86025 Energy Systems AnalysisArnulf Grubler A Primer on Energy Technologies and Technological Change

86025 Energy Systems Analysis Arnulf Grubler

A Primer on Energy Technologies and Technological Change


Energy Technologies in the US(estimates in GW)

Prime movers based on US DOC, 1975 and 1994, (year 2000 data refer to 1992)All others: zero order estimates

46 TW in 2000 equal 4.6 Trillion US$ (or 40% of US GDP in 2005) at 100 $/kW


An 75% Snapshot of Energy Technologies:Steam & Combined Cycle + Motors

ElectricityWater

Generator

Generator

Heat recovery steam generator

Cooling air

Condenser

Combustion chamber

Fuel

Air

Exhaustgases

Electricity

Gas turbine

Steam turbine

boiler

electricdrives

fuelsupply


Technology and Energy Economics

• Price data, while volatile indicate no resource scarcity/depletion

• Depletion mitigated by technological change, substitution, efficiency improvements

• Technological change originates both from within energy sector (off-shore oil) as well as from economy at large

• Productivity gains and cost declines yield macro-economic benefits


US Energy Prices over the Last 200 Years:Prices do not show depletion, nor can explain energy

transitions (wood→coal→oil/gas); Technology as main driver

US - Energy Prices in constant $ (=inflation adjusted)

1

10

100

1800 1825 1850 1875 1900 1925 1950 1975 2000 2025

US

$195

8 p

er k

Wyr

wood

coal

oil

gas

1860: 300 $/bbl


US - Crude Oil Price (at wellhead)

0

10

20

30

40

50

60

1900 1920 1940 1960 1980 2000 2020

curr

ent

and

co

nst

ant

US

$(20

00)

in constant 2000$

in current $

Extent of inflation since 1900:1900$ = 1$/bbl2000$ = 22 $/bbl

US Electricity in the 20th Century

US - Electricity Prices vs. Cumulative Generation: 1900 - 2005

1

10

100

1000

1 10 100 1000 10000 100000 1000000

cumulative TWh electricity generation

cu

rre

nt

an

d co

nsta

nt

cen

ts(2

00

0)

pe

r k

Wh

US - Electricity Prices (residential)

1

10

100

1000

1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010

cu

rren

t an

d c

on

sta

nt

cen

ts(2

000) p

er

kW

h

Falling nominal and realprices (constant 2000$)to 1970

Nominal and realprices vs.cumulative generation


Energy Economics of the 20th Century

1900-1970:

Falling real-term fuel prices

Conversion:-- Improved efficiency,-- economies of scale,-- rapid demand growth

and-- capital turnover

>1970:

Price escalation & volatility of fuel prices

Conversion:-- scale and efficiency frontier reached

for steam cycles,-- slow diffusion of gas turbines

and combined cycles,-- slower demand growth, lack of investment, slow capital turnover-- increasing environmental regulation-- deregulation of natural monopoly markets (e.g. electricity)


History of US Steam Turbine Generators

Source: T. Lee and R. Loftness,1987. IIASA WP-87-54

Drivers of lower costs:

Increasing temperature & pressure= higher efficiency.

Increasing unit size= economies of scale


US - Scale Frontier of Power Plants

Source: T. Lee and R. Loftness, 1987. IIASA WP-87-54


1990 1991 1993 1998

Diameter, m 23 31 44 63

kW 150 300 600 1500

DM per kW 2538 2410 2135 1752

Million DM .381 .723 1.281 2.638

Estimate .686 1.220 2.633

Difference (actual/estimate)

+5.4% +5.0% -0.2%

Declining Costs per kW of German Wind Turbines: Pure Economies of Scale: DMt = (kWt/kWt-1)0.84 x DMt-1


Nordex N-80: Capacity: 2.5 MWHeight: 80 mRotor: 80 mTotal: 120 m

HarknessTower: 66 m


101 of Technological Change

• Technological change is a process involving many steps and feedbacks.

• Uncertainty pervasive at all stages of technology life cycle

• Technology = combination of disembodied and embodied knowledge. Embodied TC only via (costly) investments (by technology users ≠ tech producers)

• Significant costs downstream “R” (research): Development dominates “R&D”, Deployment investments dominate R&DD

• “Value” of technology increases downstream also: Value of patent < private RoR < social RoR of innovation

• Returns to adoption: Static technologies: decreasing returns; dynamic technologies, networks: increasing returns (the more deployed the cheaper, better, more acceptable)


The “black box” of Technology

BasicR&D

AppliedR&D

Demon-stration

Nichemarkets

Diffusion

Product / Technology Push

Market / Demand PullLearning

Public Sector

Private Sector

DisembodiedTechnology(Knowledge)

EmbodiedTechnology(plant,equipment,..)

funding

funding incentives,standards, regulation,subsidies, taxes

investments,knowledge andmarket spillovers

Stages of Technology Development and the Resource Gap for Innovation

Stages of Technology Development

Research Prototype Pilot Production

Sum

Government

Industry

Relative Resourc

es Availabl

e

The “Valley of Death”

Source: M. Chertow, 2003

technology sellersAND buyers


R&D as % of Net Sales(Source: EPRI, 2005)


US – Energy R&DSource: Nemet and Kammen (in press)


US – Private Sector Energy R&D and Venture Capital

Source: Nemet and Kammen (in press)


US Electricity Sector: Investment as % of Revenues. Source: K. Yeager EPRI, 2005


The Energy “Valley of Tears”

• Declining R&D (public and private)

• Declining investments

• Declining venture capital

• Declining long-term R&D and investment incentives in deregulated markets

• Increasing needs for long-term strategic decision making (hedging portfolios to address climate change)


TECH 101 cont’dTechnological Change is…

• Uncertain (feasibility, improvement potentials, opposition, env. impacts)

• Dynamic (only certainty: Change)• Cumulative (building on past experience,

other technologies)• Systemic (no technology is an island:

Electricity+telephone+PC+www=Internet)• Actor based: Producers and Consumers• Extreme event like: Majority of benefits

from few “big hits”


Innovation Uncertainty: Patented but non-functional smoke-spark arrestors

Source: J. White, American Locomotives, 1968.


Scherer’s Rule (compounding uncertainties)

Probability an R&D project gets selected** ??

Probability of technical success (once selected)* . 57Commercialization (given technical success)* .67Financial success (given commercialization)* .74

Aggregate probability . 27

Magnitude of financial success(private AND social RoR)** ??

* Based on Mansfield et al.’s empirical study of R&D project histories in US enterprises in chemical, pharmaceutical, electronics, and petroleum industries

** Largest uncertainties!


Fat tailed distributions: Pay-offs from US Pharmaceutical Innovations:

(similar as in UK and EU “patent value” studies)

Source: Scherer, 2000.

NCE: New Chemical Entities

Cost Uncertainty: Important Information in Tails!

Strubegger&Reitgruber 1995 analysis of near-term technology cost projections


Basic Economics of PV Supply and Demand

Source: BP, 2003


Japan PV: Importance of Supply Push AND Demand Pull

Data: Yuji, 2002

Result of R&D Result of niche market filling

100

1,000

10,000

100,000

0 0.1 1 10 100 1,000

Cumulative expenditures, billion (1985) Yen

PV

cos

ts (

198

5) Y

en p

er W

1973: 30,000

y = 10 4.0 – 0.54x

R 2 = 0.989

1995: 640

Applied R&D InvestmentBasic R&D

1976: 16,300

1980: 4,900

1985: 1,200

Data source:Watanabe, 1995 &1997

Japan - PV Costs (improved knowledge) vs. Expenditures (effort): Cumulative Effects

→learning curves PR: 2-b = 2-0.54 = ~0.7 = 30% decline in costs/Wper doubling of cumulative expenditures(total: ~2.5 billion $, ~75% investment, ~25% R&D)


“Learning”: ResolvingTechnological Uncertainties

• Innovation: many are called, few are chosen• Diffusion: multiple factors, e.g.

-- Technological “Figures of Merit”-- Economics: Use = f (Costs)

- Static: Ct

- Dynamic: e.g. Ct = f (ΣtU) (LbD)• Feedbacks: e.g. spillovers, “take backs”• Externalities: e.g. networks, knowledge,• Impacts: nonlinear f(U) or

“discovery by accident”


LbD

• Vast case study literature (with possibilities of statistical interpretation, e.g. mean costs: -20% for 2x cumprod)

• Pro-innovation bias: mostly success stories (exceptions: Lockheet Tristar, French reactors)

• Quality improvements (+/-) largely ignored• Impossibility to separate R&D (innovation) and learning

in manufacturing (2-factor learning curves empirically vacuous and theoretically based on dismissed linear model of innovation)

• Intricate measurement challenges (costs vs. prices,…)• Effects of spillovers important but difficult to measure

(esp. for inter-industry and internationalspillovers, not to mention inter-technology spillovers)


Learning/Experience Curve TerminologyCosts: CLearning Rate: LR

(% cost decline per doubling of output)Progress Ratio: PR = 1 – LR

(remaining fraction of initial costs after doubling of output)

Learning parameter: bOutput: OLearning investment: Cumulative expenditures above

break-even value

Ct = C0 * (Σ0tO)-b

PR = 2-b

LR = 1 - PR

e.g. 30% cost reduction per doubling of output: Co =100 Ct = 70 Oo =1 Ot = 2 LR = .3 PR = .7 b = -.51477


Gas Turbines at GE

Source: MacGregor et al., 1991 and Rogner, 1995.


Improved Economics: Prices vs. Costs

Ford Model T

y = -0.214x + 3.8738

R2 = 0.9765

y = -0.1194x + 3.6024

R2 = 0.7333

4

5

-2 -1.5 -1 -0.5 0 0.5 1 1.5 2

log (cum. production, million)

log

(1

99

3$

/ca

r)

Prices

Costs

Source: Based on Abernathy&Ward, 1975

Learning Potentials: Number of units sold to date and cum. investments (at current replacement costs)

Cumulative number produced to date (1900-2005)

Investments, billion $

Motor vehicles/ICEs >2 10^9 42,000

Fuel cells <1 10^4 ~2

Computer microchips >1 10^11 800

PV cells >1 10^9 30

Gas turbines <1 10^5 130

Wind turbines <1 10^5 80

Nuclear reactors <1 10^3 1,000

Reminder: LbD Models need to separate economies of scale effects,i.e. consider UNITS rather than capacity


Technology Learning Curves

Windmills (Germany) (learning rate <10%)

2002

1993

1990

1998


Learning Rates of 108 Technologies

Source: Argote&Epple, 1990

Negative learning:Lockheed TristarFrench nuclear reactors

2 Uncertainties: Learning Rates and Market GrowthT

echn

olog

y dy

nam

ics

in r

espo

nse

to

R&

D o

utco

mes

, eco

nom

ies

of s

cale

, m

ater

ial c

osts

, etc

.

Demand growth, incentives, dynamics of competitors

Impacts of Uncertainty, Learning, and Spillovers (IPCC AR4 ,forthcoming in 2007)

Emissions from 130,000 Scenarios of Technological Uncertainty

Emissions by 2100, GtC

Rela

tive f

req

uen

cy (

perc

en

t)

5 10 15 20 25 30

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.05 10 15 20 25 30

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Set of 520 technology dynamics

Optimal set of 53technology dynamics

Gritsevskyi&Nakicenovic, 2000 Energy Policy 38:907-921

Global CO2 emissions by 2100, in GtC

Rel

ativ

e F

req

ue

nc

y, i

n p

erce

nt

Emissions from 130,000 Scenarios of Technological Uncertainty

Emissions by 2100, GtC

Rela

tive f

req

uen

cy (

perc

en

t)

5 10 15 20 25 30

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.05 10 15 20 25 30

5.5

5.0

4.5

4.0

3.5

3.0

2.5

2.0

1.5

1.0

0.5

0.0

Set of 520 technology dynamics

Optimal set of 53technology dynamics

Gritsevskyi&Nakicenovic, 2000 Energy Policy 38:907-921

Global CO2 emissions by 2100, in GtC

Rel

ativ

e F

req

ue

nc

y, i

n p

erce

nt

Figure 2.2. Emissions impacts of exploring the full spectrum of technological uncertainty in a given scenario without climate policies. Relative frequency (percent) of 130,000 scenarios of full technological uncertainty regrouped into 520 sets of technology dynamicswith their corresponding carbon emissions (GtC) by 2100 obtained through numerical model simulations for a given scenario of intermediary population, economic output, and energy demand growth. Also shown is a subset of 13,000 scenarios grouped into 53 sets of technology dynamics that are all "optimal" in the sense of statisfying a cost minimization criterion in the objective function. The corresponding distribution function is bi-modal, illustrating "technological lock-in" into low or high emissions futures respectivelythat arise from technological interdependence and spillover effects. Baseline emissions are an important determinant for the feasibility and costs of achieving particular climate targets that are ceteris paribus cheaper with lower baseline emissions. Source: Adapted from Gritsevskyi and Nakicenovic, 2000.


Cost Dynamics of PVs: 2 Case Studiescountry Japan USA

author C. Watanabe, 2002 G. Nemet, 2004

period 1976-1999 1975-2001

Cost decline factor 27 20

approach Top-down,econometric

Bottom-up, engineering economics

Main explanatory variable Knowledge stock (cum R&D + depreciation + firm spillovers)

1. Innovation: Module efficiency, poly-crystals,..;

2. Economies of scale,

3. “Learning”

Other variables Economies of scale,energy prices

Si-costs and use,wafer size and yield

Missing International & inter-industry spillovers,

Demand response

Intra-, inter-industry, and international spillovers,

Demand response

Role of institutions Public&private R&D funding, intra-firm spillovers

n.a.


US - PV Factors of Cost Declines (in constant $/W)Source: G. Nemet, 2004

Factor/period 1975-1979 1980-2001 1975-2001

Module efficiency - 9.9 - 7.1 - 17.0

Plant size - 4.3 - 9.2 - 13.5

Si costs - 5.0 - 2.7 - 7.7

Yield - 0.4 - 0.5 - 0.9

Si consumption - 0.4 - 0.6 - 1.0

Wafer size 0 - 0.7 - 0.7

Poly-crystal share 0 - 0.4 - 0.4

Sum - 20.1 - 21.1 - 41.2

Residual -28.6 - 0.6 -29.2

Total cost decline -48.7 -21.7 - 70.4

Innovation (R&D) Economies of Scale Experience (LbD) Other/Total


Summary• Technologies have emergent properties that are

constructed by “learning” processes (increasing returns!)

• Good empirical and theoretical understanding of “routine” innovations (e.g. incremental improvements via industrial R&D)

• Need to move beyond proxy drivers and black-box view of technology:Who learns what, when, and how?

• Need to move beyond pro-innovation biasin the empirical literature

• Technologies of greatest (economic, social, environmental) interest: low probability, extreme events which are difficult to anticipate or to model (→scenario approach)

Documents

86025 Energy Systems AnalysisArnulf Grubler A Primer on Energy Technologies and Technological Change