Bachelor's Thesis
In this work, a two-stage stochastic programming model for the optimal scheduling of a domestic electric water heater with respect to electricity costs and thermal comfort under a day-ahead real-time pricing scheme is developed. It is based on an existing deterministic linear programming model that optimizes heating on average by use of a mean hot water withdrawal profile. The performance of the stochastic model is evaluated by conducting simulative experiments using recorded hot water consumption data and electricity prices from the power exchange spot market. Three approaches are developed for use with the stochastic model that generate scenarios by (a) using historical data; (b) running Monte Carlo simulations; (c) dynamically extracting parts of a scenario tree. All tested approaches yield an average combined electricity and discomfort cost improvement of approximately 17% in comparison to the deterministic model. The scenario tree based approach achieves the best results with regard to cost improvements and runtime, despite a higher modeling complexity.
| Start date | 28. January 2018 |
| End date | 16. May 2018 |
| Supervisor |
Dr.-Ing. Tobias Lübkert
|