Vortrag auf dem VDE ETG Congress am 28.11.2017 in Bonn
The transition towards an energy system with a high share of renewables creates new requirements for flexibility in the system. Residential consumers can contribute by adjusting their consumption behaviour if they are incentivized properly. This can be achieved by several variable pricing structures, which appear to have varying effectiveness.
Simulations of the customers’ reaction to price signals can help to evaluate the effects of different structures. Previous investigations show that conventional time-of-use pricing cannot be recommended; therefore, a more dynamic approach is examined here. Since this involves a lot of variable parameters, an optimization strategy for finding the best solution is also presented.
Residential consumers are characterized by measured individual load curves in the simulation. In order to quantify their reactions to given price signals, appliances which are considered suitable for demand-side management measures are identified in these load curves via a pattern recognition algorithm, and subsequently shifted to the economically optimal time of operation within an acceptable interval.
To implement a holistic approach, a general representation of relevant variable pricing structures is constructed. This allows simulating the whole range of electricity tariffs by adjusting different parameters of this formula.
The results of these simulations are evaluated with respect to different goals, like reduction of energy purchase costs, reduction of greenhouse gas emissions or grid relief. The optimal parameter set, i.e. the optimal tariff structure, is determined by mathematical optimization. Since this problem is highly nonlinear, most traditional optimization algorithms do not yield useful results in a reasonable amount of time. Therefore, several approaches like evolutionary optimization are investigated and compared.
Results and Outlook
The simulations show that the potential of dynamic pricing structures for appropriate flexibilization of residential consumption is significantly higher compared to simple time-of-use tariffs. Additional elements that incorporate short-term influences like renewable generation, wholesale prices or grid load prove useful for all investigated goals.
The optimization process for identification of the optimal parameter set is computationally quite intensive. The right choice of an optimization algorithm leads to tolerable runtime, but for larger test cases, parallelization might be necessary and should therefore be investigated.
Future work will also expand the simulation with an agent-based approach for assessment of tariff acceptance, since the described structure renders inapplicable for this question.