127
Such a tree is shown in Annex 2, Figure 1. A square node indicates a decision (test or not), a circular
node indicates a probabilistic outcome (wet well or dry well), and a triangular node indicates the
net benefits of that particular outcome. The decision maker chooses the
path through the tree that
provides the maximum expected benefit.
The first path through the tree is one in which the test is undertaken. If a wet hole is found (assessed
at probability 0.2) then investment would take place because of the certainty about the resource—the
net benefit would be 300 - K (the net benefit of the project less the cost of the test). If a dry hole is
found (probability 0.8) no investment would take place because of certainty that a viable resource
is not present—the net benefit would be -K (the cost of the test). Taking the expected value of these
outcomes (the weighted average) leads to a value of 60 - K. Provided the cost of the test is less than
60, the outcome of testing
and then investing or not, according to the test results, yields a positive
expected return.
v
The second path through the tree involves first the decision not to test and subsequently the decision
to invest. This has two possible outcomes—a wet hole (probability 0.2) with net benefit of 300 (no cost
of testing) and a dry hole (probability of 0.8) with a net benefit of -100 (wasted investment costs). The
expected net benefit (weighted average) would be -20. Drilling without testing first would, with the
assumed
probability of success, lose money on average. The difference from the case with testing is
that investing when there will be a dry hole could have been avoided by following the indications of the
test results.
The third path through the tree would be the decision not to test and not to invest. The expected net
benefit would be zero—no costs but no geothermal plant. If testing is not done,
it is better not to invest
than to invest, since the third path has a better expected outcome than the second path. However,
provided the cost of the test is less than 60, it is best to follow the first path—test, then invest or not
according to the results.
This example makes it clear that the values of the prior probability of a viable resource being found
and the cost of the test are crucial to the decision taken. Decisions that maximize the expected net
economic benefit for different values of these parameters are shown in Annex 2, Table 1.
Lower
costs for a test make it more likely that it will be optimal to test before investing, while higher prior
probabilities of well success make it less likely that a test will be needed before investing. A high test
cost and a low probability of success leads to the decision to neither test nor invest.
The cost of the test is known from general geophysical experience, adjusted to local conditions.
Drilling an exploration well is a clearly defined activity and the cost should be calculable within a
narrow margin. However, making a prior assessment of the probability that
a viable resource will be
found is more difficult and depends, in large part, on whether any prior drilling has taken place in
the area. Bickel, Smith and Meyer (2008) discuss some aspects of arriving at an assessment of this
probability. Annex 2, Box 1 indicates that globally the probability of well success (proportion
of wells
drilled that have found viable resources) is substantial, and that the probability of a well success tends
to increase with the number of wells already drilled in a field as drillers learn about the characteristics
of the field from previous trials.
A n n e x 2
v
In practice it is likely that a program of several exploratory wells would be drilled, and that certainty would be
equated with a given number
(for example, at least two out of four) indicating the presence of a viable resource.