Transcript Work on Dealing with Uncertainty

Dealing with Uncertainty in
Energy Systems Models
Overview
• Intro: SATIM
• UNEP Project – SATIM-MC
• MAPS Project – SATIM-SP
ERC’s Bread n Butter Model:
SATIM (South African TIMES Model)
• Deterministic Least Cost Planning Model
(Similar to Model used for IRP/IEP)
• Uncertainty affecting operation/short-term decisions (dispatch):
– Unpredictable hourly fluctuation in wind regime, load
– Unpredictable chances of a large thermal unit breaking down
– Dealt with using outside model (LOLP calculator/Dispatch Model)
• Uncertainty affecting medium to long term decisions (investment):
– Demand: Economic Growth, technology and fuel costs, Behaviour
– Supply: Technology and fuel costs
– This is normally dealt with using scenarios
UNEP: SATIM-MC
Projecting South African CO2 emissions to 2050
The Model:
Demographics
Economy
SATIM energy model
Least cost energy
mix
Fuels
Technology
GHG Emissions
Projection
UNCERTAINTY
Monte Carlo
algorithm
Expert Elicitation
Combined Elicited Distributions
SATIM
Resulting GHG Emissions Projection
+ Full Story
Some of the Distributions:
Global Prices – Results of 108 runs of Imaclim-W
Without Mitigation
2055
2055
2055
2050
2050
2050
2045
2045
2045
2040
2040
2040
2035
2035
2035
2030
2030
2030
2025
2025
2025
2020
2020
2020
2015
2015
2015
0
50
100
150
200
250
5
Coal 2010 $/ton
10
15
20
50
25
100
150
200
250
Oil 2010 $/bbl
Gas 2010 $/MMBtu
2055
2055
2060
2050
2050
2055
2045
2045
2040
2040
2035
2035
2030
2030
2025
2025
2020
2020
2050
2045
2040
2035
2015
2030
2025
2020
2015
0
50
100
150
200
250
2015
5
10
15
20
25
With Mitigation (2 deg)
50
100
150
200
250
Other Distributions:
GDP/Coal price From Expert Elicitations
7%
5%
4%
3%
1%
0%
-1%
R/ton
2%
2007
2010
2013
2016
2019
2022
2025
2028
2031
2034
2037
2040
2043
2046
2049
GDP growth
6%
-2%
• GDP Growth for 10 Samples
Cumul. Mt
• Coal Price Supply Curves
Range
Stochastic TIMES
Analysis of hedging strategies
Stochastic TIMES (cont.)
• TIMES offers the possibility of doing stochastic programming with
recourse on the following parameters:
– Capacity limits (which can be used to allow/disallow techs/fuels with
different costs/prices)
– Cumulative limits on flows (reserves)
– Seasonal availabilities (useful for hydro: dry year)
– Damage Costs of emissions
– Demand Projections (growth)
• Can be used to construct up to 5 stages, with a large number of
states of the world
• Objective function can also be altered:
– Linearised expected utility criterion (where risk/variance is added to
the cost)
– MiniMax – least regret (Savage criterion, when likelihoods are not well
known)
MAPS: SATIM-SP
• Use some of the distribution data from the UNEP
project to analyse some hedging strategies:
– Nuclear Programme, given uncertainty about:
• Growth
• Gas Price
• Nuclear costs
– Other mitigation policies/targets, given uncertainty
about:
• Uncertainties listed above
+
• Damage costs?
• Global CO2 prices?
IRP Update Experiment
• Start with Big Gas scenario (Cheap gas – No Nuclear)
• Gas Price starts at reference step 5 (83 2010 R/GJ dropping to
45 R/GJ by 2035)
Gas Price Stays High
2006
Gas Price Drops
2040
2030
Prelim Results
Probability of Low Price Gas
Installed Nuclear Capacity 2030
0%
4.8 GW
25%
3.36 GW
50%
2.7 GW
75%
0.3 GW (~0 GW)
100%
0