Demand Response, Behind-the-Meter Generation and Air QualityXiyue Zhang† and K. Max Zhang*,‡

†Cornell Institute for Public Affairs, Cornell University, Ithaca, New York 14853, United States‡Sibley School of Mechanical and Aerospace Engineering, Cornell University, Ithaca, New York 14853, United States

ABSTRACT: We investigated the implications of behind-the-meter (BTM) generation participating in demand response(DR) programs. Specifically, we evaluated the impacts of NOxemissions from BTM generators enrolled in the New YorkIndependent System Operator (NYISO)’s reliability-based DRprograms. Through analyzing the DR program enrollmentdata, DR event records, ozone air quality monitoring data, andemission characteristics of the generators, we found that theemissions from BTM generators very likely contribute toexceedingly high ozone concentrations in the NortheastCorridor region, and very likely account for a substantialfraction of total NOx emissions from electricity generation. Inaddition, a companion study showed that the emissions fromBTM generators could also form near-source particulate matter (PM) hotspots. The important policy implications are that theabsence of up-to-date regulations on BTM generators may offset the current efforts to reduce the emissions from peaking powerplants, and that there is a need to quantify the environmental impacts of DR programs in designing sound policies related todemand-side resources. Furthermore, we proposed the concept of “Green” DR resources, referring to those that not only providepower systems reliability services, but also have verifiable environmental benefits or minimal negative environmental impacts.We argue that Green DR resources that are able to maintain resource adequacy and reduce emissions at the same time are key toachieving the cobenefits of power system reliability and protecting public health during periods with peak electricity demand.

■ INTRODUCTION

High electric demand days (HEDD) are typically hazy, hot andhumid summer days, and often triggered by regional heat wavesand exacerbated by urban heat island. With a changing climate,the occurrence of HEDDs is becoming increasingly frequent,1

posing tremendous challenges to maintaining power systemsand incurring high energy costs for both the system and thecustomers. Furthermore, HEDDs are typically most conduciveto air pollution formation. As a result, peak electricity demand(and the resultant emissions) generally corresponds todays when the potential for poor air quality, both locally andregionally, is greatest. Together, high temperature and highpollution levels during HEDDs are a serious threat to publichealth in the metropolitan areas, where the population densitiesare the highest. We refer to the challenges facing the energysystems and public health during HEDDs as a “peak” problem.Therefore, the key question to solve this “peak” problem is howto maintain power systems reliability and protect public healthat the same time.Peak demand often requires the system operators to call

upon all available resources to maintain a reliable power system.As all available supply side resources on the system are usuallycommitted and dispatched near their maximum operatingcapacities, system operators resort to demand response (DR)programs to reduce demand during critical hours, which typi-cally happen on only a few days in a year, but have becomeimportant reliability resources.

Participants enrolled in DR programs can respond via twogeneral mechanisms, that is, load curtailment (i.e., loadshedding and/or shifting) and behind-the-meter (BTM)generation. BTM generation refers to a generator locatedbehind the retail delivery point that can directly serve the hostcustomer’s electrical demand in lieu of or in addition toelectricity the customer takes through the transmission grid.2

BTM generators enrolled in the DR programs are almost ex-clusively diesel-fueled stationary reciprocating internal combustionengines (RICE) for backup purposes (often referred to as dieselbackup generators),3 which often have high emission factors of airpollutants due to lack of up-to-date emission control technologiescompared to natural gas-fired combustion turbines. Althoughsome studies suggest that using BTM generation to meet peakdemand is cost-effective,4 the question has been raised overwhether it can result in significant environmental impacts.5,6

The overall objective of our study is to evaluate the environ-mental impact of BTM generation participating in DRprograms. For this paper, we focus on two important aspectsin addressing the objective, described as follows.First, in order to reduce the number of assumptions we make

in our analysis, we focus on reliability-based DR programs by

Received: October 13, 2014Revised: January 1, 2015Accepted: January 4, 2015Published: January 5, 2015

Policy Analysis

pubs.acs.org/est

© 2015 American Chemical Society 1260 DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

New York Independent System Operator (NYISO), whichmakes detailed market data publicly available.Second, the main environmental impact to be addressed in

this paper is regional ozone air pollution. A companion paperexamines another environmental impact, that is, near-sourceparticulate matter (PM) pollution.7 Exceedingly high ozoneconcentrations often occur during hot summer days, whenreliability-based DR programs are likely to be initiated andBTM generators operate to respond. The New York Metro-politan Area is an 8 h ozone nonattainment area.8 We focus onNOx emissions from BTM generation and their contributionsto ozone pollution. Through analyzing energy and environ-mental data, all publicly available, from 2010 to 2013, we aim toanswer the following two questions. First, do NOx emissionsfrom BTM generators contribute to potential ozone exceed-ance? Second, do BTM generators emit significant amount ofNOx emissions? The implications on future DR programs arediscussed and the concept of “Green DR” proposed.

■ DATA AND METHODS

DR Programs and NYISO Market Data. DR programs,offered by Independent System Operators (ISO)/RegionalTransmission Organizations (RTO) and local utilities compa-nies, can be either incentive-based (IBP) or price-based (PBP).IBPs are designed to induce load reduction and/or maintain anoperating reserve in case of critical conditions for the sake ofpower system reliability, by offering incentive payments forenrollment or performance. Examples of IBPs include classicaldirect load control and interruptible/curtailable load programs,market-based demand bidding, reliability-based (or emergency)DR programs, capacity market, and the ancillary services market.PBPs, on the other hand, are based on dynamic pricing rates andaim at flattening the demand curve by offering a higher priceduring peak periods. Examples of PBPs include the time of userate, critical peaking pricing (CPP), extreme day CPP, extremeday pricing, and real time pricing.9

Figure 1 summarizes the NYISO DR programs, currentlyconsisting of the Emergency Demand Response Program(EDRP), the ICAP Special Case Resources (SCR) program,the Day Ahead Demand Response Program (DADRP), and theDemand Side Ancillary Services Program (DSASP). All fourprograms are IBPs, among which EDRP and SCR, that is,the focus of our study, require voluntary or mandatory loadreductions from participants in critical conditions for power

system reliability, also referred to as “Reliability-Based”. DADRPand DSASP provide retail customers with an opportunity to bidtheir load reductions into the wholesale markets as an operatingreserve, and offers determined to be economic are paid at themarket clearing price, also referred to as “Market-Based”.Typically industrial and commercial customers sign up to

take part in NYISO EDRP and/or SCR programs.10 The partic-ipants are paid by NYISO for reducing power consumptionwhen asked to do so, and their responses are typically manda-tory. Both programs are designed to reduce power usagethrough shutting down of businesses and large power users.11

The details of those programs are summarized in NYISO’sannual and semiannual reports to the Federal EnergyRegulatory Commission (FERC).As there is no mandatory reporting requirement, it is difficult

to determine the exact contributions of BTM generation to thetotal load reduction during DR events. NYISO has an optionfor the enrolled participants in the SCR/EDRP programsto report voluntarily the capacity of their BTM generators.Table 1 summarizes the responses from SCR/EDRP enroll-

ment data, which reveal that, out of the total SCR/EDRP MW,on average 11% was provided by enrolled BTM generatorsfrom 2010 to 2014.12−16 It should be noted that the loadreduction supported by BTM supply sources is greater than thereported. We selected 10% as the lower boundary for BTMgeneration’s contributions to total DR load reduction, and 50%as the upper boundary in constructing our scenarios.

Load, Emission and Air Quality Data. Region of Interest.Ozone Transport Commission (OTC) counties adjacent to theNortheast I-95 Corridor were selected for discussion.17 Thisregion, referred to as the “NE Corridor” for the rest of the paper,is not only heavily populated with high electricity demand, butalso is an 8 h ozone nonattainment area.8 States/districts that

Figure 1. NYISO demand response programs.

Table 1. Percentage of NYISO SCR/EDRP MW Provided byBTM Generators12−16

2010 2011 2012 2013 2014

SCR+EDRP MW ofenrolled localgenerators

304.8 247.2 162.0 127.0 124.7

total SCR+EDRP MW 2236.4 1917.3 1805.6 1239.9 1158.3percentage 14% 13% 9% 10% 11%average 11%

Environmental Science & Technology Policy Analysis

DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

1261

situate within the region include New Jersey, New York City(NYC), Long Island (LI), Connecticut, Rhode Island, andsoutheast Massachusetts,18 as encircled in the EPA 8 h OzoneNonattainment Areas map (Figure 2). These counties fall into

the control areas of three ISO/RTO, that is, NYISO, ISO NewEngland (ISO-NE) and PJM Interconnection (PJM). Our airquality analysis is based on the NE Corridor region, while theemission analysis is based on NYC and LI as well as New YorkState, taking advantage of the DR program data available atNYISO.Time Periods. The Environmental Protection Agency (EPA)

defines ozone season in the U.S. as May 1 to September 30,19

during which ground-level ozone increases significantly. Ashourly real-time load data is not available from ISO-NE priorto 2008, and electric power consumption in 2009 is notrepresentative of the nation’s normal consumption level dueto the economic recession, we select the time periods underdiscussion as May 1 to September 30 from 2010−2013.Data Sources. Hourly real-time load data were obtained

from NYISO, ISO-NE, and PJM. The data are synchronized ateach hour to find the peak daily load over the NE Corridorregion. HEDD records, according to its definition, are derivedfrom PJM’s historical load forecasts.20,21 Hourly NOx emissionsfrom electric generating units (EGUs) are obtained from EPAAir Markets Program Data,22 and are summed up across the NECorridor region. The 8 h average ozone concentrations areobtained from EPA AirData.23

Emission Factors for BTM Generators. Three emissionfactors (EF) for the BTM generators are adopted in ouranalysis, listed in Table 2. “DEC EF” is derived based on theanalysis on diesel backup emergency generators in NYC con-ducted by the New York State Department of Environmental

Conservation (NYSDEC).24 “ConEd EF” is based on theestimated BTM generator mix in New York City by Con-solidated Edison, Inc.25 In addition, we include “Tier4 EF”, thatis, assuming that BTM generators are all equipped with the bestavailable control technology and meet EPA Tier 4 standards fornonroad diesel engines.26−29 In other words, The DEC EF andConED EF are more realistic representations of current BTMemission characteristics, while Tier4 EF represents a futuristicscenario. We apply those emission factors to NYC and LI andalso the entire NYISO system.

Scenarios. Based on the two BTM penetration levels, thatis, 10% and 50%, into NYISO SCR/EDRP programs and threeemission factors, we construct six different scenarios as de-scribed in Table 3. Then we quantify the contributions of BTM

generation to both electricity load and NOx emissions for eachNYISO SCR/EDRP events day from 2011−2013 (and 2010market data are not available).

■ RESULTS AND DISCUSSIONSContribution of BTM NOx Emissions to High Ozone

Concentrations. Figure 3 illustrates the relationship betweendaily peak load and daily NOx emissions from EGUs, bothsynchronized across NYISO, PJM, and ISO-NE for the NECorridor region. The median values of 8 h average ozone con-centration among all monitoring sites across the NE Corridorare also shown in Figure 3 with the color scale upward fromblue to red capped at 75 ppb, the current limit for 8 h averageozone concentration in the National Ambient Air Quality Stan-dards (NAAQS). Furthermore, the corresponding ozone con-centrations during NYISO SCR/EDRP events are encircled inFigure 3.Figure 3 indicates that the highest peak loads, for example,

over 45 GW, always correspond to the ozone concentrationsexceeding the ozone NAAQS. Those are the periods whenSCR/EDRP events are most likely to be called and BTMgenerators are turned on to respond. In fact, out of 17 dayswhen SCR/EDRPs are called by NYISO from 2010 to2013,30−32 14 days see ozone concentrations exceeding theNAAQS level in the NE Corridor region.The close correlation between highest electricity demands

and exceedingly high ozone concentrations depicted in Figure 3is consistent with the findings reported by He et al. (2013),33

which suggests that high ambient temperatures in the EasternU.S not only lead to high energy demand and, consequently,high EGU emissions, but also intensify photochemical production

Figure 2. Region of interest: OTC counties adjacent to the NortheastI-95 Corridor and the ozone nonattainment areas.8

Table 2. NOx Emission Factors

name emission factors

DEC EF 16.00 g/kW-hr (11.9 g/bhp-hr)ConEd EF 10.63 g/kW-hr (7.9 g/bhp-hr)Tier4 EF 2.16 g/kW-hr (1.6 g/bhp-hr)

Table 3. Six Scenarios Based on Two BTM GenerationPenetration Levels and Three Emission Factors

BTM penetration levels ×2

NOx emissionfactors ×3

10% BTM generation,2.16 g/kW-hr (Tier4 EF)

50% BTM generation,2.16 g/kW-hr (Tier4 EF)

10% BTM generation,10.63 g/kW-hr (ConEd EF)

50% BTM generation,10.63 g/kW-hr (ConEd EF)

10% BTM generation,16.00 g/kW-hr (DEC EF)

50% BTM generation,16.00 g/kW-hr (DEC EF)

Environmental Science & Technology Policy Analysis

DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

1262

of tropospheric ozone, which is typically controlled by NOx.Therefore, NOx emissions from BTM generators occur whenthe atmospheric condition is most conducive to ozone forma-tion, and very likely contribute to potential ozone exceedance.However, it should be noted that emissions from BTM gener-ators are not the only contributors and that NOx emissionsfrom all sectors (e.g., the transportation sector) are most damaging

during those periods. In contrast with mobile emissions, EGUemissions are very sensitive to temperature in the ozone seasons.

BTM NOx Emissions on NYISO SCR/EDRP Event Days.Comparison to NOx Emission Factors From Peaking Units.As the DR programs have the potential to shift part of thesupply for peak power from conventional peaking units todistributed diesel-fueled BTM generators, we first compare theemission factors of two types of generation units.The maximum annual operating time for combustion

turbines to avoid LAER/BACT requirements for NOx emis-sions in NYC is 66 h,34 in which LAER stands for the lowestachievable emission rates and BACT for the best availablecontrol technology. We adopt the threshold of 66 h to identifythe peaking units located in NYC, and assume that this criterionholds true for LI. NOx emission factors for the identified peakingunits are computed from their annual NOx emissions and grossload obtained from the EPA Air Markets Program Data.Figure 4 illustrates the distribution of NOx emission factors

for peaking units located in NYC and LI from 2010 to 2013.

The bands at the top, bottom and inside of the boxes representthe third quartile (75%), first quartile (25%), and median of thedata, respectively. The bands at the ends of the extendingwhiskers represent minimum and maximum of the data withoutconsidering the outliers. The dots represent the outliers that layapproximately 2.7 standard deviations away from the mean ofthe data.Figure 4 shows that the ConEd EF for BTM generators is

similar to those for the highest emitting peaking units in NYC.Taking the peaking units in LI into consideration, the highestEF is 13.81 g/kWh recorded in 2012, which is lower than theDEC EF. Therefore, simply replacing peaking units with BTMgenerators will likely increases NOx emissions.

BTM NOx Emissions in New York City and Long Island.NYC and LI, home to ∼56.8% population of New York State,reside within the Northeast Corridor region. They are servicedas Zones J and K, respectively, by NYISO. The New YorkMetropolitan Area is also an 8 h ozone nonattainment area.8

No detailed information is publicly available for the fourNYISO SCR/EDRP called events in 2010, so we performedanalysis on the rest 13 days from 2011 to 2013,30−32 sum-marized in Table 4. For each of these NYISO SCR/EDRPevents days, we list the total MW-hr responded by the DR par-ticipants, and estimate the NOx emissions from BTM genera-tion under the six scenarios. We group the NOx emissions bythe three emission factors, and within each group the lower andupper values represent the 10% and 50% penetration levels,respectively. Furthermore, we report the NOx emissions from

Figure 3. Daily NOx emissions from electric generating units vs. peakdaily load for the NE Corridor region for four ozone seasons: (a)May 1, 2010 to September 30, 2010; (b) May 1, 2011 to September30, 2011; (c) May 1, 2012 to September 30, 2012; and (d) May 1,2013 to September 30, 2013. The color scale is capped at 75 ppb, thecurrent limit for 8 h average ozone concentration in the NAAQS.

Figure 4. NOx emission factors for peaking units in New York City(NYC) and Long Island (LI).

Environmental Science & Technology Policy Analysis

DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

1263

the identified peaking units during the same days and estimatethe fraction the total EGU NOx emissions in NYC and LI that

comes from BTM generation (across the six scenarios), whichis defined as

+×

NO emissions from BTM generationNO emissions from BTM generation NO emissions from NYC/LI EGUs

100%X

X X

The summary indicates that NOx emissions from BTMgenerators in NYC/LI vary significantly, from ∼0.2 (May 29,2012) to ∼26 (July 22, 2011) tons/day, depending on thenumber of hours called, assumed emission factors, and BTMgenerators participation. It should be noted that on July 17,2012, the SCR/EDRP was called in Zone B and a subloadpocket in Zone J, and the responses were not mandatory,resulting in very small load reduction. This day is not includedin our analysis. In comparison, NOx emissions from the peakingunits vary from ∼0.3 (July 17, 2013) to ∼23 (July 22, 2011)tons/day, which mostly fall within or even below the range ofNOx emissions from BTM generators on the same day usingthe DEC EF.It is also worthwhile comparing the contributions of BTM

generation to total electricity load during the DR event daysagainst its contributions to total NOx emissions from electricitygeneration. Here we define:

= +total load (for electricity) NYC/LI metered load BTM generation(1)

=

+

total NO emissions NO emissions from (NYC/LI EGUs

BTM generators)X X

(2)

Taking July 22, 2011 (a day with highest response from DRparticipants from 2011 to 2013) as an example, Figure 5 depictsthe two types of contributions, with Figure 5a on electricityload and Figure 5b on NOx emissions (with six scenarios).BTM generators only operate during NYISO SCR/EDRPevents called hours, from 1:00 pm to 6:00 pm. Figure 5 indi-cates that, under the 50% BTM penetration + DEC EFscenario, the BTM generators contribute 15.2% of total NOxemissions over the day, while serving only 1.5% of the totalload. Yet under the Tier4 EF scenario, NOx emissions fromBTM generators have very minor impacts. In other words, if allBTM generators were upgraded to Tier 4 or higher tier models,

their impacts would be significantly reduced from the worsescenarios estimated.

New York State. Assuming that the emission factors can begeneralized to all BTM generators in the entire New YorkState (NYS), we conduct a similar analysis at the state level (thesame as the NYISO system). Equations 1 and 2 still hold exceptthat the scale under discussion expands from NYC and LI to

Table 4. Summary of Emission Analysis for New York City (NYISO Zone J) and Long Island (NYISO Zone K)

NOx emissions from BTMgeneration (tons/day)

dateDR responded

(MW-hr)Tier4EFa

ConEdEFa DEC EFa

fraction of total EGU NOxemissionsb

NOx emissions from peaking units(tons/day)

07/21/2011 2827.7 0.6−3.1 3.0−15.0 4.5−22.6 0.7−21.6% 1.507/22/2011 3245.6 0.7−3.5 3.5−17.3 5.2−26.0 0.5−15.2% 22.705/29/2012 711.6 0.2−0.8 0.8−3.8 1.1−5.7 0.1−5.0% 20.406/20/2012 1339.8 0.3−1.5 1.4−7.1 2.1−10.7 0.3−10.9% 2.506/21/2012 2440.8 0.5−2.6 2.6−13.0 3.9−19.5 0.5−15.7% 1.306/22/2012 2667.6 0.6−2.9 2.8−14.2 4.3−21.3 0.8−22.8% 0.607/18/2012 1959.0 0.4−2.1 2.1−10.4 3.1−15.7 0.4−12.8% 9.807/15/2013 1475.0 0.3−1.6 1.6−7.8 2.4−11.8 0.3−11.0% 6.807/16/2013 1655.2 0.4−1.8 1.8−8.8 2.7−13.2 0.4−11.9% 2.907/17/2013 1546.2 0.3−1.7 1.6−8.2 2.5−12.4 0.4−11.7% 0.307/18/2013 1682.6 0.4−1.8 1.8−8.9 2.7−13.5 0.3−9.6% 10.107/19/2013 1803.3 0.4−2.0 1.9−9.6 2.9−14.4 0.3−9.9% 12.9

aLower boundary of the range determined by the 10% BTM generator participation. Upper boundary determined by the 50% participation. bFor allsix scenarios.

Figure 5. Temporal profiles of (a) load and (b) NOx emissions for theNYC-LI system on July 22, 2011.

Environmental Science & Technology Policy Analysis

DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

1264

NYS. Table 5 follows the similar format as that of Table 4,summarizing the results of our emission analysis for NYS.Figure 6 is the equivalent of Figure 5, but for NYS. TakingJuly 22, 2011 as an example, Figure 6 indicates that under the50% BTM penetration + DEC EF scenario, BTM generationcontributes 20.4% of total NOx emissions over the day, whileserving 1.8% of total load.In March 2007, six OTC states including NYS signed a

nonbinding Memorandum of Understanding (MOU) concern-ing the incorporation of HEDD emission reduction strategiesinto their ozone attainment implementation planning. NYS’commitment to NOx emission reduction on HEDDs, primarilytargeting at peaking units, is 50.8 tons/day,20 which wouldhave been totally offset by the BTM NOx emissions (i.e.,60.3 tons/day) under 50% BTM penetration + DEC EFscenario (e.g., on July 22, 2011). We have verified that all

NYISO SCR/EDRP events were called on HEDDs during theozone season, and, therefore, the DR event daysare a subset of HEDDs. Policies designed to reduce HEDDNOx emissions without properly regulating BTM generatorswill likely become ineffective.

■ IMPLICATIONSDR programs are traditionally perceived as clean resourcesof power systems services. Great advancement has beenaccomplished in load management technologies and peakload reduction. However, our study suggests that, withoutproper emission control, the participation of BTM generationin DR programs may result in significant NOx emissions duringthe periods when atmospheric conditions are most favorable forthe formation of ozone pollution. Shifting load from peakingunits to BTM generators through DR programs during thoseperiods may result in higher overall NOx emissions, becauseBTM generators typically have higher emission factors thanpeaking units as shown in Figure 4.In a companion study, we have conducted computational

fluid dynamics (CFD)-based microenvironmental simulationson the near-source air quality impacts of BTM generation.7

Two PM2.5 emission factors (one corresponding to the samegeneration mix in ConED EF, 0.5 g/kWh, and the othercorresponding to Tier-4 standards, 0.04 g/kWh), two generatorsizes (i.e., 500 kW and 1 MW), typical meteorological condi-tions for DR event days (e.g., unstable atmospheric stability andhigh ambient temperature representative of all DR event daysfrom 2011 to 20137), and a realistic urban site in NYC are usedto construct baseline scenarios. Furthermore, three additionalstreet canyon configurations and two additional meteorologicalconditions are selected for investigation. We show that thenear-source impacts have a strong dependence on urbanconfigurations. The presence of either tall upwind or downwindbuilding can deteriorate the air quality in the near-stack streetcanyon, largely due to the recirculation zones generated by thetall buildings. The near-ground PM2.5 concentration for theworst scenarios could well exceed 100 μg m−3, posing potentialhealth risk to pedestrians and residents in nearby buildings.Even generators that comply with Tier 4 standards could leadto PM hotspots if their stacks are next to tall buildings.7

In summary, the impacts of BTM emissions on regionalozone pollution and near-source PM pollution can potentiallybecome unintended consequences of DR programs. Therefore,

Table 5. Summary of Emission Analysis for New York State (i.e., the NYISO System)

NOx emissions from BTM generation (tons/day)

date DR Responded (MW-hr) Tier4 EFa ConEd EFa DEC EFa fraction of total EGU NOx emissionsb

07/21/2011 3332.7 0.7−3.6 3.5−17.7 5.3−26.7 0.4−12.8%07/22/2011 7537.6 1.6−8.1 8.0−40.1 12.1−60.3 0.7−20.4%05/29/2012 3017.6 0.7−3.3 3.2−16.0 4.8−24.1 0.5−15.2%06/20/2012 1917.2 0.4−2.1 2.0−10.2 3.1−15.3 0.3−10.1%06/21/2012 6335.3 1.4−6.8 6.7−33.7 10.1−50.7 0.8−23.4%06/22/2012 2988.4 0.7−3.2 3.2−15.9 4.8−23.9 0.5−16.3%07/18/2012 2082.5 0.5−2.3 2.2−11.1 3.3−16.7 0.2−7.9%07/15/2013 1716.1 0.4−1.9 1.8−9.1 2.8−13.7 0.2−8.0%07/16/2013 1935.3 0.4−2.1 2.1−10.3 3.1−15.5 0.3−8.6%07/17/2013 1805.3 0.4−2.0 1.9−9.6 2.9−14.4 0.2−7.6%07/18/2013 4822.9 1.0−5.2 5.1−25.6 7.7−38.6 0.5−15.6%07/19/2013 4924.8 1.1−5.3 5.2−26.2 7.9−39.4 0.5−15.7%

aLower boundary of the range determined by the 10% BTM generator participation. Upper boundary determined by the 50% participation. bFor allsix scenarios.

Figure 6. Temporal profiles of (a) load and (b) NOx emissions for theNYISO system on July 22, 2011.

Environmental Science & Technology Policy Analysis

DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

1265

there is a need to quantify the environmental impacts of DRprograms in designing sound policies related to demand-sideresources.We would like to introduce the concept of “Green” DR

resources, defined as those that not only provide power systemsreliability services, but also have verifiable environmentalbenefits or minimal negative environmental impacts. We arguethat Green DR programs are key to solving the “peak” problemdescribed in the Introduction, when almost all EGUs aredispatched to meet the demand, leaving little room for otherstrategies such as environmental dispatch. In essence, Green DRresources hold promise for ensuring resource adequacy andreducing emissions at the same time, thus achieving thecobenefits of power system reliability and protecting publichealth during periods with peak electricity demand.In general, Green DR resources include measures for load

curtailment and clean distributed generation. BTM generatorswith the latest emission control technologies, such as thosemeeting Tier 4 emission standard, and with rigorous sitingdesigns, can potentially become part of Green DR resources.Moving forward, we recommend that Green DR resourcesshould be differentiated from the polluting ones, and incen-tivized for their societal and economic benefits. Furthermore,the participation of low-emitting hydrocarbon-fueled gener-ation in DR programs should not hinder the deployment ofnonemitting generation that can reduce the electricity demandduring the “peak” periods.

■ AUTHOR INFORMATION

Corresponding Author*Phone: 607-254-5402; fax: 607-255-1222; e-mail: kz33@cornell.edu.

Author ContributionsThe manuscript was written through contributions of bothauthors. Both authors have given approval to the final versionof the manuscript.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

This study was supported by Consortium for Electric ReliabilityTechnology Solutions (CERTS) and the New York StateEnergy Research and Development Authority (NYSERDA).We greatly appreciated the discussions with Prof. Tim Mount atCornell University and Prof. Ben Hobbs at Johns HopkinsUniversity and staff at New York Independent System Operator(NYISO), Consolidated Edison, Inc. and the New York StateDepartment of Environmental Conservation (NYSDEC).

■ REFERENCES(1) Rosenzweig, C.; Solecki, W.; DeGaetano, A.; O’Grady, M.;Hassol, S.; Grabhorn, P. Responding to Climate Change in New YorkState: The ClimAID Integrated Assessment for Effective Climate ChangeAdaptation, Synthesis Report; Albany, NY, 2011.(2) U.S. Federal Energy Regulatory Commission. Demand ResponseSupporters v; New York Independent System Operator, Inc.;Washington, DC, 2014; pp. Docket No. EL13−EL74-000.(3) Northeast States for Coordinated Air Use Management. AirQuality, Electricity, and Back-up Stationary Diesel Engines in theNortheast; Boston, MA, 2012.(4) Gilmore, E. A.; Adams, P. J.; Lave, L. B. Using backup generatorsfor meeting peak electricity demand: A sensitivity analysis on emission

controls, location, and health endpoints. J. Air Waste Manage. Assoc.2010, 60, 523−531.(5) Ozone Transport Commission. RE: Proposed Rule - Docket No.EPA-HQ-OAR-2008-0708, 2012.(6) American Lung Association. Comments on the U. S. Environ-mental Protection Agency’s Proposed Revisions to National EmissionsStandard for Hazardous Air Pollutants for Reciprocating InternalCombustion Engines (“ RICE NESHAP ”) Docket ID No. EPA-HQ-OAR-2008-0708, 2012.(7) Tong, Z.; Zhang, K. M. The near-source impacts of diesel backupgenerators in urban environments. Atmos. Environ. accepted.(8) U.S. Environmental Protection Agency. 8 h Ozone Nonattain-ment Areas (2008 Standard). http://www.epa.gov/groundlevelozone/designations/2008standards/final/finalmap.htm.(9) Albadi, M. H. H.; El-Saadany, E. F. A summary of demandresponse in electricity markets. Electr. Power Syst. Res. 2008, 78, 1989−1996.(10) New York Independent System Operator. NYISO DemandResponse Programs http://www.nyiso.com/public/markets_operations/market_data/demand_response/index.jsp (accessed Octo-ber 1, 2014).(11) Hibbard, P. J.; Okie, A. M.; Darling, P. G. Demand response incapacity markets: reliability, dispatch and emission outcomes. Electr. J.2012, 25, 14−24.(12) New York Independent System Operator. Semi-AnnualCompliance Report on Demand Response Programs; Rensselaer, NY,NY, 2010.(13) New York Independent System Operator. Semi-AnnualCompliance Report on Demand Response Programs; Rensselaer, NY,2011.(14) New York Independent System Operator. Semi-AnnualCompliance Report on Demand Response Programs; Rensselaer, NY,2012.(15) New York Independent System Operator. Semi-AnnualCompliance Report on Demand Response Programs; Rensselaer, NY,2013.(16) New York Independent System Operator. Semi-AnnualCompliance Report on Demand Response Programs; Rensselaer, NY,2014.(17) Downs, T.; Fields, R.; Hudson, R.; Kheirbek, I.; Kleiman, G.;Miller, P.; Weiss, L. The Nature of the Ozone Air Quality Problem in theOzone Transport Region: A Conceptual Description; Boston, MA, 2010.(18) Northeast States for Coordinated Air Use Management. HighElectric Demand Day and Air Quality in the Northeast; Boston, MA,2006.(19) EPA’s NOx Reduction Program and Clean Air Interstate Rule:2009 Environmental and Health Results; U.S. Environmental ProtectionAgency: Research Triangle Park, NC, 2011.(20) Ozone Transport Commission. Memorandum of Understandingamong the States of the Ozone Transport Commission Concerning theIncorporation of High Electric Demand Day Emission Reduction Strategiesinto Ozone Attainment State Implementation Planning; Washington. DC,2007.(21) Ozone Transport Commission. High Electric Demand DayTrigger - Final; Washington. DC, 2007.(22) U.S. Environmental Protection Agency. Air Markets ProgramData http://ampd.epa.gov/ampd/ (accessed October 1, 2014).(23) U.S. Environmental Protection Agency. AirData http://www.epa.gov/airdata/ad_data_daily.html (accessed October 1, 2014).(24) New York State Department of Environmental Conservation.Overview of Part 222, Version MMXIII; 2013.(25) Consolidated Edison Inc. Comments by Consolidate Edison ofNY to draft Part 222, Distributed Generation Sources, 2013.(26) Nelson, B. Alternative Control Techniques Document: StationaryDiesel Engines; Chapel Hill, NC, 2010.(27) U.S. Environmental Protection Agency. Exhaust and CrankcaseEmission Factors for Nonroad Engine Modeling − Compression-Ignition,EPA-420-R-10-018; 2010.

Environmental Science & Technology Policy Analysis

DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

1266

(28) Sinha, P.; Depro, B.; Braun, F. Regulatory Impact Analysis (RIA)for Proposed Reconsideration of Existing Stationary Compression Ignition(CI) Engines NESHAP; Research Triangle Park, NC, 2012.(29) King, M. EPA’s Air Quality Regulations for Stationary Engines;U.S. Environmental Protection Agency, 2013.(30) New York Independent System Operator. NYISO 2011 AnnualReport on Demand Response Programs; Rensselaer, NY, 2012.(31) New York Independent System Operator. NYISO 2012 AnnualReport on Demand Response Programs; Rensselaer, NY, 2013.(32) New York Independent System Operator. NYISO 2013 AnnualReport on Demand Response Programs; Rensselaer, NY, 2014.(33) He, H.; Hembeck, L.; Hosley, K. M.; Canty, T. P.; Salawitch, R.J.; Dickerson, R. R. High ozone concentrations on hot days: The roleof electric power demand and NOx emissions. Geophys. Res. Lett. 2013,40, 5291−5294.(34) NERA Economic Consulting. Independent Study to EstablishParameters of the ICAP Demand Curve for the New York IndependentSystem Operator; Washington, DC, 2013.

Environmental Science & Technology Policy Analysis

DOI: 10.1021/es505007mEnviron. Sci. Technol. 2015, 49, 1260−1267

1267