Using a GAMS modelling environment to solve network scheduling problems

摘要

The network scheduling problem is formulated as a dynamic optimisation problem and then transformed into a non-linear programming problem. The optimisation problem takes into account the full hydraulic model that is used in simulation, thusenabling the optimisation and simulation models to use the same input data.The mathematical formulation shows that it is a mixed integer problem. However, the physical nature of the problem enables it to be relaxed to a continuous non-linear programming problem, which is easier to solve. The case study shows that this relaxation is valid practically.The equations in the problem formulation are directly implemented in the GAMS programming language. The language is declarative which enables a fast development cycle and lower software maintenance costs.The problems associated with non-linear programming are overcome by integrating the scheduler with a simulator in the same modelling environment. This was used to produce initial conditions for the network scheduler.Furthermore, the reservoir trajectory governs the optimality of the solution. The resulting optimal continuous schedule is easily transformed into an optimal discrete schedule using a post-processing heuristic that follows the reservoir trajectory. Thecase study shows that this is valid and only marginally increases costs. The combination of network scheduler and simulator, together with other software applications, is an advance over previous implementations. The scheduler uses simulation modelsdirectly and it is faster and easier to set up and use.