www.bartlett.ucl.ac.uk/energy
UCL Energy Institute
Projected inflow compared to historic Capacity factors
Assessing uncertainty of climate change impacts on long-term hydropower generation using the CMIP5 ensemblePablo E. Carvajal11 UCL Energy Institute, University College London, [email protected]
INTRODUCTION• Hydropower is vulnerable to long-term climate change. The Coupled
Model Intercomparison Project (CMIP5) projects large uncertainty forprecipitation.
• Large hydro suffers of significant cost and time overruns.• Long-term energy system optimisation models usually consider constant
hydro-climatic conditions and only minimise cost, not risk.
CASE STUDY - ECUADOR• Hydro share in electricity generation was 75% in 2017 – 4368MW• Six large river basins with large hydro and still untapped potential.
RESULTS
CONCLUSIONS• Power expansion plans (and energy system optimisation models) should
consider minimum cost, minimum environmental impact and also minimum risk.
• The cheapest (minimal cost) portfolio might entail large uncertainties and price volatilities.
• There are still large discrepancies and uncertainties for future hydropower production under climate change.
REFERENCESAwerbuch (2006) Portfolio-based electricity generation planning: Policy
implications for renewables and energy security. Mitigation Adaption Strategyto Glob Chang 11:693–710.
Nijs & Poncelet (2016) Integrating recurring uncertainties in ETSAP energysystem models, VITO NV for ETSAP
METHOD INTEGRATED HYDRO - POWER – RISK MODEL
6
2
4
5
1
3
Hydropower stations Gauging stations River network
Pacific Amazon
Esmeraldas Guayas Jubones Santiago Pastaza Napo
Andes Mountain Range
COLOMBIA
PERU
PACIFIC OCEAN
Toachi - A.J Pilatón
Toachi 205.4MW Pilatón 48.9MW
Mazar 170MW
Molino 1100MW
Sopladora 487MW
Coca Codo Sinclair 1500MW
Marcel Laniado 213MW
Minas San Francisco 275MW
Daule
Paute - D.J. Palmira
Jubones - D.J.San Francisco
San Francisco 230MW
Agoyán 156MW
Pastaza - Baños
Coca - San Rafael
kilometers
123456
Basins
Watersheds
0 520260130
Exploring hydropower’s largest sources of uncertainty: impacts of climate change and capital cost overruns, and how to assess them in long-term energy system models.
Runoff projections CMIP5 – RCP4.5* (2071-2100)• Large discrepancy among climate models for precipitation• Models reflect seasonal changes0
50
100
150
J F M A M J J A S O N D
Inflo
w (m
3/s)
Toachi Pilatón (1)
0
500
1000
J F M A M J J A S O N D
Marcel Laniado (2)
0
100
200
300
J F M A M J J A S O N D
Minas San Francisco (3)
0
200
400
J F M A M J J A S O N D
Inflo
w (m
3/s)
Paute (4)
0
100
200
300
400
J F M A M J J A S O N D
Agoyán (5)
0
250
500
750
1000
J F M A M J J A S O N D
Coca Codo Sinclair (6)
CMIP5 Ensemble mean Historic Individual GCMs
CMIP5 Ensemble mean ± SD
●
●
●●
●
●
●
●●
●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Capa
city
fact
or C
F
Toachi Pilatón (1)
●
● ● ●
●
●
●●
● ●
●●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Marcel Laniado (2)
●
●
● ●
●
● ●
●
●●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Minas San Francisco (3)
●
●
●
●●
● ●●
●
●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Capa
city
fact
or C
F
Paute (4)
●
●
●
●● ● ● ●
●
●
● ●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Agoyán (5)
●
●
●
●
●
● ●
●
●
●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Coca Codo Sinclair (6)
● CMIP5 Ensemble mean Historic
0
50
100
150
J F M A M J J A S O N D
Inflo
w (m
3/s)
Toachi Pilatón (1)
0
500
1000
J F M A M J J A S O N D
Marcel Laniado (2)
0
100
200
300
J F M A M J J A S O N D
Minas San Francisco (3)
0
200
400
J F M A M J J A S O N D
Inflo
w (m
3/s)
Paute (4)
0
100
200
300
400
J F M A M J J A S O N D
Agoyán (5)
0
250
500
750
1000
J F M A M J J A S O N D
Coca Codo Sinclair (6)
CMIP5 Ensemble mean Historic Individual GCMs
CMIP5 Ensemble mean ± SD
●
●
●●
●
●
●
●●
●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Capa
city
fact
or C
F
Toachi Pilatón (1)
●
● ● ●
●
●
●●
● ●
●●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Marcel Laniado (2)
●
●
● ●
●
● ●
●
●●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Minas San Francisco (3)
●
●
●
●●
● ●●
●
●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N DCa
pacit
y fa
ctor
CF
Paute (4)
●
●
●
●● ● ● ●
●
●
● ●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Agoyán (5)
●
●
●
●
●
● ●
●
●
●
●
●
0.00
0.25
0.50
0.75
1.00
J F M A M J J A S O N D
Coca Codo Sinclair (6)
● CMIP5 Ensemble mean Historic
0
1000
2000
3000
J F M A M J J A S O N D
Powe
r gen
erat
ion
(GW
h)
+1SD
0
1000
2000
3000
J F M A M J J A S O N D
Mean
0
1000
2000
3000
J F M A M J J A S O N D
−1SD
Toachi Pilatón (1)Marcel Laniado (2)Minas San Francisco (3)Paute (4)Agoyán (5)Coca Codo Sinclair (6)
HYDROLOGICAL MODEL• 40 GCMs RCP4.5 – 2001-2100 – monthly level – 0.5°grid res.• Precipitation -> Runoff -> Availability factor for hydro (%)
TIMES ENERGY SYSTEM MODELObjective function
Minimising costs and risks simultaneouslyDetail• Horizon 2060• 32 Time slices (12 months x 3 inter day periods)• Plant by plant (125 existing plants) and 15 new technologies.• Run of river and reservoir hydro by river basins and discrete sizes.
RISK MODEL Upper Absolute Deviation
𝑀𝑖𝑛(𝐸 𝑐𝑜𝑠𝑡 + 𝛾×𝑈𝑝𝐴𝑏𝑠𝐷𝑒𝑣(𝑐𝑜𝑠𝑡)𝑈𝑝𝐴𝑏𝑠𝐷𝑒𝑣 𝑐𝑜𝑠𝑡 =7 𝑝8× 𝑐𝑜𝑠𝑡8 − 𝐸 𝑐𝑜𝑠𝑡 :�
8• The UpAbsDev computes the average value of the positive total cost
deviations for all-states-of-the-world j (with probability pj).• Applied for fuel prices and capital cost of new technologies
*CMIP5 – Coupled Model Intercomparison Project IPCC AR5RCP 4.5 Representative Concentration Pathway
Available energy from hydropower system in different basins• Significant changes in expected hydropower production
Installed capacity (MW)
Observed generation (GWh/y)
Simulated annual generation (GWh/y)-1SD Ensemble mean +1SD
4368 20822 -55%↓ 6%↑ 39%↑
Next steps• Consider changes in long-
term hydro in TIMES for the whole Ecuadorian power system
• Include risk and derive a trade-off between system cost and risk
Cost – risk trade off curve (Pareto optimal)
0
50
100
150
J F M A M J J A S O N D
Inflo
w (m
3/s)
Toachi Pilatón (1)
0
500
1000
J F M A M J J A S O N D
Marcel Laniado (2)
0
100
200
300
J F M A M J J A S O N D
Minas San Francisco (3)
0
200
400
J F M A M J J A S O N D
Inflo
w (m
3/s)
Paute (4)
0
100
200
300
400
J F M A M J J A S O N D
Agoyán (5)
0
250
500
750
1000
J F M A M J J A S O N D
Coca Codo Sinclair (6)
CMIP5 Ensemble mean Historic Individual GCMs
CMIP5 Ensemble mean ± SD
