Quantifying CO2 savings from wind power

Ireland

Joseph Wheatley

October 4 2012

Abstract

The contribution of wind power generation to operational CO2 sav-ings is investigated for the Irish electricity grid Wind contributed 17of electricity demand in 2011 and reduced CO2 emissions by 9 Windenergy saved 028 tCO2MWh on average relative to a grid carbon in-tensity in the absence of wind of 053tCO2MWh Emissions savings areat the lower end of expectations It is likely that this reflects decreasingeffectiveness of wind power as wind penetration increases

1 Outline of the problem

Increase in atmospheric greenhouse gases due mainly to burning of hydrocar-bons and coal is shifting the Earthrsquos radiative balance in favour of a warmerclimate[1] Reduction of industrial CO2 emissions is a major focus of global en-vironmental policy One widely adopted policy measure are state supports forwind power[2] on the grounds that wind power displaces fossil fuel generationand so reduces emissions State supports have included mandatory targets feed-in tariffs subsidised finance for infrastructure etc As a consequence significantamounts of wind power have been embedded into electrothermal generation sys-tems It has been known for some time that thermal generation responds in anon-trivial way when operated in parallel to stochastic power sources to meetsystem demand Not all thermal plant are displaced equally with flexible andorhigh marginal cost generation being displaced the most Average efficiency isreduced and higher cycling rates occur than would otherwise be the case Allof these effects tend to reduce the effectiveness of wind power in meeting itrsquosprimary policy goal namely emissions reduction

The task of quantifying emissions savings from wind power is not straight-forward Electricity grids are complex systems with many competing com-ponents and feedbacks Moreover every grid has a different combination offuel-mix generator types wind penetration interconnection despatch prac-tices etc Estimates of emissions savings have ranged widely[3] from higher thangrid average[4 5] to near zero[6 7] Savings assumptions by public authoritieshave trended lower over time[8] Meanwhile despatch models have demonstrated

1

that marginal savings decrease as installed wind capacity increases[9 10] andthat high levels of wind penetration may even be counterproductive in terms ofemissions[11]

Empirical approaches based on real world grid data can help shed furtherlight on these issues Ireland is a good empirical test case for the followingreasons

1 high average wind penetration (17 in 2011)

2 minimal electricity exports means that virtually all wind generation mustbe accommodated on the domestic grid

3 modern thermal plant portfolio with large amounts of relatively flexiblegas generation (asymp 58 of demand) as well as coal and peat plant

4 zero nuclear and a low level of hydro (asymp 2)

5 Irelandrsquos highly volatile wind resource favours statistical even over rela-tively short timeframes such as one year

6 availability of relatively high frequency grid data and mandatory emissionsreporting at plant level under EU-ETS[12]

The Irish grid operator[13] reports approximate system demand wind gen-eration and total CO2 emissions rate every 1

4 -hour It is easy to obtain a pre-liminary estimate of emissions savings from this dataset Linear regression ofthe time-series of grid carbon intensity (emissions rate per unit demand) ontowind penetration (wind generation per unit demand) gives a zero-wind emis-sions intensity of 051tCO2MWh and wind power savings 035tCO2MWh in2011 This is equivalent to a displacement effectiveness of just 65 A plausi-ble interpretation is that wind power displaces primarily clean gas (which havetypical emissions asymp 035tCO2MWh) rather than high emissions coal or peat

While these numbers are suggestive their origin and accuracy are unclearFirstly aggregate numbers cannot show which generators or fuels are beingdisplaced by wind power Secondly cycling effects (startup and ramping ofthermal plant) are not included in the carbon emissions algorithm used by thegrid operator Thirdly the role played by interconnection (electricity importsand exports) is unclear Fourthly the result for emissions savings is sensitiveto the correlation between wind generation and system demand Spurious cor-relation may be present because approximate wind generation is used in thecalculation of system demand

In this study time-series of CO2 emissions are estimated for each grid-connected thermal unit in 2011 This calculation is based on generation dataand physical characteristics of each generator Additional emissions due to start-ups are included Based on this CO2 data and a careful treatment of the windand system demand we estimate wind savings of -028tCO2MWh with impliedeffectiveness of only 53 Some implications of these numbers are discussed atthe end of the article

2

2 Data and model

Data on the Irish electricity grid is available from a number of sources[13][14][15]The Single Electricity Market Operator (SEMO) provide 1

2 -hourly generationdata ldquoEx Post Initialrdquo metered generation data is compiled four days after gen-eration date for invoicing purposes This accurate dataset is used here SEMOprovide a loss factor adjusted total metered generation (ldquoMGLFrdquo) time-serieswhich includes all domestic generation as well as power traded via intercon-nectors to Northern Ireland Each power source is weighted by a unit specifictransmission system loss factor which also depends on time of day and seasonMGLF is a proxy for total end-user demand because accurate balance betweengeneration and demand is required at all times in order to maintain grid fre-quency stability MGLF is very similar but not identical to the real-time systemdemand recorded by the grid operator (correlation 099)[13] A time-series ofnet intra-jurisdiction importsexports is also available (ldquoNIJIrdquo)

Under the priority despatch rule for wind conventional generation mustadjust continuously to match system demand minus wind generation Fromthe point of view of the conventional generation system wind generation is anexogenous random variable which reduces effective demand but also makes itmore volatile Figure 1 The aim is to find out what impact this has on individualthermal generators SEMO provides 1

2 -hourly metered generation data (ldquoMGrdquo)data on all grid-connected generators[15] Excluding wind power 54 generatorssupply the Irish grid 35 of these are thermal the rest are small hydropumpedstorage units Descriptions of thermal generators such as fuel type maximumcapacity etc are given in Table 3[15]

Wind generation is the sum of the outputs of more than 130 wind farmsMost of these are smaller installations connected to the distribution systemSEMO also provide 1

2 -hourly MG data for wind farms but there are gaps in thisdataset Some kind of statistical modelling is required to recover total windgeneration To do this generation data for 36 of the largest wind farms whichwere fully operational during 2011 (20 transmission and 16 distribution systemconnected) were compiled The sum of the capacities of this sample amountedto asymp 2

3 of the total installed wind capacity during 2011 The time-series of totaloutput from the sample was normalised upwards to reflect the correct totalinstalled monthly wind capacity (this allows for asymp 200MW increase in windcapacity during 2011) This is expected to be a good approximation becausewind generation is highly correlated in space and in time1 The estimate of windgeneration resulting from the procedure is similar to the real-time estimate ofthe grid operator (correlation amp 099)[13]

In all the generator dataset used in this study contains close to 2 milliondata points A visualisation of the dataset showing 1

2 -hourly generation fraction

1The physical reason for this is that the maximum linear separation of wind farms on theIrish grid is much less that the extent of typical mid-latitude weather systems which bringwindy or calm conditions For example the correlation between wind generation at Meentycatin extreme North-West and Carnsore in the extreme South-East is 047 Autocorrelation isalso very strong eg 12 hour lagged correlation is 065

3

by fuel type ( ldquofuel-mix plotrdquo) is shown in 2 [17] Gas is the dominant powersource (53) and it is displaced when wind generation is present

To model CO2 emissions detailed physical characteristics of individual ther-mal generators are required Parameters listed in Tables 34 are used by electric-ity market participants to determine fuel use and support despatch decisions[16]Table 4 gives Incremental Heat Rate Slopes (GJMWh) which are defined be-tween a set of capacity points (MW) specific to each generator Together withzero-load energy use (GJh) these can be inverted to find the rate of energyuse by the generator at each capacity point Full input-output curves were thenconstructed by cubic spline interpolation These input-output curves give theemissions rate when a generator is operated in steady state at any point between0MW and MaxCap (making use of fuel CO2 emission factors[14]) In practicegenerators can only operate in steady state only between MinCap and MaxCapor as spinning reserve (at 0MW)

Once input-output curves have been constructed they can be combined withgeneration data to create emissions time-series for each unit2 Total emissionscalculated in this way include the effects of generation mix and part-load effi-ciency Emissions from this model were 1164Mt in 2011 compared to 1167Mtusing the grid operator data[13] The two emissions time-series are similar butnot identical with a correlation of asymp 097 Further tests on the validity of thismodel are described below

Additional emissions arising from cycling (start-ups) can be calculated usingparameters in Table 3 Fuel costs (in GJ) depend on thermal state of thegenerator at start-up (Hot Warm or Cold) With appropriate initial conditionsa thermal state time-series is constructed for each generator using transitiontimes between the states (HotToWarm and WarmToCold) provided in Table 3Startup fuel which may be different from the primary fuel type in some cases(eg for coal generators startup fuel is a 68-32 oilcoal mixture) Figure3 shows an example of the emissions associated with a coal generator Startupemissions in 2011 were 014MtCO2 bringing total emissions to 1178MtCO2This compares to verified emissions of 1184MtCO2

2One peat generator (ldquoED1rdquo) was 15 co-fired with biomass material in 2011 This meansthat under emissions rules itrsquos net emissions are reduced by the same factor

4

Figure

1ndash

Upp

er

Dem

and

(MG

LF

bla

ck)

and

esti

mate

dw

ind

pow

ergen

erati

on

(gre

en)

Mea

ndem

and

is2813M

W

Mea

nw

ind

was

467M

Ww

ith

capaci

tyfa

ctor

was

30

T

he

corr

elati

on

bet

wee

nw

ind

gen

erati

on

and

dem

and

iscl

ose

toze

ro(asymp

00

5)

Low

er

Eff

ecti

ve

dem

and

(=dem

and

-w

ind)

as

seen

by

conven

tional

pla

nt

Eff

ecti

ve

dem

and

isasymp

38

more

vola

tile

than

syst

emdem

and

5

Figure 2 ndash Left Half-hourly fuel mix time-series (generation fraction) for 2011derived from SEMO data Right corresponding CO2-mix time series The verti-cal scale is day of year Mean generation fractions were (1) gas (+chp) 58 (2)wind 17 (3) coal 15 (4) peat 8 Mean emissions fractions on the other handwere (1) gas (+chp) 49 (2) coal 31 (3) peat 19

6

Figure 3 ndash Top Metered generation(MW) from coal generator MP2 (Money-point power station) during 2011 The red lines correspond to maximum andminimum stable operating capacities Thermal state (Cold Warm Hot) are in-dicated by colours blue brown black Bottom corresponding CO2 emissionsrate (tCO2h) The spikes in emissions correspond to cycling (start up) eventsFor visualisation and modelling convenience all cycling emissions are assumed tooccur within a half hour interval

As a consistency test of the emissions model comparison can be made withannual power plant emissions reported under to the European Emissions TradingScheme (EU-ETS) for 2011[12] These data provide an independent test of theemissions model because they are based on actual fuel consumption informationcollected using stock accounting principles To convert to CO2 standard emis-sions factors are associated with each fuel (although in the case of coal emissionsfactors can vary significantly from year to year) To make contact with thesedata the emissions time-series of the 35 thermal generators are aggregated into11 power plant and annual emissions calculated These data provide both aconsistency check and model constraint as illustrated in 4 In all agreementis satisfactory and is particularly good for gas base-load (Combined Cycle Gas

7

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

that marginal savings decrease as installed wind capacity increases[9 10] andthat high levels of wind penetration may even be counterproductive in terms ofemissions[11]

Empirical approaches based on real world grid data can help shed furtherlight on these issues Ireland is a good empirical test case for the followingreasons

1 high average wind penetration (17 in 2011)

2 minimal electricity exports means that virtually all wind generation mustbe accommodated on the domestic grid

3 modern thermal plant portfolio with large amounts of relatively flexiblegas generation (asymp 58 of demand) as well as coal and peat plant

4 zero nuclear and a low level of hydro (asymp 2)

5 Irelandrsquos highly volatile wind resource favours statistical even over rela-tively short timeframes such as one year

6 availability of relatively high frequency grid data and mandatory emissionsreporting at plant level under EU-ETS[12]

The Irish grid operator[13] reports approximate system demand wind gen-eration and total CO2 emissions rate every 1

4 -hour It is easy to obtain a pre-liminary estimate of emissions savings from this dataset Linear regression ofthe time-series of grid carbon intensity (emissions rate per unit demand) ontowind penetration (wind generation per unit demand) gives a zero-wind emis-sions intensity of 051tCO2MWh and wind power savings 035tCO2MWh in2011 This is equivalent to a displacement effectiveness of just 65 A plausi-ble interpretation is that wind power displaces primarily clean gas (which havetypical emissions asymp 035tCO2MWh) rather than high emissions coal or peat

While these numbers are suggestive their origin and accuracy are unclearFirstly aggregate numbers cannot show which generators or fuels are beingdisplaced by wind power Secondly cycling effects (startup and ramping ofthermal plant) are not included in the carbon emissions algorithm used by thegrid operator Thirdly the role played by interconnection (electricity importsand exports) is unclear Fourthly the result for emissions savings is sensitiveto the correlation between wind generation and system demand Spurious cor-relation may be present because approximate wind generation is used in thecalculation of system demand

In this study time-series of CO2 emissions are estimated for each grid-connected thermal unit in 2011 This calculation is based on generation dataand physical characteristics of each generator Additional emissions due to start-ups are included Based on this CO2 data and a careful treatment of the windand system demand we estimate wind savings of -028tCO2MWh with impliedeffectiveness of only 53 Some implications of these numbers are discussed atthe end of the article

2

2 Data and model

Data on the Irish electricity grid is available from a number of sources[13][14][15]The Single Electricity Market Operator (SEMO) provide 1

2 -hourly generationdata ldquoEx Post Initialrdquo metered generation data is compiled four days after gen-eration date for invoicing purposes This accurate dataset is used here SEMOprovide a loss factor adjusted total metered generation (ldquoMGLFrdquo) time-serieswhich includes all domestic generation as well as power traded via intercon-nectors to Northern Ireland Each power source is weighted by a unit specifictransmission system loss factor which also depends on time of day and seasonMGLF is a proxy for total end-user demand because accurate balance betweengeneration and demand is required at all times in order to maintain grid fre-quency stability MGLF is very similar but not identical to the real-time systemdemand recorded by the grid operator (correlation 099)[13] A time-series ofnet intra-jurisdiction importsexports is also available (ldquoNIJIrdquo)

Under the priority despatch rule for wind conventional generation mustadjust continuously to match system demand minus wind generation Fromthe point of view of the conventional generation system wind generation is anexogenous random variable which reduces effective demand but also makes itmore volatile Figure 1 The aim is to find out what impact this has on individualthermal generators SEMO provides 1

2 -hourly metered generation data (ldquoMGrdquo)data on all grid-connected generators[15] Excluding wind power 54 generatorssupply the Irish grid 35 of these are thermal the rest are small hydropumpedstorage units Descriptions of thermal generators such as fuel type maximumcapacity etc are given in Table 3[15]

Wind generation is the sum of the outputs of more than 130 wind farmsMost of these are smaller installations connected to the distribution systemSEMO also provide 1

2 -hourly MG data for wind farms but there are gaps in thisdataset Some kind of statistical modelling is required to recover total windgeneration To do this generation data for 36 of the largest wind farms whichwere fully operational during 2011 (20 transmission and 16 distribution systemconnected) were compiled The sum of the capacities of this sample amountedto asymp 2

3 of the total installed wind capacity during 2011 The time-series of totaloutput from the sample was normalised upwards to reflect the correct totalinstalled monthly wind capacity (this allows for asymp 200MW increase in windcapacity during 2011) This is expected to be a good approximation becausewind generation is highly correlated in space and in time1 The estimate of windgeneration resulting from the procedure is similar to the real-time estimate ofthe grid operator (correlation amp 099)[13]

In all the generator dataset used in this study contains close to 2 milliondata points A visualisation of the dataset showing 1

2 -hourly generation fraction

1The physical reason for this is that the maximum linear separation of wind farms on theIrish grid is much less that the extent of typical mid-latitude weather systems which bringwindy or calm conditions For example the correlation between wind generation at Meentycatin extreme North-West and Carnsore in the extreme South-East is 047 Autocorrelation isalso very strong eg 12 hour lagged correlation is 065

3

by fuel type ( ldquofuel-mix plotrdquo) is shown in 2 [17] Gas is the dominant powersource (53) and it is displaced when wind generation is present

To model CO2 emissions detailed physical characteristics of individual ther-mal generators are required Parameters listed in Tables 34 are used by electric-ity market participants to determine fuel use and support despatch decisions[16]Table 4 gives Incremental Heat Rate Slopes (GJMWh) which are defined be-tween a set of capacity points (MW) specific to each generator Together withzero-load energy use (GJh) these can be inverted to find the rate of energyuse by the generator at each capacity point Full input-output curves were thenconstructed by cubic spline interpolation These input-output curves give theemissions rate when a generator is operated in steady state at any point between0MW and MaxCap (making use of fuel CO2 emission factors[14]) In practicegenerators can only operate in steady state only between MinCap and MaxCapor as spinning reserve (at 0MW)

Once input-output curves have been constructed they can be combined withgeneration data to create emissions time-series for each unit2 Total emissionscalculated in this way include the effects of generation mix and part-load effi-ciency Emissions from this model were 1164Mt in 2011 compared to 1167Mtusing the grid operator data[13] The two emissions time-series are similar butnot identical with a correlation of asymp 097 Further tests on the validity of thismodel are described below

Additional emissions arising from cycling (start-ups) can be calculated usingparameters in Table 3 Fuel costs (in GJ) depend on thermal state of thegenerator at start-up (Hot Warm or Cold) With appropriate initial conditionsa thermal state time-series is constructed for each generator using transitiontimes between the states (HotToWarm and WarmToCold) provided in Table 3Startup fuel which may be different from the primary fuel type in some cases(eg for coal generators startup fuel is a 68-32 oilcoal mixture) Figure3 shows an example of the emissions associated with a coal generator Startupemissions in 2011 were 014MtCO2 bringing total emissions to 1178MtCO2This compares to verified emissions of 1184MtCO2

2One peat generator (ldquoED1rdquo) was 15 co-fired with biomass material in 2011 This meansthat under emissions rules itrsquos net emissions are reduced by the same factor

4

Figure

1ndash

Upp

er

Dem

and

(MG

LF

bla

ck)

and

esti

mate

dw

ind

pow

ergen

erati

on

(gre

en)

Mea

ndem

and

is2813M

W

Mea

nw

ind

was

467M

Ww

ith

capaci

tyfa

ctor

was

30

T

he

corr

elati

on

bet

wee

nw

ind

gen

erati

on

and

dem

and

iscl

ose

toze

ro(asymp

00

5)

Low

er

Eff

ecti

ve

dem

and

(=dem

and

-w

ind)

as

seen

by

conven

tional

pla

nt

Eff

ecti

ve

dem

and

isasymp

38

more

vola

tile

than

syst

emdem

and

5

Figure 2 ndash Left Half-hourly fuel mix time-series (generation fraction) for 2011derived from SEMO data Right corresponding CO2-mix time series The verti-cal scale is day of year Mean generation fractions were (1) gas (+chp) 58 (2)wind 17 (3) coal 15 (4) peat 8 Mean emissions fractions on the other handwere (1) gas (+chp) 49 (2) coal 31 (3) peat 19

6

Figure 3 ndash Top Metered generation(MW) from coal generator MP2 (Money-point power station) during 2011 The red lines correspond to maximum andminimum stable operating capacities Thermal state (Cold Warm Hot) are in-dicated by colours blue brown black Bottom corresponding CO2 emissionsrate (tCO2h) The spikes in emissions correspond to cycling (start up) eventsFor visualisation and modelling convenience all cycling emissions are assumed tooccur within a half hour interval

As a consistency test of the emissions model comparison can be made withannual power plant emissions reported under to the European Emissions TradingScheme (EU-ETS) for 2011[12] These data provide an independent test of theemissions model because they are based on actual fuel consumption informationcollected using stock accounting principles To convert to CO2 standard emis-sions factors are associated with each fuel (although in the case of coal emissionsfactors can vary significantly from year to year) To make contact with thesedata the emissions time-series of the 35 thermal generators are aggregated into11 power plant and annual emissions calculated These data provide both aconsistency check and model constraint as illustrated in 4 In all agreementis satisfactory and is particularly good for gas base-load (Combined Cycle Gas

7

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

2 Data and model

Data on the Irish electricity grid is available from a number of sources[13][14][15]The Single Electricity Market Operator (SEMO) provide 1

2 -hourly generationdata ldquoEx Post Initialrdquo metered generation data is compiled four days after gen-eration date for invoicing purposes This accurate dataset is used here SEMOprovide a loss factor adjusted total metered generation (ldquoMGLFrdquo) time-serieswhich includes all domestic generation as well as power traded via intercon-nectors to Northern Ireland Each power source is weighted by a unit specifictransmission system loss factor which also depends on time of day and seasonMGLF is a proxy for total end-user demand because accurate balance betweengeneration and demand is required at all times in order to maintain grid fre-quency stability MGLF is very similar but not identical to the real-time systemdemand recorded by the grid operator (correlation 099)[13] A time-series ofnet intra-jurisdiction importsexports is also available (ldquoNIJIrdquo)

Under the priority despatch rule for wind conventional generation mustadjust continuously to match system demand minus wind generation Fromthe point of view of the conventional generation system wind generation is anexogenous random variable which reduces effective demand but also makes itmore volatile Figure 1 The aim is to find out what impact this has on individualthermal generators SEMO provides 1

2 -hourly metered generation data (ldquoMGrdquo)data on all grid-connected generators[15] Excluding wind power 54 generatorssupply the Irish grid 35 of these are thermal the rest are small hydropumpedstorage units Descriptions of thermal generators such as fuel type maximumcapacity etc are given in Table 3[15]

Wind generation is the sum of the outputs of more than 130 wind farmsMost of these are smaller installations connected to the distribution systemSEMO also provide 1

2 -hourly MG data for wind farms but there are gaps in thisdataset Some kind of statistical modelling is required to recover total windgeneration To do this generation data for 36 of the largest wind farms whichwere fully operational during 2011 (20 transmission and 16 distribution systemconnected) were compiled The sum of the capacities of this sample amountedto asymp 2

3 of the total installed wind capacity during 2011 The time-series of totaloutput from the sample was normalised upwards to reflect the correct totalinstalled monthly wind capacity (this allows for asymp 200MW increase in windcapacity during 2011) This is expected to be a good approximation becausewind generation is highly correlated in space and in time1 The estimate of windgeneration resulting from the procedure is similar to the real-time estimate ofthe grid operator (correlation amp 099)[13]

In all the generator dataset used in this study contains close to 2 milliondata points A visualisation of the dataset showing 1

2 -hourly generation fraction

1The physical reason for this is that the maximum linear separation of wind farms on theIrish grid is much less that the extent of typical mid-latitude weather systems which bringwindy or calm conditions For example the correlation between wind generation at Meentycatin extreme North-West and Carnsore in the extreme South-East is 047 Autocorrelation isalso very strong eg 12 hour lagged correlation is 065

3

by fuel type ( ldquofuel-mix plotrdquo) is shown in 2 [17] Gas is the dominant powersource (53) and it is displaced when wind generation is present

To model CO2 emissions detailed physical characteristics of individual ther-mal generators are required Parameters listed in Tables 34 are used by electric-ity market participants to determine fuel use and support despatch decisions[16]Table 4 gives Incremental Heat Rate Slopes (GJMWh) which are defined be-tween a set of capacity points (MW) specific to each generator Together withzero-load energy use (GJh) these can be inverted to find the rate of energyuse by the generator at each capacity point Full input-output curves were thenconstructed by cubic spline interpolation These input-output curves give theemissions rate when a generator is operated in steady state at any point between0MW and MaxCap (making use of fuel CO2 emission factors[14]) In practicegenerators can only operate in steady state only between MinCap and MaxCapor as spinning reserve (at 0MW)

Once input-output curves have been constructed they can be combined withgeneration data to create emissions time-series for each unit2 Total emissionscalculated in this way include the effects of generation mix and part-load effi-ciency Emissions from this model were 1164Mt in 2011 compared to 1167Mtusing the grid operator data[13] The two emissions time-series are similar butnot identical with a correlation of asymp 097 Further tests on the validity of thismodel are described below

Additional emissions arising from cycling (start-ups) can be calculated usingparameters in Table 3 Fuel costs (in GJ) depend on thermal state of thegenerator at start-up (Hot Warm or Cold) With appropriate initial conditionsa thermal state time-series is constructed for each generator using transitiontimes between the states (HotToWarm and WarmToCold) provided in Table 3Startup fuel which may be different from the primary fuel type in some cases(eg for coal generators startup fuel is a 68-32 oilcoal mixture) Figure3 shows an example of the emissions associated with a coal generator Startupemissions in 2011 were 014MtCO2 bringing total emissions to 1178MtCO2This compares to verified emissions of 1184MtCO2

2One peat generator (ldquoED1rdquo) was 15 co-fired with biomass material in 2011 This meansthat under emissions rules itrsquos net emissions are reduced by the same factor

4

Figure

1ndash

Upp

er

Dem

and

(MG

LF

bla

ck)

and

esti

mate

dw

ind

pow

ergen

erati

on

(gre

en)

Mea

ndem

and

is2813M

W

Mea

nw

ind

was

467M

Ww

ith

capaci

tyfa

ctor

was

30

T

he

corr

elati

on

bet

wee

nw

ind

gen

erati

on

and

dem

and

iscl

ose

toze

ro(asymp

00

5)

Low

er

Eff

ecti

ve

dem

and

(=dem

and

-w

ind)

as

seen

by

conven

tional

pla

nt

Eff

ecti

ve

dem

and

isasymp

38

more

vola

tile

than

syst

emdem

and

5

Figure 2 ndash Left Half-hourly fuel mix time-series (generation fraction) for 2011derived from SEMO data Right corresponding CO2-mix time series The verti-cal scale is day of year Mean generation fractions were (1) gas (+chp) 58 (2)wind 17 (3) coal 15 (4) peat 8 Mean emissions fractions on the other handwere (1) gas (+chp) 49 (2) coal 31 (3) peat 19

6

Figure 3 ndash Top Metered generation(MW) from coal generator MP2 (Money-point power station) during 2011 The red lines correspond to maximum andminimum stable operating capacities Thermal state (Cold Warm Hot) are in-dicated by colours blue brown black Bottom corresponding CO2 emissionsrate (tCO2h) The spikes in emissions correspond to cycling (start up) eventsFor visualisation and modelling convenience all cycling emissions are assumed tooccur within a half hour interval

As a consistency test of the emissions model comparison can be made withannual power plant emissions reported under to the European Emissions TradingScheme (EU-ETS) for 2011[12] These data provide an independent test of theemissions model because they are based on actual fuel consumption informationcollected using stock accounting principles To convert to CO2 standard emis-sions factors are associated with each fuel (although in the case of coal emissionsfactors can vary significantly from year to year) To make contact with thesedata the emissions time-series of the 35 thermal generators are aggregated into11 power plant and annual emissions calculated These data provide both aconsistency check and model constraint as illustrated in 4 In all agreementis satisfactory and is particularly good for gas base-load (Combined Cycle Gas

7

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

by fuel type ( ldquofuel-mix plotrdquo) is shown in 2 [17] Gas is the dominant powersource (53) and it is displaced when wind generation is present

To model CO2 emissions detailed physical characteristics of individual ther-mal generators are required Parameters listed in Tables 34 are used by electric-ity market participants to determine fuel use and support despatch decisions[16]Table 4 gives Incremental Heat Rate Slopes (GJMWh) which are defined be-tween a set of capacity points (MW) specific to each generator Together withzero-load energy use (GJh) these can be inverted to find the rate of energyuse by the generator at each capacity point Full input-output curves were thenconstructed by cubic spline interpolation These input-output curves give theemissions rate when a generator is operated in steady state at any point between0MW and MaxCap (making use of fuel CO2 emission factors[14]) In practicegenerators can only operate in steady state only between MinCap and MaxCapor as spinning reserve (at 0MW)

Once input-output curves have been constructed they can be combined withgeneration data to create emissions time-series for each unit2 Total emissionscalculated in this way include the effects of generation mix and part-load effi-ciency Emissions from this model were 1164Mt in 2011 compared to 1167Mtusing the grid operator data[13] The two emissions time-series are similar butnot identical with a correlation of asymp 097 Further tests on the validity of thismodel are described below

Additional emissions arising from cycling (start-ups) can be calculated usingparameters in Table 3 Fuel costs (in GJ) depend on thermal state of thegenerator at start-up (Hot Warm or Cold) With appropriate initial conditionsa thermal state time-series is constructed for each generator using transitiontimes between the states (HotToWarm and WarmToCold) provided in Table 3Startup fuel which may be different from the primary fuel type in some cases(eg for coal generators startup fuel is a 68-32 oilcoal mixture) Figure3 shows an example of the emissions associated with a coal generator Startupemissions in 2011 were 014MtCO2 bringing total emissions to 1178MtCO2This compares to verified emissions of 1184MtCO2

2One peat generator (ldquoED1rdquo) was 15 co-fired with biomass material in 2011 This meansthat under emissions rules itrsquos net emissions are reduced by the same factor

4

Figure

1ndash

Upp

er

Dem

and

(MG

LF

bla

ck)

and

esti

mate

dw

ind

pow

ergen

erati

on

(gre

en)

Mea

ndem

and

is2813M

W

Mea

nw

ind

was

467M

Ww

ith

capaci

tyfa

ctor

was

30

T

he

corr

elati

on

bet

wee

nw

ind

gen

erati

on

and

dem

and

iscl

ose

toze

ro(asymp

00

5)

Low

er

Eff

ecti

ve

dem

and

(=dem

and

-w

ind)

as

seen

by

conven

tional

pla

nt

Eff

ecti

ve

dem

and

isasymp

38

more

vola

tile

than

syst

emdem

and

5

Figure 2 ndash Left Half-hourly fuel mix time-series (generation fraction) for 2011derived from SEMO data Right corresponding CO2-mix time series The verti-cal scale is day of year Mean generation fractions were (1) gas (+chp) 58 (2)wind 17 (3) coal 15 (4) peat 8 Mean emissions fractions on the other handwere (1) gas (+chp) 49 (2) coal 31 (3) peat 19

6

Figure 3 ndash Top Metered generation(MW) from coal generator MP2 (Money-point power station) during 2011 The red lines correspond to maximum andminimum stable operating capacities Thermal state (Cold Warm Hot) are in-dicated by colours blue brown black Bottom corresponding CO2 emissionsrate (tCO2h) The spikes in emissions correspond to cycling (start up) eventsFor visualisation and modelling convenience all cycling emissions are assumed tooccur within a half hour interval

As a consistency test of the emissions model comparison can be made withannual power plant emissions reported under to the European Emissions TradingScheme (EU-ETS) for 2011[12] These data provide an independent test of theemissions model because they are based on actual fuel consumption informationcollected using stock accounting principles To convert to CO2 standard emis-sions factors are associated with each fuel (although in the case of coal emissionsfactors can vary significantly from year to year) To make contact with thesedata the emissions time-series of the 35 thermal generators are aggregated into11 power plant and annual emissions calculated These data provide both aconsistency check and model constraint as illustrated in 4 In all agreementis satisfactory and is particularly good for gas base-load (Combined Cycle Gas

7

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

Figure

1ndash

Upp

er

Dem

and

(MG

LF

bla

ck)

and

esti

mate

dw

ind

pow

ergen

erati

on

(gre

en)

Mea

ndem

and

is2813M

W

Mea

nw

ind

was

467M

Ww

ith

capaci

tyfa

ctor

was

30

T

he

corr

elati

on

bet

wee

nw

ind

gen

erati

on

and

dem

and

iscl

ose

toze

ro(asymp

00

5)

Low

er

Eff

ecti

ve

dem

and

(=dem

and

-w

ind)

as

seen

by

conven

tional

pla

nt

Eff

ecti

ve

dem

and

isasymp

38

more

vola

tile

than

syst

emdem

and

5

Figure 2 ndash Left Half-hourly fuel mix time-series (generation fraction) for 2011derived from SEMO data Right corresponding CO2-mix time series The verti-cal scale is day of year Mean generation fractions were (1) gas (+chp) 58 (2)wind 17 (3) coal 15 (4) peat 8 Mean emissions fractions on the other handwere (1) gas (+chp) 49 (2) coal 31 (3) peat 19

6

Figure 3 ndash Top Metered generation(MW) from coal generator MP2 (Money-point power station) during 2011 The red lines correspond to maximum andminimum stable operating capacities Thermal state (Cold Warm Hot) are in-dicated by colours blue brown black Bottom corresponding CO2 emissionsrate (tCO2h) The spikes in emissions correspond to cycling (start up) eventsFor visualisation and modelling convenience all cycling emissions are assumed tooccur within a half hour interval

As a consistency test of the emissions model comparison can be made withannual power plant emissions reported under to the European Emissions TradingScheme (EU-ETS) for 2011[12] These data provide an independent test of theemissions model because they are based on actual fuel consumption informationcollected using stock accounting principles To convert to CO2 standard emis-sions factors are associated with each fuel (although in the case of coal emissionsfactors can vary significantly from year to year) To make contact with thesedata the emissions time-series of the 35 thermal generators are aggregated into11 power plant and annual emissions calculated These data provide both aconsistency check and model constraint as illustrated in 4 In all agreementis satisfactory and is particularly good for gas base-load (Combined Cycle Gas

7

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

Figure 2 ndash Left Half-hourly fuel mix time-series (generation fraction) for 2011derived from SEMO data Right corresponding CO2-mix time series The verti-cal scale is day of year Mean generation fractions were (1) gas (+chp) 58 (2)wind 17 (3) coal 15 (4) peat 8 Mean emissions fractions on the other handwere (1) gas (+chp) 49 (2) coal 31 (3) peat 19

6

Figure 3 ndash Top Metered generation(MW) from coal generator MP2 (Money-point power station) during 2011 The red lines correspond to maximum andminimum stable operating capacities Thermal state (Cold Warm Hot) are in-dicated by colours blue brown black Bottom corresponding CO2 emissionsrate (tCO2h) The spikes in emissions correspond to cycling (start up) eventsFor visualisation and modelling convenience all cycling emissions are assumed tooccur within a half hour interval

As a consistency test of the emissions model comparison can be made withannual power plant emissions reported under to the European Emissions TradingScheme (EU-ETS) for 2011[12] These data provide an independent test of theemissions model because they are based on actual fuel consumption informationcollected using stock accounting principles To convert to CO2 standard emis-sions factors are associated with each fuel (although in the case of coal emissionsfactors can vary significantly from year to year) To make contact with thesedata the emissions time-series of the 35 thermal generators are aggregated into11 power plant and annual emissions calculated These data provide both aconsistency check and model constraint as illustrated in 4 In all agreementis satisfactory and is particularly good for gas base-load (Combined Cycle Gas

7

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

Figure 3 ndash Top Metered generation(MW) from coal generator MP2 (Money-point power station) during 2011 The red lines correspond to maximum andminimum stable operating capacities Thermal state (Cold Warm Hot) are in-dicated by colours blue brown black Bottom corresponding CO2 emissionsrate (tCO2h) The spikes in emissions correspond to cycling (start up) eventsFor visualisation and modelling convenience all cycling emissions are assumed tooccur within a half hour interval

As a consistency test of the emissions model comparison can be made withannual power plant emissions reported under to the European Emissions TradingScheme (EU-ETS) for 2011[12] These data provide an independent test of theemissions model because they are based on actual fuel consumption informationcollected using stock accounting principles To convert to CO2 standard emis-sions factors are associated with each fuel (although in the case of coal emissionsfactors can vary significantly from year to year) To make contact with thesedata the emissions time-series of the 35 thermal generators are aggregated into11 power plant and annual emissions calculated These data provide both aconsistency check and model constraint as illustrated in 4 In all agreementis satisfactory and is particularly good for gas base-load (Combined Cycle Gas

7

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

Turbine or CCGT) plant It is not surprising that errors are somewhat larger forintermittently used ldquopeakingrdquo plant but these form only a very small fractionof total emissions

Figure 4 ndash Comparison of model and reported power station emissions in 2011Carbon intensities (tCO2MWh) based on emissions reported under EU-ETS areshown in light colours while modeled values are shown in heavy colors Baseloadgas (CCGT) plant have lowest emissions intensities Emissions factors used toconvert from energy units to emissions (TWhtCO2) were coal = 899 peat= 1167 oil = 714 gas= 569 Verified EU-ETS emissions at Aughinish includeemissions unrelated to electric power generation for the grid so these have beenset to the model value

In fact the above emissions model is incomplete Ramping a thermal gen-erator burns additional fuel even when no start-ups occur This arises when agenerator is ramped between MinCap and MaxCap as in Figure 3 for exam-ple Simply summing emissions based on the generatorrsquos input-output curve asdone above does not capture this effect This additional source of emissions hasbeen emphasised particularly by le Pair Udo and de Groot[6] and is likely tobe especially significant in the case of less flexible base-load plant[18] Ramp-ing emissions were not included in the calculation described above primarilybecause no model parameterisations of this effect are available However inview of the good agreement between reported values and model there is no ev-idence of a large missing source of emissions from ramping (order of magnitudesim 01MtCO2) Emissions from CCGT gas plant in particular seem to be ac-curately described Nevertheless on physical grounds we know such emissionsdo exist and are likely to increase as wind penetration increases

3 Results and Discussion

Fuel-mix plots[13] (Figure 2) show that wind displaces gas generation and toa lesser extent coal Peat and gas CHP plant are operated on a ldquomust runrdquobasis These plant respond to system demand but they are not displaced bywind The CO2-mix plot in Figure 2 is calculated using the emissions modeldescribed above Coal and peat play a far larger role in the CO2 mix For

8

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

example peat provided only 8 of generation but produced 19 of emissionsWind power displaces gas generation and therefore it is not surprising thatperiods of high wind generation are associated with lower emissions from gasplant This can be seen clearly from Figure 5

Figure 5 ndash Left CO2 emission rates for gas coal and peat fuel generators as afunction of total system demand High wind generation is indicated in light blueand low wind generation is shown in dark blue Right Empirical probabilitydistribution for CO2 emission rates for ldquohighrdquo (green) and ldquolowrdquo (pink windgeneration ie lower or higher than the median wind generation The coal andgas plant portfolio is more likely to be found in a low emission state at high windgeneration This effect is far stronger for gas than for coal

Emissions intensity fall linearly with wind penetration A summary of thelinear regression fit is shown in Table 1 The intercept 053tCO2MWh corre-sponds to grid average emissions intensity in the absence of wind The fittedslope -028tCO2MWh corresponds to emissions savings due to wind genera-tion The displacement effectiveness of wind power is therefore just 53 (Ifstart-up emissions are omitted the fit parameters become 052tCO2MWh and03tCO2MWh respectively) If there was no wind in 2011 emissions wouldhave been 129MtCO2 versus 118MtCO2 observed ie a savings of asymp 9

It seems surprising at first sight that emissions savings of 028tCO2MWhis less than the emissions intensity of the cleanest thermal generators on thegrid Figure 4 To understand how this arises it is necessary to drill downand see how individual generators respond to wind generation In general therelationship between an individual generator and total system demand (or total

9

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

Estimate Std Error t value Pr(gt|t|)(Intercept) 05253 00008 67253 00000

Wind Penetration -02788 00035 -7938 00000

Table 1 ndash Fit parameters and standard error for grid emissions intensity(tCO2MWh) vs wind penetration

wind) is non-linear and noisy However since linear behaviour is observed for thegrid as a whole Table 1 it can be assumed that this derives from relationshipsof the form co2g = αgD + βgW at the individual generator level Here co2gis emissions rate of generator g D is system demand and W is wind powerMultiple linear regression yields the fit parameters αg and βg for the 35 thermalgenerators shown in Table Significant negative value of βg indicates thatemissions from generator g are displaced by wind It is clear from Table thatmost of the displacement of thermal plant emissions occurs at just four of themodern base-load gas units (CCGT)

With this information a simple example illustrates how efficiency losses canpush CO2 savings below the average emission intensity of the CCGT generatorswhich are displaced by wind At full-load (1600MW maximum efficiency) av-erage emissions intensity of the four units is 035tCO2MWh according to themodel input-output curves At part-load (770MW minimum stable operatingcapacity) emissions intensity rise to 04tCO2MWh Assume that initially thefour CCGT plant are operating at optimal full-load1600MW and wind gener-ation is zero If wind generation increases to 840MW their combined outputdrops to their minimum operation capacity 770MW The change in emissionsrate per unit wind power in this scenario is (035 x 1600 - 04 x 770)840 =03tCO2MWh This illustrates how efficiency losses alone can lower emissionssavings below the emissions intensity of the cleanest thermal plant

The presence of electricity imports and exports means that grid averageemissions intensity can be expressed in different ways For example carbonintensity can be expressed with respect to total system demand (as above)system demand net of exports (ie ldquodomestic demandrdquo) or system demandnet of imports (ie ldquodomestic generationrdquo) This also affects how wind powersavings are expressed In the case of Ireland in 2011 SEMO ldquoNIJIrdquo data showthat total imports and exports were 3 and 1 of demand respectively sothat the differences between these measures of carbon intensity are relativelysmall In other situations the differences may be substantial With respect toldquodomestic demandrdquo and ldquodomestic generationrdquo zero-wind grid intensity were053tCO2MWh and 055tCO2MWh respectively while wind power savingswere 026tCO2MWh an 030tCO2MWh respectively Corresponding displace-ment effectiveness were 50 and 56

10

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

generator fuel alpha betaADC CCGT 00264 -00502DB1 CCGT 00400 00029HNC CCGT 00317 -00562HN2 CCGT 00388 -00143PBC CCGT 00327 -00027TYC CCGT 00251 -00586WG1 CCGT 00432 -00528SK3 CHP 00076 00031SK4 CHP 00088 00020MP1 coal 00523 -00238MP2 coal 00569 -00414MP3 coal 00564 -00137LR4 peat 00312 -00033ED1 peat 00292 00042WO4 peat 00260 00220AD1 OGCT 00026 00126AT1 OGCT 00014 00048AT2 OGCT 00004 00021AT4 OGCT 00001 00003NW4 OGCT 00000 00000NW5 OGCT 00049 00023MRT OGCT 00012 00007GI1 OGCT 00002 -00005GI2 OGCT 00001 -00005GI3 OGCT 00007 -00020RH1 oil 00001 -00001RH2 oil 00000 -00001TB1 oil 00003 -00008TB2 oil 00002 -00004TB3 oil 00030 -00084TB4 oil 00017 -00051ED3 oil 00002 -00005ED5 oil 00002 -00004TP1 oil 00000 -00001TP3 oil 00000 -00001All 05235 -02788

Table 2 ndash Displacement of generator emissions by wind power using the lin-ear response model The multiple regression fits assume zero intercept To-tals

sumαg represents the grid average intensity and

sumβg represents wind power

savings (tCO2WMh) Most of the emissions reductions occurred at just fourmodern CCGT plants shown in boldface Aghada (commissioned in 2010)Huntstown(2002) Tynagh(2006) and Whitegate (2010) The anti-correlation be-tween wind and emissions at WO4 (West Offaly peat plant) arose because therewas an outage at this plant during the summer when wind generation is lowestThe data suggest that MP2 increased to compensate

11

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

4 Conclusion

As currently deployed wind power is a supplementary power source whose roleis to reduce fossil fuel use by displacing thermal generation The Irish situationis typical in the sense that it has rapidly growing wind penetration embedded ina diverse portfolio of thermal plant A detailed empirical model of operationalCO2 savings was developed for 2011 It is found that savings of 028tCO2MWhwere achieved versus a zero-wind emissions intensity of 053tCO2MWh Thisestimate is at the lower end of expectations[9] In particular it is significantlylower than the emissions intensity of the CCGT plant which play the primaryrole in balancing wind generation Effectiveness is likely to fall further as windpenetration increases[10 11]

Assessments of the economic or environmental benefit of wind power arenot credible unless they are based on accurate emissions (and fuel) savingsThis study suggests that savings may be lower than contemplated by publicagencies to date The Irish government has an ambitious target of meeting 37of domestic electricity demand using wind power by 2020 It is a concern that at17 wind penetration the system is already in a regime where effectiveness isapproaching asymp 50 even before significant curtailment andor exports of windpower begin to occur

Finally life-cycle estimates of CO2 emissions involved in construction andinstallation of wind power are sensitive to assumptions about the capacity factoreconomic life of wind turbines infrastructure requirement etc[19] Estimates arein the range 0002-008 tCO2MWh At the upper end of this range life-cycleemissions are a significant fraction of operational CO2 savings

References

[1] Global Warming The complete briefing Third Edition John HoughtonCambridge University Press 2004

[2] For example ldquoDirective 200928EC on renewable energy implemented byMember States by December 2010 sets ambitious targets for all MemberStates such that the EU will reach a 20 share of energy from renewablesources by 2020 and a 10 share of renewable energy specifically in thetransport sectorrdquo httpeceuropaeuenergyrenewablestargets_

enhtm

[3] Wind power the economic impact of intermittency G Cornelis van KootenLetters in Spatial and Resource Sciences (2010) 3 117

[4] Blowing Away the Myths British Wind Energy Association Report February2005 p 4 recommended use of emissions savings of 086tCO2MWhhttpwwwbweacompdfref_threepdf BWEA currently recommend useof a ldquogrid averagerdquo number 043tCO2MWh httpwwwbweacomedu

calcshtml

12

[5] Emissions savings from wind power generation Evidence from Texas Cali-fornia and the Upper Midwest Daniel T Kaffine Brannin J McBee JozefLieskovsky Working Paper Colorado School of Mines

[6] Wind turbines as yet unsuitable as electricity providers C le Pair F Udoand K de Groot Europhysics News Vol 43 No 2 2012 pp 22-25

[7] Why wind power does nor deliver the expected emissions reductions HInnhaber Renewable and Sustainable Energy Reviews 15(2011) 2557-2562

[8] Wind Power in the UK Sustainable Development Commission Novem-ber 2005 SDC used a wind savings value 035tCO2MWh (ie basedon CCGT emissions) httpwwwsd-commissionorgukdatafiles

publicationsWind_Energy-NovRev2005pdf

[9] Impact of Wind Power Generation In Ireland on the Operation of Conven-tional Plant and the Economic Implications ESB National Grid February2004

[10] System-Wide Emissions Implications of Increased Wind Power Penetra-tion Lauren Valentino Viviana Valenzuela Audun Botterud Zhi Zhou andGuenter Conzelmann Environmental Science and TechnologyEnviron SciTechnol 2012 46 (7) pp 42004206

[11] Quantifying the Total Net Benets of Grid Integrated Wind E Denny andM OMalley IEEE Transactions on Power Systems 2007 22 (2) p 605-615

[12] The EU Emissions Trading System (EU ETS) httpeceuropaeu

climapoliciesg-gasindex_enhtm

[13] wwweirgridcom

[14] Sustainable Energy Authority of Ireland Provisional Energy Balance2011 httpwwwseaiiePublicationsStatistics_Publications

Energy_Balance

[15] Single Electricity Market Operator httpwwwsemo-ocom

[16] SEM-11-052 2011-12 Validated SEM Generator Data Parameters PUBLICv1xls Single Electricity Market Operator document

[17] All-Island Wind and Fuel Mix Summary Report 2011 Eirgridand SONI operations httpwwweirgridcomoperations

systemperformancedataall-islandwindandfuelmixreport

[18] How less became moreWind power and unintended consequencesin the colorado energy market httpwwwbentekenergycom

WindCoalandGasStudyaspx

[19] Life Cycle Greenhouse Gas Emissions of Utility-Scale Wind Power StaceyL Dolan and Garvin A Heath Journal of Industrial Ecology 16(s1) p136-154 April (2012)

13

Source

IDNam

eFuel

Min

Cap

MaxCap

Ram

pUp

Ram

pD

own

Cold

Start

Warm

Start

HotStart

HotToW

arm

Warm

ToCold

Aughin

ish

SK

3Sealrock

3(Aughin

ish

CHP)

Gas

400

0830

060

060

012000

010000

08000

080

0240

0Aughin

ish

SK

4Sealrock

4(Aughin

ish

CHP)

Gas

400

0830

060

060

012000

010000

08000

080

0240

0Edenderry

ED

1Edenderry

PeatBio

mass

410

01176

017

617

623080

010840

04360

040

0480

0ESBPG

AA1

Ardnacrusha

Unit

1120

0210

060

060

000

000

000

0120

0480

0ESBPG

AA2

Ardnacrusha

Unit

2120

0220

060

060

000

000

000

0120

0480

0ESBPG

AA3

Ardnacrusha

Unit

3120

0190

060

060

000

000

000

0120

0480

0ESBPG

AA4

Ardnacrusha

Unit

4120

0240

060

060

000

000

000

0120

0480

0ESBPG

AD

1Aghada

Unit

1G

as

360

02580

051

051

043020

021850

012730

050

0670

0ESBPG

AT1

Aghada

CT

Unit

1G

as

150

0900

050

050

0630

0630

0630

0120

0480

0ESBPG

AT2

Aghada

CT

Unit

2G

as

150

0900

050

050

0630

0630

0630

0120

0480

0ESBPG

AT4

Aghada

CT

Unit

4G

as

150

0900

050

050

0630

0630

0630

0120

0480

0Endesa

GI1

Great

Isla

nd

Unit

1O

il250

0540

007

010

05620

04490

02180

0120

0480

0Endesa

GI2

Great

Isla

nd

Unit

2O

il250

0490

007

010

05620

04490

02180

0120

0480

0Endesa

GI3

Great

Isla

nd

Unit

3O

il300

01130

007

015

07430

06000

02930

0120

0600

0ESBPG

MP1

Moneypoin

tUnit

1FG

DSCR

Coal

990

02850

030

040

0146200

069200

043600

0120

0600

0ESBPG

MP2

Moneypoin

tUnit

2FG

DSCR

Coal

990

02850

030

040

0146200

069200

043600

0120

0600

0ESBPG

MP3

Moneypoin

tUnit

3FG

DSCR

Coal

990

02850

030

040

0146200

069200

043600

0120

0600

0ESBPG

MRT

Marin

aNo

ST

Gas

200

0880

050

050

0630

0630

0630

0120

0280

0ESBPG

NW

4Northwall

Unit

4G

as

873

01630

037

8120

0800

0800

0800

0120

0600

0ESBPG

NW

5Northwall

Unit

5G

as

40

01040

080

080

0500

0500

0500

0120

0480

0ESBPG

PBC

Poolb

eg

Com

bin

ed

Cycle

Gas

2320

04800

0110

0110

028000

018000

015000

080

01120

0ESBPG

WO

4W

est

Offaly

Power

Peat

450

01370

007

007

07500

06000

04500

0120

0600

0Synergen

DB1

Dublin

Bay

Power

Gas

2090

04150

0100

090

077000

026040

026000

0120

0600

0Tynagh

TYC

Tynagh

Gas

1940

03885

0125

0125

050330

029440

020170

0120

0480

0Virid

ian

HN2

Huntstown

Phase

II

Gas

1940

04040

0200

0200

06440

05310

03180

0120

0600

0Virid

ian

HNC

Huntstown

Gas

1840

03430

070

070

047720

028030

08350

0120

0600

0ESBPG

AD

CAghada

CCG

TG

as

2130

04350

0124

3124

324000

018000

012000

0120

0600

0ESBPG

LR4

Lough

Rea

Peat

680

0910

013

713

75000

04000

03000

0120

0480

0Bord

Gais

WG

1W

hitegate

Gas

1800

04450

0300

0300

04370

03300

02230

0120

0600

0Endesa

RH1

Rhode

1D

istilla

te

120

0520

050

0100

0240

0240

0240

0120

0600

0Endesa

RH2

Rhode

2D

istilla

te

120

0520

050

0100

0240

0240

0240

0120

0600

0Endesa

TB1

Tarbert

Unit

1O

il200

0540

010

010

05620

04490

02180

0120

0600

0Endesa

TB2

Tarbert

Unit

2O

il200

0540

010

010

05620

04490

02180

0120

0600

0Endesa

TB3

Tarbert

Unit

3O

il349

02400

017

022

031800

019340

010720

0140

01200

0Endesa

TB4

Tarbert

Unit

4O

il349

02400

017

022

031800

019340

010720

0140

01200

0Endesa

TP1

Tawnaghm

ore

1D

istilla

te

120

0520

050

0100

0240

0240

0240

0120

0600

0Edenderry

ED

3Cushaling

Distilla

te

200

0580

050

050

0200

0200

0200

005

010

0Edenderry

ED

5Cushaling

Distilla

te

200

0580

050

050

0200

0200

0200

005

010

0Endesa

TP3

Tawnaghm

ore

3D

istilla

te

120

0520

050

0100

0240

0240

0240

0120

0600

0ESBPG

ER1

Erne

Unit

140

0100

050

0100

000

000

000

0120

0480

0ESBPG

ER2

Erne

Unit

240

0100

050

0100

000

000

000

0120

0480

0ESBPG

ER3

Erne

Unit

350

0225

0100

0225

000

000

000

0120

0480

0ESBPG

ER4

Erne

Unit

450

0225

0100

0225

000

000

000

0120

0480

0ESBPG

LE1

Lee

Unit

130

0150

024

0150

000

000

000

0120

0480

0ESBPG

LE2

Lee

Unit

210

040

006

040

000

000

000

0120

0480

0ESBPG

LE3

Lee

Unit

330

080

015

080

000

000

000

0120

0480

0ESBPG

LI1

Liffe

yUnit

130

0150

050

0100

000

000

000

0120

0480

0ESBPG

LI2

Liffe

yUnit

230

0150

050

0100

000

000

000

0120

0480

0ESBPG

LI4

Liffe

yUnit

404

040

020

020

000

000

000

0120

0480

0ESBPG

LI5

Liffe

yUnit

502

040

000

320

000

000

000

0120

0480

0ESBPG

TH1

Turlo

ugh

HillUnit

150

0730

02100

02700

000

000

000

0120

0480

0ESBPG

TH2

Turlo

ugh

HillUnit

250

0730

02100

02700

000

000

000

0120

0480

0ESBPG

TH3

Turlo

ugh

HillUnit

350

0730

02100

02700

000

000

000

0120

0480

0ESBPG

TH4

Turlo

ugh

HillUnit

450

0730

02100

02700

000

000

000

0120

0480

0

Table

3ndash

(a)

The

50

gri

d-c

onnec

ted

non-w

ind

gen

erato

rsuse

din

this

study

Min

Cap

and

MaxC

ap

are

min

imum

and

maxiu

mum

stable

op

erati

ng

capaci

ties

(MW

)R

am

pU

pand

Ram

pD

own

are

maxiu

mum

ram

pup

rate

s(M

Wm

in)

Cold

Sta

rt

Warm

Sta

rtand

HotS

tart

are

ther

mal

state

dep

enden

tst

art

up

ener

gy

cost

s(G

J)

HotT

oW

arm

and

Warm

ToC

old

are

cooling

dow

nti

mes

(hours

)A

dapte

dfr

om

Ref

[1

6]

14

IDNam

eFuel

NoLoad

Cap1

Cap2

Cap3

Cap4

IHRS01

IHRS12

IHRS23

IHRS34

SK

3Sealrock

3(Aughin

ish

CHP)

Gas

1000

0400

0830

050

050

0SK

4Sealrock

4(Aughin

ish

CHP)

Gas

1000

0400

0830

050

050

0ED

1Edenderry

PeatBio

mass

4976

0880

01120

01180

039

389

589

5AD

1Aghada

Unit

1G

as

1828

31200

01900

02580

078

184

585

2AT1

Aghada

CT

Unit

1G

as

2950

5400

0900

081

0100

5AT2

Aghada

CT

Unit

2G

as

2950

5400

0900

081

0100

5AT4

Aghada

CT

Unit

4G

as

2950

5400

0900

081

0100

5G

I1G

reat

Isla

nd

Unit

1O

il510

7250

0450

0540

0135

9135

9136

7G

I2G

reat

Isla

nd

Unit

2O

il510

7250

0450

0490

0135

9135

9136

7G

I3G

reat

Isla

nd

Unit

3O

il1026

5300

0980

01130

0108

8108

8109

8M

P1

Moneypoin

tUnit

1FG

DSCR

Coal

1737

71950

02800

02850

094

895

8140

9M

P2

Moneypoin

tUnit

2FG

DSCR

Coal

1737

71950

02800

02850

094

895

8140

9M

P3

Moneypoin

tUnit

3FG

DSCR

Coal

1737

71950

02800

02850

094

895

8140

9M

RT

Marin

aNo

ST

Gas

2588

4470

0810

0880

086

694

8114

1NW

4Northwall

Unit

4G

as

3513

4870

01150

01620

060

768

482

3NW

5Northwall

Unit

5G

as

3336

1500

01040

090

397

0PBC

Poolb

eg

Com

bin

ed

Cycle

Gas

4521

14800

05100

061

369

9W

O4

West

Offaly

Power

Peat

1186

71370

090

0D

B1

Dublin

Bay

Power

Gas

4793

42030

04150

050

051

6TYC

Tynagh

Gas

6040

01940

02500

04040

040

054

657

4HN2

Huntstown

Phase

II

Gas

6036

01950

02300

04120

040

056

257

4HNC

Huntstown

Gas

5412

02000

02300

02500

03520

040

051

959

960

1AD

CAghada

CCG

TG

as

5923

92130

04300

04350

040

054

055

6LR4

Lough

Rea

Peat

852

9910

086

5W

G1

Whitegate

Gas

6666

81850

03990

04600

040

052

661

1RH1

Rhode

1D

istilla

te

850

1120

0520

098

298

2RH2

Rhode

2D

istilla

te

850

1120

0520

098

298

2TB1

Tarbert

Unit

1O

il446

6200

0460

0540

0116

3116

3117

5TB2

Tarbert

Unit

2O

il446

6200

0460

0540

0116

3116

3117

5TB3

Tarbert

Unit

3O

il2476

1350

01000

01800

02400

080

780

790

691

5TB4

Tarbert

Unit

4O

il2476

1350

01000

01800

02400

084

084

094

396

4TP1

Tawnaghm

ore

1D

istilla

te

866

2120

0520

0100

095

9ED

3Cushaling

Distilla

te

850

0580

090

0ED

5Cushaling

Distilla

te

850

0580

090

0TP3

Tawnaghm

ore

3D

istilla

te

866

2120

0520

0100

095

9

Table

4ndash

(b)

Addit

ional

para

met

ers

for

35

gri

d-c

onnec

ted

ther

mal

gen

erato

rsuse

din

this

study

NoL

oad

isth

eex

trap

ola

ted

zero

load

(Cap0)

ener

gy

use

(in

GJh)

Capaci

tyP

oin

ts(C

ap1

etc

MW

)are

use

dto

spec

ify

the

hea

tra

tecu

rve

IHR

Sva

lues

are

Incr

emen

tal

Hea

tR

ate

Slo

pes

(GJM

Wh)

defi

ned

bet

wee

nea

chpair

of

capaci

typ

oin

ts

Thes

edata

are

use

dto

crea

test

ati

cC

O2

emis

sions

vs

gen

erati

on

curv

es(C

O2h

vs

MW

)or

ldquoin

put-

outp

utrdquo

curv

esfo

rea

chgen

erato

rA

dapte

dfr

om

Ref

[1

6]

15

• Outline of the problem
• Data and model
• Results and Discussion
• Conclusion

IDNam

eFuel

NoLoad

Cap1

Cap2

Cap3

Cap4

IHRS01

IHRS12

IHRS23

IHRS34

SK

3Sealrock

3(Aughin

ish

CHP)

Gas

1000

0400

0830

050

050

0SK

4Sealrock

4(Aughin

ish

CHP)

Gas

1000

0400

0830

050

050

0ED

1Edenderry

PeatBio

mass

4976

0880

01120

01180

039

389

589

5AD

1Aghada

Unit

1G

as

1828

31200

01900

02580

078

184

585

2AT1

Aghada

CT

Unit

1G

as

2950

5400

0900

081

0100

5AT2

Aghada

CT

Unit

2G

as

2950

5400

0900

081

0100

5AT4

Aghada

CT

Unit

4G

as

2950

5400

0900

081

0100

5G

I1G

reat

Isla

nd

Unit

1O

il510

7250

0450

0540

0135

9135

9136

7G

I2G

reat

Isla

nd

Unit

2O

il510

7250

0450

0490

0135

9135

9136

7G

I3G

reat

Isla

nd

Unit

3O

il1026

5300

0980

01130

0108

8108

8109

8M

P1

Moneypoin

tUnit

1FG

DSCR

Coal

1737

71950

02800

02850

094

895

8140

9M

P2

Moneypoin

tUnit

2FG

DSCR

Coal

1737

71950

02800

02850

094

895

8140

9M

P3

Moneypoin

tUnit

3FG

DSCR

Coal

1737

71950

02800

02850

094

895

8140

9M

RT

Marin

aNo

ST

Gas

2588

4470

0810

0880

086

694

8114

1NW

4Northwall

Unit

4G

as

3513

4870

01150

01620

060

768

482

3NW

5Northwall

Unit

5G

as

3336

1500

01040

090

397

0PBC

Poolb

eg

Com

bin

ed

Cycle

Gas

4521

14800

05100

061

369

9W

O4

West

Offaly

Power

Peat

1186

71370

090

0D

B1

Dublin

Bay

Power

Gas

4793

42030

04150

050

051

6TYC

Tynagh

Gas

6040

01940

02500

04040

040

054

657

4HN2

Huntstown

Phase

II

Gas

6036

01950

02300

04120

040

056

257

4HNC

Huntstown

Gas

5412

02000

02300

02500

03520

040

051

959

960

1AD

CAghada

CCG

TG

as

5923

92130

04300

04350

040

054

055

6LR4

Lough

Rea

Peat

852

9910

086

5W

G1

Whitegate

Gas

6666

81850

03990

04600

040

052

661

1RH1

Rhode

1D

istilla

te

850

1120

0520

098

298

2RH2

Rhode

2D

istilla

te

850

1120

0520

098

298

2TB1

Tarbert

Unit

1O

il446

6200

0460

0540

0116

3116

3117

5TB2

Tarbert

Unit

2O

il446

6200

0460

0540

0116

3116

3117

5TB3

Tarbert

Unit

3O

il2476

1350

01000

01800

02400

080

780

790

691

5TB4

Tarbert

Unit

4O

il2476

1350

01000

01800

02400

084

084

094

396

4TP1

Tawnaghm

ore

1D

istilla

te

866

2120

0520

0100

095

9ED

3Cushaling

Distilla

te

850

0580

090

0ED

5Cushaling

Distilla

te

850

0580

090

0TP3

Tawnaghm

ore

3D

istilla

te

866

2120

0520

0100

095

9

Table

4ndash

(b)

Addit

ional

para

met

ers

for

35

gri

d-c

onnec

ted

ther

mal

gen

erato

rsuse

din

this

study

NoL

oad

isth

eex

trap

ola

ted

zero

load

(Cap0)

ener

gy

use

(in

GJh)

Capaci

tyP

oin

ts(C

ap1

etc

MW

)are

use

dto

spec

ify

the

hea

tra

tecu

rve

IHR

Sva

lues

are

Incr

emen

tal

Hea

tR

ate

Slo

pes

(GJM

Wh)

defi

ned

bet

wee

nea

chpair

of

capaci

typ

oin

ts

Thes

edata

are

use

dto

crea

test

ati

cC

O2

emis

sions

vs

gen

erati

on

curv

es(C

O2h

vs

MW

)or

ldquoin

put-

outp

utrdquo

curv

esfo

rea

chgen

erato

rA

dapte

dfr

om

Ref

[1

6]

15

• Outline of the problem
• Data and model
• Results and Discussion
• Conclusion