The project concerns the development of generic software for large-scale multi-objective optimal decision problems under uncertainty. The main aim is to provide fundamental decision making methodology in conjunction with a natural problem specification framework. Uncertainty underlying the system, specifically random effects such as price or demand forecast errors, is addressed using a scenario based and mean-variance framework. The multistage stochastic mean-variance approach is extended to find optimal investment strategies after the specific tax rules are taken account. The worst case design for robust investment strategies is also developed.

The technical issues that will be addressed are:

• Definition of performance optimisation problems in terms of queues with non-standard features of the relevant novel queueing systems;
• Derivation of appropriate static and dynamic optimisation algorithms, with transient analysis for the incorporation of higher moments to account for quality of service;
• Worst case analysis of systems operating under different environments, de feasibility envelopes;
• Resource pricing, e.g. in grid-like models;
• Validation including pilot applications to industrially relevant systems.