the CPUC method. It shows that avoided capacity costs represent the largest share of total avoided costs, followed by avoided T&D costs and avoided energy costs. The GHG benefit is minor compared to other avoided costs (though of course, this depends on the carbon-intensiveness of the system – California’s is relatively low – and the value given to reductions in carbon emissions). Programme administrator expenses represent the largest share in terms of the DR programme cost. Costs also include operation and maintenance costs, DR system operation and communication costs, marketing and outreach costs, and evaluation, measurement and verification costs. ‘Program Administrator Capital Costs’ include the costs of information technology equipment and demand control technology. The losses in production due to participation in a DR programme are also included in participant annual expenses. This DR programme also involves expenses on hardware, software and equipment; overall, the benefits of DR outweigh its costs. It is noticeable that the net benefits are relatively small, in part because this is only part of the Californian system. But this is a reminder that DR programmes have costs and benefits, for the system as a whole and also for specific agents.

Figure 10 Costs and benefits of DR in a California electric utility

Source: Woolf et al. (2013), Figure 6-1, page 61.

2.4 Enabling frameworks for developing the DR market

This section summarises a number of policies and regulatory provisions that can support the development of DR and its contribution to the system operation. While it draws mainly on examples where wholesale electricity markets exist and final prices are determined through competition in retail markets (e.g. the US and Europe), most of the considerations discussed below are also relevant to other electricity system structures (e.g. vertically integrated utilities). These policy and regulatory ‘enablers’ are organised under several themes, reflecting business opportunities for DR, incentives for supply-side agents and consumers to engage in DR programmes, and technical and commercial pre-requisites for successful DR implementation. It should be noted that these themes represent general principles of ‘good practice’ for DR development, and any policies and regulatory provisions should be considered by reference to specific conditions and context of the electricity system concerned.

Apart from the themes to be discussed below, some implementation and governance issues are nonetheless important to successful customer participation and thus worth considering for the long-term viability of DR programmes. These issues include the legal and administrative framework to enforce contracts (especially for incentive-based programmes), governance transparency, fair treatment of DR resources in system operation and planning (e.g. through independent regulation and system operation), and pricing regimes that reflect the time-specific cost of electricity generation and transmission.

2.4.1 Business models of DR programmes

There should be opportunities and mechanisms for DR resources to provide system services as discussed in Section 2.2. Depending on the ownership structure and characteristics of electricity sector, different business models can be used to procure DR resources, including retail price programmes, dispatchable incentive-based


programmes managed by system operators, and participation of consumers in wholesale electricity markets (Table 9). Price-based DR such as TOU and CPP is typically non-dispatchable, since utilities or system operators are not certain about how much customers will deliver in response to pricing signals; dispatchable DR usually takes the form of contracts for peak demand reduction, in return for financial payment, or participation in the wholesale market. Note that aggregators can play important roles in the procurement and implementation of DR programmes under almost all market structures, which are further discussed in Section 2.4.3.

Table 9 Business models of DR under different electricity industry structures

Market Structure Sponsoring Entities

DR Programme Offerings1 Non-dispatchable Programmes Dispatchable Programmes Types Implementation Types Procurement

Vertical Integration

Vertically integrated utility




Single-Buyer Model

‘Single-buying entity’ (or system operator)




Wholesale Competition

System operator Distribution utilities

Demand-based tariff TOU, CPP and RTP2




Full Competition

System operator Distribution network operators Retail suppliers

Demand-based tariff TOU, CPP and RTP Time-based network charges




1Generally speaking, non-dispatchable DR refers to price-based DR, while dispatchable DR typically means incentive-based DR. However, the degree of ‘dispatchability’ may differ amongst resource types, depending on factors such as the stringency of the non-performance penalty. 2 In principle, real-time pricing may also be offered under vertical integration or single-buyer model but this is not common.

Sponsoring entities

The ownership structure of electricity industry determines who can administer DR programmes. Under vertical integration of generation, transmission, distribution and retail activities, or a single-buyer model where the single buyer procures electricity from generators and other supply sources, the entity tasked with system dispatch and planning (e.g. vertically-integrated utility or ‘single-buyer’) is often the organisation to procure DR resources.

Where wholesale electricity markets are established, the system operator (e.g. ISO/RTO) may allow DR resources to participate, to varying degrees, in different wholesale markets (e.g. energy, ancillary or capacity markets). As shown in Section 2.3, distribution utilities or retail suppliers may also retain ‘legacy DR programmes’ introduced before the restructuring, or procure DR resources as a strategy to avoid high wholesale electricity prices. If full wholesale and retail competition is introduced, independent distribution network operators may also provide DR programmes as an alternative to traditional network reinforcement or constraint management.

Programme offerings

Price-based (non-dispatchable) DR

Price-based DR programmes use demand-related tariffs or time-varying prices to incentivise customers to reduce or manage their electricity demand, especially during system peak periods.

§ Demand-related tariffs charge customers based on their energy use (kWh) and maximum demand (kW). Economists usually consider such two-part tariffs to be more efficient than volumetric tariffs (i.e. charging based on kWh only) to the extent that the kW demand charge reflects the fixed costs of system, with the kWh charge reflecting the variable costs. A typical approach involves contracting for a maximum level of demand (kW), with any extra demand above this level being interrupted53. For large customers, an

53 This is the approach in Spain, where the fixed component of tariff accounts for an important and growing share of electricity bill.



alternative arrangement can be that maximum demand is measured and charged for directly54. Demand-related tariffs can be introduced in most electricity systems, regardless of the degree of restructuring.

§ Time-varying tariffs encourage customers to reduce their demand using higher electricity prices at times of system stress:

• Time-of-use (TOU) tariffs typically raise prices55 during pre-set hours of high electricity demand, and are generally considered to be economically efficient, provided they are designed to reflect the system marginal cost at different times of day. TOU tariffs (e.g. Economy 7 in the UK) may also be packaged with the provision of electric storage heating or other types of storage.

• Critical peak pricing (CPP) raises prices by a significant amount (e.g. 5-10 times of non-CPP price) during hours of maximum system demand for a given number of days in a year. The system operator will notify customers in advance when CPP will be in effect. For instance, in France, there is a ‘red light tariff’, which was introduced to help balance the French system where inflexible nuclear power takes up a high proportion of generation capacity. Under this tariff, consumers have a lower than normal price at most times; however, at times of system stress, they pay a much higher price. These times are notified to consumers by means of a red light on the consumer’s premises. CPP tariffs are generally more effective in reducing demand at times of system stress than TOU tariffs. One study estimated that TOU can reduce peak demand by about 5%, whereas CPP could achieve reductions as high as 20% (Fox-Penner, 2014). A related tariff design involves a critical peak rebate, which does not increase prices during system peak hours but offers a rebate if consumers reduce their demand to a predetermined level.

• Real-time or dynamic prices (RTP) change frequently (e.g. hourly) to reflect real-time system marginal cost or wholesale market price. These prices may be set one day ahead (via day-ahead market) and made public. Consumers may be offered this as a pricing option, or it may be the default for all consumers unless an alternative is chosen (as will be the case in Spain soon). RTP induces efficient DR to the extent that consumers are responding to short-term system marginal cost or price signals. However, there are a number of drawbacks:

o RTP is complex and volatile, and consequently it is typically offered to large customers. This may change as smart devices capable of responding to market prices become more common amongst smaller customers56;

o RTP requires a well-functioning wholesale market (or transparent and accurate short-term system marginal costs) to provide short-term pricing signals. Implementing RTP is problematic where there is no spot market, where wholesale markets are distorted by subsidies (e.g. for renewable technologies, the costs of which are recovered outside the wholesale market) or where the short-term system marginal costs are not transparent or accurate.

While the options of time-varying tariffs depend on the structure and characteristics of a given system, all of them benefit from having a good measure of the short-term system marginal cost in order to pass on appropriate signals to consumers. This requires not only enabling technology (e.g. smart metering and management device), but also an appropriate framework for price regulation.

In several regions (see Section 2.2.1), DR resources, usually from large customers, can offer into wholesale energy market via demand bidding. Once their bids are accepted in the wholesale energy market, these DR resources are required to deliver a pre-determined amount of demand reduction during applicable hours, and thus are dispatchable. This is different from responses to demand-based or time-varying tariffs, which are non-dispatchable due to the uncertainty of customer responses.

Incentive-based (dispatchable) DR

Incentive-based DR programmes typically offer financial payment in return for a customer’s obligation to reduce, by a particular amount, the electricity demand of specific end-uses (e.g. through direct load control) or overall

54 Some systems differentiate between generation and network capacity. In the UK, some large ‘energy only’ customers pay for transmission capacity through TRIAD. 55 Peak price is usually up to twice the level of the non-peak price. 56 The Economist recently wrote that “Nest is selling its programmes all over North America, and more recently in Britain too. Customers of its “Rush Hour Rewards” programme can choose between being given notice a day in advance of a two- to four-hour “event” (meaning their thermostat will be turned down or up automatically) or being told ten minutes ahead of a 30-minute one. This can cut the peak load by as much as 55%. In another scheme customers agree to a change of a fraction of a degree over a three-week period.” (The Economist, January 7th 2015)


usage (e.g. interruptible contracts), if pre-defined system reliability or other conditions occur. This can be done through a variety of mechanisms.

§ Direct procurement of system operator/utilities with administratively determined payments or rebates. This model can be implemented in most electricity systems, whether vertically integrated or unbundled. For instance, the Hawaii Electricity Company controls 250,000 electric water heaters. When power is short, they can turn off some or all of the water heaters for brief periods.57 In Austin, Texas, the local utility has signed up 7,000 consumers for a scheme in which they get an $85 rebate on an internet-enabled thermostat. This enables the local power company to reduce summer peak demand by 10 MW by automatically lowering the temperature on thermostats58. In other programmes, the system operator sends instruction to customers for reducing demand, and the customers are penalised if their demand is not reduced to the level previously specified. Examples can be found in France and Spain, where large customers are compensated for agreeing to reduce their demand when the system operators instruct them to do so.

§ Participation in wholesale market. For regions with wholesale markets (Section 2.3.1), DR may be allowed to compete against supply-side and other demand-side resources in ancillary or capacity market. If DR resources clear in these markets, customers will commit to deliver demand reduction of a specified amount in accordance with market rules. Similarly with DR directly procured by system operator or utilities, upon receiving dispatch instruction from the system operator, customers with an obligation to deliver demand reduction are required to do so, in compliance with their contract, or face penalties.

The conditions for interruptions, either through direct control or in response to instructions, are normally specified in the contract with the system operator or local utility. Customers typically receive compensation. For instance, in Spain, large customers received over €700 million in 2013 as payment through the interruptible tariff. Depending on the programme design, compensation can take different forms:

§ Regular payment (e.g. monthly) for the obligation to deliver demand reduction when needed, or the availability for providing demand reduction (e.g. forward capacity markets in ISO-NE, PJM and GB, Special Case Resource in the Installed Capacity Market of NYISO);

§ Payment when DR resources are utilised, meaning that providers of DR are paid based on the amount of demand reduction they have actually delivered (e.g. Emergency DR Programme in NYISO);

§ Combination of availability and utilisation payment (e.g. STOR market and frequency response programme of National Grid and C&I DR pilots of Low Carbon Network Fund in the UK).

Procurement mechanisms for incentive-based DR

There are three main mechanisms for procuring incentive-based DR resources:

§ Bilateral contracts can be used to procure DR resources, which typically specify the requirements and compensation for delivering demand reduction but do not involve competition between resources (e.g. Frequency Control by Demand Management of National Grid in the UK);

§ Administrative offerings provide pre-determined standard financial compensation to customers that sign up to these programmes. Whether regulated payments for interruptible service are economically efficient depends on the compensation level, programme design and market conditions (e.g. electricity price and benefits of electricity demand reduction);

§ Dedicated interruptible service auctions can be used to determine who provides this service and at what price. The Spanish government recently decided to replace the administrative interruptible tariff with an auction, in which industrial customers bid to supply interruptible services. The government auctioned 2000 MW of interruptible load, including two products, each with its own auction: 1,190 MW from 238 blocks of 5MW and 810MW from 9 blocks of 90 MW. The resulting cost was €500 million, with more companies providing the service than had been the case for the administrative tariff (which was restricted to very large customers). The entire system should benefit from this sort of auction because it can set price at a competitive level if the auction is well designed. The disadvantage is that designing auctions is not easy and there is room for gaming (anti-competitive behaviour) and hidden subsidies. In the GB Capacity Market, the long-term idea is to make DR directly compete with generation capacity by 2018. However, until then, the

57 Peter Fox-Penner, Op. Cit., page 260. 58 The Economist, Special Report: Energy and Technology: Energy Efficiency, January 17, 2015, page 2.



transitional arrangement is to reserve part of the capacity procurement targets specifically for DR. Because this is a new concept for customers, the regulator is considering ways to encourage consumers to participate, for instance by reducing the penalties for not meeting commitments, defining products that can be easily met by DR (e.g., those of relatively short duration), and offering a service whereby DR can be matched with generators to provide a combined service.

§ Multiple resource auctions (MRA) are the model employed in wholesale electricity markets whereby demand-side resources like DR compete against supply-side resources to meet short-term energy and system needs (e.g. demand bidding in wholesale energy market, ancillary services) and/or long-term resource adequacy requirements (e.g. capacity market). To the extent that DR and other demand-side resources such as energy efficiency are less costly than supply solutions, MRA may be an economically efficient way to meet system needs. However, these auctions are complex to design and organise, and typically require well-functioning wholesale markets. This approach also requires a substantial rethink of the way in which markets work once consumers become an important source of DR and other demand-side resources (e.g. distributed generation and storage). Some of the reforms are relatively straightforward, such as facilitating aggregation of loads so that consumers can participate in existing markets. Other proposals are more innovative. For instance, the regulator in New York State has proposed a new system for integrating demand-side resources into wholesale markets59. Colleagues at OIES have also proposed an innovative solution that distinguishes between “as available” energy (typically zero marginal cost) and “on demand” energy (requiring fossil fuel energy, with marginal costs). This is designed to encourage consumers to reflect prices in their consumption and investment decisions60.

2.4.2 Regulatory incentives for industry stakeholders to promote DR

The governments or market regulators in many jurisdictions globally require electric utilities to reduce system peak demand (e.g. quantifiable targets) or integrate cost-effective DR and other demand-side resources in their resource portfolio (e.g. resource adequacy and cost-effectiveness requirement). In restructured electricity markets, there is also high-level policy intention to support the participation of DR and other demand-side resources. Regardless of the market structure, one vital prerequisite for the long-term development of DR is that key stakeholders recognise the wide value of DR (e.g. economic, reliability and environmental benefits) and face appropriate incentives to capture these benefits. As explained in Section 2.3, the benefits of DR are diffused across different segments of the electricity system, which implies that the regulatory framework for promoting DR may also vary depending on the ownership structure and conditions of any specific market. Based on the experiences in the US and the UK, key regulatory provisions to incentivise regulated utilities and wholesale markets to promote DR and other demand-side resources (e.g. DSM projects) are summarised here.

Regulated utilities

Recognition of the value of DR resources (e.g. economic, reliability and environmental benefits) underpins the interest of regulated utilities in promoting DR over the long term. In light of this, the regulatory framework should put in place adequate incentives for utilities to consider DR as an economic efficient, alternative class of system resource. This can be done by introducing requirements on the resource procurement of utilities (e.g. cost-effectiveness, environmental impacts and management of risk in resource procurement).

However, given the likely impact of DR on the business of utilities, additional regulatory enables are necessary to engage utilities (e.g. vertically-integrated utilities, ‘single-buyers’ and regulated distribution utilities/network operators in restructured markets) in promoting DR in the long run. These impacts may include:

§ Programme cost recovery – available funding for DR programmes is essential. As costs are incurred in the design and operation of DR programmes (e.g. communication and management infrastructure and customer payment), it is vital that regulated utilities have the opportunity and measures to recover investment and operating costs, including a reasonable rate of return.

§ Lost revenues – This is mainly because DR and other demand-side resources (e.g. energy efficiency) can reduce the need for investment in the infrastructure of generation, transmission and distribution, on which utilities are usually allowed to earn a rate of return. Moreover, DR and other DSM programmes may even lead to revenue losses from a decline in electricity sales, thus impacting the recovery of utility fixed costs.

59 http://www3.dps.ny.gov/W/PSCWeb.nsf/All/26BE8A93967E604785257CC40066B91A?OpenDocument 60 Malcolm Keay, John Rhys and David Robinson, “Electricity Markets for a Distributed Generation Era”, in F.P. Sioshansi (ed) Distributed Generation for the Utility Industry, Academic Press, 2014, page2 177-182.


A number of policy options have been used in some jurisdictions of the US to address the issues of programme cost recovery and lost revenues, and even to give financial incentives like decoupling for regulated utilities to procure and invest in demand-side resources (Figure 11). These have been designed primarily to incentivise utility investment in energy efficiency, in which price-regulated electric utilities are the major actors in most of the US. However, some of the same issues apply to DR. It should be noted that the exact form of these tools varies among jurisdictions and is determined by a variety of local factors.

§ Programme cost recovery

Regulated utilities can recover the prudently incurred programme costs (e.g. clearing cost-effectiveness test) in three major ways:

• Surcharge on customer electricity bills. System benefit charges (SBC) or tariff riders, which are levies on electricity bills and earmarked for programmes including DR, energy efficiency and/or renewable energy development, are the most widely used approach to recovering programme costs. SBC is often charged as a percentage of electricity bills or a fixed charge per unit of electricity use, while a tariff rider is a surcharge item on the electricity bill. SBC and tariff riders are typically subject to adjustment on an annual basis to balance the collected surcharges and actual programme expenses.

• Treatment as utility revenue requirement in the rate case proceedings. This approach incorporates the programme costs into the revenue requirements during periodic ratemaking proceedings (e.g. to set retail electricity prices). However, the delays between regulatory ratemaking proceedings constitute a disincentive for regulated utilities to increase their spending on DR and other demand-side resources.

• Deferral accounting. Under this approach, programmes costs can be amortised as regulatory assets in ratemaking proceedings over time. This addresses the disincentive to increase programme spending between ratemaking proceedings. Moreover, perceived risks in future ratemaking proceedings or market conditions may raise concerns regarding the recoverability of such regulatory assets.

§ Lost revenues

Utility regulators in the US have designed three types of regulatory provisions to address the disincentive of utilities due to the potential impacts of DR and other demand-side programmes on electricity sales.

• ‘Decoupling’ approach – this has been introduced in some jurisdictions, notably California, to break the connection between utility revenues and electricity sales. It typically involves determination of the revenues that can be recovered, based on the eligible or allowed fixed costs and the variable costs. If the actual revenues are lower (higher) than allowed, the shortfall (excess) can be adjusted by collecting (returning) the difference from (to) customers.

• Lost revenue adjustment – similar with some approaches in recovering programme costs, this method involves calculating the lost revenues that can be attributed to the DR or other demand-side programme and recovering the lost revenue in future ratemaking proceedings or from surcharges on electricity bills.

• Straight fixed-variable pricing (SFV) – as the lost revenues affect the recovery of utility fixed costs, this approach recovers such costs from a separate charge per customer, which is independent from the actual electricity sales.



Figure 11 Regulatory mechanisms for recovering DSM costs in the US

Source: Adapted from Hedman and Steiner (2013)


§ Incentive for performance

While the utility often earns a return on supply-side investment, DR or DSM does not produce similar financial incentives, which creates a bias towards supply-side solutions. A number of states in the US have introduced mechanisms to allow utilities to earn some return on the investment in demand-side resources.

• Shared-saving mechanism – this provision allows utilities to retain part of the net benefits (i.e. avoided system costs minus programme costs) of demand-side programmes. Shared-saving mechanisms usually specify a minimum level of programme achievement before the net benefits can be shared between utilities and customers.

• Bonus payment for performance – another mechanism will reward utilities if they have achieved a pre-defined minimum demand-side programme performance. The bonus payment can be awarded based on programme spending, achievement of programme goals or net benefits.

• Rate of return – in ratemaking proceedings, the investment in DR and other demand-side resources may be treated as regulatory assets and thus allowed a rate of return. In some regions, higher rates of return are allowed for demand-side programmes than supply-side resource investment.

While the electricity sector in the UK is highly deregulated, transmission and distribution are regulated network utilities subject to price-control processes. To support the use of DR as an alternative solution to distribution network operation and management, a new regulatory price-control framework, RIIO-ED1, has been established to incentivise DNOs in GB to find innovative, cost-effective solutions for the challenges associated with a low-carbon energy transition. The RIIO-ED1 price control model, which sets out the revenues DNOs are allowed to collect during 2015-2023, integrates incentives for innovative network solutions in long-term distribution network planning (i.e. revenues are a function of output, innovation and incentives). Over the price control period of 2015-2023, the DNOs have noted in their business plans the drivers for developing DR, including opportunities to defer network reinforcement investment and to manage network constraints or planned outages.

Regulation also has an important role to play in promoting energy efficiency, which has the potential to reduce peak demand. Regulators can often require utilities or energy suppliers to procure energy efficiency resources (e.g. supplier obligation for subsidising energy efficient equipment or undertaking retrofits) or mandating more efficient buildings and electrical appliances. This approach can be applied in most electricity systems, regardless of the degree of liberalization. The main economic attraction is the potential reduction in system costs. Specific consumers may also benefit directly if more efficient equipment reduces their own electricity consumption and/or improves their level of comfort (e.g. warmer house), or if they receive subsidies. However, energy efficiency improvements typically lead to rebound effects, where a lower cost of energy services leads to an increase in consumption. Furthermore, these programmes can be expensive, and in the case of supplier obligations, some consumers may win and others lose, depending on how the costs and benefits are allocated. It is therefore difficult to make a general assessment of the long-term efficiency or distributional impact of this sort of measure.

Wholesale markets and restructured electricity industry

For wholesale markets and restructured electricity industries, predicting and capturing the value from DR may be problematic as there may be a number of value streams, some of which (e.g. network benefits) may not naturally come directly to investors. In that case, even cost effective DR may not happen because no single actor can capture its value. If there isn’t a market in the relevant value, it will be difficult to know what the real value is, and certainly difficult for any private actor outside the utility to capture it. This points to the importance of a stable, transparent market structure that allows DR to be valued and the benefits to be captured. Regulatory efforts should remove barriers for DR resources to participate in these markets, ensure that pricing signals and market rules provide the stakeholder incentives for engaging in DR, and support the opportunities for DR to earn appropriate returns.

Price in the electricity wholesale markets can reflect the value of different resources to the system. For the DR resources bid into wholesale markets (e.g. through a regulated utility or CSP), the payment and revenues from participating in wholesale markets may well allow programme sponsors to recover their programme costs and even earn some profit margin. The system operator maintaining wholesale markets is able to pass on the costs of procuring DR resources to final consumers as part of the wholesale market charges.



As wholesale markets have traditionally focused on the participation of supply-side resources, there are barriers that undermine the incentives of industrial parties to develop DR resources and offer them into the wholesale markets (FERC, 2009). In accordance with the Energy Policy Act of 2005 that emphasized the importance of DR programmes, FERC has introduced several orders to remove the barriers for DR participation and to strengthen the incentives for industrial parties to engage in the DR programmes (Hurley et al., 2013):

§ Order No.890 (February 2007) to allow non-generation resources to provide specific ancillary services, and to require DR to be treated as comparable to traditional resources in the transmission planning;

§ Order No.719 (October 2008) to require the energy market price to reflect the value during the system shortage so as to promote DR and other new resources, and to allow participation of DR resources in ancillary services and aggregators to bid into wholesale markets (unless this is forbidden by law or regulation);

§ Order No.745 (March 2011) to require DR resources to be compensated ‘the full market price of energy’61, the cost of which will be recovered from customers benefiting from lower energy market price due to DR participation, so as to strengthen the economic case for DR62;

§ Order No.755 (October 2011) to require compensation for DR resources providing the regulation service to reflect the difference in response speed and performance.

In the UK, since electricity retail supply is not regulated (i.e. only networks are), there is traditionally little regulatory involvement with demand-side resources controlled by consumers. Electricity supplier energy efficiency programmes have been significant, but driven by specific obligations on suppliers rather than price incentives. Since 2012, these obligations have been focused almost entirely on gas. So despite there being a very large potential for efficiency improvement in electricity use (DECC, 2013), the only relevant policy instrument is the Electricity Demand Reduction pilot, which may be integrated into the GB Capacity Market.

For DR services procured by the National Grid or DNOs, the associated cost can be integrated into the wholesale or network charges and passed on to customers. However, as noted by Ofgem (2013), some disincentives exist for industry stakeholders in furthering DR development, for which reason a number of regulatory changes are proposed. For example, besides the GB Capacity Market and RIIO-ED1 price control framework for distribution networks as discussed earlier, another change is reform in the electricity balancing code and settlement arrangement. In the GB wholesale electricity market, an imbalance price (or ‘cash-out’ price) is to be charged to generators and energy suppliers if their generation or demand is different from the contracted position. In essence it is seen as an incentive for generators or energy suppliers to ‘balance’ their trading position in the GB market. However, due to the lack of sufficient short-term price signals to fully reflect the ‘value of flexibility and security of supply at times of system stress’ (Ofgem, 2013), incentives for DR are sub-optimal. For this reason, an Electricity Balancing Significant Code Review has been launched by Ofgem to find solutions. Moreover, as most small customers’ electricity demand is settled in the wholesale market based on estimated profiles, energy suppliers have little incentive to encourage DR as a trading option for these customers. With the rollout of smart meters63, it would be possible to use accurate real-time demand data to improve the settlement process, and create opportunity for DR.

2.4.3 Roles of aggregation

Aggregation by organisations other than electricity utilities (i.e. stand-alone or utility-owned commercial companies) is not common for energy efficiency programmes in North America and Europe, where regulation tends to focus on requiring distribution utilities or retail suppliers themselves64 to procure energy efficiency improvements. By contrast, aggregation of DR resources is more common, as the activity tends to benefit system operators and network companies with no direct access to customers.

61 If the DR resource can help the supply-demand balance and is cost-effective 62 This Order has been overturned by the D.C. Circuit court on the grounds that the FERC did not have jurisdiction to pass it, and is now the subject of an appeal. http://breakingenergy.com/2015/01/27/another-step-forward-for-demand-response-ferc-order-745-case/ 63 According to the latest Impact Assessment DECC, 2014a. Impact Assessment (IA) for Smart meter roll-out for the domestic and small and medium non-domestic sectors (GB). Department of Energy and Climate Change., the smart metering roll-out in residential and small and medium non-residential sectors in GB costs £10.9bn with the benefits of £17.1 over the period to 2030. 64 In some states in the US (e.g. New York, Vermont and Oregon), a state agency or appointed contractor administers ratepayer-funded energy efficiency programmes.


Business model of aggregators

Aggregation can build on innovative business models, delivering ‘value-adding services’ to system operators, utilities and customers. There are three main categories of service that aggregators may provide:

§ Curtailment service providers (CSPs). In some regions in the US (e.g. ISO-NE, NYISO and PJM), where retail and wholesale competition is introduced, non-utility companies like CSPs can aggregate customer DR resources and bid them directly into wholesale markets. A number of CSPs such as EnerNOC and Kiwi Power are also active in the DR markets of the UK.

§ Support of direct DR procurement of system operator/utilities. Aggregators may collaborate with system operators and/or utilities in administering DR programmes, under a variety of market structures. In some cases, this means the system operator and/or utilities ‘buying’ DR resources developed by aggregators.

§ Provision of other energy-related services. For customers, these services can include solutions for energy use and cost management, energy efficiency, performance optimisation and information-enabled analytics. For system operators and/or utilities, aggregators may be able to provide support for customer engagement, advanced analytics for monitoring programme performance and identifying potential opportunities, and energy efficiency programme expertise and tools.

Benefits of aggregation

The participation of aggregators in DR programmes or markets can bring about multiple benefits to programme sponsors and customers, which support the long-term development of DR resources.

§ Tapping into small DR resources. The potential for delivering DR is not limited to large C&I customers. In fact, as DR becomes more important in system operation and planning, it is necessary to capture the demand flexibility of smaller customers. Since many DR programmes and wholesale markets require a minimum capacity of DR resources (e.g. for administrative reasons), aggregators can ‘pool’ these resources from smaller customers and offer them into DR programmes or markets. This allows smaller customers to access DR revenues that would otherwise be unobtainable.

§ Customer engagement and market development. The success of DR programmes depends on the ability of programme sponsors (e.g. system operators or utilities) to procure adequate resources. This is a significant commercial undertaking, and in restructured markets, it is likely that some programme sponsors like system or network operators may not have the means for engaging customers. As aggregators typically have dedicated marketing and customer teams, they will be able to reach a wide customer base, and deliver economies of scale and cost-effective procurement. Moreover, it should also be in the interest of aggregators to allocate resources to identifying the DR potentials of various customer categories and helping individual customers with diagnostics and customised DR plans, thus adding value to the consumers’ and programme sponsors. There is evidence that CSPs have been valuable in recruiting new DR resources through their marketing and service offerings (Cappers et al., 2010). This may be partly due to the incentive for the aggregator to tap into the DR market as much as possible, while traditional utility programme sponsors often face ‘conflicting financial incentives’ (Hurley et al., 2013).

§ Performance risk management. Like other resources, including even traditional supply-side resources, DR resources may not always deliver fully their demand reduction commitments. This creates a performance risk for system operators or utilities that rely on such resources. Since aggregators typically have a portfolio of resources to call upon, they have an important role in managing commercial contracts with individual DR providers and promoting the reliability of DR resources (at least from the perspective of DR programme sponsors). In essence, this means the transfer of performance risk from programme sponsors to aggregators, which arguably should have better knowledge of market and customer needs.

§ Scope for innovation. Since the contract between customers and aggregators is separate from that with the DR programme sponsor, there may even be scope for innovation in how the agreement is constructed and how DR resources are called (e.g. delivery of contracted DR to the programme sponsor by dispatching a large number of customers on the rolling basis).

§ Incentive to maximise DR procurement. More importantly, unlike regulated utilities, aggregators should have the ‘natural’ incentive to develop the DR market, by offering innovative products to meet the needs of customers and necessary support services, and to increase the procurement of DR resources as long as there is a market for them.



Supporting aggregation

To support the aggregation business model, a number of provisions in regulatory frameworks and market conditions are very important.

§ Access to customers and wholesale markets. First and foremost, aggregators should be allowed to sign commercial agreements with individual DR providers and to package and sell the DR resources to parties needing them (e.g. utilities or independent system operators). This could include providing aggregated DR resources to utilities through bilateral or other arrangements (e.g. signing up to standard programme offerings, or participation in dedicated DR auctions), or eligibility for bidding into the wholesale markets where DR resources are allowed to participate.

§ Market size and opportunities. On one hand, DR market size must be large enough to justify the aggregation business model; on the other, there should be recognition from DR programme sponsors of the value of aggregation. With higher targets for procuring DR and other demand-side resources, utilities may see the benefits in ‘outsourcing’ to CSPs that can deliver targets in a ‘cheaper, faster and better’ way (Cappers et al., 2010). Aggregators may make it easier for the DR programme sponsors to access particular customer segments, provide customised solutions to fit specific needs, or offer expertise in programme design and operation. As DR resources become more important in system planning, the reliability of committed load reduction or other response that aggregators can provide may become more attractive.

§ Financial viability of aggregation business model. The aggregation business model relies on adequate revenues in offering resources to DR programmes or markets, and/or other services to system operators or utilities (e.g. administering DR programmes, customer marketing and development) and customers (e.g. analytics and DR plan development, other energy management services).

• Prospects for earning reasonable profits. Aggregator revenues should recover the payment to customers providing DR, and justify the services provided to them and the costs in marketing and procurement. For this reason, it is perhaps not surprising if aggregation business focuses on certain DR markets. For instance, the CSPs in the US have concentrated on opportunities in capacity and other programmes that can provide up-front or on-going payment for the committed demand reduction (Cappers et al., 2010). In other words, DR programme design will also influence the revenues and financial position of aggregation business models.

• Other sources of revenue. As discussed earlier, many aggregators also provide services to customers (e.g. energy efficiency, energy analytics and performance optimisation, other energy management services), which utilise their market and customer expertise to provide additional sources of revenue. This will strengthen the value proposition to customers, thus increasing likely customer engagement and driving market demand for aggregator services. It is also possible for aggregators to provide services to system operators or utilities to support their DR programmes (e.g. technologies to support DR programme implementation, performance monitoring).

2.4.4 Customer technical capability, incentive and education

For customers to engage with and participate in DR programmes in the long term, they need the technical capability to deliver DR and appropriate financial benefits to participate in DR programmes. Customer education is important, not least to address concerns they may have on the risks of participation.

§ Technical potential for different DR services. Customers should have an electricity-use pattern suitable for DR programmes. There should be end-uses that are technically flexible and that customers are willing to shift or reduce, especially during the time windows most valuable to an electricity system (e.g. system balancing, peak periods for distribution or transmission networks). Potential end-use loads should also meet the parameters of specific DR programmes (e.g. response time, duration of DR event, frequency of calls), and consumers must be willing to accept these parameters.

• Assessment of DR potential is important. Customers should have resources to assess the feasibility of individual sites or end-uses for DR programmes, and formulate appropriate DR operational plans based on their specific requirements and characteristics. While many large C&I customers tend to have dedicated energy managers who will be able to help with diagnostics, assistance from third parties such as aggregators and other energy service providers can be valuable.

• Enabling technologies are valuable for DR capabilities. As discussed in Section 2.4.6, a variety of technologies can support customers in their participation in DR programmes. For example, automated


DR and energy management systems that can centrally monitor and control individual end-use loads for different sites can enable customers to respond easily to DR dispatch instructions. This may allow customers to participate in DR programmes that require rapid and frequent response, and to monitor performance in delivering load reduction. For small commercial and residential customers, the use of feedback technologies such as in-home displays allows for ‘visualisation’ of electricity demand, and the engagement of customers in DR, by allowing them to see when demand response is required/being provided, and to link any reduction in peak demand with the actions they have taken. Moreover, the availability of on-site generation can increase the (grid) demand reduction customers can deliver, provided that the benefits of providing DR services outweigh the costs of on-site generation. As seen in the Customer-Led Network Revolution DR trial for commercial and industrial customers, the availability of on-site generation provides a successful ‘entry point’ for new customers to engage in DR (MacLennan, 2014). However, using on-site generation may increase CO2 emissions, air pollution and/or other environmental impacts. Moreover, since on-site generation just replaces part of the bulk generation capacity, the potential capacity savings on the level of whole electricity system would not be as high if the peak demand reduction is achieved from load shedding or shifting. Depending on programme characteristics and market conditions, offering financial incentives for enabling technologies may be a viable option to promote their market penetration and strengthen the business case and technical ability of customers to participate in DR.

• Diversifying DR programme offerings. As the characteristics of electricity use vary greatly amongst customers, they may have varying needs and abilities for providing DR services. Meanwhile, as noted earlier, the parameters of DR services vary depending on how they are used to meet different system needs (e.g. capacity resource or ancillary service). For these reasons, it is desirable for the DR portfolio in the long term to offer diverse DR programmes to cater to the heterogeneous customer base (as in the US and the UK), so as to capture greater DR potential. However, the available programme types must not overwhelm customers, especially when their knowledge and experience with DR are limited. As customers develop some experience of delivering DR services and gain more confidence in the value they can get, it should be easier to introduce more sophisticated or complex products and programmes.

Financial benefits for customers in delivering DR services. For the long-term viability of DR programmes, customers must be appropriately motivated, especially financially, to commit to providing DR services on an enduring basis. One implication is that there needs to be a reasonably large demand, from the system operator or wholesale market, for DR resources so as to support the DR market.

• Customer benefits need to be greater than costs. For example, financial compensation for providing DR services should be sufficient to allow customers to recover the capital investment needed for participation (e.g. energy control or management system, operational cost for delivering DR), and to incentivise them to provide DR (e.g. significant drop in electricity expenses), even after any transaction costs involved in participation (e.g. DR plan design for individual sites, compliance with measurement and verification (M&V) requirements). Understanding how benefits and costs play out for individual customers is very important in helping them find a business case for participating in a specific DR programme. This may be different from the cost-benefit analysis from a system perspective, and could require deep knowledge of the needs and capabilities of different customer types.

• Programme payment should reflect the value of DR. In terms of compensation level, DR resources should not be discriminated against when compared to other resources that provide equivalent system services. This requires a means of estimating the value of system services provided by DR (e.g. avoided generation capacity or network investment, or ancillary services). Moreover, to the extent that DR resources provide different system services, the compensation level should differentiate to reflect the value of these services (e.g. three different Emergency DR programmes in the forward capacity market of PJM). This is important for the long-term engagement of customers and viability of DR programmes.

• Programme design matters. First, as discussed earlier, there are many different ways of making payment (e.g. availability and utilisation payment, or capacity payment for delivering load-following DR services). While the design of payment method depends on the specific programme, it should give customers the confidence that they will be able to gain meaningful financial compensation for their participation, and it should also be transparent and easy to understand. As shown in ISO-NE, FCM and PJM Economic DR programmes, the level of compensation for DR resources can be an important factor influencing participation (Hurley et al., 2013). In the PJM system, demand-side services earned approximately $700 million in 2013, mostly in the capacity market. As with other resources, DR



providers are more interested if payment is reliable and adequate. Another consideration is the length of agreement for participating in DR programmes is a consideration. Longer agreements and more predictable revenues will give customers much-needed certainty and attract higher participation (e.g. Hurley et al., 2013). However, these should be considered in combination with the potential value of the DR services provided to the programme sponsor or the system.

• Electricity retail pricing should reflect short-term system marginal costs. Another well-known barrier to consumer engagement has to do with retail pricing. As most customers do not face retail prices that reflect the real-time costs of generation and transmission, there is inadequate incentive for them to engage in DR services, especially if retail rates are very low. However, changes to the retail pricing structure alone may not be sufficient to ‘kick start’ DR development, and must be combined with the technical conditions and additional incentives discussed earlier. Even then, pricing and technology alone may not have much impact on household-level demand response: for that, a carefully-thought-out consumer engagement programme is usually required, in which customers are offered clear, useful feedback on their consumption and guidance about how they can contribute to demand response, and why it matters (VaasaETT, 2010, 2011).

Customer education and engagement. Since DR is a new concept to many customers, besides technical capability and financial incentive, sustained customer education and engagement are vital for the success of DR programmes and thus the long-term development of DR market. As the customer needs may differ depending on their experience with DR, the concept of ‘customer journey’ is valuable in helping identify priorities for customer engagement activities (VaasaETT, 2011). Especially at early stages of the ‘customer journey’, education should address concerns about any risks of participating in DR programmes. For instance, the Low Carbon London C&I DR trial in the UK found that perceptions of the potential risks to service level, comfort, and equipment were greater barriers than technical and financial factors for the C&I DR (UKPN, 2014). For this reason, it is worthwhile to organise dedicated customer education, based on specific customer characteristics, and to disseminate any well-validated ‘success stories’ to promote confidence.

• Sustained customer engagement may provide opportunities for further developing DR capability. For those customers who have gained some experience with DR and a better understanding of the benefits from participation, sustained engagement can be helpful in ‘nudging’ customers to develop their technical or operational capability for DR even further. As the DR market develops and the programme portfolio diversifies, utilities and other programme sponsors should also be able to build upon prior positive customer experience to increase their interest in new programmes. Moreover, regular communication can allow customer problems to be identified, which informs the improvement of programme design and implementation.

• Energy feedback has an enabling role to play. Besides dedicated education and outreach activities, feedback technologies, by which customers can monitor their electricity demand and delivery of DR, can act as a lever for engaging customers. Evaluations of smart metering and residential DR programmes65 show how consumer responses can vary greatly, depending on the amount of ‘shiftable’ load (especially thermal loads), pricing structures, and the extent to which customers can see and understand their consumption (Darby and McKenna, 2012). They also show the value of customer information displays66 in conjunction with smart meters: these typically increase the response to time-of-use pricing (e.g. Faruqui et al., 2010; Jessoe and Rapson, 2012).

2.4.5 Reliability of DR resources

Industry stakeholders, particularly electricity utilities and system planning authorities, need to have confidence in the reliability of DR resources, with specific requirements depending on the service characteristics (e.g. frequency response or as a capacity resource). This is essential for treating them as alternatives to supply-side solutions. Partly due to the insufficient historical evidence of DR programme impacts, especially those on a full deployment scale, there are some concerns about whether DR resources can be expected to deliver reliably (FERC, 2009). While specific programme designs (e.g. penalty for under-performance in delivering DR commitment) can offer incentives for reliability, technologies that promote end-use controllability also have an enabling role. Moreover, it is equally important for programme sponsors to undertake regular evaluation, measurement and verification (EM&V) to strengthen the evidence for the reliability of DR resources. 65 Note that many of these programmes do not require aggregators: they are simply arrangements between the customer and the retailer or network operator. 66 These take many forms, from ‘orbs’ that change colour according to the unit price at any time, to digital displays that show consumption in real time and over previous periods.


While some factors related to a DR programme (e.g. response speed, duration and frequency, unscheduled unavailability of on-site generation) can affect the reliability of individual DR providers, from the perspective of programme sponsors who are interested in the aggregate impacts, it is perhaps more important to strengthen the evidence basis for aggregate programme-level reliability performance. This is especially valuable for non-dispatchable DR programmes, since they do not typically involve incentive for reliability. There are some industry efforts to improve the evidence base for the reliability of DR resources. For instance, the North American Electric Reliability Corporation has established the Demand Response Availability Data System (DADS), which specifies data collection requirements for measuring the performance of DR resources. There is now substantial evaluation evidence for energy efficiency programmes in North America and Europe, showing that energy savings are achieved. Historically, the delivery has been less than ex ante estimates, but this is improving as evaluation evidence matures (Wade and Eyre, 2015).

For DR, evaluation evidence is more recent, but has been subject to greater early scrutiny, where used in capacity markets. Cappers et al. (2010) assessed the reliability of DR resources in NYISO and ISO-NE, by comparing actual load reduction to the programme subscribed amount. The reliability level of DR resources in capacity programmes with non-compliance penalties of NYISO and ISO-NE (2001-2007) was 64% and 77% respectively, while the voluntary emergency DR programme in NYISO had 52% reliability. The reliability level of economic DR programme in ISO-NE ranged between 9% and 53%, with an average of 32%, which can be explained by reasons such as customers’ lack of experience with the programme and absence of non-performance penalties.

DR resources in capacity programmes also exhibit less variance in reliability than in voluntary or economic programmes. Hurley et al. (2013) found a very high reliability level (90-100%) for the DR resources in the forward capacity markets in PJM and ISO-NE, and in the emergency interruptible load service of ERCOT. The North American Electric Reliability Corporation (NERC) has set up the Demand Response Availability Data System Working Group for reporting and measuring the performance and other issues regarding DR resources. In its first report that assessed the DR reliability based on the data collected from 126 entities in North America, it was found that the DR resources achieved high reliability of 87% and 79% for the 2011 summer and 2011-2012 winter respectively (NERC, 2013). Future efforts will be made to analyse how the reliability performance would differ between seasons, time, sectors, service type and other parameters. Regarding the STOR market in the UK, the availability of DR resources is 67-85% (UKPN, 2014). In the Low Carbon London C&I DR trial, embedded generation and real load reduction achieved the reliability of 78% and 62% respectively (UKPN, 2014). To keep these in perspective, the equivalent availability factor (EAF)67 of generation capacities in the PJM RPM is 82.3% in 2014, down from slightly less than 88% in 2007 (Figure 12).

One observation from these findings is that formal non-compliance sanctions/performance rewards may be important design features to promote the reliability of DR resources (e.g. capacity programmes with contracts for

67 Fraction of a given operating period in which overall generating units are available without any forced, maintenance and planned outages

Figure 12 Generator Equivalent Availability Factor (EAF) in PJM RFM from 2007 to 2014

Source: Monitoring Analytics (2015)



performance, as opposed to voluntary programmes). Other considerations such as the experience and expertise of customers in identifying, delivering and measuring DR services, and the economic value of participation, may also influence customers’ reactions to dispatch calls (Cappers et al., 2010). This highlights the importance of customer engagement, especially for programme performance in the longer term.

2.4.6 Enabling technologies

Advanced Metering Infrastructure (AMI) is one essential technology for the implementation of DR programmes on a wider scale, especially for price-based programmes, largely due to its ability to enable interval metering, frequent information communication and remote management/control. In the UK, the government has established a programme to roll out smart metering to households and small commercial customers by 2020. Consideration of the need for DR implementation is integrated into the rollout programme and the technical specification. Considerable progress in rolling out smart meters has now been made in the US (e.g. FERC, 2014) and in countries such as Italy, Sweden, Finland and Spain68.

Energy management and control technologies (e.g. smart controls and gateways, energy monitoring or management systems or process control systems for C&I customers) can be enablers for DR programme, in particular those that require quick response and automation. Moreover, granular data about real-time demand for various end uses may also be useful in increasing the ‘visibility’ of the impacts of DR actions, and in identifying the potential to provide different DR programmes. For customers to invest in these enabling technologies, the benefit-cost case for doing so must be strong. The benefits could be improved opportunities to manage or optimise business activities (e.g. energy management) or the revenue prospects for participating in DR programmes, while the technologies themselves should also be cost-effective. As proposed in FERC (2009), a number of options exist to increase market penetration of these technologies, including marketing partnerships between equipment manufacturers and retailers/installation contractors, and the approaches used in energy efficiency programmes (e.g. financial incentives for particular technologies, educational and awareness campaigns).

A number of considerations pertain to the technical aspects of enabling technologies. With technological development, appropriate measures must be put in place to reduce the risk of assets becoming obsolete and ‘stranded’, which may undermine cost recovery and customer interest in making the investment. Secondly, there is value in employing technical protocols and standards that are based on interoperability rules and open standards, so as to promote product or service innovation and competition. For consumer appliances that are likely to be important in delivering DR programmes (e.g. refrigeration, water heating and HVAC equipment), product standards will need to include provision of the necessary switching (e.g. automatic and/or remote), as well as the usual minimum standards for energy efficiency.

2.5 Concluding remarks and implications for China

2.5.1 International experience

This section reviews the background and experience of DR used to provide different system services in the US and the UK, research efforts to assess the potential and benefits of DR resources, and policies and regulatory considerations central to the long-term development of DR market. Key points are that:

§ Technical and policy developments drive growing worldwide interest in DR. Factors including the penetration of intermittent renewables and electric vehicles, increasing availability of AMI, environmental concerns and economic efficiency objectives highlight the benefits DR can bring to system operation and resource planning. These considerations are highly relevant to China, especially against the backdrop of low-carbon energy transformation, economic rebalancing and ongoing power industry reform. For this reason, it is vital that policymakers and industry stakeholders seriously consider the value of DR and make concrete actions to promote it where this is economically efficient.

§ DR from different sectors and end-uses can provide a variety of system services. International experience shows the feasibility of DR resources to contribute to short-term system balancing and long-term resource adequacy. They may also improve the economic efficiency of wholesale electricity markets. Opportunities for DR can be found in certain sectors, customer segments and end-use categories, which

68 See http://www.escansa.es/usmartconsumer/documentos/USmartConsumer_Landscape_2014_Final_pr.pdf


may be better suited than others to provide specific DR services. For the moment, the most important value of DR seems to relate to the deferment or avoidance of new capacity and overall resource savings.

§ The nature and importance of DR largely depends on the specific characteristics of electricity systems, including the drivers of system peak demand, integration of intermittent renewables and network constraints. These have an important bearing on the design and implementation of DR programme and supporting policies. For instance, they may determine end-uses and customer segments to target, incentives, system services to be provided, requirements regarding dispatch, speed, duration and frequency of response, and enabling technologies, as well as the entities to procure and utilise DR resources. Therefore, it is vital that system operators and government agencies overseeing the electricity system operation and planning in China undertake rigorous assessment of the system needs (e.g. resource adequacy, ancillary service) in the medium and long term, and whether DR offers cost-effective solutions to these system needs. This requires system operators and government agencies to improve their definition and governance of resources that can support these needs (e.g. ancillary service), especially in the context of likely changes in electricity system (e.g. increasing integration of renewables).

§ Technical and socioeconomic enablers are important for the long-term development of DR markets. To create markets for DR, utilities and system operators should face incentives for using DR as a resource and have confidence in its reliability, while customers need to have sufficient benefits and support for their long-term participation. Key aspects include the business opportunities for DR, the regulatory framework to incentivise utilities and system operators to procure DR as well as to trial innovative DR, customer technical capability, incentive and education, support for aggregation businesses that can promote customer participation and commercial innovation, confidence in the reliability of DR and availability of enabling technologies. To support the market for DR, it is essential that industry stakeholders should see the value of using it as a resource. Moreover, programme sponsors should focus on removing the barriers to customer participation in DR programmes, and designing programmes that can deliver adequate value to customers.

§ Experience has been gained internationally in estimating DR potential and quantifying its value. As the essence of DR is about changing how customers use electricity, its potential largely depends on the patterns of electricity use across different customer segments during particular times (e.g. system peak period). With the growing availability of interval metering data (especially at the end-use level) and evidence about the performance of DR programmes, it is possible to undertake detailed DR potential assessment. There is also some valuable international experience in calculating the benefits of DR (e.g. avoided capacity costs, avoided energy and ancillary service costs, and avoided T&D costs), and in designing regulatory and market mechanisms to encourage DR. This exercise can assist utilities and government planning authorities in China in making informed decisions regarding DR development. Given the heterogeneity of regional or provincial electricity systems in China, it is important that such assessments be carried out for specific individual systems and updated periodically to make contributions to DR policy-making.

2.5.2 Implications for China

While the drivers for promoting DR and the above lessons for developing the DR market are relevant to China, the experience from OECD countries may not be directly transferable, essentially because the Chinese power system has developed very differently from systems in OECD countries. This difference highlights the need to design DR programmes to reflect the specific regulatory context in China, as well as the need to consider changes in the practices and institutional framework governing electricity system operation. It also implies several practical considerations in estimating the potential and value of DR resources. Here a few key points are summarised.

§ Administrative demand planning not compatible with market-based DR. While originating from the 1970s when electricity supply was short, the practice of annual demand planning, whereby provincial-level planning departments allocate supply to their jurisdictions and oversee load management and rationing programmes69, still exists now. This, in principle, represents a divergent approach from market-based DR where customers respond to incentives to change their electricity use. There are a number of limitations to demand planning. First, administrative measures for load management can cause significant economic losses. Secondly, demand planning does not give customers adequate incentives for voluntarily changing their electricity demand. Thirdly, demand planning is limited in its scope for developing DR. For example, the administrative load management programmes have focused on large industrial customers, thus leaving the potentials in other customer segments (e.g. residential and commercial) almost untapped. Moreover, the

69 Load management includes load shifting and load avoidance, while load rationing refers to restricted use and load curtailment.



programmes are mainly designed for addressing system emergency of resource inadequacy, and may not be flexible enough to support other system services (e.g. ancillary services for promoting the integration of renewables). If DR is to play a more important role in the system operation and planning, it is necessary to explore the potential of different sectors to provide various DR services. Fourthly, administrative demand planning and its effect of demand suppression can undermine the value of market-based DR. For these reasons, it is important to make the transition from administrative approach to a model based on market mechanism. However, this requires not only regulatory changes, but also well-designed DR programmes to reflect the needs and characteristics of customer segments, and to provide necessary support.

§ Rigid institutional and governance framework for resource planning and dispatch can create constraints for DR. As discussed further in Kahrl and Wang (2014), annual generator output plans and unit commitment plans, together with grid operating plans, form the basis for generation dispatch plan70. Annual generator plans typically require the maintenance of certain hours of operation for generation capacities. Together with the fixed schedule for interregional and interprovincial power exchange71, this dispatch model risks constraining the scope for using DR resources, to the extent that the deployment of DR can have significant impacts on the operation hours of generation capacities. Moreover, the multi-level hierarchy of dispatch model may also introduce complexity in the potential sharing of DR resources, as well as the costs and benefits amongst provinces.

§ Inadequate drive for economic efficiency in resource planning and dispatch may disadvantage DR. Unlike the UK and the US, the incentive for economic efficiency is largely absent in the electricity system of China. For example, the lack of optimised economic dispatch (e.g. day-ahead and operating reserves) across all generation types (e.g. coal, gas, hydro and renewables), or even the ad hoc approach for dispatch in some cases, means that less economically efficient units may be running at the expense of more efficient ones. The implication for DR is two-fold: the potential benefits of cost-effective DR may not be fully realised in the existing model of system dispatch and operation, and there is little incentive for considering DR as an alternative resource in electricity system planning.

§ Lack of pricing signal in electricity system operation makes it difficult to assess DR value. It is often possible to value DR resources in the UK and some US states by reference to what they (or an alternative resource) can earn in competitive wholesale markets for capacity, energy and ancillary services, financial transmission rights and CO2 emission permits. For systems with competitive market mechanisms, DR resources may compete directly against supply-side and other demand-side resources, or be procured by a competitive mechanism to determine the price for specific DR services. For systems without competitive market mechanism, there is usually some economic information that allows the administrative price for DR to reflect an estimate of avoided costs. By contrast, the electricity system in China is still mostly subject to central planning, with prices, operating hours of power plants and peak demand pre-defined. Since the competitive market mechanism does not exist to determine prices, DR forms part of the planning process rather than being driven by price signals. The inadequate pricing signal makes it difficult to assess the full economic benefits of DR. Moreover, as customers do not face marginal prices that reflect system conditions and economic costs, they may see less benefit than otherwise from participating in DR, and have limited incentives.

Even with these caveats, the electricity system in China is changing and DR may well become more valuable resources. Economic rebalancing towards a service-based economy will drive the growth in residential and commercial electricity demand, suggesting the need and opportunity of tapping into the DR potentials in these sectors. Similarly with other countries, the integration of renewables and growing penetration of EVs underline more important roles DR can play in the system operation.

More importantly, on the regulatory level, there are a series of new developments that should pave the way to support DR in the future. For example, in the Opinions on Further Deepening the Power System Reform as issued in 2015, the State Council has not only heightened the importance of DR and other demand-side solutions in ensuring the supply-demand balance, but also highlighted its objectives for electricity pricing reform and introducing market-based mechanisms. Further development in these areas will contribute to the strengthening of pricing signals in system operation, and the flexibility of resource planning and dispatch. Moreover, there is also an intention to cut back on the practice of administrative demand planning in an orderly manner, and to allow voluntary interruptible contracts between customers and utilities. In principle, these efforts should create a more

70 Annual generator output plans are drawn by provincial-level planning agencies to guarantee hours of operation for generators. For provinces using energy efficiency dispatch, unit commitment plan is made based on the dispatch order table mandating the order for dispatching generation. Grid operating plan incorporates transmission system security considerations and constraints. 71 Interregional and interprovincial power exchange schedule is fixed before provincial dispatch organisation makes the dispatch plan.


enabling environment for DR. NDRC also issued a notice in 2015 to the four DSM pilot cities in China (Beijing, Suzhou, Tangshan and Foshan) to require the implementation of market-based DR pilots, based on the international experience and that of Shanghai DR pilot. Therefore, there is a strong case for an experiment in the area of DR, which can ensure that demand does not impose unnecessary strains or impose undue cost in the future.

A ‘phased’ approach for developing the DR market looks appropriate, given the regulatory conditions of electricity system in China. At the early stage, DR programmes may consider simple designs (e.g. curtailable programmes for resource adequacy to targeted customer groups), even on a pilot basis, to promote customer interest and understanding of this market-based approach, which is different from the administrative demand planning. Table 9 lists several types of DR, particular TOU, CPP, DLC and curtailable programmes, which can be implemented under the current structure of electricity industry in China. Once customers gain more knowledge of participating in DR programmes, opportunities may arise for introducing more DR programmes (e.g. with more differentiated response requirements and payment levels) to fit with the diverse customer characteristics and different system needs.

From the viewpoint of the ‘customer journey’, this approach has value in enhancing customer engagement and learning, and sustained customer education and assistance should support it. Utilities and aggregators have very important roles to play in this. Moreover, it is important to conduct regular programme evaluation to develop the evidence of DR reliability, to share ‘best practices’ and to understand where improvement is needed. Meanwhile, on the regulatory level, it is advisable to leverage the opportunities of ongoing electricity market reform to ‘familiarise’ customers with ‘market characteristics’ of electricity prices. Examples can include the introduction of time-of-use (TOU) tariffs or critical peak pricing (CPP) to increase the potential benefits for delivering DR during high-priced periods. Regulators and utilities should also improve the accounting system and system service definitions to enable accurate assessment of the value of DR to system operation. This should complement the regulatory incentives for more cost-effective resource procurement and strong engagement of utilities in DR.

60 Assessment of Demand Response Potential and Benefits in Shanghai DR Market Potential and Valuation in Shanghai

3. DR MARKET POTENTIAL AND VALUATION IN SHANGHAI

3.1 Introduction

Estimating the potential and benefits of DR resources is an important part of efforts to promote the utilisation of such non-supply-side resources to provide system services. The analysis can help utilities, system operators, government and other key stakeholders in identifying the scale and source of DR resources that can be expected, and their potential values to the electricity system. It can inform the decision-making process of setting targets and strategies for DR development (e.g. programmes to offer and customer segments to target) .

This section aims to assess the potential and value of DR resources in Shanghai. It summarises the methodology and key considerations in conducting analyses, and provides an estimate of the potential and benefits of DR resources in Shanghai. The assessment should be seen as indicating a scale of DR potential and its value, based on the availability of locally specific data and information (e.g. load profiles for different customer segments or end-uses, factors influencing the participation in DR programmes). It is our intention to demonstrate the process of conducting the potential and valuation analysis so as to support relevant stakeholders in undertaking similar exercises in the future, should more locally specific information and data become available. This section first introduces the framework and methodology for assessing the potential and benefits of DR, by reviewing the existing literature and international DR potential studies. It then presents and discusses the results of assessment drawing upon the evidence in international DR programmes and the pilot project in Shanghai as well as valuable local expert knowledge.

3.2 Framework and Methodology

3.2.1 Assessment of DR potential

Defining the DR potential

International literature of DR potential studies has used a few concepts to distinguish the levels of potential that can be expected when specific factors are considered:

§ Technical potential. For specific DR programme or overall market, this term essentially refers to the level of DR potential, if all electricity end-uses regarded as suitable for providing DR for all eligible customers are procured. In other words, it does not consider any economic or other factors (e.g. cost-effectiveness of DR, marketing and customer engagement, regulatory and market conditions, characteristics of electricity use for individual businesses) that may affect the extent to which customers are willing to participate in DR programmes or be flexible with their electricity demand.

§ Market potential. Different from technical potential, market potential (or ‘achievable potential’ as it is known in some studies) accounts for practical considerations (e.g. programme design/incentive, economic considerations for participating in DR programmes, customer engagement, characteristics of electricity use, regulatory and market conditions) that can affect the participation of customers in providing DR resources and/or the level of response in reducing the electricity demand. As discussed later, the assessment of market potential is subject to assumptions on important parameters, which can be made based on international and local evidence as well as dedicated modelling work. This study focuses on the market potential of DR.

General framework and methodology

The international studies for DR market potential are heterogeneous in their specific analytical approach. However, they share some common features in the overall framework for conducting the potential analysis, one of which is the ‘bottom-up’ approach is assessing the DR potential. As noted in FERC (2009), there are strong

Assessment of Demand Response Potential and Benefits in Shanghai DR Market Potential and Valuation in Shanghai 61

reasons to support the use of such ‘bottom-up’ approaches. For example, the potential of various customers to reduce demand and thus deliver DR largely depends on the characteristics of their electricity use during system peak periods, which are likely to vary across different customer segments. The existence of specific end-uses (e.g. AC or other electric equipment with substantial load impacts) and the ownership of enabling technologies (e.g. back-up generation, and energy control and management system) can influence the amount of DR individual customers can potentially deliver. Moreover, customer segments may have different capabilities and willingness to participate in DR programmes, given their specific participation model, requirements for response to DR calls, financial compensation and other programme design characteristics.

Figure 13 shows the general framework and key steps for the ‘bottom-up’ approach in estimating the DR market potential. Serving as guidance, the following section describes these key steps in detail and some examples of analytical techniques as well as their strengths and weaknesses.

Figure 13 Key steps in analysing the DR market potential

Step 1: Determine scope of analysis

The analytical scope should be determined at the outset of DR potential study, following these two dimensions:

§ Type of DR programmes. The potential study should specify the type of DR programmes to be considered (e.g. price- or incentive-based) or even the system services to be provided by DR resources (e.g. capacity, ancillary or energy resources). This is largely because the type and design of DR programmes may affect the participation of customers and/or their level of response, thus making it necessary to distinguish various DR types. A few considerations may inform the decision on what DR types to consider:

• Interest of system operators and policymakers – e.g. the system services which DR is expected to provide (e.g. DR as capacity or ancillary resources), and when DR resources are likely to be needed (e.g. weekday and/or weekend, summer and/or winter, depending on the specific characteristics of electricity system);

• Specific market structure and conditions – e.g. the existence of wholesale electricity market or retail choice, whether there is a way to reflect the marginal cost in the electricity system, and feasibility of introducing pricing-based DR programmes;

• Availability of enabling infrastructure – e.g. whether advanced metering infrastructure (AMI) or smart metering, which is necessary alongside utility data management/billing system for pricing-based DR programmes, is available or planned for rollout.

§ Target customer population. This involves segmenting the customer population based on their sectors (e.g. residential, commercial or industrial) and/or level of electricity demand (e.g. peak electricity demand). As a given DR type (e.g. real-time pricing) may be more appropriate for certain customer segment (e.g. large C&I customer), defining target customer population is valuable in informing the choice of DR types and focusing the research effort. Factors influencing the decision to target specific customer population may include the contribution of customer segments to the system peak demand, the technical capability of customers in delivering DR (e.g. whether business or industrial activities are flexible enough to provide DR services), market and regulatory conditions and the interest of system operators and policymakers (e.g. whether or not to offer specific DR programmes to a given customer segment).

• For programmes intended to target specific end-uses (e.g. DLC for AC programme or water heating), market penetration of these end-uses is an important factor in determining who will be eligible to participate in the DR programme.

• For pricing-based DR programmes, the progress in deploying AMI or smart metering should be considered in identifying the customers eligible to participate in pricing-based DR programmes.

Determine scope of analysis

Segment customers

Develop average per-

customer load profile

Estimate participation

rate

Estimate average load

impact DR market potential


Step 2: Segment customers

The target customer population may need to be segmented, to the extent that customers have varying degrees of suitability for particular DR programmes or can be expected to respond in different ways to DR. In other words, different customer segments may vary in their willingness to participate in a given DR programme or in their level of response in reducing electricity demand. If the potential analysis is interested in specific DR programme designs, it is necessary to assess how the programme requirements (e.g. response frequency and duration, trigger conditions for DR, compensation and penalty, technical requirements) and characteristics (e.g. ‘transaction costs’ for participation, contractual arrangements) may influence different customer segments regarding their participation rate and potential capability of load reduction. Moreover, customer segmentation may also be necessary for practical considerations (e.g. programme strategies and marketing). As noted in Goldman et al. (2007), since customer segments are considered separately in the market potential analysis, it is important that the segmentation process ‘captures significant trends’ influencing the participation of customers in DR and their level of response. At the same time, practical considerations (e.g. availability of detailed data for developing the average load profile for each customer segment) should also inform the decisions on segmenting customers. In the international DR potential studies, common measures to segment customers include:

§ Classification of business activities (e.g. manufacturing, commercial or government), since the business activities are ‘often strongly correlated’ with the willingness of customers to participate in certain DR programmes and their level of response (e.g. Goldman et al., 2005). Existing customer classification (e.g. standard industrial classification) can be used to group customers based on their key businesses.

§ Size of peak electricity demand. Based on their level of peak demand, customers can be segmented into small, medium or large customers. A number of international DR potential studies have used this indicator to reflect the different characteristics of electricity use across customers (e.g. FERC, 2009). Depending on the availability of data, peak electricity demand of customers can be estimated by analysing interval meter data or applying assumed load factors72 to the average electricity demand73 per customer for a given rate class.

Step 3: Develop average per-customer load profile

For each delineated customer segment, the average load profile (i.e. level of electricity demand) per customer without the load response under DR programmes needs to be identified. At the minimum, the average load per customer for the time periods when the DR resources are likely to be used (e.g. afternoons of summer weekdays or early evenings of winter weekdays, depending on the electricity system characteristics) should be estimated. Subject to the requirement of the DR potential study and data availability, average load profile can be depicted as average hourly load for each hour of the time period of interest or average load across the entire time period of interest. As an example, FERC (2009) took the average of per-customer hourly electricity load across 2pm-6pm of days with the 15 highest system peak demands (i.e. typical DR event days), and used it as the average per-customer load during the time period of interest for estimating the DR potential.

Ideally, the average per-customer load profile can be constructed from metering studies of a representative sample of customers in each customer segment. The sample should be large enough and often stratified (e.g. based on indicators including level of electricity consumption, sector and key business activities) to reflect the diverse electricity use patterns of customers. Customers should be randomly sampled and metered for a reasonable length of time to capture the temporal difference in load pattern (e.g. weekly, monthly or seasonal). Detailed guidance for developing customer load profiles is readily accessible from various organisations such as the wholesale market settlement body74. Peak load factor (PLF) can be calculated as the ratio of average annual demand (i.e. annual electricity in kWh as averaged over 8,760 hours in a year) and the estimated peak demand. Alternatively, load profile coefficients may also be developed to represent the ratio between the instantaneous electricity demand for a given hour and the annual electricity consumption. As the data for total electricity consumption (in kWh) is more readily available, the PLF or load profile coefficient can be applied to the average per-customer annual electricity use to estimate the average per-customer load profile.

However, as hourly load data from a dedicated metering study for a representative customer sample may not be available, international DR potential studies have relied on other techniques, including small customer sample (e.g. Element Energy, 2012) and regression analysis based on existing load profile information (e.g. FERC, 2009), to estimate the average per-customer load profile for the time period of interest. While these techniques

72 Based on load shape analysis 73 Annual electricity consumption averaged over all hours in a year 74 Detailed guidance for doing load profile research can be found at Elexon (https://www.elexon.co.uk/reference/technical-operations/profiling/) in the UK and many other international utilities and/or system operators.


make the potential study possible, this may inevitably introduce uncertainty in the analysis, and highlights the importance of load research.

Step 4: Estimate participation rate by customer segment and programme

For a given DR programme, the estimated participation rate refers to the percentage of eligible customers that can be expected to take part. It reflects the level of willingness that can be expected of different customer segments in providing DR services. This recognises the practical consideration that not all customers eligible for particular DR programmes will end up taking part75. There are several aspects worth bearing in mind when estimating the participation rate.

§ Factors likely to influence the participation rate of DR programmes. Considering these factors means estimating the participation rate from the perspectives of customers (e.g. cost-benefit case for taking part), and paying attention to how this may differ between customer segments:

• The design of programme is a very important determinant, which includes the process of participation, requirements of response (e.g. frequency and duration), financial incentive levels and rules for non-compliance (e.g. under-delivery of DR commitment);

• The analysis may consider the penetration of necessary enabling technologies for DR (e.g. energy management or automated DR system for more frequent and quicker DR) and their cost-effectiveness, and the opportunity and cost-benefit of fuel switch (subject to environmental regulation);

• It may be necessary to consider whether any factors external to the programme are relevant (e.g. electricity retail price, regulatory conditions, customers’ technical expertise for participating in specific DR programmes);

• For programmes expected to undergo significant changes, either within themselves or externally, it is necessary to consider how these changes may affect the factors influencing the participation rates of different customer segments.

§ Ramping-up of new DR programmes. Between the inception of DR programmes and when they become more ‘mature and established’, the participation rate may gradually increase as the customer population gets familiar with programme rules, and shares the technical ‘know-how’ in delivering DR services. However, it is important to note the possibility of participation rates fluctuating during the ramping-up period, due to factors internal or external to DR programmes.

§ Potential interactions between different DR programmes. There are at least two dimensions in considering the relationship of different programmes:

• Eligibility of customers to participate in multiple DR programmes. It may be necessary to establish an ‘order for participation’ (i.e. which programme will be offered to customers first and, if declined by the customer, which programme will be offered next, should the two be mutually exclusive), particularly when the sum of estimated participation rates for different DR programmes is greater than one for a given customer segment.

• Whether simultaneous participation in multiple DR programmes influences the participation rate. If participating in different DR programmes simultaneously is allowed, analysis is needed to determine whether this will influence the participation rate for each of the programmes individually. When such influence is deemed to be significant, this should be taken into account in estimating the participation rate.

Several analytical techniques can be used to estimate participation rates (Table 10). These techniques tend to make different assumptions about factors that could influence participation. For example, benchmarking implicitly assumes that the evidence observed for other programmes or even jurisdictions is applicable to the system of interest. Given the likely differences in programme design and regulatory and market conditions (e.g. whether customers face marginal pricing signals, electricity prices, prior experience with DR, availability of enabling technology or assistance, market barriers to participation), it is advisable to adjust assumptions for participation rate based on judgement of specific conditions in the system of interest. Similarly, other techniques such as benefit threshold or choice model tend to assume economic factors play a key role in influencing customer participation, which may miss the potential impacts of non-economic factors.

75 Even in cases where particular DR programmes (e.g. TOU tariff) are offered as the default option, customers typically have the option to opt out of these programmes.


At the same time, it is important to note that these techniques require different levels of data input and research effort. For instance, an estimate based on benefit threshold (i.e. cost-benefit case for participating in specific DR programmes) requires locally specific evidence or assumptions about the level of payment customers can expect to get and the minimum cost-benefit requirements (e.g. maximum simple payback period, minimum savings in energy expenses) for different customer segments. If there is no locally-specific evidence available to show the relationship between benefit threshold and participation rate, one possible strategy is to undertake a customer survey, adjusting any assumptions based on the survey as firmer local evidence starts to emerge.

These considerations suggest that there is uncertainty in the estimation. Therefore, it is valuable to ‘triangulate’ by using a combination of different techniques (e.g. benchmarking together with expert judgement and/or customer surveys) to reduce bias or to estimate the DR potential by assuming different levels of participation rates. Given the often-limited resources for conducting DR potential studies, the choice of analytical techniques should be based on the specific needs and practical constraints for the potential study (e.g. availability of data to allow analysis specific to the market and customer population of interest).

Table 10 Examples of techniques in estimating customer participation rate

Technique Description Strength Weakness

Expert judgment or ‘Delphi’ approach

Consulting expert for assumed participation rates for different DR options or customer types

• Simple to implement

• Appropriate when it is not possible to do a detailed participation analysis

• Subjective assessment • Requires expertise with DR

programmes and target customer population

Benchmarking Applying participation rates as observed in other jurisdictions to the target customer population

• Simple to implement

• Appropriate when it is not possible to do a detailed participation analysis

• Using real participation rate data in other markets

• Assumes that market and regulatory condition, DR options, characteristics of electricity demand and customer population are similar to the target customer population

Customer survey Surveying target customer population on the likely interest in participating in DR programme

• Providing information specific to the market and target customer population of interest

• Not real participation rate for DR programme, especially when customers lack direct experience with DR

• Requires time and resources

Benefit threshold

Estimating whether the benefits to individual customers by taking part in DR programmes exceeds a certain level (e.g. minimum payback period, amount of energy expense saving), with the assumption that participation is driven by the expected benefits to individual customers in doing so


• Systemic way of estimating participation rate

• Requires high-level assumptions

• Does not account for non-economic factors influencing participation, e.g. customer engagement, technical capability of customers to participate

Choice model

Building statistic models for factors influencing the level of customer participation, based on evidence from similar markets/customer segments or from market survey

• Systemic way of estimating participation rate

• Depending on how the evidence is collected, potential issues include the relevance of evidence from similar markets or customer segments, or the hypothetical nature of market survey


Sources: Based on Goldman et al. (2007) and Faruqui et al. (2014)

Step 5: Estimate average load impact by customer segment and programme

The average load impact refers to the average per-customer reduction in electricity demand, compared to the determined baseline, during the period of interest (e.g. system peak period). It can be represented as an absolute figure of load reduction (kW) or a percentage of reduction from the determined baseline demand. In the analysis of DR potential, average load impact should be estimated separately for different customer segments and DR programmes.

International DR potential studies have used different techniques for estimating the average load impact of DR programmes (Table 11). As with estimating the participation rate, these techniques vary in their data


requirements, scope for analysis specific to the market and customer population of interest, and flexibility to account for factors likely to influence the average load impact (e.g. availability of enabling technologies and/or back-up generation, incentive level, characteristics of electricity use). While the analysis should ideally be specific to the market and customer population of interest (e.g. using existing local DR evidences or pilot outcomes), this needs to be balanced with other practical concerns including the availability of data and resources for the DR potential study.

These techniques may differ in their applicability to specific DR types. While an engineering approach is appropriate for DLC programmes, it is limited in its ability to estimate the average load impact of time-based pricing programmes, especially under different pricing differentials. For pricing-based DR, some international potential studies have estimated the price elasticity of customers, which represents the percentage change in electricity load for each percentage change in electricity prices. Then the estimated price elasticity is multiplied with the pricing differential between peak and non-peak periods to determine the percentage reduction from baseline electricity load during the period of interest (e.g. system peak). Given the limited evidence in most cases, especially that specific to the market and customer population of interest, and the uncertainty in analysis, it is valuable to use a combination of techniques (e.g. benchmarking coupled with reviewing the evidence in local DR or similar projects) or to estimate DR potential by assuming a range of average load impact levels.

Table 11 Examples of techniques in estimating average load impact

Technique Description Strength Weakness

Engineering approach

Applying average load impact as estimated from expert judgment or benchmarking to the data of electricity demand for specific end-use loads or equipment stock (e.g. AC).

• Providing information and analysis specific to the market and target customer population of interest

• Typically requiring detailed load profiles for different end-use loads during the period of interest (e.g. system peak period)

• Does not account for the likely impacts of different incentive levels, thus not suitable for price-based programmes

Customer survey

Surveying target customer population on their expected actions under hypothetical DR programmes, so as to formulate the average load impact likely to emerge


• Not real average load impact, especially when customers lack direct experience with DR programmes


Benchmarking

Applying average load impact as observed in other jurisdictions to the target customer population

• Simple to implement • Appropriate when information specific to the

target market and population is not available • Using real data of average load impact in

other markets

• Does not account for the likely impacts of difference in market and regulatory condition, DR programme design, characteristics of electricity demand and customer population

Elasticity approach

Combining the estimated price elasticity and expected pricing differential to estimate average load impact

• Using real data to estimate the relationship between price and average load impact

• Allowing for adjustment for factors specific to the market and customers of interest that may influence the average load impact (e.g. penetration of AC, climate differences, availability of enabling technologies like programmable communicating thermostats)

• Depending on the approach of choice (e.g. price elasticity of demand, elasticity of substitution or arc price elasticity), estimating price elasticity may have a high requirement of data

Source: Based on Goldman et al. (2007)

Another factor worth considering in the estimate of average load impact is the response rate of customers to the DR event calls. With the exception of DR resources that system operators can directly control (e.g. DLC), it is likely that not all customers called upon to reduce their electricity demand will be willing or able to do so (due to technical or operational issues), although this may depend on programme design and rules (e.g. penalty for non- or under-performance) or the characteristics of participating customers (e.g. operational requirements for business activities). The response rate, as estimated from benchmarking or evidence from local DR projects and other relevant experience, can be used to adjust the average load impact by customer segment and DR programme.


Step 6: Estimate DR market potential

The final step is to bring together evidence and assumptions resulting from the previous five steps, and calculate the estimated DR market potential for different customer segments and programmes. If the intention of the assessment is to consider the DR potential for different programme years in the future, assumptions should be made about the key parameters considered in the previous five steps (e.g. average per-customer load profile, size of eligible customer population, participation rates during programme ‘ramping-up’, and average load impacts) for different years. These assumptions can be made based on observed trends of parameters underpinning the DR market potential, policy objectives and expert expectations. Regardless of the exact assumptions, it is important to be transparent with each of them, in order to facilitate any adjustment should new evidence emerge to necessitate revision.

3.2.2 Assessment of DR benefits

The benefits flowing from DR are calculated by reference to the costs which the DR programmes enable the utility to avoid (known as avoided cost) – i.e. those that it would have incurred in meeting the extra demand which would have existed in the absence of the programmes. In the following section, we introduce different types of avoided costs that are considered when evaluating DR benefits in the UK and US76.

Avoided cost of new generation capacity

The avoided cost of new generation capacity is one of the most significant and most difficult to calculate. As was mentioned in Section 2.3.3, avoided capacity cost represents the majority of the total benefits of using DR, as reflected in the example of a California electric utility (See Figure 12). In principle, calculating this cost should be straightforward: avoided capacity cost is derived from the reduction in the generation capacity that would have been needed to satisfy peak demand (including a planning reserve margin77 – PRM – and taking account of the avoided incremental T&D network losses) without DR. In a simple case, the estimated DR potential (usually in MW) can be used as a reference to calculate the relevant capital costs of power plant that would provide equivalent generation capacity; the value of the demand resource is in the avoidance of these costs.

𝑨𝒗𝒐𝒊𝒅𝒆𝒅 𝒄𝒂𝒑𝒂𝒄𝒊𝒕𝒚 𝒄𝒐𝒔𝒕𝒔 = 𝑮𝒆𝒏𝒆𝒓𝒂𝒕𝒊𝒐𝒏 𝒂𝒗𝒐𝒊𝒅𝒆𝒅 𝒄𝒂𝒑𝒂𝒄𝒊𝒕𝒚 𝒗𝒂𝒍𝒖𝒆×(𝟏 + 𝑷𝑹𝑴)/(𝟏 − 𝑻&𝑫 𝒍𝒐𝒔𝒔𝒆𝒔)

However, in practice, the calculation is not straightforward. A number of possible methodologies are available, depending on such matters as the nature of the generating system, the new plant options, and the structure of the DR programme concerned. In a situation where there are fully developed wholesale electricity markets, prices in those markets can often be used as a reference point:

§ Where capacity markets are in place they can in appropriate circumstances be taken to reflect the value of capacity and be used as a benchmark in the calculation above.

§ Where there are energy-only markets that can be considered robust enough to internalize generation capacity costs, there may be no need to calculate a separate capacity element – the avoided energy costs can be taken to incorporate both the avoided marginal cost of generation and the cost of generation capacity. The problem with this approach is that it is often unclear whether capacity costs have in fact been fully internalized (hence the global trend towards separate capacity markets). Another approach is to try to separate the energy and capacity elements in the energy market price (Woolf et al, 2013).

There are many situations in which robust markets do not exist, and in some cases, even where markets exist, a reference point is needed for capacity or reliability markets. In such cases (for instance, in the UK capacity market and the PJM reliability market) a benchmark known as CONE is often used. CONE stands for Cost of New Entry. It represents the annual revenue a new generation source needs to earn to cover its capital investment and fixed costs (The Brattle Group, 2014). It is normally calculated via a bottom-up analysis of costs such as power generation equipment, plant design, emission control equipment, plant construction and so on, which account for the majority of the total generation capacity cost, along with fixed costs such as property taxes, insurance and fixed manpower requirements. The CONE value is then calculated by taking the annual average of these costs, plus a return on capital, over the assumed lifetime of the plant concerned78. The calculation is often then translated into “netCONE”, which is arrived at by subtracting assumed margins for energy and ancillary

76 Section 2.3 provides a background on how avoided costs are used in DR programmes. 77 We have added a reserve margin to the calculation of capacity avoided cost but that in future analyses the question would require more detailed consideration in the context of the planning and operation of the Shanghai system. 78 For more details of the CONE method, refer to Section 3.2.2.


services from the estimated CONE. The resulting number gives an indication of the annual revenue required (over and above income from the other sources mentioned) in order to remunerate the provision of capacity. The calculation depends on various assumptions, including the choice of reference generation plant, regional construction costs and so on. For example, The Brattle Group (2014) provides estimates for the CONE value of open and combined cycle gas turbines in five different areas, ranging from $138 to $204 per kW per annum. By comparison, the UK valuation of netCONE, for the purposes of its capacity market and using a somewhat different methodology, was £49/KW per annum (DECC, 2014b). If we had sufficient information to calculate it for Shanghai, we would propose following this methodology. However, taking account of the information available, we decided instead to adopt the generation cost model developed by Energy and Environmental Economics (E3) to calculate avoided generation capacity cost (as a proxy of CONE). This model is used to calculate the fixed and variable costs of building and operating different types of thermal power plants in China.79

Avoided energy costs

Avoided energy costs due to a reduction in the amount of electricity generation required can similarly be calculated in various ways – for instance, by using historic and estimated future load profiles along with a forecast of the average value of wholesale energy (or of wholesale market prices where these exist). In a region without a wholesale energy market, avoided energy costs can be determined by comparing energy costs under two circumstances: with and without DR. However, it is difficult and controversial to estimate avoided energy costs because there is no true counterfactual: we simply do not know what the demand would have been without the DR program80.

Avoided costs of ancillary services

Avoided costs of ancillary services (e.g. secondary and tertiary reserves) are another important benefit from DR and they are also subject to the absence of a clear counterfactual. However, in many countries, there are markets for these services; the prices in these markets may provide a reasonable estimate of their value to the system.

Avoided transmission and distribution costs

There are potential savings from avoided transmission and distribution costs as well as from avoided line losses. Most of the value of avoided costs from transmission and distribution are in capacity rather than operations. In practice, avoided transmission costs are difficult to estimate because they depend on the time period, specific location and overall configuration of the system. The starting point is usually to identify future potential network congestion and consider the capacity needed to relieve that congestion in the absence of DR. In some situations, DR may help to avoid or defer very large system costs.

As with generation capacity, the calculation for avoided network costs may be made by reference to market prices where these exist. For instance, analysts sometimes measure the avoided cost of transmission by reference to the difference in locational marginal prices. However, this is a relatively short-term measure and only relevant to some systems. Avoided distribution investment costs are even more complicated and depend on the specific circumstances. Avoided losses can be determined by referencing to the additional costs resulting from line losses between the point of generation and the point of retail delivery.

Overall, while all of the avoided costs are relatively easy to identify in principle, DR programme benefits can be difficult to calculate in practice. They are dependent on the nature, timing and uncertainties of the DR programmes (e.g. direct load control may provide more certainty in providing capacity resources than real time pricing) as well as the configuration of the power system, and whether robust markets are in place as reference points.

Potential costs of implementing DR

The discussion above focuses on the benefits of DR rather than on the level and allocation of costs. However, it must be borne in mind that there are also costs. For instance, there is capital expenditure involved in the installation of demand response equipment such as smart meters and demand control technologies. It is therefore important when working out the details of a DR programme to clarify the costs and benefits by 79 For details of the E3 model, refer to <https://www.ethree.com/public_projects/generation_cost_model_for_china.php> 80 Avoided energy costs are a small component of the total avoided costs related to DR in the US. However, China is likely to be different as the number of hours for DR calls is probably greater than other countries.


reference to the different stakeholders, including the system operator, the participants in the DR programme and other consumers. The balance between costs and benefits, and the allocation of the costs and benefits, will depend on the nature of the system and regulations concerned.81

3.3 Approach for Assessing DR Potential and Values in Shanghai

Following the general framework as discussed above, this section describes the specific analytical approach for estimating the potential and benefits of DR resources in Shanghai. One challenge is the limited data availability for some of the key parameters underpinning the potential of DR and its benefits. In these cases, this study has used assumptions and techniques to approximately estimate these parameters. Since this introduces uncertainty, future studies of the potential and benefits of DR will benefit from stronger evidence specific to Shanghai.

3.3.1 Assessment of DR potential

Background of electricity system in Shanghai

Electricity consumption in Shanghai has increased considerably in recent years, from 56 TWh in 2000 to 141 TWh in 2013 with an annual growth rate of 7.4% on average (Figure 14). Industrial electricity use has exhibited an annual growth rate of 6% on average between 2000 and 2013. Commercial and residential electricity consumption shows an annual growth rate of 11% on average during the same period of time, while agricultural electricity use has declined annually by 2% on average. Industrial electricity use takes up the largest share of total electricity consumption in Shanghai, while that share has been declining gradually, from 70% in 2000 to 57% in 2013. By contrast, the share of commercial and residential electricity use has risen from 19% to 28% and from 10% to 15% respectively during 2000-2013. Electricity imports from other jurisdictions have made a notable contribution to the electricity supply in Shanghai, with its share around 30% in recent years.

Figure 14 Total electricity consumption by sector in Shanghai for recent years

Source: Shanghai Statistical Yearbook 2014, 2013 and 2011

81 Section 2.3.3 includes a discussion on the potential benefits from a stakeholder perspective, as well as identifying the costs of DR from the perspective of a California utility.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0

20

40

60

80

100

120

140

160

2000 2010 2012 2013 Agricultural 0.9 0.6 0.7 0.7

Residential 5.3 16.9 18.7 20.5

Commercial 10.4 33.4 37.3 39.9

Industrial 39.3 78.7 78.6 79.9

Net Import 27% 28% 31%

% of Im

port from O

ther Jurisdictions TW

h


Figure 15 Historical growth and estimate of highest system peak demand in Shanghai

Dark blue line represents the forecast highest system peak demand by the State Grid in Shanghai for 2015-2017 The installed generation capacity in Shanghai for 2006 is not available. Source: Shanghai Statistical Yearbook 2014, 2011 and 2006; data from the State Grid in Shanghai

Figure 16 Forecast for peak demand in future years (2020, 2025 and 2030)

The electricity system in Shanghai experiences the highest annual peak demand in summer months. The typical system peak period82 in Shanghai is 1p.m.-3p.m. of weekdays in summer months. Between 2002 and 2013, the highest annual peak demand has exhibited significant growth (Figure 15). The annual growth rate of the annual peak demand is 8.2% on average for the same period of time. As for years 2015, 2016 and 2017, the State Grid in Shanghai forecasts system peak demand of 30GW, 31.5GW and 33GW respectively. As discussed below, this study considers the DR market potential for milestone years of 2020, 2025 and 2030. Given the lack of peak demand forecast for years beyond 2017, the analysis makes very simple estimate of system peak demand for 2020, 2025 and 2030, by assuming the same growth rate of system peak demand (i.e. annual growth of 5%) as

82 Time period when system peak demand reaches the highest level

12.4 13.6

15.0 16.7

19.5 21.2

22.4 23.8

26.2 25.5 25.9

29.6

26.8

30.0 31.5 33.0

11.4 11.1 12.0

13.4 14.4

16.8 16.6 18.6

19.6 21.5 21.6

0

5

10

15

20

25

30

35

40

45

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

GW

Highest Peak Demand Installed Generation Capacity in Shanghai

30

38

49

62

0

10

20

30

40

50

60

70

2015 2020 2025 2030

GW


implied in the forecast by State Grid (Figure 16). However, it should be noted that growth of demand usually slows down as economies and markets become more developed.

While the installed generation capacity in Shanghai has been growing at an average annual rate of 6%, it can only meet 70-80% of the peak demand. The 21.6GW of installed generation capacity in 2013 was roughly made of 12GW of coal-fired capacity, 7GW of gas-fired capacity and 2.6GW of on-site generation at customers’ premises. At the same time, external generation capacity totaling 14GW from other jurisdictions is accessible to Shanghai, about two-thirds of which is hydropower. The load-following practice in Shanghai primarily relies on installed thermal generation capacity, with gas-fired generation capacity being given the priority.

The State Grid, which is responsible for the electricity T&D and retail services, has provided the aggregate load profile for the whole system in Shanghai on a typical weekday in summer (Figure 17). For the summer weekday chosen by the State Grid, the average system peak demand for 1p.m.-3p.m. reaches 26.1GW. The air conditioning (AC) load makes a considerable contribution to the system load83, constituting 45-50% of the system load between 1p.m. and 3p.m. on the chosen typical summer weekday.

Figure 17 Aggregate hourly load of electricity demand in Shanghai on a typical summer weekday (6 August 2014)

Source: Data provided by the State Grid in Shanghai

The electricity demand in Shanghai in the summer exhibits a significant daily peak-valley difference84. For instance, in 2013, the peak-valley difference on the day with highest annual system peak demand was 12GW, which is 41% of the highest peak demand that day. Similarly, for 2009-2011, the peak-valley difference on the day with the highest system peak demand is 40-50% of the highest system peak demand on that day. The large peak-valley difference contributes to lower utilisation hours for generation capacities. In 2013, while the highest annual peak demand in Shanghai is 29.4GW, the duration of system demand over 24.4GW is only 367 hours in a year. This leads to a low load factor and underutilisation of generation capacity, thus creating economic inefficiency in the system operation. Moreover, the widening peak-valley difference also increases strain on thermal generation capacity to follow load. With the growing contribution of external hydropower generation capacity, most of which is available to Shanghai in evening hours when the system demand is at its low point, the pressure of load-following becomes more intense.

Scope of analysis

In assessing the market potential of DR in Shanghai, this study considers commercial, industrial and residential sectors. The agricultural sector is excluded since compared with other sectors it is small in terms of its electricity use. Moreover, this study focuses on the dispatchable incentive-based DR opportunities in Shanghai. The analysis expects these programmes to provide resource adequacy services, largely because such programmes

83 The State Grid provides the data of AC load, which is estimated by the difference between the 15-minute load data for 24 hours of the chosen summer weekday and that of another weekday in April. This is based on the assumptions that 1) non-AC load remains the same between the two chosen days and 2) there is no AC load on the chosen weekday in April. 84 The difference between highest and lowest electricity demand on a particular day

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0

5

10

15

20

25

30

% of A

C Load in H

ourly Electricity Dem

and Lo

ad o

f Ele

ctric

ity D

eman

d (G

W)

Non-AC Load AC Load % of AC Load


constitute a large share of the existing DR resources on the international level (see Section 2). Given the existing structure of electricity industry in China, dispatchable incentive-based DR for resource adequacy services appears viable in the near future.

As for non-dispatchable price-based DR, while the absence of wholesale market and thus the mechanisms to discover marginal cost makes it hard to introduce RTP, TOU tariffs are available to customers in Shanghai for a long time and CPP is to be trialled out in other parts of China. However, the limited evidence showing the price elasticity of electricity consumption in China does not allow this study to undertake rigorous analysis. With the progress of electricity market reform in China, particularly the direction towards a more market-based model, considering the potential for price-based DR programmes is nonetheless worthwhile in future research.

The analysis considers two types of dispatchable incentive-based DR programmes, both of which should be viable in Shanghai:

§ Direct load control (DLC) – for residential customers, this analysis assumes their use of air conditioning (AC) to be remotely controllable by the system operator or sponsoring entity of such DLC programmes, when the pre-defined system conditions materialise (e.g. shortfall in system reserve). Other end-uses (e.g. electric water heating) are excluded from this analysis, largely due to the lack of data showing the usage pattern (e.g. diversity and coincidence factors) and market penetration of these end-uses. For small and medium C&I customers, DLC programmes may also be offered to control end-uses such as AC.

§ Curtailable programmes – these programmes typically make payment to customers in return for accepting the obligation to reduce their overall electricity demand by a specific amount or to a pre-defined level when pre-defined conditions are triggered. In other words, curtailable programmes focus the reduction in overall electricity demand without targeting specific end-uses. Depending on their design and purpose, these programmes can take many forms (e.g. interruptible tariffs, emergency DR, load as a capacity resource, or ancillary service resource) that may vary depending on the system and market characteristics. This analysis assumes curtailable programmes can be offered to C&I customers.

The analysis also assumes the estimated DR market potential to be achieved by customers from load shedding only, largely because of the limited evidence showing the propensity of customers for different DR strategies (e.g. load shedding, load shifting or use of on-site generation). It focuses on the 1p.m.-3p.m. of weekdays in summer, since it is the typical time period when the system in Shanghai experiences the highest peak demand in a day. This study also assumes a 15-year ‘ramping-up’ period for the incentive-based DR programmes being considered. This is primarily because such programmes, particularly those operating in the market-based model, have very limited experience in Shanghai. The utility company and customers need time and resources to build or strengthen their capabilities for the wide rollout of these DR programmes. A number of milestone years (i.e. 2020, 2025 and 2030) are chosen to mark the progress towards achieving specific market penetration levels for DR programmes (i.e. participation rate).

Given the constraints of data and time, this study considers only the ‘overall scale’ of DR market potential but not how such potential may be distributed amongst DR products with different participation requirements (e.g. duration and frequency of response, maximum response time and financial penalty for non-performance). Moreover, while the analysis takes into account the contribution of potentially flexible end-use loads (e.g. AC) wherever possible in making assumptions on the DR load impact85, the lack of detailed load profile for various end-uses makes it very difficult for this analysis to consider the potential of different end-uses in providing DR services. However, with the development of DR experience and the increasing availability of detailed metering data (e.g. for a larger customer population and various end-uses), future research would be able to make more detailed assessment and provide more targeted estimates to inform the design of DR programmes and policies.

Customer segmentation

This study relied on the classification system of the State Grid to group customers based on their sectors and major business activities. Customers are segmented into residential, agricultural, and industrial and commercial sectors, which are further divided into sub-sectors. The analysis estimated the annual electricity use and number of customers in each sub-sector for future years. Table 12 summarises the estimated annual electricity use and number of customers for 2020, 2025 and 2030.

85 For instance, the assumed DR load impact (% load reduction) may not exceed the contribution of flexible end-use load (e.g. AC).


§ The annual electricity use in each sub-sector for future years was estimated by applying the historical average growth rate to the sub-sector-level electricity consumption data for 2014, which was supplied by the State Grid in Shanghai. The historical average growth rate for each sub-sector was estimated based on the annual electricity consumption data by sector86 for Shanghai during 2011-2013.

§ The number of customers for each sub-sector in Shanghai for 2008-2010 is available from the Compilation of Statistical Materials of Electric Power Industry 2009-2011. Given the lack of recent data, the analysis first calculated the average percentage change in the number of customers for residential, commercial, industrial and agricultural sectors87. It then applied the average percentage of change to the number of customers in each sub-sector in 2010 to gauge the number of customers for 2020, 2025 and 2030. As the ongoing economic transition unfolds in China, it seems reasonable to assume that the number of industrial customers would decrease, while that for commercial and residential customers may go up. This has can have important implication for the DR market potential for each customer segment (e.g. more DR resource to come from commercial and residential sectors) as well as the design and implementation of DR programmes. However, given the lack of dedicated study on the future trend of customer population, this analysis made very rudimentary estimate, which may not reflect the actual trend88.

Many international DR potential studies have considered the impact of customers’ peak demand size on their potential and capabilities in providing DR services. It is also common to limit the participation in specific DR programmes based on individual customers’ highest peak demand or potential to deliver a minimum level of load reduction. These considerations make it necessary to segment customers based on their peak demand. Given the absence of empirical evidence on how customers’ peak demand size or other characteristics of electricity use may influence their potential to deliver DR in Shanghai, this study followed FERC (2009) in distinguishing small (<20kW89), medium (20-200kW) and large C&I customers (>200kW). As the per-customer coincident peak demand in each sub-sector is not readily available in Shanghai, the analysis further considered the customer segmentation in the next step of estimating the average per-customer load profile. It should be noted that, in this study, the average per-customer load profile only represents an average per-customer level of peak demand amongst the customers in each sub-sector. While data available to this study does not allow further segmentation of customers based on their peak demand within each sub-sector, future research would benefit from detailed information on the peak demand of individual customers to refine the potential assessment in each sub-sector.

The analysis also assumed the following programme eligibility for customers with different peak demand size:

§ DLC for AC programmes are offered to residential, and small and medium commercial customers. One key factor determining the eligibility in such programmes is the ownership of AC so it is very important to estimate the penetration rate of AC. For residential customers, the statistics from the National Statistics Bureau in China estimate the average ownership rate of AC in households to be 172 pieces per 100 households in Shanghai in 2013. Assuming that households owning AC have 2 pieces of AC equipment on average, the penetration rate is estimated at 86%. For C&I customers, the AC penetration rate is assumed at 100% in the analysis.

§ Curtailable programmes are available to small, medium and large C&I customers. Due to the data limitation, the analysis assumes that all customers and their entire load in these categories are eligible to participate in curtailable programmes. However, in reality, the eligibility depends on the specific programme design and the objectives of programme sponsors.

Therefore, future research could benefit from more evidence in Shanghai that shows:

§ Penetration of particular electricity end-uses (e.g. air conditioning, water heating) that can be targeted for DLC programmes;

§ Any electricity end-uses not suitable for DR and their share in electricity demand in different customer segments;

§ Penetration of enabling technologies that are necessary for participation in DR programmes.

86 Since the data does not divide the industrial annual electricity consumption by sub-sector, the analysis only determined the average annual growth rate for the whole industrial sector. 87 This reduces the impact of significant annual variability on the estimation. 88 Based on experience in the US, it seems reasonable to expect the number of industrial consumers will taper off in the later years and that incremental DR will come from growing numbers of commercial and residential consumers. 89 Referring to individual customer’s average peak demand coincident with the system peak period in summer


Table 12 Estimates of annual electricity use and number of customers by sub-sector in 2020, 2025 and 2030

Total Annual Electricity Use (TWh) Number of Customers (in thousands) Average % of Growth*

2014 2020 2025 2030 Average % of Change**

2010 2014 2020 2025 2030

Residential 3% 17.4 20.8 24.1 27.9 2% 7,947 8,757 10,129 11,436 12,911 Agriculture 6% 0.7 1.0 1.4 1.8 -4% 87 75 59 49 41 Industrial -0.1% 106 105 105 104 103

Mining 1% 0.0 0.0 0.0 0.0 -0.1% 0.2 0 0.2 0.2 0.2 Food/Beverage/Tabaco Manufacturing 1% 1.7 1.8 1.9 2.0 -0.1% 8.2 8 8.1 8.1 8.0 Textile 1% 1.1 1.2 1.3 1.3 -0.1% 4.1 4 4.0 4.0 4.0 Clothing Manufacturing 1% 0.8 0.8 0.9 0.9 -0.1% 5.2 5 5.2 5.2 5.1 Timber and furniture 1% 0.6 0.7 0.7 0.8 -0.1% 4.1 4 4.1 4.1 4.1 Paper Manufacturing 1% 0.8 0.8 0.9 0.9 -0.1% 1.1 1 1.1 1.1 1.1 Printing Industry 1% 0.4 0.5 0.5 0.5 -0.1% 1.3 1 1.3 1.3 1.3 Stationary Manufacturing 1% 0.2 0.2 0.2 0.2 -0.1% 0.5 1 0.5 0.5 0.5 Petrochemical/Coking/Nuclear Material 1% 2.4 2.6 2.7 2.8 -0.1% 0.4 0 0.4 0.4 0.4 Chemical Materials and Products 1% 8.5 9.0 9.5 9.9 -0.1% 2.5 3 2.5 2.5 2.5 Pharmaceutical Manufacturing 1% 0.9 0.9 1.0 1.0 -0.1% 0.7 1 0.7 0.7 0.7 Chemical Fibre Manufacturing 1% 1.1 1.2 1.2 1.3 -0.1% 0.3 0 0.3 0.3 0.3 Rubber and Plastics Manufacturing 1% 3.7 4.0 4.2 4.4 -0.1% 7.2 7 7.2 7.1 7.1 Non-Metallic Mineral Product 1% 1.5 1.6 1.7 1.8 -0.1% 2.1 2 2.1 2.1 2.0 Ferrous Metal Smelting and Rolling 1% 15.2 16.2 17.0 17.9 -0.1% 0.1 0 0.1 0.1 0.1 Non-Ferrous Metal Smelting and Rolling 1% 0.6 0.7 0.7 0.7 -0.1% 0.4 0 0.4 0.4 0.4 Metalwork Manufacturing 1% 4.6 4.9 5.1 5.4 -0.1% 14.6 15 14.5 14.4 14.3 Machinery Manufacturing 1% 6.1 6.5 6.8 7.2 -0.1% 19.2 19 19.0 18.9 18.8 Transport/Electronics/Electrical

Equipment 1% 10.6 11.3 11.9 12.5 -0.1% 7.4 7 7.3 7.3 7.2

Craftwork and Other Manufacturing 1% 3.3 3.5 3.7 3.9 -0.1% 12.1 12 12.0 11.9 11.9 Waste Management and Recycling 1% 0.0 0.0 0.0 0.0 -0.1% 0.3 0 0.3 0.3 0.3 Electricity/Gas/Water

Generation/Supply 1% 14.2 15.1 15.9 16.7 -0.1% 5.7 6 5.7 5.6 5.6

Construction 1% 1.4 1.4 1.5 1.6 -0.1% 7.8 8 7.8 7.7 7.7 Commercial 7% 697 907 1,345 1,867 2,593

Transport, Warehouse and Post 10% 4.1 7.3 11.8 19.0 7% 6 7 11 15 21 IT 5% 1.8 2.4 3.1 3.9 7% 24 31 46 65 91 Commercial and Hospitality 3% 7.6 9.1 10.6 12.3 7% 111 146 219 307 431 Finance, Real Estate and Business 11% 16.5 30.9 52.0 87.6 7% 450 590 885 1,242 1,742 Utilities and Public Governance Bodies 6% 8.8 12.5 16.7 22.4 7% 107 140 210 294 413

*Based on the data of annual electricity use by sector for years 2008-2014 available from Shanghai Statistical Yearbooks 2013, 2012 and 2010, and from the State Grid in Shanghai. The growth rate for IT sub-sector is determined based on the data for 2013 and 2014 since previous data reporting does not report the annual electricity use for this sub-sector. **Based on the data of number of customers in each sub-sector in Shanghai for 2008, 2009 and 2010 available from Compilation of Statistical Materials of Electric Power Industry 2009-2011. Negative average percentage of change indicates a decrease in the number of customers. Source: Own calculation based on the available data

Average per-customer load profile

Ideally information would be available on the load profile for a typical individual customer in different segments. This may be generated by conducting metering studies for a representative sample of customers in each segment. Alternatively, it can be estimated by dividing the aggregate load coincident with system peak period for each customer segment by the number of customers in each segment. However, neither the typical load profile as determined from a metering study or the aggregate coincident peak demand for each customer segment in Shanghai was available to this study. For this reason, this analysis tries to roughly estimate the average peak load profile for each customer segment, defined as the average per-customer electricity demand during 1p.m.-3p.m. on weekdays in summer.

The analysis relied on the load profiles of a sample of customers to estimate the average load profile for different customer segments. The load profiles for the sampled C&I customers are from the State Grid in Shanghai and the Shanghai DR pilot in the summer of 2014, while those for the sampled residential customers are from Hangzhou Telek Technology Co., Ltd. To start with, the analysis estimated the average peak load factor (PLF) for each customer segment. Given the quality of data, different techniques are used to estimate the average PLF for C&I and agricultural customers, and residential customers.


§ C&I and agricultural customers

The State Grid in Shanghai provided the electricity demand of 60 C&I customers at a 15-minute interval during 1p.m. and 3p.m. of two weekdays in summer90. The load profiles of 1 agricultural, 25 commercial, 6 industrial customers on a weekday without load response in the 2014 Shanghai DR pilot were also available to this study. Table 13 shows the distribution of sampled customers amongst different sub-sectors. Given the customer number in each segment, the sample size is very small and does not cover all the sub-sectors. Moreover, the load data is only for one or two days, which may not accurately reflect an average or typical load pattern on weekdays for the whole summer. These considerations highlight the uncertainty of this analysis in estimating the average load profile, which can be reduced in future research with a larger sample size and longer metering duration.

For each individual customer in the sample, the average electricity demand during 1p.m. and 3p.m. of the chosen day (i.e. average peak demand) is first calculated. In the case of the customer sample provided by the State Grid in Shanghai, the average peak demand of individual customer is estimated by averaging the electricity demand for 1p.m.-3p.m. on the two days chosen. Subsequently, the peak load factor (PLF) is calculated for each individual customer by dividing the annual average demand (i.e. annual electricity consumption averaged over 8760 hours in one year) by the estimated average peak demand. Following that, the average PLF is determined for the sub-sector from all sampled individual customers in that sub-sector. As a number of sub-sectors are not covered in the sampled customers, a PLF of 70% is assumed for these sub-sectors following literature evidence and practices in other DR potential assessments (e.g. Chen et al., 2011; FERC, 2009).

Table 13 Distribution of sampled customers with load profile information

State Grid in Shanghai 2014 Shanghai DR Pilot Total Industrial

Textile 1 0 1 Petrochemical/Coking/Nuclear Material Processing 0 1 1 Chemical Materials and Products 4 0 5 Chemical Fibre Manufacturing 2 0 2 Ferrous Metal Smelting and Rolling Processing 1 1 2 Non-Ferrous Metal Smelting and Rolling Processing 1 1 3 Metalwork Manufacturing 2 0 1 Machinery Manufacturing 7 1 7 Transport, Electronics and Electrical Equipment 12 1 14

Commercial Transport, Warehouse and Post 1 0 1 IT 0 1 1 Commercial and Hospitality 5 3 8 Finance, Real Estate and Business 16 20 36 Utilities and Public Governance Bodies 8 1 9

Agricultural 0 1 1

Source: Data provided by the State Grid in Shanghai and from the 2014 Shanghai DR pilot

§ Residential customers

The load profiles of 100 households in Shanghai for 1p.m.-3p.m. on two weekend days91 in August 2014 were made available to this study. Since no information was provided for the annual electricity use for sampled individual customers, the analysis first determined the average of electricity demand amongst the 100 households for every 15-minute interval during 1p.m.-3p.m. on these two days (i.e. 18 average values in total). Then the average of the 18 average values was taken to represent the average per-customer peak demand. Subsequently, the average per-customer annual electricity use in 2014 was calculated by dividing the total annual electricity use for residential customers with the number of residential customers in 2014, which were already estimated in the preceding ‘customer segmentation’ section. Finally, the average per-customer annual electricity use in 2014 was divided by 8,760 hours in a year and the estimated average per-customer peak demand to roughly determine the average PLF for residential customers.

Table 14 lists the assumed PLFs for different sub-sectors. Since these PLF were estimated based on the very limited load data available to the analysis, it is perhaps not surprising to see any discrepancy with results or assumptions in other research and studies. For example, based on the load analysis for over 100 customers in cities like Shanghai, Beijing and Tianjin, Chen et al. (2011)estimated PLFs of 0.7-0.95 for industrial customers

90 6th and 29th August in 2014 91 It is impossible for this study to assess how representative these sampled households are of the entire residential sector in Shanghai. Load profiles for weekdays were not made available to the analysis.


and 0.57-0.83 for commercial customers, while that for residential customers was estimated to be 0.74. At the same time, FERC (2009) assumed the PLF for C&I customers to be within 0.6-0.7. While these figures may not be directly comparable, future research will benefit from PLFs estimated in a more rigorous way (e.g. larger, more representative customer sample).

The PLFs were applied to the annual electricity consumption in 2014 for each of the customer sub-sectors to tentatively determine the average peak demand for each sub-sector. The estimated average peak demand for each sub-sector was summed up and then scaled proportionately to fit with the highest system peak demand of 26.8GW in 2014 (i.e. applying the same scaling factor to each sub-sector). For each sub-sector, the number of customers divides the scaled average peak demand in 2014 to estimate the average per-customer peak demand (or ‘load profile’). As for future milestone years (e.g. 2020, 2025 and 2030), the analysis assumed the PLFs in Table 15 for respective sub-sectors and the same ‘scaling’ process was followed to fit with the forecasted system peak demand and determine the average per-customer peak demand for each sub-sector in future years. Table 15 summarises the estimated average per-customer peak demand for each of the sub-sectors in 2014, 2020, 2025 and 2030. For C&I sectors, it forms the basis for determining whether an average customer in each sub-sector can be classified as a small, medium or large customer.

Table 14 Estimate and assumption of Peak Load Factors (PLFs) by sub-sector

Sub-Sectors Estimated Peak Load Factor (PLF) Residential 35% Industrial

Mining 70%* Food/Beverage/Tabaco Manufacturing 70%* Textile 74% Clothing Manufacturing 70%* Timber and furniture 70%* Paper Manufacturing 70%* Printing Industry 70%* Stationary Manufacturing 70%* Petrochemical/Coking/Nuclear Material Processing 86% Chemical Materials and Products 68% Pharmaceutical Manufacturing 70%* Chemical Fibre Manufacturing 58% Rubber and Plastics Manufacturing 70%* Non-Metallic Mineral Product 70%* Ferrous Metal Smelting and Rolling Processing 30% Non-Ferrous Metal Smelting and Rolling Processing 28% Metalwork Manufacturing 86% Machinery Manufacturing 54% Transport, Electronics and Electrical Equipment 53% Craftwork and Other Manufacturing 70%* Waste Management and Recycling 70%* Electricity/Gas/Water Generation/Supply 70%* Construction 70%*

Commercial Transport, Warehouse and Post 73% IT 51% Commercial and Hospitality 47% Finance, Real Estate and Business 41% Utilities and Public Governance Bodies 34%

Agriculture 37%

* Indicates assumed PLF Source: Estimation based on the load profile data to this study and balanced consideration of other evidence in literature


Table 15 Estimated total peak demand and average per-customer peak demand by sub-sector for 2020, 2025 and 2030

Total Peak Demand (MW) Average Per-Customer Peak Demand (kW) 2014 2020 2025 2030 2014 2020 2025 2030

Residential 4,530 6,154 7,269 8,259 0.5 0.6 0.6 0.6 Agriculture 179 289 393 516 2 5 8 13 Industrial

Mining 5 5 6 6 26 32 34 35 Food/Beverage/Tabaco Manufacturing 219 265 284 292 27 33 35 36 Textile 141 170 182 187 35 42 45 47 Clothing Manufacturing 101 122 130 134 19 23 25 26 Timber and furniture 83 101 108 111 20 25 26 27 Paper Manufacturing 101 122 131 135 93 113 121 126 Printing Industry 57 69 73 76 42 51 55 57 Stationary Manufacturing 25 30 32 33 48 58 62 64 Petrochemical/Coking/Nuclear Material 257 311 333 343 708 860 926 959 Chemical Materials and Products 1,136 1,372 1,470 1,514 453 550 592 613 Pharmaceutical Manufacturing 114 138 148 152 170 206 222 230 Chemical Fibre Manufacturing 172 208 223 229 526 639 687 712 Rubber and Plastics Manufacturing 487 588 629 648 67 82 88 91 Non-Metallic Mineral Product 200 241 258 266 96 117 126 130 Ferrous Metal Smelting and Rolling 4,634 5,597 5,993 6,174 33,714 40,963 44,086 45,642 Non-Ferrous Metal Smelting and Rolling 206 248 266 274 469 570 614 635 Metalwork Manufacturing 485 585 627 646 33 40 44 45 Machinery Manufacturing 1,031 1,245 1,333 1,373 54 66 71 73 Transport/Electronics/Electrical Equipment 1,828 2,208 2,365 2,436 249 302 325 337 Craftwork and Other Manufacturing 430 519 556 573 36 43 47 48 Waste Management and Recycling 4 5 5 5 13 16 17 18 Electricity/Gas/Water Generation/Supply 1,853 2,238 2,396 2,469 326 396 426 441 Construction 177 214 229 236 23 28 30 31

Commercial Transport, Warehouse and Post 515 1,039 1,704 2,690 71 96 112 126 IT 322 490 638 798 10 11 10 9 Commercial and Hospitality 1,483 2,015 2,380 2,704 10 9 8 6 Finance, Real Estate and Business 3,668 7,806 13,402 22,135 6 9 11 13 Utilities and Public Governance Bodies 2,360 3,809 5,193 6,811 17 18 18 16

Source: Own calculation based on available data

Participation rate estimation by customer segment and programme

Given the limited empirical evidence in Shanghai in recruiting customers for DR programmes92, the analysis used international benchmarking as the main technique in making assumptions for the customer participation rate for different DR programmes. This study reviewed the results of FERC DR Survey in 2012 and several international DR potential assessment studies and evaluation reports to form assumptions about customer participation rate. As discussed in Section 3.2.1, the approach of benchmarking has its limitation in fully capturing the locally specific factors (e.g. level of electricity price, existence of incentives for customers to become flexible with their electricity use, availability of enabling technology/infrastructure) that may influence the participation rate. Thus the analysis considered the results of international benchmarking and the inputs from local and international experts to make assumptions.

In analysing the results of FERC DR Survey in 2012, this study focused on the results of Q8 in the FERC-731 Survey Form, since it is the only form that contained information needed for estimating participation rate (e.g. number of customers enrolled in DR programmes). For each C&I DR programme listed in the Q8 form, the analysis first calculated the average per-customer peak demand, by dividing the maximum demand and the number of enrolled customers, and classified DR programmes into those targeting small (<20kW), medium (20kW-200kW) or large C&I (>200kW) customers. Then, for each state in the US as listed in the Q8 form, the analysis summed up the numbers of enrolled customers for various DR programmes separately for different customer segments (e.g. residential, small/medium/large C&I) and programme types (i.e. DLC or curtailable programmes93). The summed up number of enrolled customers by customer segment and programme for each state was divided by the total number of customers in respective customer segment94 to calculate the participation rate by customer segment and programme. Finally, for a given DR programme type, different levels of participation rates (e.g. 25th, 50th and 75th percentiles) were identified across the estimated participation rates

92 Especially for market-based programmes beyond the administrative load management programme 93 Including interruptible tariffs, demand bidding & buy-back, emergency DR programmes, load as a capacity resource and other incentive-based DR programmes but excluding DLC programmes 94 This was estimated based on the EIA data in on the number of customers in residential, commercial and industrial sectors in 2011. To estimate the share of customers in small/medium/large C&I sectors, the relative share of these sub-categories as estimated in FERC (2009) was applied to the total number of C&I customers in each state.


across the US states covered by the survey to reflect the varying participation rates that had been achieved in different jurisdictions. Since the Q8 form only included the DR programmes offered by retail utilities (i.e. not including the programmes directly bid into wholesale markets) and the information for many programmes was not complete, the analysis might have underestimated the participation rate.

In formulating the assumed participation rates, this study considered the high-end levels of participation rate in the international benchmarking (e.g. synthesis of international evidence and review of other potential studies), and assumed such levels of participation rate be achievable in the long term (i.e. 2030). Since a programme ‘ramping-up’ period of 15 years was assumed, the participation rate would incrementally increase from 2015 to reach the assumed level by 2030. Three scenarios were made to reflect the different levels of participation rate to be expected by 2030. Table 16 summarises the assumed participation rates for milestone years under these three scenarios. Appendix details the synthesis of international benchmarking on the participation rate.

§ ‘Top-performing’ scenario assumes a level of participation rate by 2030 at the high end of the participation rates achieved in international DR programmes or assumed by other DR potential assessments;

§ ‘Moderate’ scenario assumes a level of participation rate by 2030 at ½ of the level assumed in the ‘top-performing’ scenario; and

§ ‘Basic’ scenario assumes a level of participation rate by 2030 at ¼ of the level assumed in the ‘top-performing’ scenario.

Future estimate of participation rate in Shanghai should benefit from strong evidence (e.g. from programme evaluation) in these following aspects:

§ Relationship of participation rate with specific electricity end-uses, or business or industrial activities;

§ Influence of elements of programme design (e.g. response frequency and duration, speed of response, level of financial incentive, non-compliance penalty, customer engagement and education) on the participation rate of various customer segments;

§ Roles of enabling technologies (e.g. energy management or automated DR system, on-site generation, customer interfaces for visualisation and communication) and their penetration rate in various customer segments; and

§ Impacts of the ongoing electricity price reform on the willingness of customer segments to participate.

Table 16 Assumptions for customer participation rate in future milestone years (2020, 2025 and 2030)

Programme Type Sector Scenarios Participation Rate (% of eligible customers) 2020 2025 2030

Direct Load Control

Residential Top Performing 7% 13% 20% Moderate 3% 7% 10% Basic 2% 3% 5%

Small C&I Top Performing 2% 3% 5% Moderate 1% 2% 3% Basic 0% 1% 1%

Medium C&I Top Performing 3% 7% 10% Moderate 2% 3% 5% Basic 1% 2% 3%

Curtailable Programmes

Small C&I Top Performing 2% 5% 7% Moderate 1% 2% 4% Basic 1% 1% 2%

Medium C&I Top Performing 3% 7% 10% Moderate 2% 3% 5% Basic 1% 2% 3%

Large C&I Top Performing 10% 20% 30% Moderate 5% 10% 15% Basic 3% 5% 8%

Source: Own assumption based on analysis

Load impact estimation by customer segment and programme

To estimate the average per-customer DR load impact, the analysis drew upon international benchmarking and the evidence in 2014 Shanghai DR pilot. Apart from the evidence in 2014 Shanghai DR pilot, international benchmarking (especially the FERC DR survey) does not allow comprehensive synthesis of the possible impact of DR duration length (e.g. 1 or 2 hours) on the average load impact over that time period. The analysis assumed


that the average per-customer load impact as identified in the international benchmarking could be maintained for 2 hours, unless otherwise noted by available information.

§ International benchmarking

Similarly with participation rate, the estimation of average per-customer load impact relied on reviewing the results of FERC DR Survey in 2012 and several international DR potential assessment studies and evaluation reports. In analysing the results of FERC DR Survey in 2012, this study focused on the results of Q8 in the FERC-731 Survey Form, since it is the only form that contained information needed for estimating load impact (e.g. number and maximum demand of enrolled customers, and potential peak reduction). For residential DLC for AC programmes, the average per-customer load impact was estimated by dividing the potential peak reduction (MW) with the number of enrolled customers. For other DLC for AC programmes and curtailable programmes, the analysis divided the potential peak reduction with the maximum demand of enrolled customers to estimate the average percentage (%) reduction in peak demand. Subsequently, for a given DR programme type, different levels of average per-customer load impact (e.g. the 25th, 50th and 75th percentiles) were identified across programmes listed in the survey by customer segment and programme type to reflect the varying levels of load impact achieved by different programmes.

§ Evidence in 2014 Shanghai DR pilot

The outcomes of 2014 Shanghai DR pilot (e.g. load reduction achieved by individual participants during the system peak period) were also considered to complement the international benchmarking:

• Commercial building participants – the DR event on 29 August 2014 requested 27 large commercial buildings to respond in two rotations, which achieved a 9% reduction overall in the total demand of these customers during system peak period. Further analysis of the load profiles of individual participants which were made available to this study demonstrated that 1) the average load reduction over 1p.m.-3p.m. of an individual customer could reach 15% of the baseline demand, while the instantaneous load reduction in particular 15-minute intervals could be nearly 30%; and 2) some individual participants successfully reduced their AC load by an average of up to 15% over 1p.m.-3p.m. and some of them even achieved a nearly 50% reduction in AC load in some 15-minute intervals.

• Industrial participants – only 7 industrial customers participated in the 2014 Shanghai DR pilot. For participants that ‘successfully delivered DR’95, the average load impact over 1p.m.-3p.m. of the DR event day was 11-29%, while some of them achieved over 40% of load reduction for some 15-minute intervals.

As the analysis considers the potential for DLC for AC programmes, the share of AC load in the estimated average per-customer peak demand was also estimated for commercial and residential customers. The 2014 Shanghai DR pilot (26 commercial customers) and the dataset from Hangzhou Telek Co., Ltd (100 households) included the end-use-level load information at 15-minute intervals for 1p.m.-3p.m. on one and two metered days respectively in summer. To estimate the average share of AC in the electricity load, the analysis calculated the ratio between the sum of AC load and the sum of total load for all sampled customers in the same sector (i.e. commercial or residential) for each of the 15-minute intervals on the metered days. Then the share of AC load in average per-customer peak demand was estimated by taking an average of the ratios for all 15-minute intervals on the metered days:

§ Residential customers – 86% of the average per-customer peak demand (or around 0.4kW). The analysis assumed that up to 0.4kW could be reduced per eligible customer via a ‘DLC for AC’ programme.

§ Commercial customers – 43% of the average per-customer peak demand. As for other end-uses, lighting and plug loads take up 42%, while motors and other categories constitute 4% and 11% respectively of the average per-customer peak demand. Assuming that commercial customers can reduce up to 15% of their AC load (as seen in the 2014 Shanghai DR pilot), the analysis assumed that up to 5% of the average per-customer peak demand could be reduced per eligible customer via the DLC for AC programme.

This study also formulated three scenarios to reflect different average per-customer load impact levels that may be expected for various customer segments:

95 Actual electricity demand is lower than the established baseline


§ ‘Top-performing’ scenario assumes a level of average per-customer load impact at the high end of the load impact achieved in international DR programmes and the 2014 Shanghai DR pilot or assumed by other DR potential assessments;

§ ‘Moderate’ scenario assumes a level of load impact at ½ of the level assumed in the ‘top-performing’ scenario;

§ ‘Basic’ scenario assumes a level of load impact at ¼ of the level assumed in the ‘top-performing’ scenario.

While the average per-customer load impact may differ between customer sub-sectors or segments (e.g. size of peak demand of individual customers), the lack of end-use-level load data for these sub-sectors and segments makes it practically difficult to make very detailed assessments. Thus, under the same scenario, the analysis assumed the same level of average per-customer load impact (% of baseline load) for different customer sub-sectors and segments. This assumption should be refined when more locally specific evidence becomes available to show how the potential load impact may differ between customers.

Table 17 Assumptions for average per-customer load impact*

Programme Type Sector Scenarios DR Impact (kW or % of load reduction) Response Rate

Direct Load Control

Residential Top Performing 0.4kW 90% Moderate 0.2kW 90% Basic 0.1kW 90%

Small C&I Top Performing 5% 90% Moderate 3% 90% Basic 2% 90%

Medium C&I Top Performing 5% 90% Moderate 3% 90% Basic 2% 90%


Small C&I Top Performing 40% 60% Moderate 20% 80% Basic 10% 90%

Medium C&I Top Performing 40% 60% Moderate 20% 80% Basic 10% 90%

Large C&I Top Performing 40% 60% Moderate 20% 80% Basic 10% 90%

*Assumed to be average per-customer load impact for 1p.m.-3p.m. of weekdays in summer months

Source: Own assumption based on analysis

Another consideration is the response rate in DR programmes (i.e. the percentage of customers signing up for the DR programmes that actually respond to DR calls by reducing their electricity load). As a matter of reference, the administrative load management programme for industrial customers in Shanghai observed these response rates:

§ 80-90% of response rate for a requested load reduction of 10%

§ 70% of response rate for a requested load reduction of 30%

§ 30% of response rate for a requested load reduction of 50%

Apart from DLC programmes (for which the response rate was assumed to be 90%), the response rate for curtailable programmes was assumed at different levels based on the assumed average per-customer load impact. Table 17 summarises the assumed average per-customer load impact for DR programmes under these scenarios. Appendix details the synthesis of international benchmarking on the load impact.

Given the very limited end-use-level customer load data that is available to the study, it is very difficult to assess the load impact and thus the DR market potential for each end-use. However, as DR is essentially about changing electricity end-use, there is value in assessing in detail the potential of various end-uses in delivering DR, with regard to programme offering and customer engagement. For future research to estimate the DR market potential at the end-use level, it is important to develop evidence in these areas:


§ Detailed end-use-level load profiles for a representative sample for each customer segment, for a reasonable length of time, to show the usage pattern of various end-uses;

§ Technical suitability of each end-use for DR, and how this may differ between customer segments;

§ Potential of load reduction for each end-use that is suitable for DR, for different customer segments.

3.3.2 Assessment of DR benefits

The aim of this avoided cost calculation is to capture the overall benefit to society in Shanghai associated with demand response, including both direct savings and externality values of un-priced CO2 emissions. We are specifically interested in estimating the benefits to society from incremental DR in the summer peak hours.

It is always difficult to calculate avoided costs because we are talking about costs that are not actually incurred. Looking backwards, we can estimate the impact of DR, but we do not really know what the counterfactual is. In other words, we do not really know what would have happened to costs and demand without the measures to encourage demand response. We may therefore attribute a decline in demand and system costs to DR when there are other explanations, including the possibility that consumers would have behaved this way without being encouraged to do so. Looking forward is even more hazardous due to the difficulty of defining a baseline against which to compare the predicted impact on demand and system cost of DR programmes.

It is especially difficult to estimate avoided costs in systems, like China’s, where prices are set administratively rather than through markets, especially where administrative prices do not reflect avoided costs96. Within market systems, at least we have a benchmark for the economic value of alternative sources of supply for capacity, energy and ancillary services. In China, we have few robust market prices to guide us on what economic costs can be avoided through using demand response. Furthermore, we have limited information on electricity generation, transmission and distribution costs in China. Due to the absence of market-based prices and cost information, the estimates of avoided cost are very approximate and are reported as ranges.

We should add that we are not fully confident in transferring OECD experience, at least in terms of detail, to China. Regarding the DR value of avoided capacity costs, one argument in the OECD is that demand may be “peakier” than is optimum, as prices do not generally reflect the marginal cost of supply at peak times. For OECD countries, one of the justifications for DR is therefore that it remedies the price deficiencies and enables electricity systems to avoid the unnecessary construction of new capacity; the savings are then taken into account in the calculations of the value of DR. In China, there has traditionally been a different approach to pricing with less emphasis on cost-reflectivity; instead, there has been more emphasis on administrative arrangements for orderly electricity use to match demand and supply. The implications for DR and optimum capacity are difficult to identify with any certainty, but it is at least possible that there has been under-investment, rather than over-investment, in peak capacity, by OECD standards. In that case, the avoidance of capacity construction would be less central to the case for DR. However, from a wider perspective, there are two key features of the Shanghai system looking forward – that it is becoming peakier and that it is moving towards the greater use of market mechanisms. So over time it is likely to develop in a way that brings it closer to OECD models. Given other factors, such as the need to incorporate more intermittent renewable sources, we believe that the case for examining the potential for DR in China is robust and that there will be clear economic benefits from lowering demand at peak and to encouraging DR to provide flexibility. However, it should be borne in mind that because of the different conditions in China (and the limitations of data availability) any calculation of these benefits must be subject to a high degree of uncertainty.

Therefore, we propose to consider the potential value of demand response within the limits of the available information in Shanghai, taking account of the difficulty of doing so in all systems. In the calculations below, we distinguish between avoided generation capacity costs in the long term, and the other avoided costs (including avoided energy costs, avoided CO2 emission costs, avoided T&D losses) in the shorter term.

§ The reference plant used to estimate avoided generation capacity costs is gas-fired, as explained further below. We understand that this is the technology that would be built in Shanghai under a least cost expansion plan and is in any event the technology mandated by municipal regulations. We offer an estimated range of avoided generation capacity cost in Shanghai, in RMB/kW-yr. To estimate the total savings derived from DR with respect to generation capacity, the avoided cost per kW is multiplied by the potential kW reduction in peak demand. When calculating the system benefits of DR, the avoided capacity

96 The avoided costs are meant to guide demand response programmes and need to be reasonably accurate, even if they are not precise.


costs should also include a reserve margin as well as the marginal T&D losses. We first estimate the avoided cost of an additional kW of capacity. We incorporate the reserve margin and the avoided T&D losses in our overall assessment of benefits. As suggested by international experience, avoided generation capacity costs account for most of the potential savings from DR.

§ In the shorter term, avoided costs are related to the existing system, which is predominantly coal-based. Our estimate for these avoided costs – notably for avoided energy supply, T&D losses and CO2 emissions - is reported in RMB/kWh. To estimate the savings derived from DR with respect to the existing system, the avoided cost per kWh is multiplied by the potential kWh reduction in peak demand. Given the uncertainty about which plant will be operating at the margin, we have carried out this analysis for different types of coal-fired and gas-fired plant. More detailed estimates of the avoided costs would require information on the value of the marginal resources. These could well change over time, including the possibility that a combination of gas and coal-fired power stations will be the appropriate reference since demand response may well lead to lower output from a combination of plants on the system.

The methodology and results are further explained below.

Long term analysis: avoided generation capacity cost

As section 3.2.1 of this report discusses, one methodology for calculating capacity savings in OECD countries is via a calculation of the cost of new entry (CONE), which is regarded as a reasonable proxy for the avoided generation capacity cost, expressed as an annual cost. If we had sufficient information to calculate it for Shanghai, we would propose to follow this methodology. However, taking account of the information available, we have decided instead to adopt the generation cost model developed by Energy and Environmental Economics (E3) to calculate the avoided generation capacity cost. We describe the steps for calculating generation capacity costs in Figure 18.

Figure 18 Steps for calculating generation capacity cost

Identify reference plants and relevant technical specifications

The avoided generation capacity cost depends on the choice of reference plant, its size, the investment costs and other required fixed costs (i.e. not related to generation output). Normally, the reference plant would be built in accordance with a least-cost expansion plan to enable the system to meet peak demand within the system’s resource adequacy standards. It is a hypothetical plant, whose capacity costs might be avoided through DR.

In Shanghai, we think the appropriate reference plant for this study is gas-fired plant, most likely peaking plant. Since this is a hypothetical plant, there can be no definitive answer as to the technology to be used; our choice of a gas-fired plant is based on the following considerations:

§ First, the Shanghai Air Pollution Mitigation Target states that the construction of new coal power plant is not allowed apart from combined heat and power plant and IGCC plant (Ministry of Environmental Protection and Shanghai government, 2014).

§ Second, the Shanghai Clean Air Action Plan (2013 – 2017) limits the use of natural gas for chemical purposes and encourages the development of gas power plant for peaking services (Shanghai government, 2013).

§ Third, DR may make it unnecessary to build this plant. Nevertheless, due to the increasing ‘peakiness’ of the Shanghai system to meet air conditioning demand, we anticipate that a least cost expansion plan would probably include gas-fired plant.97 We certainly see no case for using conventional coal-based generation as the reference plant.

97 Ministry of Environmental Protection and Shanghai Government (2014) Shanghai Air Pollution Mitigation Target (上海市大气污染防治目标责任书). Available from: http://www.mep.gov.cn/ztbd/rdzl/dqst/mbzrs/201401/P020140127516343078870.pdf [in Chinese]. Shanghai Government (2013) Shanghai Clean Air Action Plan (2013 – 2017) (上海市清洁空气行动计划（2013 – 2017）, available from: http://www.shanghai.gov.cn/shanghai/node2314/node2319/node12344/u26ai37377.html [in Chinese]

Identify reference power plant and relevant technical

specifications Bottom-up estimate of

investment costs Estimate fixed operation and

maintenance costs Estimate avoided cost of

generation capaciy


We have chosen two types of gas-fired plant, namely open cycle turbine (CT) and combined cycle gas turbine (CCGT) in order to determine the avoided gas-fired plant capacity cost. However, there is limited information of gas-fired plant in Shanghai, which does not allow us to report the specific plant characteristics. For that reason, we calculate the capacity cost of gas-fired plant based on the technical characteristics (such as plant heat rate, own use, and others) provided in the E3 model.

Bottom-up estimate of investment costs

The bottom-up estimate of investment costs should consider the costs of equipment, material and labour, as well as other costs such as engineering, procurement, construction contracting costs (for plant design and other matters), land, financing fees, and so forth.

We do not have the information to undertake a full bottom up estimate, but report here calculations made by other organizations, focusing on CCGT units (See Table 18). We include data on a CCGT station in Shanghai, which has three units of 400MW combined cycle gas turbines installed with an average unit cost at 2332 RMB/kW98. The E3 data shows unit investment cost of 3249 RMB/kW for a 300 MW CCGT gas plant. For comparative purposes, we also include cost data for a CCGT plant built by the Eastern Mid-Atlantic Area Council (EMAAC) described in the Brattle Group report (2014), which has 668 MW of installed capacity and an average unit cost of 7502 RMB/kW (or 1210 dollar/kW, at an exchange rate of $1 = 6.2 RMB). The Brattle Group estimate includes various costs that are not included in the other calculations mentioned above; for instance, investment in the natural gas and electricity networks accounts for about 10% of the Brattle figure. Because the Brattle costs are not comparable, we do not use them in our modelling of avoided capacity costs in Shanghai.

Table 18 installed capacity and investment of selected gas-fired plant or unit

Source Installed capacity (MW) Investment (RMB/kW)

CT Plant E3 50 2697 The Brattle Group 396 4706

CCGT Plant China Huaneng Group 400 2332 E3 300 3249 The Brattle Group 668 7502

Estimate fixed operation and maintenance costs

Fixed O&M costs include labour, materials, property tax, insurance, asset management costs, and working capital. They include inter alia the costs of permanent employees, materials, property tax, insurance, asset management, working capital and other costs that do not vary with generation output. We have very little information about these other costs in Shanghai. However, the E3 model has included a number of cost parameters and tax rates (including insurance rate, maintenance rate, labour cost, property tax, land use tax and others, see Table 19 below), which we have used to calculate the other fixed costs.

Table 19 Cost parameters and tax rates of CT and CCGT plants

Items CT CCGT Insurance rate (%) 0.25% 0.25% Maintenance rate (%) 2.0% 2.0% Employee salary (RMB/year) 50,000 50,000 Number of employees 4 50 Non-wage costs uplift 60% 60% Property tax (%) 1.2% 1.2% Land use tax (RMB/m2) 10 10

Source: the E3 calculator

Estimate the avoided cost of generation capacity for reference plant in Shanghai

The avoided capacity cost is estimated by using the capital investment costs and the fixed O&M costs together with financial parameters (such as own capital ratio, loan interest rate, after-tax IRR, system lifetime, and depreciation). It also considers the project's risk and the expected lifespan of the asset. The estimated capacity cost range is shown in Table 20. As explained earlier, we are estimating a cost range for CCGT plants because

98 The unit investment is reported by China Huaneng Group (2010) Huaneng Shanghai Gas Fired Power Plant, Available from: http://www.chng.com.cn/n93521/n93759/n93873/c94360/content.html [in Chinese]


we do not have information that enables us to be certain about the appropriate reference plant, nor do we have reliable information about the cost elements for each of these potential technologies in Shanghai. We also include an estimated avoided cost for single cycle combustion turbine (CT) capacity based on E3 data.

Table 20 Estimated capacity cost (RMB/kW-year)

Estimated capacity cost CT Plant 406

CCGT Plant 351 (based on the China Huaneng Group number) 487 (based on the E3 data)

Short term analysis: Other Avoided Costs

In the shorter term, DR may reduce the costs of the existing system, in particular by avoiding costs related to energy, CO2 emissions and T&D losses99. These avoided costs depend mainly on the plant that will be operating at the margin during the peak periods, and specifically on the carbon content of the fuel, the costs of the fuel, the plant efficiency and the shadow price for CO2 emissions that could be avoided by generating less. Table 21 below lists a number of parameters that are used in this study.

Table 21 Parameters used in the estimation of other avoided costs

Parameter Value % of peak demand hours in one year100 4.2% Coal cost (RMB/t) 560 Fuel cost for coal (RMB/tonnes of coal equivalent) 785 Gas cost (RMB/m3) 2.0 Fuel cost for gas (RMB/tonne of coal equivalent) 1629 CO2 price (RMB/t) 40

Source: the proportion of peak demand hours is based on the Shanghai information from NRDC which shows that the peaking load hours were 367 hours in 2013; coal cost and gas cost are from E3; CO2 price is from the Financial Times (2014)101; the fuel cost of coal and gas is based on low heat values (5,000 kcal/kg for coal and 8,600 kCal/m3 for gas) and other conversion factors (one tonne of coal equivalent equals to 29.31 GJ and one calorie equals to 4.184 joules).

We have chosen three types of coal-fired generation plant, i.e. supercritical (SPC), ultra-supercritical (USC) and subcritical (SBC) and two types of gas-fired generation plant, i.e. CT and CCGT, as the reference plants in Shanghai for this analysis because we understand that they are the potential sources of incremental generation in the current system. Adjustment can be made if more information about the future Shanghai power system becomes available, specifically on the nature of demand response (i.e. load shifting or non-load shifting) and which plants are affected (coal-fired, gas-fired, or a mixture). Technical characteristics of these coal-fired plants are provided in the E3 model (See Table 22).

Table 22 Technical characteristics of plant used to estimate other avoided costs

SPC USC SBC CT CCGT Gross heat rate (kgce/MWh) 315 285 330 368 246 Plant own use (%) 8% 9% 10% 3.4% 4.0% Net heat rate (kgce/MWh) 342 313 367 381 256 Fuel use (kgce/MWh) 367 336 393 408 275 CO2 emission factor (tCO2/TJ) 94.5 94.5 94.5 56 56 Plant CO2 emission factor (kg CO2/kWh) 1.02 0.93 1.09 0.67 0.45

Note: Fuel use avoided includes an element for T&D losses (6.68%)

The discussion above focuses on the benefits of DR rather than the costs, but it must be borne in mind that there are also potential expenses when implementing DR - for instance, the capital expenditure involved in the installation of demand response equipment such as smart meters and demand control technologies. As is discussed in Section 2.3.3, the potential expenses to implement DR programmes can be significant, which will decrease the net benefits of DR. The balance between costs and benefits will depend on the nature of the system concerned. Due to the absence of relevant data in Shanghai, we do not include the costs of DR in this report.

99 We do not consider T&D investment cost in this report due to limited information. 100 This is a relatively large number of hours of DR. It raises questions about whether it is better to reduce demand or to increase capacity to meet demand, or to invest in energy efficiency. Note that the costs of DR and other programmes to deal with rising peak demand could raise the average level of tariffs required to recover system costs. 101 The Financial Times (2014) http://www.ft.com/cms/s/0/c9b0faf8-d9e1-11e3-b3e3-00144feabdc0.html?siteedition=uk#axzz3bouqvySA


3.4 Results of Assessment

3.4.1 Estimate of DR potential

Figure 18 shows the assessment results of DR market potential in future milestone years (2020, 2025 and 2030) under different scenarios for customer participation rate and average per-customer load impact. Under the ‘top-performing’ scenario, the analysis shows that the market potential of DR resources could reach 2.5GW in 2030, representing 4% of the forecast peak demand in that year. The ‘moderate’ scenario assessed the market potential at 790MW in total or 1% of the forecast peak demand in 2030. As for the more conservative ‘basic’ scenario, the DR market potential was estimated at a low end of 214MW or 0.3% of the forecast peak demand in 2030.

Figure 19 Assessment of DR market potential in Shanghai

There are a number of key observations from the analysis:

§ C&I curtailable programmes contribute a predominant share in the estimated DR market potential. In the assessment, around 64-73% of the DR market potential is coming from curtailable programmes in the C&I sectors in Shanghai. Industrial curtailable programmes, in particular, can contribute 43-59% of the total DR market potential as estimated for individual milestone years under different scenarios. Under the ‘top-performing’ scenario, out of the 2.5GW of total DR market potential in 2030, curtailable programmes for commercial and industrial customers may deliver 0.5GW and 1.1GW of peak demand reduction respectively.

§ Residential DLC for AC programmes can make a significant share of the contribution. While DLC programmes for AC for small- and medium-sized C&I customers may make only a minimum share, residential DLC for AC programmes constitute 23-33% of the estimated DR market potential for individual milestone years under different scenarios. Under the ‘top-performing’ scenario, residential DLC for AC programmes may be reduced peak demand by 0.8GW in in 2030.

In interpreting the assessment results, it is important to note that the analysis focused exclusively on incentive-based DR options. While TOU tariffs have been offered to various customers in Shanghai for some time, there are still opportunities for price-based DR programmes, especially given the regulatory efforts in electricity market reform and growing policy interest in using pricing signals to influence the electricity consumption of individuals.

13 30 50 52 118 200 209 472

799

1 1 2 4

4

14

21

1 3

6 5 15 22

15

45

100

31

66 102

110 236 364

330

707

1,092

8 24 55 29 89

199

80

242

536

0.0%

0.5%

1.0%

1.5%

2.0%

2.5%

3.0%

3.5%

4.0%

4.5%

-

500

1,000

1,500

2,000

2,500

3,000

2020 2025 2030 2020 2025 2030 2020 2025 2030

Basic Moderate Top Performance

MW

Residential DLC Industrial DLC Commercial DLC Industrial Curtailable Commercial Curtailable % of Forecast Peak Demand


The high penetration rate of smart metering infrastructure in Shanghai also suggests the technological readiness for more innovative price-based DR programmes in the future. In other words, the market potential for both price- and incentive-based DR programmes may well exceed the figures in this study. Moreover, there are also a few other caveats that should be borne in mind.

§ This study only considered dispatchable incentive-based DR programmes for resource adequacy services, largely because 1) these programmes are deemed to be viable under the existing structure of electricity industry in China and could be offered to customers in Shanghai; and 2) the lack of evidence on price elasticity of electricity use to allow the potential estimate for non-dispatchable price-based DR. However, with the new developments in the power industry reform in China, price-based DR, especially beyond the TOU tariff that is already available in some parts of China, may become a viable option and have significant market potential. This means that the assessment may have underestimated the DR market potential. In fact, evidence in the US (e.g. FERC DR survey as discussed in Section 2.2.3) and the results of some DR market potential studies (e.g. FERC, 2009) have shown the sizeable market potential of price-based DR programmes.

§ DLC programmes considered in the analysis only target the AC load, largely due to the lack of data showing the average load contribution of other end-uses and the market penetration of relevant equipment. Since other end-uses such as electric storage water heater and refrigerators (e.g. residential and C&I uses) may also provide DR services, it may be valuable to consider the potential of these end-uses in the future.

§ Given the lack of detailed projection of peak demand and electricity use (e.g. for 2020-2030 for different customer segments), this study made very simple estimates, which may not accurately reflect the future trend of peak demand and electricity use in Shanghai.

§ The limited data availability makes it difficult for this study to conduct detailed assessment for different sub-sectors and end-use categories.

o In estimating the average per-customer load impact, information on the contribution of major end-use categories to the peak demand for different commercial and industrial sub-sectors is insufficient. While the analysis relied on international benchmarking and some locally specific evidence in Shanghai, it did not consider how the load impact potential may differ between sub-sectors with heterogeneous electricity use patterns.

o The limited data showing the contribution of different sub-sectors to the system peak demand and the composition of customers in terms of their peak demand size may have impacted the assessment results. For example, although the commercial building participants in the 2014 Shanghai DR pilot are large customers (with their peak demand >200kW), the analysis estimated the average per-customer peak demand in commercial sub-sectors at a level of small- or medium-sized customers. While there is uncertainty in the estimation, this observation does highlight the diversity within each sub-sector. As the assumptions for participation rate differ between small, medium and large C&I customers, the inability to distinguish customer sizes within sub-sectors may have a bearing on the results of the DR market potential assessment.

§ Given the data limitation and tight project timeline, this study does not consider the difference between DR products (e.g. various ancillary products and capacity resources) that could be offered under curtailable programmes. It indicates a scale of DR market potential without making detailed analysis of the potential for individual DR products with specific technical and commercial characteristics. With the development of DR experience and evidence in Shanghai, there is value in conducting further research on the potential of specific DR products, especially if DR becomes an important resource for electricity system operation.

§ The potential analysis did not consider the cost-effectiveness of DR programmes, which is one measure often employed to gauge whether to include specific DR programmes in the potential analysis. This is mainly because of the difficulty in obtaining data to show the scale of likely programme cost (e.g. investment cost in enabling technology for specific DR programmes like DLC). With the evidence of DR programmes (e.g. load reduction impacts and associated benefits and costs) specific to Shanghai growing, future research could consider the cost-effectiveness of individual DR programmes or products to refine the potential analysis.

3.4.2 Estimate of DR benefits

The total avoided costs of DR in Shanghai include both the avoided generation capacity costs associated with the hypothetical gas-fired plant and the other avoided costs (including avoided energy costs, avoided CO2

emission costs and avoided T&D costs) associated with the reference plant. A discount rate of 7% is used to


determine the net present value102 and applied to both avoided capacity costs and other avoided costs, unless specified otherwise.

Range of avoided generation capacity costs

To estimate the value of DR, we multiply the avoided generation capacity costs (per kW) by the estimated DR potential under the different scenarios that were presented in Figure 19. We include the planning reserve margin103 (assumed at 15%) and the average T&D losses (6.68%) in our estimation. Figure 20 presents the estimated avoided capacity cost between 2020 and 2030. It shows that the avoided capacity costs could reach 554.2 million RMB in 2030 with the estimated top performing DR potential of 2.5 GW and highest CCGT capacity cost of 487 RMB/kW-yr. However, the range of avoided capacity costs varies significantly. For example, the lowest avoided capacity cost is 33.5 million RMB in 2030 based on the lowest CCGT capacity cost (351 RMB/kW-year) and the basic DR potential (0.2 GW). We also present Figure 21 and Figure 22 to show the avoided capacity cost using the discount factor of 0% and 10%, respectively. It shows that the avoided capacity costs can vary significantly with different discount factors.

Figure 20 Avoided capacity cost between 2020 and 2030 with 7% discount factor


102 Future cash flows are discounted to reflect the time value of money. Adjustment of discount factor can be made in the Excel Spreadsheet to reflect the real situation in Shanghai. 103 Planning reserves represent the capacity used to ensure the sufficient energy supply to meet peak demand and the required operating reserves.

0

100

200

300

400

500

600

2020 2025 2030 2020 2025 2030 2020 2025 2030


Mill

ion

RM

B p

er y

ear

CT CCGT Low CCGT High

0 200 400 600 800

1000 1200 1400 1600 1800

2020 2025 2030 2020 2025 2030 2020 2025 2030


Mill

ion

RM

B p

er y

ear




Range of other avoided costs (RMB/kWh)

Other avoided costs104 are determined by the potential reduction in kWh of generation during the peak hours. The hours of peak demand are assumed to be the same as in 2013 (i.e. 4.2% of the 8,760 hours in one year). Thus, the avoided costs are estimated by multiplying the potential reduction in kWh during the peak by the associated costs per kWh (e.g. energy costs, CO2 emission costs and T&D losses, see Table 21 above).

Figure 23 shows that the avoided energy cost could reach between 89.5 and 226 million RMB in 2030 under the top performing scenario, depending on the use of different types of gas fired and coal-fired plants105. The avoided energy costs for CT plant are much higher than other generating plants because CT plant has the highest fuel consumption per unit of power output (See Table 22) and the cost of gas is higher than the cost of coal for the same amount of thermal output (See Table 21). As for the avoided CO2 cost in Figure 24, the annual savings could reach between 6.1 and 14.8 million RMB in 2030, if the CO2 price is 40 RMB per tonne. Besides avoided energy and CO2 emission costs, there are potential savings associated with T&D losses. Figure 25 presents the avoided T&D costs (given the average T&D losses of 6.68% in Shanghai). As the amount of T&D losses (in kWh) is identical for different types of power plant, the higher cost per unit of power output will lead to higher value for T&D losses. For that reason, CT plant has higher avoided T&D costs than other plants.

Taking into account avoided energy costs, avoided CO2 emission costs and avoided T&D losses, the range of other avoided costs can be found in Figure 26. The range reflects the use of different coal power plants in the estimation, but not the use of coal with different calorific value. It shows that the other avoided costs could range between 108.5 and 251.3 million RMB in 2030 under the ‘top performing’ scenario, but be significantly lower in the other two scenarios.

Figure 23 Avoided energy cost

104 Note that our estimation of other avoided costs refers to load shedding. If DR were load shifting, the estimated avoided costs would be less since some or all those costs would be incurred off-peak. 105 We have listed the avoided energy costs from both the coal-fired plants and the gas fired-plant as we do not have the information about the marginal resource in Shanghai. The operating hours for all different types of plant are assumed to be identical.

0.0 50.0

100.0 150.0 200.0 250.0 300.0 350.0 400.0

2020 2025 2030 2020 2025 2030 2020 2025 2030

Basic Moderate Top performing Mill

ion

RM

B p

er y

ear


0

50

100

150

200

250

2020 2025 2030 2020 2025 2030 2020 2025 2030


Mill

ion

RM

B p

er y

ear

SPC USC SBC CT CCGT


Figure 24 Avoided CO2 cost

Figure 25 Avoided T&D cost

Figure 26 Range of all other avoided costs (energy, CO2 and T&D)

Total avoided costs

Adding up all the avoided costs (long term capacity costs and short term other costs), Figure 27 presents the range of estimated annual savings due to DR in 2020, 2025 and 2030. It suggests that total avoided costs could reach 811.2 million RMB in 2030, when discounted at 7%. Avoided generation capacity costs contribute most to the total avoided costs (its share ranges between 68.3% and 79.7%), and avoided energy costs come second (between 17.8% and 28.1% of total avoided costs). Avoided T&D costs and avoided CO2 costs together account for between 2.5 % and 3.8% of the total avoided costs.

0 2 4 6 8

10 12 14 16

2020 2025 2030 2020 2025 2030 2020 2025 2030


Mill

ion

RM

B p

er y

ear

SPC USC SBC CT CCGT

0 2 4 6 8

10 12 14 16 18

2020 2025 2030 2020 2025 2030 2020 2025 2030


Mill

ion

RM

B p

er y

ear

SPC USC SBC CT CCGT

0

50

100

150

200

250

300

2020 2025 2030 2020 2025 2030 2020 2025 2030


Mill

ion

RM

B p

er y

ear

SPC USC SBC CT CCGT


Figure 27 Total avoided costs between 2020 and 2030

There are a number of caveats. On the one hand, the assessment only considers dispatchable DR (i.e. incentive-based programmes). Furthermore, due to inadequate information, we have not attempted to estimate the avoided costs of network expansion. For these reasons, our results may under-estimate DR market potential and value. On the other hand, there are reasons why our analysis may overstate the value of DR. First, since the valuation assumes that all market potential is achieved from load shedding only, the avoided costs106 may be lower if part of the potential DR is actually achieved from load shifting or on-site generation respectively. Second, the practice of administrative demand planning has suppressed demand. This may well reduce the ‘real’ economic benefits of DR since capacity may be less than the optimal level to meet peak demand. Third given the lack of detailed projections of peak demand and electricity use (e.g. for 2020-2030), this study made very simple estimates, which may not accurately reflect the future trend of peak demand and electricity use. Finally, we have not estimated the costs of introducing DR.

3.5 Recommendations for further research

Rigorous assessments for the potential and value of DR resources are central to the regulatory efforts or policy decision-making for promoting DR programmes, and the industry processes for system operation and planning. Besides the general need for improved data accessibility, this study identifies a number of areas where more work needs to be done so as to refine the potential and valuation analyses in future research.

Strengthen load profile research. One key challenge of this study is the lack of typical load profiles for different customer segments. Future research will benefit from more rigorous analysis of customer load profiles, especially covering a large sample and maintaining long metering duration. This will help researchers in identifying key patterns in electricity use, and thus refining the customer segmentation. Since AMI has a very high penetration in Shanghai, there are great opportunities to leverage it to improve the understanding of customer load profiles. The load profiles of various end-uses (e.g. lighting, AC, plug loads and refrigerator) will also facilitate the analysis of customer potentials to deliver DR.

Develop a more robust evidence base for participation rate and DR load impact. Many factors may influence the acceptance and capability of customers in delivering DR, including the flexibility potential of business activities and their electricity use, availability of enabling technologies and their functionality, cost-benefit case for participating in DR programmes and other solutions (e.g. on-site generation or fuel switching). As the DR pilot develops, more locally specific evidence or data will emerge to show how these factors may differ

106 From a system perspective

0

100

200

300

400

500

600

700

800

900

low high low high low high low high low high low high low high low high low high

2020 2025 2030 2020 2025 2030 2020 2025 2030


Mill

ion

RM

B p

er y

ear

Avoided capacity costs Avoided energy costs Avoided CO2 costs Avoided T&D costs


amongst various customer segments. This will inform future research in considering the market potential for different customer segments. To achieve this goal, comprehensive and regular evaluation of DR programmes is necessary.

Analyse empirically the results of the 2015 summer DR pilot programme in Shanghai. The data from this pilot programme could be very useful in identifying the impact of DR on different customer categories and specific end uses. This information could be studied in detail to guide the design of DR programmes in Shanghai and other cities.

Undertake analysis on the relative potential of different DR strategies. There is value from future research to consider the strategies customers are likely to take to deliver peak demand reduction (e.g. load shedding, load shifting or use of on-site generation) in Shanghai. Such information will support the valuation analysis in terms of making informed assumptions about avoided energy and CO2 costs and system-level avoided capacity costs. This involves better understanding of the characteristics of business activities and their electricity use, the role of enabling technologies in supporting different DR strategies, and the cost-benefit considerations for customers.

Detailed assessment at the end-use level and for other DR types. Due to data limitations, this study could not undertake an assessment of the potential for different end-use categories (e.g. air conditioning, industrial process or refrigeration) or other DR types (e.g. non-dispatchable price-based DR). Future research will benefit from the availability of end-use-level granular load profiles for different customer segments, and more evidence showing the price elasticity of customers regarding their electricity use.

Further define DR products. With the role of DR in system operation and planning becoming more important, it is useful to assess the potential and value of individual DR products, thus offering more a detailed gauge of the DR potential and value for system planning and operation. However, this may add to the need for evidence or data to indicate how different customers are likely to participate and deliver response for various DR products. In other words, this requires a more robust understanding of customers as noted above.

Develop long-term peak demand forecast. The largest benefits of DR are most likely to be the avoided costs in generation and network capacity over the long term and DR programmes typically have a ‘ramping-up’ period before certain levels of participation rate or load impact can be achieved. For these reasons, it is worthwhile to consider the DR potential and valuation over the medium or long term. This requires extending the timeframe for peak demand forecast to a longer term, by taking into account the likely changes that may be expected to materialise in the electricity system and influence peak demand.

Develop a customer engagement strategy. As indicated above, sustained customer education and outreach are essential for the long-term development of DR programmes, and there needs to be provision for them in any comprehensive DR plan. A review of international experience in customer engagement would be a useful initial contribution to this.

Develop better information on system costs in order to better estimate system benefits. More accurate measures of avoided cost require more information on which generation plant will be affected by DR and on the specific incremental cost of those plants. In the longer term, in a market-based system or one where prices reflect marginal costs and dispatch is based on marginal cost, the system’s avoided costs could be calculated by forecasting the hour by hour avoided system electricity costs. Furthermore, it is important to determine whether and to what extent DR would contribute to the avoidance of network capacity expansion.

Develop estimate of DR programme cost. Assessing the cost of DR programmes is essential for understanding the true value they can bring to the system operation and planning. It allows us to estimate the net value of DR programmes and make informed decisions as to how programmes should be designed (e.g. how to reduce programme cost) or whether particular programmes should be launched at all (e.g. cost-effectiveness of DR).

Consider adding other environmental externalities. The current estimates include CO2 as potentially important avoided cost. Future research might include other avoided environmental externalities, including PM10 emissions.

Assessment of Demand Response Potential and Benefits in Shanghai References 91

REFERENCES

Cappers, P., Goldman, C., Kathan, D., 2010. Demand response in U.S. electricity markets: Empirical evidence. Energy 35, 1526-1535.

Chen, W., Yue, L., Cui, K., Li, J., 2011. Analysis on Power Load and Consumption Characteristics of Representative Consumers. Energy Technology and Economics 23.

Conchado, A., Linares, P., 2012. The Economic Impact of Demand-Response Programs on Power Systems. A Survey of the State of the Art, in: Sorokin, A., Rebennack, S., Pardalos, P.M., Iliadis, N.A., Pereira, M.V.F. (Eds.), Handbook of Networks in Power Systems I. Springer Berlin Heidelberg, pp. 281-301.

Darby, S.J., McKenna, E., 2012. Social implications of residential demand response in cool temperate climates. Energy Policy 49, 759-769.

DECC, 2013. Impact Assessment (IA) for Electricity Demand Reduction - Amendment to Capacity Market Clauses.

DECC, 2014a. Impact Assessment (IA) for Smart meter roll-out for the domestic and small and medium non-domestic sectors (GB). Department of Energy and Climate Change.

DECC, 2014b. Impact Assessment of Electricity Market Reform - Capacity Market. Department of Energy and Climate Change.

DECC, 2014c. Implementing Electricity Market Reform (EMR) - Finalised policy positions for implementation of EMR.

ECCC, 2015. Implementation of Electricity Market Reform: Eighth Report of Session 2014-15. The Energy and Climate Change Committee, House of Commons.

EIA, 2011. About 60% of the U.S. electric power supply is managed by RTOs. U.S. Energy Information Administration.

EIA, 2013. Electric Power Annual 2011.

Element Energy, 2012. Demand Side Response in the Non-Domestic Sector. Prepared for Ofgem.

Element Energy and RedPoint, 2012. Electricity System Analysis - future system benefits from selected DSR scenarios. Department of Energy & Climate Change.

Energy and Environmental Economics, 2011. Energy Efficiency Avoided Costs 2011 Update. California Public Utilities Commission.

Faruqui, A., Hledik, R., Lineweber, D., 2014. Demand Response Market Potential in Xcel Energy's Northern States Power Service Territory. The Brattle Group and YouGov America. Prepared for Xcel Energy.

Faruqui, A., Sergici, S., Sharif, A., 2010. The impact of informational feedback on energy consumption—A survey of the experimental evidence. Energy 35, 1598-1608.

FERC, 2006. Assessment of Demand Response and Advanced Metering Staff Report.

FERC, 2008. Assessment of Demand Response and Advanced Metering Staff Report.

FERC, 2009. A National Assessment of Demand Response Potential. Federal Energy Regulatory Commission.

FERC, 2011a. Assessment of Demand Response and Advanced Metering Staff Report - February 2011. Federal Energy Regulatory Commission.

FERC, 2011b. Assessment of Demand Response and Advanced Metering Staff Report - November 2011. Federal Energy Regulatory Commission.

FERC, 2012. Assessment of Demand Response & Advanced Metering Staff Report - December 2012. Federal Energy Regulatory Commission.

FERC, 2014. Assessment of Demand Response and Advanced Metering Staff Report - December 2014. Federal Energy Regulatory Commission.

Fox-Penner, P., 2014. Smart Power: Climate Change, the Smart Grid and the Future of Electric Utilities. Island Press.

Goldman, C., Hopper, N., Bharvirkar, R., Neenan, B., Boisvert, R., Cappers, P., Pratt, D., Butkins, K., 2005. Customer Strategies for Responding to Day-Ahead Market Hourly Electricity Pricing. Ernest Orlando Lawrence Berkeley National Laboratory.

92 Assessment of Demand Response Potential and Benefits in Shanghai References

Goldman, C., Hopper, N., Bharvirkar, R., Neenan, B., Cappers, P., 2007. Estimating Demand Response Market Potential among Large Commercial and Industrial Customers: A scoping study. Ernest Orlando Lawrence Berkeley National Laboratory.

Hedman, B., Steiner, J., 2013. DSM in the Rate Case: A regulatory model for resource parity between supply and demand. Public Utilities Fortnightly January 2013.

Hesmondhalgh, S., Owen, G., Pooley, M., Ward, J., 2014. GB Electricity Demand - 2012 and 2025: Impacts of demand reduction and demand shifting on wholesale prices and carbon emissions. Results of Brattle modelling., in: Sustainability First (Ed.), GB Electricity Demand Project - realising the resource.

Hurley, D., Peterson, P., Whited, M., 2013. Demand Response as a Power System Resource: Program Designs, Performance, and Lessons Learned in the United States. The Regulatory Assistance Project (RAP).

Jessoe, K., Rapson, D., 2012. Knowledge is (less) power: experimental evidence from residential energy use. NBER Working Paper No. 18344.

Kahrl, F., Wang, X., 2014. Integrating Renewables into Power Systems in China: A Technical Primer - Power System Operations. Regulatory Assistance Project, Beijing, China.

MacLennan, M., 2014. CLNR Industrial & Commercial DSR Trials. Customer-Led Network Revolution.

Monitoring Analytics, 2015. 2014 State of the Market Report for PJM.

National Grid, 2013a. Fast Reserve Service Description 2013.

National Grid, 2013b. Short Term Operating Reserve (STOR) - Frequently Asked Questions.

National Grid, 2014. Final Auction Results: T-4 Capacity Market Auction 2014.

National Grid, 2015. Firm Frequency Response - Frequently Asked Questions.

NERC, 2013. 2011 Demand Response Availability Report. North American Electric Reliability Corporation.

Ofgem, 2013. Creating the right environment for demand-side response.

PJM, 2013. 2012 Load Response Activity Report: April 2013. PJM Demand Side Response.

PJM, 2014. 2013 Demand Response Operations Markets Activity Report: April 2014. PJM Demand Side Response Operations.

PJM, 2015. 2014 Demand Response Operations Markets Activity Report: March 2015. PJM Demand Side Response Operations.

SEDC, 2014. Mapping Demand Response in Europe Today: Tracking Compliance with Article 15.8 of the Energy Efficiency Directive. Smart Energy Demand Coalition.

SmartGrid GB, 2013. Energy Bill 2013: Demand Side Response and the Capacity Market in focus. SmartGrid GB.

The Brattle Group, 2007. Quantifying Demand Response Benefits in PJM. PJM Internconnection, LLC and the Mid-Atlantic Distributed Resources Initiative (MADRI), Cambridge, MA.

The Brattle Group, 2014. Cost of New Entry Estimates for Combustion Turbine and Combined Cycle Plants in PJM with June 1, 2018 Online Date. Prepared for PJM Interconnection, L.L.C.

Torriti, J., Hassan, M.G., Leach, M., 2010. Demand response experience in Europe: Policies, programmes and implementation. Energy 35, 1575-1583.

UKPN, 2014. Industrial and Commercial Demand Response for Outage Management and as an Alternative to Network Reinforcement. UK Power Networks.

VaasaETT, 2010. Empower Demand: The potential of smart meter enabled programs to increase energy and system efficiency: a mass pilot comparison, http://www.cse.org.uk/pdf/vaasa_ett_2011_potential_of_smart%20meter_enabled_programs_etc.pdf

VaasaETT, 2011. Empower Demand 2: Energy Efficiency through Information and Communication Technology - Best Practice Examples and Guidance, http://esmig.eu/sites/default/files/final_empower_2_demand_report_final_distr2.pdf.

Wade, J., Eyre, N., 2015. Energy Efficiency Evaluation: the evidence for energy savings from energy efficiency programmes in the household sector. UK Energy Research Centre.

Ward, J., Pooley, M., Owen, G., 2012. What Demand-Side Services can Provide Value to the Electricity Sector?, in: Sustainability First (Ed.), GB Electricity Demand- realising the resource.

Woolf, T., Malone, E., Schwartz, L., Shenot, J., 2013. A Framework for Evaluating the Cost-Effectiveness of Demand Reponse, National Action Plan on Demand Response. U.S. Department of Energy and Federal Energy Regulatory Commissio

Assessment of Demand Response Potential and Benefits in Shanghai Appendix 93

APPENDIX Participation rates and DR load impacts in international programmes and potential studies

Programme Sector Participation Rate (% of accounts) DR Impact (kW or % of load reduction) Direct Load Control

FERC 2012 Survey on Demand Response and Advanced Metering

Residential 0-72% (1st Quartile: 0.1% 2nd Quartile: 4% 3rd Quartile 9%) 1st Quartile: 0.7kW 2nd Quartile 1.2kW 3rd Quartile: 3.6kW

Small C&I (<20kW) 0-24% (1st Quartile: 0.2% 2nd Quartile: 0.4% 3rd Quartile: 1.1%) 1st Quartile: 1kW 2nd Quartile: 1.6kW 3rd Quartile: 4.5kW

Medium C&I (20-200kW) 1st Quartile: 1.3% 2nd Quartile: 2.4% 3rd Quartile: 5.6% 1st Quartile: 5kW 2nd Quartile: 23kW 3rd Quartile: 57kW

Large C&I (>200kW) 1st Quartile: 2.4% 2nd Quartile: 4.2% 3rd Quartile: 8.7%

1st Quartile: 1.3kW 2nd Quartile: 250kW 3rd Quartile: 345kW

FERC 2009 National Assessment of Demand Response Potential

Residential 25% (3rd Quartile) 19-52% If data is missing from the FERC 2008 Survey on Demand Response and Advanced Metering, 1kW is assumed.

Small C&I (<20kW) 1% (3rd Quartile) 7-17%

Medium C&I (20-200kW) 7% (3rd Quartile) 2-5%

Assessment of Demand Response and Energy Efficiency Potential - Eastern Interconnection, 2010

Residential 15-25%

Assumptions of FERC 2009 National Assessment of Demand Response Potential

C&I 2.5-5%

Assessment of Portland General Electric's Demand Response Potential, 2012

Residential AC 20% Water Heating 30%

Small C&I (<30kW) 20%

Medium and Large C&I 18%

Comprehensive Assessment of Demand-Side Resource Potentials (2014-2033) for Puget Sound Energy, 2013

Residential Central Heating 20% Water Heating 20% Central Heating 1.74kW Water Heating 0.58kW

Tennessee Valley Authority Potential Study: Volume 3 Demand Response Potential Study, 2011

Residential 11-21% AC - 1kW Water heating - 1kW

Small C&I 7.5-8.5% AC - 1kW Water heating - 1kW

Brattle Energy Efficiency and Demand Response in 2020: A survey of expert opinion, 2011

Residential 10-15% (5-10%, 5-30% and 10-30% depending on region)

C&I 5-15% (2-10% and 8.5-32.5% depending on region)

Potential for EE, DR and Onsite Solar Energy in Pennsylvania, 2009

Residential 15-35% 0.6-1kW

Load Impact Evaluation for PG&E Smart AC Programme, 2013

Residential 3% 0.41kW (or 16-18% of average household load)


FERC 2012 Survey on Demand Response and Advanced Metering

Interruptible tariffs Small C&I (<20kW) 0-0.3% (average 0.1%) 67%-100% (only 4 programmes)

Medium C&I (20-200kW) 0-30.6% (1st Quartile: 0.1% 2nd Quartile: 0.3% 3rd Quartile: 1.3%) Average 82% (17%-100%, only 22 programmes)

Large C&I (>200kW) 0-33.4% (1st Quartile: 0.7% 2nd Quartile: 1.9% 3rd Quartile: 5.5%) 1-100% (1st Quartile 50% and 2nd Quartile 75%)

Other incentive-based DR Small C&I (<20kW) n/a n/a

94 Assessment of Demand Response Potential and Benefits in Shanghai Appendix

Programme Sector Participation Rate (% of accounts) DR Impact (kW or % of load reduction)

Medium C&I (20-200kW) 0.2%-23% (1st Quartile: 0.4% 2nd Quartile: 1% 3rd Quartile: 13.8%)

1-100% (1st Quartile: 19% 2nd Quartile: 25% 3rd Quartile: 38%)

Large C&I (>200kW) 0-44% (1st Quartile: 0.3% 2nd Quartile: 1% 3rd Quartile: 3.4%)

5-100% (1st Quartile: 22% 2nd Quartile: 38% 3rd Quartile: 98%)

FERC 2009 National Assessment of Demand Response Potential

Interruptible tariffs Small C&I (<20kW) n/a n/a

Medium C&I (20-200kW) 2% (3rd Quartile) 27-100% Large C&I (>200kW) 17% (3rd Quartile) 13-100%

Other incentive-based DR Small C&I (<20kW) n/a n/a Medium C&I (20kW-200kW) 0% (75th Percentile) 39%-100%

Large C&I (>200kW) 19% (75th Percentile) 10-100% 2013 California Statewide Base Interruptible Programme

Medium and large C&I 3% 84%

2012 California Statewide Base Interruptible Programme

Medium and large C&I 3% Average 75%

2011 California Statewide Base Interruptible Programme

Medium and large C&I 3% Average 63%


C&I 0.10% Assumptions of FERC 2009 National Assessment of Demand Response Potential

Assessment of Portland General Electric's Demand Response Potential, 2012

Large C&I (>200kW) 17%

Comprehensive Assessment of Demand-Side Resource Potentials (2014-2033) for Puget Sound Energy, 2013

Medium and large C&I 20% 27-34%

Potential for EE, DR and Onsite Solar Energy in Pennsylvania, 2009

Commercial 10-30% (% of aggregate load in relevant sector) 15-25%

Industrial 20-40% (% of aggregate load in relevant sector) 20-40%

Brattle Energy Efficiency and Demand Response in 2020: A survey of expert opinion, 2011

C&I 3-15% (3-4% and 40-60% depending on region)


C&I 0.14-0.23% Assumptions of 2009 FERC National Assessment of DR Potential

Tennessee Valley Authority Potential Study: Volume 3 Demand Response Potential Study, 2011

Small C&I 3-7.5% 5-30%

Medium C&I 3-7.5% 12-40%

Large C&I 3-19% 39-40%

Brattle Energy Efficiency and Demand Response in 2020, 2011

C&I 4-15% (2-9% and 17.5-55%)

Assessment of Demand Response Potential and Benefits in Shanghai Appendix 95

Programme Sector Participation Rate (% of accounts) DR Impact (kW or % of load reduction) 2013 California Statewide Demand Bidding Programme

C&I 9% Average 8% (4-39.4%)

2012 California Statewide Demand Bidding Programme

C&I 10% Average 7% (4.6-8.1%)

2011 California Statewide Demand Bidding Programme

C&I 10% Average 7% (4.6-8.1%)

2013 California Statewide Aggregator Demand Response Programme

Medium and Large C&I 0.30% Capacity Bidding Average: 19% (14-31%) Aggregator Managed Portfolio: 27% (9-31%)


Medium and Large C&I 0.40% Capacity Bidding Average: 23% (3.3-43.6%) Aggregator Managed Portfolio: 30% (27.7-39%) DR Resource Contract Programme: 31% (29.1-65.8%)


Medium and Large C&I 0.30% Capacity Bidding: Average 26% (16-44%) Aggregator Managed Portfolio: Average 30% (27-34%) DR Resource Contract Programme: Average 27%

Contact us Natural Resources Defense Council Taikang Financial Tower, Suite 1706 No. 38 Dongsanhuan Bei Road, Chaoyang District Beijing, China 100026 Tel: +86-10-5927 0688 www.nrdc.cn

Assessment of Demand Response Market Potential and Benefits … of Demand... · July 2015...

Documents

Transcript of Assessment of Demand Response Market Potential and Benefits … of Demand... · July 2015...