142
5. Sensitivity
of Planning Decisions in River Basin
Management
terministic, and the sensitivity of its objective function and outputs to
small variations in four inputs and parameters (each at two levels) was
measured. The inputs and parameters chosen were (I) system hydrology,
(II) reservoir costs, (III) future irrigation demand, and (IV) interest rate.
A linear penalty function was used to punish noncompliance with irrigation
demands.
In all, sixteen sets of trials were run resulting in no change in the dam
construction schedule for 11 of the cases. When both the interest rate and
the demand schedules were simultaneously at their lower levels the construction schedule was altered for the two later dams, but the same dam
was chosen to be built first. In four cases no solution was possible for
scheduling the dam construction because no positive net benefit accrued
from the construction of dams 9-14.
Finally there is a discussion of stochastic analysis of water resources
systems. Three strategies, all based on simulation, are discussed. It is concluded that method 3, which combines optimization and regression analysis
with simulation, is the most reliable available.