U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change...

15
U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand Modeling and Capacity Planning USAEE/IAEE North American Conference, Washington DC Nidhi R. Santen, Massachusetts Institute of Technology ([email protected] ) Mort D. Webster, Massachusetts Institute of Technology David Popp, Syracuse University/National Bureau of Economic Research

Transcript of U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change...

Page 1: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change

Tuesday, October 11, 2011Session 31: Electricity Demand Modeling and Capacity Planning

USAEE/IAEE North American Conference, Washington DC

Nidhi R. Santen, Massachusetts Institute of Technology ([email protected])Mort D. Webster, Massachusetts Institute of Technology

David Popp, Syracuse University/National Bureau of Economic Research

Page 2: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

2

Introduction (1 of 2)

EIA, AER 2009; EIA 2011

Page 3: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

3

Introduction (2 of 2)

Power System Technology R&D

(Public and Private)

Government

Makes Environmental Policies

Electric Utilities

Build Power Plants Using Available Generation Technologies

NaturalEnvironment

1. Constraining Regulations2. Production Support

Direct R&D Support

New or Improved

Generation Technologies

Increased Demand for

Technologies

CO2 Emissions

Two main policy pathways to reduce cumulative power sector emissions

“Now v. Later”“Adoption v. Innovation”

Page 4: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

4

Research Question and Outline

Research Question:What is the socially optimal balance of inter-temporal regulatory policy andtechnology-specific R&D expenditures for the U.S. electricity generation sector, given aspecific cumulative climate target?”

Outline for Today’s Presentation1. Overview of existing electricity sector planning models’ capabilities2. Introduction of the current modeling framework3. Snapshots from first results4. Future research5. Summary

Page 5: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

5

1. Overview of Existing Numerical Power Generation Expansion Models (1 of 2)

Top-Down v. Bottom-up Models• Top-Down: Use Average Costs and Assume Capacity Factors• Bottom-Up: Use Specific Costs (e.g., Capital, O&M, Fuel) and Solve for Capacity Factors

Rigorously studying emissions potentials from the power sector requires modeling operational details of the physical system (more easily resolved in bottom-up models).

Including Operational

Realism Matters!

Resu

lts

Pre

vie

w –

Less

Deta

il

Resu

lts

Pre

vie

w –

More

Deta

il

Page 6: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

6

1. Overview of Existing Numerical Power Generation Expansion Models (2 of 2)

Common Methods to Model Technology Change and Learning Dynamics

Decision Variables: Capacity Additions

1. (Exogenous) Fixed Trend:CapCostt,g = CapCostt-1,g*(1+ α)

2. (Endogenous) Learning-by-Doing:CapCostt,g = InitialCapCostg / (CapitalStockt,g)LBDCoeff

Decision Variables: Capacity Additions + R&D Investments

3. (Endogenous) Learning-by-Searching:CapCostt,g = InitialCapCostg / (KnowledgeStockt,g)LBSCoeff

KnowledgeStockt,g = δΣ1:t-1R&D$t,g + R&D$t,g

4. 2-Factor Learning Curves (2FLC):CapCostt,g = InitialCapCostg / [(CapitalStockt,g)LBDCoeff2 *

(KnowledgeStockt,g)LBSCoeff2]

KnowledgeStockt,g = δΣ1:t-1R&D$t,g + R&D$t,g

Numerical Models

Representation of Technology Improvement

Fixed Time Trends (Exogenous)

Learning by Doing

(Endogenous)

Learning by Researching

(Endogenous)

NREL ReEDS X X

Gen Star Lite (Ramos et.

al./IIT)X

SMART (Powell et. al) X

MIT EPPA (Electricity

Sector)X (Modified)

IIASA ERIS (Electricity

Sector)X (Limited)

EPA/MIT MARKAL

(Electricity)X

Page 7: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

7

Knowledge Stock (H)

R&D$

New Knowledge (h)

Generation Planning Inputs

Generation Technology

Costs ($/MWh)

Electricity Demand

(MW/time)

Generation Technology Availability

(Year)

Learning by Experience

Technology Change Module

“Innovation Possibilities

Frontier”

ht = aRD$bHΦ

Environmental Policy

New Power Plant Additions

(GW)

Production (GWh)

Learning by Researching

2. Modeling Framework for this Research

Generation Planning Model

CO2 Emissions (Million Metric Tons)

Generation Planning Model

Ht,g = (1-δ)Ht-1,g + ht,g

Page 8: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

8

2. Modeling Framework for this Research

Structural Details

• Centralized, social planning (decision-support model)• Representative technologies of the U.S. system• Representative U.S. load duration curve• 50-year planning horizon, 10-year time steps

• Objective

• Decision Variables (per period)(1) R&D $ (by Technology)(2) Carbon Cap(3) Generation Expansion(4) Generation Operation

• Key Constraints(1) All traditional generation expansion constraints (e.g., demand balance,

reliability, non-cycling nuclear technology, etc.)(2) Cumulative carbon cap(3) Cumulative R&D funding account balance

Generation Technologies

CoalSteam Gas

Wind

Advanced CoalGas CCNuclear

Solar

Coal w CCSGas CTHydroOther

Page 9: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

9

3. First Results: With and Without Learning-by-Searching (under a Medium Cumulative Emissions Target)

No LBS

With LBS (NPVLBS < NPVNoLBS)

Page 10: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

10

3. First Results: Medium v. Strong Cumulative Emissions Target

Medium Target

Strong Target

Page 11: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

11

3. First Results: Sensitivity of Innovation Possibilities Parameters (Strong CCS Possibilities under a Medium Emissions Target)

Base Case Innovation Possibilities

Strong CCS Innovation Possibilities

Page 12: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

12

4. Future Research

Model optimal generation (carbon cap distribution) and R&D investment decisions under multiple uncertain innovation possibilities using stochastic dynamic programming

Page 13: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

13

Summary

• Studying how to balance regulatory efforts and R&D efforts for the electricity generation sector requires a decision model where the capital costs of technology change endogenously with respect to new builds (adoption) and new research (innovation)

• Rigorous study of emissions management from the power sector requires operational details of the physical system, embodied within bottom-up type models.

• Results confirm both a “tradeoff” and “interaction” between adoption v. innovation for technologies with strong learning potentials (dynamics that are popular in the theoretical literature)

• More research needs to be done to 1) understand the sensitivity of innovation parameters on decisions, 2) compare these results with more traditional knowledge stock formulations, and 3) model the effect of uncertainty of returns to research on near-term regulatory and R&D decisions.

Page 14: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

Thank You

14

Source: US EPA E-Grid Database & NPR.org

Page 15: U.S. Electric Power Generation Planning under Endogenous Learning-by-Searching Technology Change Tuesday, October 11, 2011 Session 31: Electricity Demand.

Barreto, L. and S. Kypreos. (2004). “Endogenizing R&D and market experience in the "bottom-up" energy-systems ERIS model,” Technovation, 24(8):615-629.

Fischer, C. and R. G. Newell. (2008). “Environmental and technology policies for climate mitigation.” Energy Economics 55: 142-162.

Hobbs, B. F. (1995). “Optimization methods for electric utility resource planning.” European Journal of Operational Research 83:1-20.

Ibenholt, K. (2002). “Explaining learning curves for wind power,” Energy Policy 30: 1181-1189. Jaffe, A., and M. Trajtenberg. (2002). Patents, citations, & innovations: a window on the knowledge economy. MIT Press:

Cambridge, MA, 478pp. Johnstone, N., Hascic, I, and D. Popp. (2010). “Renewable Energy Policies and Technological Innovation: Evidence Based

on Patent Counts,” Environmental Resource Econ, 45: 133-155. Messner, S. (1997). “Endogenized technological learning in an energy systems model,” J Evol Econ 7: 291-313. Miketa, A. and L. Schrattenholzer. (2004). “Experiments with a methodology to model the role of R&D expenditures in

energy technology learning processes.” Energy Policy, 32(15):1679-1692. Popp, D. (2002). “Induced Innovation and Energy Prices.” American Economic Review 92(1): 160-180. Popp, D. (2006). “ENTICE-BR: Backstop Technology in the ENTICE Model of Climate Change.” Energy Economics 28(2):

188-222. Popp, D. (2006b). “They Don't Invent Them Like They Used To: An Examination of Energy Patent Citations Over Time.”

Economics of Innovation and New Technology 15(8): 753-776.

15

References

Title Slide Photo Credits (from left to right): (1) www.scientificamerican.com (2) http://www.pelamiswave.com (3) Sandia National Labs (4) http://www.metaefficient.com (5) http://img.dailymail.co.uk (6) https://inlportal.inl.gov