EnergySystemsResearchGroup-WorkingPaper

Evolution,AssumptionsandArchitectureoftheSouthAfricanEnergySystemsModelSATIM

Authors:AlisonHughes,BrunoMerven,BryceMcCall,FadielAhjum,TaraCaetano,FaaiqaHartley,GregoryIreland,JesseBurton,AndrewMarquard,

Abstract:

SouthAfrica facesmany energy challenges and opportunities. Amongst the challenges are constraints toelectricitysupplyandarisingcostofelectricity toconsumers,high levelsofenergypoverty,andaheavyreliance on coal in the supply and demand sectors. Energy systems models are useful tools to exploredevelopment pathways that address these and other challenges. This paper describes SATIM, the SouthAfricanTIMESmodel,intermsofitsmethodology,structure,andapplication.SATIMisafullsectormodelofSouthAfricathatincludesadetailedrepresentationofkeycomponentsoftheenergysystem,suchascoalsupplytopowerplantsfrommines,liquidfuelsproductionusingCoaltoLiquidandGastoLiquidplants,therepresentation of intensive energy industries in terms of their process flows and the representation ofresidentialenergyconsumptionfordifferentincomelevelsandelectricityaccess.Duetothetechnicalandprocessdetailinthemodelitissuitedtoawiderangeofapplicationsandhasalreadybeenappliedtothedevelopment of national policy responses to climate change. In addition to providing an overview ofimportantfeaturesofSATIM,thispapersupportsthemovetowardsopenandtransparentuseofSATIMandthedevelopmentofsimilarmodelsinothercountries.

1. Introduction Energy systems models provide insights into complex energy challenges such as climate mitigationresponses, ensuring energy security andmaintaining grid stability as renewable energy (RE) integrationincreases(Li,Bataille,Pye,&Sullivan,2019;Lopion,Markewitz,Robinius,&Stolten,2018).Energysystemsmodelsthatincludesupplyandtechnologydetailaswellasdemandforenergyservicesarepowerfultoolscapableofdeliveringvaluableinsightsintooptimalenergysystemconfigurationandhowthismightchangeasnewtechnologiesorfuelsareintroduced,orbehaviourandneedschange.

Developingcountriesfacemanycomplexchallengesrelatedtoenergysupplyanddemand. Onthedemandsidechallengesincludeaspectssuchaslowelectrificationrates,theuseofinferiorfuelsthathaveharmfulhealth impacts, un-serviced demand and inefficient servicing of demand. On the supply side challengesinclude the need to increase supply (often rapidly) within constrained budgets to meet economicdevelopment,uncertaintiesaroundeconomicgrowthandhowthistranslatestoindividualwelfare,carbonintensiveenergysupplysystemsandarapidlyevolvingenergytechnologylandscape.Inadditiontothesechallengestheanalysisofbothsupplyanddemandishamperedbythelackofdata,andpolicyuncertainty.

SouthAfrica(SA),likeotherdevelopingcountries,facesmanychallengesontheroadtoachievingitsenergysystemaspirations.SAisahighGHGemitterandheavilycoalreliant.REisbeingintroducedintotheelectricitysupplysystem,butonlycomprised3.7GWoutofatotalinstalledcapacityofover50GWin2018.Atransitionawayfromcoalislikely,however,concernsaroundcoalsectorjoblossesandtheirassociatedwelfareimpactsmaydelayimplementation.CTLandGTLplantssupply16percentofliquidfuels.

Inadditiontothesupplysidechallenges,SouthAfricahashighinequalityandpoverty.Householdenergypoverty is around50% (StatsSA, 2019). Energy services andmodern fuels areoftenunaffordable to lowincomehouseholds,whichrelyonmultiplefuels,includingcollectedfuelwoodandcoaltosupplytheirenergyneeds(DOE,2013).Energysectorchallengesincludetheneedtoincreaseelectrificationfromitscurrentlevelof around 85%, a need to reduce energy poverty and the use of solid fuels by households, realisingopportunities to improve energy efficiency in all sectors, realising thepotential fordemand shifting, andtransitioningawayfromcoal.

Systemsmodelsareessentialtoolsforunderstandingthechallengesandopportunitiessurroundingenergyplanning.Energysystemsmodelscontributetowardsunderstandingtheenergysystemineconomic,socialandenvironmentalterms,andthecomplexsynergiesandtrade-offsbetweeneachofthesewhenprovidingforfutureenergyneeds.Energysystemsmodelscantakeintoconsiderationvariousconstraintsortargetswhichincludeenergyaccesstargets,energytransitiongoals,distributedgenerationasanalternativewhereuseislowanddemandissparse,nexuschallengesforcompetingresourcesandnexustrade-offs.

Manyreviewsofenergysystemsmodelsareavailable,demonstratingtheirrelevanceandapplicationoverdifferent spatial and temporal scales in developed and developing countries (see Hiremath, Shikha, &

Ravindranath,2007;Jebaraj&Iniyan,2006;Laha&Chakraborty,2017;Lopionetal.,2018;Neshat,Amin-Naseri,&Danesh,2014).Theyhighlighttherangeofmethodologiescurrentlyemployedtoanalyseenergysystemschallenges.Energyoptimisationmodels,developedinresponsetothe70’soilcrisisare“howto”toolsandarecriticalforevaluatingdecisionsthathavelongtermimplicationsfortheenergysystemunderanalysis(DeCarolis,Hunter,&Sreepathi,2012;Hiremathetal.,2007).Optimisationmodels,withtechnologyrichrepresentationsofdemandandsupply,havebeenappliedtoawiderangeofpolicyquestionsinmanycountries.

In SouthAfrica severalmodellingapproacheshavebeenused to answerkeyquestions related to energyprovisioning.Forexample,PLEXOShasbeenusedinthedevelopmentof integratedResourcePlans(DOE,2018)andOSeMOSYShasbeenusedtodeveloptheIntegratedEnergyPlan.TheMARKALmodelhasbeenapplied to theanalysisof climate response, forexample in thedevelopmentof theLongTermMitigationScenarios(LTMS).LEAP,asimulationmodel,hasbeenappliedtothedevelopmentoftheNationalEnergyEfficiencyStrategy. Inadditioncomputablegeneral equilibriummodels (CGE)havebeenused toaddressenergyrelatedquestions,forexampleE-SAGEandUPGEMhavebeenappliedtoexploretheimpactsofcarbontaxes (Alton, 2014; Blignaut, 2016), and TERM-SA has been used to explore the economic impacts oftransitioningtoalowcarbonenergysupplymix(Bohlmann,2019).

Amongst themodels developed to analyse the South African energy system and its interactionwith theeconomyandenvironmentistheSouthAfricanTimesModel–SATIM.SATIMisafullsector,nationalenergysystem,leastcostoptimisationmodelthatcanbeappliedtotheplanningproblemofmeetingprojectedfutureenergydemandgiven assumptions such as the retirement schedule of existing infrastructure, future fuelcosts,futuretechnologycosts,learningrates,andefficiencyimprovements,aswellasanyconstraintssuchastheavailabilityofresources.TheentireenergyserviceneedsoftheeconomyarerepresentedinSATIMalongwith technologies that supply energy and energy services. In this way demand within the model isendogenouslyshapedbyefficiency,fuelandtechnologycosts,theformationoftheeconomyandeconomicinteractions,populationgrowthandincomegrowth,amongstothers.

SATIMisuniqueinitsrepresentationoftheSouthAfricanenergysystem.Itincludesadetailedrepresentationofalleconomicsectors.Eachsectorisadaptedtobestrepresentenergydemandorsupplywithinthesector.For example, the residential sector captures the inequality in energy access, use ofmultiple fuelswithinhouseholdstosupplythesameenergyservice,fuelsubsidiesandtechnologyefficiency.Thesupplyofliquidfuels includes a representation of the GTL and CTL plants aswell as the large and complex coal supplynetworksintopowerplants.

SATIMhasbeenappliedtodevelopinganalternativeintegratedresourceplanforSouthAfrica(Mccalletal.,2019),explorationofcoaltransitions(Burton,Caetano,&Mccall,2018),analysisofuncertaintyincoalpricesandclimatecommitments(Merven,Durbach,&McCall,2016),andtheimplicationsofwaterconstraintsforenergyplanninginSouthAfrica(Ahjumetal.2018).SATIMhasalsobeenlinkedtoaCGEmodeltoexplore,forexample,themacroandsocio-economicbenefitsofincreasingREinthepowersector(Mervenetal,2018)andmitigationresponsesthatalignwithdevelopmentobjectives(Altierietal.,2016).

Thereisagrowingemphasisontransparency,accessibilityandreplicabilityinenergymodels.InordertoimprovethetransparencytoSATIM,whichhasalreadybeenusedforpolicyandpolicycritique,thispapersetsouttoprovideadescriptionofSATIM,intermsofitsdesign(structure),application,andthestrengthsandlimitationsofthemodel. Indoingso, italsoprovidesanexampleoftheuseoftheTIMESplatformtoservicetheenergyplanningneedsofadevelopingcountry,andhopefullycontributestowardsthefurtherdevelopmentanduseofenergysystemsmodelsindevelopingcountries.

Thepaperproceedsasfollows,sectiontwoprovidesanoverviewofthesalientfeaturesoftheSouthAfricanenergysystem.ThisisfollowedbyanoverviewofTIMESandthestructureofSATIMinsectionthree.Sectionfourprovidesanoverviewofthesectorsintermsofthemodelstructureandderivationofenergyservicedemands. Section five presents a discussion of model applications and the strengths andlimitations/weaknessesofSATIM.

2. Overview of South Africa energy system Figure1providesaSANKEYdiagramoftheenergysysteminSouthAfrica.SouthAfricaconsumesaround6.5TJofprimaryenergyayear(DMRE,2017).Mostoftheenergycomesfromcoal,supplieddomestically.Coal,whichaccountsforover85%ofdomesticprimaryenergyproductionisusedprimarily inelectricitygeneration(70%)andinliquidfuelsproduction(21%)(NationalPlanningCommission(NPC),2018).Importsareprimarilyoilandpetroleumproducts,althoughthebulkofpetroleumproductsaremanufacturedlocallyfromimportedcrudeoil.Exportsofenergyarepredominantlycoal,around75Mt/year.

In2016,coalaccountedfor74%oftotalelectricitygenerationcapacitywiththebalancecomprisingofhydroandpumpedstorage,peakingplantssuchasOpenCycleGasTurbines(OCGTs),nuclearandrenewables.Totalcapacityin2016was49.8GW(Wrightetal.2017).Around92%ofelectricitygenerationisfromcoal(DOE,2016).Theagingcoalfleetmeansthatexpansionofgenerationcapacityisneededtoreplacebothretiringcapacityaswellastomeetanticipateddemandgrowth.Thenew-buildmustalsobelesscarbonintensiveto

meet South Africa's international climate change commitments as the electricity sector accounts for asignificant proportion of national emissions. There is significant potential for a transition towardsrenewables.SouthAfricapossessessomeofthebestsolarandwindresourcesintheworld,withvastareasofthecountrybeingsuitableforgeneratingelectricityusingrenewableenergy(Hagemann,2008;Fluri,2009;WASA,2015;CSIR2016).

Onthedemandside,transportisthehighestenergyconsumer.Liquidfuelsusedfortransportarearound40%ofTotalFuelConsumption(TFC).Petrolanddieselconsumptionforroadvehiclesaretheprimaryfuelsused in theTransport sectoraccounting forapproximately80%of thesector’s totalenergyconsumption(ERC,2017).Industryhasthesecondhighestenergydemandataround44%ofTFC,usingaround8%ofcoaland62%ofelectricity.Theresidentialsector,andcommercialsectorusearound15%ofTFC.Commercialsector energy consumption ismainly electricity,which comprises almost90%of total energyuse. In thecommercial sector, coal is the second largest sourceof energy, followedbyLPGandparaffin.Very smallamountsofdiesel,naturalgasandHFOareused(<1PJ).Householdfueluseisdiverseandincludestheuseofbothcoalandwoodfuel,primarilyinlowincomeruralorperi-urbanhouseholds.

Figure 1 The South African Energy System in 2015

3. The South African TIMES model 3.1. TIMES

TheIntegratedMARKAL-EFOMsystem(TIMES)wasdevelopedbytheEnergyTechnologySystemsAnalysisprogramme(ETSAP)atechnologycollaborationprogrammeoftheInternationalEnergyAgency(IEA).TIMESis a partial equilibrium, techno-economic, optimisation model generator. The objective in TIMES is tomaximise welfare by minimising total discounted system cost. In TIMES, the user has the flexibility tostructure the energy system components,which include technologies, commodities, resources, trade andemissions,tosuittheenergysystemunderanalysis.TIMESallowstheusertochoosearepresentativenumberoftimeperiodsintheyeartocharacteriseandaccountforthetemporalnatureofdemand.TIMESalsoallowsthespatialrepresentationofdemand,supplyandtradebetweenregions.Constraintscanbeappliedtoforcethesystemtobehaveincertainways,allowingadeviationfromtrueleastcostsolutions.

1.1. Overview of sectoral and temporal representation of demand and supply in SATIM

SATIMisasingleregion,multi-sector,multitimeperiod,bottomup(enduse)TIMESmodel.Itcapturesthefulleconomyandit’senergyrelatedemissions(NOX,SOX,CO2,CH4,N2O,CH4,CO,NMVOC’s,PM10),allowingthemodellingofcarbontaxesandcarbonbudgetsinadditiontoenergysupply,demand,resourceutilisation,importsandexports.

WithinSATIMtherearethreesupplysectors(electricityandliquidfuels)andfivedemandsectors(industry,agriculture,residential,commercial,andtransport).Eachsectorisfurtherdisaggregatedintosubsectorsasappropriate.IncludingadetailedrepresentationofboththesupplyanddemandsectorsinSATIMexplicitlycapturestheimpactofstructuralchangesintheeconomy(i.e.differentsectorsgrowingatdifferentrates),processchanges,fuelandmodeswitching,andtechnologyimprovementsandefficiencygains.

EachdemandsectorwithinSATIMisgovernedbyanumberofparametersandgeneralassumptionsrelatingto(a)thestructureofthesectoranditsenergyserviceneeds;(b)thebaseyeardemandforenergybyfueltype;(c)technicalandcostparametersofthetechnologiesavailabletosatisfythedemandforenergyservicesin thebase year andover themodel horizon and; (d) thedemand for energy services over theplanninghorizon.

Baseyeardemandandprojecteddemandisexogenouslyspecifiedintermsofthedemandforusefulenergyin each sector. This allows fuel substitution, as well as technology improvements, which influence theefficiencyatwhichenergyservicesaremet,toendogenouslydeterminethefinalenergydemandinthesector.

Thelevelofdetailinwhicheachsectorismodelledintermsoftechnologyandsectordisaggregationreflectsthe relative contributionof the sector to total consumption aswell as the level of funding thathasbeenavailablefordevelopingthatsectorinthemodel.ThustheTransportsector,whichisalargeenergyuserandhasbeenafocusofmodeldevelopmentforseveralyears,isquitedetailedcomparedtotheAgriculturalsector,whichaccountsforarelativelysmallpercentageofTFCandemissionsandisquitesimplisticallyrepresentedinSATIM.ThedisaggregationofthesupplyanddemandsectorsisdiscussedfurtherinSection4.

Resourcesincludelocalextraction,importsandexportsandrenewableenergy(RE).CoalminessupplyingESKOMpowerstationsarerepresentedindividually.Thisallowscoaltransportandcoalcoststoeachpowerstationtobeexplicitly included ineachscenario.Resourceextractionalso includesoilandgas fromlocalfields,aswellasfuturepotentialforcoal,oil,andgasextraction.

Figure 2 depicts the SATIMmodel components while Table 1 summarises the economic demand sectorrepresentationintermsofthesub-sectorandendusedisaggregationandsectordrivers.

Themodelplanninghorizonspansfrom2012(thebaseyear)to2050.ThebaseyearreplicatesSouthAfrica’senergysystemin2012and2017.Adetaileddescriptionofthe2012yearenergybalancecalibrationcanbefoundin(Hartley,2018),the2017documentationstillinprogress.

SATIMuses timeslices tomodel typical enduse loadprofilesandRE resourceavailability forwinterandsummer.Theloadprofilesincludeamorningandeveningpeakandnighttimeperiodwithloweraveragedemand.Thewintereveningpeaktimesliceisusedtodeterminethesystempeakdemanduptowhichthesystemmustbuildfirmdispatchablecapacitywitha15%reservemargin.

Due to the increased computational time needed with increased timeslice resolution, SATIM has beendevelopedtoeasilyaccommodatedifferenttimesliceresolutionsinordertoaccommodatearangeofresearchquestions.TemporalchangesindailyandseasonaldemandinthelowestresolutionversionofSATIMarerepresented by two seasons and one day type. Winter days consist of 5 timeslices and summer threetimeslices.Thehighestresolutionversioncurrentlyallowsfor72timeslices,representing2winterdaysand2summerdays.ThetemporalresolutionisappliedtoendusedemandandRE,apartfromtransportwheredemandisdefinedannually.

Figure 2: Schematic Summary of the South African Times Model (SATIM)

Table 1: Summary of Economic Sector Representation in SATIM and their Main Drivers

Sector SubsectorDisaggregation EnduseDisaggregation UEdemand

Driver

Agriculture NoneIrrigation,heating,processing,

traction,other. SectoralGDP

Residential

High,mediumandlow-incomeelectrified

households

Cooking,waterheating,spaceheating,refrigeration,lighting,

otherelectric

Population,Household-income,

electrificationrateMediumandlow-

incomenon-electrifiedCooking,waterheating,space

heating,lighting

Commercial None

Cooling,spaceheating,cooking,lighting,refrigeration,waterheating,publiclights,public

water

SectoralGDP,buildingstock

Industrial

IronandSteel,Ferroalloys,

Aluminium,Non-metallicmineralsand

pulpandpaper

TonnesproducedbyIndustrialprocesses

SectoralGDPmining,chemicals,foodbeveragesandtobacco,preciousandnon-feroorusmetalsandgereralmanufacturing

Boilerandprocessheating,cooling,HVAC,lighting,fans,pumping,compressedair,electrochemicalandother

electricalservices

Transport

Air,FreightandPipeline

Freighttonnekmbyrailandroad(1lightvehicleclass,9heavyvehicleclasses)

TransportGDP,

Populationand

householdcategoryandincome

PrivatepassengerPassengerkilometretravelby

Cars,SUV,motorbikesHouseholdincome

PublicpassengerPassengerkilometretravelbybus,train,minibus,BRT

Householdincome

1.2. Drivers of energy demand within SATIM Primarydrivers of demand arepopulation andGDP. Population growthdirectly impacts thedemand forenergyintheresidentialandtransportsectorsandindirectlyinallotherdemandsectors.Populationgrowthfollowsthecentral(median)casefromtheWorldPopulationProspects2019(DESA,2019),increasingto75.5millionin2050.

Whilstpopulationforecastsaregenerallykeptconstantinallscenarios,GDPgrowthcanchangeinresponsetopricesorpolicytargetsetc.GDPgrowthineacheconomicsector,alongwithanticipatedhouseholdincomelevels,isdrawnfromaCGEmodel.WhileCGEmodelsarenotforecastingtools,theydogenerateeconomicallyconsistentpathsof economic growth, including sector growth, employment andhouseholdwelfare, for agivensetofassumptions.

Useful Energy (UE) demand (the demand for energy services) is exogenously specified. UE demandprojections to 2050 are calculated based on the growth of primary drivers and are therefore based onassumptionsaroundhowanticipatedpopulationandGDPgrowthwilltranslateintoagrowthincommercialfloorarea,householdincomelevels,industrialsectorproduction,demandfortransportpassengerkmandfreighttonnekmindifferenttransportmodes,andagriculturalproduction.

2. SATIM supply and demand sector descriptions 2.1. Representation of electricity and liquid fuels supply in SATIM

Thesupplysectorincludestheextractionofprimaryresources,productionofelectricityandliquidfuels,aswellasimportsofelectricity,oilandotherliquidfuelsproducts,andexportofcoal,electricityandliquidfuelsproducts.Exportofelectricityisexogenouslyandfixedatthebaseyearvalue.Coalexportsareflexiblebutsubjecttoaconstraintongrowth(decayorincrease)rateof5%peryear.Overproductionofliquidfuelsisexported.ImportofelectricityisconstrainedtoregionalprojectslistedintheSouthAfricanDepartmentofEnergyIntegratedResourcePlans,itislargelyhydrofromInga.

Electricity

WithinSATIM,thepowersectorissplitintoGeneration,TransmissionandDistribution,showngraphicallyinFigure3.TheGenerationcomponentismodelledinthemostdetail.OperatingpowerplantsarerepresentedindividuallyandthepowersectorinSATIMthereforeincludestheexpecteddecommissioningscheduleofcoalfiredpowerplants,allplannednewbuilds,plannedretrofitsaswellasplanttechnologycharacteristics(efficiency,capacityfactorsetc).Theexpecteddecommissioningscheduleoftheexistingcoalfleet,airqualityplantretrofitschedulesforfabricfilterplants(FFP)andlowNOxburners(LNB)areassumedtofollowTable2,althoughitispossibletorunscenariowithendogenousretirementofexistingandnewcoalpowerplants.Fluegasdesulfurization(FGD)retrofitsarenotaccountedfor.MedupiandKusilearedecommissionedafter2050whichfallsoutsideoftheplanningwindow.

Newbuildoptionsincludenuclear,wind,solar,coalandgas,hydro,pumpedstorageandbatterytechnologies.InSATIMsolarPVandwindareconservativelyassumedtobeunabletocontributetopeakdemand(0%‘capacitycredit’).Dispatchablegenerationandstorageisbuilttomeeta15%reservemarginfirmcapacityrequirement.Unscheduledoutagesarealsoaccommodatedforthroughthisreservemargin.

Althoughtheremaybepracticalbuildratelimitationsforexamplebecauseoftransmissionexpansion,portlimitations,localskillscapacity,manufacturingetc.,noupperlimitsonannualnewbuildforwindandsolarareincludedintheSATIMreferencecase,allowingthesystemtochoosetheleast-costoptimalenergymixwithoutartificialconstraints.

SinceSouthAfricaismodelledasasinglenodeinSATIM,transmissionanddistributionlossesareappliedasaggregatelossesinthemodel.Underthisconfiguration,shiftinggenerationfromcurrentlocationstomoredispersed locationsdoesnot impact results for instanceasa resultofdevelopinga significantamountofrenewableenergycapacity.

Transmissioncapacityismodelledasasingletechnologylinkingcentralised/highvoltageelectricity(ELCC)to medium voltage (MV) electricity lines (ELC). Transmission of electricity between ELCC and the MVdistributionnetworkincursalossof4%.MVelectricityisdistributedtoeachsectorinawaythatcapturesthe different levels of losses that are incurred distributing electricity to the sector. For example, thedistribution of electricity to residential households is assumed to incur a loss of 19.7% whereas thedistributionofelectricitytotheindustrialsectors(excludingmining)incursasmalleraveragelossof6.%.Notethatforpurposesofsimplicity,thesupplytechnologiesandsectordistributionlegsareaggregatedinFigure3.

Figure 3: Simplified Schematic of the South African Power Sector in SATIM

Technologiesthatliewithinthetransmissionanddistributionnetworkarepumpedstorage,cogenerationand distributed renewables. Losses in the transmission of electricity to and from these technologies isaccountedforasfollows:

● Pumpedstoragestationsliebetweengenerationanddistribution.Theyuseelectricitydistributedthrough the transmission grid and feed electricity back into the transmission grid, these plantsthereforeincurtransmissionlossestwice.

● Cogeneration occurs after transmission and electricity is fed into the distribution grid tomeetindustrialelectricitydemand.Theheatgeneratedisusedtomeetindustrialheatdemand.

● Distributedenergytechnologiesbuiltwithinasectorsuchasembeddedsolarorstoragearenotsubjectedtodistributionlossesandarecurrentlynotabletofeedelectricityintothecentralnetworktoreachtheothersectors.

Table 2: Eskom plant decommissioning and air quality retrofitting schedules1 (Sources: IRP 2016, Eskom, DEA)

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

2026

2027

2028

2029

2030

2031

2032

2033

2034

2035

2036

2037

2038

2039

2040

2041

2042

2043

2044

2045

2046

2047

2048

2049

2050

Majuba Kendal Matimba Lethabo Tutuka Duvha Matla Kriel Arnot

Hendrina Camden Grootvlei Komati

Emissionabatementretrofit

50yearlifedecommissioning

Liquid Fuels

Commodities produced by the South African liquid fuels sector at present are petrol (gasoline), diesel,kerosene(jetfuel/illuminatingparaffin),aviationgasoline,liquidpetroleumgas(LPG),heavyfueloil(HFO)andmethane-richgas(producedbythesyntheticfuelsindustryonly).InSATIM,liquidfuelscanbesupplied

1

U235

Coal

ELCC ELC

NuclearPlant

CoalPlant

OCGTPlant

HydroPlant

Solar/WindPlant

Diesel

RE

GENERATION

Transmissiont

PumpedStorage

CoGen

DistributiontoResidentialt

DistributiontoIndustryt

Distributionto

Commercialt

ResidentialElectricity

HeattoIndustry

IndustrialElectricity

CommercialElectricity

TRANSMISSION DISTRIBUTION

fromthreesourcesnamely:crudeoilrefineries;syntheticfuelsmanufacturedfromcoalandgas,asshowninFigure4;andbiofuels(bioethanolandbiodiesel)asshowninFigure5.

SouthAfricahas3conventionalcrudeoilrefineriessituatedatthecoast,andonethatissituatedinland.ThecoastalrefineriesaregroupedtogetherinSATIM,astheyhavesimilarproductdistillateandoperatinginputs.Theinlandcruderefineryhasamorediesel-andkerosene-heavyproductdistillate,andthetwosyntheticrefinerieshaveagasoline-heavydistillate,thustheyarecharacterizedseparatelyinSATIM.Productslatesaremodelledbyapplyinganassumedupperboundoutputcommodityshareforeachcommodity.Inexistingtheproduct remains the same over the planning period, new refineries are modelled in a way that canaccommodateavaryingproductslate.

Refineriescanhavevariouscommodityinputs,whichcanincludecrudeoil,coal,gas,methanerichrefinerygasandsteam.InSATIMSteamismodelledasanancillaryinputservicetoCTLplants.Steamissuppliedbyboilertechnologiesusingenergyasaninput.Themodellingofsteamasanancillaryinputserviceallowsthemodeltooptimisethemostcosteffectivefuel(e.g.coal,gas)andtechnology(e.g.existingvsnewandmoreefficientboilervsCHP)toprovidethesteamneededforprocessheat,aswellasforfeedstockinCTLplants.Thislatteruseismuchgreaterthanprocessheatrequirements.SteamisalsousedincruderefineriesandGTL plants but further data is required for the characterization and therefore this consumption is notcurrentlyreflectedinSATIM.ThesteamconsumptionofcrudeandGTLrefineriesishoweversignificantlylowerandtheabsenceofthisdetailisassumedtonothaveaverysignificantimpactontheoverallresults.

TheenhancedfuelsstandardscurrentlycontemplatedbytheDepartmentofEnergywillrequireconsiderableupgradingofexistingliquidfuelsinfrastructure.Inthereferencecaseitisassumedthatnewfuelsstandardswillberolledoutby2025duetothedelaysinimplementation,bywhichtimeexistingrefinerieswilleitherhave to make significant investments, or shut down. The estimated capital investment for refineries tomigratefromCF1tofuelqualitycompatiblewithEuro5emissionsstandardsis$3.2billionto$3.7billion(2009dollars).

Figure 4: Simplified Schematic of the South African Liquid Fuels Refinery Sector in SATIM

Biofuelsareincludedasasupplychaininthemodeltoallowforamaximumsubstitutionbyvolumeof10%ofpetrolwithbioethanol;and100%ofconventionaldieselwithbiodiesel.HighervolumesofbioethanolareaccountedforbytheinclusionofE85biofuel-readyvehicles;whichallowforasubstitutionofupto85%ofpetrolwithbioethanol.AsshowninFigure5,agriculturallandandwaterarealsoincludedascommoditiesinthebiofuelsupplychainastheserepresentlimitingconstraintstodomesticproduction;notingthatthewatersupplychainisstillunderdevelopment.

Refineriesnewandexiting

Oil

Electricity

Gas

CoalCTL

newandexiting

GTL

newandexiting

Electricity

Electricity

Oil

Gas

Steam

HFO(Ag,Com,Ind,Tra)

Gasoline(Ag,Com,Ind,Tra)

Diesel(Ag,Com,Ind,Tra,Power)

Kerosene(Ag,Com,Res,Tra)

LPG(Ag,Com,Ind,Res)

AvGas(Tra)

Figure5:SimplifiedSchematicof1stGenerationLiquidBiofuelsProductioninSATIM

Hydrogen

HydrogenasacommodityisalsoincludedinSATIMwiththreeprocessroutesmodelledasshowninFigure6. Coal gasification, natural gas or steammethane reformation (SMR), and electrolysis ofwater are theproductionroutesincluded.FortheSMRroute,inlandandcoastalproductionisdistinguishedonaccountoflanded gas prices.Water electrolysis is presently presumed via the platinum based polymer electrolytemembrane(PEM)process.AlsodepictedinthefigurearetheassociatedCO2emissionswithaproductionroute.Aswithbiofuelssupply,theinclusionofthewatersupplychainisunderdevelopment.

Figure6:SimplifiedSchematicofHydrogenProductioninSATIM

2.2. Representation of Energy Demand in SATIM Residential

Householdsarethebasisofdemandintheresidentialsector.Householdenergyuseisverydiverse,andthereare several distinguishing features that have been found to be important in determining the quantity of

energyusedbyhouseholds,theirfuelchoiceanddemandprofiles.Amongsttheimportantinfluencersareincome,accesstoelectricity,accesstowaterandhousingtype.Thereisadistinctincreaseinbothoverallfuelconsumption,aswellasapplianceownershipandtheuseofmodernfuelsasincomeincreases.

Toaccountfortheinfluenceofincomeandelectrificationonhouseholdconsumption,SATIMincludesthreecategoriesofhouseholdsdistinguishedbytheirlevelofincome.Thelowestincomegrouprepresents47%ofhouseholds,manywhichusemultiplefuelswithalowoverallenergyconsumption.Themiddleincomegrouprepresentsaround30%ofhouseholds.Middleincomehouseholdareprimarilyelectricityusers,buthavealower overall appliance ownership and consumption levels compared to high income households whorepresent23%ofhouseholds.Eachhouseholdgroupingisdistinctinitsdemandforenergyservices,andthetechnologypenetrationusedtosupplyenergyservices.

Driversofdemand in theResidentialSectorarepopulationandhousehold incomegrowth.Projectionsofhouseholdgrowth ineachhousehold category in themodel relyonassumptionsaround theevolutionofpopulation growth, household size, household income levels, and the success of the electrificationprogramme. An exogenous projection of the electrification rate is required to derive a household splitbetween the electrified and unelectrified households in the two lower income groups. This is based onhistoricaltrendsandfuturegovernment/Eskomtargets,andisshowninTable3until2050.

Table 3: Assumed Future South African Electrification Rates by Income Group

IncomeGroup 2010 2020 2030 2040 2050

LowIncome 71% 80% 85% 90% 95%

MiddleIncome 83% 90% 95% 95% 100%

HighIncome 100% 100% 100% 100% 100%

OverallElectrification 81% 90% 95% 97% 99%

Changesindemandoccurashouseholdsshiftgroups,orthroughexogenousresponsestosectormarketshifts.Useful energy demand remains the same in the lower two income groupswheremedian income is heldconstant,butinthehighestincomegroupthemedianincomecanchangeandthedemandforusefulenergyservicesislinkedtoincomeusingelasticities.Bydisaggregatinghouseholdsaccordingtotheirincomeandelectrification status, the model can incorporate and react to policy interventions which target certainhouseholds(i.e.anincreaseinresidentialelectricitytariffforhighenergyconsumers)aswellasmitigationactionsrelevant toaspecifichouseholdgroups (i.e.a solarhotwaterheatingprogrammeon low incomehouseholds).

Thesplittingofhouseholdsintoincomebandsislargelygovernedbyapplianceownership,(Beute,2010),(Dekenah,2010)(Gertler,Shelef,Wolfram,&Fuchs,2012).Nationalsurveysaretheprimarysourceofdataforthesplit(IESandCensusorcommunitysurvey).Thebaseyearcalibration,intermsoffueluseandenergyservicedemands,reliesonacombinationoflargenationalsurveysthatmatchincomegrouptofueluseforlighting,heating,andcookingandcaptureapplianceownership(SWh,TV)etc,andbottomupsurveys,whichcanbelargeorsmallscaleandcaptureuserbehavioursortiequantitiesofenergytoenergyservicesthatcanbematchedtotheincomegroups.Thedemandforenergyservicesineachgroupisthereforeanindicativeestimateandtherepresentationofthesectorinfurtherdetailislimitedbydataavailability.

ThebaseyearcalibrationofUEattributedtoeachhouseholdgroupiscalculatedasfollows:

Uijn= EIijnXHinWhereUijnistheusefulenergydemandforincomegroupiusingenergyservicejinyearn(GJ),EIijnisthe

energyIntensityforincomegroupiusingenergyservicejinyearn(GJ/perhousehold/year)andHinisthenumberofhouseholdsiinyearn(households)

Transport

ThetransportsectorinSATIMincludesenergyusedforpassengerandfreighttransportbyroad,rail.Italsoincludesenergyusedinpipelinetransfersandaviation.Energydemandforpassengerandfreighttransportisdrivenprimarilybytwofactors,a)vehicle-kilometerstravelledandb)theefficiencyoftravel.Thevehicle-kilometerstravelledaredrivenbytheneedsofsocietyandtheeconomytomovepeopleandgoodsaround.Conversionefficiencydifferswithvehicletype,fueltypeandtheageofthevehicleparcandtosomedegreethepatternsofutilisationofdifferentvehicletypes.TheenergyservicedemandinSATIMisdefinedintermsofpassengerkilometresandtonnekilometres.

Acalibratedvehicleparcmodel2 is used to estimatedemand forpassengerkilometres (pkm).Aspectsoftransportincludedintheparcmodelarethesizeoftheexistingvehiclefleet,annualvehiclesales,annualvehiclescrapping,distancetravelledpervehicle,fuelsalesandvehiclefuelefficiency.Outputsofthevehicleparcmodelare totalkilometres travelled, theaverageageofvehicles in thevehicle fleetandtheaverageefficiencyof thevehicle fleet.Thesecomponentsallowefficiencyor intensityof transport tochangewithvehiclestockchangesandanincreaseordecreaseinvehicleownershipinresponsetopopulationandincomechanges.Certainfactorsaffectingthedistancetravelledandfuelefficiency,forinstancetrafficcongestion,aredifficulttoquantifyastheyarenotwellunderstood.Toaccommodatetheunknownimpactoftangibleandintangibleinfluencesonefficiency,thevehicleparcmodeliscalibratedbyadjustingvariablessuchasvehicleoccupancy and ownership assumptions until the output (annual distance travelled by vehicles) incombinationwithvehicle fuelefficiencymatchesknown fuel salesdata.Theannualdistance travelledbyvehicles,istranslatedintoademandforpkmbyassumingaverageoccupancyratesforthedifferentvehicletypesinSATIM.Arecentupdateofthevehicleparcmodelcalibrationcanbefoundin(Stoneetal.,2018).

Thedemandforpkmissplitbetweenincomegroupsandacategory“other”toaccountforcommercially-andgovernment-owned cars. The passenger demand projection model (in excel) uses assumptions aroundprivatevehicleownershipbyincomegroup,vehiclemileage,vehicleoccupancy,publicmodeshares,averagemode speeds, and a travel time budget, to derive vehicle-km demand by passenger vehicle class forhouseholds.Thisiscombinedwithatransport-GDPlinkedprojectionofthenon-householdownedcarstogiveatotalpassengervehicle-kmdemandprojectionforroadvehicles.Thepassenger-kmprojectionsforrailare derived from assumptions around futuremode shares i.e. the degree towhich passengers are usingprivateorpublictransport.ThetransportmodelinSATIMisthusimplementedasdepictedinFigure6.

Figure 7: An overview of the SATIM transport sector model (Merven et al 2012).

Demand for freight vehicle-kilometers is linked to sectoral GDP growth as shown in Figure 7, For bothpassengerandfreightroadvehicles,constantoccupancyandloadfactors,respectively,areassumedoverthetimehorizonwhilevehicleefficienciesarepresumedtoimproveatarateof1%perannum.

AsdescribedbyStoneetal.(2018)scrappagefactorsareappliedtotheexistingvehiclepopulationwhichremovesexistingvehiclesfromtheactivefleet;andintandemvehicleactivitydeclineswithage(i.e.decayratesareappliedtoactivevehicles).Onthisbasisasthefleetpopulation(capacity)andactivitydecreases,newvehicles(capacity)arepurchased,informedbythetotaldiscountedcost(vehicleandfuelcost),tomeetthedemandforvehicle-kilometres.At thisdecisionpoint, themodelselects fromtheportfolioofvehiclesavailable(e.g.diesel,petrol,gas,oil-hybrid,battery-electricetc.)Theportfoliooftechnologiesthatthemodelisabletoselectfrom,withinavehicleclass,isshowninFigure4.NotechnologiesaremodelledforaviationbutratherfueldemandiscorrelatedwithGDP.

2 For more information see (Ahjum et al. 2018)

(1):BRT:BusRapidTransport;(2):Metro:Metropolitani.e.intra-city;(3):UsedforCoastal&InlandNavigation;*InternalCombustionEngine;#:BatteryElectricforRoadVehicles;HCV1:Mediumcommercialvehicleof3000–7500kgGVM;HCV2:Heavycommercialvehicleof7501–12000kgGVM;HCV6:Heavycommercialvehicleof24001–32000kgGVM.SUV:SportUtilityVehicle(usually4X4and>1toninmass)

Table 4: Transport Technologies in SATIM

Commerce

The commercial sector includes energy used by the wholesale, retail and motor trade services andaccommodation (trade); finance, real estate and business services (finance); government; and personalservices sub-sectors. Commercial sector demand for energy services is estimated based on an assumedenergyintensityofenergyservicesneededasafunctionoffloorarea(PJ/m2);thegrowthinfloorareaovertime;andimprovementsinenergyefficiencyasaresultofbuildingcoderegulations.FinalenergydemandisendogenousinSATIMandiscalculatedbasedontheefficiencyoftechnologiessupplyingenergyserviceneedsandtheirpenetrationinthesector.

The 2012 USA Commercial Buildings Energy Consumption Survey (2012 CBECS) (EIA, 2016) providesinformation on electricity usage per square meter by end use for commercial building types. The non-residentialbuildingcategory‘othernon-residentialspace’isnotrepresentedinthe2012CBECSsurvey.AweightedsumoftheaverageelectricityintensityforallCBECSbuildingtypeslocatedintheclimatezonemostsimilar to South Africa (i.e. climate zone <4000HDD& <2000CDD)3 was used to calculate the electricityintensityofenergyservicesperm2.TheweightsareestimatedusingStatsSAcommercialfloorareachangeswhichareappliedto1990stockestimatesfromdeVilliers(2000).Thisinformationisthenusedtocalculatetheshareofeachenergyservicedemandintotalelectricityusage(perm2).Thisisappliedtoelectricityusageperm2data forSouthAfrica todetermineSouthAfricaspecificelectricityusageperm2 foreachenduse.Energyefficiencyimprovesovertimeasbuildingcodesareupdated.

Totalcommercial floorareawasestimatedtobearound88millionm2in2012. ThisestimateisbasedonStatisticsSouthAfrica’smonthlyandannualstatisticsonthechangeinfloorareabybuildingtype(STATSSA,n.d.) anddeVilliers (2000) estimateof commercial floor area excludingwarehouses in1990,whichwasreported to have been 57.9millionm2. Growth in floor area over themodel horizon is informed by therelationshipbetweenhistoricgrowthinfloorareaandgrowthinGDP.AnanalysisofthechangeinfloorareaasreportedbySTATSSAandthegrowthinrealGDPshowsthatthereisaclosecorrelationbetweenthetwosetsofinformation,althoughthechangeinfloorarealagschangesinrealeconomicgrowthbyabout2years.Thechangeinfloorareaisequivalenttoaround70%ofthegrowthinrealGDP.Thisrelationshipisassumedtoholdinthefuture.

The existing floor area is based on total commercial stock in 2012with the share of existing floor areadecreasingby1%from2013.Newbuildingfloorareathereforeincreasesfrom2013attherateofcommercialfloorareagrowthplusthelossinexistingfloorarea.By2050,54%ofthetotalfloorareaconsistsofnewbuildingstock.Itisassumedthatenergyservicesinnewbuildingscanbemetmoreefficientlywithnewertechnologies,whilstresidualcapacityrestrictstheuseofnewertechnologiesinolderbuildings.

Industry

Theindustrialsectorconsistsofseveralenergyintensivesectors,suchastheironandsteelsectorsaswellassmallerproducerssuchastheautomotiveindustryandothermanufacturing.InSATIM,twomethodologiesareappliedtomodelindustrialenergyconsumption.Thefirstmethod(approachone)reliesonestimatesofenergyserviceneeds(lighting,coolingetc),thesecondmethod(approachtwo)utilisesanestimateoftheenergy intensity of industrial technology processes. Approach two is typically applied to sectors whereproductsaremoreuniformandtheenergyintensityofproductionishighsuchastheironandsteelsector.Bothapproachesaredescribedinmoredetailbelow.

Approachone isappliedtothemining,chemicals, foodbeveragesandtobacco,preciousandnon-ferrousmetals(excludingaluminiumproduction)andgeneralmanufacturing. In thisapproachanestimateof thetotalusefulenergyservicerequirement(e.g.Processheat,compressedair,etc),perunitofoutputorvalueadded,andtheefficiencyatwhichenergyservicesaremetareexogenouslyspecifiedandallowthemodeltoendogenouslydeterminefinalenergyconsumptionforeachenergyserviceinthesub-sector.ThedriverofindustrialenergyconsumptionisthereforethedemandforusefulenergyservicesU.ThisisdemonstratedinFigure8,wherethelevelofusefulenergyservicesneeded,inthiscaseprocessheating,andtheefficiencyoftheboiler,determinestheamountoffinalenergy(coal)consumed.

3 HDD and CDD refer to heating and cooling degree days and indicate the energy demand needed to heat and

cool a building, respectively, to get to the base temperature 65°F (18°C).

Figure 8: Technology representation in approach one subsectors in industry – example of a boiler

technology

Theestimateofbaseyeardemandforusefulenergyservices!!,#,$,%iscalculatedforeachindustrysubsectori,technologyT,fuelorcommodityf,andenduseserviceu(heating,lightingetc.)usingtheequationbelow:

!!,#,$,% = #!,#,$,% × %!,#,$,%Where#!,#,$isthetechnologyefficiency,andthefinalenergyconsumedtosupplytheenergyserviceiscalculatedaccordingtotheshareoftechnologiessupplyingtheenergyservice

%!,#,$,% ='!,#,$ ×!'$,!,% × ($,!where

($,!istotalfinalconsumptionoffuelfbysubsectori(Joules),and!'$,!,%isthe%shareoffinalenergysupplyingenduseserviceu'!,#,$isthe%shareoftechnologyTassumedtosupplytheenduseservice

Assumptions around technology efficiency and energy service levels aredrawn from literature and fromenergyauditsandsurveys.

ApproachtwoisusedfortheIronandSteel,Ferroalloys,Non-metallicmineralsandpulpandpapersectors.Thedemand for final energy in these sectors is calculated endogenouslybasedon the energy intensitiesspecifictotechnologyprocessesandtheirlevelofproduction.ForexampleinFigure9,thedemandforcoalandcokebyblastfurnacesintheproductionofironiscalculatedbasedonthetechnologyspecificenergyintensity(GJ/t)ofironproductioninSouthAfricanblastfurnaces.Inordertoapplyapproachtwo,theshareofproductionbytechnologytypeandtheenergyintensityofproductioninSouthAfricamustbeknownorestimated.

Figure 9: Example of technology representation in approach two subsectors in industry

TheproductionoutputObyproductPisgivenas:

)( =*#

+#,(

where+#,(istheactivity4ofatechnologyTproducingproductP.FinalenergyconsumptionFoffuelfforanindustryisthen:

%$ =*#

(+#,( × -#,$!)

where-#,$! istheenergyintensityfortechnologyT,and/'representsthefuels.Figures10showhowexistingtechnologycapacity(blue)andnewtechnologyadditions(green)areexplicitlyrepresentedinthemodel.Theironandsteelsector,showninFigure10,currentlyproducessteelusingarangeoftechnologiesandfuels.Existingtechnologiesandplantscanberetrofittoimproveefficiencies.Newtechnologiesarealsoavailable,allowingtheefficiencyofproductioninnewplantstobehigher.Thenon-

4 The activity of a technology is bound by limits by the user, but is determined in future years by

the optimisation program

Coalboiler

(%Efficiency)ProcessHeat-PJCoal-PJ

BlastFurnace

Intensity:CoalGJ/t

CokeGJ/t

Iron-tonnes

Coal-PJ

Coke-PJ

Figure 10: Structure of the a) Iron and Steel b)Non-metallic minerals c) Pulp and Paper industry and d) Ferroalloys representation in SATIM

a) b)

c) d)

metallicmineralssector(Figure10)includestheproductionofcement,glass,lime,andbricks.Thepulpandpaper(P&P)(Figure10)sectorproducespaper,paperproducts,anddissolvingpulp.Steamisacritical componentof theenergyandprocesssystems for theP&Psector,andbiomassresidueandproductionwasteareoftenusedasfeedstocksforboilersinthepulpingprocess.TheFerroalloyssector(Figure 11) includes the production of Ferro-Chrome, Ferro-Silicon, Ferro-Manganese, and otherManganesealloysinSouthAfrica.Theproductionisferro-chrome,whichisusedinthemanufactureofstainlesssteel,dominatedferro-alloyproductioninSouthAfrica.Themajorityofenergyconsumedintheproductionofferro-alloysisusedinelectricarcfurnaceswhichheatamixtureofore,fluxesandreducing agents (coke and/or coal) for smelting into metal alloy product. Due to the difficulty ofaccessinginformationanddataforthissector,theferroalloyindustryinSATIM,apartfromaluminium,isrepresentedbyasingletechnology–theelectricarcfurnace(showninFigure10).TheAluminiumsectorisalsocarvedoutofthenon-ferrousmetalssectorandischaracterisedasasingletechnologyconsumingelectricitytoproduceAluminium,withassociatedprocessemissions.ThishasbeendoneastheAluminiumsectorisarelativelylargeconsumerofelectricityanditsactivityisnotstrictlylinkedtotherestofthenon-ferrousmetalsindustry,norwiththerestoftheSouthAfricaneconomy.Untilfurtherupdates,itisassumedthattheproductionofaluminiuminSouthAfricaremainsconstant-asmostoftheproductisexported,thebalancegoingtolocaldemand.Fluctuationsinexportandlocaldemandareassumedtobalanceeachotherout.

3. Accounting for emissions Fugitiveemissionsfromextractionandrefineries,aswellasemissionsfromcombustion,areaccountedfor in SATIM. Combined they account for 80% of overall GHG emissions in South Africa. Fugitiveemissionsareaccountedforatthesourcei.e.emissionsfromminingareaccountedforunderextractionandrefineryemissionsareanoutputofrefineries.Botharedirectlyrelatedtooutput.Emissionsfromcombustionareallocated tosectoral fueluse.Theyare thereforeaverageemissions thatoccurasaresultofthecombustionofcoal(ofdifferentgrades),gas,liquidfuelsetcineachsector.Process(IPPU)emissionsareaccountedforattheprocesslevel,exceptinthecaseofthe2D:Non-Energyproductsfromfuelsandsolventuse,and2F:Productusesassubstitutesforozonedepletingsubstances,whicharespecifiedexogenouslyasafixedtimeseries.AFOLUandWasteemissionsarecurrentlyaccountedforoutsideofSATIM.

4. Discussion of strengths and limitations of the SATIM modelling approach

AstrengthoftheapproachtakenindevelopingSATIMliesintheinclusionoftheentireenergysystemin themodel. This allows emissions, including fugitive emissions, and fuel use to be accounted forexplicitlyforallsectors.Thisinturnallowsfortrade-offbetweensectorswhenmitigationisrequiredandthetargetingofspecificsectorsinscenariosforpolicyinterventions.Therepresentationofdemandforenergyservicesinallsectors,allowsefficiency,processeschangesandfuelswitchingonthedemandsidetocompletewithchangesonthesupplysidetoprovideanoptimalenergypathwayforSouthAfricatomeetfutureenergyserviceneedsunderpolicyorotherconstraints.Forexample,itispossibleforelectricitygenerationtoincreasetheuseofRE,orforthedemandsectorstoincreasetheiruseofRE,ortoutilisemoreefficienttechnologiestoachieveanoverarchingobjectiveofreducingemissions.

ThehighlevelsectoraldetailinSATIMmatchestheenergybalancespublishedbytheDepartmentofEnergyofSouthAfrica.Energyconsumption in thebaseyeardiffers fromtheDOEenergybalanceswhereotherpublishednational statistics indicate that adjustmentsarenecessary.This ensures theaccuracyofbaseyeardemand.However,duetothedetailedrepresentationofdemandattheenergyservicelevel,aswellasthedelaysinpublishingtheenergybalancesandnationalstatisticsneededtoupdate themodelbaseyear,baseyearupdatesare irregular.Thecurrentbaseyear is2012withadetailedupdate for2017.A featureofTIMESwhichhelps in this regard is thatmodelyearscanbeindependent of data years, allowing sector updates to be unconstrained by data availability in allsectors.

Incomechangesovertimeplayaroleindeterminingthedemandfortransport,influencingthelevelsofbothtonneandpassengerkilometres.Intheresidentialsectorincomeandelectricityaccessplayaroleinenergyservicedemands,fueltypeandefficiencyoftechnologies.Householdincomegrowthandelectrificationthereforeneedtobeaccuratelyreflectedinanenergysystemsmodelinawaythatisconsistentwitheconomicgrowth.Inaddition,incomegrowthreflectedinthetransportsectormustbe

consistentwiththatoftheresidentialsector.ThisisdoneinSATIMbylinkingthemodeldriverstotheoutputofaCGEmodeltodevelopeconomicallyconsistentscenarios.

Mileagedecayandefficiencyimprovementduetovintagingisexplicitlyincludedinthetransportsector.Similarly,improvementsintheefficiencyofbuildingstockovertimeareincludedinthecommercialsector and in the intensive industries process efficiency of plant can improve with additionalinvestment.

Coalsupplytopowerplantsandtheplantstheysupplyarerepresentedindividually.Thisallowscoalqualityandcosttobecapturedagainsteachpowerplantindividually.Italsoallowsinvestigationoftheoptimalpathfortransformationinthepowersectorawayfromcoal,basedonplantefficiency,costandotherobjectives. This is an important componentof SATIMas coal is a largeportionofTPES, andcompetitionbetweencoalfiredplantsandREmustberepresentedasaccuratelyaspossible.

AnimportantstrengthofSATIMisitsmodularity,allowingeasyexpansionofsectorsandsubsectorsand energy service representation. The time slice resolution in SATIM can also easily be adjustedaccording to researchneeds, and expansionof tradeof energywithneighbouring countries canbeaccommodated, through the utilisation of the regional trade features in TIMES. The flexibility ofTimesliceresolutionisimportantasithasimplicationsforREintegration,andmodelruntime.Regionaltradeislikelytoexpandinfutureandhavingtheflexibilitytoincludedemandandsupplyoptionsinneighbouringcountriesexplicitly,atasimilarlevelofdetailtoSATIM,willallowexplorationofoptimalenergypathwaysfortheregion.

An additional important development in recent years is the soft linking of SATIM to the SAGEComputableGeneralEquilibriummodel.Thelinkedmodelcaptureseconomicresponsetofuelpriceswhichoccurforexample,asaresultofdifferentenergyinvestmentpaths,andchangestoenergyserviceneedsinresponsetoGDPgrowth.EconomicconsistencyinscenariosanddemandgrowthisaconcernasgrowthinallsectorsisdrivenbyGDPeitherdirectly(e.g.commerce)orindirectly(e.g.residential).TheuseofaCGEmodeltodevelopeconomicallyconsistentscenariosaswellasthelinkingofSATIMSAGEhasallowedtheexpansionofresearch toareassuchas just transitionsandeconomicallyandclimateoptimaldevelopmentpathways.

AsSATIMrepresentsdemandinSouthAfricaasasinglenodethespatialdiversityofdemandisnotcaptured.SimilarlyREandfossilfuelresourcesarealsospatiallydiverse.Notcapturingthisdiversityhasimplicationsfortheoptimisationoftheenergysystem.Forexample,powerplantsthatarelocatedfarfromdemandcentresdonotincuratransmissionpenalty.

TheTIMESapproachisnormative:optimizedfromasingleagent’spointofview.Thisisfineforsupply,andindustry,whichnormallybehaveinthisway.However,withoutextensivefurtherdisaggregationoftheresidentialdemandsector,heterogeneityofpreferencesisnotcaptured.Thecheapesttechnologywilltendtodominatethesolutionevenifthereisanotherproductavailablewithslightlyhighercostbutothernon-monetaryadvantages.Themodelwouldthereforebenefitfromsoftlinkingwithmoredetailedchoicemodels.

Anotherweaknessisthefactthatmodechoiceintransportisexogenous.Toaccuratelyreflectmodalshift inthetransportsectorwithinSATIMwouldrequire furtherdisaggregationtoreflect transportdemandinthelargercitiesofSouthAfrica.Eachcityhasitsownpeculiarities,forexamplewhetherrailtransportisavailable,whichwouldhavetobecaptured,inorderformodalswitchtobeendogenized.

5. Conclusions

ThispaperpresentsanoverviewofSATIM,a fullsector, technologyrich,optimisationmodelof theSouthAfricaEnergysystem.SATIMisuniqueinitsrepresentationaspresentedinthispaper.ImportantfeaturesinSATIMarethatithasbeendevelopedinawaythatallowschangestoitsmodularstructureandtemporalresolutionallowingshiftsintechnology,sectorexpansion,orincreaseddataresolutiontobeincorporatedastheneedarises.ThesefeatureshaveallowedSATIMtobeadaptedtobeutilisedtoinformawiderangeofcriticalpolicyquestionsinSouthAfrica,includingclimatechangeresponses.

PresentingSATIMinanopenmannerinthispaperhasthreeadvantages:firstlyitallowscritiqueofthemodelling system and assumptions; secondly it allows others facing similar challenges related tocharacterisingadevelopingcountryenergysystemanddatascarcitytoreviewtheapproachtakentoovercometheseissuesinthedevelopmentandmaintenanceofanenergysystemsmodeland;thirdlyasSATIMhasalreadybeenused,andwillcontinuetobeused,todeliverinsightsintopolicyquestionsitisimperativethattheassumptionsbehindthemodelareinthepublicdomainandcanbeinterrogated.

InordertoensureSATIM’srelevancethebaseyearandmodelstructurehavebeenupdatedasoftenaspossible.InorderforSATIMtoremainrelevanttheupdatecycleneedstocontinue.SATIMisnowstoredanopenaccessversion-controlsystem.Flexiblespatialdisaggregationintoprovinces/citieswouldalsobeagreatenhancement.

6. Acknowledgments ThispaperiswrittenwiththesupportofSouthAfricanNationalEnergyDevelopmentInstitute,whohavealsocontributedtothedevelopmentofSATIM,thecalibrationofbaseyeardata,andthelinkingofSATIMtoSAGE.

7. References AhjumF.,MervenB.,CullisJ.GoldsteinG.,DeLaquilP.andStoneA.2018.DevelopmentofaNationalWater-Energy SystemModel with Emphasis on the Power Sector for South Africa. EnvironmentalProgress&SustainableEnergy,Wiley,USA.https://doi.org/10.1002/ep.12837

Ahjum,F.,Merven,B.,Stone,A.,&Caetano,T.(2018).RoadtransportvehiclesinSouthAfricatowards2050 :Factorsinfluencingtechnologychoiceandimplicationsforfuelsupply.JournalofEnergyinSouthernAfrica,29(3),33–50.

Altieri,K.E.,Trollip,H.,Caetano,T.,Hughes,A.,Merven,B.,&Winkler,H.(2016).Achievingdevelopmentand mitigation objectives through a decarbonization development pathway in South Africa.ClimatePolicy,16(sup1),S78–S91.https://doi.org/10.1080/14693062.2016.1150250

Alton, T., Arndt, C., Davies, R., Hartley, F.,Markelov, K., Thurlow, J., Ubogo, D., (2014): IntroducingCarbonTaxesinSouthAfrica.AppliedEnergy,116344–354.Web.

Blignaut,Jamesetal.(2016)“TheEconomicandEnvironmentalEffectsofaCarbonTaxinSouthAfrica:A Dynamic CGE Modelling Approach.” South African Journal of Economic and ManagementSciences19.5.

Bohlmann,H.R.etal.“RegionalEmploymentandEconomicGrowthEffectsofSouthAfrica’sTransitiontoLow-CarbonEnergySupplyMix.”EnergyPolicy,128(2019):830–837.Web.

Burton, J., Caetano, T., & Mccall, B. (2018). Coal Transitions in South Africa: Understanding theimplications of a 2°C-compatible coal phase-out for South Africa. Retrieved fromhttps://coaltransitions.files.wordpress.com/2018/09/coaltransitions_finalreport_south-africa_2018.pdf

DeCarolis,J.F.,Hunter,K.,&Sreepathi,S.(2012).Thecaseforrepeatableanalysiswithenergyeconomyoptimization models. Energy Economics, 34(6), 1845–1853.https://doi.org/10.1016/j.eneco.2012.07.004

deVilliers,M.2000.Greenhousegasbaselineandmitigationoptionsforthecommercialsector.ReportnumberEDRC/00/R5,EnergyandDevelopmentResearchCentre,UniversityofCapeTown:CapeTown,SouthAfrica.

DESA,U.,2019.WorldPopulationProspects2019.UnitedNations.DepartmentofEconomicandSocialAffairs.WorldPopulationProspects2019.

DMRE(2017)CommodityFlowandEnergyBalance2017,Pretoria,SouthAfrica,Retrievedfromhttp://www.energy.gov.za/files/energyStats_frame.html

DOE.(2013).Asurveyofenergy-relatedbehaviourandperceptionsinSouthAfrica2013.Pretoria,SouthAfrica. Retrieved from http://www.energy.gov.za/files/media/Pub/Survey of Energy relatedbehaviourandperceptioninSA-ResidentialSector-2012.pdf

DOE.(2018).IntegratedResourcePlan(draft).Pretoria,SouthAfrica.

EnergyInformationAdministration(EIA).2016.2012CommercialBuildingsEnergyConsumptionSurveyData.Available:http://www.eia.gov/emeu/cbecs[May2019].

Hiremath,R.B.,Shikha,S.,&Ravindranath,N.H.(2007).Decentralizedenergyplanning ;modelingandapplication — a review. Renewable and Sustainable Energy Reviews, 11, 729–752.https://doi.org/10.1016/j.rser.2005.07.005

Jebaraj,S.,&Iniyan,S.(2006).Areviewofenergymodels.RenewableandSustainableEnergyReviews,

10,281–311.https://doi.org/10.1016/j.rser.2004.09.004

Laha,P.,&Chakraborty,B.(2017).Energymodel–Atoolforpreventingenergydysfunction.Renewableand Sustainable Energy Reviews, 73(April 2016), 95–114.https://doi.org/10.1016/j.rser.2017.01.106

Li,F.G.N.,Bataille,C.,Pye,S.,&Sullivan,A.O.(2019).Prospectsforenergyeconomymodellingwithbigdata :Hype , eliminatingblind spots , or revolutionising the stateof the art ?AppliedEnergy,239(February),991–1002.https://doi.org/10.1016/j.apenergy.2019.02.002

Lopion,P.,Markewitz,P.,Robinius,M.,&Stolten,D.(2018).Areviewofcurrentchallengesandtrendsin energy systems modeling. Renewable and Sustainable Energy Reviews, 96(July), 156–166.https://doi.org/10.1016/j.rser.2018.07.045

Mccall,B.,Burton, J.,Marquard,A.,Hartley,F.,Ahjum,F., Ireland,G.,&Merven,B.(2019).Least-costintegratedresourceplanningandcost-optimalclimatechangemitigationpolicy :Alternativesfor

the South African electricity system. Retrieved fromhttp://www.erc.uct.ac.za/sites/default/files/image_tool/images/119/Papers-2019/Alt IRPfinal07022019_2.pdf

Merven,B.,Stone,A.,Hughes,A.,&Cohen,B. (2012).Quantifying theenergyneedsof the transportsectorforSouthAfrica:Abottom-upmodel.

Merven, B., Durbach, I., & McCall, B. (2016). Obtaining long-term forecasts of the key drivers ofgreenhousegasemissionsinSouthAfrica.

Merven, B., Ireland, G., Hartley, F., Arndt, C., Hughes, A., Ahjum, F., Mccall, B., Caetano, T., (2018).Quantifyingthemacro-andsocio-economicbenefitsofatransitiontorenewableenergyinSouth

Africa,SA-TIEDworkingpaper.

NationalPlanningCommission(NPC).(2018).NPCEconomySeries:Energy.

Stone,A.,Merven,B.,Maseela,T.,&Moonsamy,R.(2018).Providingafoundationforroadtransportenergydemandanalysis:AvehicleparcmodelforSouthAfrica.JournalofEnergyinSouthernAfrica,29(2),29-42.

Neshat, N., Amin-Naseri, M. R., & Danesh, F. (2014). Energy models : Methods and characteristics.JournalofEnergyinSouthernAfrica,25(4),101–111.