Development of new building HVAC control algorithms has grown due to needs for energy efficiency and operational flexibility. However, case studies demonstrating new algorithms are largely individualized, making algorithm performance difficult to compare directly. In addition, the effort and expertise required to implement case studies in real or simulated buildings limits rapid prototyping potential. Therefore, this paper presents the Building Optimization Testing Framework (BOPTEST) and associated software for simulation-based benchmarking of building HVAC control algorithms. A containerized run-time environment (RTE) enables rapid, repeatable deployment of common building emulators representing different system types. Emulators use Modelica to represent realistic physical dynamics, embed baseline control, and enable overwriting supervisory and local-loop control signals. Finally, a common set of key performance indicators are calculated within the RTE and reported to the user. This paper details the design and implementation of software and demonstrates its usage to benchmark a Model Predictive Control strategy.
27 May 2020 in Journal of Building Performance Simulation
Predictive controllers can greatly improve the performance of energy systems in buildings. An important challenge of these controllers is the need of a building model accurate and simple enough for optimization. Grey-box modelling stands as a popular approach, but the identification of reliable grey-box models is hampered by the complexity of the parameter estimation process, specifically for multi-zone models. Hence, single-zone models are commonly used, limiting the performance and applicability of the predictive controller. This paper investigates the feasibility of the identification of multi-zone grey-box building models and the benefits of using these models in predictive control. For this purpose, the parameter estimation process is split by individual zones to obtain an educated initial guess. A virtual test case from the BOPTEST framework is contemplated to assess the simulation and control performance. The results show the relevance of modelling thermal interactions between zones in the multi-zone building.
Optimal controllers can enhance buildings’ energy efficiency by taking forecast and uncertainties into account (e.g., weather and occupancy). This practice results in energy savings by making better use of energy systems within the buildings. Even though the benefits of advanced optimal controllers have been demonstrated in several research studies and some demonstration cases, the adoption of these techniques in the built environment remains somewhat limited. One of the main reasons is that these novel control algorithms continue to be evaluated individually. This hampers the identification of best practices to deploy optimal control widely in the building sector. This paper implements and compares variations of model predictive control (MPC), reinforcement learning (RL), and reinforced model predictive control (RL-MPC) in the same optimal control problem for building energy management. Particularly, variations of the controllers’ hyperparameters like the control step, the prediction horizon, the state-action spaces, the learning algorithm, or the network architecture of the value function are investigated. The building optimization testing (BOPTEST) framework is used as the simulation benchmark to carry out the study as it offers standardized testing scenarios. The results reveal that, contrary to what is stated in previous literature, model-free RL approaches poorly perform when tested in building environments with realistic system dynamics. Even when a model is available and simulation-based RL can be implemented, MPC outperforms RL for an equivalent formulation of the optimal control problem. The performance gap between both controllers reduces when using the RL-MPC algorithm that merges elements from both families of methods.
Building predictive control has proven to achieve energy savings and higher comfort levels than classical rule-based controllers. The choice of the model complexity needed to be used in model-based optimal control is not trivial, and a wide variety of model types is implemented in the scientific literature. This paper shares practical aspects of implementing different control-oriented models for model predictive control in a building. A real thermally activated test building is used to compare the white-, grey-, and black-box modeling paradigms in prediction and control performance. The experimental results obtained in our particular case reveal that there is not a significant correlation between prediction and control performance and highlight the importance of modeling the heat emission system based on physics. It is also observed that most of the complexity of the physics-based model arises from the building envelope while this part of the building is the most sensitive to weather forecast uncertainty.