REPORTOFTHEMINISTERIALADVISORYCOUNCILONENERGY(MACE)WORKINGGROUPON
ANALYSISANDRECOMMENDATIONSONTHEASSUMPTIONSANDMETHODOLOGIESADOPTEDIN
THEIRP2016BASECASESCENARIO
31October2016
MembersoftheWorkingGroup
MikeLevington(Chairperson)
ProfessorAntonEberhard
Dr.TobiasBischof-Niemz
ProfessorJohanvanDyk
ExecutiveSummary
The Minister of Energy appointed a Working Group on 16 September 2016 to analyse certain
concernsexpressedbymembersoftheMinisterialAdvisoryCouncilonEnergy(MACE)onanumber
ofassumptionsusedtoderivetheIRPScenario“BaseCaseA1”andtoreportbacktotheMinisteron
theirfindings.
TheWorkingGrouphasrequestedcertaininformationfromtheDepartmentofEnergy(DOE),which
was partially received. Based on this information, the Working Group’s findings and its
recommendationstotheMinisterareasfollows:
Mainfinding:
A least cost IRP model, free of any artificial constraints and before any policy adjustments does
notincludeanynewnuclearpowergenerators.TheoptimalleastcostmixisoneofsolarPV,wind
andflexiblepowergenerators(withrelativelylowutilisation).
RecommendationsoftheMACEWorkingGroup:
Recommendation1
ConsistentwiththeapproachusedinIRP2010,thescenariothatformstheBaseCasemustbe
leastcostandfreeofanypolicyadjustments.TheWorkingGroupthereforerecommendsthat
theannualnew-buildlimitsimposedonsolarPVandwindareremovedandthisunconstrained
scenario(presentedinTable2inthisdocument)formstheBaseCasefortheIRP2016.
Recommendation2
The input costs assumed for solar PV and wind in the IRP2016 are significantly higher (in real
terms)thanwhatwasassumedintheIRP2010,despitethefactthattariffsactuallyachievedin
the Renewable Energy IPP Procurement Programme (REIPPPP) are lower than what IRP2010
1
had assumed. This apparently is a result of technical mistakes made when converting average
tariffs achieved in Bid Window 4 of the REIPPPP into model input costs, combined with a
reduced cost reduction potential for solar PV compared to IRP2010. It is therefore
recommendedtoadjustthecurrentlyassumedcostsofbothsolarPVandwinddownwardsto
correctlyreflectSouthAfricanactualtariffsaswellasanticipatedcostreductionsasperIRP2010
inthecaseofsolarPV.1
Recommendation3
Any policy-adjustment to or the imposing of a constraint on the least-cost unconstrained Base
Case will increase the total cost of the power system and therefore the average tariff to the
consumer. The Working Group recommends that the cost differences between the least-cost
unconstrainedBaseCaseandeachalternativescenariosbereportedsothatavalueformoney
case can be assessed by all stakeholders when a certain policy decision or a constraint is
proposed.
CommentsreceivedfromMACEmembers
MACEWorkingGrouprecommendations
Recommendation1
Least cost should be adopted as the base case
beforeanypolicy-relatedconstraints
Recommendation2
IRP input costs for solar and PV are higher than
they should be and the IRP should adopt lower
cost estimates which are proven through
contractualagreementsrecentlyconcluded
MACEcomments
No MACE member has argued against the least
costmodelrunasbeingtheadoptedBaseCase
NoMACEmemberhascontestedthis
• Somehavequestionedthecomparability
of some prices, but not in a way that
detracts from the Working Group's
recommendations.
• There is also a call for the use of a
consistent
methodology
of
benchmarking prices when deviating
fromstandardpricingmodels.
Recommendation3
The cost differential between any policy-related No MACE member has contested this
constrained scenarios and the base case should recommendation.
be clearly documented and taken into account
bydecision-makers
Conclusion
TheWorkingGroupisoftheviewthatiftheaboverecommendationsareappliedtotheIRPprocess,
itwillresultinamethodologythatisconsistent,willallowtheenergyplannerstoachievethemost
efficientpricepathandwillleadtoanoutcomethatMACEasawholewillbeabletoendorseand
defend in the public participation phase of the IRP to current and future electricity consumers in
SouthAfrica.
1
ItisworthnotingthatnothwithstandingtheneedtorecalibratetherelativecostingofsolarPVandwind,the
least-costBaseCasestilldoesnotincludeanynewnuclearpowerplants.
2
ThisprocesshasraisedanumberofotherissuesinrespectoftheIRPwhichshouldreceiveattention
infuture:
a) IRP treatment of the risk associated with cost and time overruns exhibited in the
constructionofmegaprojects.
b) Treatmentofgridconstraintsatbothmunicipalandnationallevels.
c) Introduction ofa full macro-economic cost-benefit analysis to assess socioeconomic tradeoffs(employment,localmanufacturingetc.)
END
3
1 KeyrelevantextractsfromIRP2010
IRP2010,page62
“Toaccountfortheuncertaintiesassociatedwiththecostsofrenewablesandfuels,anuclearfleetof
9,6GWisincludedintheIRP;”
Themain"policyadjustment"awayfromthemathematicalleast-costoptionintheIRP2010wasthe
introduction of a 9.6 GW nuclear fleet to cater for uncertainties (at the time) around the (at the
time)forecastedcostreductionsofrenewables.
IRP2010,page10
1) In 2010, a number of scenarios were tested, based on the input assumptions that were
applicableatthattime.Allscenarioswereoptimisedtobe"leastcost"withintheboundary
conditionsimposedonthem.
2) The "Base Case" was the least-cost scenario without a CO2-emission limit imposed on the
model. It included neither nuclear new-build nor renewables. The "Base Case" was almost
exclusivelycoal-basedandreflectedthestatusquo.
3) The"EmissionLimit"scenariointroducedaCO2-emissionlimittothemodelandasaresult
themodelbuiltanuclearfleetof9.6GW(6x1.6GWunits)forthepowersystemtobeable
tostaywithintheimposedCO2-emissionlimit.
4) TheRevisedBalancedScenario(RBS)wasthesynthesisofallleast-costoptimisedscenarios
asofOctober2010.Itincludedanuclearfleetof9.6GW.
5) AsecondroundofpublicconsultationswasthenconductedfromOctober2010toJanuary
2011.Followingthepublicinput,nuclearandrenewablescostswereadjusted(nuclearup,
2
http://www.energy.gov.za/IRP/irp%20files/IRP2010_2030_Final_Report_20110325.pdf
4
renewables down) and solar technologies were disaggregated. The revised nuclear cost
assumptionwas5000$/kW.
6) Asaresult,thenewleast-costbasecasewasnowthescenario"AdjustedEmission",which
still had the same CO2-emission limit as per the RBS, but did not include the nuclear fleet
anymore. The nuclear fleet of the RBS was replaced by a mix of solar PV, wind, CSP and
naturalgas.
7) Themodelonlybuiltnuclearinthetwoscenarios"RiskAverse"and"PeakOil".
a. The "Risk Averse" scenario limited the amount of allowable imported electricity,
whichinreturnrequiredmoredomesticpowergeneration.Becausewindandsolar
PVwereartificiallylimitedto1.6GW/aand1GW/aspeedofroll-outandto10GW
oftotalinstalledcapacityeach,themodelhadnootherchoicebuttobuildnuclear
astheonlydomestic,carbon-neutralsourceofelectricity(moredomesticcoalwas
notanoptionforthemodelbecauseoftheCO2limit).
b. The"PeakOil"scenarioassumedhighercostofcoal,whichmadebothnuclearand
renewablescostcompetitivetonewcoal.Thereforelessnewcoalplantswerebuilt
in this scenario. The gap in electricity could not be filled with wind and solar PV
(becauseoftheannualnew-buildlimitsandthetotalcapacitycap),andhencethe
modelhadnootherchoicebuttobuildnuclear.
8) Thesynthesisofthenewleast-costscenariosfromFebruary2011,whichallimposedaCO2emission limit, was the Policy-Adjusted IRP. The policy adjustment was to keep a 9.6 GW
nuclear fleet in the plan (as per RBS), despite the fact that the new least-cost scenarios
"AdjustedEmission","HighEfficiency","LowGrowth"and"EarlierCoal"alldidnotplanany
newnuclear,andtheonlytwoscenarioswithanuclearfleetwerethosewhererenewables
ranintoartificiallyimposednew-buildconstraints.
QuoteIRP2010,page11
“PolicyIssue1:Nuclearoptions
4.2
Thescenariosindicatedthatthefuturecapacityrequirementcould,intheory,bemet
without nuclear, but that this would increase the risk to security of supply (from a
dispatchpointofviewandbeingsubjecttofuturefueluncertainty).
4.3
Threepolicychoiceoptionswereidentified:
4.4
a)
CommittothenuclearfleetasindicatedintheRBS;
b)
Delaythedecisiononthenuclearfleetindefinitely(andallowalternativesto
beconsideredintheinterim);
c)
Committotheconstructionofoneortwonuclearunitsin2022-4,butdelaya
decision on the full nuclear fleet until higher certainty is reached on future
costevolutionandriskexposurebothfornuclearandrenewables.
The Department accepted option 4.3a, committing to a full nuclear fleet of 9600
MW.Thisshouldprovideacceptableassuranceofsecurityofsupplyintheeventofa
peakoil-typeincreaseinfuelpricesandensurethatsufficientdispatchablebase-load
capacityisconstructedtomeetdemandinpeakhourseachyear.”
5
IRP2010,page41
Conclusion:
•
•
•
•
InIRP2010,nuclearwasnottheleast-costoption.
Itwasapolicydecisiontoincludenuclearintheplantocaterforuncertaintiesaroundthe
forecastedcostreductionofrenewables,asatthetimeitwasunclearwhethertheywould
materialiseinthemagnitudeandasquicklyasanticipatedintheIRP2010.
Annualnew-buildlimitswereimposedonsolarPVandwind,inordertolimitthebuild-out
ofthesetwotechnologies.
Theseadjustmentsledtothemodelbuildingnuclearunderveryspecificconstraints,where
theamountofrequiredCO2-neutralelectricitycouldnotbesuppliedentirelybyrenewables
becauseoftheseannualnew-buildlimits.
6
2 IRP2016
Twomainareasofconcernarosefromthepresentationofthedraftassumptionsandresultsatthe
MACEmeetingon16September2016,andfromthesubsequent“assumptionsmeeting”attheDOE
on14October2016.Thesewere:
1) CostassumptionsforsolarPVandwind;and
2) Annualnew-buildlimitsforsolarPVandwind.
2.1 CostAssumptions
ThefigurebelowshowsthecostassumptionspathforsolarPVintheIRP2010/2013aswellasthe
cost-assumptions path for IRP 2016, derived from the assumptions presented to MACE. It
furthermoreshowstheactualresultsoftheREIPPPPforthefirstfourBidWindows(aswellasthe
ExpeditedRound).
R/kWh
(Apr-2016-Rand)
4.0
3.6
Assumptions:IRP2010- high
3.5
Assumptions:IRP2010- low
Assumptions:IRP2013- high
3.0
2.5
Assumptions:IRP2013- low
Assumptions:IRP2016- high
2.2
Assumptions:IRP2016- low
Actuals:REIPPPP(BW1-4)
2.0
1.5
1.2
0.9
1.0
0.5
0.6
0.0
2010
2012
2014
2016
2018
2020
2022
2024
2026
2028
2030
Assumptions:CPIusedfornormalisationtoApr-2016-Rand;LCOEcalculatedforIRPwith8%discountrate(real),25yrs lifetime,costandloadfactorassumptionsasperrelevantIRPdocument;
“IRPTariff”thencalculatedassuming90%oftotalprojectcoststobeEPCcosts,i.e.dividetheLCOEby0.9toderiveatthe“IRPTariff”
Sources:IRP2010;IRP2013;http://www.ipprenewables.co.za/gong/widget/file/download/id/279;IRP2016draftasofSeptember2016
BothIRP2010andIRP2013forecastasignificantcostreductionforsolarPVuntil2030.Actualcosts
ofsolarPV(derivedfromactualtariffsachievedintheREIPPPP)approachedtheforecastcostpathof
IRP2010/2013veryquicklyandarenowsignificantlylowerthanthecost-assumptionfunnelforIRP
2010/2013.
TheIRP2016effectivelyincreasesthecostassumptionsforsolarPVbya)choosingahigherstarting
pointforsolarPVcostsin2015andb)assumingamuchlowercostreductionratefromtodayuntil
2030thanwhatIRP2010/2013hadassumed.ItisillogicaltoassumehigherpricesforPVthantheIRP
2010assumed,whileinrealitypriceshavefallensubstantially,asshownbythelatestREIPPPprices.
7
Recommendation:
ItisadvisedtokeepthesolarPVcostassumptionsexactlythesameaswasassumedinIRP2010
and IRP 2013, with the only difference that the cost funnel should be adjusted downwards for
early years such that actual tariffs from BW4 Expedited lie at the lower boundary of the cost
funnelin2015.
Thenextfigureshowsthecost-assumptionspathforwindintheIRP2010/2013aswellasthecost
assumptions path for IRP 2016, derived from the assumptions presented to MACE. It furthermore
showstheactualresultsoftheREIPPPPforthefirstfourBidWindows.
R/kWh
(Apr-2016-Rand)
Assumptions:IRP2010
Assumptions:IRP2013- high
1.75
Assumptions:IRP2013- low
1.5
Assumptions:IRP2016- high
1.50
Assumptions:IRP2016- low
Actuals:REIPPPP(BW1-4)
1.25
1.2
1.00
0.9
0.75
0.7
0.6
0.50
0.25
0.00
2010
2012
2014
2016
2018
2020
2022
2024
2026
2028
2030
Assumptions:CPIusedfornormalisationtoApr-2016-Rand;LCOEcalculatedforIRPwith8%discountrate(real),20yrs lifetime,costandloadfactorassumptionsasperrelevantIRPdocument;
“IRPTariff”thencalculatedassuming90%oftotalprojectcoststobeEPCcosts,i.e.dividetheLCOEby0.9toderiveatthe“IRPTariff”
Sources:IRP2010;IRP2013;http://www.ipprenewables.co.za/gong/widget/file/download/id/279;IRP2016draftasofSeptember2016
Both IRP 2010 and IRP 2013 forecast a relatively moderate cost reduction for wind until 2030.
IRP2013madeamistakeinthatitincreasedthecostassumptionsforwindsignificantlycomparedto
IRP2010.
Actualcostsofwind(derivedfromactualtariffsintheREIPPPP)approachedtheforecastcostpathof
IRP2010/2013veryquicklyandthemostlatestwindcost(ExpeditedRound)isalreadywellbelow
theforecastcostassumptionofIRP2010fortheyear2030.
TheIRP2016moreorlessrevertsbacktotheIRP2013costassumptionpathforwind,whichwas
toohigh.
Recommendation:
It is advised to reduce the cost assumption for wind to the actual achieved average tariff of Bid
Window4Expeditedandtokeepitconstantatthatleveluntil2030.
8
2.2 AnnualNew-buildLimitsforSolarPVandWind
Table1belowshowsthecurrentbasecaseoftheIRP2016assenttotheMACEWorkingGroup.Itis
similartothe“BaseCaseA”presentedtoMACEon16September2016.Ascanbeseenfromthat
table,themodelisabletoremaingenerallywithintheCO2capof275milliontonnesperyearuntil
thelate2030s(whichisthetimewhentheCO2capstartstoreducefromitsplateauintodecline).It
even allows new coal to be built. One can also see that the wind and solar PV build-out rates are
running into a limitation from 2030 onwards. The exact new-build limit for both technologies per
year is unknown to the authors at the time of writing. The annual energy balance as well as the
annualtotalcostsarealsounknowntotheauthors.
TheexperiencefromthefirstfourroundsoftheREIPPPPaswellastheglobalannualdeploymentof
windandsolarPVindicatethatitisillogicaltoimposeannualbuildlimitstosolarPVandwind,and,
moreso,toonlylimitthesetwotechnologiesandnoneoftheothers.
9
Table2 shows a scenario of the IRP 2016 that was only sent to the MACE Working Group. It was
requested by MACE on 16 September 2016 but not presented to the full MACE. This scenario
imposesnoannualnew-buildlimits(neitherforsolarPVnorforwind,norforanyothertechnology
forthatmatter).Ascanbeseenfromthattable,themodelhasnoproblemstoremainwithinthe
CO2capof275milliontonnesperyear.Infact,from2028onwardsthemodeliswellbelowtheCO2
capandtheannualCO2emissionsevengodownto155milliontonnesperyearby2050,i.e.they
are 40 million tonnes per year lower than in the scenario of Table1. This is achieved without any
nuclearnewbuild.Theannualenergybalanceaswellastheannualtotalcostsareunknowntothe
authors.
10
NeithertheannualdeploymentofsolarPVandwindnorthetotaldeploymentofsolarPVandwind
in this scenario are unrealistic in South African or in global terms. The planned solar PV and wind
penetrationinthisscenarioby2050isinlinewithactualpenetrationlevelsofanumberofcountries
today.
Conclusions:
1) ItisclearfromtheanalysisofthetwopresentedscenariosthatthemixofsolarPV,wind
andnaturalgasisthecheapestnew-buildmix(refertoTable2),evenwithincorrectlyhigh
costassumptionsforsolarPVandwind(seeconcernnumber1)
2) Inan“unconstrained”casewithoutannualnew-buildlimitsforsolarPVandwind(referto
Table2), the model only builds solar PV, wind and gas-fired power stations. Only small
11
amountsofnewcoal(andthatonlybecauseofthetoohighcostassumptionsforsolarPV
andwind)arebuilt.Nonewnuclearisbuilt
3) Thisunconstrainedleast-costmixleadstolowerCO2emissionsthantheCO2cap.Thecap
isthereforeirrelevantandmathematicallyit'snotaconstraintanymore
4) Specificsoftheconstrainedmodel(Table1):
a. Intheconstrainedmodel,themodelcannotmeetallthedemandwithwind/PV/gas
alone(becauseoftheannualsolarPVandwindnew-buildlimits).Itthereforebuilds
thesecondcheapestoption:coal
b. ThemodelcandothatupuntilitrunsintotheCO2cap.Fromthereonitthenhasto
buildnuclearastheonlyremainingCO2-neutraloption
c. ThetimingofthereductionoftheCO2cap,theleveloftherenewablesannuallimits
and the projected demand post 2030 together determine how much nuclear the
modelbuildsandwhen.
Recommendation:
Itisadvised:
•
•
•
•
Theannualnew-buildlimitsforsolarPVandwindshouldberemoved
This unconstrained scenario (Table 2), re-run with correct solar PV and wind cost
assumptions,shouldformtheleast-costBaseCaseofIRP2016
Thetotalcostperscenarioperyearshouldbecalculated
Step by step constraints or policy adjustments scenarios (for example annual new-build
limitsforsolarPVandwind)shouldalsobefinanciallymodelledandthetotalcostperyear
ofsuchconstrainedscenarioscomparedwiththerevisedleast-costBaseCasetoassessthe
costeffectivenessofsuchinterventions
12