This report addresses the implementation of an optimised control of the HYBUILD system energy flows in the residential buildings by considering internal and external requests.
The main aim of this document is to provide a detailed description of the energy management approaches adopted for the Building Energy Management System (BEMS) proposed within the HYBUILD project.
It presents the procedures and the features of the two control systems developed for addressing:
- the optimisation of the energy management for the provisioning of flexibility services to grid operators (provided by ENG);
- the minimisation of the energy operational costs (provided by UDL).
The two implemented optimisation processes adopt two different methods and pursue different objectives. The optimiser provided by ENG relies upon a multi-objective optimisation framework able to handle two or more objectives at the same time; this has been performed by the implementation of a heuristic algorithm, the Non-dominated Sorting Genetic Algorithm II (NSGA II). The appointed objectives are: the provision of flexibility services to the grid operators, the economic management of the energy operations, and the users’ comfort satisfaction.
UDL’s control strategy implements a reinforcement learning technique, a Deep Learning Control (DLC) algorithm characterised by a three-layer fully connected network. This tool focuses on the internal cost management of the energy operation of each device inside the building for reducing costs.
This report shows how it is possible to adopt different approaches for addressing the same energy operations from two different standpoints. Mostly, the reduction of the costs related to the energy flows among the systems and devices inside the building is always taken into account. The comfort of the building inhabitants is always one of the most referenced constrain of the processes, as well. In this case, the solutions proposed allow also to leverage on the storage systems, in particular the electrical battery and the latent storage, not only for handling internal energy management but also for addressing requests from electric and district heating grid operators for the provision of a flexibility service. This is part of a wider framework of Demand Response (DR) implementation in the field of building energy management.
In this view, a typical DR mechanism has been envisioned within the BEMS optimiser: a grid operator sends a flexibility service request signal that triggers the external request optimisation module of the BEMS. This signal consists of a power profile to be followed by the building controlled by the BEMS whilst absorbing electricity or district heating; moreover, a reward value is provided with it, in order to encourage economically the Energy Manager of the building because this reward corresponds to the economic prize granted if the service is actually delivered. This service request drives the optimisation: the algorithm will leverage on the capabilities of the storage systems installed within the HYBUILD buildings for optimising the provision of the services requested while taking into account the energy operations costs, thatare related with this service, and the building inhabitants’ comfort. At the end of the process, the Energy Manager has the possibility to choose between a set of optimised solution that are put at disposal by a Decision Support System. This tool indeed selects the most convenient solution for each aspect taken into account by the multi-objective framework and shows it in a dedicated dashboard where Energy Manager can assess the building performances and select one solution to be implemented in the building.
The delivery of the software reported in this document demonstrates how the participation to Demand Response (DR) programs could be feasible in this context, exploiting the flexibility allowed by the adoption of a HYBUILD solution for the energy management of the building.
According to the user preferences and demonstration needs, the DLC or the NSGA-II approach can be selected. In this perspective, the BEMS can be considered as the harmonization of these two different control strategies. The final version of the BEMS will integrate the final models of the building devices and its user interface. The implemented software will be tested by means of the simulation environment based on TRNSYS developed inside the same WP. This will give the possibility of testing and comparing the two control systems even before their development into the demo pilots. The results will be shown in a next report to be released in a few months (D4.4 – Report on system performance).
Read the full report
Download here the full report.