Slide 1

Fakulteit Ingenieurswese Faculty of Engineering CSP energy systems modelling in STERG Paul Gauch SA Energy Modelling Colloquium 31 July 2012

Slide 2

Agenda Introduction to STERG Why we do CSP systems modelling How we do plant and systems modelling What we can do and dont/wont do How we can collaborate 2

Slide 3

STERG INTRODUCTION 3

Slide 4

STERG fits in here 4 Stellenbosch University Engineering Mechanical Engineering STERG NEW: Eskom chair Sasol researcher DST/NRF spoke STERG NEW: Eskom chair Sasol researcher DST/NRF spoke DST/NRF CRSES (Renewable Centre)

Slide 5

STERG research structure 5 STERG Holistic/Multidisciplinary Research Social & Political Sciences Engineering Economic Sciences System R&D (Modelling, Techno-economic, Resources, etc) System R&D (Modelling, Techno-economic, Resources, etc) Component R&D: Eg. Dry Cooling Component R&D: Eg. Thermal Storage Component R&D: Eg. Heliostats, Receivers Solar Resource Measure & R&D SUNSTEL Stellenbosch University Solar Thermal Electricity Project (Primary projects: SUNSPOT, LFR) SUNSTEL Stellenbosch University Solar Thermal Electricity Project (Primary projects: SUNSPOT, LFR) SWH, Process Heat, Desalination etc.

Slide 6

Technology focus areas for R&D (and modelling) 6 11+ Projects from distribution to system to components focused on SUNSPOT

Slide 7

Experimental foundation 7 18 m tower Solar resource station

Slide 8

WHY WE DO CSP SYSTEMS MODELLING 8

Slide 9

SA background SA learned good lessons in last 15 years Struggle to bring IPPs and renewables onto grid Introduced the Integrated resource plan, a robust planning process as law 9 | Basic IRP timeline structure | 20 years 2012 2013 2014 201x Tender yearIRP horizon Tender: 200 MW Total: 1,000 MW Tender: 100 MW? Total: 1,000 MW Tender: 100 MW? Total: >1,000 MW? IRP 2010 IRP 2 IRP 3 On-going CSP Allocation

Slide 10

IRP summary 10 Capacity Electricity produced

Slide 11

Background 11 CSP Status Just entering growth phase of tech lifecycle Largely unknown in SA (no plant experience) ~1% of installed capacity by 2030 (IRP) GAP Sources: Grobbelaar, S., A road map for CSP industry development in South Africa: current policy gaps and recommended next steps for developing a competitive CSP industry, Essay, University of Cambridge, 2011. IRP2010. 2011. Integrated Resource Plan for Electricity 2010-2030. Government Gazette, Republic of South Africa, 6 May, 2011. Winkler (ed) 2007. Long Term Mitigation Scenarios: Technical Report. Prepared by the Energy Research Centre for Department of Environment Affairs and Tourism, Pretoria, October 2007. CSP Need LTMS, IEA, (Eskom) see CSP as foundation post fossil Climate change & fossil resources suggest crisis Large wind and PV allocation in IRP require 100% capacity backup not accounted for

Slide 12

Wind and solar in symphony (Denholm & Mehos - NREL) 12 ? ?

Slide 13

SA background CSP potential has been investigated by Fluri (short term) and Meyer & van Niekerk (longer term) Short term multi-constraint potential (500GWe+) vastly exceeds current or future electricity needs IRP 2010/11 allocates generously to renewables but not CSP we see this as risk for baseload or peaking. This work extends previous work to explore full potential 13

Slide 14

Rutledge coal model Based on Hubbert peak model finite resources follow a normal distribution production curve. It works very well. Would have forecast British coal depletion to within months 100 years earlier. 14

Slide 15

South African coal 15

Slide 16

South African coal 16 Source Peak year (and peak production) 90% year (and/or total cumulative extraction) Mohr & Evans (2009)2012 (258 Mt/y)18.6 Gt Rutledge (2011) Similar to others but prefers not to comment due to peak year volatility 2048 (18 Gt) Patzek & Croft (2010)2007 (478.6 EJ calculated as 17.15 Gt) Hartnady (2010) & (2012) 2020 (284 Mt) 2012/2013 (254.3 Mt/yr) 23 Gt 18.675 Gt What are these models saying? Peak coal: Now 2020 Then its downhill to about mid century What are these models saying? Peak coal: Now 2020 Then its downhill to about mid century

Slide 17

Other resources (worldwide) 17 Conventional uranium: ~2065 Other conventional and unconventional fuels also limited Conventional uranium: ~2065 Other conventional and unconventional fuels also limited

Slide 18

Wind, water and solar Note: 2030 IRP annual power need =~ 500 TWh The wind resource is about 80 TWh Hydro is not a major source in SA Wave and ocean current is for the future Solar resource is immense and vastly exceeds future needs Both are intermittent and a problem This concludes the major energy sources 18

Slide 19

Making sense of it all 19 2030 energy needs ~500 TWh Coal 300 TWh Nuclear 77 TWh CCGT 10 Hydro 15 OCGT 10 PV 900 Wind 80 CSP no storage 900 CSP w Storage 900 CSP Future >> 900

Slide 20

HOW WE DO CSP SYSTEMS MODELLING 20

Slide 21

Introduction Dispatchability = storage + low inertia = CSP value prop 20 MWe Gemasolar plant demonstrated 24h full load 21

Slide 22

Method: Plant Based on the Gemasolar plant Approximated optical performance + Chambadal- Novikov engine (modified Carnot) + inertia capacitance + storage capacitance Model validated using eSolar measured data (Gauch et al. SolarPACES 2011) NREL predicted annual electricity generation for this plant (110 vs. 115 GWh/yr) 22 ItemValue Country, RegionSpain, Seville Andaluca Location 3733 44.95 North, 519 49.39 West Land area195 Ha Solar resource2,172 kWh/m 2 /yr Electricity Generation110 GWh/yr (planned) Cost230,000,000 Euro O&M jobs45 Heliostat aperture area304,750 m 2 Number of heliostats2,650 Heliostat size120 m 2 Tower height140 m Heat transfer fluidMolten salt Receiver outlet / inlet temperature 565 C / 290 C Turbine capacity (gross)19.9MWe CoolingWet Storage2 tank, 15 hours

Slide 23

Method: Plant Heliostat field Receiver balance Inertia & storage model Heat engine 23

Slide 24

Method: Spatial solar and weather data Plant model only requires 3 parameters for each hour for dry cooled plant (DNI, Tamb, wind) Grid of points for all South Africa: 0.375 increments latitude and longitude 823 points in the boundaries of SA 24

Slide 25

25 Johannesburg Pretoria Bloemfontein Cape Town Durban 823 Grid points (uniform / unbiased)

Slide 26

Method: Spatial solar and weather data Plant model only requires 3 parameters for each hour for dry cooled plant (DNI, Tamb, wind) Grid of points for all South Africa: 0.375 increments latitude and longitude 823 points in the boundaries of SA Helioclim-3 data derived from Meteosat Real 2005 data (not TMY) Point validation of wind and ambient temperature using SA weather data. Sensitivity analysis to DNI, Tamb, wind showed strong sensitivity to DNI and very weak sensitivity to wind and Tamb. Helioclim DNI data has issues. The method is still demonstrable. 26

Slide 27

Method: The spatial analysis 823 grid points * 3 parameters * 8760 hours = 21.6 million inputs 1 output parameter (power) = 7.2 million outputs Proxy for testing dispatchability Run plant as-is (generates power when it can) Half size power block (emulates half the 823 plants attempting to run at any 1 time) Quarter size power block (emulates quarter of 823 plants attempting to run at any 1 time) Some other combinations were tried 27

Slide 28

Results and analysis: Time plots 28 8 January days 8 June days

Slide 29

Results and analysis: Time plots 29 1 out of 4 plants running at a time practically demonstrates baseload Data anomaly

Slide 30

Results and analysis: Spatial 30

Slide 31

What we can do and dont/wont do STERG centric (CRSES to some degree, but not SI) Can do in future Through partnerships: Real and TMY solar, wind* and weather data multi year CSP, PV & wind spatial and time modelling GIS modelling for multi-criteria spatial type analysis Develop and improve underlying technology models Dont / wont do (as far as I can tell) ERC-like TIMES modelling (stochastic, complex multi-criteria systems considerations) Climate and climate change models Anything in the policy or social space 31

Slide 32

Areas for collaboration Collective database of Discount rate sets for RE technologies (scenarios) Capacity and capacity factor scenario sets for all options Technology models Conventional resource estimation scenarios (fossil and fissile) Common solar, wind and weather data sets (real and TMY) Demand profiles at least to hourly demand (historical and forecast) Other For IRP Set of assumptions on demand per year and finer resolution Recognition of non electric energy needs that transition to electricity particularly transport Other 32

Slide 33

33 Thank you!