High-renewables scenarios Thought experiments for the South African power system

Transcription

1 High-renewables scenarios Thought experiments for the South African power system CSIR Energy Centre Pretoria, 22 August 216 Dr Tobias Bischof-Niemz Chief Engineer Dr Tobias Bischof-Niemz Cell: Crescent Mushwana Cell:

2 CSIR s new Energy Centre streamlines and expands CSIR s energy research offerings in five areas today: 25 employees, growing CSIR Energy Centre research areas Energy Efficiency & Demand Response Renewable Energy Technologies Energy Storage and Hydrogen Energy-System Planning & Operat. Energy Markets and Policy Energy Efficiency in all end-use sectors Demand forecasting Demand response Energy statistics Solar Biomass/-gas Liquid Biofuels Small Hydro Ambient Heat Energy Storage (Power-to-Power, Power-to-Heat) Power-to-Hydrogen Power-to-Gas Power-to-Liquids Electric Mobility Energy Planning Grid Planning Micro and Island Grids System Operations Smarter Grids Macro-and Energy Economics Clean Energy Markets (RE and Natural Gas) Regulatory Environment and Market Design Energy-Autonomous Campus (EAC) Programme 2 Sources: CSIR Energy Centre analysis Five year objective: approx staff to be able to address all relevant dimensions of RSA s energy transition

3 In 215, 12 GW of wind and solar PV newly installed globally Global annual new capacity in GW/yr Solar PV 4 4 Subsidy-driven growth triggered significant technology improvements, mass manufacturing and subsequent cost reductions Consequence Renewables are now cost competitive to alternative new-build options in South Africa Total South African power system (approx. 45 GW) Sources: GWEC; EPIA; BNEF; CSIR analysis This is all very new: Almost 9% of the globally existing PV capacity was installed during the last five years alone!

4 Renewables until today mainly driven by US, Europe, China and Japan Globally installed capacities for three major renewables wind, solar PV and CSP end of Operational capacities in GW end of USA Canada 15 2 Americas w/o USA/Canada Europe Middle East and Africa 25 4 India.2 4 China 4 7 Japan 5 Australia 7 Rest of Asia Pacific Solar CSP PV Total World Sources: GWEC; EPIA; CSPToday; CSIR analysis Total RSA power system (45 GW) South Africa is rapidly picking up with 4./2.8/1.1 GW new capacity by 218-2

5 Agenda Renewables in South Africa potential in South Africa Extreme renewables scenarios 5

6 Integrated Resource Plan 21 (IRP 21): Plan of the power generation mix for South Africa until 2 Installed capacity Energy mix Total installed net capacity in GW Electricity supplied in TWhper year Solar PV CSP Hydro Nuclear Peaking Gas % 5% % 1% 5% 46 5% 2% 1% Solar PV CSP Hydro Nuclear Peaking Gas Renewable TWh'sin 2 (14%) Carbon free TWh's in 2 (4%) Coal % 65% Coal Share new renewables % 9% 6 CO 2 intensity Implementation of the IRP is done by Department of Energy through competitive tenders ( REIPPPP for renewables) Note: hydro includes imports from Cahora Bassa Sources: Integrated Resource Plan 21, as promulgated in 211; CSIR Energy Centre analysis

7 Competitive tender outcome: new wind/solar PV projects very cheap First four bidding windows results of Department of Energy s RE IPP Procurement Programme (REIPPPP) Average tariff in R/kWh (May-216-R) Solar PV Bid ow 1 (4 Nov 211) Bid submission dates Bid ow 2 (5 Mar 212) =. GW Bid ow (19 Aug 21) Bid ow 4 & additional (18 Aug 214) = 2. GW? Bid ow 4 expedited (early 215) Notes: For CSP Bid ow and.5, the weighted average of base and peak tariff is indicated, assuming 5% annual capacity factor; BW = Bid ow; Sources: Department of Energy s publications on results of first four bidding windows StatsSA on CPI; CSIR analysis

8 Consequence of renewables cost reduction for South Africa: Solar PV and wind are the cheapest new-build options per kwh today Lifetime cost per energy unit (LCOE) ZAR/kWh (May-216-R) Renewables Conventional new-build options Capital Fixed O&M Fuel (and variable O&M) Bid ow Variable: Solar PV Bid ow 1.69 Variable: Baseload: Coal Baseload: Nuclear Mid-merit: Gas (CCGT) Mid-merit: Coal Peaking: Gas (OCGT) Assumed capacity factor 85% 1% Peaking: Diesel (OCGT) 92% 5% 5% 1% 8 Note: Changing full-load hours for conventional new-build options drastically changes the fixed cost components per kwh (lower full-load hours higher capital costs and fixed O&M costs per kwh); Assumptions: Average efficiency for CCGT = 55%, OCGT = 5%; nuclear = %; IRP costs from Jan-212 escalated to May-216 with CPI; assumed EPC CAPEX inflated by 1% to convert EPC/LCOE into tariff; Sources: IRP 21 Update; DoE REIPPPP; StatsSA for CPI; Eskom financial reports for coal/diesel fuel cost; EE Publishers for Medupi and Kusile cost; CSIR analysis

9 Agenda Renewables in South Africa potential in South Africa Extreme renewables scenarios 9

10 The CSIR conducted a and Solar PV Resource Aggregation Study CSIR, SANEDI, Eskom and FraunhoferIWES conducted a joint study to holistically quantify the wind-power potential in South Africa and the portfolio effects of widespread spatial wind and solar power aggregation in South Africa Atlas South Africa (WASA) data was used to simulate wind power across South Africa Solar Radiation Data (SoDa) was used to simulate solar PV power across South Africa Output: Simulated time-synchronous solar PV and wind power production time-series 5 km x 5 km spatial resolution Almost 5, pixels covering entire South Africa 15-minute temporal resolution 5 years temporal coverage (29-21) 1 Sources:

11 South Africa has wide areas with > 6 m/s average wind 1 m Average wind speed at 1 meter above ground for the years from for South Africa Polokwane Latitude -26 Johannesburg Upington Bloemfontein Durban Port Elizabeth Cape Town Longitude 11 2 Mean wind speed (29 to 21) at 1 m above ground [m/s] 12

12 Five different generic wind turbine types defined for simulation of wind power output per 5x5 km pixel in South Africa (~5 pixels) High-wind-speed turbine Low-wind-speed turbine Turbine type no Nominal power [MW] Selection criterion Blade diameter [m] Hub height[m] Space requirement.1km²/mw max. 25 MW per pixel

13 Distribution of turbine types according to mean wind speeds Latitude -26 Johannesburg -28 Upington Bloemfontein Durban Cape Town -4-6 Port Elizabeth Longitude 1 1 Turbine type no. Polokwane -24

14 One outcome of the study: More than % capacity factor achievable almost everywhere in RSA Achievable average wind capacity factors for for turbine types Latitude Upington Polokwane Johannesburg Bloemfontein Durban Average capacity Load factor Actuals Spain Actuals Germany -2-4 Cape Town Port Elizabeth Longitude

15 Even when placing only high-wind-speed turbine types (1, 2, ) in each pixel shows: more than % capacity factor achievable in wide areas Achievable average wind capacity factors for for turbine types Latitude Upington Polokwane Johannesburg Bloemfontein Durban..2.1 Load factor -2-4 Cape Town Port Elizabeth Longitude

16 On almost 7% of suitable land area in South Africa a 5% capacity factor or higher can be achieved (>5% for turbines 1-) Share of South African land mass less exclusion zones with capacity factors to be reached accordingly 16 Installable capacity [GW] Electricity generated per year [1 TWh] Percentage of South African land mass less exclusion zones Turbine types 1-5 Turbine types Load factor Installing turbine type 4 and 5 will cause higher costs but also increase capacity factors and electricity yield whilst consuming the same area

17 Achievable capacity factors in all turbine categories significantly higher than in leading wind countries Achievable capacity factor distribution per pixel per turbine type.5 # pixels: 51 # pixels: 4 64 # pixels: 2 18 # pixels: # pixels: Total # of pixels: Load factor Spain (installed capacity: 2 GW) Germany (installed capacity 46 GW) Turbine type 17

18 18 A single wind farm changes its power output quickly Simulated wind-speed profile and wind power output for 14 January 212

19 19 Aggregating just 1 wind farms output reduces short-term fluctuations Simulated wind-speed profile and wind power output for 14 January 212

20 2 Aggregating 1 wind farms: 15-min gradients almost zero Simulated wind-speed profile and wind power output for 14 January 212

21 Agenda Renewables in South Africa potential in South Africa Extreme renewables scenarios 21

22 Thought experiment: Build a new power system from scratch Base load: 8 GW Annual demand: 7 TWh/yr (~% of today s South African demand) Questions Technical: Can a wind & solar PV blend, mixed with flexible dispatchable power to fill gaps supply this? Economical: If yes, at what cost? Assumptions/approach 1 16 GW R/kWh (Bid ow 4 average tariff in May-216-Rand) 2 6GW solar R/kWh (Bid ow 4 average tariff in May-216-Rand) 8 GW flexible power generator to fill the 2. R/kWh (e.g. high-priced 11. $/MMBtu) 15-minute solar PV and wind data from recent CSIR study, covering the entire country Check out the results: 15-minute simulation of supply structure for three consecutive years (21-212) 22 Notes: ZAR/USD = 14. Sources: IRP; REIPPPP outcomes; CSIR analysis

23 Thought experiment: assumed 8 GW of true baseload (constant load) GW System Load 6h 12h 18h 24h Hour of the day 2 Sources: CSIR analysis

24 A mix of solar PV, wind and flexible power can supply this baseload demand in the same reliable manner as a base-power generator GW Excess energy Residual Load (flexible power) Solar PV 1 16 GW Residual load supply options: gas, biogas, (pumped) hydro, R/kWh 2 R/kWh Excess solar PV/wind energy curtailment assumed (no 2. R/kWh 8 GW h 6h 12h 18h 24h 24 Sources: CSIR analysis Hour of the day Total installed capacity: GW to supply 8 GW baseload does this make sense? Yes, it s about energy, not capacity!

25 On the lowest-wind day the residual load is large Simulated wind and solar PV power output for a 16 GW wind and 6 GW solar PV fleet on 21 July 211 GW Excess energy Residual Load (flexible power) Solar PV 1 16 GW 2 6GW 8 GW h 6h 12h 18h 24h Hour of the day 25 Sources: CSIR analysis

26 On the lowest-solar-pv day the wind fleet still contributes a lot Simulated wind and solar PV power output for a 16 GW wind and 6 GW solar PV fleet on 21 June 212 GW h Excess energy Residual Load (flexible power) Solar PV GW 6h 6GW 12h 18h 24h 8 GW Hour of the day 26 Sources: CSIR analysis

27 On a high-wind and solar day the amount of excess energy is large Simulated wind and solar PV power output for a 16 GW wind and 6 GW solar PV fleet on October 212 GW Excess energy Residual Load (flexible power) Solar PV 1 16 GW 2 6GW 8 GW h 6h 12h 18h 24h Hour of the day 27 Sources: CSIR analysis

28 During low-wind periods, fuel for flexible generator must be stocked Simulated 15-minute solar PV and wind power supply for the week from July 211 GW Excess energy Residual Load (flexible power) Solar PV Monday Tuesday Wednesday Thursday Friday Saturday Sunday Day of the week 28 Sources: CSIR analysis

29 Technical feasibility in two key dimensions more analyses ongoing Ramping Maximum 15-minute ramp of residual load from 21 to 212:.9 GW/(15-min) 12% of installed flexible capacity of 8 GW per 15-min Minimum 15-minute ramp of residual load from 21 to 212: -1. GW/(15-min) -12% of installed flexible capacity of 8 GW per 15-min Open-Cycle Gas Turbines can ramp up or down with 5-1% output change per minute (Pumped) hydro plants can ramp up and down even faster Plus, a down-ramp of the residual load can always be catered for by short curtailment of wind/pv Fuel-storage The flexible power generator of 8 GW installed capacity requires a fuel-storage capacity of 1 days Eskom currently stocks coal at power stations for more than 5 days on average Buffer capacity of a LNG landing terminal is 4-6 weeks at the minimum 29 Additional analyses required: stable operations of power-electronics-based power systems

30 On average, solar PV and wind supplies 8% of the total demand Average 15-minute solar PV and wind power supply calculated from simulation for years from GW Excess energy Residual Load (flexible power) Solar PV 1 16 GW 2 6GW 8 GW flexible fleet runs at annual average capacity factor of 17% 8 GW h 6h 12h 18h 24h Hour of the day Sources: CSIR analysis

31 Mix of solar PV, wind and expensive flexible power costs 1 R/kWh (excess thrown away) same level as alternative baseload new-builds TWh/yr (15%) Requires ~11-1 PJ/yr of natural gas 2 mmtpa of LNG-based natural gas; roughly what Sasol converts to liquid fuels today New-build baseload coal: R/kWh 48 (68%) 11.5 TWh/yr *.87 R/kWh TWh/yr *.69 R/kWh TWh/yr * 2. R/kWh = 1.1 R/kWh 7.1 TWh/yr 1 System Load Solar PV Excess PV/wind Solar PV 12 (17%) Residual Load (flexible power) Pessimistic assumptions No value given to 6.1 TWh/yr of excess energy (bought and thrown away ) Bid ow 4 costs for PV/wind (no further cost reduction assumed) Very high cost for flexible power of 2. R/kWh assumed Sources: EE Publishers; CSIR analysis

32 Backup The mix of solar PV, wind and a variable power generator would cost R69 billion per year R52 billion fixed cost and R17 billion variable Solar PV and Annual Solar PV tariff payments (fixed): 11.5 TWh/yr *.87 R/kWh = R1. billion/yr Annual wind tariff payments (fixed): 52.6 TWh/yr *.69 R/kWh = R6. billion/yr Flexible power generators Annualised CAPEX and fixed O&M (fixed): R7. billion/yr Fuel cost and variable O&M (variable): 12.1 TWh/yr * 1.4 R/kWh = R17. billion/yr Total Fixed cost: (R1. + R6. + R7.) billion/yr = R5.6 billion/yr Variable cost: R17. billion/yr ======================================================================================= Total R7.6 billion/yr 2 Flexible power 17% capacity factor: assuming.6 R/kWh to be fixed (capital and fixed O&M) and 1.4 R/kWh variable (fuel)

33 1% less load: excess energy increases, need for flexible power reduces Average hourly solar PV and wind power supply calculated from simulation for the entire year GW Excess energy Residual Load (flexible power) Solar PV 1 16 GW 1% reduced demand 8 GW flexible fleet now runs at annual average capacity factor of 12% 2 6GW h 6h 12h 18h 24h Hour of the day 8 GW Sources: CSIR analysis

34 Low sensitivity to demand change (-1%): unit cost goes up by only 2% TWh/yr 1% reduced demand TWh/yr *.87 R/kWh TWh/yr *.69 R/kWh TWh/yr * 2.2.R/kWh = R/kWh TWh/yr 4 System Load Solar PV Excess PV/wind Solar PV Sources: CSIR analysis 8 Residual Load (flexible power) Pessimistic assumptions No value given to 9.1 TWh/yr of excess energy (bought and thrown away ) Bid ow 4 costs for PV/wind (no further cost reduction assumed) Very high cost for flexible power of 2. R/kWh assumed

35 Backup With a 1% reduction in demand, annual costs of power generation go down by R5.6 billion (mainly savings in expensive fuel) Solar PV and Annual Solar PV tariff payments (fixed): 11.5 TWh/yr *.87 R/kWh = R1. billion/yr Annual wind tariff payments (fixed): 52.6 TWh/yr *.69 R/kWh = R6. billion/yr Flexible power generators Annualised CAPEX and fixed O&M (fixed): R7. billion/yr Fuel cost and variable O&M (variable): TWh/yr * 1.4 R/kWh = R billion/yr Total Fixed cost: (R1. + R6. + R7.) billion/yr = R5.6 billion/yr Variable cost: R billion/yr ======================================================================================= Total R billion/yr 5 A 1% reduction in demand reduces total costs by more than 8% unit cost in R/kWh go up only slightly by ~2%

36 Thought experiment: Build a new power system from scratch Load profile: As per South African system load from , scaled to 4 GW peak demand Annual demand: 261 TWh/yr (~1% more than today s South African demand) Questions: Technical: Can a blend of wind and solar PV, mixed with flexible dispatchable power to fill the gaps supply this? Economical: If yes, at what cost? 6 Assumptions/approach 1 65 GW R/kWh (Bid ow 4 average tariff in May-216-Rand) 2 25 GW solar R/kWh (Bid ow 4 average tariff in May-216-Rand) 5 GW flexible power generator to fill the 2. R/kWh (e.g. high-priced 11. $/MMBtu) 15-minute solar PV and wind data from recent CSIR study, covering the entire country Check out the results: 15-minute simulation of supply structure for three entire years (21-212) Notes: ZAR/USD = 14. Sources: IRP; REIPPPP outcomes; CSIR analysis

37 South African actual system load on 1July GW System Load h 6h 12h 18h 24h Hour of the day 7 Sources: CSIR analysis

38 A mix of solar PV, wind and flexible power can supply this load Actual RSA demand and simulated wind/solar PV power output for a 65 GW/25 GW fleet on 1July 21 GW Residual load supply options: gas, biogas, (pumped) hydro, CSP Excess solar PV/wind energy curtailment assumed (no 2. R/kWh GW R/kWh 2 25 R/kWh h 6h 12h 18h 24h 8 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

39 On the lowest-wind day the residual load is large Actual RSA demand and simulated wind/solar PV power output for a 65 GW/25 GW fleet on 21 July 211 GW GW GW 2 25 GW h 6h 12h 18h 24h 9 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

40 On the lowest-solar-pv day the wind fleet still contributes a lot Actual RSA demand and simulated wind/solar PV power output for a 65 GW/25 GW fleet on 21 June 212 GW GW GW 25 GW h 6h 12h 18h 24h 4 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

41 On a high-wind and solar day the amount of excess energy is large Actual RSA demand and simulated wind/solar PV power output for a 65 GW/25 GW fleet on October 212 GW GW GW 25 GW h 6h 12h 18h 24h 41 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

42 During least windy hour, output from 65 GW wind fleet is 1.5 GW Actual RSA demand and simulated wind/solar PV power output for a 65 GW/25 GW fleet on 14 Sept. 211 GW Lowest -year output of wind fleet would have been 1.5 GW around 1h on 14 Sept GW GW 25 GW h 6h 12h 18h 24h 42 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

43 The highest residual load in the three years would have been 4 GW Actual RSA demand and simulated wind/solar PV power output for a 65 GW/25 GW fleet on 22 June 21 GW Highest -year residual load would have been 4 GW around 18h on 22 June GW GW 25 GW h 6h 12h 18h 24h 4 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

44 During low-wind periods, fuel for flexible generator must be stocked Actual RSA demand and simulated 15-minute solar PV/wind power supply for the week from July 211 GW Monday Tuesday Wednesday Thursday Friday Saturday Sunday 44 Sources: CSIR analysis Excess PV/wind Residual Load (flexible power) Solar PV Day of the week

45 On average, solar PV and wind supply 86% of the total system load Average actual RSA demand and average simulated solar PV/wind for years from GW GW flexible power: 12% average annual capacity factor GW GW 2 25 GW h 6h 12h 18h 24h 45 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

46 Mix of solar PV, wind and expensive flexible power costs 1 R/kWh (excess thrown away) much cheaper than mix of base- and mid-merit New baseload coal: R/kWh TWh/yr (16%) 214 New mid-merit coal: R/kWh 18 (7%) 48 TWh/yr *.87 R/kWh TWh/yr *.69 R/kWh + 6 TWh/yr * 2. R/kWh = 1. R/kWh 261 TWh/yr 46 System Load Solar PV Excess PV/wind Solar PV Sources: CSIR analysis 6 (14%) Residual Load (flexible power) No value given to 6 TWh/yr of excess energy (bought and thrown away ) Bid ow 4 costs for PV/wind (no further cost reduction assumed) Very high cost for flexible power of 2. R/kWh assumed

47 1% less load: excess energy increases, need for flexible power reduces Average actual RSA demand less 1% and average simulated solar PV/wind for years from GW % reduced demand 5 GW flexible power: 8% average annual capacity factor GW GW 2 25 GW h 6h 12h 18h 24h 47 Sources: CSIR analysis Excess PV/wind Residual Load Solar PV Hour of the day

48 Low sensitivity to changes in demand (-1%): unit cost increases +4% TWh/yr 1% reduced demand TWh/yr *.87 R/kWh TWh/yr *.69 R/kWh + 624TWh/yr * 2.2.R/kWh = R/kWh TWh/yr 48 System Load Solar PV Excess PV/wind Solar PV Sources: CSIR analysis 24 Residual Load (flexible power) No value given to 5 TWh/yr of excess energy (bought and thrown away ) Bid ow 4 costs for PV/wind (no further cost reduction assumed) Very high cost for flexible power of 2. R/kWh assumed

49 What we have learned from having high-fidelity wind data available Before high-fidelity data collection and after resource in South Africa is not good resource in South Africa is on par with solar There is not enough space in South Africa to supply the country with wind power >8% of the country s land mass has enough wind potential to achieve % capacity factor or more power has very high short-term fluctuations On portfolio level, 15-minute gradients are very low power has no value because it is not always available On average, wind power in South Africa is available 24/7 with higher output in evenings and at night In a mix with cheap solar PV and expensive flexible power it is cheaper than dispatchable alternatives 49 analyses to be continued

50 Ha Khensa Re a leboha Siyathokoza Enkosi Thank you! Re a leboga Ro livhuha Siyabonga Dankie 5