Transmission Planning Under Uncertainty: A Stochastic Two ...hobbsgroup.johnshopkins.edu/docs/papers/OH_FERC_TransmissionUn… · Transmission Planning Under Uncertainty: A Stochastic
Post on 28-Mar-2018
217 Views
Preview:
Transcript
Transmission Planning Under Uncertainty: A Stochastic Two Stage Modelling ApproachA Stochastic Two‐Stage Modelling Approach
H d W ijdHarry van der WeijdeVU Amsterdam & EPRG, University of Cambridge | hweijde@feweb.vu.nl
Benjamin F HobbsBenjamin F. Hobbs Johns Hopkins University, EPRG, & CAISO| bhobbs@jhu.edu
The Next Generation of Electricity Planning ModelsThe Next Generation of Electricity Planning ModelsFERC, Nov. 4, 2010
Making networks fit for renewables …www.eprg.group.cam.ac.uk
Overview
The problemThe problem
Existing studies
O d lOur model
– How it works
– Data it needs
– Data sources + assumptions
Some results
Our conclusions
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
Our conclusions
2
D ti h i !
The Problem: The Problem: HyperuncertaintyHyperuncertainty!!What’s a Poor Transmission Planner to do?What’s a Poor Transmission Planner to do?• Dramatic changes a‐coming!
• RenewablesH h?– How much?
– Where?
– What type?yp
• Other generation– Centralized? Do these uncertainties – Distributed?
• Demand
o ese u ce a eshave implications fortransmission investments now?
– New uses? (EVs)
– Controllability?
• PolicyMaking networks fit for renewables …
www.eprg.group.cam.ac.uk
• Policy
The problem, Cont.The problem, Cont.
Transmission planning– Generators respond: multi‐level – Decisions can be postponed: multi‐stage– Uncertainties & variability: stochastic
Important questions:– Optimal strategy under uncertainty?– Value of information? (EVPI)– Cost of ignoring uncertainty? (ECIU)– Option value of being able to postpone?p g p p
Deterministic planning can’t answer these!
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
• Stochastic can!
4
Decision making uDecision making undernder uncertaintyuncertainty-----------Previous Work-----------
Real options analysis of single lines, usually
Single-stage trans-mission planning under uncertainty
Two-stage trans-mission planning
----------- Ours-----------
based on exogenous price processes (Hedman et
al. 2005; London Economics 2003; Fleten et al. 2009; Parail 2009)
under uncertainty with generator response (Awad et al. 2009;
Crousillat et al. 1993; De la Torre et al. 1999; Oolomi Buygi et al. 2004; Oliveira et al. 2007; Hyung Roh et al 2009; Sauma & Oren 2009)
mission planning under uncertainty with generator response
Hyung Roh et al. 2009; Sauma & Oren 2009)
Invest inline now?
Uncertainprices
(Some:Invest in
line later?)
Investtrans.now
Uncertainties(usually load)
Gen.operation
(& sometimes
Invest/ operatetrans. /
gen
Investtrans./gener.
Uncertainties(policy,
load,
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
line later?) (&, sometimes, Investment)
gen.later
gnow
load,technology)
Our model: timelineStage 1 Stage 2
2010 2020 2030
3. Dispatch 6. Dispatch
1. Transmissioninvestment
4. Transmissioninvestment
2. Generationinvestment
5. Generationinvestment
Objective: min total costs (investment + generation) s.t. power flow constraints, wind availability, build limits,
bl t t
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
renewables targets
6
Mathematical Schematic– Math programming with recoursep g g
• scenarios s=1,2,..,S, each with probability PRs
– Simplest: Assume 2 decision stages:Simplest: Assume 2 decision stages:1. Choices made “here and now” before future is
known – E.g., investments in 2010g ,– These are x1
2. “Wait and see” choices, which are made after the future s is known.
d h/– E.g., dispatch/operations, investments in 2020– These are x2s (one set defined for each scenario s)
– Model:– Model:MIN C1(x1) + Σs PR
s C2s(x2s)s.t. A1(x1) = B1
A2s(x1 x2s) = B2s ∀s
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
A2s(x1, x2s) = B2s ∀s
Some assumptions
Alignment of generation and transmission objectivesAlignment of generation and transmission objectives
– e.g., nodal pricing + perfect competition
Generation
– Constant variable costs
– No start‐up costs, min run levels, ‘lumpy investment’
– No ramping constraints
Demand:
– No short‐term demand flexibility, demand‐side management
R bl t t t i t ffi i tMaking networks fit for renewables …
www.eprg.group.cam.ac.uk
Renewables targets met in most efficient way8
Data necessary
regions+ transmission
scenarios(2020 2030) &
generator types + current capacities + maximum+ transmission
constraints+ losses
wind output and demand
(2020, 2030) & probabilities:
generation costs (i l b i )
capacities + maximum build limits + costs
time series (1 year)+ interconnector flows
(incl. carbon price), transmission
investment costs,demand,
renewable targets,nuclear feasibilitynuclear feasibility
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
investment alternatives9
Data sources
Regional wind output: Neuhoff et al. (2007)
Hydro output: Duncan (2010)Hydro output: Duncan (2010)
Regional demand data: National Grid
BritNed Flows: Parail (2010)BritNed Flows: Parail (2010)
Maximum build limits: Various
Regions + trans. constraints: NG 7‐year statement (2009)g y ( )
Transmission losses: own calculations
Investment alternatives + costs: KEMA (2009)
Generation costs: NEA and IEA (2005), US DOE, own calculations
S i V i (Di LENS R d i t t )
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
Scenarios: Various (Discovery, LENS, Redpoint, etc.)
10
Alternatives (overnight construction cost)
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
11
ScenariosGen. inv. cost Var. gen cost Trans. inv.
cost
Demand CO2
price
Others
Status Quo CCGT/OCGT/DG: + + +/‐ No RT
Low cost DG DG: ‐‐ CCGT/OCGT: ‐
DG: ‐‐
+ ++ RT: +
Nuclear replacement onlyDG: ‐‐ Nuclear replacement only
Low Cost
Large Scale
G
Renewables : ‐‐ CCGT/OCGT/DG: ++ ‐‐ +++ RT: +++
Green
Low Cost
Conventional
Conventional: ‐ CCGT/OCGT/DG: ‐ ++ + No RT
Paralysis All except
offshore: +++
CCGT/OCGT/DG: + Onshore: +++
Others +
++ ++ RT: +
Nuclear replacement only
Techno+ All : ‐ CCGT/OCGT/DG: + ‐ ++ ++ RT: ++
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
12
Some results
Disclaimer: the following results areDisclaimer: the following results are preliminary and based on restrictive
assumptionsassumptions.
They cannot be used to evaluate proposed transmission investmentstransmission investments.
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
13
Optimal stochastic solution
Onshore CCGTwind
Off h
CCGT
OCGTOffshore wind
OCGT
Nuclear
BiomassBiomass
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
14
Cf. Traditional robustness analysis
2020 Installations by Scenario “Robust”
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
15
Value of perfect information
How much average savings if we knew which scenario would happen?scenario would happen?
1. Solve stochastic model
2 Solve deterministic model for each scenario2. Solve deterministic model for each scenario
3. Compare objectives (1) and (2)
• Results:
– For gen & transmission: £3,729M (3%)
– For trans alone: £101M (0.1%)
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
16
Cost of ignoring uncertainty
How much would costs go up if we naively plan for one scenario but other scenarios can happen?
1.Solve stochastic model
2.Solve naïve (deterministic) model for each2.Solve naïve (deterministic) model for each scenario
3 Solve stochastic model imposing first‐stage3.Solve stochastic model, imposing first‐stage transmission decisions from step 1
4 C bj ti (1) d (3)Making networks fit for renewables …
www.eprg.group.cam.ac.uk
4.Compare objectives (1) and (3)17
Cost of ignoring uncertainty
Scenario planned for ECIU (Transmission)(Present worth)(Present worth)
Status Quo £432MLow Cost DG £0 Low Cost Large Scale Green £29MLow Cost Conventional £196M
lParalysis £221M Techno+ £0 Average £146M = 0 12% ofAverage £146M = 0.12% of
expected costs (stochastic solution)
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
18
Option value of waiting
How much would costs go up if we had toHow much would costs go up if we had to make all decisions now?
1 Solve stochastic model1.Solve stochastic model
2.Solve stochastic model, imposing same i i i l f ll itransmission expansion plan for all scenarios
3.Compare objectives (1) and (2)
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
19
Option value of waiting
2020 2020
Example: Paralysis
SCO
2020stoc
SCO
no option
UNO UNO
p
NOR
MID
NOR
MIDMID
CEN
SWE
ESTCEN
SWE
EST
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
20
SWE
Option value of waiting
Option value (transmission only):Option value (transmission only):
= £102M present worth= 0.08% of total costs (stochastic)(stochastic)
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
21
Conclusions
For transmission planning:
I i i k h ifi bl i– Ignoring risk has quantifiable economic consequences
Option al es can be significant– Option values can be significant
– Approach useful for policy/planning questions
kFuture work
– Demand response
– Bi‐level formulation
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
22
References E. O Crousillat, P. Dörfner, P. Alvarado, and H. M. Merrill, “Conflicting Objectives and Risk in Power System Planning,” IEEE Trans. Power Systems, vol. 8, pp. 887‐893, 1993.,N. Duncan, “”,2010.S. ‐E. Fleten, A. M. Heggedal, and A. Siddiqui, “Transmission Investment under Uncertainty: The Case of Germany‐Norway,” presented at the 1st International Ruhr Energy Conference, Essen, Germany.gy , , yK. W. Hedman, F. Gao, and G. B. Sheble, “Overview of Transmission Expansion Planning Using Real Options Analysis,” in Proc. IEEE North American Power Symposium, 2005.J. Hyung Roh, M. Shahidehpour, and L. Wu, “Market‐Based Generation and y g , p , ,Transmission Planning With Uncertainties,” IEEE Trans. Power Systems vol. 24, pp. 1587‐1598, 2009.KEMA “Assessment of overall robustness of the transmission investment proposed for additional funding by the three GB Electricity Transmission Owners”, 2009.London Economics, London, “Economic Evaluation of the Path 15 and Path 26 Transmission Expansion Projects in California”.National Grid, “Seven‐Year Statement”, 2009.
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
23
References (cont’d)NEA and IEA, “Projected Costs of Generating Electricity – 2005 Update”, Nuclear Energy Agency and International Agency, OECD, Paris, France, 2005. K Neuhoff J Cust L Butler K Keats H Hoexter A Kreckzo G Sinden and AK. Neuhoff, J. Cust, L. Butler, K. Keats, H. Hoexter, A. Kreckzo, G. Sinden, and A. Ehrenmann, “Space and Time: Wind in an Investment Planning Model”. EPRG Working Papers 0603, 2006.G. C. Oliveira, S. Binato, and M. W. Pereira, “Value‐Based Transmission Expansion Planning of Hydrothermal Systems Under Uncertainty,” IEEE Trans. Power Systems, g y y y, y ,vol. 22, pp. 1429‐1435, 2007.M. Oloomi Buygi, M. Shahidehpour, H. M. Shanechi, and G. Balzer, “Market Based Transmission Planning Under Uncertainties,” Proc. 2004 Int. Conf. on Probabilistic Methods Applied to Power Systems, pp. 563‐568.V. Parail, “Can Merchant Interconnectors Deliver Lower and More Stable Prices? The Case of NorNed,” EPRG Working Papers 0926, Nov. 2009.V. Parail, “Properties of Electricity Prices and the Drivers of Interconnector Revenue”, 2010.E. E. Sauma, and S. S. Oren, “Proactive Planning and Valuation of Transmission Investments in Restructured Electricity Markets,” Journal of Regulatory Economics 30, pp. 261‐290, 2006.
Making networks fit for renewables …
www.eprg.group.cam.ac.uk
24
top related