Worldwide electricity consumption and the extent of peak electrical energy utilization are on the rise as societies everywhere become more and more developed. The increase of distributed renewable energy production, electric vehicles and large load consuming appliances generates unprecedented demand and volatile supply, posing a potential threat to the stability of electrical power grids. Demand side management (DSM) is a potential way to postpone grid reinforcements, and to prepare electrical power grids for the challenges that they face. DSM can be implemented by including a management system at the consumer end of the energy network, which automates the control of loads and power production within the household power network. Subsequently, electricity costs for the end user, total electricity demand or loads at peak load times can be reduced through ideal management. To be able to manage energy production, storage and significant loads as crucial components to the system, ideally, the components are analysed with respect to the underlying physical processes that occur during their operation. This analysis, combined with research into already existing home energy management systems (HEMS) found in literature, is then used to formulate a framework for testing and validating HEMS approaches. An agent-based architecture is proposed for the core of the evaluation system. This offers flexible addition of applications as agents, and provides a simple solution for building a system based on the framework. The management software is based on the open-source home automation platform Home Assistant, which is written in the Python programming language. Python offers a large ready-to-use library of modules and is an ideal tool to create distributed software for scientific applications. Therefore, it is the first choice for programming a flexible optimization and management module which lies at the heart of the HEMS evaluation framework. The final system is then validated in a simulation environment comprising a simulated photovoltaic module, an electric vehicle charging point, a battery energy storage system and simulated loads. Three different management approaches are illustrated using distinct management algorithms. These approaches are consequently compared using several key performance indicators. In a final step, it is shown that the proposed agent-based architecture can offer means for evaluating different HEMS approaches and algorithms.