102 Pages

R & D planning under uncertainty


Gain access to the library to view online
Learn more


A feasibility project
Research policy and organisation


Published by
Reads 8
Language English
Document size 2 MB

Commission of the European Communities
A Feasibility Project
EUR 9196/1, EN
Blow-up from microfiche original Commission of the European Communities
A Feasibility Project
and Y. SMEERS1
1 Université Catholique de Louvain - CORE
1, Place de l'Université
2 Facultés Universitaires N.D. de la Paix
Contract No. EM2-049-B 1983
Directorate-General for Science, Research and Development
1984 EUR 9196/1, EN Published by the
Information Market and Innovation
Bâtiment Jean Monnet
Neither the Commission of the European Communities nor any person acting on behalf
of then is responsible for the use which might be made of the following
©ECSC — EEC — EAEC, Brussels · Luxembourg, 1984 III
The Energy Research and Development Programme of the European Communities,
approved by the Council on 11th September 1979 is one of the actions with
which the Community replied to the challenge arising from the energy price
crisis. This four-year programme extending from 1 July 1979 until 30 June
1983 covers the following sub-programmes ("objectives") :
- Energy Conservation
- Production and Utilization of Hydrogen
- Solar Energy
- Geothermal Energy
- Energy System Analysis and Strategy Studies
The programme is implemented by means of so-called "indirect actions",
i.e. by concluding research contracts, on a cost-sharing basis, between
the Commission of the European Communities and research laboratories, industrial
firms, universities or other third parties in the Member Countries.
The results of the research projects carried out in this programme become
available in the final reports delivered at the end of each contract, and
diffused by the Services of the European Communities in the form of the
present report. This report results from a contract of the sub-programme
on "Energy System Analysis and Strategy Studies".
Further diffusion of the knowledge gathered through the implementation of the
programme is obtained by means of :
- status seminars in which the content and the progress of the work is described
and discussed,
- symposia and conferences (either organized by the Commission or by third
parties), in which work of EC contractors is presented in "free competition"
with scientific papers submitted from all over the world.
- a yearly status report of the whole Energy Programme.
All the reports published directly by the Commission are generally quoted with
a short summary in "EURO-ABSTRACTS" and can be obtained in full text or micro­
fiche by everyone interested from the
Boite Postale 1003,
Further information on the whole Energy Programme or on any one of the sub-
programmes can be obtained from the
Directorate-General for Science, Research and Development
200, rue de la Loi,
Uncertainty pervades the whole field of energy planning : it is par­
ticularly relevant in R-D decision problems which involve long term horizon.
Because of this one can expect that the scenario approach, commonly found
for treating situations of uncertainty will prove inadequate for dealing
with energy R-D projects. In particular it is likely that equally plausible
assumptions can be made on future parameters that may lead to drastically
different conclusions about the viability of new technologies.
This paper reports on a feasibility study on the use of statistical
decision analysis for exploring R-D energy planning problems. We depart
from the pure scenario approach which examines the advantages of a given
technology under different assumptions and consider a decision tree where
we structure the set of possible R-D decisions and random events relevant
for the technologies considered. Each path of the tree corresponds to a
usual scenario which is then studied by a deterministic model. The value
of the objective function of the model is used for valuating the terminal
•codee of the tree which is then analysed by the usual methods of statistical
decision analysis. - 1 -
Uncertainty pervades all aspects of energy planning.
Demand, fuel prices and investment costs exhibit considera­
ble vagaries and are difficult to forecast. More recently
the changing economic conditions and the resulting finan­
cial difficulties encountered by firms and governments
have added a new dimension to the problem: financing dif­
ficulties are faced now in the investment process that
would simply never have been envisaged five or ten years
Uncertainty is usually tackled by scenarios: educated
guesses are made on several parameters so as to arrive at
a coherent set of assumptions. This provides the framework
in which the evolution of the energy system in then studied.
This approach, although quite common is not always satis­
factory: modelers rightly insist that their scenarios should
not be taken as forecasts but are essentially aimed at
bounding uncertainty . This point of view is certainly
quite reasonable , but does not help in sorting out
the numerous energy outlooks constructed on these scenarios:
bounding uncertainty is of no use when too many different
and sometime incompatible decisions can be justified on
the basis of equally plausible scenarios.
These difficulties, although recognized a long time
ago, are still largely present today. The reason is con­
ceptually simple but sometimes difficult to explain. Pre­
sent decisions are not made with perfect foresight, con­
trary to what many optimisation energy models assume.
They are not done either on the sole basis of past expe­
rience as represented in recursive simulation models.
They result from past experiences and future expectations:
more specifically they are made, at least in principle,
on the basis of the current situation, with a feeling for - 2 -
what the future could be; hopefully they also take into
account the possibility of adapting to those different
possible evolutions. In term of modeling this implies
that actions to betaken now should account for what deci­
sion will be under different future circonstances. This
necessary inclusion in the current decision process of
a representation of a set of future choices (one for each
possible future) is what is technically, the core of the
Uncertainty crucially depends on the scope of the
problem on hand. The present work deals with R-D in new
energy technologies and hence with the longer term problem.
Needless to say, the longer the horizon the more numerous
the uncertainties and the more elusive the scenario approach,
This is well known: long term outlooks are often generated
to be superseded in a few months by new, sometimes contra­
dictory studies; completely different recommendations find
equally sound scientific support and very little informa­
tion can be gathered from scattered of conflicting results,
Whatever the efforts, uncertainties cannot be reduced beyond
a certain level for which the purely deterministic analysis
is unable to sufficiently discriminate between alternatives.
This paper reports on a feasibility study commissioned
by DGXII for investigating the use of stochastic decision
methods for R-D energy analysis. A small set of technolo­
gies has been selected as the object of the study. VThile
most of the R-D energy studies try to achieve a ranking
of technologies, we instead concentrate on theg of
the current R-D decisions related to these technologies.
This approach is in the line of statistical decision me­
thods and should help illustrate the difficulties and
poddibilities of the approach.
The paper is organised a follow: the structuring of
the R-D process is discussed after this introduction and - 3 -
some methodological aspects of decision under uncertainty
recalled in section 3. The energy model constructed in
this context is presented next and a summary of the results
of the analysis is given in section 5. Finally recommen­
dations about possible developments of the work are stated
last. This paper is a summary of the study: more detailed
developments are presented in appendices. - 4 -
The analysis of R-D energy strategies has been the
focus of many studies. In this work we shall follow the
structuring of the R-D process presented in [ 1 ] and [2 1.
Very similar analysis has already been conducted for the
European Commission [ 3 ] and would be most relevant in
future work on the subject.
Our modeling of the R-D decision process is directly
taken from the now classical field of statistical decision
theory. One decomposes the R-D phase into individual steps
corresponding to prototype, pilot or demonstration plants.
In previous work done for the Commission [4 ] Love and
Michel similarly advocate this decomposition as valid for
most of the new technologies. In addition they introduce
the notion of generic technology which does not correspond
to the construction of specific plants but to a continued
R-D effort on a family of processes which hopefully should
lead to the emergence of a commercial technology. These
two different concepts are of direct applicability in this
work: while photovoltaic cells belong to the class of ge­
neric technologies, HTR, FBR, underground gasification,
solar tower and hard coal liquefaction can easily be cast
in the more common framework of technologies that require
a sequence of prototype, demonstration... plants before
reaching maturity.
A main aspect of the work is the representation of
uncertainty. In this project experts were asked about
technico-economic characteristics of the technologies.
More global information about future market share were
purposedly excluded from the discussions: these are indeed
the result of market operations, and as such, should be
an outcome of the analysis. It has been argued [4 ]
that imperfection in the economic system could seriously