OCR Text |
Show reliabilities, and operation over a record of flows. This is the course adopted in the model presented in this chapter. In a later section procedures for circumventing the need for pre-specification of yield reliabilities are discussed. It is also worth noting that the MSP procedure is capable of determining the desired storage capacities for any pre- specified operating rule that the decision maker may desire to use in conjunction with a yield specification. For example, an operating rule that solves for the hydropower release as a function of available head could easily be specified as part of the recursive MSP procedure, and be related to the firm release Hs by requiring that the average release ( or the minimum) equal the candidate yield Hs specified by the optimization model. Explicit constraints in the optimization model or implicit constraints through the MSP procedure could be used to achieve such an equivalence. ( 2) FAS - The reservoir should be full at the end of the operation period. STs - Ss > 0 ( 5.14) where Sys is the active storage at the last operation period ( T), and is also evaluated as a function of the decision variables as part of the MSP procedure. Note that it is difficult to end up with the reservoir full exactly on the last month of the operation period, where the yield model is used. In the applications performed, the maximum active storage in the last 36 months ( out of 432 or 660) was used instead of Sjs. ( 3) DSC - The dead storage Ds should fall within prescribed bounds, specified as permissible ratios of dead storage to total storage. D DR . < - i < DR ( 5.15) s^ nin T s, max v ' s where DRS j ^ and DRS m2LX are lower and upper bounds of the ratio of dead storage to total storage. This constraint is also a nonlinear function of the decision variables. If Ts is zero ( i. e. no reservoir) the ratio of dead storage to total storage is set to its lower bound. ( 4) GSC - Limits on Generator Capacity Gs. The generator capacity Gs is restricted to upper and lower limits GSjiriax and Gs m^, respectively. G £ G < G ( 5.16) s, imn s s, max v ' Since Gs is not a decision variable of the optimization model, and is evaluated through simulation considering all the model decision variables, this constraint is nonlinear in terms of the decision variables. Upper and lower bounds are specified on all decision variables. hf . < Mr < M1" m= 1... M, r= 1.... R ( 5.17) sm. min sm sm. max ' v ' 111 |