MULTICOMMODITY NETWORK FLOW MODEL FOR LONG-TERM HYDRO-GENERATION OPTIMIZATION

The solution to the long-term hydro-generation optimization problem finds out, for each interval into which is divided a long time period, the amounts of hydro-generation at each reservoir that minimize the expected cost of thermal generation over the period. It must take into account that water inflows are olny known as probability density functions. The nonlinear network flow technique already employed in the case of deterministic inflows, is here extended to the case of inflows given as probability density functions by using multicommodity network flows. It is shown that problems involving several reservoir systems with partially dependent inflows, can also be adequately modelled.