Decision making in a complex environment, characterized by multiple, incommensurate and conflicting criteria, is influenced by sources of uncertainty that stem from the imprecision and changes associated with the input data, the unavoidable imprecisions in the modelling phases, or the evolutionary characteristics of the decision maker preference system throughout the interactive decision process.

This R&D project achieved the following main aims:

- to make a comparative analysis of the distinct approaches (probability theory, fuzzy set theory, interval programming and sensitivity analysis) to deal with uncertainty and imprecision in multiple objective linear programming (MOLP) models;

- to make theoretical and methodological advances in the conceptual framework of these approaches, particularly as far as sensitivity analyses and interval programming is concerned;

- to develop computer applications suitable to provide decision support to tackle uncertainty in MOLP.

4 Ph.D. dissertations and 2 M.Sc. have been initiated in the framework of this project.

A new routing algorithm has been developed considering the routing problem in integrated communication networks as a multiple objective shortest path problem, therefore explicitly incorporating the distinct evaluation aspects into the formulation. The developed algorithm computes nondominated paths by optimising weighted-sums of the multiple objective functions to determine vertex solutions (that is, those which belong to the boundary of the convex hull), and using a computationally effective k-shortest path algorithm to search for unsupported nondominated solutions within duality gaps (that is, solutions located inside the convex hull). QoS requirements may be expressed as additional (soft) constraints on the objective function values in terms of requested and acceptable thresholds for each metric, which define preference regions in the objective function space.

An innovative multicriteria model has been developed aimed at studying the evolution scenarios to deploy new supporting technologies in the access network to deliver broadband services to individuals and small industries.

2 M.Sc. dissertations have been concluded in the framework of this project.

Models of final use electric loads have been developed, which are based on the physical phenomena related to its functioning. Thee are mainly associated with heat transfer, consumption and conservation of thermal energy (generically called physically-based load models).

A simulation method has been established for a large number of loads of the same type, which enables to define the consumption diagrams of loads subject to similar environmental conditions, with or without remote load control actions by the electricity distribution company.

A multiple objective mixed integer linear programming decision support model has been developed, which enables to take into account the main concerns in a remote electric load control program: to minimize the peak of the load diagram as perceived by the distribution company, to maximize the revenue obtained with selling the energy necessary to maintain the energy services provided by the controlled loads, to maximize the quality of service.

1 Ph.D. dissertation has been concluded in the framework of this project.