assumptions and methodologies in the south african times ... · assumptions and methodologies in...

`

Assumptions and Methodologies in the South African TIMES (SATIM) Energy Model

EnergyResearchCentreSystemsAnalysis&PlanningGroup

Version2.12013/04/08

`

ExecutiveSummary

TheEnergyResearchCentreat theUniversityofCapeTownmaintainsaTIMESenergymodel(SATIM)which isnow in its3rdgeneration.TIMES isapartialequilibrium linearoptimisationmodel capable of representing thewhole energy system, including its economic costs and itsemissions, and is thus particularly useful in modelling potential mitigation policies. Theapproach is fundamentally sectoral as would be the case with other models, even simplespreadsheetmodels likeMAED, and so the analysis of the structure of energy demand fromsectors is a fundamental building block of the modelling process, regardless of the toolsselected.ThisreportdescribesthemodellingframeworkdevelopedbytheEnergyResearchCentre(ERC)fortheSouthAfricaenergysector.Themodellingframeworkisbasedonaseriesofnation‐wideenergymodellingtasks/projectsthathavebeencarriedbyERCsincetheLongTermMitigationScenarios (LTMS)modelling process and themodel is in its third generation since then. Themodellingsystem,themethodologyanditsvariouscomponentsarebrieflydescribed,followedbyamoredetaileddescriptionofthedataandassumptionscurrentlyemployedinthemodelonasectorbysectorbasis.TheSouthAfricanTIMESmodel(SATIM)isstructuredintofivedemandsectorsandtwosupplysectors– industry,agriculture, residentialcommercialand transporton thedemandside,andelectricityandliquidfuelsonthesupplyside.Thesectorsvarysomewhatintheirshareoffinalconsumptionandconsequentlygreenhousegasemissions,withTransportaccountingfor27%ofTotalFinalConsumptionofEnergy in2006, the IndustrySectoraccounting fornearly40%andtheAgricultureSectoronlyaccountingfor2.6%Giventhesedisparities,notallsectorsinSATIMhaveenjoyedthesameinvestmentofresearch,andresearchfundinghastendedtoconcentrateonsectorshavingahighenvironmentalimpactand profile like the Transport Sector and Electricity Supply sector. The Transport Sector inparticulariswelldocumentedandincludesahighlevelofdetailduetooutsideinvestmentfromrecentprojects.Thisdocumentshouldbeausefulreferenceforreadersinvolvedinsettingupnationalenergymodelsforinfrastructureplanningoremissionsmitigationpurposes.Whilecertainaspectsareparticular to SouthAfrica,muchof the approach and solvingof problemswill haveuniversalapplication.The detailed assumptions of the model in numbers are maintained in a Microsoft Excelspreadsheetwhichformstheappendicesofthisdocument.ThespreadsheetispostedwiththisdocumentonthewebsiteoftheEnergyResearchCentre,UniversityofCapeTown.

`

TableofContents

ExecutiveSummary.........................................................................................................................................................2

TableofContents.............................................................................................................................................................3

ListofTables.......................................................................................................................................................................5

ListofFigures.....................................................................................................................................................................7

Glossary................................................................................................................................................................................8

Introduction........................................................................................................................................................................1

StructureoftheSATIMModel.....................................................................................................................................2

GeneralMethodologyforSectorAnalysisinSATIM..........................................................................................4

TheUseofsub‐Sectors................................................................................................................................................5

CompilingaReferenceCase......................................................................................................................................5

KeyDataSources–SouthAfrica’sIntegratedResourcePlan(IRP).........................................................5

ExogenousProjectionofDemand–GeneralApproachandDataSources...............................................6

EmissionsandSectoralEnergyModels...................................................................................................................8

IndustrySector...............................................................................................................................................................12

StructuralAssumptions&ModellingDecisions............................................................................................12

CompilingtheBaseYearConsumptionData‐AssumptionsandIssues.............................................17

AssumptionsCharacterisingTechnologiesandEnergyServices...........................................................19

ValidationoftheSectorModel..............................................................................................................................20

FutureDemandforEnergyfromtheIndustrySector................................................................................22

AgricultureSector.........................................................................................................................................................23

ModelStructure..........................................................................................................................................................23



FutureDemandforEnergyfromtheAgriculturalSector..........................................................................28

CommercialSector........................................................................................................................................................30

ModelStructure..........................................................................................................................................................30



FutureDemandforEnergyfromtheCommercialSector..........................................................................32

TransportSector............................................................................................................................................................34

ModelStructure..........................................................................................................................................................34

Sub‐ModelforProfilingBaseYearTechnologies–TheVehicleParcModel.....................................36

VehicleParcModelInputDataandAssumptions 39 CharacterisingNewVehicleTechnologies.......................................................................................................50

Futureimprovementsinvehiclefueleconomy 50

`

InvestmentcostsforNewTransportTechnologiesandConstraintsonMarketPenetration 52 Sub‐ModelforProjectingPassengerCarMotorisationandtheDemandforPassengerTravel–TheTimeBudgetModel...........................................................................................................................................52

Background‐Thedailytraveltimebudget 53 UsingtheTimeBudgetModeltoProjectDemandforPassengerTravelonLand 54 ProjectingDemandforFreightonLand............................................................................................................58

CalculationofFutureEnergyDemandfromRoadVehicles......................................................................59

ResidentialSector..........................................................................................................................................................61

ModelStructure..........................................................................................................................................................61

Householdclassificationprocess 61 CompilingtheBaseYearConsumptionData‐AssumptionsandIssues.............................................64

Assumptionsandconstraintsforreferencescenariointheresidentialsector................................65

ProjectingtheEvolutionofHouseholdIncome 65 Assumptionsonfuturetechnologyshare 67 FutureDemandforEnergyfromtheResidentialSector............................................................................67

SupplyofEnergyfromtheElectricitySupplySector......................................................................................70

ModelStructure..........................................................................................................................................................70

Transmission................................................................................................................................................................70

Distribution...................................................................................................................................................................71

Generation.....................................................................................................................................................................71

ExistingPowerPlants 72 ParameterizationofExistingPowerPlants 72 AdditionalParameterizationforCombinedHeat&Power(CHP)Plants 74 AdditionalParameterizationforExistingPumpStorageSystems 75 BasicStructure&AssumptionsforNew/FuturePowerPlants 75 ParameterizationofNewTechnologyPowerPlants 77 ParameterisationofNewPowerPlantTechnologiesintheProcessofBeingImplemented 79

SupplyofEnergyfromtheLiquidFuels&GasSupplySector.....................................................................80

ModelStructure..........................................................................................................................................................80

BaseYearData.............................................................................................................................................................80

AssumptionsCharacterisingRefineriesandRefiningProcesses...........................................................81

ParameterizationofRefineryTechnologies 81 CharacterizationofCTLTechnology 84 ModellingtheSupplyofAncillarySteamInputServicestoRefineries 84

SupplyofPrimaryEnergy..........................................................................................................................................85

References........................................................................................................................................................................86

Appendices.......................................................................................................................................................................89

`

ListofTables

TABLE1:GLOBALWARMINGPROPERTIESOFMAJORGREENHOUSEGASES.....................................................................................9 TABLE 2: EXAMPLE OF IPCC EMISSION FACTORS ‐ STATIONARY COMBUSTION OF SUB‐BITUMINOUS COAL BY ENERGY

INDUSTRIES(KG/TJONANETCALORIFICBASIS)......................................................................................................................11 TABLE 3: MEASURED EMISSION FACTORS FOR SELECTED SOUTHERN AFRICAN POWER PLANTS COMPARED TO IPCC

DEFAULTFACTORS(ZHOU,ETAL.,2009)................................................................................................................................11 TABLE 4:DISAGGREGATION OF THE SATIMMODEL BY INDUSTRY SUB‐SECTOR COMPARED TO THENATIONAL ENERGY

BALANCEANDLTMSMODEL.......................................................................................................................................................12 TABLE5:DISAGGREGATIONOFTHESATIMMODELBYENERGYSOURCECOMPAREDTOTHENATIONALENERGYBALANCE

...........................................................................................................................................................................................................13 TABLE6:ENDUSEDEMANDSMODELLEDFORTHEINDUSTRYSECTORINSATIMANDTHEENERGYSOURCE–TECHNOLOGY

CHAINTOSUPPLYTHEM...............................................................................................................................................................14 TABLE 7: ESTIMATES OF ATTAINABLE OVERALL EFFICIENCY IMPROVEMENTS FOR ELECTRICITY END‐USES IN THE

INDUSTRYSECTOR..........................................................................................................................................................................16 TABLE8:COMPARISONOFESTIMATESOFTHEAGGREGATECONSUMPTIONOFFUELS(PJ)FORTHESATIM,DOEANDIEA

ENERGYBALANCESFORSOUTHAFRICA‐2006.......................................................................................................................17 TABLE9:EVOLUTIONOFSUB‐SECTORESTIMATESOFCOALCONSUMPTIONFORTHEBASEYEAR(2006)...........................18 TABLE10:SECTORADJUSTMENTSTOCOALCONSUMPTIONINTHESATIMMODELRELATIVETOTHENATIONALENERGY

BALANCE..........................................................................................................................................................................................19 TABLE11:ASSUMEDSHAREOFELECTRICITYCONSUMPTIONBYENERGYSERVICESWITHININDUSTRYSUB‐SECTORS......19 TABLE12:ASSUMEDEFFICIENCYOFTECHNOLOGIESININDUSTRYSECTOR.................................................................................20 TABLE13:COMPARISONOFENERGYCONSUMPTIONBYSOURCE(PJ)FORSATIMANDTHEDOEENERGYBALANCEFOR

SOUTHAFRICA‐2006..................................................................................................................................................................25 TABLE14:IEPASSUMEDSHAREBYENERGYSERVICESOFUSEFULAGRICULTURALELECTRICALENERGYCONSUMPTION26 TABLE15:LTMSASSUMEDBREAKDOWNOFAGRICULTURALTOTALFINALCONSUMPTIONBYENERGYSERVICEFOR2001

...........................................................................................................................................................................................................27 TABLE16:EVOLUTIONOFASSUMPTIONSOFENERGYSUPPLYSHAREOFIRRIGATIONFINALENERGYBYENERGYSERVICES

...........................................................................................................................................................................................................27 TABLE17:SATIMASSUMEDENERGYSERVICESHARESOFAGRICULTURESECTORTFCBYENERGYSUPPLY......................28 TABLE18:ASSUMEDAGRICULTURALENERGYSERVICEEFFICIENCIESINTHESATIMMODEL................................................28 TABLE19:ASSUMPTIONSOFCHANGEINAGRICULTURALENERGYINTENSITYFORTHELTMS................................................29 TABLE20:COMMERCIALBUILDINGCATEGORIESUSEDIN2006STUDYCOMPAREDWITHPREVIOUSLTMSASSUMPTIONS30 TABLE21:COMPARISONOFESTIMATESOFTHEAGGREGATECONSUMPTIONOFFUELS(PJ)BYTHECOMMERCIALSECTOR

FORTHESATIMMODELANDTHEDOEENERGYBALANCEFORSOUTHAFRICA‐2006.................................................30 TABLE22:ESTIMATEDENERGYCARRIERSHAREOFENERGYCONSUMPTIONBYENDUSEANDFUEL....................................31 TABLE23:EXISTINGTRANSPORTSECTORTECHNOLOGIESINSATIMBASEYEAR.....................................................................34 TABLE24:KEYDATAINPUTSTOTHEVEHICLEPARCMODEL.........................................................................................................38 TABLE25:VEHICLECLASSESADOPTEDFORTHEVEHICLEPARCMODEL.......................................................................................39 TABLE26:VEHICLECLASSWEIBULLCOEFFICIENTS&RESULTINGAVERAGEAGESFORTHECALIBRATEDVEHICLEPARC

MODELCOMPAREDTOOTHERSTUDIES&SOURCES................................................................................................................42 TABLE27:ASSUMEDAVERAGEVEHICLEMILEAGE(KM/ANNUM)....................................................................................................45 TABLE28:AVERAGEANNUALMILEAGEPERVEHICLEFORPASSENGERMODESINVARIOUSAFRICANCITIES...........................46 TABLE29:IMPROVEMENTSINPASSENGERCARFUELECONOMYINWORLDMARKETS2000–2010.......................................47 TABLE30: CALIBRATEDMODELFUELECONOMY(L/100KM)BYVEHICLECLASSCOMPAREDTOOTHERSTUDIESAND

SOURCES–PART1..........................................................................................................................................................................48 TABLE31:AVERAGEOCCUPANCYPERVEHICLEFORPASSENGERMODESINVARIOUSAFRICANCITIES.....................................49 TABLE32:MODELOCCUPANCYANDLOADFACTORBYVEHICLECLASSCOMPAREDTOOTHERSTUDIESANDSOURCES..........49 TABLE33:ADDITIONALTECHNOLOGIESAVAILABLETOTHEMODELFORMEETINGFUTUREDEMANDFORTRANSPORT...50

`

TABLE34:IMPROVEMENTSINPASSENGERCARFUELECONOMYINWORLDMARKETS2000–2010........................................50 TABLE35:HYPOTHETICALFUELECONOMYIMPROVEMENTSCENARIOSFORGASOLINEPASSENGERCARSCOMPAREDTOTHE

CURRENTEFFICIENTNON‐HYBRIDGASOLINEPASSENGERCARS.............................................................................................51 TABLE36:ASSUMPTIONSANDCHECKSUSEDTOESTIMATEACCESSTOPRIVATEVEHICLESBYINCOMEGROUPS–BASEYEAR

(2006).............................................................................................................................................................................................55 TABLE37:TOTALDISTANCE,TIMEANDSPEEDFORTHENEWEUROPEANDRIVECYCLE.............................................................56 TABLE38:TIMEBUDGETMODELASSUMPTIONSFORCALCULATINGPASSENGERTRAVELDEMANDFORPUBLICANDPRIVATE

MODES...............................................................................................................................................................................................56 TABLE39:ORGANISATIONOFROADANDRAILFREIGHTDEMANDSUCHTHATTHEROAD/RAILSPLITCANBEKEPTCONSTANT

FORTHEBASECASE.........................................................................................................................................................................59 TABLE40:HOUSEHOLDSGROUPINGINCURRENTSOUTHAFRICANTIMESMODEL(SATIM)..................................................61 TABLE41:ENERGYINTENSIVEAPPLIANCES.......................................................................................................................................63 TABLE42:COMPARISONOFESTIMATESOFTHEAGGREGATECONSUMPTIONOFFUELS(PJ)BYTHECOMMERCIALSECTOR

FORTHESATIMMODELANDTHEDOEENERGYBALANCEFORSOUTHAFRICA‐2006.................................................64 TABLE43:MAPPINGOFSTATSSAINCOMEANDEXPENDITURESURVEYCATEGORIESANDCGEMODELLINGRESULTSTO

THREEINCOMECATEGORIES........................................................................................................................................................66 TABLE44:GENERICEXAMPLEOFASSUMEDTECHNOLOGYLIMITSHARES.....................................................................................67 TABLE45:ILLUSTRATIVEEXAMPLEOFASSUMPTIONSOFTHEELASTICITYOFUSEFULENERGYUSEBYHOUSEHOLDWITH

RESPECTTOHOUSEHOLDINCOMEFORINCREASINGHOUSEHOLDINCOME...........................................................................68 TABLE46:ASSUMEDFUTURESOUTHAFRICANELECTRIFICATIONRATESBYINCOMEGROUP..................................................68 TABLE47:DISTRIBUTIONTECHNOLOGIESINSATIM.......................................................................................................................71 TABLE48:SATIMAGGREGATIONOFEXISTINGPOWERPLANTS(BEFORE2010)INTOTECHNOLOGIES...............................72 TABLE49:SATIMPARAMETERIZATIONOFPOWERPLANTTECHNOLOGIES................................................................................72 TABLE50:ADDITIONALPARAMETERIZATIONINSATIMFORCHPPOWERPLANTS..................................................................74 TABLE51:ADDITIONALPARAMETERIZATIONINSATIMFORPUMPEDSTORAGEPOWERPLANTS.........................................75 TABLE52:SATIMPARAMETERIZATIONOFNEWPOWERPLANTTECHNOLOGIES......................................................................78 TABLE53:SATIMREFINERYOUTPUTS...............................................................................................................................................81 TABLE54:SATIMPARAMETERISATIONOFREFINERYTECHNOLOGIES........................................................................................81 TABLE55:ASSUMEDUPPERBOUNDSONOUTPUTCOMMODITYSHARESFORREFINERYTECHNOLOGIES..............................82 TABLE56:SUMMARYOFEXISTINGANDNEWREFINERYTECHNOLOGYCHARACTERISTICS.......................................................83 TABLE57:CURRENTREFINERYSTEAMBOILERTECHNOLOGIESIMPLEMENTEDINSATIM.....................................................84

`

ListofFigures

FIGURE1:SHAREOFTOTALFINALENERGYCONSUMPTIONBYSECTOR–DOEENERGYBALANCE2006................................3 FIGURE2:SHAREOFTOTALNATIONALENERGYSECTORGREENHOUSEGASEMISSIONS2000(MWAKASONDA,2009)......3 FIGURE3:TYPICALFRAMEWORKFOLLOWEDFORENERGYANALYSIS................................................................................................4 FIGURE4:LTMSGDPGROWTHRATEPROJECTION..............................................................................................................................7 FIGURE5:SATIMGDPGROWTHRATEPROJECTION............................................................................................................................7 FIGURE 6: SECTORAL SOURCE APPORTIONMENT OF GLOBAL GREENHOUSE GAS EMISSIONS – CO2 EQ TERMS (IPCC C,

2007)..................................................................................................................................................................................................8 FIGURE7:SOURCEAPPORTIONMENTOFGLOBALGREENHOUSEGASEMISSIONS–CO2EQTERMS(IPCCC,2007)...............9 FIGURE8:LTMSMETHODOLOGYFOR INCLUDINGEFFICIENCYPROGRAMSINANOPTIMISATIONMODELOFASIMPLIFIED

INDUSTRYSECTORWITHAGGREGATETECHNOLOGIESHAVINGUNKNOWNCOSTS............................................................15 FIGURE9:INDEXEDINTENSITYOFELECTRICITYCONSUMPTIONFORINDUSTRYSUB‐SECTORS1998–2009......................21 FIGURE10:INDEXEDPHYSICALOUTPUTFORINDUSTRYSUB‐SECTORS1998–2009..............................................................21 FIGURE11:RESDIAGRAMSHOWINGTHESTRUCTUREOFTHEAGRICULTURESECTORASREPRESENTED INTHESATIM

MODEL..............................................................................................................................................................................................24 FIGURE12:ENERGYSOURCESHARESOFAGRICULTURALENERGYCONSUMPTION(DOEENERGYBALANCE,2006).........24 FIGURE 13: IEP ASSUMED USEFUL AGRICULTURAL ENERGY CONSUMPTION DISAGGREGATED BY ENERGY SERVICE

DEMANDS(HOWELLS,KENNY,&SOLOMON,2002)...............................................................................................................26 FIGURE14:COMPARISONOFENERGYCONSUMPTIONBYENDUSEFOR2001WITHUPDATEDBASEYEAROF2006............32 FIGURE15:GENERALISEDSYSTEMOFPRE‐PROCESSINGTRANSPORTSECTORINPUTSFORSATIMUSINGSUB‐MODELS...36 FIGURE16:SYSTEMDIAGRAMANDDATATABLEFEATURESOFANALYTICALPLATFORM............................................................37 FIGURE17:SCHEMATICREPRESENTATIONOFTHEVEHICLEPARCMODELANDITSDATAINPUTSANDVALIDATIONS............38 FIGURE18:BASEYEARSCRAPPINGCURVESFORTHEVEHICLETECHNOLOGYTYPESINTHEVEHICLEPARCMODEL...............43 FIGURE19:EPAMOBILE6ANNUALMILEAGEDECAYASSUMPTIONSCOMPAREDTORESULTSOFVEHICLEACTIVITYSTUDY

FORNAIROBIKENYA......................................................................................................................................................................44 FIGURE20:ASSUMEDHISTORICALEVOLUTIONOFVEHICLEFUELECONOMYINTHEMODEL.......................................................47 FIGURE21:GOMPERTZCURVEMODELOFTHECORRELATIONOFMOTORISATIONWITHINCOMEPERCAPITADARGAYETAL.

(2007).............................................................................................................................................................................................52 FIGURE22:AVERAGEPER‐CAPITATRAVELBUDGETFORVARIOUSLOCALITIESANDREGIONSACROSSTHEGDPSPECTRUM

SCHAFER&VICTOR(2000)..........................................................................................................................................................53 FIGURE 23: CGE MODEL OUTPUT FOR ANNUAL HOUSEHOLD CONSUMPTION 2005‐2030 BY DECILES OF HOUSEHOLDS

AUTHOR’SCALCULATIONSUSINGDATAFROMALTONETAL.(2012).......................................................................................58 FIGURE24:APPLIANCESATURATIONLEVELS......................................................................................................................................62 FIGURE25:INDEXEDHOUSEHOLDSAPPLIANCEOWNERSHIP............................................................................................................63 FIGURE26:PROJECTEDHOUSEHOLDINCOMECATEGORYSHARE...................................................................................................67 FIGURE27:SIMPLIFIEDSCHEMATICOFSATIMMODELOFTHESOUTHAFRICANPOWERSECTOR.........................................70 FIGURE28:ASSUMEDREFURBISHMENT/RETIREMENTPROFILESFOREXISTINGESKOMPLANT...........................................74 FIGURE29:SCHEMATICOFPUMPEDSTORAGESUB‐MODELINSATIM.........................................................................................75 FIGURE30:NEWELECTRICITYPRODUCTIONTECHNOLOGIESAVAILABLEINSATIM................................................................76 FIGURE31: ILLUSTRATIVECOMPARISONOF SATIMCAPACITYAVAILABILITYFACTORS BYMODELTIMESLICE FORNEW

SOLARCENTRALRECEIVERTECHNOLOGIESHAVING3‐14HOURSOFSTORAGE–WINTERSEASON..........................79

`

Glossary

BRT:BusRapidTransportCSIR:CouncilforScientificandIndustrialResearchCTL:CoaltoLiquid(SyntheticFuelRefinerywithCoalFeedstock)GTL:GastoLiquid(SyntheticFuelRefinerywithNaturalGasFeedstock)HCV:HeavyCommercialVehicles(>8500kgGVM)IRP:IntegratedResoourcePlan–TheDepartmentofEnergy’slongtermelectricitysupplyplanLCV:LightCommercialVehicle(>3500kgGVM)MBT:Minibus‐TaxiMCV:MediumCommercialVehicle(3500–8500kgGVM)NAAMSA:NationalAssociationofAutomobileManufacturersofSouthAfricaNatis/eNatis:SouthAfricanNationalTrafficInformationSystempkm:passenger.kilometresSAPIA:SouthAfricanPetroleumIndustryAssociationSATIM: The South African Times Model, a TIMES‐MARKAL linear optimisation model of theSouthAfricanenergysystemSOL:StateofLogisticsReports(AnnuallypublishedbytheCSIR)Statistics South Africa (Stats SA): The government agency responsible for measuring andcollatingnationalstatisticsforSouthAfricaSUV:SportUtilityVehicle.Usuallyalarge4wheeldrivepassengervehiclet.km:ton.kilometres

1

Introduction

The Energy Research Centre’s TIMES energy model is now in its 3rd generation since thecompletion of the LTMS project. The current model is known as SATIM – the South AfricanTimes Model. This document aims to provide some insight into energy models throughaugmenting a brief overview of general approaches with a fairly detailed description of theSATIM model and its methodology, drawing not only on its current form but also, whererelevant,fromtheLTMSmodelswherethatapproachdiffered.This document aims to describe the sector analysis methodology of the SATIM model withparticularattentionto thedevelopmentofassumptionsthataccount forshortcomings indataand the practical necessity of simplifying complex economic and industrial structures. Thedocumentisstructuredsuchthatdedicatedsectionsarepresentedforeachprimaryeconomicsector prefaced by a general overview of sector analysis. Fundamentally, the modellingmethodologyofSATIMcharacterisesthedemandforenergybytheenergyservicesrequiredbya sector.These services are suppliedby technologies that require energy and thequantity ofthat energy supplywill dependon the efficiencyof the technology.The cost of supplying theservicewilldependonthecostoftheenergycarrier(fuels,electricityetc.)andthecostofthetechnology over time which together can be calculated as a levelised cost of supplying theservice.Themodelwillselecttechnologiestominimisethiscostsubjecttoconstraints.Inordertoarticulate theassumptionsrequiredbythismethodology foreachof the5demandsectors,Industry, Agriculture, Commercial, Transport and Residential, this document attempts tobroadlycoverthefollowing: Thestructureofthesectoranditsenergyservicesasitimpactsonthedemandforenergy Theestablishmentofbaseyeardemandforenergyinthesector

Technical and cost parameters of the technologies available to satisfy the demand forenergyservicescurrentlyandinthefuture.Technologycostsandtheprojectionoftheseareclearlycriticaltoanoptimisationmodelasareconstraintsonthepenetrationratesofnewtechnologies to reflect realistic replacement rates of existing technologies and non‐costbasedchoices.

TheprojectionoffuturedemandforenergyservicesThe level of detail for a sector depends on the relative contribution of the sector to totalconsumptionandalsoonhowmuchfundinghasbeenhistoricallyreceivedfordevelopingthatsector in the model. Thus Transport is quite detailed but Agriculture is not and is quitesimplisticallyrepresentedinSATIMbecauseinSouthAfricatheAgriculturesectoraccountsforrelativelysmallenergyconsumptionandlowemissions.Thetwosupplysectorsinvolvedintransformation,ElectricityandLiquidFuelsProductionaredescribedinsomedetailasregardstheirstructureandthetechnicalandcostparametersofthetechnologies available to supply the energy required by the demand sectors in the form ofelectricity,heatandliquidfuels.Thesesupplysectorsrequireprimaryenergyandthisispartof

2

the energy chain that must be costed to determine a levelised cost of supply. Assumptionsregardingprimaryenergysupplyarethereforebrieflydiscussedintherelevantsection.InSATIM,usefulenergyisanexogenousinputdisaggregatedbyenergycarrier,foreachdemandsector.Finalenergydemandisdeterminedendogenouslyusingtheassumedefficienciesoftheleastcostdemand‐sidetechnologiesselectedbythemodel.Thetwosupplysectorsandprimaryenergysourcesmustmeetthesumofthesedemands.Thesupplysectorsdonotthereforehavetheir own demand projections and projection of demand is not discussed in the supply‐sidesections.Themodeloptimiserwillselectsupply‐sidetechnologiestomeetthedemandforfinalenergyatleastcost.Somegeneraldiscussiononsectoralmodelling,theprojectionoffuturedemandforenergyfromsectorsforexogenousinputtothemodelandtheinclusionofemissionsinthemodelprecedesthesectorbysectordocumentation.

StructureoftheSATIMModel

Theeconomyofanationorregionconsumesenergy in the formofanumberofprimaryandsecondarysourceswhichdeliverservicesbymeansofamyriadoftechnologieslargeandsmall.A model of the demand for energy needs to capture this complex structure and thus thesesourcesandtechnologiesneedtobeorganisedinsomelogicalway.Optionsincludebysourceor by technology but more commonly demand models organise the demand for energy byeconomicsector.Thisisforanumberofreasons,particularlybecausedatatendstobecollectedbyeconomicsectorbutalsobecauseeconomicsectorswillhavemanyconsistencieswithregardtotechnologiesandenergysourcesthatfacilitatesimilartreatment.TIMES, the platform for SATIM, is a partial equilibrium linear optimisationmodel capable ofrepresenting thewhole energy system, including its economic costs and its emissions, and isthus particularly useful in modelling potential mitigation policies. The approach is howeverfundamentally sectoral as would be the case with other models, even simple spreadsheetmodels like MAED, and so the analysis of the structure of energy demand from sectors is afundamentalbuildingblockofthemodellingprocess,regardlessofthetoolsselected.The SATIM energy model has been constructed on a modelling platform called TIMES,developedbyETSAP,oneoftheInternationalEnergyAgency’simplementingagencies,andisasuccessor to MARKAL. TIMES is a partial equilibrium linear optimisation model capable ofrepresenting the whole energy system of a country, including its economic costs and itsemissions,andisthusparticularlyusefulinmodellingpotentialmitigationpolicies.Themodeliscapableofsolvingforavarietyofconstraintsincludingemissionsconstraints,bysector,forthewholeeconomy,orcumulativelyoveraperiod,andcanbeusedto identify themorecomplexconsequencesofmitigationactions.TheSouthAfricanTIMESmodelstructureiscontainedinadatabase,andconstructedviaauserinterfacecalledANSWER,whichprovidesaframeworkforbothstructuringthemodelandscenarios,andalsoforinterpretingresults.ANSWERcompiles

3

themodeldataintoasetof linearequations,whicharethensolvedbya linearsolversuchasCPLEX.TheSouthAfricanTIMESmodel(SATIM)isstructuredintofivedemandsectorsandtwosupplysectors– industry,agriculture,residentialcommercialandtransportonthedemandside,andelectricityandliquidfuelsonthesupplyside.Thesectorsvarysomewhatintheirshareoffinalconsumptionandgreenhousegasemissionsasshownbelow.

Figure1:ShareofTotalFinalEnergyConsumptionbySector–DOEEnergyBalance2006

Figure2:ShareofTotalNationalEnergySectorGreenhouseGasEmissions2000(Mwakasonda,2009)

4

Giventhesedisparities,notallsectorsinSATIMhaveenjoyedthesameinvestmentofresearch,andresearchfundinghastendedtoconcentrateonsectorshavingahighenvironmentalimpactand profile like the Transport sector and Electricity Supply sector. The Transport sector inparticulariswelldocumentedandincludesahighlevelofdetailduetooutsideinvestmentfromrecentprojects.Theassumptionsformodellingthissectorandtheirdevelopmentarethereforedescribed with more rigour than some of the other sectors. Agriculture on the other hand,accounting for less consumption in the energy balance than the unspecified sectors and lessthan5%oftotalgreenhousegasemissionsin2000(Mwakasonda,2009)isdealtwithrelativelysimplisticallyinSATIM.Inothercountries,theAgriculturesectorcanbeamajor,evendominantemitter, and a quite different approach to that of SATIMwould bewarranted. The industrysectorwarrantsagreaterlevelofdetailandthisispartoftheongoingprogramofimprovementin SATIM. The structure of each sector and assumptions around its technologies and energyservicesarecoveredintherespectivesectionsbelow.

GeneralMethodologyforSectorAnalysisinSATIM

Theenergymodellingprocess involvesthreefundamentalsteps,beingdatabasedevelopment,

energyanalysisandreviewevaluation,asdepictedinFigure3 below. These steps can be done iteratively, without following any order. The database development deals with assembly of all necessary data required to conduct an energy analysis. This step also involves calibrating the data for input into energy model.

Figure3:TypicalFrameworkfollowedforenergyanalysis

Source:IAEA,1984TheSouthAfricanTIMESmodelhasrelativelydetailedcharacterisationofthetechnologiesusedinboththedemandandsupplysectorsandthisthereforemakesupasubstantialportionofthedataandassumptionsrequiredtobuildthemodel.

5

TheUseofsub‐Sectors

Asub‐sector isaneconomicclassificationmoredisaggregatedthansector level.ExamplesareMining and Iron & Steel Manufacture within the Industry sector, and Passenger and Freightwithin the Transport ector. Organising demand and technologies by sub‐sectors is useful astechnologiesmaybeuseddifferentlyandhavedifferentcharacteristicswithinsub‐sectorsofasector.

CompilingaReferenceCase

Sincethekeydriversofenergydemandarefutureeconomicandpopulationgrowth,projectionof these are key assumptions for an energy demandmodelling exercise. On top of these twodrivers,thereareothervariablessuchassharesoftechnologiesused,changesintechnologicalefficiency and sectoral disaggregationwhich also need to be projected into the future. In anoptimisationmodel it is frequently necessary to constrain these parameters in themodel toactivelypreventtherapiddominanceofthecheapesttechnologyoptiontoreflecttherealitiesofconsumerpreference,policyandshortageofcapitalforreplacement.Scenario based energy modelling requires development of differing scenarios which areunderpinnedbyassumptions in the formof “What If?” analysis. In suchamodellingexercise,thereisalwaysareferencecase(alsotermedbusiness‐as‐usualscenario)whichwillbechangedormodifiedtoconstructotherscenarios.Thisisessentiallythedefaultversionofthefuture.Ascenario ismadeupof alternativeassumptionson factorssuchas future fueluse, fuelprices,and technology costs in different sectors, technology efficiencies and changes in technologymarketshares.Thispaperwill attempt tobrieflydescribe thecriticalassumptionsandconstraints thatwereusedtoconstructthereferencescenarioandthebasisforselectedassumptionsforeachoftheeconomicsectorsmodelledinSATIM.

KeyDataSources–SouthAfrica’sIntegratedResourcePlan(IRP)

The South African Department of Energy has undertaken an extensive study that includedmodelling toproducean IntegratedResourcePlan (IRP) forElectricity (DOE,2011).The firstreportwaspublishedinOctober2010withrevisionsreleasedin2011.Theoutputsinclude: Projectionsofeconomicgrowthfortheperiodextendingfrom2010to2035

A detailed review of Power Generation technology undertaken by the Electric PowerResearchInstitute(EPRI).Thisincludedthepublicationofdetailedtechnologydataandcostdataincludingprojectionsoflearningratesontechnologycosts

Variousfundamentalassumptionsrequiredformodellingsuchasdiscountrates.This work provides a rich source of data for SATIM particularly as regards costs in theElectricitySupplysectorandisreferredtofrequentlyinthisdocument.

6

Exogenous Projection of Demand – General Approach and DataSources

Thegeneralapproachusedinprojectingenergydemandfromaneconomicsectoristodefineanindicator/sthatreflectsthedemandforenergyservices.Forexample, inthecaseoftransportthiswouldbepassenger.kmandton.kmandforthecommercialsectorthismightbefloorarea.Ifabottomupmodelwithsufficienthistoricaldataisavailabletheindicatorcanbecorrelatedwith drivers like GDP or population using that data. Alternatively a rate of change in energyintensity(MJ/pass.kmforpassengertransportsay)needstobeestimatedandhistoricalvaluesfortheindicatorgenerated.The relationship derivedwith drivers like GDP, population and prices can then be projectedoverthestudyperiod.Theenergyservicesofabottom‐upmodelwillneedtomeetthisdemandfor the indicator, given assumptions of activity. To project future demand for useful energy,assumptions of energy intensities need to be derived, usually from historical trends, andmultipliedbytheprojectedindicator.

InSATIMthedemandforusefulenergyisexogenouslyinputforeachofthe5demandsectors.Thismaytaketheformofnaturalunits(sayPetajoules)oranindicatorthatreflectsthedemandfor energy services as yet unconverted into energy units. Currently in SATIM the demandprojectionsforallsectorsareinnaturalunitsexceptfortheTransportsectorwherethedemandindicatorbillionvehiclekilometresisinputtothemodel.Inthiscasethetechnologyefficiencyisnotinunitsof%asitwouldbeifusefuldemandwasinnaturalunitsbutinunitsofkm/MJandfinalenergyandtheusefulenergyservicefromthedemandprojectionrelateasfollows:

U=FX Equation1

U=UsefulEnergyService(billionvehiclekm)F=FinalEnergy(PJ)=TechnologyEfficiency(km/MJ)In undertaking demand projections for SATIM, gross domestic product and population projections were used for the demand sectors as these are the main drivers of energy demand. The population forecasts were adopted from the Centre for Actuarial Research (CARe) at the University of Cape Town which conduct demographic modelling, which is encapsulated in a model known as the Actuarial Society of South Africa (ASSA) population model. The GDP projection data was sourced

from Statistics South Africa. Figure 4 shows the GDP projection that was adopted in the LTMS

modelling. In LTMS the GDP was assumed to follow a GDP growth rate exhibited by most developed countries where GDP growth rate was assumed to rise continuously peaking in 2020 and then declining thereafter.

In SATIM, adifferentapproachwasadoptedtoobtaintheGDPforecastovertheperiod.TheE‐SAGE(EnergyextensiontotheSouthAfricanGeneralEquilibrium)wasusedinordertoprovidea more consistent framework for the growth of the economy as a whole, as well as for the

7

varioussectorsinthemodel.TheE‐SAGEmodelisusedtoprovidetheprojectionsforGDPandsectorgrowthfrom2010to2030(Arndt,Davies,&Thurlow,2011).InSATIM,theGDPgrowthrate projections are initially close to those of themoderate projectionsmade in the IRP (seeabove).HowevertheprojectionsusedremainmuchlowerthanthatofthemoderateprojectionsfromtheIRP2010,peakingat4.27%in2035.ThelowerGDPforecastusedinSATIM,showninFigure5,ismorerealisticconsideringthecurrentstateoftheglobaleconomy.

Figure4:LTMSGDPgrowthrateprojection

Figure5:SATIMGDPgrowthrateprojection

Another importantparameter toconsiderwhendoingenergymodelling that involvescosts isthediscountrate.InSATIM,theTreasury’sadvocatedrealdiscountrateof8%isusedandthesameratewasusedinIRP2010.

8

EmissionsandSectoralEnergyModels

In global terms all the economic sectors are significant contributors to anthropogenicgreenhousegasemissionsandofferpotentialformitigationstrategiesasindicatedbelow.

Figure6:SectoralSourceApportionmentofGlobalGreenhouseGasEmissions–CO2eqterms(IPCCc,2007)

Energy models and particularly optimisation models were originally designed aroundinfrastructureplanningbutwithincreasingconcernsaroundclimatechangeandairqualitytheynow find use in evaluating emissions mitigation interventions or planning infrastructureconstrained by limits on emissions. If a carbon tax is part of one of themodelling scenarios,emissions will need to be calculated to determine the levelised cost of new infrastructure.Energymodelsclearlylendthemselvestotrackingemissionsfromthecombustionoffuelsusedin transformationor toprovideenergyservicesbecausetheyaccount for thequantityof finalenergysuppliedandconsumed.Emissions however also arise from the extraction of energy commodities, fugitive emissionsfromcoalminingforinstance,aswellasothernon‐energyemissionsarisingfromthestorageoffuels or from chemical processes in industry. Agriculture can be a significant contributorthroughland‐usechange,entericfermentationordirectemissionsfromcultivation,particularlyrice.Wastefromallsectorsisthefourthmajoraggregatesourceofanthropogenicgreenhousegases.One of the decisions that needs to be made early on in using sectoral energy models inmitigation studies iswhether to try track all these emissions in the energymodel by linkingthem to the activity of technologies or whether to account for energy emissions only in themodelandaccountfortheotheremissionselsewhere,aninventoryforexample.Caremustbetakeninthe lattercase toadjustconstraints likenationalgreenhousegasemissiontargets torepresentjusttheenergyemissionportionthereofinthemodel.All atmospheric gases contribute to global warming to a greater or lesser degree and theIntergovernmental Panel on Climate Change publish data on 63 gases common to industrialemissions for the purpose of compiling inventories (IPCC b, 2007). Typically however, the

9

criticalemissionstoconsiderarethethreemajorgreenhousegasesCO2,CH4andN2Oforwhichsomebasicdataispresentedbelow.Table1:GlobalWarmingPropertiesofMajorGreenhouseGases

CarbonDioxide

(CO2)Methane(CH4)

NitrousOxide

(N2O)

Chlorofluoro‐

carbons(CFC’s)

AtmosphericLifetime(years)

50–200* 12 114 1.3–1,700

GlobalWarmingPotential

1 21 310 90–8,100

Sourceexceptwhereindicated:(IPCCb,2007)*:(Gutknecht&Akos)–theIPCCsourceusesarelativelycomplexresponsefunctiontodefineatmosphericlifetimesothisalternativesourcewasusedtorathershowanindicativerange

DataforCFC’sareincludedforcomparisontoshowthewiderangeofglobalwarmingpotentialsandatmosphericlifetimesofgreenhousegases.CO2,theemissionofgreatestconcernhasalowglobal warming potential and a fairly average lifetime in the atmosphere but its dominantcontribution stems from the massive rate of production by anthropogenic sources. Thedisproportionate contribution of greenhouse gases per unit quantity is dealt with by aconvention of expressing quantities as CO2 equivalent or CO2 eq for short. When annualemissionsofallgreenhousegasesareconvertedtoequivalentterms,CO2accountsforover75%oftheglobalanthropogenicgreenhousegasloadasshownbelow.

Figure7:SourceApportionmentofGlobalGreenhouseGasEmissions–CO2eqterms(IPCCc,2007)

Modellers therefore need to be careful in tracking whether emission factors reflect tons ofpollutantortonsCO2equivalentofthatpollutant.Intheformercasethedifferentgasesneedtobe converted to CO2 eq before being compared to a constraint or target. National targetsthemselvesmayonlyapplytoCO2emissionsinwhichcasetheothergasesneedtobeexcludedfrommodelconstraints.

10

Consistency across sectors is also a potential pitfall. If the Electricity Supply sector isrepresented in themodelandemissions fromelectricityproductionareaccounted for in thatmodule,theyshouldnotbedoublecountedintheconsumingsectors,Industryforexample.This discussion centres on greenhouse gases but if air quality is also a focus of amodellingstudy, the toxic emissions regulated in most countries are also often modelled. A detaileddiscussionisbeyondthescopeofthisdocumentbuttypicallythesewillincludethefollowing:

OxidesofSulphur(SOX) OxidesofNitrogen(NOX) CarbonMonoxide(CO) Non‐MethaneVolatileOrganicCompounds(NMVOC) ParticulateMatter(PM10andPM2.5)

Dependingon the levelofdetailof theassessmentNMVOC’smaybe furtherdisaggregatedbyspecies, benzene for example, or by organic family for example polycyclic aromatichydrocarbons (PAH’s)whichareknowncarcinogens.Airquality is oftena concern in studiesthatincluderoadtransportanddetailedresourcesforemissionfactorsoftheabovepollutantsexistinthepublicdomainformobilesources,forinstancetheEuropeanUnion’sCOPERTmodelortheUSEPA’sMobile6model.BothofthesesourcesarealsoreferencedintheIPCCemissionfactordatabase.In order to calculate total emissions in ourmodel emission factor data needs to be collectedalong with the efficiency and activity data for each technology. Emissions factors can beexpressedinanumberofways:

Typicallyforstationarycombustiononperunitenergybasis(tonspollutant/TJoffuelenergy),

Typicallyforproductionemissionsonapermassbasis(tonspollutant/tonoffuel), Typicallyformobilecombustionsourceslikecarsortrucksitcanbeactivitybased(kg

pollutant/kmtravelled).

Clearlyforanenergymodelthemostusefulformisonaperunitenergybasisbecausethisisthecommon quantity for all sectors, technologies and services in the model and this keepscalculationssimple.Activitybasedemissionsformotorvehicles,forexample,canbeconvertedtoaperenergybasisinapre‐processingexerciseusinganassumedfueleconomy.AtypicalsourceofgreenhousegasemissionsfactorsistheIPPCNationalGHGGuideline(IPCCa,2006) which includes an exhaustive public database of emission factors for sectors andactivities.Insomecasesthesearedisaggregatedbycountrybutpredominantlythesearedefaultfactorssuchasthefollowing:

11

Table 2: Example of IPCC Emission Factors ‐ Stationary Combustion of sub‐Bituminous Coal by EnergyIndustries(kg/TJonanetcalorificbasis)

IPCC2006Source/SinkCategory Gas Fuel2006 Value Unit

1.A.1‐EnergyIndustries CARBONDIOXIDE Sub‐BituminousCoal 96100 kg/TJ

1.A.1‐EnergyIndustries METHANE Sub‐BituminousCoal 1 kg/TJ

1.A.1‐EnergyIndustries NITROUSOXIDE Sub‐BituminousCoal 1.5 kg/TJ

Source:2006IPCCGuidelinesforNationalGreenhouseGasInventories,Volume2:Energy,Tables1.4and2.2ThesehavebeenextensivelyusedinSATIMbuttheguidingprincipleofemissionsinventoriesshouldbeappliedwhereverpossible.

Local emission factors should be usedwherever they are available from creditablesources

This is particularly true for a fuel like coal forwhich the chemical composition and calorificvalue varies markedly by region. Measurements by Zhou, et al., 2009 for power stations inSouthernAfricashowedsignificantdeviationsfromIPCCdefaultemissionfactors,inthecaseofone plant, even falling outside the IPCC range of uncertainty. The CO2 results,while in somecasesextreme,arenotinconsistentwiththehighashcontentandlowcarboncontentofmanycoalsusedforpowerproductionintheregion.Table3:MeasuredEmissionFactors forSelectedSouthernAfricanPowerPlantsCompared to IPCCDefault

Factors(Zhou,etal.,2009)

Measured IPCCDefaultFactorsIPCCUncertainty

Range

Country PlantName

CO2(kg/GJ)

NOx(kg/TJ)

CO2(kg/GJ)

NOx(kg/TJ)

CO2

(kg/GJ)NOx

(kg/TJ)

SouthAfrica

Kendal 96.3 0.446 94.6 1.5 89.5‐99.7 (0.5‐5)

Kendal(Repeat) 97.4 0.21 94.6 1.5 89.5‐99.8 (0.5‐5)

Lethabo 99.6 0.583 94.6 1.5 89.5‐99.7 (0.5‐5)

Arnot 95.3 0.28 96.1 1.5 92.8‐100 (0.5‐5)

Zimbabwe Bulawayo 97.5 0.07 94.6 1.5 89.5‐99.7 (0.5‐5)

Botswana Morupule 103.0 0.32 96.1 1.5 92.8‐100 (0.5‐5)Whilelocalemissionfactorsshouldtakeprecedenceoverageneralisedsource,thisshouldnotprecludeathoroughassessmentof theirveracity.BeforeadoptingtheveryhighCO2valuefortheBotswanaplantortheverylowNOxvaluefortheZimbabweanplant,forexample,thedatashould be subjected,wherever possible, to a scrutiny of its repeatability and a reviewof theinstrumentquality,calibrationproceduresandcertification.CO2emissionsshouldbalancethe

12

carbon content and calorific value and these should have been repeatably determined formultiplesamples.

IndustrySector

TheIndustrySectorinSouthAfricaaccountedfor40%offinalenergyconsumptionin2006andis, ingeneral,energyintensive,beingdominatedbyheavyindustriesintheminingandmetalsrefining sub‐sectors. Mitigation actions designed around energy efficiency programs wereidentified in the LTMS to be amongst the potentially most cost effective ways to reduceemissionsalthoughthenetreductionsattainablewereestimatedtobesignificantbutnotlarge(Hughes, Haw, Winkler, Marquard, & Merven , 2007). This sector is therefore an importantcomponentoftheSATIMmodel.

StructuralAssumptions&ModellingDecisions

Modelling thestructureof theenergychain from fuel input through toenergyservices in theindustry sector of any country will present a challenge because of the great diversity ofactivities and processes. Data collection is further complicated, relative to say the similarlycomplex but centrally regulated transport sector, by the large number of regulatory andindustry bodies. The challenges and expense involved with detailed disaggregation of theIndustrySectorhasresultedinafairlysimplestructuralapproachintheSATIMmodelwhichisdiscussedinmoredetailbelow.SATIM is more disaggregated by sub‐Sectors within the Industry Sector than the LTMSgenerationofthemodelasshownbelowandnowdisaggregatesallthemajorenergyconsumingIndustry sub‐Sectors. The smaller sub‐Sectors included in the national energy balancepublishedbytheDepartmentofEnergy(DOE)thatarereportedtocontributeonlyafractionofapercenttototalconsumptionarehoweveraggregatedintothesub‐Sector“Other”inSATIMasshownbelow.Table4:DisaggregationoftheSATIMModelbyIndustrysub‐SectorComparedtotheNationalEnergyBalance

andLTMSModel

DOE3EnergyBalance

DOEShareofTotal

Consumption(2006)

SATIM1 LTMS

MiningandQuarrying 16.9% MiningSIC2GoldMiningOtherMining

IronandSteel 28.0% Iron&Steel‐351

Manufacturing

ChemicalandPetrochemical 13.3% Chemicals‐33

Non‐FerrousMetals 4.5% Precious&Non‐Ferrousmetals‐352Non‐MetallicMinerals 4.8% N.M.MProducts‐34FoodandTobacco2 0.3% Food,Beverage&Tobacco‐30PaperPulpandPrint2 0.6% Pulp&PaperProducts‐323Construction 1.1% Other

13

Machinery 0.2%TextileandLeather 0.1%WoodandWoodProducts 0.1%TransportEquipment 0.02%Non‐specified(Industry) 30.2%1:NumbersareStandardIndustrialClassification(SIC)codes2:Theseaccountfor5%and7%ofconsumptioninSATIMafterrevisionofcoalconsumptiondata3:DOE–DepartmentofEnergy,RepublicofSouthAfrica

Similarly the disaggregation by energy source in SATIM accounts for 90% of energyconsumption by the Industry Sector according to the Department of Energy (DOE) energybalanceconsumptionasshownbelow.LPGandHFO(residualoil),althoughaccountingforonlyafractionofconsumptionin2006,areincludedbecausetheyareboilerfuelsandthisallowstheoptimisationmodelsomescopetoselectbetweenfuels,givenemissionsconstraints.WoodandBagassebiomassareincludedinSATIMbecause,asdiscussedinSection0below,theyrepresentafargreatershareofconsumptionthanindicatedbytheDOEenergybalance.Table5:DisaggregationoftheSATIMModelbyEnergySourceComparedtotheNationalEnergyBalance

DOEEnergyBalance DOEShareofTotalConsumption(2006)

SATIM

Electricity 39.15% ElectricityBituminousCoal 38.54% CoalGasworksGas 9.77% GasCokeovencoke 4.69% CokeovencokeGasDiesel 3.36% OilDieselCokingCoal 1.93%

BlastFurnaceGas 1.47%Bitumen# 0.68%Lubricants# 0.29%

OtherKerosene 0.06%MotorGasoline 0.05%ResidualFuel 0.02% OilHFO*WhiteSpirit 0.0026%

LPG 0.0004% OilLPG*ParaffinWax 0.0002%

Renewables&Waste ‐BiomassWoodBiomassBagasse

*:Includedasboilerfuelsforoptimisationwithemissionsconstraints#:non‐EnergyIntheSATIMmodeltheseenergysourcessupplytheenergychainthatmeetsthedemandforuseful energy services in each Industry sub‐Sector. The services currently included are asfollows:

14

Table6:EndUseDemandsModelled forthe IndustrySector inSATIMandtheEnergySource–TechnologyChaintoSupplyThem

EnergySource Technology EnergyServiceDemandsElectricity ElecHeating‐Electricity ElecHeatingElectricity Compressedair‐Electricity CompressedairElectricity Lighting‐Electricity LightingElectricity Cooling‐Electricity CoolingElectricity HVAC‐Electricity HVACElectricity Pumping‐Electricity PumpingElectricity Fans‐Electricity FansElectricity Othermotive‐Electricity OthermotiveElectricity Electrochemical‐Electricity ElectrochemicalElectricity boiler/processheating‐ Electricity boiler/processheatingCoal boiler/processheating‐ Coal boiler/processheatingGas boiler/processheating‐ Gas boiler/processheatingHFO boiler/processheating‐ HFO boiler/processheatingLPG boiler/processheating‐ LPG boiler/processheatingBiomassBagasse boiler/processheating‐ BiomassBagasse1 boiler/processheatingBiomassWood boiler/processheating‐ BiomassWood2 boiler/processheatingDiesel Transport–Diesel3 TransportNote:Theenergyservicedemandslistedapplytoallsub‐Sectorsexceptthosewhereindicatedotherwise1:Food,Beverage&Tobaccosub‐Sectoronly2:Pulp&PaperProductssub‐Sectoronly3:MiningandOthersub‐Sectorsonly

Thefollowingassumptionsapplytothemodelstructurebrieflydescribedabove:

IntheSATIMmodeloftheIndustrySectorthereisnodisaggregationoftechnologiesusingaparticularfuelforaparticularenduse.Alltechnologyshareswithinafuelandend‐usearetherefore100%andthetechnologycharacteristicsrepresenttheaverage.Thisreflectsthepracticallimitsoflocaldataandmaybechangedshouldbetterinformationbeavailable.

Currently, no capital, fixed or variable costs are loaded for these generic fuel specifictechnologies.Onlythefuelcostwillthereforecontributetocostingtheenergychain.

All thermal fuels except diesel are used for high temperature process heat. SATIM alsoassumesthatthesethermalfuelsarenotusedfortransformation.

DieselisassumedtobeconsumedbytheMiningand‘Other’(Construction)sub‐sectorsfortransport/traction.

Improvement in efficiency in industry is assumed to be slow and is only reflected inimprovedenergy intensity in theexogenousdemandcalculationandnotendogenouslybymoreefficientendusetechnologies.

15

Currently all sectors are modelled to experience electricity supply at national averagedistribution losses (10%). In general however large industries have their own substationsuppliedattheveryhighestdistributionvoltagesolosseswouldbelowerthaniscurrentlymodelledinSATIM.

The first two of these structural assumptions would seem to render an optimisation modelsterile and without scope for modelling mitigation actions because there is no competitionbetweentechnologiestomeetend‐usedemand,onlyfuelcompetitionbasedonfuelprice.Thereis however scope to add efficiency improvement programs to themodel by loading them asdummytechnologies,allocatingavariablecost(orcapitalcostwithalifetime)thatiscalculatedtogiveanappropriatepaybackperiodrelativetothemarginalcostofthefuelbeingsaved.Itisimportanttobearinmindthatthemodelcostofelectricity,beingthemarginalcostofsupplycalculatedby themodel,will likelybedifferent toareal life tariff. Suchdummytechnologieswere used for the LTMS which defined a number of industry efficiency ‘wedges’ in themitigationscenarios.Themodellingmethodologyisillustratedbelow.

Figure 8: LTMSMethodology for including Efficiency Programs in anOptimisationModel of a Simplified

IndustrySectorwithAggregateTechnologieshavingUnknownCosts

16

Clearly, the futurepenetration rate of thedummyEfficiencyProgram technologyneeds tobeboundedto themaximumpercentageenergysavingthat isestimatedtobepossiblegiventhenatureoftheintervention.An alternative to loading dummy technologies to reflect efficiency programs is to developscenarioswherebyfuturetechnologyefficienciesimproverelativetothebaseyear.Theprofileof these improvements will usually be derived from a look‐up table of projected net energysavingsperend‐use.Themodeldoesofcoursenotselecttheefficiencymeasuresendogenouslyinthiscase.SuchatablewouldlookmuchliketheattainableefficiencygainstablethatcameoutoftheLTMSstakeholderprocessshownbelow.Table7:Estimates ofAttainableOverall Efficiency Improvements forElectricityEnd‐Uses in the Industry

Sector

End‐Use 2008 2015 2030 2050

Boilersandsteamsystems 0% 10,10% 16,16% 20,20%

Compressedair 0% 7.5,7.5% 16,16% 20,20%

Processheat 0% 3,‐% 4,‐% 5,‐%

HVAC 0% 12,‐% 18,‐% 25,‐%

HVACwithwasteheat 0% 0% 10% 30%

Lighting 0% 30,10% 70,10% 75,10%

Othermotive 0% 9% 11% 15%

Pumping,fans(processflow) 0% 10% 25% 40%

Processcooling 0% 5% 7% 10%

Note:The tabledistinguishesbetween technologicalefficiencyand systems savings.TechnologyefficiencyimprovementsarelistedfirstandcommaseparatedfromsystemefficiencyimprovementsSource:(Winkler,H(ed.),2007)

Futuretechnologyefficienciescanbeestimatedusingthe followingequation forascenarioofenergyefficiencyimprovementlikethoseabovethatweredevelopedforscenarioassessmentintheLTMS.If:i: FutureTechnologyEfficiencyinYeari

0: BaseYearTechnologyEfficiencySi: Overall%EnergySavingsAchievedbyYeariFEi: FinalEnergyrequiredforanenergyserviceinyeariSi = (FEi–FE0)/FE0 Equation2

i = 0/(1‐Si) Equation3

17

CompilingtheBaseYearConsumptionData‐AssumptionsandIssues

ThedatasituationfortheindustrysectorinSouthAfricacanbesummedupasgoodelectricityconsumptiondatatopersub‐sectorlevelforcustomersofthenationalelectricitysupplyutility,ESKOM,butquestionableaggregatedataforotherenergysourcesandnon‐existentorveryoutofdateend‐usedatawithinsub‐sectors.Municipaldataisgenerallyunavailableevenatsector(eg: industry, commerce etc.) level. There is a lack of good studies on the Industry Sector inSouthAfricaexacerbatedbyageneralcultureofsecrecyenforcedbyconfidentialityclausesthatwithsomeexceptions,keepstheinformationcollectedbyindustriesfortheirownplanningoutofthepublicdomain.Themainsourcesforindustrialdatarelatedtoenergyconsumptionandoutputare:

ESKOM‐ElectricitysalesbySICCODE(2ndand3rdlevelwhereneeded),

DOEEnergyBalance‐Industrialconsumptionbyfueltype

NERSA‐Municipalelectricitydistributedtoindustry

STATSSA ‐Physical output from industry [STATTSA 2002], Value added by industrialsectors[STATSSA,

IndustryAssociations

Annualreports

Personalcommunicationandinternetsearches

This means a bottom‐up model can be fairly well calibrated for electricity but attributingconsumptiontoendusesreliesheavilyonassumptions.Forotherenergysourcestheaggregateconsumptiondata from thenational energybalance is sometimes also problematic as can beseen fromTable8belowwhichcontrasts theaggregateSATIMsub‐sectorconsumptionswiththe official national energy balance and the IEA energy balance for the base year 2006. Thecompilersoftheenergybalancefacethesamechallengesasmodellersincollatingdataacrossthe disparate activities in industry and gaps are possible where energy like biomass is nottraded through a traceable retail structure by a small number of entities in the way thatelectricityis.Table8:ComparisonofEstimatesof theAggregateConsumptionofFuels (PJ) for theSATIM,DOEand IEA

EnergyBalancesforSouthAfrica‐2006

Fuel Coal Electricity Gas OilDiesel

OilHFO

OilLPG

BiomassWood

BiomassBagasse

SATIM 536 429 105 36.0 0.17 0.004 41.2 10.0DOEEB1 413 420 105 36.0 0.17 0.004 0.00 0.00IEA2 328 408 94 35.6 72.39*

*Biofuels&Waste1:(DOE,2009)–SouthAfricanDepartmentofEnergy2:(IEA,2009)–InternationalEnergyAgency

DifferencesarenotableforWood,BagasseandCoalconsumption.Thefirstoftheseisbasedonindustry information that indicates that wood waste supplies a large portion of the energy

18

needsofthePulpandPaperIndustry.SouthAfricahasalargesugarindustryandwhilebagasseiscertainlyanenergysourceinthatindustry,theSATIMbagasseconsumptionvalueiscurrentlyaroughorderofmagnitudeestimate.ThetotalbiomassestimateishoweverinfairagreementwiththatoftheIEA.Coal used for process heating is however the dominant non‐electricity source and thereforemostoftheeffortinmitigatingerrorsinbaseyearconsumptiondatahasfocussedonimprovingthedataforcoalusebyindustry.UnfortunatelycoalconsumptiondatainSouthAfrica ispoorandalternativefigureshavegenerallybeenderivedbymeansofinferenceandassumption.Theassumptionsofsub‐sectorcoalusefortheSATIMmodelrelyheavilyonthefollowing:

Nationalstatisticsforexpenditureflowsbetweensub‐sectors,called‘SupplyandUse’tablespublished by Statistics South Africa (Stats SA). Because the prices which individualindustriespayforcoal isunknown,energyconsumptionestimitesderivedfromthetablesaredoneusingthesamecoalpriceforallsectorsandgradesofcoal.

This data is supplemented with Annual reports to shareholders published by dominantfirms in industries and a newly emerged corporate governance document, the so‐called‘SustainabilityReport’.

The resulting base year data is compared to historical estimates from publications andpersonalcommunications(confidential)fromindustryinsiderstotryandpreventanygrosserrors.

Thus, although the national energy balance attributes no coal consumption to the Pulp andPaperindustry,thesupplyandusetablesindicatethatthisindustryspendsalargeamountofmoneyoncoal.Thequantitycanbeestimatedfromthatexpenditureor fromdata in industryreportsfromthetwolargepapermanufacturers,beingSappiandMondi.Table9belowshowshow a sub‐sector disaggregation of coal use has developed in contrast to the DOE’s nationalenergybalance.Table9:EvolutionofSub‐SectorEstimatesofCoalConsumptionfortheBaseYear(2006)

Sub‐Sector DOE SATIMIronandsteel 117.4 125.4ChemicalandPetrochemical 50.8 50.8Non‐FerrousMetals 0.0 0.8Non‐MetallicMinerals 20.3 54.0FoodandTobacco 0.0 32.4PaperPulpandPrint 0.0 64.8Other 171.5 154.5Mining 53.3 53.3Total 413.3 536.0Note:WhereLTMSandSATIMvaluesagreewithDOE(italics)themodelwas/iscalibratedtotheDOEEnergyBalancevalue for thebaseyearwithoutamendment.Alternativeestimatesderivedfromsupplyandusetablesorotherresearchareinbold

19

To some extent these efforts to attribute coal to industry sub‐sectors reflects the need toreallocate excess coal from other sectors where the energy balance consumption seemsimplausible, for instance in the Residential Sector. The total national consumption of coal ishowever fairlywellknownand thenationalenergybalancecanbeconsideredreliable in thiscase.ThusthesumofcoalconsumptionfromallsectorsinSATIMshouldbalancethisnumber.TheSATIMbottom‐upcalculationofhouseholdcoaluseintheResidentialSectoryieldedamuchlowernumber than thenational energybalanceand thisdifferencehasbeenbalancedby theextracoalallocatedtotheIndustrySectorintheSATIMmodelasshownbelowinTable10.Table10:SectorAdjustmentstoCoalConsumptionintheSATIMModelRelative

totheNationalEnergyBalance

BalanceIndustrySector(PJ)

ResidentialSector(PJ)

NationalEnergyBalance 413.3 152.6SATIMModel 536.0 26.8

Difference 122.7 ‐125.8

AssumptionsCharacterisingTechnologiesandEnergyServices

There have been relatively few studies in SouthAfricawhich have looked at energy serviceswithin industry. The earliest notable study and to date themost complete is that of Bennett(Bennett, 1975). Whilst several sectors still rely on the same processes, due to continuedprocessefficiencyimprovements,thisstudyisoutofdateandthereforeindicativeatbest.Otherthanthese,thereareisolatedexamplesrelatedtoenergyservices,forexampleprocessheatingin the food and beverage industry. These studies were used along with international datarelatedtoenergyservices.Thebestsourceof internationalendusedatawasfoundontheUSEIAwebsite(EIA,1998).TheapproachtakenwastosupplementtheSouthAfricandataavailablefromlocalstudieswithinternationalbenchmarksandpersonalopinionswherenootherdatawasavailable.Table11belowshowstheassumedpercentageofelectricityattributedtoeachenergyservice.Table11:AssumedShareofElectricityConsumptionbyEnergyServiceswithinIndustrySub‐Sectors

EnergyServiceFractionalSharesbySub‐sector

Mining Iron&Steel

Chemi‐cals

Precious&Non‐Ferrousmetals

N.M.MProducts

Food,Beverage&Tobacco

Pulp&PaperProducts

Other

ElecHeating 2.0% 40.0% 2.0% 1.0% 23.0% 7.0% 2.0% 9.7%

Compressedair

18.6% 5.0% 15.0% 0.0% 13.0% 4.0% 35.0% 10.9%

Lighting 4.5% 3.5% 4.0% 1.0% 5.0% 5.0% 10.0% 8.1%

Cooling 8.1% 1.0% 5.0% 0.0% 0.0% 23.0% 0.0% 5.1%

HVAC 8.0% 2.0% 2.0% 1.0% 2.0% 6.0% 4.0% 8.5%

20

Pumping 17.9% 2.5% 35.0% 0.0% 8.5% 28.0% 35.0% 13.0%

Fans 6.9% 4.5% 8.0% 0.0% 9.0% 4.0% 0.0% 5.5%

Othermotive 33.8% 41.5% 20.0% 7.0% 39.5% 21.0% 14.0% 36.9%

Electrochemical

0.2% 0.0% 8.0% 90.0% 0.0% 0.0% 0.0% 1.3%

Boiler/processheating

0.0% 0.0% 1.0% 0.0% 0.0% 2.0% 0.0% 1.0%

Total 100% 100% 100% 100% 100% 100% 100% 100%

Theefficienciesofenergyserviceswereestimatedtobeasfollows:Table12:AssumedEfficiencyofTechnologiesinIndustrySector

Technology Efficiency

ElecHeating‐Electricity 100%

Compressedair‐Electricity 5%

Lighting‐Electricity 30%

Cooling‐Electricity 200%

HVAC‐Electricity 90%

Pumping‐Electricity 80%

Fans‐Electricity 80%

Othermotive‐Electricity 80%

Electrochemical‐Electricity 76%

boiler/processheating‐Electricity 76%

boiler/processheating‐Coal 64%

boiler/processheating‐Gas 72%

boiler/processheating‐HFO 68%

boiler/processheating‐LPG 72%

boiler/processheating‐BiomassBagasse 60%

boiler/processheating‐BiomassWood 60%CurrentlySATIMdoesnotintroducenewtechnologiesinthefuturefortheIndustrySectorandfor the reference case efficiencies are assumed to stay constant. The model can only selectbetween technologies for boiler/process heating and so upper and lower bounds to preventimprobableratesofpenetrationhavebeendefinedforthesetechnologiesonly.

ValidationoftheSectorModel

Theendusedataandassumptions inSATIM forall sectors, including the IndustrySectorarecalibratedtotheaggregateconsumptiondatafromtheenergybalanceoranadjustmentofthatfigure.Thisistheprimaryvalidationofanybottom‐upmodel.For the industry sector an additional check is performed at sub‐sector level for electricityconsumption whereby electrical energy intensity is calculated for an extended period.Electricity intensity is calculatedbydividing thephysicaloutputofgoodsproducedbya sub‐

21

sectorasreportedbyStatsSAbyanestimateofelectricalenergyconsumedbasedonESKOMandmunicipalitysalesdata.Theobjectiveistocheckthatenergyintensityforeachsub‐Sectorissensible and continuous. As can be seen from Figure 9 below energy intensity formost sub‐Sectorsapparentlydecreasedgraduallyorwasstableuntil2006andthenrose,rapidly in thecaseofIron&Steel,till2009.

Figure9:IndexedIntensityofElectricityConsumptionforIndustrysub‐Sectors1998–2009

The rapid rise in energy intensity between 2006 and 2009 relates to the rapid drop inproduction in response to the effects of the credit crunch which affected Iron & Steel mostseverelyasshownbyFigure10below.Energyintensitytendstocorrelatewithoutputbecausea drop in production equates to plant running at low capacity factor which tends to beinefficient.

Figure10:IndexedPhysicalOutputforIndustrysub‐Sectors1998–2009

0.5

0.6

0.7

0.8

0.9

1.0

1.1

1.2

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Indexed Electricity Intensity

Iron and steel

Chemical and Petrochemical

Non‐Ferrous Metals

Non‐Metallic Minerals

Food and Tobacco

Paper Pulp and Print

Mining

0.8

0.9

1.0

1.1

1.2

1.3

1.4

1.5

1.6

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Indexed Physical Output

Iron and steel

Chemical and Petrochemical

Non‐Ferrous Metals

Non‐Metallic Minerals

Food and Tobacco

Paper Pulp and Print

Mining

22

FutureDemandforEnergyfromtheIndustrySector

TheprojectionofdemandfortheindustrysectorinSATIMreliesonthreeinputs:

Analysis of the projections for year on year GDP growth for industry sub‐sectors outputfromaComputableGeneralEquilibrium(CGE)model.Theseresultsweremadeavailablebya collaborating research group, the University of WIDER in Finland. This CGEmodel forSouthAfricawasdevelopedtostudytheeconomicimplicationsofintroducingcarbontaxesinSouthAfricaforthepurposeofgreenhousegasemissionsmitigation(Alton,etal.,2012).

TherelationshipbetweenGDPprojectionsforthesectorandsectoraloutputwherephysicalquantitiyisusedastheunitforenergyintensity.

Estimates of historical energy intensity of industry sub‐sectors in terms of either valueaddedorphysicalquantityofproductdependingonthesub‐sector.

Thefuturedemandforenergyservicesintheindustrysectoristhencomputedinthefollowingsteps:

1. Thegrowthinactivity(Randsofoutputperyear) ineachsubsector isprojectedusingthe

economicCGEmodel,withsectorsaggregatedtomatchthoseintheenergymodel.Activitygrowthisrecordedeitherasvalueaddinthesectororphysicaloutput.

2. Theintensity(energy/randofoutputorenergy/physicalunitofproduction)ofeachsectorisprojectedbyextrapolatinghistoricalobservations.

3. Final energy is calculatedbymultiplying the intensity in each sectorby theoutput of thesectorovertime,finalenergyisattributedtoendusesusingtheestimatesinTable11

4. Usefulenergydemandiscalculatedforeachenduse inthesub‐sectorsbymultiplyingthefinalenergycalculatedbytheassumedefficiencyofelectricalandthermalconsumption.

23

AgricultureSector

Although relatively small in monetary and energy terms, the agricultural sector plays animportant role in the South African economy. Primary agriculture was estimated by theDepartmentofEnergy(DOE)toaccountfor2.6%ofTotalFinalConsumption(TFC)ofEnergyin2006(DOE,2009)whichcloselymirrorsthe2.4%contributionofthesectortogrossdomesticproduct(GDP)inthatyear(StatsSA,2012).Energyemissionsfromthesectorwereestimatedtocontribute less than 1% to total greenhouse gas emissions in 2000 (Mwakasonda, 2009) butwhenother agricultural sourcesofGHGemissions likeenteric fermentation,biomassburningandN2Oemissionsfrommanagedsoilsareconsidered,thisshareoftotalemissionsapproachesamoresignificant6%.SouthAfricahasatotallandareaof122millionhectares,ofwhich82%(100millionhectares)isfarmland.Farmlandisprimarilyusedforlivestockrearingandcropproduction.SouthAfrica’slandresourcewithsufficient rainfall tobeconsideredarable isestimated tobe14percentoffarmland(14millionha).Drylandfarmingispractisedon11.2millionhawithonly1.2millionha under irrigation. The latter area nonetheless produces 25 to 30% of the country’sagriculturaloutput (Gbetibouo&Ringler,2008).The implication is thatagrowingpopulationmaydrivesignificantgrowthinenergyconsumptionbyirrigation.

ModelStructure

Fiveenergyserviceshavebeenidentifiedinagricultureforanalysispurposes(Winkler,Hetal,2006): Traction Irrigation

Heating Processingand ‘Other’purposessuchaslightingandcoolingGiven the relatively small contribution to TFC of the Agriculture Sector, this module of theSATIMmodeliskeptquitesimple,andthewholeenergychainofthemodelcanberepresentedbytheReferenceEnergySystem(RES)diagramshownbelow.

24

Figure11:RESDiagramShowingtheStructureoftheAgricultureSectorasRepresentedintheSATIMModel


Dieselisthemostimportantenergysourceintheagriculturalsectorandaccountsformorethanhalfoftheenergyconsumed,orjustover38PJin2006.Dieselisprimarilyusedtofuelvehiclessuch tractors and combine harvesters. Electricity is also a significant source of energy in theagricultural sector, accounting for 30%of the energy consumed by the agricultural sector in2006.Motorgasoline,otherkerosene,heavyfueloilsandcoalaccountfortheremainingenergyconsumed.

Figure12:EnergySourceSharesofAgriculturalEnergyConsumption(DOEEnergyBalance,2006)

Coal1,1%

Motor Gasoline7,2%

Other Kerosine3,6%

Gas Diesel55,2%

Heavy Fuel Oil2,5%

Electricity30,4%

25

Comparison of historical statistics has shown good agreement between the DOE electricityconsumption estimate and ESKOM’s sales data. Likewise thermal fuels estimates show goodhistorical agreement with the sectoral estimates of the South African Petroleum IndustryAssociation(SAPIA).AtthistimethereforetheDOEenergybalanceisassumedcorrectandthedataapplieddirectly inSATIM.Ascanbeseen fromTable13below, theenergyconsumptiondisaggregatedbyenergy source inSATIMagrees closelywith thenational energybalance forthebaseyear2006,omittingonlynon‐energysourcesandaccountingfor98.5%ofagriculturaltotalfinalconsumption.Table13:ComparisonofEnergyConsumptionbySource(PJ)forSATIMandtheDOEEnergyBalanceforSouth

Africa‐2006

EnergySource DOEEB(PJ) SATIM(PJ)

OilDiesel 38.23 38.23Electricity 21.03 21.03OilGasoline 4.96 4.96OilParaffin 2.51 2.51OilHFO 1.76 1.76Coal 0.76 0.76OilLPG 0.001 0.001Lubricants 0.85 ‐WhiteSpirit 0.19 ‐Total 70.28 69.25


Thestudysupportingthe2003SouthAfricanIntegratedEnergyPlan(IEP)(Howells,Kenny,&Solomon, 2002) modelled energy services by sector for input to a MARKAL model, apredecessor of the current SATIM model. This study derived the following breakdown ofelectricityusebytheagriculturesectorusingrelativelydisaggregateenergyservicessimilarindetailtotheindustrialsector.

26

Figure13:IEPAssumedUsefulAgriculturalEnergyConsumptionDisaggregatedbyEnergyServiceDemands

(Howells,Kenny,&Solomon,2002)

Thetablebelowshowsthepercentageshareofeachenergyserviceofusefulelectricalenergyconsumptionforthebaseyear2001.Table14:IEPAssumedSharebyEnergyServicesofUsefulAgriculturalElectricalEnergyConsumption

Pumping 29.0%

Process 10.0%

OtherMotive 3.2%

MaterialsHandling 16.0%

Lighting 4.1%

Homes&Hostels 20.4%

HVAC 6.4%

Fans 1.5%

Cooling 9.5%ThelaterLTMSmodelsimplifiedthenumberofenergyservices,presumablybecauseofthelackofdataandstudies,aggregatingtheaboveasfollows:

Electrical, Diesel and Gasoline Powered ‘Pumping’ became ‘Irrigation’ retaining theassumption fromtheIEPthat IrrigationFinalEnergyShare is76%Electricity,12%Dieseland12%Gasolinewhichtranslates toausefulenergyshareof91%Electricity,5%Dieseland 4% Gasoline based on efficiencies of 95% for electrical pumping, 35% for dieselpumpingand25%forgasolinepumping.

‘Process’and‘MaterialsHandling’wereaggregatedinto‘Processing’ AllremainingIEPelectricalenergyserviceswereaggregatedinto‘Other’

0

2

4

6

8

10

12

14P

J U

sefu

l

2001 2003 2005 2007 2009 2011 2013 2015 2017 2019

Year

Pumping

Process

Other motive

Mat handling

Lighting

Homes & hostels

HVAC

Fans

Cooling

27

This yielded the following share of energy services for the agricultural analysis base year of2001 with diesel and gasoline not used in ‘Irrigation’ attributed to ‘Tractors, harvesters &transport’andallotheruseofpetroleumproductsintheenergybalanceassumedtobeusedinheating.Table15:LTMSAssumedBreakdownofAgriculturalTotalFinalConsumptionbyEnergyServicefor2001

EnergyService ShareofFinalEnergy ShareofUsefulEnergyIrrigation 11.5%1 16.2%Tractors,harvestersandtransport 59.5% 36.6%Processing 7.8%2 13.9%Heat 7.7% 9.4%Other 13.5%3 24.0%Total 100.0% 100%1:Calculatedas(29%XAgriculturalElectricityTFC)+12%DieselShare+12%GasolineShare2:Calculatedas[16%+10%]XAgriculturalElectricityTFC3:Calculatedas(1‐[29%+16%+10%])XAgriculturalElectricityTFC

The 2003 IEP assumptions of the share of energy services of total electrical energyconsumption have therefore essentially been retained. The only major change to the LTMSapproachtotheagriculturesectorisinthesplitoffinalenergyallocatedto‘Irrigation’byenergysupply as shown below. This reflects the electrification of rural areas and the increasingdominance of large industrialised commercial farms such that diesel or petroleum poweredpumpingforirrigationisalmostnegligible.Table16:EvolutionofAssumptionsofEnergySupplyShareofIrrigationFinalEnergybyEnergyServices

Fuel IEP(2002) LTMS(2007) SATIM(2011)Irrigation\Elec 76% 76% 95%Irrigation\Diesel 12% 12% 4%Irrigation\Petrol 12% 12% 1%TheenergyservicesharesoffinalenergyforSATIMaredeterminedbythefollowingstepsforthebaseyear2006,assumingthe2006DOEenergybalancedatatobecorrect.

Allthermalfuelsotherthandieselandpetrolareallocatedto ‘Heat’.ThisyieldsashareofTFCof7.3%ratherthanthe7.7%inTable15sotheothersharesareadjustedslightlyup.

This adjustment yields an irrigation share of TFC of 11.5%which is then split by energysupplyusingtheassumptionsinTable16.

Allremainingdieselandgasoline in theenergybalance isassumedtobeused inTractionandTransport.

All‘Processing’isassumedtobesuppliedbyelectricityandiscalculatedusingtheadjustedshare inTable 15.Aswehave seen this rests on the assumption that ‘Process’ is 26%ofElectricityTFCaspertheIEP,thesumof10%'Process'and16%'MaterialsHandling'.

28

All'Other'energyservicesareassumedtobesuppliedbyelectricity.Thiswouldbelighting,fans, cooling etc. As we have seen this rests on the assumption that ‘Other’ is 45% ofElectricityTFCasperanaggregationofenergyservicesassumedintheIEP.

ThisyieldsthefollowingbreakdownofagriculturalTFCbyenergysupplyandenergyservice:Table17:SATIMAssumedEnergyServiceSharesofAgricultureSectorTFCbyEnergySupply

EnergyService

Coal OilDiesel

Elec‐tricity

OilGasoline

OilHFO OilParaffin

OilLPG

Heating 100.0% 100.0% 100.0% 100.0%Processing 26.1%Traction 99.2% 98.4%Irrigation 0.8% 36.1% 1.6%Other 37.8%SUM 100.0% 100.0% 100.0% 100.0% 100.0% 100.0% 100.0%The following efficiencies are assumed in converting final to useful energy for agriculturalenergyservicesintheSATIMmodel.Table18:AssumedAgriculturalEnergyServiceEfficienciesintheSATIMModel

EnergyService EnergySupply Efficiency

Heating

Coal 65%OilHFO 70%OilParaffin 70%Electricity 75%

Processing Electricity 100%1

TractionOilDiesel 35%Electricity 95%OilGasoline 25%

IrrigationOilDiesel 35%Electricity 95%OilGasoline 25%

Other Electricity 100%1

1:Theseaggregateenergyservicesaretreatedasabstractwithanefficiencyof100%

FutureDemandforEnergyfromtheAgriculturalSector

Although the balance has steadily swung in favour of imports over the last 50 years, SouthAfricaisstillreportedtobeanetfoodexporter(SAIRR,2012)(FAO,2011),exporting30–47%morefoodthanitimportedin2006/2007(FAO,2011).Giventhereforethatnotallproductionis consumed locally, Value Added (GDP) is used as the driver for energy demand in theagriculturalsector.

IntheLTMS,assumptionsweremadeastothefuturechangeinenergyintensityofservicesintheAgriculturalSectorasfollows:

29

Table19:AssumptionsofChangeinAgriculturalEnergyIntensityfortheLTMS

Intensity[GJ/2003Rands]

EnergyServiceReductionfactor 2001 2020 2030

Irrigation 0.5% 0.201 0.220 0.232Traction ‐0.5% 0.453 0.412 0.392Processing ‐0.5% 0.172 0.156 0.149Heat ‐1.0% 0.117 0.096 0.087Other 0.5% 0.298 0.328 0.344

In SATIM however, until such time as better data can be acquired on which to base theseassumptions, energy intensityhasbeenassumed constant over theprojectionperiod and theelasticityofenergydemandwithrespecttovalueaddedhasbeensetto1.InSATIMcurrentlytherefore, agricultural energy demand is assumed to grow directly proportional to GDP. InSouthAfrica’scasewheretheAgricultureSectoraccountsforlessthan3%oftotalfinalenergyconsumption,thisisunlikelytobeamajorsourceoferror.Inothereconomies,especiallythosewhere agriculture is a dominant contributor to the economy, energy intensity and elasticitywouldhavetobemorepreciselydetermined.

ThegrowthinValueAddedfortheAgriculturalSectorisderivedfromtheoutputoftheE‐SAGEComputable General Equilibrium (CGE) model which includes projections for GDP by sectorfrom2010to2030(Arndt,Davies,&Thurlow,2011).ThisprojectsAgriculturalGDPgrowthofaround3.5%annualisedand isgenerally lower thanmostothersectorsandsub‐sectorsovertheprojectionperiod.

30

CommercialSector

ThefollowingdescriptionofthemodellingoftheCommercialSectorinSATIMdrawsheavilyonthe equivalent section of the Low Emissions Pathways Technical Report (Energy ResearchCentre,2011)

ModelStructure

InSATIMthecommercialsectorincludesallnon‐residentialbuildings,excludingbuildingsusedfor industrial and agricultural activities. The sector is further divided into three buildingactivities based on Statistics South Africa building data categories (StatsSA, 2010a). Eachbuilding activitywas linked to the Statistics SouthAfrica economic data categories (Stats SA,2010b). The percentage share of floor area for each building activity in 2006was based onbuildingscompletedbetween1993and2006(Table20).Table20:Commercialbuildingcategoriesusedin2006studycomparedwithpreviousLTMSassumptions

Buildingactivity EconomicsectorPercentageshareoffloorarea2001

Percentageshareoffloorarea2006

ShoppingspaceWholesale,retail,motortradeandaccommodation 37% 36%

Officeandbankingspace

Finance,realestateandbusinessservices 30% 39%

Othernon‐residentialspace

Personalservices 33% 24%

Note:Thebuildingcategoriesof 'industrial&warehouse'and 'additionsandalterations'wereexcludedastheycouldnotbedisaggregatedbysector.


Aside fromminordifferences forelectricity,paraffinandLPG, theaggregateconsumptionsbyfuel assumed for the Commercial Sector in SATIM have been sourced from the DOE EnergyBalancefor2006asshownbelow.Table21:ComparisonofEstimatesoftheAggregateConsumptionofFuels(PJ)bytheCommercialSectorfor

theSATIMModelandtheDOEEnergyBalanceforSouthAfrica‐2006

Fuel Coal OilDiesel Elec‐tricity Gas OilGasoline OilHFO OilParaffin OilLPG

SATIM 76.30 1.74 114.66 0.86 0.24 17.77 1.33 2.22

DOEEB 76.30 1.74 103.80 0.86 0.24 17.77 1.01 0.00

The electricity discrepancy arises from the redistribution of the ‘General/Unspecified’ share(35.2PJor5%)ofsalesin2006asreportedbytheNationalRegulator(NERSA,2006)betweenthe Industry, Commercial and Residential sectors. The Paraffin and LPG differences are areallocationofsomeoftheResidentialSectorshareofthesefuelsinthenationalenergybalanceof99.96%and84%respectively.TheenergybalanceallocatesnoLPGtotheCommercialSectorbutit’susedasacookingfuelinhospitality,cateringandrestaurantbusinesses.

31


Themost recent published data found on Commercial Sector energy consumption by energyserviceinSouthAfricawasagreenhousegasmitigationstudyforthesectoralsoreferencedbythe LTMS (deVilliers, 2000). Calculatingweighted values of end‐use energy consumption foreachbuildingactivityandfuelgaveanestimateofenergyconsumedbyeachenduseforeachfuel. Since then thepercentage shareof floor areaof eachbuildingactivityhasbeenupdated(Stats SA, 2010a). This required reducing the number of building activities considered in thepreviousLTMSfromeighttothreemajorgroups(Table20).TheUSAcommercialbuildingsenergyconsumptionsurveytableE6A(CBECS,2003)providedelectricity energy use intensities that related to the Stats SA economic sectors of 'office andbanking space' and 'shopping space'. For 'other non‐residential space' the average electricityenergy intensity for all types of buildings located in the climate zone most similar to SouthAfrica fromCBECS table E6Awasused. Theweighted electricity energy‐intensity by end usefromthesethreecategorieswasthencalculated.Theshareofnon‐electricityenergycarriersconsumedbycommercialenduseswasbasedoninformation from the annual survey of registered industrial and commercial fuel burningappliancesundertakenbytheCityofCapeTownairqualitydepartmentin2007andsomaynotberepresentativeofthenationalbuildingstock(CCT,2007).Theendusesharesshownbelowwereestimatedbyreviewingthecommentsinthedatabaserelatingtoenergyconsumptionandconsidering theactivitiesbeingundertakenwithin thebuildings.Thecontributionofdifferentenergycarriers todeliveringspaceheatingandhotwaterwereobtained fromtheLTMSas itwasnotpossibletodeterminethesplitfromtheavailabledata.Table22:EstimatedEnergyCarrierShareofEnergyConsumptionbyEndUseandFuel

Lighting Space heating Water heating Cooling &

ventilation Refrigeration Cooking Other

Electricity 40% 5.82% 2.18% 30% 7% 14%

LPG 100%

Paraffin 100%

Coal 54% 46%

Residual oil 0% 100%

Town gas 0% 100%

Thefigurebelowcomparestheenergyconsumptionbyenduseforthebaseyearof2001fromthepreviousLTMSstudywiththeestimatedenergyconsumptionbyenduseforthebaseyearof2006.

32

Figure14:Comparisonofenergyconsumptionbyendusefor2001withupdatedbaseyearof2006

FutureDemandforEnergyfromtheCommercialSector

Theenergydemandinthecommercialsectorisbasedonthefloorspaceforagivencommercialactivity.The increase inenergydemand ismodelledon increasing floorspacearearelative tothe base year. Floor space projections are generated using an elasticity derived from aregression of floor area against GDP between 1993 and 2006. The projection of CommercialSectorGDPinSATIMisbasedonthesameCGEmodeloutputasisusedfortheIndustrySectorasdescribedabove.The futuredemandforenergyservices in thecommercialsector iscomputed in the followingsteps:

1. The growth of the tertiary sector (Communications, Financial Services, BusinessServicessub‐sectorsetc.aggregated)isprojectedbymeansofaneconomicCGEmodel.

2. Thetotal floorspace isprojectedbyusinganelasticityof0.64,derivedfromobservedhistorical data, to relate floor area to the tertiary sector growths using the followingequation:

FAi =FAi-1*(1+elasticity*(GDPIi/GDPIi-1-1)) Equation 4 Where: FAi = Floor Area in year i GDPIi = Indexed GDP of tertiary Commercial Sector in year i elasticity = %change in floor area / % change in indexed GDP as indicated by historical data

3. The floor space is split into two classes, and an estimate of the share of each class is

estimatedoverthestudyhorizon:

33

Oldbuildingcode(2006:95%droppingto2030:50%) Newbuildingcode(2006:5%risingto2030:50%)

4. Forthefloorspaceusing“oldbuildingcodes”,theusefulenergyintensityforeachend‐

use is set at the calibrated useful energy intensities (correcting for the 5% newbuildings).

5. For the floor spaceusing “newbuilding codes”, theuseful energy intensity is set20%lower than the “old building code” from 2010 onwards. Energy intensity for newbuilding codes is at this time assumed to remain at this level for the model horizon(typicallyto2050).

6. Theusefulenergyisthenprojectedbymultiplyingtheintensitybythefloorareaforthetwobuildingcodeclassesandthensummingthemupforeachyearintheprojection.

Theelasticityof0.64iscurrentlyunderreviewduetothesmallnumberofdatapointsusedtodetermineit.The20%efficiencygainofnewbuildingsisalsounderreviewandremainsafairlyspeculativeassumptionatthistime.Errorisboththeseassumptionscanresultinfutureenergydemandbeingsignificantlyunderoroverprojected.

34

TransportSector

TheenergyconsumptionofthetransportsectorinSouthAfricaislarge,totalingaround28%oftotalfinalconsumption(TFC)inthenationalenergybalances.Thebulkofthisenergydemand(97%)isintheformofliquidfuels,andthistransportsectorshareofliquidfuelsis84%ofthenationalliquidfueldemand.(DoE,2009)(IEA,2011).Theevolutionoftransportdemand,bothin terms of itsmagnitude and form (carrier) is very uncertain andhas large implications forinfrastructurerequirements.Thishasledtothetransportsector,alongwiththeelectricitysupplysector,beingmodelledinrelativedetail compared toother sectors.Funding for researchon this sectorhasalsodrivendevelopmentof thispartof theSATIMmodeland itsdetaileddocumentationso thestructureandassumptionsforthetransportsectordescribedbelowundertakeahigherlevelofdetailandrigourthantheothersectors.

ModelStructure

The Transport Sector input data of the SATIM model disaggregates road vehicles by basicvehicletypeandfueltypebutdoesnotatthisstagedisaggregatebytechnicalspecificationslikeengine size or emissions regulation compliance level.Whiledisaggregationbyengine size forinstance can be useful, this level of detail would create problems in a cost optimisedmodelbecausetypicallythechoiceoflargermoreexpensivecarsbytheconsumerisnotcostbased,oratbestonlypartiallyso,andboundsonpenetrationwouldhavetobecarefullyconstructedtoproduce realistic results. Rail is disaggregated by its broad energy services but inter‐citypassengerrailhasnotyetbeenincludedasthereisnodatainthepublicdomain.Atthisstagepipeline freight,passengerand freightaviationandnavigation fuelledbyheavy fueloil (HFO)are combined in one sub‐Sector called ‘Other’ and are represented by generic fuel basedtechnologieswithoutefficiencyandcostdetail.Table23:ExistingTransportSectorTechnologiesinSATIMBaseYear

sub‐Sector EnergySource Technology EnergyServicePassenger Diesel SUVPriv.Veh.OilDiesel

PassengerTransportbySUVPrivateVehicle

Passenger Gasoline SUVPriv.Veh.OilGasoline

Passenger DieselSUVPriv.Veh.OilDieselHybrid

Passenger GasolineSUVPriv.Veh.OilGasolineHybrid

Passenger Diesel CarPriv.Veh.OilDiesel

PassengerTransportbyCarPrivateVehicle

Passenger Gasoline CarPriv.Veh.OilGasoline

Passenger DieselCarPriv.Veh.OilDieselHybrid

Passenger GasolineCarPriv.Veh.OilGasolineHybrid

Passenger Electricity CarPriv.Veh.ElectricityPassenger Gas CarPriv.Veh.Gas

35

Passenger GasolineMotoPriv.Veh.OilGasoline

PassengerTransportbyMotorcyclePrivateVehicle

Passenger Diesel BusOilDieselPassengerTransportbyBusPassenger Gas BusGas

Passenger Diesel MinibusOilDiesel

PassengerTransportbyMinibusPassenger Gasoline MinibusOilGasolinePassenger Diesel MinibusOilDieselHybridPassenger Gas MinibusGasPassenger Diesel BRTOilDiesel(1)

PassengerTransportbyBRTPassenger Gas BRTGasPassenger Electricity BRTElectricity

Passenger Electricity MetroRailElectricity(2)PassengerTransportbyMetroRail

Passenger ElectricityHighSpeedMetroTrainElectricity

PassengerTransportbyHigh‐SpeedMetroTrain

Freight Diesel LCVOilDieselFreightTransport‐LCVFreight Gasoline LCVOilGasoline

Freight Gas LCVGasFreight Diesel MCVOilDiesel

FreightTransport‐MCVFreight Gasoline MCVOilGasolineFreight Gas MCVGasFreight Diesel HCVOilDiesel

FreightTransport‐HCVFreight Gas HCVGasFreight Diesel RailCorridorDiesel

FreightTransport‐RailCorridorFreight Electricity RailCorridorElectricityFreight Diesel RailOtherDiesel

FreightTransport‐RailOtherFreight Electricity RailOtherElectricity

Freight ElectricityRailExport(bulkmining)Electricity FreightTransport‐RailExport

(bulkmining)

Freight ElectricityRailExport(bulkmining)Diesel

Other Electricity PipelineElectricity TransportOther‐Pipeline

Other JetFuel AviationJetFuelTransportOther‐AviationJetFuel

Other AviationGasoline AviationGasolineTransportOther‐AviationGasoline

Other HFO(ResidualOil) HFO(3) TransportOther‐HFO(1):BRT:BusRapidTransport(2):Metro:Metropolitanieintra‐city(3):UsedforCoastal&InlandNavigationSUV:SportUtilityVehicle(usually4X4and>1toninmass)Priv.Veh.:PrivateVehicle

36

Externally funded projects since the LTMS have led to a great deal of refinement in roadtransportmodellinginSATIM.Muchhaveofthisefforthasgoneintothedevelopmentofsub‐modelsthatestimatethebasedataandexogenousdemandforinputintoSATIM.Thefunctionsofthesesub‐modelsandtheirplatformsaresummarisedbelowinFigure15.TheVehicleParcModel,Time‐budgetmodel,freightdemandmodelandroadtransportportionofthepassengerdemandmodelarefurtherdescribedintheirrespectivesectionsbelow.

Figure15:GeneralisedSystemofpre‐ProcessingTransportSectorInputsforSATIMusingsub‐Models

Sub‐ModelforProfilingBaseYearTechnologies–TheVehicleParcModel

Avehicleparcmodel,calibratedoversevenmodelyearsfrom2003to2009,wasdevelopedtoprovide a comprehensive picture of the baseline vehicle parc, the disaggregation of vehicleclassesandtechnologiesandtheactivitylevelofthoseclassesandtechnologies.TheAnalytica software platformwas selected for developing themodel as it lends itself to asystems approach, catering for feedback relationships between system elements and bothdeterministic and stochastic outputs from elements. A graphical interface allows the systemelementstobesetoutinacomprehensiblenetworkandthusaddingnewsystemelementstoanexistingmodeliseasierthanin,sayaspreadsheet,becauseinAnalyticaitonlyrequiresasimpleadditiontothemodelframework.Someoftheadvantagesofaspreadsheetareretainedinthatdatatables,forinstanceofbaseyearvehiclecategoryfueleconomy,canbequicklyenteredoraccessedfromthesystemdiagramasshowninFigure3.

Vehicle Parc Model ‐ Analytica

Time Budget Model ‐

CGE Model (External)

Freight Demand Model ‐ Excel

SATIM Model ‐ TIMES

Passenger Demand Model ‐

Excel

Base Year Public & Private pkm

Base Year tkm

Private / Public Split by Income Group

Projected Income Group

Share of Population

Base Year Stock, Mileage & Efficiency

Road ‐ Proj. veh‐km

Projected vehicle‐km

37

Figure16:Systemdiagramanddatatablefeaturesofanalyticalplatform

Fueldemandwascalculatedbymultiplyingthekilometerstravelled,thevehicletechnologyfuelefficiency and the number of vehicles in the vehicle technology segment as shown in theequationbelow.Thetechnologysegmentfueldemandsweresummedtoyieldthevehicleparcdemandandcomparedtohistoricalfuelsalesforcalibrationpurposes.

Equation5

Df,k = DemandforfuelfinyearkNi,j = Thenumberofvehiclesintechnologysegmentiwithmodelyearj(Y1beingthe

firstmodelyear),wheretechnologiesnumbered1toCallusefuelf.FCi,j = EstimatedfuelconsumptionfortechnologysegmentiwithmodelyearjVKTi,j = Vehiclekilometrestravelledpervehicleintechnologysegmentiwithmodelyear

jThe fuel demand calculation and model calibration process therefore required a number ofassumptionstopopulatethethreevariablesinEquation1,Nthenumberofvehicles,VKT,theirmileageandFCtheirfueleconomy:

1. Avintageprofilederivedfromrealisticscrappingcurvesthatenabledvehiclestocktobeestimatedfromhistoricalvehiclessalesdisaggregatedbyvehicletype.Thecurveswerecalibratedsothatthestockestimatecloselymatchedavehicleregistrationdatabase.

2. Anassessmentofannualvehiclemileageforeachvehicleclassandtherateatwhichthisdecaysasthevehicleages.

3. Estimatesofthefueleconomyofeachvehicleclassandhowthiswillchangeovertime.

38

The fuel demand was calibrated to match the known fuel sales data by first iterating tillapproximateagreementbymeansofscalingthekilometrestravelledpervehicleandthenfinetuningwithadjustmentstothefueleconomyassumptions.

A schematic representation of the vehicle parc model and its data inputs and validations isshowninFigure17withakeytothedatainputsinTable24.

Figure17:Schematicrepresentationofthevehicleparcmodelanditsdatainputsandvalidations

Table24:KeyDataInputstotheVehicleParcModel

NAAMSANationalAssociationofAutomobileManufacturersofSouthAfrica

Industryassociationthatcollatesdetailedvehiclesalesdata

eNatisNationalTrafficInformationSystem

Nationalvehicleregistrationdatabase

NatmapNationalTransportMasterPlan

Wide‐rangingDepartmentofTransportstudybetween2008and2010intendedforpolicydesign

SOL StateofLogistics

CollaborationbetweenCouncilforScientific&IndustrialResearch(CSIR)andtheUniversityofStellenboschthatproducesanannualanalysisoffreighttransportinSouthAfrica

SAPIASouthAfricanPetroleumIndustryAssociation

Industryassociationthatcollatespetroleumproductsretaildata

EB EnergyBalanceNationalEnergyBalancepublishedbytheDepartmentofEnergy

39

VehicleParcModelInputDataandAssumptions

DevelopingtransportsectormodelsandprojectingdemandintothefutureischallengingintheSouth African context because there is a paucity of data on vehicle utilisation and thereforeassumptionshadtobemadearoundthescrappingfactors,vehiclemileage,occupancyandfueleconomyinputs.ThevehicleparcmodeldevelopedforSATIMrequireddisaggregateddataonthe current vehiclepopulation, vehicle efficiencydata andutilisationdata forbothpassengerandfreighttransport.Data on the total registered vehicle population in South Africa is captured by the electronicnational administration traffic information system (eNaTiS). The eNaTiS vehicle registrationdata includes seven vehicle classes, namely: Motorcars; minibus; buses and midi‐buses;Motorcycles,lightdutyvehicles,panelvansandotherlightvehicles(lessthan3500kg);truckslargerthan3500kg;andotherself‐propelledvehicles.TheeNaTiSdataismoreaggregatedthanthe historic vehicle sales data available from the National Association of AutomobileManufacturers of South Africa (NAAMSA). NAAMSA publishes 14 vehicle classes crossreferencedtotechnicaldatathatcouldbeusedtodisaggregatetheeNaTiSdataintoadditionalsubcategories but excludes motorcycles. The NAAMSA data records only vehicle sales andthereforedoesnotdirectlytranslateintovehiclesontheroad.DeterminingthecountofvehiclesinthevehicleclassesshowninTable25thereforerequiredthe14NAAMSAvehicleclassestobemapped to the 7 eNaTiS classes so that stock could be determined at the higher level ofdisaggregationbutcalibratedtoregistrationdatabase.

Table25:VehicleclassesadoptedfortheVehicleParcModel

Vehicletypes Fueltype ModelID*

Passengercar Diesel CarDieselPassengercar Gasoline CARHybridGasolinePassengercar Gasoline CarGasolineBus Diesel BusDieselHeavycommercialvehicle Diesel HCVDieselMediumcommercialvehicle Diesel MCVDieselMediumcommercialvehicle Gasoline MCVGasolineLightcommercialvehicle Diesel LCVDieselLightcommercialvehicle Gasoline LCVGasolineMinibustaxi Diesel MBTDieselMinibustaxi Gasoline MBTGasolineSportutilityvehicle Diesel SUVDieselSportutilityvehicle Gasoline SUVHybridGasolineSportutilityvehicle Gasoline SUVGasolineMotorcycle Gasoline MotoGasoline*TheseIDsareusedingraphlegendsbelow

Estimatesoffreightutilisationinton.kmhavebeenavailableinthepublicdomainthroughtheannually published State of Logistics reports (Havenga, Simpson, & van Eeden, 2011) since2004. Estimates of the demand for passenger transport in passenger.km are not readily

40

availablebutcouldpotentiallybeinferredbyanalysis,forexamplefromthetripdatageneratedbytheNationalTransportMasterPlanmodel(DoT,2009).In order to check the model calibration, regional data on fuel sales was required. Whileaggregate fuel consumption by the transport sector is available through the national energybalances published by the DoE, there were challenges in apportioning fuel consumption topassenger and freight transport as fuel use in transport is not disaggregated in the energybalances.TheSouthAfricanPetroleumIndustryAssociation(SAPIA)recordsdisaggregatedfuelsalesdatabyprovinceunderseveralcategories,oneofwhichis‘Freight’howeverthiscontainsonlydieselsalestodepots.Longhaultrucksfrequentlyobtainfuelfromretailoutlets(classifiedas ‘retail’ by SAPIA) and therefore the recorded use of diesel for the ‘freight’ categorydesignatedbySAPIAaccountsforlessthanhalfoftheactualfreightconsumptionofdiesel.ThefueldemandcalculationandmodelcalibrationprocessrequiredanumberofassumptionstopopulatethethreevariablesinEquation1,Nthenumberofvehicles,VKT,theirmileageandFCtheirfueleconomy.Theassumptionsrequiredare:1. Avintageprofilederivedfromrealisticscrappingcurves;2. Anassessmentof annual vehiclemileage for eachvehicle class and the rate atwhich this

decaysasthevehicleages;and3. Estimatesofthefueleconomyofeachvehicleclassandhowthisischangingwithtime.

VintageprofileToprojecttheenergyconsumptionofavehicleparcandhowitmayevolveovertime,avintageprofile of the current vehicle parcwas established. This is important, as newer vehiclesmayhavebetterfueleconomyandhighervehiclemileagethanoldervehiclesand,asnewervehiclesentertheparcandolderonesaredrivenlessandarescrapped,theaveragefueleconomyoftheparcchanges.Therateatwhichvehicleshavebeenscrappedwasdefined in themodelbyscrappingcurveswhichestimatetheprobabilityofavehiclesurvivingasafunctionof itsage.Thisallowsustoconvert historical sales data into stock data. The Weibull cumulative distribution function,shownbelow,wasusedforthispurpose.If: x = ageofthevehicle f(x) = theprobabilityofthevehicleremainingoperational α = aconstant β = aconstant

Equation6

Multiplying the total sales of a vehicle type in a particular year (vintage) by the appropriatescrapping factor on the curvewill yield the probable population in a future base year. Thus

α

βx

exf

41

historical sales data can be converted to an approximation of stock in the vehicle parc for agivenyearbysubstitutingtheresultofEquation6,theprobabilityofavehiclebeingscrapped,inEquation7.If: YS = Theyearofsale YP = Theyearforwhichthevehicleparkisbeingcharacterised VP = ThestockofvehiclesinthevehicleparcinyearYPsoldinyearYS VS = ThenumberofvehiclessoldinyearYS(YP–YS) = Thefunctionestimatingtheprobabilityofthevehiclebeingscrapped

VP = (YP–YS)VS Equation7

Thescrappingcurveswerecalibratedbyiteratingtheparametersforthescrappingcurvesuntila targetpopulationwasreached.Thiswasdoneuntil theconvertedhistoricaldetailedvehiclesales data from NAAMSA matched the more aggregated total vehicle population data fromeNaTiS for a calibration year while maintaining an average vehicle age for the model thataccordedwith published data andwas continuouswith other calibration years. TheWeibullconstantsusedforthevehicleparcmodelandtheresultingaverageageofvehiclecategoriesinthe model for the 2010 calibration year are presented below including data from previousstudiesforcomparison.

42

Table26:VehicleClassWeibullCoefficients&ResultingAverageAgesfortheCalibratedVehicleParcModelComparedtootherStudies&Sources

Source CalibratedVehicleParcModel

SAnationaloctanestudy

Belletal(2003)

Moodley&Allopi(2008)

Stone&Bennett(2001)

Year 2010 2002 2005 2000Vehiclecategory β Avg.Age β Avg.Age Avg.Age Avg.Age

Dieselcar 22 3.0 5.0 20.2 3.2 4.2 10Gasolinecar 23 2.0 11.8 20.2 3.2 10.4Hybridgasolinecar 22 3.0 2.2 DieselSUV 22 3.0 5.2 HybridgasolineSUV

22 3.0 0.7

GasolineSUV 22 3.0 6.9 DieselLCV 22 3.0 7.8 20.2 3.2 7.2

9.3GasolineLCV 22 1.4 12.4 20.2 3.2 9.9DieselMCV 24 3.0 8.5

12 11.9GasolineMCV 24 3.0 19.1DieselHCV 24 3.0 9.6DieselMBT* 23 3.0 3.5

13.0GasolineMBT* 23 3.0 13.0 20.0 3.2 11.3Dieselbus 30 3.0 15.4 11Motorcycle 16 3.0 5.5 *MBT:MinibusTaxi

There is a wide range of average ages between vehicle classes but the younger classes, forexampledieselcars,dieselminibus‐taxisandhybridsreflect recentsales thatarea lothigherthanhistorical sales.Theestablishedvehicle classes suchasgasoline cars, LCVsandHCVsallhaveaverageagesofaround10ormoreyears.ThescrappingcurveforeachvehicleclassinthemodelplottedusingtheWeibullcoefficientsisshowninFigure18below.

43

Figure 18: Base year scrapping curves for the vehicle technology types in the

vehicleparcmodel

VehiclemileageTheprocessofdevelopingmileageassumptionsforthemodel,thatwouldbebothplausibleandallowforthecalibrationofmodelfueldemandwithfuelsales,requiresanassumptionaroundthe initialannualmileageof ‘newvehicle’annualmileage.Theassumed ‘newvehicle’mileagewas based on national and international literature. The annualmileage of vehicles has beenobservedto,onaverage,decaysteadilyfromthisinitialvalueforeachyearofoperation.TheUSEPA’sMOBILE6modelassumesaconstantrateofdecaycompoundingannuallythatisspecifictovehicletype(Jackson,2001)asshowninFigure19.Ingeneral(busesbeingtheexception),the rate of decay assigned is higher for vehicles with a higher initial mileage, heavy truckmileageforexampledecaysat10.9%perannumwhileforlight‐dutyvehiclesthedefaultrateinMobile6is4.9%annualdecayinannualmileageperannum.AlthoughthelatterratehasbeenobservedtobebothhigherandlowerforspecificareaswithintheUnitedStates(Yu,Qiao,Li,&Oey,2002),goodagreementwasshownwithaparkedcarstudycoveringanumberofsitesinNairobi (University of California at Riverside, Global Sustainable Systems Research, 2002)whereinitialmileagewaslowerbuttherateofdecayverysimilar.

44

Figure19:EPAMobile6annualmileagedecayassumptionscomparedtoresultsofvehicleactivitystudyfor

NairobiKenya

Lackingeven rudimentarymileageaccumulationdata forSouthAfrica, thevalueof4.9%wasusedas therateofmileagedecay for theSouthAfricanvehicleparcmodelacrossallvehiclesclasses.Therateofdecaycombinedwiththeassumptionofaninitial‘newvehicle’mileageandthe age profile of themodel parc resulting from the scrapping assumptions discussed aboveresultsinanestimateofaverageannualmileageforeachvehicleclass.Clearly,ifrecentvehiclesales have been low then this will reduce the average mileage of that class because oldervehicleswhichcoverlessmileagecontributedisproportionately.Aftermodelcalibration,theseassumptions resulted in average mileages for the model vehicle classes that are reasonablyconsistentwithpreviousstudiesandlocalandforeigndataasshownbelowinTable27.

0

20000

40000

60000

80000

100000

120000

140000

160000

0 5 10 15 20 25 30

Annual M

ileage (km

)

Vehicle Age

EPA Light‐duty Vehicle(Diesel & Gasoline)

EPA Light GasolineTruck

EPA Light Diesel Truck

EPA Heavy DieselTruck (15 ‐ 27t)

EPA Diesel Bus

Nairobi Passenger Cars

45

Table27:Assumedaveragevehiclemileage(km/annum)

Region SouthAfrica NorthAmerica

OECD–Europe&Pacific

non‐OECD

Source SATIM–newvehiclemileage

SATIM –averagemileageofstock

SAPIAPDSA1

RTMC2 LTMS3 NationalOctaneStudyModel–45km/h4

NationalOctaneStudyModel–34km/h4

Stone&Bennett5

Stone‐CoastalKZN6

IEA/SMPModel(2010)7

Year 2006 2006 2008 2007 2003 2002 2002 1998 2002 2010 2010 2010

Dieselcar 24000 21254 19000 14644 15000 23467 19778 18873 17600 11250 10875

Gasolinecar 24000 16169 19000 14575 17647 14872 14016 17600 11250 10875

Hybridgasolinecar 24000 23678

DieselSUV 24000 20314

HybridgasolineSUV 24000 24000

GasolineSUV 24000 19128

DieselLCV 25000 19202 19500 18806 15000 25196 21143 20577

GasolineLCV 25000 16662 19500 14575 21046 17660 16552

DieselMCV 45000 33417 42901 39933 34211 32000 25000 21125

GasolineMCV 25000 13575 38229 32000 25000 21125

DieselHCV 70500 48403 52583 72354 60000 60000 50000

DieselMBT 50000 43474 27480 70000 35000 35000 40000

GasolineMBT 50000 30927 30000 70000 92365 92365 70332 35000 35000 40000

Dieselbus 40000 22072 35227 28912 61985 60000 60000 40000

Motorcycle 10000 8340 6124 5000 7500 75001:(NAAMSA/SAPIAWorkingGroup,2009)2:(RoadTrafficManagementCorporation,2009)3:(DEAT,2007)4:(Bell,Stone,&Harmse,2003)–ThismodelusedthespeeddependentCOPERTequationstocalculatefueleconomysothecalibrationwithfuelsalesrequiredadjustmentofannualmileageifaveragespeedwaschanged.5:(Stone&Bennett,2001)6:(Stone,2004)7:(IEA,2011)

46

Thelargerangeofminibus‐taxiannualmileageestimatesisofinterestbecausethisvehicleclasshas a large effect on the model calibration due to its high modal share. The discrepancy inmileagebetweenthevariousstudiesreflects tosomedegreetherespectiveauthor’sstruggleswith calibrating their models in the absence of good activity data for these vehicles. Recentpublished data for African cities (International Association of Public Transport & AfricanAssociation of Public Transport, 2010) presented in Table 28 suggestsminibus‐taximileagesarehigh.Table28:AverageannualmileagepervehicleforpassengermodesinvariousAfricancities

CityPassengercar(km/annum)

Dieselbus(km/annum)

Minibus‐taxi(km/annum)

Abidjan 12 000 60 049 86400Accra 19 200 29 952 79872AddisAbaba 25 357 53 924 57350Dakar 7 500 45 582 58006DaresSalaam 25 000 70000Douala 15 000 40 000 50000Johannesburg 21900 27260 64680Lagos 4 260 73 920 72000Nairobi 8 133 15 000 18000Windhoek 15 863 15 000

Thevalueofvehicleparcmodelswouldbegreatlyenhancedifreliabledatafortheminibustaxiindustrywasavailable,arelativelylow‐costexercisegiventhattherearerelativelyfewvehiclesofthistypewhichoperatewithinacommercialstructure,albeitsometimessemi‐formal.Gooddataforminibustaxiswouldimprovethecalibratedmodeloutputsforothervehicleclasses.

FueleconomyFueleconomiesforeachvehicleclassandeachmodelyearweregeneratedbyassuminga1%annualimprovementinfueleconomyofthevehicleclassesinthevehiclefleetrelativetotheir2010fueleconomyaccordingtotheaggregatemanufacturer’sdataavailableforrepresentativecarmodelsineachvehicleclass.Averagevehiclefueleconomyisafactorofseveralvariables,asvehicles age the efficiencydecreases, but the fuel economyofnewvehicles tends to improveovertimeasshowninTable29below.Thisistheresultnotonlyoftechnologybecomingmoreefficient but alsobecause regulation is reducing vehiclemass and engine capacity. The SouthAfricanvehicleparcisdominatedbymodelsfromEuropeandJapan,sogiventhedatashowninTable29ahigherannualimprovementmightbeexpectedbutgiventheslowerrateofscrappingin SouthAfrica and the lowervalueof0.4% for the shortperiod reviewedby the IEA, itwasdecided that 1% is a reasonable historical improvement in the absence of local reliable data.This assumption is also supportedbyaBritish study (Kwon,2006),which suggests thatnewpassenger vehicles and light commercial vehicles had an improved vehicle efficiency of 0.9%perannumbetween1979and2000inBritain,amuchlongerperiodthantheperiodcovedbythestudiesreviewedbelow.

47

Table 29: Improvements in passenger car fuel economy in world markets2000–2010

(ICCT,2011)

(Cuenot&Fulton,2011)(InternationalEnergyAgency)

Country Period Annualfueleconomy

improvement

Period Annualfueleconomyimprovement

US 2000–2010 1.60% 2005–2008 1.90%Canada 2000–2008 1.28% ‐ ‐EU 2000–2010 1.90% 2005–2008 1.90%Japan 2000–2009 2.81% 2005–2008 2.20%SouthAfrica ‐ ‐ 2005–2008 0.40%

Dataforthefueleconomyimprovementofheavy‐dutyvehiclesoverthecalibrationperiodwasnotfoundandthereforeanassumptionof1%wasappliedtothesevehicleclassesaswell.TheresultinghistoricalfueleconomytrajectoryforthevehiclesclassesinthemodelispresentedinFigure20.Giventheblanket1%assumption,thefueleconomyofallvehicleclassesincreasesbyjustover22%over the20yearperiodshown. Clearly, incertain instances the fuel economydata for some technologies are extrapolated back to before those technologies entered themarket,gasolinehybridSUVsforinstance,butthisdoesnotaffectthemodelifnostockofthesevehiclesexists.

Figure20:Assumedhistoricalevolutionofvehiclefueleconomyinthemodel

Thecalibrationprocessinvolvedfirstadjustingtheinitialannualmileageassumedtothefinalvaluesshownabove inTable27and thenadjusting the2010 fueleconomyestimatesslightlyuntil a good match was obtained between the data for historical fuel sales to the transportsector and the fuel demand of the vehicle parc model. The adjusted new vehicle 2010 fueleconomy assumptions and the resulting fuel economyof stock in that year for the calibratedmodelarecomparedtootherlocalandinternationalstudiesandmodelsinTable30.

48

Table30: Calibratedmodelfueleconomy(l/100km)byvehicleclasscomparedtootherstudiesandsources–Part1

Region SouthAfrica NorthAmerica

OECD –Europe&Pacific

non‐OECD

Source Thismodel–new

vehiclefueleconomy

ThisModel –averagefueleconomyof

stock

Vander‐schuren1

SAPIAPDSA2

LTMS3 NationalOctaneStudyModel–45km/h4

NationalOctaneStudyModel–34km/h4

Stone&Bennett

4

Stone‐CoastalKZN5

IEA/SMPModel(2010)6

Year 2006 2006 2010 2008 2003 2002 2002 1998 2002 2010 2010 2010

Dieselcar 7.5 7.7 8.2 6.3 7.7 5.9 6.7 6.8 9.5 7.4 9.1

Gasolinecar 8.3 9.1 10.5 8.4 9.3 7.5 8.6 10.8 11.6 8.9 11.1

Hybridgasolinecar 6.4 6.4

DieselSUV 11.5 12.0

HybridGasolineSUV 7.3 7.3

GasolineSUV 13.0 13.7

DieselLCV 11.5 12.2 10.5 11.2 7.7 9.0 8.7 10.6

GasolineLCV 13.0 14.2 13.8 14.7 10.8 13.3 12.5

DieselMCV 28.1 30.0 17.4 17.2 25.6 23.7 28.0

GasolineMCV 33.3 38.7 31.4

DieselHCV 37.5 40.7 31.6 47.5 41.9 36.1 33.1

DieselMBT 11.4 11.8 10.5 11.2

GasolineMBT 13.5 15.1 11.4 14.4 12.7 15.4 15.4 16.0 18.0 18.0 16.0

Dieselbus 31.2 35.5 36.0 36.1 27.9 33.0 33.0 28.0

Motorcycle 5.2 5.4 4.5 3.5 2.31:(Vanderschuren,2011)2:(NAAMSA/SAPIAWorkingGroup,2009)3:(DEAT,2007)4:(Bell,Stone,&Harmse,2003)–ThismodelusedthespeeddependentCOPERTequationstocalculatefueleconomysothecalibrationwithfuelsalesrequiredadjustmentofannualmileageifaveragespeedwaschanged.5:(Stone&Bennett,2001)6:(Stone,2004)7:(IEA,2011)

49

OccupancyandloadfactorLessdatawasavailabletoguidetheassumptionsforvehicleoccupancyandloadfactorwhicharecritical forcalculatingthedemandforpassenger.kmandton.kminthemodel.AstatisticalreviewoftransportinAfricancitiespublishedbytheInternationalAssociationofPublicTransport(2010)offersaregionalperspectiveandsuggests,giventhefiguresforJohannesburgcomparedtoothercitiespresentedbelow,thatoccupancyinSouthAfricaisgenerallylowerthantherestofAfrica.

Table31:AverageoccupancypervehicleforpassengermodesinvariousAfricancities

City

Passengercar(pass/veh)

Dieselbus(pass/veh)

Minibus‐taxi(pass/veh)

Abidjan 2.0 60 18Accra 2.0 68 18AddisAbaba 3.7 80 11Dakar 2.0 66 35DarEsSalaam 1.9 45 29Douala 2.3 45 17Johannesburg 1.4 37.1 8.5Lagos 1.8 43 18Nairobi 1.7 70 18Windhoek 1.3

Thefinaloccupancyandloadfactorsselectedforthemodelarecomparedtootherstudiesandsourcesbelow.Table32:Modeloccupancyandloadfactorbyvehicleclasscomparedtootherstudiesandsources

Region

SouthAfrica NorthAmerica

OECD–Europe&Pacific

non‐OECD

Source Thismodel

Vander‐schuren1

LTMS2 IEA/SMPModel3

Year Units 2006 2010 2003 2010 2010 2010

Dieselcar pass/veh 1.4 1.40 2.10 1.47* 1.61* 1.77*

Gasolinecar pass/veh 1.4 1.40 2.10 1.47* 1.61* 1.77*

Hybridgasolinecar pass/veh 1.4

DieselSUV pass/veh 1.4

HybridgasolineSUV pass/veh 1.4

GasolineSUV pass/veh 1.4

DieselMBT pass/veh 14 12 35 6.00 8.40 10.70

GasolineMBT pass/veh 14 12 15 6.00 8.40 10.70

Dieselbus pass/veh 25 40 35 12.00 16.70 22.00

Motorcycle pass/veh 1.1 1.20 1.20 1.40

DieselLCV ton/veh 0.5 2.10

GasolineLCV ton/veh 0.5 2.10

DieselMCV ton/veh 2.5 2.20 1.60 1.70

GasolineMCV ton/veh 2.5 2.20 1.60 1.70

DieselHCV ton/veh 15.00 10.00 8.00 6.30*Data for LDVs which include cars and light trucks/vans/suvs 1: (Vanderschuren, 2011) 2: (DEAT, 2007) 3:(IEA,2011)

50

CharacterisingNewVehicleTechnologies

Inordertomeetfuturedemandforfuturetransportsectordemandforpassengerandfreighttransportenergyservices,theSATIMmodelhasthefollowing‘new’technologiesavailableinadditiontothosepresentedaboveinTable23.Table33:AdditionalTechnologiesAvailabletotheModelforMeetingFutureDemandforTransport

Technology Mode FuelPassengerCarElectricity PrivatePassenger ElectricityPassengerCarGas PrivatePassenger NaturalGasPassengerCarDieselHybrid PrivatePassenger DieselPassengerCarGasolineHybrid PrivatePassenger GasolineSUVDieselHybrid PrivatePassenger DieselSUVGasolineHybrid PrivatePassenger GasolineHighSpeedRailElectricity1 PublicPassenger ElectricityMinibusGas PublicPassenger NaturalGasMinibusDieselHybrid PublicPassenger DieselLargeBusGas PublicPassenger NaturalGasBRTElectricity2 PublicPassenger ElectricityBRTGas2 PublicPassenger NaturalGasBRTDiesel2 PublicPassenger DieselHCVGas Freight NaturalGasMCVGas Freight NaturalGasLCVGas Freight NaturalGas1:AhighspeedraillinkbetweenthecitiesofJohannesburgandPretoria,the‘Gautrain’,openedinAugust20112:BusRapidTransitSystems(BRT)beganoperatinginanumberoflocationsaroundthetimeofthe2010SoccerWorldCup

Futureimprovementsinvehiclefueleconomy

Averagevehiclefueleconomyisafactorofseveralvariables,asvehiclesagetheefficiencydecreases,butthefueleconomyofnewvehicleshasinrecentyearstendedtoimproveovertime,asshowninthetablebelow.Table34:Improvementsinpassengercarfueleconomyinworldmarkets2000–2010

ICCT(2011) Cuenot&Fulton(2011)(InternationalEnergyAgency)

Country Period Annualfueleconomy

improvement

Period Annualfueleconomyimprovement

US 2000–2010 1.60% 2005–2008 1.90%Canada 2000–2008 1.28% – ‐EU 2000–2010 1.90% 2005–2008 1.90%Japan 2000–2009 2.81% 2005–2008 2.20%SouthAfrica ‐ ‐ 2005–2008 0.40%

Thisistheresultnotonlyoftechnologybecomingmoreefficientbutalsobecauseregulationisreducingvehiclemass and engine capacity. The2008European Impro‐Carproject (Nemry, Leduc,Mongelli,&Uihlein, 2008)

51

indicatedthatfueleconomyimprovementsof7%(theyquoteCO2emissionreductions)couldbeattainedbya12%vehiclemassreductionandimprovementsof18%fora30%vehiclemassreduction.Aggressivereductionofenginecapacityby30%butmaintainingpowerbyturbochargingwasexpectedtoreducefueleconomyby7%fordieselcarsand12%forgasolinepassengercars.Thesetwomeasurescombinedcouldthereforeaccountfor a 1% annualized improvement for about 20 years for gasoline cars. Givingway on engine capacity andpowertogether,concedingperformancewillofferevengreaterpotentialbenefitsifthisweretobedrivenbylegislation.AJointTransportResearchCentrestudy(JTRC,2008)thatmodelledCO2emissionsfromtheworldtransportfleetusingtheIEA’sMoMomodeladopteda29%fueleconomyimprovementbetween2005and2050asthemostlikelyscenariowhichequatestoa0.75%annualizedimprovement.Theirmodelhoweverindicatedthatfor world vehicle fleet CO2 emissions to stabalise a 56% improvement between 2005 and 2050 would benecessary,whichisequivalenttoa1.8%annualizedimprovement.Forthepurposesofageneralperspective,aseriesofhypotheticalfueleconomyimprovementsbetween2012and 2050 are tabulated below, showing net and annualised quantities and the end‐point in 2050 given arepresentativefleetfueleconomyforgasolinepassengercarsin2012of8.6l/100km.Thesearecomparedtothecombinedcyclefueleconomyfortwoofthecurrentmostefficientsmallcarmodels,scaledupby10%toaccountforthereporteddifferenceoftheEuropeantestcyclewithrealworldfueleconomy(Pelkmans&Debal,2006)(Kwon,2006).Table35:Hypotheticalfueleconomyimprovementscenariosforgasolinepassengercarscomparedtothecurrentefficient

non‐hybridgasolinepassengercars

Annualisedfueleconomy

improvement

Totalimprovement2012–2050

2012(l/100km)

2050(l/100km)

‐0.5% ‐17% 8.6 7.0‐1.0% ‐32% 8.6 5.8‐1.5% ‐44% 8.6 4.7‐2.0% ‐54% 8.6 3.8‐2.5% ‐62% 8.6 3.1

Fiat500(1.2l)+10%a 5.6 ?KiaRio(1.2l)+10%b 5.9 ?a:(Fiat,2012)b:(Kia,2012)The SouthAfrican vehicle parc is dominatedbymodels fromEurope and Japan, so given the data for thosemarketsshowninTable29abovewemightexpectahigherannualimprovementthanthelowvalueof0.4%fortheshortperiodreviewedbytheIEA.ThismayreflectaslowerrateofscrappingandapreferenceforlargervehiclesinSouthAfrica.Itwasdecidedthat1%wasareasonablehistoricalimprovementforthecalibrationofthevehicleparcmodelintheabsenceoflocalreliabledata.ThisassumptionisalsosupportedbyaBritishstudy(Kwon, 2006),which suggests that new passenger vehicles and light commercial vehicles had an improvedvehicleefficiencyof0.9%perannumbetween1979and2000inBritain,amuchlongerperiodthantheperiodcovedbythestudiesreviewedabove.Itwas similarlydecided thata future sustainedannualised improvementof1% in fuel economy forexistingtechnology typeswasa reasonable assumption for the reference case.As shownbyTable35 this is feasiblewithcurrenttechnologyforgasolinenon‐hybridsgivenacompleteshiftinconsumerpreferencetosmallcars.

52

A 2% annualised improvement seems far more challenging but plausible and so this was deemed anappropriaterateforahighefficiencyimprovementscenario.

InvestmentcostsforNewTransportTechnologiesandConstraintsonMarketPenetration

The investmentcostassumptions fornewtransport technologiesareallderived fromtheLTMSprojectwiththeexceptionofthecostsforelectricpassengercarswhichhavebeenupdatedtoreflectdevelopmentofthistechnology.There has not been a heavy investment in time in refining cost estimates because this sector of the SATIMmodel is not solved as a pure least cost optimisation with technologies being fairly tightly constrained bybounds on penetration. In practice then this part of themodel is used as a scenariomodelwith alternativepenetration rates being tested against one another. The rail freight and ‘other’ technologies have abstractnumbers,usually1,forcostsandefficienciesandhavethereforenotbeenincluded.Thesplitbetweenroadandrailisexogenouslypre‐processedinthedemandcalculationdiscussedbelow.ThusthetotalenergydemandforthesectorissplitbyenergyserviceexogenouslyandthisshareisinputtoTIMES.InSATIMcurrentlyallfreightmodes output a different energy service for instanceHeavy Commercial Road Freight,Medium CommercialRoadFreight,RailcorridorFreightandsoon.Thereforetheroadandrailfreightmodeswillsupplyenergytomeetthepre‐processedenergyservicesharewithouttheinfluenceoftheoptimiser.Whiletheoptimisationcapabilityofthemodelisnotreallybeingusedthesystemwideimpactsarestillseen.Thisapproachisausefulcompromiseintransportwheremodeshiftandtechnologyselectionarenotoriouslyirrationalinpractice.

Sub‐ModelforProjectingPassengerCarMotorisationandtheDemandforPassengerTravel–TheTimeBudgetModel

The relationship between motorisation (vehicles per 1000 population), particularly passenger carmotorisation,andGDP/capitaiswelldocumented.IngeneralmotorisationincreasesmoreorlesslinearlywithGDP/capita until saturating and is thus usuallymodelledwith the s‐shaped Gompertz curve, an example ofwhich is shown below fitted to the historicalmotorization and per‐capita income data for some developedeconomies(Dargay,Gately,&Sommer,2007).

Figure21:GompertzCurvemodelofthecorrelationofmotorisationwithincomepercapitaDargayetal.(2007)

53

Thus far, it seems that increasing incomepercapitahas inexorably led toagrowth inprivateasopposed topublicmode travel. This has only been tempered in extremely dense cities likeHongKongwhere a privatevehicleoffersnoparticulartimesavingorconvenienceadvantage.Understandingthelikelyevolutionofprivatevehicle motorisation with changing per capita income and the impact of travel time is therefore critical toprojectingthedemandforpassengertravel.SouthAfricancitiesarenotnotablydenseandhighspeedpublictransport isonlybeginningtoemergesoitseemslikelythat fortheforeseeable future,peopleascendingtheladderofpercapitaincomearelikelytoassumetheestablishedappetiteforprivatetraveloftheirnewincomegroup.InSATIM,thisprinciple,alongwithassumptionsaroundaveragetrafficspeedsandtheprevailingtraveltimebudgethavebeenleveragedinasub‐modeltoprojectthedemandforpassengertravelwhichisdescribedinmoredetailbelow.

Background‐Thedailytraveltimebudget

Time‐useandtravelsurveysfromnumerouscitiesandcountriesthroughouttheworldsuggestthatthetraveltimebudget isonaverageapproximately1.1hperpersonperdayacross thespectrumofpercapita income(Schafer&Victor,2000).

Figure 22: Average per‐capita travel budget for various localities and regions across the GDP spectrum

Schafer&Victor(2000)

South African cities are not notably dense, and the poor tend to live in satellite ‘townships’ far fromemployment nodes,which suggests that the travel time budget in SouthAfricamay be different. Victor andSchaferhoweverarguethattheshareoftravellerstototalpopulationtendstobelowerinlow‐incomegroupsandthereforethetimebudgetwhenconvertedtoanaverageperpersoninthepopulationissimilartohigh‐incomegroups.ThishasbeeneloquentlyexpressedbySchaferasfollows(Schafer,2006):

“Although theamountof time spent traveling ishighly variableonan individual level, largegroupsofpeoplespendabout5percentoftheirdailytimetraveling…..Onaverage,residentsinAfricanvillages,thePalestinianTerritories,andthesuburbsofLimaspendbetween60and90minutesperdaytraveling,thesameasforpeoplelivingintheautomobiledependentsocietiesofJapan,WesternEurope,andtheUnitedStates.”

54

SchaferandVictorraisethecaveatthatthestabilityofaveragetraveltimebudgetholdsonlyfortravelbyallmodesandthattimespentinmotorisedmodesriseswithincomeandmobilityaspeopleshiftfromslownon‐motorised modes to motorised travel. As this shift tends to completion, however, total motorised travelapproaches1.1hours.Thusananalysisacrossincomegroupsmustconsiderthatatthelowerincomeendthetimebudgetwillincludesomenon‐motorizedtransport,walkingforinstancewhichwouldbeless,attheupperincomeendofthescalebutforalargepopulationtheaveragetimebudgetforallincomegroupswillbearound1.1hours.

UsingtheTimeBudgetModeltoProjectDemandforPassengerTravelonLand

Theprojectionofenergydemandforlandtransportrequiredanexogenousinputoffuturepassengertravel,inpassenger.km,intothetimebudgetmodel.Themethodusedtocalculatethedemandforpassengertravelcanbesummedupasfollows:

Passengerdemandforroadandrailwasmodelledforthreeincomegroupsrepresentinglow‐,medium‐andhigh‐incomehouseholds.Thiswasdonebecausegrowthofprivatetransportoverlowspeedpublicmodesisstronglyrelatedtohouseholdincome.Thelow‐incomegroupincludeshouseholdswithincomeofuptoR19 000 per annum (in 2007 rands), the middle‐income group includes households with an incomebetweenR19000andR76800andthehigh‐incomegrouptheremainder.

Motorisation (carownership)per capita foreachof the incomegroupswasestimated for thebaseyearusingsurveydata(DoT,2005).

Assumptions around ratios of public and private transport, average speed and travel time budgetweremadeforeachoftheincomegroupsandusedtocalculatetheirnetdemandforpassengertravel.Duetothesparsenessofactivitydatathemodelwascalibratedtomatchthevehicleparcmodelforonlytwomodes,privateandpublic,forthebaseyearof2006.

Mobilitynotmetbyprivate transportwasassumed tobemetbypublic transport (forexampleminibus,bus,metrorail,BRTorrapidtrain)anddistributedbetweenmodesaccordingtoanticipatedinvestmentininfrastructuresupportingeachofthemodes.

Populationprojectionsundereach incomegroupwereusedtoproject thetravel timebudgetofeach incomegroupinthefuture.Thecalculationoftraveldemandwasmadeasfollowswhereiisoneof3incomegroupsandjisoneoftwomodes,publicorprivate:

∑ ∑ Equation8

PKM = Totalpassenger.kmperyearFij = Fractionofannualtimebudgetspenttravellinginmotiononmodejbyincomegroupi.This

excludestimespentwalkingandwaitingtomakeuseofmodej.Sj = AveragespeedofmodejTi = TraveltimebudgetperpersonperyearforincomegroupiNij = Numberofpeopleinincomegroupiusingmodej

ThevariableFijisthefractionoftimespentactuallytravellingandmustthereforebeadjustedtoexcludetimespentwalkingandwaitingtomakeuseofthemodeasfollows:

1 _ _ Equation9

Fij = Fraction of annual time budget spent travelling inmotion onmode j by income group i. Thisexcludestimespentwalkingandwaitingtomakeuseofmodej.

55

Fw_ij = FractionoftimespentwalkingandwaitingwhenusingmodejFT_ij = TotalFractionofannualtimebudgetspenttravellinginonmodejbyincomegroupi.ThenumberofpeoplewithaccesstoaprivatevehiclesineachincomegroupwhichisthevariableNij, fortheprivatetravelmode,wasestimatedbasedontheNationalHouseholdTravelSurveyundertakenin2003(DoT,2005).Thesurvey foundthat75%ofhouseholds in thehigh‐incomegrouphadaccess toacarbut thatonly26%ofhouseholdsacrossall incomegroupshadaccess toacar.Thecriticalassumptions inTable36beloware:

Carownership/access–the%ofpeopleinanincomegroupwithaccesstoacar. Carsperpersonwithaccesstoacar–thenumberofcarsperpersonhavingaccesstoacar.

Ifweassumetheseremainconstantover theprojectionperiodwecanestimate thenumberofcars foreachfutureyearasthepopulationineachincomegroupchangesovertime.

Table36:Assumptionsandchecksusedtoestimateaccesstoprivatevehiclesbyincomegroups–baseyear(2006)

Travellergroup

Modelvariable Unit Low‐income

Middle‐income

High‐income

Comp./gov/carrental

Total/avg

Calib.check

Annualincome(percentile) (%) 50% 30% 20%

Population million 23.44 14.06 9.37 46.88

Carownership/access %ofpers 7% 23% 78%a 26% 26%b

Carsperpersonwithaccesstocar cars/pers 0.25 0.30 0.40

Calculatedtotalcars mveh 0.41 0.96 2.94 0.68 4.987 4.987c

M/cycleownership %ofpers 0.1% 0.4% 1.4%

Totalm/cycles mveh 0.03 0.06 0.13 0.05 0.27 0.27c

Personswithaccesstoaprivatevehicle

7% 23% 79%

Privatevehicles/personwithaccesstovehicle

veh/pers 0.25 0.31 0.41

Personswithaccesstoaprivatevehicle

millionpers. 1.67 3.24 7.44

Personswithoutaccesstoaprivatevehicle

millionpers. 21.77 10.82 1.94

a:TheNHTS(DoT,2005)had5incomecategoriesratherthanthe3ofthisstudy.The2highesttogetheraccountedforjustoverthe20thpercentileofthepopulationofhouseholdsandhadanaverageaccessratetoacarof75%whichhasbeenincreasedslightlytotakeaccountofthelapseof3yearssincethestudy.b:NHTS(DoT,2005)c:FleetsizeoftheVehicleParcModel.

ThereisasparsityofdatawhichshowsrepresentativeaveragespeedsoftrafficonSouthAfricanroads.Clearlylocal circumstances as regards congestion and road type are enormously variable, and establiishing averagespeedsonall roadswouldbean immenseundertaking. In a studywhichdevelopeda speedbasedemissioninventorymodelfortheCityofJohannesburgGoyns(2008)sampledthespeedofthirtyvehiclesforatotalof716hourscovering29587kmin2006/2007.Theresultsindicatedanaveragetripdurationof17minutes,anaveragetripdistanceof11.5kmandanaveragespeedof41km/h(Goyns,2008).Thetimebudgetmodelwas,however, as onewould expect, very sensitive to average speed and the speedsmeasured by Goyns did notallowforatimebudgetashighas1.1hours.AcompromisewasmadebyassumingtheaveragespeedSprivateof

56

privatetransporttobe34km/h,theaveragespeedoftheNEDCemissionstestcycle(Dieselnet,2000)usedinEuropewhichismeanttoberepresentativeofurbandrivingincludingaportionofhighwaydriving.

Table37:Totaldistance,timeandspeedforthenewEuropeandrivecycle

Phase Distance Time AverageSpeed(km/h)(km) [sec]

1(ECE15) 4.052 780 18.7

2(EUDC) 6.955 400 62.6

NEDC:ECE15+EUDC 11.007 1180 33.6

Thespeedofpublic transportSpublicwasassumedtobesignificantly lowerat20km/h.Thus tocalibrate thetimebudgetmodel,Fij the fractionof theannual timebudget spentonmode jby incomegroup iwasvarieduntilthepassengertraveldemandandtotalvehiclekmforeachmodematchedthatofthevehicleparcmodel.Table38presentstheestimatesofthevariablesforEquation4andEquation5.AgoodcalibrationwasattainedforreasonablevaluesofFgivenourassumptionsfortimebudgetTandaveragespeedS. Itmaywellbethataveragespeeds inSouthAfricaarehigherandthat for the timebeingthetimebudgetofprivate travelers inparticularismoderatelylessthan1.1hoursperdaybutitisproposedthatacalibratedtimebudgetmodelwithat least plausible estimates for the variableswill give a better estimate of future passenger.km than simpleextrapolationofdemandintothefuture.

Table38:Timebudgetmodelassumptionsforcalculatingpassengertraveldemandforpublicandprivatemodes

Variable Travellergroup Total Calib.check

Low‐income

Middle‐income

High‐income

Comp./gov/carrental

1.Allmodes

Ti(hours/day/person)b 1.1 1.1 1.1

Annualtraveldaysc 300 300 300

Ti(hours/year/person) 330.00 330.00 330.00

2.Public – noaccesstocar

Fw_ijd 42% 42% 42%

FT_ij 100% 100% 100%

Fij 58% 58% 58%

Sj(km/h) 20 20 20

Ni(106persons)e 21.77 10.82 1.94

PKMij(109p.km) 83.34 41.42 7.41 132.17

3.Public – accesstocar

Fw_ijd 42% 42% 42%

FT_ij 75% 48% 11%

Fij 44% 28% 6%

Sj(km/h) 20 20 20

Ni(106persons)e 1.67 3.24 7.44

PKMij(109p.km) 4.79 5.96 3.13 13.9

TotalpublicPKM(109p.km) 88.13 47.38 10.54 146.1 147.0a

57

Variable Travellergroup Total Calib.check

Low‐income

Middle‐income

High‐income

Comp./gov/carrental

4.Private

Fw_ij 0.00 0.00 0.00

FT_ijd 25% 52% 89%

Fij 25% 52% 89%

Sj(km/h) 34 34 34

Ni(106persons)e 1.67 3.24 7.44

PKMij(109p.km) 4.6830 18.92 74.26 18.87f 116.7 118.6a

TotalPKM(109p.km) 92.81 66.30 84.80 18.87 262.8 265.5a

a:p.kmoutputoftheVehicleParcModelfor2006.b:See(Schafer&Victor,TheFutureMobilityoftheWorldPopulation,2000)c:Author’sassumptiond:Author’sassumptionadjustedsothattimebudgetmodelcalibratestovehicleparcmodelfor2006e:SeeTable36abovefortheestimateofthesenumbersf:Estimatedfromthenumberofthesevehiclesinthevehicleparcmodelandfixedassumptionsof18,000kmannualmileageandoccupancyof1.4persons/vehicle

To summarise: the timebudgetmodel allowedus to generatepassenger traveldemand for each yearof theprojectionfromexogenousvaluesoffuturepopulationandincome.Essentiallythepremiseofthemodelisthatprivatevehiclepassengerdemandwillincreasenotonlywithpopulationgrowthbutalsoastheproportionofthepopulationinthemiddleandhigherincomegroupsincreases.Therateofincreasewillsaturateifthehighincomegroupbecomesdominant.The timebudgetmodel also affords the flexibility to investigate scenariossuchas:

ascenarioofreducedaveragespeedsduetocongestion; ascenarioofhigherpublictransportspeedduetoimplementationofbrtandhighspeedrail; timebudgetsspecifictoincomegroups;and ascenarioofpublicmodewalkingandwaitingtime.

Whilethissub‐modelcouldarguablybemoredisaggregateditisappropriatetolocaldataavailabilityandisagreat improvement on the assumption of simple linear growth of passenger travel demandwith populationgrowththatunderpinsmanyprojectionsoftransportenergydemand.The mobility of the future population between the income groups defined above is clearly critical to theprojection of demand for passenger travel. As shown in Figure 15 above this input was derived from ananalysis of the results of a Computable General Equilibrium (CGE)modelmade available by a collaboratingresearch group. This CGE model for South Africa was developed to study the economic implications ofintroducingcarbontaxesinSouthAfricaforthepurposeofgreenhousegasemissionsmitigation(Alton,etal.,2012).OurstudyusedthisCGEmodeltoestimatetheprobablefutureevolutionofhouseholdincomeinSouthAfrica,forthe3incomegroupsusedbythetraveltimebudgetmodel,givenamoderateandstableGDPgrowthof3.9%between2010and2030.The samemethodwasused toproject future applianceownership for theResidentialsectorandisdiscussedinmoredetailwithregardtothatinSection0below.

58

ProjectingDemandforFreightonLand

ThesectorGDPprojectionsoftheCGEmodelalsoformedthebasisforthefreightmodel.ClearlyasGDPgrowsthequantityofgoodsthatmustbetransportedgrowsproportionallyandwecanmodelthissimplisticallyasfollows:

TKMi = ei X (1 + GRGDP)i X TKMi-1 Equation10

Where:TKM = demandforfreighttransportinunitsofton.kmei = theelasticityoffreightdemandwithrespecttotransportGDP = (1+%changeinfreightdemand)/(1+%changeinGDP)GRGDP = transportGDPgrowthrateThesectorsrelevanttofreightdemand(transport,miningandironandsteel)wereprojectedtogrowasshowninFigure23bytheCGEmodelbasecaseuntil2030,theCGEprojectionsareextrapolatedfrom2030to2050.

Figure 23: CGE model output for annual household consumption 2005‐2030 by deciles of households

Author’scalculationsusingdatafromAltonetal.(2012)

Theirdifference ingrowth is small in theprojectionwindow,onlydivergingafter2030.Therefore transportsectorGDPgrowthwastakenasGRGDPforthelandfreightmodes,roadandrail.Datawasnot available for theelasticityofdemand for freight ei and thiswas taken as0.8.This results in athreefold increase inthedemandfor freighttransportbetween2005and2050.This isanareawherefutureresearch can contribute to better estimates of growth in freight demand. A particular concern is that atransitiontoa lessenergyintensiveserviceeconomycouldtranslatetoa long‐runelasticitysignificantly lessthan0.8.Thefreightdemandforrail inthebaseyearof2006wasassumedtobethatpublishedintheCSIR’sStateofLogisticsReport (Ittmann,King,&Havenga,2009)disaggregated intocorridor, rural,urbanandbulkminingfreight.Thiswascombinedwiththeroadfreightdemandestimatedbythevehicleparcmodelandtheroad/railsplits for corridor and for rural/urban freight estimated as shownbelow.Thebase caseassumed that these

59

remainthesametill2050andanenergyefficientscenariothatincludedashiftbackfromroadtorailcouldbemodeled.Table39:Organisationofroadandrailfreightdemandsuchthattheroad/railsplitcanbekeptconstantforthebasecase

Mode Freight2006(t.km)

Road/railsplit2006(%)

Assumedbasecaseroad/

railsplit2050(%)

Transportfreightbyroad–LCV 14Roadvsrail–rural/urbanfreight#

Transportfreightbyroad–MCV 9 26% 26%Transportfreightbyrail–other 26 74% 74%

Roadvsrail–corridorfreight#TransportfreightbyroadHCV* 120 81% 81%Transportfreightbyrailcorridor 28 19% 19%Transport freight – rail export (bulkmining)

67

Total 265*InpracticeaportionofthisNAAMSAcategoryarerigidtrucksactiveinurban/ruralfreight.#ThetotalofroadandrailfreightisgrownwithexogenousGDPestimateandthendisaggregatedbytheroad/railsplit

CalculationofFutureEnergyDemandfromRoadVehicles

First a projection for vehicle‐km is calculated for each demand using an occupancy (passenger/veh forpassengervehicles)orloadfactor(tons/vehforfreight):

Equation11

, Equation12

Where:VKMti = vehicle‐km projection for demand i in year t, PKMti = passenger‐km projection for passenger demand i in year t, TKMti = ton‐km projection for freight demand i in year t, Oti = vehicle occupancy for passenger demand i in year t, Lti = loading for freight demand i in year t.

Then,eachyear,theshortfallinvehicle‐kmcapacityforeachdemandi(e.g.privatecars)iscalculatedbytakingthedifferencebetweenthecapacityofthevehicleparcfordemand i inthatyearandthevehicle‐kmdemandprojection for that year, and used to calculate the total vehicle sales for vehicles that year. Since vehiclevintagesaretracked,the‘sales’calculationisforthevintageforthatyear.

Equation13

Where:Sti = Salesinyeartforvehiclesvintagetthatcanmeetdemandi,Ptiv = Populationofvehiclesvintagevinyeartthatcanmeetdemandi(thischanges eachyearasvehiclesfromthevintagearescrapped),

60

AFtiv = Averagekm/yearthatvehiclesvintagevisexpectedtodrive(thisdecreasesasthevehiclesgetolder).

Thensharesof thesales for technologiescompeting foraparticulardemandare imposed(differentones fordifferentonesfordifferentscenarios)exogenously.Assumingthatalltechnologiesforaparticularvintagehavethesamecapacity(drivethesamenumberofkmperyear)andthattheircapacitywillbefullyutilizedallowsustocalculatethefuelusedbyeachtechnologyasfollows:

Equation14Where:Ftijv = Fuelusedinyearttomeetdemandibytechnologyj,vintagev,Etijv = Fueleconomy(l/100km)fortechnologyj,vintagev,inyearttomeetdemandi.

61

ResidentialSector

Themodelling process in theResidential sector is facedwith both energy consumptiondata challenges andstructural challenges due to the great diversity in energy consumption patterns of households, the basicbuildingblocksofthesector(Senatla,2012).Householdsvarywidelyintheirenergyprofileandthesectorismade up of many of these small diverse users. The challenge is therefore to trade‐off model detail,computational effort and data availability. Typically national surveys are the primary source of data onhouseholdsandingeneralthesecharacterisehouseholdsbytheirincome,dwellingtype,demographicprofileandotherhouseholdscharacteristics.IncomeappearstobethebestdeterminantofboththetypeandquantitiyofenergyusedbyhouseholdsinSouthAfricaandhouseholdincomethereforeformsthefirstlevelaggregationofendusers.ThecompromisethatwasreachedwithSATIMwastogrouphouseholdincomegroupsthathavesimilarenergyconsumptioncharacteristics.Thepurposeofhouseholdgroupingsistobeabletobuildamodelwhichcanincorporateandreacttothefollowing:

Energyuseprofilesandconsumptionlevelsacrossincomegroups Themovementofhouseholdsbetweenincomegroups Policyinterventionswhichtargetcertainhouseholds(i.e.anincreaseinresidentialelectricitytariffforhigh

energyconsumers)

Mitigationactionsrelevanttoaspecifichouseholdgroup(i.e.asolarhotwaterheatingprogrammeonlowincomehouseholds)

ModelStructure

Householdclassificationprocess

Determining thehouseholdgroupings involvesbalancingmodel simplicityagainstmodelversatilityallowingpolicy options and future scenarios to be captured (Senatla, 2012). Each iteration of SATIM provides apossibility forsectoral improvementalthoughproblemswithdatahave incertaincasesrequireddevolution.The LTMS’s disaggregation of households by income, electrification status and rural/urban divide has, forinstance, generally been considered themost comprehensive grouping as itwas informed by a stakeholderengagementprocess.Thecurrentmodelhowevergroupshouseholdsbyincomeandelectrificationstatusonly

sinceStatisticsSouthAfrica1nolongerprovideshouseholdsdatabyurbanorruralclassificationandthereisnointentionofsodoinginthefutureasthisclassificationisarguedtobeunhelpfulfordevelopmentpurposes.Therefore,basedondatafromtheStatisticsSouthAfrica’s2007CommunitySurvey,NationalIncomeDynamicsStudy (NIDS) and All Media and Products Survey (AMPS), households were classified into the 5 groupingsshownbelowinTable40.Table40:HouseholdsgroupingincurrentSouthAfricanTIMEsModel(SATIM)

Householdgrouping Numberofhouseholds Percentageshare(%)

LowIncomeelectrified 4,089,009 33%Lowincomenonelectrified 1,654,030 13%Middleincomeelectrified 3,553,678 28%Middleincomenonelectrified 747,667 6%Highincome 2,367,588 20%Total 12,500,610 100%

1OrganisationinSouthAfricathatisresponsibleforlargescalehouseholdssurveys

62

Low‐incomehouseholdshaveahouseholdincomeofunderR19000in2007,middle‐incomehouseholdshadan incomebetweenR19000andR76800 in2007andhigh‐incomehouseholdshad incomeaboveR76800.Thechoiceofthreeincomebandsandtheincomesplitwasinfluencedtoalargeextentbyelectricalapplianceownership. Electricity is the dominant fuel in the residential sector and consumption of electricity byhouseholds in the incomebands is driven by appliance ownership (Beute, 2010), (Dekenah, 2010) (Gertler,Shelef,Wolfram,&Fuchs,2012).High incomehouseholds tend toownmoreappliancesandasa result theyconsumemoreelectricity.Usingapplianceownershiptodisaggregatehouseholdsbyincomeseemsreasonableas appliance ownership is itself strongly related to household disposable income. Electricity usage data andelectricalappliancedatacamefromdifferingsourceswithdifferingincomebandclassifications;thereforetheapproximateincomebandsfromtheapplianceownershipanalysiswereused.The energy intensive appliances shown in Table 41 were used for the analysis of categorising households.These appliances were used because they contribute significantly to the electrical load profile while lessintensiveappliancessuchastelevisionsetsandradiosdonotcontributesignificantly.Theanalysisstartedwithanalysingsaturationlevelsinthresholdsof10%‐90%indifferentincomebands.Forexampleifanalysisisat10th percentile, households in any particular income bandwhich showed 10% ownership of the appliancesshowninTable41,that incomebandwillhaveatallyof1,else0.Figure24showstheresultsof10%‐90%saturation thresholds. The 70th percentilewas found to be similar to the indexed average ownership trendshowninFigure25,anditwaschosenastheappropriatepercentileforcategorisinghouseholds.

Figure24:Appliancesaturationlevels

By using the 70th percentile, it was found that low income households earned an income that is less thanR19188(whichisveryclosetoR19200forlowincomehouseholdsofthisstudyfromStatisticsSouthAfrica),middle income households earned an income betweenR19 200 andR71 988 (which is close to themiddleincomehouseholdscut‐offpointofR76000fromStatisticsSouthAfrica)andhighincomehouseholdsearned

63

aboveR720002.Figure25showstheindexedaverageownershipofallthresholdsofappliancesshowninTable41.

Figure25:Indexedhouseholdsapplianceownership

Source:DerivedfromAllMediaProductsSurvey,2007andStatisticsSouthAfrica2007

Table41:EnergyIntensiveAppliances

ElectricStoveOtherGasOrCoalStove

MicrowaveOvenRefrigerator

FreeStandingDeepFreezerVacuumCleanerDishwasher

AutoFrontLoadingWasherAutoTopLoadingWasherSemi‐AutoTwinTub

TumbleDryer Source:chosenfromAMPS(2007)Having decided on the households’ groupings, it is essential to delve into end use fuel consumption andtechnologiesusedforthebaseyearcalibration.Theresidentialsectorismodelledwithdemands(enduses)for:

Lighting, Cooking, Spaceheating, Waterheating, Refrigeration “other”endusesSomeoftheseendusesaremetwithdifferentfuels.Forbaseyearcalibration,themodelneeds:

Energyintensitiesofenergyservices Technologiesandassociatedfuelsusedfortheenduses

2AllRandsarein2007Rands

Low Income Middle Income High Income

64

Technologysharesinlow,middleandhighincomehouseholds Thedifferenthouseholdgroupings(low,middleandhighincomehouseholds) Technologyinvestmentandoperatingcosts TechnologyefficienciesandlifetimesThenationalincomedynamicstudy(NIDS),AMPSdatasetsandothersmallstudiedwerecollatedtoestimatethetechnologyownershipsharesofallappliancesintheresidentialsector.Informationonthemainfuelusedforheating,lightingandcookingacrossdifferentincomegroupsiscollectedroutinelybyStatisticsSouthAfricainthecensus,communitysurveys,andthegeneralhouseholdsurvey.Thesesurveysdonothoweverprovideaquantitativeoverviewofenergyconsumptionpatternsandenergyend‐usecharacteristicsofdifferenthouseholds.Therefore,quantitativedataonenergyconsumption inSouthAfricanhouseholds is limited. This is partly due to historically poor energy data collection systems which haveworsenedoverthepastdecade,theinaccessibilityofhouseholdspecificelectricityconsumptiondata,aswellaschallengesrelatedtotherecordingandquantifyingofmultiplefuelusebyhouseholds.Thislackofquantitativeenergy consumption data required some consolidation of end use energy data from a range of studies,includingmicro‐levelstudies(communities,villagesetc)fromvarioussources.Thedatacollatedfrommicro‐levelstudiesonhouseholdenergyusewascrosscheckedwithenergyconsumptiondatafromenergybalancesfrom theDepartmentofEnergy,LoadResearchdatabyEnerweb, theLTMSstudy,ESKOM,NationalEnergy

RegulatorofSouthAfrica(NERSA)andtheSouthAfricanPetroleumIndustryAssociation (SAPIA).


Themainsourcesforresidentialdatarelatedtoenergyconsumptionandoutputare:

2007CommunitySurveydata, NationalIncomeDynamicsStudy(NIDS), AllMediaandProductsSurvey(AMPS) DepartmentofEnergy fromthepreviousLTMS, ESKOM,inparticulartheDomesticLoadResearchDatabase(NRS034).

NationalEnergyRegulatorSouthAfrica(NERSA), SouthAfricanPetroleumIndustryAssociation(SAPIA).

TheresidentialsectorbaseyeardatainSATIMhasagreaternumberofdiscrepancieswiththenationalenergybalancethantheothersectorsandestimatesofalltheenergycarriersexceptparaffindiffermarkedlyasshownbelow.Table42:ComparisonofEstimatesoftheAggregateConsumptionofFuels(PJ)bytheCommercialSectorfortheSATIMModel

andtheDOEEnergyBalanceforSouthAfrica‐2006

Fuel Electricity OilParaffin Coal BiomassWood

OilLPG

SATIM 158.9 22.6 26.8 125.8 1.9DOEEB 142.8 22.9 152.6 190.4 13.4Thesedifferencesareaccountedforasfollows:

65

As is the case with the Industry and Commerce sectors, a portion of the 35PJ or 5% attributed to the‘General/Unspecified’Categoryof sales in2006by theNationalRegulator’s statistics (NERSA,2006)hasbeenapportionedtotheResidentialSector.

As described in the review of the Industry Sector, a bottom‐up calculation of household coal use in theResidentialSectoryieldedamuchlowernumberthanthenationalenergybalanceandtheexcesscoalhasbeenattributedtotheIndustrySector.

Similarly, a bottom‐up calculation of biomass use in the Residential Sector, including extensive surveysundertakenbytheERCitself,yieldedamuchlowernumberthanthenationalenergybalance.AportionofthedifferenceisaccountedforbythePulpandPaperandFoodandBeveragesub‐SectorsoftheIndustrySectorasdescribedpreviously.

NodataforLPGusagehasbeenfoundinthepublicdomainforSouthAfricaandatthistimeonlynominalamountshavebeenattributed toLPGconsumptionbyall the sectors in SATIM.LPG technologies canbeselectedinthefuturebythemodelbutthebaseyearsituationhasyettoberesolved.

Assumptionsandconstraintsforreferencescenariointheresidentialsector

Scenario based energy modelling requires a development of differing scenarios which are underpinned byassumptions in the form of “What If?” analysis. In such a modelling exercise, there is always a referencescenario (also termed business‐as‐usual scenario) which will be changed or modified to construct otherscenarios.Ascenarioismadeupofassumptionsonfuturefueluse,fuelprices,andtechnologycostsindifferentsectors, technology efficiencies and changes on technologymarket shares. This section explains the criticalassumptionsandconstraints thatwereused toconstruct thereferencescenarioandhowthoseassumptionsweremadeforeachofthesectorsthatweremodelled.To consume energy, a wide range of technologies is used and it is vital to predict how the technologyavailabilitiesandusethereofwillevolvethroughoutthemodellingperiod.Indisaggregatedsectorssuchastheresidentialsectorandtransportsectors,mobilityofhouseholdsfromanyofthethreehouseholdcategories(asdefinedabove)hastobepredictedthroughoutthemodellingperiod.Intheresidentialsector,thekeyenergydemanddriverispopulationgrowthbutsinceenergyisconsumedinhouseholdsandthemodelisconfiguredaroundhouseholds,thekeyassumptionstobedefinedissplittingthegrowingpopulationpopulationgrowthintothethreehouseholdcategoriesoverthemodellingperiod.

ProjectingtheEvolutionofHouseholdIncome

Totranslatepopulationintohouseholds,populationconversionfractiontohouseholdfractionisrequired.Thisproposedconversionfractionshavetotakeintoaccountthehouseholdsizesofdifferenthouseholdcategories.UsingdemographicdatafromStatisticsSouthAfrica’scensuses(1996and2001)andCommunitySurvey2007,it was observed that high income households had a household size of 2.41 people per household, middleincome2.97peopleperhouseholdand low incomehouseholdanaveragehousehold sizeof4.34peopleperhousehold in 2007. For simplicity, household sizeswere assumed to be constant throughout themodellingperiod,but themovementofhouseholds fromone incomebandwasassumedtochangerelative toassumedchangesinGDPgrowth,labourmarketgrowth.There are two significant assumptions that underpin mobility of households, and these assumptions arecentredonthevariablesthatareusedtocategorisehouseholds.Thesevariablesaretherateofelectrificationinnon‐electrified households andmobility of households from one income band to another income band.For

66

household income mobility, the computable general equilibrium (CGE) model ESAGE (which is a detailed economic model) is used to estimate the probable future evolution of household income in South Africa, for the 3 income groups used. CGE assumed a stable and moderate GDP growth of 3.9% between 2010 and 2025, allowing the labour employment to grow at 2.6 per cent per year, which then leads to a gradual decline in national unemployment. The CGE model deals with actual expenditure within households and households are divided into deciles with the highest

income decile split into quintiles as shown in Table43.

Table43:MappingofStatsSAIncomeandExpenditureSurveyCategoriesandCGEModellingResultstoThreeIncome

Categories

SATIM

StatsSAIncome&ExpenditureSurveys CGEmodel

IncomeBandFractionofhouseholds Aggregation

Averagehouseholdconsumption

Fractionofhouseholds

Householdcategories(Share)

LowIncome

Noincome 9.10%

46%

R13662.00 10%

50%R1‐R4800 5.50% R23492.00 10%R4801‐R9600 9.90% R28384.00 10%R9601‐R19200 21.10% R32426.00 10%

MiddleIncome

R19201‐R38400 21.40%34%

R38645.00 10%R44464.00 10%

30%R38401‐R76800 13.00%

R53566.00 10%R74354.00 10%

HighIncome

R76801‐R153600 8.70%

20%

R141306.00 10%

20%

R153601‐R307200 6.10% R208067.00 2%R307201‐R614400 3.30% R244258.00 2%R614401‐R1228800 1.10% R292088.00 2%R1228801‐R2457600 0.40% R364406.00 2%R2457601ormore 0.30% R596546.00 2%

To match the actual expenditure results from the CGE model with the ANSWER-TIMES energy model, it was necessary to map the CGE results for 14 income groups to the three household categories required for residential modelling. Thus in the first year of the CGE model our 3 income groups contain 50%, 30% and 20% of all households but as the consumption of deciles rises in future years and they exceed the fixed threshold of the income group, more households will migrate from the lower to the middle group and from the middle income to the higher income group as shown in Figure 26.

67

Figure26:ProjectedHouseholdIncomeCategoryShare

Assumptionsonfuturetechnologyshare

AsSATIMisanoptimisationmodel,whichaimstomeetenergyservicesataleastcost,theanalysthastodefineconstraints and bounds on technology uptake in the future, because if no bounds are made, the cheapesttechnologywillalwaysbechosenbythemodel.Upperandlowerlimitboundsareascribedtotechnologiesinthe residential sector, these are defined for each technology individually, a generic example of the waytechnologysharesareassignedisshowninTable44.Theupperandlowerlimitcanbedefinedasthemaximumandminimumtechnologysharethataparticulartechnologycantake.

Table44:GenericExampleofAssumedtechnologylimitshares

Years 2010 2030 2050DefaultUpperLimitShare(%) 5% 25% 50%DefaultLowerLimitShare(%) 50% 25% 5%

FutureDemandforEnergyfromtheResidentialSector

Projections of future demand are done in terms of the demand for energy services or “Useful energy” (e.g.cooking,lighting,etc.)ratherthanintermsof“finalenergy”(e.g.paraffin,LPG,etc.)toallowforabetterstudyof the substitution between alternative fuels, as well as an appraisal of the effect that evolution of thetechnologicalimprovementshasonprojectionsoffuelrequirements.

In SATIM thedrivers of demand for energyby theResidential Sector are population andhousehold incomegrowth. The critical assumptions are the future population in each of our three income groups, the futureincomeof those groups and the elasticity of energy consumptionwith respect to income for a given energyservice. The Centre for Actuarial Research (CARe) at the University of Cape Town conduct demographic

68

modelling, and they produce amodel of population growth known as the ASSA (Actuarial Society of SouthAfrica)model.InSATIMthereferencescenariooftheASSAmodelisusedtoforecastthenationalpopulation.

Elasticitiesareassumedtodecreasewithincreasingincomewhichisintuitivebecauseatacertainincomelevela household will become saturated with energy intensive appliances. The following table illustrates thestructureoftheelasticityassumptions:

Table45:IllustrativeExampleofAssumptionsoftheElasticityofUsefulEnergyUsebyHouseholdwithrespecttoHouseholdIncomeforincreasinghouseholdincome

EnergyServiceHouseholdIncomeThreshold(000Rands)

15 40 100 400 >400Lighting 0.7 0.6 0.3 0.2 0.1Cooking 0.5 0.4 0.1 0.05 0SpaceHeating 0.7 0.6 0.4 0.2 0.1WaterHeating 0.7 0.6 0.3 0.1 0Refrigeration 0.8 0.7 0.4 0.1 0.05Other 0.7 0.6 0.4 0.2 0.1

NonEnergyUses* 0.7 0.6 0 0 0*:Mostlyuseofparaffintopolishcementfloors

Thus given thatwe can estimate an energy intensity (per capita energy usage) for our income groups andenergyservicesfromourbaseyeardata,wecanprojectfutureenergyintensitiesbyiteratingfromestimatedbaseyearenergyintensitiesasfollows:

EIijn = EnergyIntensityforincomegroupiusingenergyservicejinyearn(GJ/perperson/year)

ij = ElasticityofUsefulHouseholdEnergyConsumptionwithrespecttoHouseholdIncomeofIncome

Groupiforenergyservicej

Iin = Growthinincomeofincomegroupiinyearn(%)

EIijn = EIij(n‐1)X(1+ijXIin) Equation15

The demand for residential energy, as represented by this model, is therefore strongly dependent on twoassumptions,thefuturetransitionofpeopletoanincomegroupwithahigherenergyintensityandthegrowthof average real incomewithin an income groupwhichwill increase the energy intensity of the people thatremain in that group. Currently however in SATIM only the High income group is assumed to have anincreasingrealincomeandtheLowandMiddleIncomegroupsareassumedtohaveconstantincome.TheprojectedtransitionbetweenthreeincomegroupsLow,MiddleandHighusingtheoutputofaCGEmodelfor SouthAfrica is discussed in detail above in Section 0. For the purposes of estimating residential energydemand however whether a household is electrified has a large impact and therefore the Low andMiddleIncome groups are further split into electrified and non‐electrified. Thus an exogenous projection of theelectrification rate is required toderiveapopulation splitbetween the five incomegroups.This isbasedonhistoricaltrendsandfuturegovernment/Eskomtargetswhichyieldsthefollowingassumption:Table46:AssumedFutureSouthAfricanElectrificationRatesbyIncomeGroup

IncomeGroup 2006 2010 2020 2030 2040 2050LowIncome 71% 71% 80% 85% 90% 95%

69

MiddleIncome 83% 83% 90% 95% 95% 100%OverallElectrification 80% 81% 90% 95% 97% 100%These shares can be applied to the 3 group income split shown in Figure 26 to estimate a projected futurepopulation share for the 5 income/electrification groups as used inTable 40. The resulting population splitacross 5 income categories can be calculated by multiplying the this population share with the projectedpopulationandthiscanbecombinedwiththeotherassumptionsdescribedtocalculate futureusefulenergydemandasfollows:

Uijn = Usefulenergydemandforincomegroupiusingenergyservicejinyearn(GJ)

EIijn = EnergyIntensityforincomegroupiusingenergyservicejinyearn(GJ/perperson/year)

Pin = Populationofincomegroupiinyearn(persons)

Uijn = EIijnXPin Equation16

70

SupplyofEnergyfromtheElectricitySupplySector

ElectricityproductioninSouthAfricawasestimatedtoaccountfor50%oftotalgreenhousegasemissionsin2000,byfarthelargestsectoralsource(Mwakasonda,2009).Thisstemsfromarelianceoncoal‐firedpowerplants and has resulted in South Africa being one of the world’s most carbon intensive economies.Unsurprisingly,giventhisprofile,thesector’scriticaleconomicimportanceandhighcostsinpublicfunds,theelectricity supply systemhasbeenmodelled (withvarious limitationsand constraints)more thananyotherpart of the energy system in SouthAfrica, and data is relatively goodon the activity, stock and costs of thesystemanditsexistingtechnologiesaswellastechnologyoptionsforthefuture.

ModelStructure

ThepowersectorissplitintoGeneration,TransmissionandDistribution.InSATIM,theGenerationcomponentismodelled in themostdetail. Since thecountry ismodelledasa singlenode, transmission ismodelledasasingle technology linking centralised/high voltage electricity (ELCC) to medium voltage levels (ELC). Themediumvoltageelectricity isdistributed to each sectorusingadifferent technology to capture thedifferentlossesincurredinthedifferentsectors.Thisisdepictedinthefigurebelow.Notethatforpurposesofsimplicity,thesupplytechnologiesandsectordistributionlegshavebeenshowninaggregateform

Figure27:SimplifiedSchematicofSATIMModeloftheSouthAfricanPowerSector

Technologies such as pumped storage aremodeled as consuming power from the transmission system andfeedingoutputbackintothesystem;cogenerationoptionsaremodeledaftertransmissionandfeedpowerbackonto the grid at that point (industrial electricity demand), and also generate heat which contributes to theindustrialheatdemand.Aggregate lossesaremodeledfortransmissionanddistribution.Themodel isaone‐regionmodel and so there is no impact on results from shifting generation from current locations tomoredispersedlocations(forinstanceasaresultofdevelopingasignificantamountofrenewableenergycapacity).Thestructureandassumptionsunderlyingthetransmission,distributionandgenerationcomponentsofSATIMarebrieflydiscussedbelow.

Transmission

Inbrief,thetransmissionsystemismodelledasfollows:

71

TheTransmissiontechnologycapturesthenationalaveragetransmissionlosses,inconvertingELCCtoELC.Currentlythisisestimatedat3.8%

Anexistingstock/capacityforthetransmissiontechnologyisestimatedbasedonthepeakdemandinthebaseyearplusareservemargin.Forthebaseyearof2006,thisiscurrentlyestimatedas31.8GW.

Atthetimeofwritingthefollowingadditionsarebeingmadetothemodel:

o Theoperatingcostofissettoreflecttheestimatedrunningcostofthisinfrastructure.o The investment cost of new transmission capacity is set to reflect estimated average cost of

investment.o AreserveconstraintisimposedonELCtoensurethereserveismaintained.

Distribution

Thelegsofthedistributionsystemsupplyingeachsectoraremodelledastechnologiesasfollows:Table47:DistributionTechnologiesinSATIM

DistributionLeg TechnologyCodeinSATIMElectricitytoAgriculture AGRELCElectricitytoCommerce COMELCElectricitytoIndustry INDELCElectricitytoResidential RESELCElectricitytoTransport TRNELCElectricitytoUpstream(Refineries) UPSELCInbrief,thesedistributionlegsaremodelledasfollows:

The distribution technologies capture the average losses incurred by each sector. The capacity is setaccordingtothepeakdemandineachsectorplusthereservemargin

Areserveconstraintisimposedoneachdistributiontechnologytoensurethereserveismaintained. Atthetimeofwritingthefollowingadditionsarebeingmadetothemodel:

o Theoperatingcostof thedistribution technology isbeingset to reflect theaverage tariff seenby

eachsectoro Theinvestmentcostofnewdistributioncapacityissettoreflectaveragecostofinvestmento Currentlyallsectorsaremodelledtoexperienceelectricitysupplyatnationalaveragedistribution

losses(10%).Largeindustriesmayhoweverhavetheirownproximatesubstationsuppliedatthevery highest distribution voltage so losseswould be lower than is currentlymodelled in SATIM.Thisisintheprocessofbeingamended.

Generation

The structure and assumptions used tomodel existing and future generation technologies are dealtwith ingreaterdetailbelow.Therearetwomaingroupsofpowerplantsinthemodel:

Existing/oldPowerPlants(before2010) New/FuturePowerPlantTechnologies(from2010onwards)availabletothemodeloptimiser.

AsnotedabovetheSouthAfricanpowersectorhasbeenextensivelymodelledandthedataexiststomodelthesystem in some detail. Furthermore TIMES offers the potential for a high degree of parameterization,

72

particularlyoftechnologies.ThecurrentlevelofparameterizationinSATIM,usedtomodeltheexistingpowerplantsandnewtechnologiesavailabletothemodeloptimiserarediscussedbelow.

ExistingPowerPlants

Existing power plants are aggregated into 14 representative technologies as shown below in Table 48. Table48:SATIMAggregationofExistingPowerPlants(before2010)intoTechnologies

TechnologyGroup ExamplesInstalledCapacity2006(GW)

Coal‐firedEskomLarge(>500MW)Wet‐cooled

Arnot,Duvha,Kriel,Lethabo,MajubaWet,Matla,Tutuka 21.09

Coal‐firedEskomLargeDry‐cooled

Kendal,MajubaDry,Matimba 9.38

Coal‐firedEskomSmall Camden,Grootvlei,Komati,Hendrina 2.78Coal‐firedMunicipal KelvinA&B,PretoriaWest,Rooiwal 0.44Coal‐firedSasolSSF 0.52Coal‐firedSasolInfrachem 0.13Diesel‐fiedOCGT* Acacia,Atlantis/Ankerlig,MosselBay/Gourikwa,PortRex 2.40HydroSouthAfrica Gariep,Vanderkloof,andminihydros 0.67HydroRegional CahoraBassa 1.50PWRnuclear Koeberg 1.80PumpedStorageturbine Drakensberg,PalmietandSteenbrass 1.58Biomass/CoalCHPs‐PulpandPaperIndustry

Mbashe,SappiStanger,MondiMerebank,MondiFelixton,MondiUmlhlathuze 0.23

BiomassBagasse Sugarmills 0.10GasCHPs Mossgas 0.10TOTAL 42.71*OCGT: Open-cycle Gas Turbine

ParameterizationofExistingPowerPlants

Thetechnologiesinthedemand‐sidesectorsofSATIMarequiteuniforminanabstractsense,beingdefinedbya limitednumberofparameterssuchasefficiency,demandshareandcost.Thesupply‐side technologiesarenecessarilymorecomplexandagreatmanyparametersexist inTIMESthatcanbeusedtoprofile themandintroduce constraints into the model. Thus it makes sense to explore the methodology for this sector bydetailingtheparameterizationofthetechnologies.Existingpowerplanttechnologiesaremodelledinasimilarwayandallhavethefollowingbasicparameterization.CHPplantsandPumpedStorageplantshaveadditionalparametersinSATIMwhicharediscussedseparatelybelow.Thebasicparametersinputtothemodelarelistedbelow.Table49:SATIMParameterizationofPowerPlantTechnologies

ParameterEnergyInputCommodityorFuelWaterConsumption*EfficiencyOutputcommodityEnergyavailabilityCapacityavailability

73

CapacitycreditFixedoperatingandmaintenancecostVariableoperatingandmaintenancecostRefurbishment/retirementprofile“SEASON”&“DAYNITE”operatingcategories*Water is tracked as an emission rather than an input commodity The above parameters have the following attributes: TheEnergyInputCommodityorFueliscoal,dieselandotherfuels Waterconsumption[FLO_EMIS]istrackedinthemodelasanemissionratherthananinputcommodity.

This isbecause if the latterwere implemented,all technologieswouldhaveat least2 inputcommoditiesand themodelwouldapply theplant efficiency toboth, complicating implementation.By trackingwaterconsumptionasanemission,aconsumptionrateoflitres/MWhcanbeloadedlikeanemissionfactor,acostcanbeallocatedandconstraintsplacedonconsumptionwithoutthecomplicationsarisingfrommodellingwaterasaninputcommodity.

Efficiency,relatestheenergyinputcommoditiestothesumofenergyoutputcommodities. TheOutputCommodityis“centralizedelectricity”(ELCC). Energyavailabilityrelatesthetotalannualoutputtotheinstalledcapacity.Theavailabilityislessthanone

toaccountformaintenanceandunplanneddown‐time.Availabilityiscalculatedasfollows:

[NCAP_AFA] = (1‐POR)X(1‐FOR) Equation17

Where:[NCAP_AFA] = AvailabilityFactorPOR = PlannedOutageRateandFOR = ForcedOutageRate.[NCAP_AFA]

Capacityavailability:Sinceallexistingplantsaredispatchable,thisparametersimplyde‐ratesthepowerofaplantby(1‐FOR).Forexample,a100MWplantwitha5%FOR,willhaveacapacityavailabilityof95%,whichmeans thatatanygivenpoint in time, themodelledsystemcanonlyrelyonup to95MWfor thatplant.

CapacityCredit gets used in the ReserveMargin Constraint calculationwhich is similar to the ReserveMargin calculation. Reserve Margin (RM) is the capacity in percent above peak demand required tocompensate fortheoutageratesdiscussedaboveandisconventionallycalculatedforsystemsdominatedbybaseloadasfollows:

RM=(InstalledCapacity)/(Peakdemand)‐1. Equation18

WhereRM=ReserveMargin

The growth in intermittent generation capacity has resulted in a modified form of this equation, theReserveConstraintcomingintocommonuse.

∑ Equation19

WhereCCi=CapacityCreditoftechnologyi

74

Capacitycredit issetto1fordispatchableplantssuchthatforndispatchableplantstheequationsaboveare equivalent. For an intermittent technology likewindhowever capacity creditwill be less than1anddependonthelocalresource,spatialdistributionofsitesandinstalledcapacity.ThefiguresinSATIM

Fixed andVariableoperatingandmaintenancecostexcludesfuelcostsexceptinthecaseofKoebergnuclearplant.

A refurbishment/retirement profile specifies the capacity that is initially available and how thatincreasesinthecaseofrefurbishmentsanddecreasesasplantsorpartsofplantsretireinthefuture.TheassumedprofilesforEskomplantareshownbelowinFigure28.

Figure28:AssumedRefurbishment/retirementProfilesforExistingESKOMPlant

Base‐loadplants:Koebergnuclearandtheexistingcoalplants,areput inthe“SEASON”category,andalltheotherexistingplantsinthe“DAYNITE”category.Apowerplantinthe“DAYNITE”categorycanrespondtoloadchangeswithinthe“representativeday(s)”ofeachseason,whereasoneinthe“SEASON”categoryisconstrainedtomaintainaconstantlevelofoutputthroughoutthe“representativeday(s)”ofeachseason.

AdditionalParameterizationforCombinedHeat&Power(CHP)Plants

Combined Heat & Power Plants (CHP) have additional parameters as follows: Table50:AdditionalParameterizationinSATIMforCHPPowerPlants

ParameterIndustrialProcessHeatOperationinBackPressureAdditionalinputfuel

0

5

10

15

20

25

2006 2016 2026 2070

InstalledCapacity(GW)

Year

CoalEskomLargeExisting

CoalEskomLargeDryExisting

CoalEskomSmallExisting

75

In addition to Industrial SectorElectricity (INDELEC), CHP’s aremodelled asproducing a secondoutputcommodity, IndustrialProcessHeat [IPPS] pertaining to the sub‐sector (eg. Iron&Steel) inwhich theCHPisfound.

CHP’s are modelled as operating in back‐pressure, with a constant Electricity to Heat output ratio[NCAP_CHPR]. In back pressuremode you cannot bypass the heat load so the electricity to heat ratio isconstant. Alternatively in ‘Condensing’ mode, the electricity to heat ratio is determined endogenouslywhichrequiresanumberofadditionalparametersandisappropriatewhenCHPhasa largeshareof thesystemwhichisnotthecaseinSATIM.

SomeCHP’saregiventheoptiontouseanalternatefuel(e.g.biomassandcoal).Theshareofbiomass/coalisfixedbasedonestimatesofhistoricalconsumption.

AdditionalParameterizationforExistingPumpStorageSystems

Existingpumpstorage systemsaremodelledusing threedifferent technologies inTIMES, apump, a storagedamandaturbine,asshowninFigure29below.InadditionpumpedstoragesystemsinSATIMhavetheirownadditionalparameterizationasfollows:Table51:AdditionalParameterizationinSATIMforPumpedStoragePowerPlants

ParameterNightStorageTechnologyInputCommodity–DownstreamElectricityOutputCommodity–UpstreamElectricity*Technicallythepumpstoragedamtechnologyisamemberofthenightstoragetechnology(NST)setThe above parameters have the following attributes: InTIMESyouneedtodeclarestoragetechnologiesassuchandthestoragedamtechnologyisallocatedas

amemberoftheset[NST]whichsignifiesitasanightstoragetechnology.

Thecycleefficiencyofthesystemisattributedthepumptechnology.Thecapacityandcostparametersareattributedtotheturbine.

The input commodity is electricity downstream of transmission (ELC), and the output commodity iselectricityupstreamoftransmission(ELCC).

Figure29:SchematicofPumpedStoragesub‐ModelinSATIM

BasicStructure&AssumptionsforNew/FuturePowerPlants

Aswiththetransportsector,agreatmanytechnologiesarepossiblecandidatesforfuturesupplyofelectricityin the power sector. The level of detail selected here is a trade‐off between capturing as many of thesepossibilities and not wasting effort on verymarginal future prospects or on disaggregation that is still notusefulbecausethepowersectorinonlyslowlydiversifyingnowandtrendsareunclear.Newpowerproductiontechnologies neednot only to reflect themany emerging renewable and clean coal technologies but also allpossibleoptionsforimportfromneighbouringregions.SouthAfricahas,forinstance,arichsolarresourceand

Pump Storage (DAM)

Turbine ELCDummy ELCDummy

ELC ELCC

76

therefore there is reasonable disaggregation of these technologies. The following newelectricity productiontechnologiesarecurrentlymodelledinSATIMtobeavailabletothemodel’soptimserfrom2010onwards:Figure30:NewElectricityProductionTechnologiesAvailableinSATIM

Technology Efficiency(%)Availabil‐ityfactor(%)

Wateruse(litres/MWh)

UpperCapacity(GW)*

IndigenousCoal,Gas,Nuclear&Hydro

SupercriticalDry‐CooledCoal 37% 92% 229.10

FluidisedBedCombustionCoal 36% 90% 33.3

IntegratedGasificationCombinedCycleCoal 37% 86% 256.8

CombinedCycleGasTurbine 48% 89% 12.8

Open‐CycleGasTurbinediesel 30% 89% 19.8

Open‐CycleGasTurbinegas 30% 89% 19.8

NuclearPWRhighercost 33% 92% 0 10

Landfillgas 100% 50% 0 0.5

Microhydro 100% 50% 0 0.5

Imports

HCBNorthhydroimport 100% 38% 0 0.85

Boroma‐QuedasOcuahydroimport 100% 42% 0 0.16

ItheziTezhihydroimport 100% 64% 0 0.12

Kafuehydroimport 100% 46% 0 0.75

KaribaNorthBankextensionhydroimport 100% 38% 0 0.36

MphandaNkuwahydroimport 100% 67% 0 1.125

Kudugasimport 48% 89% 12.8 0.711

Mmamabulacoalimport 37% 85% 100 1.2

Moatize‐Bengacoalimport 35% 85% 100 1

Solar&Wind

SolarCentralReceiver12hrsstorage 100% 47% 245





SolarParabolicTrough0storage 100% 25% 245

SolarParabolicTrough3hrsstorage 100% 31% 245



SolarPVcentralisedconcentrated 100% 27% 0

SolarPVcentralisednon‐concentrated 100% 19% 0

SolarPVrooftopcommercial 100% 18% 0

SolarPVrooftopresidential 100% 18% 0

Windhighresource 100% 29% 0 10

Windmediumresource 100% 25% 0 15

PumpedStorage

PumpedStorageNew 73% 94% 0

PumpedStorageNewpump 73% 94% 0

PumpedStorageNewdam 73% 94% 0

77

PumpedStorageNewturbine 73% 94% 0

Biomass&CombinedHeatandPower

Biomassmunicipalwaste 19% 85% 200 0.1

Biomassforestrywaste(CHP) 51% 90% 210

Biomassbagasse(CHP) 37% 200

Cogen‐coal(CHP) 56% 229.1*UpperLimitonTotalInstalledCapacityin2010(GW)Briefly,thefollowingimportantassumptionsunderpinthismodelofavailablenewelectricitytechnologies:

All new coal technologies available to themodel are dry‐cooled because of increasingwater scarcity inSouthAfrica,particularly atpotentialnewbuild sites.Current committed coal‐firedbuild is supercriticaldry‐cooledandthusconsistentwiththis.Asisevidentfromtheabove,waterconsumptionisnotzerobutoftheorderofsolarthermaltechnologiesand lowcomparedtotheSATIMassumptionof1873litres/MWhforexistinglargewet‐cooledstations.

Therearealargenumberofregionaloptionsavailableforhydropowerimportbutthenetcapacity,whilesignificant, is relativelysmallasshownby thecapacityconstraintsabovewhichsumto3.4GW, just lessthan 8%of existing installed capacity. Grand Inga in the Congo, the largest hydropower resource in theregion remains excluded from the SATIM model because of concerns around political instability andeconomiccapacityissueslimitingtheprobabilityofreliableplantexpansiononthatsite.

Nuclearislimitedto10GWwhichisanassumptioninheritedfromtheSouthAfricanIntegratedResourcePlan(IRP)project.InSATIMtheintentionistoonlyextendthisassumptionto2030afterwhichnuclearwillbeunbounded.

Twowindclassesrepresentingthequalityofavailablesitesaremodelled,onewithacapacityfactorof29%andonewith25%basedonawindmapforSouthAfrica(Hageman,2008).Basedonthisresearch,thebestclass of (29%) is limited to 10 GW and any if wind is economical beyond that capacity, the additionalcapacitywillhaveacapacityfactorof25%whichinturnislimitedto15GW.

3differentPVclassesaremodelled:o CentralizedPVwithoutstoragewhichoutputselectricitytothetransmissionsystem(ELCC)o RooftopPV forcommercial/residentialbuildingswithoutbatterywhichoutputselectricity to the

residential(RESELC)andcommercial(COMELC)distributionlegs.o Rooftop PV for commercial/residential buildings with battery which outputs electricity to the

residential(RESELC)andcommercial(COMELC)distributionlegs.

Different Solar thermal options are considered, eachwith different storage options and associated costsanddiurnalavailabilities.ThefollowingnewsolarthermaltechnologiesarecurrentlymodelledinSATIM:

o SolarCentralReceiver(Tower)3hrsstorageo SolarCentralReceiver(Tower)6hrsstorageo SolarCentralReceiver(Tower)9hrsstorageo SolarCentralReceiver(Tower)12hrsstorageo SolarCentralReceiver(Tower)14hrsstorageo SolarParabolicTrough0storageo SolarParabolicTrough3hrsstorageo SolarParabolicTrough6hrsstorageo SolarParabolicTrough9hrsstorage

ParameterizationofNewTechnologyPowerPlants

TIMES offers a powerful level of parameterisation for new technologies allowing the modeller to not justcharacterisethetechnicalspecificationsandcostsbuthowthesemaychangeinthefuture.Thelearningrateof

78

renewabletechnologycostsis,forinstance,acrucialassumptionaffectingthecostoptimumfuturetechnologymix.Table52:SATIMParameterizationofNewPowerPlantTechnologies

ParameterLimitsoncapacityInvestmentcostTechnologylifeTechnologylead‐timeLowerboundonnewcapacityLowerboundoncapacityfactorBoundsonWindClassesWindIntermittencyCapacityCreditofWindDiurnalProductionofSolarwithandwithoutstoragebytimesliceNewpowerplantsaremodelledinasimilarwaytoexistingoneswiththefollowingdifferences:

Thereisnoinitialcapacityorretirementprofilefornewtechnologiesbutfuturecapacityisboundedbyacapacitylimitparameter.

Aninvestmentcostisspecified.Renewabletechnologieshaveinvestmentcoststhatdecreaseovertimetocapturetheeffectof learning fromglobal installedcapacity(notethat in thismodelthisparameter issetexogenously).

Atechnologylifeinyearsisspecified. Atechnologylead‐timeisspecifiedtocapturetheconstructionduration. In the case of committedbuild, a lowerbound is imposedon thenewcapacity as per thebuildplan,

takingintoaccounttheconstructionduration.

Sometechnologieshavealowerboundontheloadfactor(alsocalledcapacityfactor)tocharacterizefuelcontracts. This is defined by setting up an inequality relationship between special algebraic parametersratherthantheparameter‐value/spairmethodofmostotherparameterisationinTIMES.Thismeansthattheplantsmustgenerateaminimumofelectricitytoattainthiscapacityfactor.CurrentlyinSATIMthisonlyappliestogas‐fuelledgasturbineplantwhichmustmaintainaloadfactorinexcessof20%intermsoffuelsupplycontracts.

As discussed above, SATIMhas twowindclasses,with capacity factors of 29%and 25%, to reflect thevaryingqualityofsites.Eachwindclassisboundedbythepotentialfordeploymentineachclass,thatis,theminimumormaximumlimitoncapacitybuiltspecificallyinthatyear.CurrentlyinSATIMthisisonlyusedtoreflecthistoricalbuildofthehighresourcewindclassasfollows:

o 2009‐0.2GWo 2010‐0.3GWo 2011‐0.3GW

Theintermittencyofwindiscapturedbysettingthecapacityavailabilityequaltotheenergyavailability.For instancea100MWwind farmat29%estimated capacity factorwill bemodelledasonlybeable toproduce29MWofoutputinanytime‐slice.

Thecapacitycreditofwindisfixedataconservativevaluebelowthecapacityfactor.CurrentlyinSATIMthecapacitycreditforbothwindclassesissetto0.23.

In SATIM, solarPVwithout battery storage can only output power during daytimehours. This isaccomplishedbysettingthecapacityfactorpertimeslice.Timeslicesnotindaylightaresettozero.

79

SolarPVwithbattery storage can also provide somepower for thewinter evening peak as shownbyFigure31below:

Figure31:IllustrativeComparisonofSATIMCapacityAvailabilityFactorsbyModelTimesliceforNewSolarCentralReceiver

TechnologiesHaving3‐14HoursofStorage–WinterSeason

ParameterisationofNewPowerPlantTechnologiesintheProcessofBeingImplemented

Insomescenarios it isuseful fornewNuclear reactors tobemodelledwith the “lumpy investment” feature.Typically, this feature would be implemented when the power sector is modelled alone because it’s verycomputationallyintensive.Forfullsectormodelrunsthedemandtypicallyincreasesintheregionof1GWperannumwhichaccommodatestheminimumbuildsizeofmosttechnologiesandthe‘lumpyinvestment’featureisnotnecessary.

80

SupplyofEnergyfromtheLiquidFuels&GasSupplySector

This sector includes the supply of liquid fuels like diesel and gasoline as well as gas to the South Africaneconomy.ConventionallytheseproductswouldbederivedfromcrudeoilbutSouthAfricahasalargeso‐calledsyntheticfuelindustrythatproducesliquidfuelsfromgasandcoalfeedstocks.Thisindustry,whichincludesacoal to liquid refinery at Secunda, operatedby Sasol and a gas to liquid refinery atMosselBayoperatedbyPetroSA, complicates the modelling of this sector somewhat because these plants add a number of inputcommoditiestotheenergychain.Thesupplyofgasbypipelinefromrichdepositsinneighbouringcountries,bythe import of LNG or locally mined from recently discovered shale gas deposits are competing options forindustrial andeven residential energy supply. Including theseprimaryenergy supplies and the technologieslikepipelinesandterminalsthatwillsupplythemispartofthecurrentresearchprojectonthissector.

ModelStructure

Petroleum refining (refining crude oil, natural gas and coal3) is an extremely complex process which hasnumerousdiscreteprocessingunitsoperating inclose interaction.Thenumerousprocessingunitsproducearange of energy and non‐energy products and they also have unique energy requirements (in the form ofancillaryenergyservices)andasaresultcautionhastobetakenwhenmodellingtheoperationofrefineries.InSATIM, there are three distinct modelling processes associated with refinery modelling, namely energymodelling(whichconsiders inputcommoditiesand theassociatedoutputproducts),steamsupplymodellingandmodellingofnon‐energyoutputproducts.Beforedelvingintothethreedistinctmodellingprocesses,itisworthhighlightingthetypesofrefineriesthataremodelled.Modellingliquidfuelssupplyinvolvesaccountingforexistingandpossiblefuturerefineries.TheExistingRefineriesareregroupedinto4technologiesasfollows:

RefineryCrudeOilCoastalExisting(Sapref,Enref,Chevref) RefineryCrudeOilInlandExisting(Natref) RefineryGas‐to‐Liquid(GTL)Existing(PetroSA) RefineryCoal‐to‐Liquid(CTL)Existing(Secunda)

Themodel also accounts for two future refineries, namelyNewCTL andNewCrudeOilRefinerywhichwillbothcomeonlinein2018,withalifespanof50years.Thesefuturerefineriesaregenericrefineriesincludedinthemodeltoaccountfortheshortfalloffutureliquidfuelsupply.TheNewCrudeoilRefineryhasemulatedthecapacity data for the plannedMthombo refinery that is planned to be built in the Eastern Cape in the nearfuturewhiletheNewCTLrefineryhasemulatedthecapacitydatafromthediscontinuedMafuthaCTLrefineryproject.

BaseYearData

Therefineryslatedataforexistingrefineriesisclearlycriticalinconstructingamodelthatrealisticallyreflectsthefuelsactuallyproducedbytheliquidfuelssectorwhichcanbalancethedemandsoftheconsumingsectors.ForSATIMthisdatarestsonwhatisnowquiteanoldstudy(Lloyd,2001)buttheonlylargescalechangestoSouthAfrica’srefineriessincethenhasbeentheincreaseincapacityofEnrefin2003from100,000barrels/dayto 125,000 barrels/day. The finding of this study,with coastal crude refineries aggregated, is presented inTableH.2inAppendixHinthespreadhsheetappendedtothisdocument.ThelevelofproductdisaggregationissomewhatlowerinSATIMwhichmodelsthefollowingoutputcommoditiesfromrefineries:

3SouthAfricanliquidsupplyindustryusescoal,gasandcrudeoilastheirfeedstock

81

Table53:SATIMRefineryOutputs

Commodity

AviationGasoline

DieselOil

Gasoline

MethaneRichGas

Kerosene

LiquifiedPetroleumGas

OtherOil‐derivedProducts

AssumptionsCharacterisingRefineriesandRefiningProcesses

Publicdata fordetailing theexisting fleetof liquidfuelrefineries isscarceandindustryreportstowhichtheEnergyResearchCentrehaveaccess(egLloyd,2001)arenowquiteold.Theplanningoffutureinfrastructurehas also been a less public process in the semi‐regulated privately owned liquid fuels sector than in theelectricitysupplysectorwhich isdominatedbyapubliccompany.Therefore there isnot thesamewealthofdataforcharacterisingfuturetechnologiesinSATIM.Neverthelessduetoitsimportancethesectorismodelledinsomedetailasdescribedbelow.

ParameterizationofRefineryTechnologies

Thefundamentalparametersrequiredforanenergymodeloftheliquidfuelsupplysectorcanbesummarisedasfollows:

inputandoutputcommodities, investmentcosts(newrefineriesonly)andrunningcosts refineryavailabilitiesandefficiencies.

A more detailed parameterisation of refinery technologies in SATIM is presented below in Table 54. Table54:SATIMParameterisationofRefineryTechnologies

ParameterType Parameters DescriptionEnergyInputCommodities(feedstockand ancillary services) for crude oilrefineries

Crudeoil Theseare the feedstocks for the refineries.The natural gas that is imported fromMozambiqueundergoesconversionprocesstomakeitamaterialgaswhichcanserveasinputintotheCTLrefineries. Themethanerich gas is produced by SASOL in its CTLrefineries.

MethaneRichGasElectricity

EnergyInputCommodities(feedstockand ancillary services) for Gas toLiquidrefineries

NaturalGas

SteamEnergyInputCommodities(feedstockand ancillary services) for Coal toLiquidrefineries

CoalNaturalGasElectricitySteam

Set of Energy Output Commodities:products (including commodities fornon‐energyapplicationse.g.bitumen)

AviationGasoline Thesearethefuelsthatareproducedbytherefineries.

Dieseloil

Gasoline

HeavyFuelOil

82

Kerosene

LiquefiedPetroleumGas(LPG)

OtherOil(non‐energy)

MethaneRichGas

SetofNon‐energyoutputcommodities:processemissions

CH4,CO2,N2O,PM10setc. Greenhouse gas and toxic regulatedemissionsfromtherefinery

Efficiency Efficiency Relates the sum of energy inputcommodities(includingbothfeed‐stockandancillary services) to the sum of energyoutputcommodities

Availability Availability This constrains the total annual output tothecapacity.Theavailabilityislessthanonetoaccount formaintenanceandunplanneddown‐time

Inputshareconstraint Constraint(%) Aconstraintfixingtheshareofthedifferentenergy input commodities based onhistoricalobservations

Upper and lower constraints on theshare of the different energy outputcommodities.

Constraint(%) The lower andupper shares are set to 5%oneithersideoftheproductslateobservedin (Lloyd 2001) to give the existingrefineries some range of manoeuvreallowedby theequipment inplacewithoutrequiringfurtherinvestment.

Operating cost based on reportedcosts.

Costs(Rands)

Processemissionfactor CO2(kton/TJ) Relating the CO2 emission to the totalenergy output of the refinery. Energyemissions (e.g. arising from theproductionof steam) are accounted for at the level ofthe supply of the fuels as done in all theothersectors.

Retirementprofile

Time(years) Specifies how much capacity is initiallyavailable andhow that decays asplantsorpartsofplantsretireinthefuture.

CurrentlytheproductionofexistingrefineriesinSATIMissettotheproductslatedeterminedbyLloyd(2001)with product shares as anupper boundonly. Currently in SATIM commodity output shares are fixed for alltechnologiesexceptnewtechnologycrudeoilrefinerieswherethere issomeflexibility in theslateasshownbelow:Table55:AssumedUpperBoundsonOutputCommoditySharesforRefineryTechnologies

OutputProductCrudeCoastalExisting

CrudeInlandExisting

GTLExisting CTLExisting CTLNew CrudeNew

AvGasoline 0% 0% 0% 0% 0% 0%

Diesel 33% 39% 29% 24% 73% 50%

Gasoline 29% 32% 50% 54% 24% 50%

HFO 23% 4% 0% 0% 0% 0%

Kerosene 11% 21% 12% 4% 0% 20%

83

LPG 2% 0% 4% 1% 4% 3%

Other 2% 4% 5% 7% 0% 5%

MethaneRichGas 0% 0% 0% 8% 0% 0%

TOTAL 100% 100% 100% 100% 100% 128%Note:Ifthesumofupperboundssumto100%thesharesareeffectivelyfixed

Futureworkneedstoextendtheserefineryslateassumptionsfromjustanupperboundtoanupperandlowerboundandusethistodefinearealisticband,consultationwithrefineryexperts,ofoutputcommoditysharesfor existing refineries and new CTL refineries rather than the current fixed slate. Although non‐energyproducts (commodity “Other’ above) do not contribute to the refinery energy outputs, they are includedbecausethisallowsthemodeltoaccountforenergyusedtoproducetheseproducts.Table56belowpresentsasummaryoftheassumptionsregardingrefinerytechnologycharacteristicsandcostsasusedinSATIMcurrrently.Table56:SummaryofExistingandNewRefineryTechnologyCharacteristics

ExistingTechnologies NewTechnologies

Units

SasolCTL

InlandCrudeExisting

CoastalCrudeExisting

PetroSAGTL

NewCTL1

NewCrude2

Capacity bbl/day 150000 108000 405000 45000 Capacityintermsofoutputs

PJ/annum 246 212 874 59

OverallEfficiency % 44% 93% 95% 78% 49% 97%

Availability % 96% 96% 96% 96% 96% 96%

PlantLife Yrs 50 50RunningCostsperunitofoutput3

mR/PJ 30 11 11 25 30 11

InvestmentCost3mR/[PJ/annum]

0 0 0 0 305 66

Lev.CostofProduction(in2006)

R/GJ 41 88 87 46 66 91

Lev.Costofproduction(withIRP/100$/bbl)

R/GJ 57 128 121 95 80 129

CO2emissions (kt/PJ) 118.88 6.87 2.90 0.004 118.88 6.18

CH4emissions (kt/PJ) 1.49 0.00 0.00 0.004 1.49 0.001:BasedondataforproposedMthomboproject2:BasedondataforproposedMafuthaproject3:mR:currencyunit‐millionSouthAfricanRands4:Theseareunknownbutthoughttobelow

As canbe seen above, theCTL refinery technologyhas very high greenhouse gas emissions and low energyconversionefficiencybutoperatesatalowcostofproductionbecausecoalislocallycheap.Thisalongwiththedominance of coal in the electricity supply sector is one of the determining factors in SouthAfrica’s carbonintensive economy and presents both a challenge and an opportunity to mitigation initiatives. The modeloutputsforthissectorwillthereforebeverysensitivetoanemissionsconstraint.

Another future improvement envisioned is to specify a minimum unit size for new technologies. Currently, so-called ‘lumpy’ investments are not implemented in SATIM for the liquid fuels sector. A caveat is that this feature can extend the time to find a solution significantly.

84

CharacterizationofCTLTechnology

ModellingCTLrefineriesrequiressomeadditionaldatatodeterminethesharesofinputsandoutputstothatrequired for crudeoil refineries.Fortunatelymuchof thisdata is available indocumentspublishedbySasolthemselves,thecompanyoperatingSouthAfrica’sCTLrefinerylocatedatSecunda.Inbriefthistechnologyischaracterisedasfollows:1. Requiredforconstraintsonoutputshares:o Theproductslateisderivedfrom(Lloyd,2001)o TheMethaneRichGasoutputisdeterminedfromSasol’sfinancialstatementsaspublishedinthe“Analyst

BookDec2006”(SASOL,2007)2. Requiredforconstraintsoninputshares:o TheTotalCoalusebySasol isdetermined fromSasol’s financial statements aspublished in the “Analyst

BookDec2006”(SASOL,2007).o The Coal formaterial use for the base year (feedstock excl. steam generation) is published in the Sasol

Sustainabilityreport2009(SASOL,2009)expressedinktondryashfree(DAF)o Thedryash free (DAF) coal to runofmine coal (ROM) ratioused to convert thisnumber is0.65asper

personalcommunicationwithSasol.o Thetotalcoalforenergyuse(inTJ)ispublishedintheSasolSustainabilityreport2009(SASOL,2009).o The Energy content of steam used in the Sasol process of 2,627 MJ/ton comes from a personal

communicationwithSasol.

ModellingtheSupplyofAncillarySteamInputServicestoRefineries

Refinerieshavevariouscommodityinputswhichcanincludecrudeoil,coal,gas,methanerichrefinerygasandsteam,allofwhichcomplicatescostingtheenergychain.Steamismodelledasanancillaryinputservicetotherefinerybycreatingboilertechnologiesthatoutputsteamwithanenergycommodityasaninput.Themodellingof steamas an ancillary input service allows themodel topotentially optimise themost costeffectivefuel(coalvsgas)andtechnology(e.g.existingvsnewandmoreefficientboilervsCHP)toprovidethesteamneededforprocessheat,aswellasforfeedstockinCTLplants.Thislatteruseismuchgreaterperunitofrefineryoutputthanprocessheatrequirements.SteamisalsoconsumedbycruderefineriesandGTLplantsbutfurtherdata is required for thecharacterizationand therefore this consumption isnotcurrently reflected inSATIM.ThesteamconsumptionofcrudeandGTLrefineriesishoweversignificantlylowerandsotheabsenceofthisdetailisassumedtonothaveaverysignificantimpactontheoverallresults.Whiletheframeworkforoptimisingrefinerysteamproductionisinplace,thecurrentversionofSATIMthereforeonlyhasthefollowing2technologiesimplemented:

Table57:CurrentRefinerySteamBoilerTechnologiesImplementedinSATIM

BoilerTechnology InputCommodity OutputCommodity EfficiencyRefineryCTLBoilerExisting CoalExisting SteamExisting 72%RefineryCTLBoilerNew CoalNew SteamNew 77%

Competing technologieshave thusnotyetbeendefinedandcostsnotyet researched.This formspartof thework on the on‐going SATIM project andwill allow formore rigorous treatment of the liquid fuels supplysector.

85

SupplyofPrimaryEnergy

TheANSWERinterfaceandTIMESmodelaredesignedaroundtheprincipleofareferenceenergysystemthatdepicts the whole energy chain, so it is important to account for all the energy used in the model fromextractionto finalconsumption.Primaryenergyresourcessuchascoalareusedtoproducefinalenergyandeachstepalong thedifferent transformationstagesundergo losses.These losses (in the formofefficiencies)havetobeaccountedforinthemodel.Sincetheresourcesarefiniteinnature,themodelhastomeetthedemandforenergywithinthatconstraint.Tocater for limitations on resources, the analyst has to use constraints to limit an evermountingusage of theresource.IntheSATIMmodel,resourceconstraintshavebeenimplementedforfossilfuelsandforrenewableenergy. It is assumed that there is no resource limit on nuclear fuel during the modelling period. The keyresourceconstraintsonfossilfuelsarethesupplyofnaturalgasfromregionalsources(thesouthernandwestcoastgas‐fieldsinSouthAfricanwaters,andtheNamibianandMozambicangas‐fields)andthesupplyofcoalfromwithinSouthAfrica.SouthAfricanoilfieldsareveryinsignificantandareassumedtodepleteveryrapidly.It isassumedthat there isno limit to importedoiland liquifiednaturalgas,andglobalscarcity isexpressedthroughscenariosofhigherprices.CoalbedmethaneisassumedtohavenoenergypotentialforSouthAfrica,andthesameassumptionismadeaboutshalegasintheKaroobasinforthereferencecasealthoughthelatteristhesubjectofalternativescenariosforcurrentSATIMprojects.No resource limit was assumed for the LTMS for coal. In contrast to the LTMS MARKAL model, SATIMconstrainscoalreservestobe27billiontonscurrently(StatsSA2010)andseveralscenariosforcoalpriceandexport ratios have been explored. Coal reserves are however poorly documented in the public domain andexpertopinionhassuggestedthatthisassumptionofreservesmaybeasignificantunderestimate.Incommonwith theLTMS, fourmarket segments are assumed– existingandnewcoal for electricity, coal for syntheticfuels,coalfordirectindustrialuse,andcoalforexport,withcalorificvaluesrisinginthesameorder.Limitsonrenewableresourcesprimarilyapplytowindresources,whereithasbeenassumed,basedonanalysiscarriedoutbyHagemann,thatthelimitonwindgenerationpotentialforsiteswithcapacityfactorsgreaterthan30%is10GW,andforthosewithcapacityfactorsgreaterthan25%is80GW.Thepricesofproducttypicallyrefinedfromcrudeoilsuchasgasoline,dieselandkeroseneareassumedtobelinearlylinkedtothecrudepricebyafixedconstant.Thisallowslocalandimportpricesforalloilderivedfuelstobeautomatically generated foroneoil priceprojectionwithout thenecessityof separateprojections.ThepriceofoilproductsprojectionsarebasedontheIEAWorldEnergyOutlook2011projectionsfortheoilpriceinthe“currentpolicies”scenario.

86

References

Alton,T.,Arndt,C.,Davies,R.,Hartley,F.,Makrelov,K.,Thurlow,J.,etal.(2012).TheEconomicImplicationsof

IntroducingCarbonTaxesinSouthAfrica.WorkingPaper2012‐46.Helsinki,Finland:WorldInstituteforDevelopmentEconomicsResearch(WIDER),UnitedNationsUniversity.

Arndt, C., Davies, R., & Thurlow, J. (2011).Energy Extension to the SouthAfricaGeneral Equilibrium (SAGE)Model.UnitedNationsUniversity'sWorldIntituteforDevelopmentEconomicsResearch(UNU‐WIDER).

Bell,A.,Stone,A.,&Harmse,B.(2003).FINALREPORTINVESTIGATION(DESKTOPSTUDY)INTOTHEOPTIMUMFUTUREOCTANEGRADESTRUCTUREFORSOUTHAFRICA‐ExcelSpreadsheetModelaccompanyingmainreport.Pretoria:DepartmentofMineralsandEnergy,RepublicofSouthAfrica.

Bennett,K.(1975).EnergyUtilisationinSouthAfrica.EnergyResearchInstitute,UniversityofCapeTown.Beute,N.(2010).ElectrificationinSouthAfricaoverthelast20years.DomesticUseofEnergy.CapeTown:IEEE.CBECS.(2003).CBECStableE6AendusedatakWhSqf.CCT.(2007).Annualsurveyofregistered industrialandcommercial fuelburningappliances.CapeTown,South

Africa:CityofCapeTownairqualitydepartment.Cuenot, F., & Fulton, L. (2011). International Comparison of Light‐Duty Vehicle Fuel Economy and Related

Characteristics.Paris:IEA.Dargay,J.,Gately,D.,&Sommer,M.(2007,January).VehicleOwnershipandIncomeGrowth,Worldwide:1960‐

2030.EnergyJournal,28(4).de Villiers, M. (2000). Greenhouse gas baseline and mitigation options for the commercial sector. Energy

ResearchCentre,UniversityofCapeTown.DEAT. (2007).LongTermMitigationScenarios2007.Retrieved fromDepartmentofEnvironment&Tourism,

Republic of South Africa:http://www.environment.gov.za/HotIssues/2009/LTMS2/LTMSTechnicalSummary.pdf

Dekenah,M.(2010).AloadprofilepredictionmodelforresidentialconsumersinSouthAfrica.DomesticUseofEnergyConference.CapeTown:EnergyInstitute,CapePeninsulaUniversityofTechnology.

Dieselnet. (2000). Emission Test Cycles ECE 15 + EUDC / NEDC. Retrieved July 2012, from Dieselnet:http://www.dieselnet.com/standards/cycles/ece_eudc.php

DoE. (2009). Digest of South African Energy Statistics 2009. Directorate: Energy Information Management,ProcessDesignandPublications.Pretoria:DepartmentofEnergy,RepublicofSouthAfrica.

DOE. (2009). Energy Balances 2006 V1 ‐ Microsoft Excel File. Retrieved from Department of Energy:http://www.energy.gov.za/files/media/media_energy_balances.html

DOE. (2011). IntegratedResourcePlan forElectricity2010 ‐ 2030. Department of Energy, Republic of SouthAfrica.

DoT.(2005).KEYRESULTSOFTHENATIONALHOUSEHOLDTRAVELSURVEY‐TheFirstSouthAfricanNationalHouseholdTravelSurvey2003.Pretoria:DepartmentofTransport,RepublicofSouthAfrica.

DoT. (2009). National Transport Master Plan (NATMAP) 2050 Modelling Report. Pretoria: Department ofTransport,RepublicofSouthAfrica.

EnergyResearchCentre.(2011).SouthAfricanLowEmissionsPathwaysProject,FinalTechnicalReport.EnergyResearchCentre,UniversityofCapeTown.

FAO. (2011).Why has Africa Become a Net Food Importer? Explaining Africa Agricultural and Food TradeDeficits.Rome:TradeandMarketsDivision,FoodandAgricultureDivisionoftheUnitedNations.

Gbetibouo,G.,&Ringler,C.(2008).MappingSouthAfricanFarmingSectorVulnerabilitytoClimateChangeandVariability ‐ A Subnational Assessment. IFPRI Discussion Paper 00885. International Food PolicyResearchInstitute(IFPRI).

Gertler,P.,Shelef,O.,Wolfram,C.,&Fuchs,A. (2012).Poverty,GrowthandtheDemand forEnergy.California:UniversityofCaliforniaatBerkeley.

87

Goyns,P.(2008).Modellingreal‐worlddriving, fuelconsumptionandemissionsofpassengervehicles:acasestudy in Johannesburg. Thesis submitted in fulfillment of the Requirements for the degree DoctorPhilosophiæ in Energy Studies. Department of Geography, Environmental Management and EnergyStudies,FacultyofScience,UniversityofJohannesburg.

Gutknecht, B., & Akos, I. (n.d.). 1Climate in danger: Facts and Implications of the Greenhouse Effect”. SwissAgencyfortheEnvironmentForestsandLandscape.

Hageman,K.(2008).Appendix3 ‐Newassessmentofwindresources forSouthAfrica inMarquardA,MervenBandTylerE (2008) 'Costinga2020Targetof15%RenewableElectricity forSouthAfrica'. CapeTown:EnergyResearchCentre,UniversityofCapeTown.

Havenga,J.,Simpson,Z.,&vanEeden,J.(2011).TheStateofLogisticsinSouthAfrica‐SustainableImprovementsorContinuedExposuretoRisk.CentreforSupplyChainManagement,UniversityofStellenbosch.

Howells,M.,Kenny,A.,&Solomon,M.(2002).EnergyOutlook2002‐ModellingEnergyinSouthAfrica.EnergyResearchCentre,UniversityofCapeTown.

Hughes,A.,Haw,M.,Winkler,H.,Marquard,A.,&Merven,B.(2007).EnergyModeling:AmodellinginputintotheLongTermMitigationScenariosprocess,LTMSInputReport1.CapeTown:EnergyResearchCentre,UCT.

ICCT. (2011). Global Comparison of Light‐Duty Vehicle Fuel Economy/GHG Emissions Standards August 2011Update.InternationalCouncilonCleanTransportation.

IEA.(2009).EnergyBalancesofnon‐OECDCountries.Paris:InternationalEnergyAgency.IEA.(2011).EnergyStatisticsfornon‐OECDCountries.Paris:InternationalEnergyAgency.IEA.(2011).SustainableMobilityProject(SMP)Model‐Excelspreadsheetforwardedbyemail.International Association of Public Transport & African Association of Public Transport. (2010). Report on

StatisticalIndicatorsofPublicTransportPerformanceInAfrica.IPCC a. (2006). 2006 IPCC Guidelines for National Greenhouse Gas Inventories. Intergovernmental Panel on

ClimateChange.IPCC b. (2007). Climate Change 2007:Working Group I: The Physical Science Basis ‐Direct GlobalWarming

Potentials.IntergovernmentalPanelonClimateChange.IPCC c. (2007). IPCC FourthAssessmentReport: Climate Change 2007 ‐ SynthesisReport". Intergovernmental

PanelonClimateChange.Ittmann,H.,King,D.,&Havenga, J.(2009).TheStateOfLogistics‐AFive‐YearReview.SAPICS31STANNUAL

CONFERENCEANDEXHIBITION.SunCity,SouthAfrica:SAPICS.Jackson,T.(2001).FleetCharacterizationDataforMOBILE6:DevelopmentandUseofAgeDistributions,Average

AnnualMileageAccumulationRates,andProjectedVehicleCounts forUse inMOBILE6.AssessmentandModelingDivisionOfficeofTransportationandAirQuality.U.S.EnvironmentalProtectionAgency.

JTRC. (2008). Transport Outlook 2008 Focusing on CO2 Emissions froom Road VehiclesDiscussion PaperNo.2008‐13.Paris:JointTransportResearchCounciloftheOECDandtheInternationalTransportForum.

Kia. (2012). Brochure ‐ Introducing the New Rio. Retrieved from Kia South Africa:http://www.kia.co.za/Content/Uploads/documents/0398GLE%20Kia%20Rio%20Brochure_small.pdf

Kwon, T.‐H. (2006). The Determinants of the Changes in Car Fuel Efficiency in Great Britain (1978‐2000).EnergyPolicy,34,2405‐2415.

Lloyd, P. (2001). The South African Petroleum Industry: A Review. Cape Town: Report by Industrial &PetrochemicalConsultants(Pty)Ltd‐selectedextractsforwardedinpersonalcommunicationbyPhilipLloyd.

Mwakasonda, S. (2009). Greenhouse Gas Inventory South Africa 1990 to 2000. Pretoria: Department ofEnvironment&Tourism,RepublicofSouthAfrica.

NAAMSA/SAPIAWorkingGroup.(2009).ExcelSpreadsheetoftheNAAMSA/SAPIAWorkingGroupVehicleCarParcasUsedfortheSAPIAPetrolandDieselStudy,forwardedbyemailAugust2009.

88

Nemry,F.,Leduc,G.,Mongelli, I.,&Uihlein,A.(2008).Environmental ImprovementofPassengerCars(IMPRO‐car). Seville, Spain: European Commission, Joint Research Centre, Institute for ProspectiveTechnologicalStudies.

NERSA. (2006). 2006 Energy Supply Statistics for South Africa. National Energy Regulator of South Africa(NERSA).

Pelkmans, L., & Debal, P. (2006). Comparison of on‐road emissions with emissions measured on chassisdynamometertestcycles.TransportationResearch,11(PartD),233–241.

Road Traffic Management Corporation. (2009). Road Traffic Report – Year 2008. RTMC, an Agency of theDepartmentofTransport,RepublicofSouthAfrica.

SAIRR.(2012,February1).SouthAfricaremainsanetfoodexporter.RetrievedfromSouthAfricanInstituteofRace Relations: http://www.sairr.org.za/media/media‐releases/SA%20remains%20a%20net%20food%20exporter.pdf

SASOL.(2007).AnalystBookfortheHalfYearendedDecember2006.SasolLimited.SASOL.(2009).PositiveActions‐SasolSustainableDevelopmentReport2009.SasolOil.Schafer,A.(2006).Long‐TermTrendsinGlobalPassengerMobility.TheBridgeLinkingEngineeringandSociety,

24‐32.Schafer,A.,&Victor,D.(2000).TheFutureMobilityoftheWorldPopulation.TransportationResearch,34(Part

A),171‐205.Senatla, M. (2012). Periodic modelling in a highly dynamic sector: The case of updating Long Mitigation

Scenarios' (LTMS)MARKALmodel in the residential sectorof SouthAfrica.CapeTown,PresentedatInternationalEnergyWorkshop2012.

StatsSA. (2012).GrossdomesticproductAnnualestimates2002–2010Regionalestimates2002–2010Thirdquarter2011.Pretoria:StatisticsSouthAfrica.

StatsSA.(2010a).StatisticsSouthAfrica‐Timeseriesdata,buildingstatistics.Stone,A. (2004).CreatingaNationalDatabaseofTrafficBasedVehicleEmissionsFactorsandVehicleParc ‐

Excelspreadsheetmodelsupportingthispublication.NationalAssociationofCleanAir(NACA)WesternCapeSymposium.CapeTown.

Stone,A.,&Bennett,K.(2001).ABULKMODELOFEMISSIONSFROMSOUTHAFRICANDIESELCOMMERCIALVEHICLES.NationalAssiciationofCleanAirConference(NACA)2001.PortElizabeth.

University of California at Riverside, Global Sustainable Systems Research. (2002). Nairobi, Kenya VehicleActivityStudy.

Vanderschuren,M.(2011).PersonalCommunication‐Excelspreadsheetofvehiclemodeloutputs.DepartmentofCivilEngineering,UniversityofCapeTown.

Winkler,H(ed.). (2007,October).LongTermMitigationScenarios:TechnicalReport.Preparedby theEnergyResearchCentreforDepartmentofEnvironmentAffairsandTourism.Pretoria.

Winkler,Hetal.(2006).EnergyPoliciesforsustainableDevelopmentinSouthAfrica‐OptionsfortheFuture.EnergyResearchCentre.UniversityofCapeTown.

Yu,L.,Qiao,F.,Li,G.,&Oey,H.(2002).ForecastingTrafficCharacteristicsforAirQualityAnalyses.Austin:TexasDepartmentofTransportation.

Zhou,P.,Yamba,F.,Lloyd,P.,Nyahuma,L.,Mzezewa,C.,Kipondya,F.,etal.(2009,November).DeterminationofregionalemissionfactorsforthepowerSectorinSouthernAfrica.JournalofEnergyinSouthernAfrica,20(4).

89

Appendices

ThedetailedassumptionsofthemodelcanbeaccessedintheappendiceswhichcanbefoundintheMicrosoftExcelfile‘SATIMMethodologyAppendicesv1.0.xlsx’postedontheERCwebsitewiththisdocument.

AppendixA. Comparison ofAggregateDemand SectorEnergyConsumptionby Fuel for 2006 ‐ ComparisonbetweenSATIMandDOEEnergyBalance

Appendix B1. Industrial Sector ‐ Summary of SATIM Energy Consumption Estimates by Primary Fuel orEnergyCarrier,Sector&EnergyServiceforBaseYear2006

AppendixB2.IndustrialSector‐SelectedReferenceCaseAssumptions

Appendix C1. Agriculture Sector ‐ Summary of SATIM Energy Consumption Estimates by Primary Fuel orEnergyCarrier,Sector&EnergyServiceforBaseYear2006

AppendixD1. Residential Sector ‐ Summary of SATIM Energy Consumption Estimates by Primary Fuel orEnergyCarrier,Sector&EnergyServiceforBaseYear2006

AppendixD2.ResidentialSector‐ReferenceCaseAssumptions

Appendix E1. Transport Sector ‐ Summary of SATIM Energy Consumption Estimates by Primary Fuel orEnergyCarrier,Sector&EnergyServiceforBaseYear2006

AppendixE2.TransportSector‐ReferenceCaseAssumptions

Appendix F1. Commercial Sector ‐ Summary of SATIM Energy Consumption Estimates by Primary Fuel orEnergyCarrier,Sector&EnergyServiceforBaseYear2006

AppendixF2.CommercialSector‐ReferenceCaseAssumptions

AppendixG1ElectricitySupplySector‐ReferenceCaseAssumptionsforExistingPowerPlants

AppendixG2ElectricitySupplySector‐ReferenceCaseAssumptionsforNewTechnologyPowerPlants

AppendixH.LiquidFuelsSupplySector‐ReferenceCaseAssumptions

AppendixI.PrimaryEnergySupplySector‐ReferenceCaseAssumptions

assumptions and methodologies in the south african times ... · assumptions and methodologies in...

Documents