Resumo:
Advanced computer models are commonly used to simulate reservoir system's performance. If the number of possible management scenarios is large, it can be extremely difficult to follow related system's operation and get a valuable picture on its spatial and temporal behavior. The decision maker or analyst can be overburdened by quantity and complexity of information generated by model, particularly if system operation is repeatedly simulated for multiyear periods. Related problem is how to select the scenario with most desired long-term consequences. Possible approach is to use selected parts of model's output and re-interpret system behavior by means of certain performance indicators, create appropriate decision matrix and perform multi-criteria analysis to rank decision alternatives (scenarios). The paper proposes a methodology that includes: (1) multiyear simulations of system operation; (2) computing spatially and temporally distributed systemperformance indices such as supply reliability, resiliency and vulnerability; (3) unbiased entropy-based weighting the importance ofperformance indices; and (4) final ranking of scenarios by means of multi-criteria analysis. The number of scenarios and number of performance indices is not restricted, and to account for possibly large sets of scenarios, an ideal-point-distance multi-criteria method TOPSIS is suggested. Proposed methodology appeared to be confident and robust in proof-of-concept application in Brazil.
