A comparison of thermal power plant modeling approaches and their influence on electricity storage demand is provided.
Mixed-integer unit-commitment was found to be superior in scenarios with low shares of variable, renewable electricity.
Linear programming with merit order dispatch is sufficient in highly renewable and granular power plant capacity mixes.
Flexibility requirements in prospective energy systems will increase to balance intermittent electricity generation from renewable energies. One option to tackle this problem is electricity storage. Its demand quantification often relies on optimization models for thermal and renewable dispatch and capacity expansion. Within these tools, power plant modeling is typically based on simplified linear programming merit order dispatch (LP) or mixed-integer unit-commitment with economic dispatch (MILP). While the latter is able to capture techno-economic characteristics to a large extent (e.g. ramping or start-up costs) and allows on/off decision of generator units, LP is a simplified method, but superior in computational effort.
We present an assessment of how storage expansion is affected by the method of power plant modeling and apply a cost minimizing optimization model, comparing LP with MILP. Moreover, we evaluate the influence of wind and photovoltaic generation shares and vary the granularity of the power plant mix within MILP.
The results show that LP underestimates storage demand, as it neglects technical restrictions which affect operating costs, leading to an unrealistically flexible thermal power plant dispatch. Contrarily, storage expansion is higher in MILP. The deviation between both approaches however becomes less pronounced if the share of renewable generation increases.
- Renewable energy;
- Storage demand;
- Economic dispatch;
- Merit order;
- Expansion planning
© 2016 Elsevier Ltd. All rights reserved.