Validation of the SHARM model

This project has attempted to calculate the added benefit of applying stochastic programming to the short-term hydropower scheduling problem. Even in the short run, the important input parameters of electricity prices and inflow to the reservoirs are best modelled as stochastic processes. The results from deterministic models in use today require manual adaptation in order to compensate for the often too bold reservoir management strategy chosen by a decision support tool where uncertainty is not adequately modelled. These adjustments are based on expert knowledge of how the water courses should be operated under uncertain conditions, and require extra resources for the planning process in terms of time and skilled operators. The uncertain parameters could be considered in the optimization model by using stochastic programming, and such tools therefore have the ability to yield added value for producers in the sense of more robust production schedules that balance profit and risk. A stochastic model for short-term optimization for hydropower, the SHARM model, has been developed by SINTEF Energy Research. The gains obtainable to producers from using this model instead of the currently used deterministic model has been tested together with a group of Norwegian hydropower producers.

In the KMB project "Optimal Short-term Scheduling of Wind and Hydro Resources" a prototype for stochastic short-term optimization (SHARM) was developed. SHARM is a stochastic formulation of the successive linear programming method used in SHOP, where SHOP is the short-term scheduling tool used by most Norwegian hydropower companies. The SHARM model takes into account uncertainty in inflow to the reservoirs and in the spot price, and is capable of providing decision support and robust planning where severa l strategies must be weighted against each other. This is directly aimed towards the use of hydropower flexibility to balance larger volumes of intermittent generation. The project builds the bridge between the research prototype SHARM and the implemen tation projects required for operationalization of SHARM in the generation companies. Economic feasibility is one of the answers addressed in the project but mitigation of the obstacles for getting the SHARM concept operational is also included. A key act ivity in the proposal is the industry tests performed by and in the partner companies. The tests will be carried out in close cooperation with SINTEF, who will develop a test framework and implement updates to the SHARM model throughout the test phase. R&D challenges in the project are related to the design and development of the test framework that substitutes testing over many years of climate data. In addition the large amount of input and output data is a new challenge, in particular how to interpr et the results from a stochastic model. In the project, a decision will be taken regarding whether to continue the SHARM development towards an operational decision support tool, or towards an improved research tool. In any case, the project is expected to increase the competence pool regarding challenges in operational hydropower scheduling.