1 Utility Analysis of Future Power Generation Based on the Choice Experiment Japan-US Workshop on Fusion Power Plants and Related Advanced Technologies.

1

Utility Analysis of Future Power Generation Based on the Choice Experiment

Japan-US Workshop onFusion Power Plants and Related Advanced Technologies with participation of EU

March 5-7, 2008 at UCSD , US

Ryoji Hiwatari*, Kunihiko Okano, Yoshiyuki Asaoka, Kohji NaganoCentral Research Institute of Electric Power Industry(CRIEPI)

Yuichi OgawaHigh Temperature Plasma Center, the Univ. of Tokyo

Takaaki KatoFaculty of Environmental Engineering, The University of Kitakyushu

Kenji TobitaNaka Fusion Energy Establishment, JAEA

Takayoshi NorimatuInstitute of Laser Engineering Osaka University

Committee for an assessment on advanced energy technologiesOrganized by CRIEPI

http://www.kitakyu-u.ac.jp/index.html

http://www.jaea.go.jp/english/index.shtml

2

Y. Ogawa (University of Tokyo, Chairman)K. Okano, Y. Asaoka, R. Hiwatari K. Nagano, T. Nanahara, N. Goto (CRIEPI)K. Itou (The Institute of Energy Economics, Japan) T. Kato (University of Kitakyushu)R. Oomori (Toyota Central R&D Labs., Inc.)) S. Konishi, Y. Oomura (Kyoto University) K. Itou (Kyushu University)K. Tokimatsu, T. Oosumi (Research Institute of Innovative Technology for the Earth) H. Suzuki (University of Tokyo)T. Norimatsu, H. Azechi (ILE, Osaka University)K. Tobita (Japan Atomic Energy Agency)A. Sagara (National Institute of Fusion Science)A. Hibiki (National Institute for Environmental Studies)

Members of Committee for an assessment on advanced energy technologies



3

Summary (1)

•The key to choose a best-mix of energy systems is how to evaluate and compare quantitatively the merits and the demerits of each energy system.

•There is no single energy system to satisfy all the three aspects of sustainability, i.e. energy, economy, and environment. Hence, the well-balanced combination, i.e. best-mix, of several energy sources should be pursued

•In this study, a methodology to evaluate the merits and demerits of energy technology is developed based on the conjoint analysis technique.



4

Summary (2)

•Utility values for energy abundance, environmental load (i.e. CO2 emission), stability of supply, sense of security, and other features as well as economic performance, are estimated from several thousands of choice experiments

•Difference of utility between gender, generations, and specialists on energy systems is discussed, and how to apply those utility values to energy development is proposed.



5

Contents

1.Choice experiment and Conjoint analysis• Concept of conjoint analysis• Attribute and its levels on the future electric generatio

n system for choice experiment • Choice experiment applied in this study

2.Results• Utility values and those reproducibility• Difference between gender, generations in general pu

blic• General public v.s. specialists on energy system

3.Application of utility values• Development priority from the public viewpoint• Education tools (risk communication)



6

Concept of the conjoint analysis

Example: a choice of your personal computer Profile example•Pentium D 3.0GHz•512MB•10GB•98000 Yen•HP, Slim type, etc

•The conjoint analysis is one of techniques to survey on public opinions.•Profile Cards, where various attributes of the target product or concept are listed, are used.

•The respondents choose the most preferable Profile Card from the set of the Cards (two or more) which are shown in each question.

-> Choice Experiment•The utility-function for the figure of merit on the individual attribute can be determined through a kind of statistical treatment.

Attributes*CPU speed*RAM size*Hard-disk capacity*Price*Brand image, design etc.



7

Attributes for Electric Generation System (1)

Attribute Level

1. Resource① nearly infinite② 100 times as large as petroleum③ 10 times as large as petroleum④ equal to petroleum

2. Resource Distribution

① every country②most of countries③localized like petroleum④localized more than petroleum

3. CO2 emission

① 1/100 as low as petroleum② 1/10 as low as petroleum③ 1/2 as low as petroleum④ equal to petroleum⑤ twice as large as petroleum

4. Electricity expense in Japan

① -2400 billion $/year (-200$ /year per person)② -1200 billion $/year (-100$ /year per person)③ equal to present (total 14 trillion $ in Japan)④ +2400 billion $/year (+200$ /year per person)⑤ +10 trillion $/year (+800$ /year per person)



8

Attributes for Electric Generation System (2)

Attribute Level

5. Interruption due to natural phenomena

① no effect② interrupted but regularly③ interrupted irregularly and suddenly

6. Interruption due to social phenomena

① highly tolerant② vulnerable to domestic socio-event only③ vulnerable to international socio-event only ④ vulnerable to both societies

7. Impact in the worst accident case

① no impact to periphery② no impact to health of public with evacuation③ injury and health damage can be possible④ some fatalities

8. Abuse by terrorism ① less useful② useful

Further consideration on waste and time (e.g. when does a new energy system becomes available ?) is required.



9

Choice Experiment on Future Electric Generation System

Impact at accident

Impact at accident

No impact to health

Some fatalities

Electricity expense in Japan

Electricity expense in Japan

Partial profile method of 3 attributes, which are chosen from all 8 attributes at random for each question, has been used. The levels of hidden attributes are assumed to be equal. 30 questions per person. The number of respondents is 1300. Then we got 39,000 answers.



10

Choice Probability from Utility Values

Utility value of A　Ｕ A = [a1iX1i + a2iX2i + a3iX3i + a4iX4i + -----]A

Xki is the i-th level value for the attribute k.aki is the coefficients for the effect-function at level i

Logit modelThe probability of choosing card-A of Lth question is

1exp1

expexpexp

AUBUBUAU

AUAP LLLL

LL



11

Contents






12

Distribution of respondents　 - public -

Male50.6%

Female49.4%

Male n=682＜＞11.1%

22.6%

22.7%

21.6%

22.0%10's 20's 30's40's 50'sOver 60's

Female n=667＜＞21.4%

19.8%22.6%

13.6%22.5%

10's 20's 30's40's 50'sOver 60's

<n=1349>

13

Results and Reproducibility of Utility Values

In this method, the differential of effects for every attributes can be compared each other.Resource 100 times as large as petroleum has a same effect to resource distributed in most of country. The increment of effect by “every country” is small.

An effect reduction due to 200$ increment in cost can be canceled by an increment of effect due to a reduction of CO2 emission by half.

Comparison between 1st (262persons) and 2nd (1349persons) shows the clear reproducibility



14

Difference between Gender

Utility effects for Impact the worst accident case is clearly different between Male and Female

Male

Female

Male’s utility for resource distribution is lager.



15

Difference between Generations

Utility effects for Impact the worst accident case is also clearly different between 10 and 20’s and over 60’s10・

20’s

Over 60’s

Utility effects of over 60’s on CO2 emission is clearly larger than that of 10 and 20’s

10・20’s

Over 60’s



16

Distribution of respondents – specialists -<n=352>

Male97.2%

Female2.8%

30's18.5%

20's1.1%

40's32.4%

50's31.0%

Over 60's17.0%

government1.7%

othercompany

16.5%

others6.0%

Energycompany

22.2%

University38.4%

Institutes15.3%

17

Difference between public and specialists

specialists

public

Utility effects of specialists on CO2 emission is clearly larger than that of general public

Utility effects for Impact the worst accident case is also clearly different between specialists and general public.

This is partially because 97% of specialists are Male, however, that is not enough to explain the difference.specialists

public

Male



18

Contents






19

Choice Probability from Utility Values




1exp1

expexpexp

AUBUBUAU

AUAP LLLL

LL



20

Improvement on electricity expense

BaselineImprovement on resource

How to take public preference into account

Interruptiondue tonatural

phenomena10%

Interruptiondue tosocial

phenomena10%

Impact inthe worstaccident

case21%

Abuse byterrorism

10%

Resource9% Resource

Distribution12%

CO2emission

16%

Electricityexpense

12%

1.Assume the baseline of a new energy system( In this case, all utility effects are Zero)

2.Consider the improvement of each attribute from the baseline to the best level.

3.Effectiveness on public preference is shown as the choice probability



21

public

45%55%

•Good in CO2 emission•Good in Impact in the worst accident case

Choice Probability - Public v.s. Specialists -

specialist

75%25%

Public v.s. Specialists



22

Summary

•The key to choose a best-mix of energy systems is how to evaluate and compare quantitatively the merits and the demerits of each energy system.

•There is no single energy system to satisfy all the three aspects of sustainability, i.e. energy, economy, and environment. Hence, the well-balanced combination, i.e. best-mix, of several energy sources should be pursued

•In this study, a methodology to evaluate the merits and demerits of energy technology is developed based on the conjoint analysis technique.

•Utility values for energy abundance, environmental load (i.e. CO2 emission), stability of supply, sense of security, and other features as well as economic performance, are estimated from several thousands of choice experiments•Utility difference between gender, generations, and specialists on energy system is discussed, and how to apply those utility values to energy development is proposed.



23



Utility value and effective functions -1


The likelihood function L = PA ・ PB ]N Choice A Choice B

N: Number of respondents

Person N’s choices: A A B A B ------------ (total 30 answers)

Probability of this phenomenon : PA1PA2PB3PA4PB5--------------------

All respondents

1exp1

expexpexp

AUBUBUAU

AUAP LLLL

LL



24


We get a lot of data-set for (UA, X1i,X2i,X3i,X4i,----), (UB, X1i,X2i,X3i,X4i,----) 　 and the result of respondent’s choice. 　

A set of aki’s which give the largest likelihood, can be determined by maximizing the function L.

The effect-function for the attribute k is shown

as a line graph by a plot of {akiXki }.

Utility value and effective functions - 2

The likelihood function L = PA ・ PB ]



1 Utility Analysis of Future Power Generation Based on the Choice Experiment Japan-US Workshop on Fusion Power Plants and Related Advanced Technologies.

Documents