OCR Text |
Show sizes, it is difficult to gain any intuitive feeling about the system, from the model solutions. The incorporation of system peculiarities in the classical formulations is also not an easy matter. A decomposition of the problem by implicitly satisfying all the operational constraints through a simulation, and treating the reservoir sizing problem in a stimulus ( yield) - response ( required storage capacity) problem with the response function evaluated through simulation, leads to the very compact, implicitly stochastic, nonlinear programming formulation presented. Procedures for the optimal and conjunctive operation of the reservoir system through the specification/ determination of appropriate yield reliabilities are also discussed. A number of heuristics and simulation strategies are also developed to efficiently address the determination of the response functions. These strategies are likely to be of use even as stand alone analysis procedures. Refinements of the general optimization model presented, particularly in the area of conjunctive operation of reservoirs to maximize system yield reliability should be pursued. The development of more specialized solution algorithms as discussed in the previous section should also prove beneficial. Two sets of parametric applications - one to the Bear River basin, and the other to the Jordan River basin - were performed. The features modeled in the two sets of applications were somewhat different. In either case, the model performed very well. Reservoir capacities and yields by purpose were identified in a manner largely consistent with expectation from simulation studies performed by local and state planning agencies. While procedures for reservoir operation optimization and rule curve derivation were developed in chapter 6 as part of the same modeling framework, no extensive applications of these procedures were performed with data from the two reservoir systems studied. These procedures were however tested successfully with data from City Creek. These results are not presented in chapter 6 for brevity. Further applications of the operation models with existing local reservoirs may be performed if an interest in such an exercise is indicated by local agencies. 8.3 Conjunctive use models There is a dearth of optimization models in the literature that consider a distributed parameter ( and detailed) representation of the surface and groundwater systems in determining reservoir sizes, groundwater development levels and operation policies. Applications of such models to real world situations are even harder to find. As discussed in the previous section, the reason for this is the curse of dimensionality associated with the classical formulations. The yield formulation developed for reservoir screening lends itself very well to the inclusion of groundwater issues. Two models were developed to analyze surface water - groundwater development. The first model developed considered a lumped parameter representation of the groundwater system through an aggregate groundwater firm yield in conjunction with the reservoir screening formulation presented. The parameters of the aggregate groundwater yield were determined from an analysis of the optimal solutions to the distributed parameter groundwater management model described in chapter 3. Ground water and surface water operation was not considered to be complementary - failure of one in some years was not compensated for by an increase in the utilization of the other. The model simply addresses the question of whether groundwater or surface water should be developed, and what is the effect on reservoir selection if an alternate source ( groundwater) is available. The formulation is thus a very simple extension of the reservoir screening model. It should be noted that both in this model and the second one developed, the reservoir screening model developed in chapter 5 retains all its attributes - multi- purpose, multi- reservoir, with yield 179 |