Use of risk premiums in chance‐constrained dynamic programing

Stochastic dynamic programing offers a powerful means of deriving optimum operating policies for water resource systems. In order to limit the probability of system failure associated with such policies it is necessary to amend the standard algorithm and use some form of chance‐constrained dynamic programing. This technique may involve an iterative search based on variations in such parameters as the penalty for failure or the discount rate. Use of the latter parameter is found to be quite feasible and is analogous to the imposition of risk premiums when one is investing in less secure ventures: the greater the probability of failure, the greater the premium demanded.