Artificial-neural-network-based model predictive control to exploit energy flexibility in multi-energy systems comprising district cooling

The integration of multi-energy systems to meet the energy demand of buildings represents one of the most promising solutions for improving the energy performance of the sector. The energy flexibility provided by the building is paramount to allowing optimal management of the different available resources. The objective of this work is to highlight the effectiveness of exploiting building energy flexibility provided by thermostatically controlled loads (TCLs) in order to manage multi-energy systems (MES) through model predictive control (MPC), such that energy flexibility can be regarded as an additional energy source in MESs. Considering the growing demand for space cooling, a case study in which the MPC is used to satisfy the cooling demand of a reference building is tested. The multi-energy sources include electricity from the power grid and photovoltaic modules (both of which are used to feed a variable-load heat pump), and a district cooling network. To evaluate the varying contributions of energy flexibility in resource management, different objective functions—namely, the minimization of the withdrawal of energy from the grid, of the total energy cost and of the total primary energy consumption—are tested in the MPC. The results highlight that using energy flexibility as an additional energy source makes it possible to achieve improvements in the energy performance of an MES building based on the objective function implemented, i.e., a reduction of 53% for the use of electricity taken from the grid, a 43% cost reduction, and a 17% primary energy reduction. This paper also reflects on the impact that the individual optimization of a building with a multi-energy system could have on other users sharing the same energy sources.

Smart controls for heat pumps are required to harness the full energy flexibility potential of building thermal loads. A literature review revealed that most strategies used for this purpose can be classified in two categories: simpler rule-based control (RBC), and model predictive control (MPC), a more complex strategy based on optimization and requiring a prior model of the systems. Both RBC and MPC can use external penalty signals to prompt their actions. The price of electricity is most often used for this purpose, leading to strategies of cost reduction. As an alternative penalty signal, a novel marginal CO2 emissions signals was also conceived. In this thesis, both an RBC and an MPC controllers were developed as supervisory controls for an air-to-water heat pump supplying the heating and cooling needs of a residential building type from the Mediterranean area of Spain. The RBC strategy modulates the temperature set-points, while the MPC strategy minimizes the overall summed penalties (costs or emissions) due to the heat pump use, while balancing with comfort constraints and a proper operation of the systems. The MPC controller in particular required the development of a simplified model of the building envelope and of the heat pump performance, both adjusted differently for heating or cooling. The MPC included several novelties, such as the mixed-integer formulation, the heat pump simplified model based on experimental data and the consideration of its computational delay. The developed controllers were then tested, firstly in an experimental “hardware-in-the-loop” setup, with a real heat pump installed in the laboratory facilities, and connected to thermal benches that emulated the loads from a building model. Implementing the control strategies on a real heat pump enabled to highlight some practical challenges such as model mismatch in the MPC, communication issues, interfacing and control conflicts with the heat pump local controller. Secondly, a simulation-only framework was developed to test other configurations of the controllers, with TRNSYS as the main dynamic building simulation tool, coupled with MATLAB for the MPC controller. In that case, the real heat pump was replaced by a detailed model which was specially developed for this purpose. It is based on static tests performed in the laboratory, and therefore reproduces the dynamic behavior of the heat pump with high fidelity. The results from experimental and simulation studies revealed the ability of both types of controllers to shift the building loads towards periods of cheaper or less CO2-emitting electricity, these two objectives being in fact contradictory. In the cases where the reference control presented a large margin for improvements, the RBC and MPC controllers performed equally and provided important savings: around 15% emissions savings in heating mode, and 30% cost savings in cooling mode. In the cases where the reference control already performed close to optimally, the RBC controller failed to provide improvements, while the MPC benefitted from its stronger optimization and prediction features, reaching 5% cost savings in heating mode and 10% emissions savings in cooling mode. The research carried out in this thesis covered many aspects of energy flexibility in buildings: creation of input penalty signals, graphical representation of flexibility, development of controllers, performance in realistic experimental setup, fitting of appropriate models and compared performance in heating and cooling. The development efforts and barriers hindering the deployment of MPC controllers at large scale for building climate control have additionally been discussed. The performance of the developed controllers was evidenced in the thesis, proving their potential for load-shifting incentivized by different penalty signals: they could become a strong asset to unlock demand-side flexibility and in fine, help integrating a larger share of RES in the grid.

Decarbonization plans promote the transition to heat pumps (HPs), creating new opportunities for their energy flexibility in demand response programs, solar photovoltaic integration and optimization of distribution networks. This paper reviews scheduling-based and real-time optimization methods for controlling HPs with a focus on energy flexibility in distribution networks. Scheduling-based methods fall into two categories: rule-based controllers (RBCs), which rely on predefined control rules without explicitly seeking optimal solutions, and optimization models, which are designed to determine the optimal scheduling of operations. Real-time optimization is achieved through model predictive control (MPC), which relies on a predictive model to optimize decisions over a time horizon, and reinforcement learning (RL), which takes a model-free approach by learning optimal strategies through direct interaction with the environment. The paper also examines studies on the impact of HPs on distribution networks, particularly those leveraging energy flexibility strategies. Key takeaways suggest the need to validate control strategies for extreme cold-weather regions that require backup heaters, as well as develop approaches designed for demand charge schemes that integrate HPs with other controllable loads. From a grid impact assessment perspective, studies have focused primarily on RBCs for providing energy flexibility through HP operation, without addressing more advanced methods such as real-time optimization using MPC or RL-based algorithms. Incorporating these advanced control strategies could help identify key limitations, including the impact of varying user participation levels and the cost-benefit trade-offs associated with their implementation.

Economic evaluation of a hybrid heating system in different climate zones based on model predictive control

The transition towards air-to-water heat pumps in European buildings supports decarbonization goals but poses challenges to electrical grid infrastructure during cold periods. Distribution system operators in several European countries, e.g., Germany and Switzerland, may impose power constraints to maintain grid stability, affecting heat pump operation and potentially thermal comfort. To mitigate thermal discomfort during these periods, buildings’ thermal mass and a buffer storage can be exploited for flexibility. To evaluate their flexibility potential, we combine a high-order model of a single-family house in EnergyPlus with a monovalent air-to-water heat pump system model in Modelica using Spawn of EnergyPlus. We compare a state-of-the-art rule-based control (RBC) strategy with a model predictive control (MPC) under varying buffer storage sizes and grid-constrained periods. For the MPC implementation, we use a simplified one-zone building model in Casadi as the process model of the building. The MPC optimizes control decisions based on perfect weather forecasts and scheduled grid-constrained periods in a co-simulation with the detailed building energy system. Results show that systems sized according to standards considering grid-constrained periods, e.g., VDI 4645, ensure thermal comfort with RBC while recommendations for buffer storage sizing can potentially be reduced by 16% to 50%. MPC can maintain thermal comfort during heat pump deactivation periods of up to 6 h each morning, even with buffer storage sizes that follow standards which do not consider grid-constrained periods, e.g., EN 15450. MPC requires only 76 L for 5 h and 223 L for 6 h heat pump deactivation time.

The aim of this paper is to mitigate the problem of high power demand peak and load oscillations in the operation of a large population of thermostatically controlled loads (TCLs) operated by model predictive control (MPC) at the TCL level. Two desynchronized MPC schemes are introduced: 1) adding random delays in reference signals and 2) extra penalizations on MPC objective functions. For characterizing and validating the proposed desynchronization MPC schemes, a partial differential equation (PDE) model is developed to represent the evolution of the operational states of the TCLs controlled by MPC in a population. The focus of this paper is put on the control of cooling fans in server racks of datacenters, whereas the proposed approach is applicable to other types of TCLs. Numerical simulation studies are carried out and the obtained results confirm the validity and the applicability of the developed approach.

The scheduling of Thermostatically Controlled Loads (TCLs) in a residential demand response (DR) context requires scalability as TCLs come in many forms. Model-free control algorithms, such as Fitted Q-iteration (FQI), can provide such scalability. They accommodate the variety in e.g. electric water heaters or heat pumps available on the market and thermal characteristics of residential buildings. However, these approaches require a significant amount of data. In the building simulation literature, Model Predictive Control (MPC) with grey-box models provides a technique that connects domain knowledge with experimental data, which results in algorithms with a higher sample efficiency. This work proposes to combine grey-box MPC and FQI in the Informed FQI method and shows how to use MPC to provide domain knowledge to the model-free FQI algorithm. The new approach is compared to FQI and MPC with a linear RC-model in terms of cost and user comfort.

In this paper, we present a flexibility estimation mechanism for buildings’ thermostatically controlled loads (TCLs) to enable the distribution level consumption of the majority of solar photovoltaic (PV) generation by local building TCLs. The local consumption of PV generation provides several advantages to the grid operation as well as the consumers, such as reducing the stress on the distribution network, minimizing voltage fluctuations and two-way power flows in the distribution network, and reducing the required battery storage capacity for PV integration. This would result in increasing the solar PV generation penetration levels. The aims of this study are twofold. First, spectral (frequency) analyses of solar PV power generation together with the power consumption of multiple building TCLs (such as heating, ventilation, and air conditioning (HVAC) systems, water heaters, and refrigerators) are performed. These analyses define the bandwidth over which these TCLs can operate and also describe the PV generation frequency bandwidth. Such spectral analyses, in frequency domain, can help identify the flexible components of PV generation that can be consumed by the various TCLs through optimal building load utilization. Second, a quadratic optimization problem based on model predictive control is formulated to allow consuming most of the low and medium frequency content of the PV power locally by building TCLs, while maintaining occupants’ comfort and TCLs’ physical constraints. The solution to the proposed optimization problem is achieved using optimal control strategies. Numerical results show that most of the low and medium frequency content of the PV generation can be consumed locally by building TCLs. The remaining high-frequency content of the PV generation can then be stored/offset using energy storage systems.

This study presents an innovative optimization method for resource scheduling in multi-energy storage systems, focusing on improving resource allocation while considering supply-demand flexibility and renewable energy integration. As renewable energy gains popularity and multi-energy systems become more complex, effective utilization of energy storage to achieve supply-demand balance, optimize energy scheduling, and maximize renewable energy integration is crucial. To address this challenge, a Markov dynamic model is developed to capture the dynamic changes in energy supply and demand within the multi-energy storage system. The model is then solved using a reinforcement learning approach to optimize resource scheduling decisions. Numerical simulations and case studies are conducted to validate the effectiveness and feasibility of the proposed method, showcasing its potential to enhance operational efficiency and reliability in multi-energy storage systems amidst constantly changing energy patterns. This research provides valuable insights and decision support for the design and operation of multi-energy storage systems, contributing to the advancement of sustainable energy utilization and promoting sustainable development in the energy sector.

This article presents a 125-day experiment to investigate model predictive heat pump control. The experiment was performed in two parallel operated systems with identical components during the heating season. One of the systems was operated by a standard controller and thus represented a reference to evaluate the model predictive control. Both test rigs were heated by an air-source heat pump which is influenced by real weather conditions. A single-family house model depending on weather measurement data ensured a realistic heat consumption in the test rigs. The adapted model predictive control algorithm aimed to minimize the operational costs of the heat pump. The evaluation of the measurement results showed that the electrical energy demand of the heat pump can be reduced and the coefficient of performance can be increased by applying the model predictive controller. Furthermore, the self-consumption of photovoltaic electricity, which is calculated by means of a photovoltaic model and global radiation measurement data, was more than doubled. Consequently, the energy costs of heat pump operation were reduced by 9.0% in comparison to the reference and assuming German energy prices. The results were further compared to the scientific literature and short-term measurements were performed with the same experimental setup. The dependence of the measurement results on the weather conditions and the weather forecasting quality are shown. It was found that the duration of experiments should be as long as possible for a comprehensive evaluation of the model predictive control potential.

A multi-energy system that controls the generation and storage of different types of energy offers great potential for improving the overall system efficiency and operating cost. The economic dispatch problem for multi-energy systems is to find the optimal system configuration that satisfies the demands under a host of constraints. The problem is often formulated as a mixed integer programming problem, which is very hard to solve since the number of variables involved could be in the order of hundreds. The difficulty of the problem is further compounded when model predictive control is implemented because the set points for controlling the multi-energy system at each time step have to be modeled, drastically increasing the number of variables in the optimization problem. In this paper, a large-scale hybrid optimization strategy is proposed to solve the economic dispatch in a multi-energy management system. Mixed-integer linear programming is used to solve a linearly approximated problem to determine the discrete set points followed by nonlinear programming for the continuous variables. We compare different optimization methods for the nonlinear programming problem with respect to the computational time and cost savings. The cost savings is also compared to a rule-based economic dispatch as the baseline. The simulation is performed using the data obtained from an office building in an R modeling and data management environment. We show that the proposed optimization strategy is able to solve the economic dispatch problem of a multi-energy management system with 720 mixed-integer variables within a short timeframe while still accounting for the nonlinearity of the optimization problem.

Model predictive control (MPC) is promising for optimizing building's operation but high hardware, software and know-how requirements impede its commercialization. Therefore, rule-based controllers (RBC) are state-of-the-art. Approximate MPC (AMPC) can help bridge this gap by replacing the optimization with an explicit functional relation called approximator. Literature lacks reproducible use cases and benchmarks and a comparison of sophisticated and traditional approximators. This study aims to close this gap by applying AMPC to BOPTEST's two-zone heat pump testcase. The BOPTEST testcase includes predefined RBCs and KPIs promoting repeatability. Then, a comparison was made between artificial neural networks (ANNs), random forest (RF), linear, and logistic regression. The results show that feature selection significantly affects the performance. After adapting the features, the ANNs and RF outperform the RBC with cost savings of up to 33% and discomfort reductions of 70%, while requiring 15% of the MPC's computation time. The traditional approximators fail to outperform the RBC.

In the energy transition, multi-energy systems are crucial to reduce the temporal, spatial and functional mismatch between sustainable energy supply and demand. Technologies as power-to-heat (PtH) allow flexible and effective utilisation of available surplus green electricity when integrated with seasonal heat storage options. However, insights and methods for integration of PtH and seasonal heat storage in multi-energy systems are lacking. Therefore, in this study, we developed methods for improved integration and control of a high temperature aquifer thermal energy storage (HT-ATES) system within a decentralized multi-energy system. To this end, we expanded and integrated a multi-energy system model with a numerical hydro-thermal model to dynamically simulate the functioning of several HT-ATES system designs for a case study of a neighbourhood of 2000 houses. Results show that the integration of HT-ATES with PtH allows 100% provision of the yearly heat demand, with a maximum 25% smaller heat pump than without HT-ATES. Success of the system is partly caused by the developed mode of operation whereby the heat pump lowers the threshold temperature of the HT-ATES, as this increases HT-ATES performance and decreases the overall costs of heat production. Overall, this study shows that the integration of HT-ATES in a multi-energy system is suitable to match annual heat demand and supply, and to increase local sustainable energy use.

Fifth Generation District Heating and Cooling (5GDHC) networks, in which low temperature water is distributed to water-source heat pumps (WSHPs) in order to meet thermal demands, are expected to have a significant impact on the decarbonisation of energy supply. Thermal storage installed in these networks offers operational flexibility that can be leveraged to integrate renewable electrical and thermal energy sources. Thus, when considered as part of a smart multi-energy district, 5GDHC substation devices (e.g., WSHPs, storage) may be optimally operated using Model Predictive Control (MPC) in order to match demand with low-cost supply of electricity. However, the application of MPC requires the ability to model 5GDHC networks within the context of a multi-energy system. Hence, this paper extends an existing, generalised control-oriented modelling framework for multi-energy systems to accommodate 5GDHC networks. Additions include the ability to represent hydraulic pumps, thermodynamic cycle devices (such as WSHPs) and multi-energy networks within the framework. Furthermore, an economic MPC (eMPC) scheme is proposed for energy management of 5GDHC-based smart districts. Finally, a case study is presented in which the proposed eMPC controller is compared with rule-based control for economic operation of a smart district.

Model predictive controls for residential buildings with heat pumps: Experimentally validated archetypes to simplify the large-scale application