An agent based model of a win-win Demand Side Management program for Energy Communities

The effects of climate change are becoming more and more evident and a fossil-fuel based economy is no longer sustainable. Thus, it is important to abandon traditional power plants in favor of renewable energy sources, trying to change the demand in such way that it follows the available supply. The presence of renewable energy sources that depend on the weather creates uncertainty and can cause large energy imbalances. Moreover, different consumers' behavior, usually uncontrollable, are not easy to be managed and require energy management strategies. Among these demand management mechanisms, it is possible to find the demand-side management (DSM) and the demand-response (DR). In both cases, a key role is played by the willingness of the customers. Needless to say, that without their consensus, none of these methods can be applied. Regrettably, however, some of the studies focus on technical and economic aspects only, addressing this problem in the future. Others deal with the problem of user's discomfort without taking into account individuals' diversity. Still others do, but they do not include upstream cognitive processes. Indeed, a copious amount of articles does not consider that before deciding to accept a program event request, the customers must decide first whether to sign up for a DSM program. On the other hand, different papers conduct a survey to discover what factors affect the decision to participate or not to a DSM project but then they are not used in simulations . Thus, in this thesis, besides exploring possible optimization methods that allow RES exploitation and cost reduction, a special attention is given to the analysis of prosumers’ behavior, their mutual influence and the individuals’ response to requests at different times. The focus is only on the residential field which is particularly challenging since many households have to participate to have a remarkable impact on the market. The main goal is to learn the different prosumers’ tolerance to their loads shifting in order to propose shifts that will be accepted by the prosumers. Thus, to have a win-win situation where the aggregator maximizes its own profit and the prosumers maximize their utility, which is a combination of monetary gain and comfort. Therefore, an agent based framework based on MOSAIK and AIOMAS has been developed to simulate the several interactions among the aggregator agent, the market agent, the DSO agent and many prosumer agents. First, by cross-referencing cadastral data, census data and other information, households’ profile has been deduced. Second, the Theory of Planned Behavior has been applied to determine the signing of the DSM contract. Once in the program, requests’ acceptance has been studied through an algorithm inspired by Q-Learning. Pyomo and CPLEX have been used for the optimization. Thanks to the algorithm, prosumers’ participation and acceptance are enhanced.