"The Choice of a Power Reactor"

It was proposed in:

Z. Covaliu and R. M. Oliver, "Representation and solution of decision problems using sequential decision diagrams", Management Science, vol. 41,

pp. 1860–1881, 1995.

And it is described as follows:

"Consider the problem of choosing a nuclear power reactor to provide future energy needs. If the decision maker decides to build such a reactor he or she can either select a "conventional" type or an "advanced" type. The latter, if successful, is more promising in its economic return, but having larger uncertainties is riskier. Before making the main decision, the decision maker can conduct tests, at a known cost, of critical components of the advanced reactor, the results of which will reduce some uncertainties related to the prospects of this, but not of the conventional, reactor. There are at most two decisions to be made denoted by D1 and D2. Initially, one of three alternatives has to be chosen: doing nothing (Di = dn), building with no test (Dl = nt), and testing (Dl = t). If the initial decision is to build without testing, a second decision must be made on the reactor type: conventional (D2 c) or advanced (D2 = a). If D1 = t, the test outcome, denoted by T, may be excellent (T = e), good (T = g), or bad (T = b). In the latter case, the advanced option is not viable, and the conventional reactor will be constructed."

It is worth noting we have added the possibility of not building any reactor at all.