On Multistage Stochastic Mixed 0–1 Optimization with Time-Consistent Stochastic Dominance Risk Averse Management

A new time-consistent risk averse measure is considered, so-called Expected Conditional Stochastic Dominance (ECSD), for multistage stochastic mixed 0–1 optimization, where first- and second-order stochastic dominance risk averse functionals are taken into account. As a result of the ECSD modeling, its problem solving is much more difficult than the Risk Neutral counterpart, so, it is unrealistic to solve the problem up to optimality by plain use of MIP solvers. Instead of it, decomposition algorithms of some type should be used. Computational results are reported for instances of a well-known real-life problem, where a decomposition matheuristic algorithm is tested in its efficiency and computing effort, having the plain use of a MIP solver as a benchmark for computational purposes.