Smart Controllers Enable Innovative and Sustainable Residential Thermal Network Design and Operation

Vincent Bellinkx, Stef Gijsbregts, Lucas Verleyen, Lieve Helsen, KU Leuven, Belgium

Smart controlled thermal networks offer a promising pathway to sustainable residential heating and cooling. This article shows how model predictive control is used to optimise performance across cost, emissions, and efficiency in four collective thermal energy systems. The results reveal that a smart controlled two-pipe network with variable temperature operation achieves optimal performance, reducing network heat losses and investment costs. Simulation-based sizing using smart control demonstrates a remarkable 69% reduction in required heat pump capacity compared to conventional methods. These findings show that smart control is an important technology for developers implementing collective thermal energy systems in residential developments.

Introduction

The newer generation district heating networks are designed to operate at lower temperatures, with some systems even approaching ambient temperatures. These lower temperatures facilitate the integration of a broader range of distributed renewable energy sources and the integration of prosumers. At the same time, the relative share of energy demand for domestic hot water is expected to rise due to increased levels of renovation and insulation. Consequently, the importance of domestic hot water production grows, necessitating a careful balance: network temperatures must remain sufficiently high to enable effective temperature boosting for domestic hot water supply, while also being low enough to maximise overall system efficiency and minimise heat losses.

Designing integrated residential energy systems is complex and challenging, particularly when selecting the optimal system configuration. Modern district heating concepts span a wide spectrum: from ambient-temperature networks relying on decentralised heat pumps, to higher-temperature systems capable of directly supplying domestic hot water, and multiple-pipe networks separating space heating, space cooling, and domestic hot water. Additionally, controlling these complex systems is very challenging. Fortunately, smart (heat pump) controllers enable the coordination of multiple heat sources and the anticipation of demand patterns, weather predictions, and energy costs, significantly enhancing system performance.

Four System Configurations Are Investigated

Although numerous system configurations could be considered, four main concepts are investigated. Concept 1, illustrated in Figure 1, features a network with one supply and one return pipe operated at ambient temperature. This network serves as a heat source for the decentralised heat pumps that create domestic hot water and space heating in the dwellings. The central heat pump maintains the network temperature within its boundaries. The main advantage of this configuration is its ability to directly provide space cooling through the network, while simultaneously repurposing the heat extracted during cooling to support domestic hot water production.

Concept 2, represented in Figure 2, features an intermittent-temperature network consisting of one supply and one return pipe with varying temperature levels. For most of the day, the network operates at a temperature suitable for space heating. However, at specific times, the smart controller increases the temperature to optimally charge the domestic hot water tanks in the dwellings. In this configuration, the central air-source heat pump and ground-source heat pump are installed in parallel before the central tank. This setup allows each heat pump to operate during its most efficient period to provide higher supply temperatures: the ground-source heat pump primarily operates in winter thanks to higher ground temperatures, and the air-source heat pump mainly operates in summer thanks to higher air temperatures. A major drawback of this concept is its inability to provide space cooling through the network, necessitating individual air-conditioning units in each dwelling.

Concept 3, illustrated in Figure 3, features a three-pipe network consisting of one supply pipe for space heating, one supply pipe for space cooling, and one return pipe. The network is connected to two central tanks: one tank is maintained at a temperature suitable for space heating, and the other tank at a temperature level appropriate for space cooling. The space heating supply also serves as the heat source for the decentral heat pumps in the dwellings to produce domestic hot water.

Concept 4, represented in Figure 4, features a four-pipe network consisting of one supply pipe for space heating, one supply pipe for space cooling, one supply pipe for domestic hot water, and one return pipe. In this configuration, all energy services are supplied via the network using three central tanks, each maintained at a temperature suitable for space heating, space cooling, or domestic hot water, respectively. Consequently, no individual production units are required within the dwellings, limiting the investment costs at the building level. The combination of a central air-source and ground-source heat pump is also possible in this case; however, it is not shown to limit the complexity of the drawing.

MPC as a Smart Controller

The collective thermal energy systems are controlled using Model Predictive Control (MPC), a smart control strategy that leverages the dynamic behaviour of system components to optimise performance. Unlike traditional rule-based control approaches that react to current conditions, MPC anticipates future events and optimises control decisions accordingly.

MPC offers several key advantages for controlling complex thermal systems. It can handle non-linear, physics-based models that accurately represent the dynamic behaviour of heat pumps, thermal storage tanks, and building envelopes. Additionally, MPC incorporates predictions of future disturbances and demands, enabling proactive system adjustments rather than reactive responses.

Traditional MPC uses a feedback loop to correct for prediction and model uncertainties. However, this study implements a specific variant of MPC that assumes perfect predictions and a perfect system model, eliminating the need for this feedback loop. This leads to a best-case scenario, allowing focus on fundamental performance differences between system configurations under optimal control conditions without uncertainties.

The MPC controller minimises a multi-criteria objective function that simultaneously maximises energy efficiency, minimises operational costs, and minimises CO₂ emissions, while maintaining thermal comfort. The controller thereby shifts operation to periods of higher heat pump COPs, lower electricity prices, and cleaner grid electricity. The controller incorporates real-time predictions of user demand profiles, weather conditions, electricity prices from the Belgian day-ahead market, and grid CO₂ intensity. By anticipating periods of optimal conditions, the system can preheat thermal storage and the thermal mass of buildings and the network during periods of low-cost or clean energy availability, thereby reducing both costs and emissions while maintaining user comfort.

The control strategy is implemented in Modelica using component models from the IDEAS and Buildings libraries, with the Toolchain for Automated Control and Optimisation providing the optimisation framework. A schematic representation of this smart control is shown in Figure 5.

Operation of the Smart Controller

The smart controller analysis reveals strong correlations between cost minimisation and CO₂ emissions reduction. For this purpose, three different objective functions were tested: one aimed at only minimising cost, another at only minimising CO₂ emissions, and a third at only maximising efficiency, all while maintaining thermal comfort. Figure 6 illustrates that the variations in key performance indicators remained minimal, typically within 1-2% of each other. Consequently, the smart controller aligns the cost minimisation of developers with the emission reduction goals.

This strong correlation in the results stems from two reasons. First, the operational costs and CO₂ emissions are inherently linked to the electricity generation mix. During periods of high renewable energy generation, both electricity prices and the grid’s CO₂ intensity are low. Second, the smart controller exploits the system dynamics to shift energy demand in time. Figure 7 demonstrates the controller’s ability to shift the heat pumps’ operation to low-price periods when optimising for costs.

Smart controller operates a three-pipe network as a two-pipe network

The analysis of the smart controller behaviour showed that the three-pipe network (concept 3) could be effectively simplified to a two-pipe configuration. In the residential use case studied, space heating and space cooling demands rarely occurred simultaneously, making the separate space cooling pipe redundant. Furthermore, the space heating pipeline naturally operated at lower temperatures suitable for cooling during summer, since the recovered heat from space cooling could be reused for domestic hot water production, while increasing energy efficiency.

These findings led to concept 3*, illustrated in Figure 8, which is an improved version of concept 3. Concept 3* features a network with a single supply pipe that switches supply temperatures between a temperature adequate for heating in winter and cooling in summer. This configuration reduced network losses by approximately 25%, while also reducing investment costs by 20%. The smart controller’s ability to anticipate weather, demand patterns, electricity prices, and emissions enabled this simplification without compromising thermal comfort or system performance.

Simulation-based sizing of the central heat pumps reduces installed capacity by 69%

In collective energy systems, investment costs constitute a significant portion of total system costs, making appropriate component sizing crucial for economic viability. The central heat pumps were sized through a simulation-based approach that systematically evaluated different sizes through full-year simulations, including the smart controller, offering substantial advantages over conventional rule-based methods described by standards.

For the improved configuration (concept 3*), conventional sizing would require approximately 65 kW of thermal production capacity for three dwellings, while simulation-based sizing demonstrated that only 20 kW was sufficient to maintain thermal discomfort, leading to a 69% reduction in heat pump capacity and significant cost savings. As mentioned before, perfect predictions and a perfect system model were assumed. For real-world implementations, safety factors or stress testing under extreme conditions should be included to account for uncertainties.

In hybrid heat pump setups, the sizing method also identifies how the capacity is divided between the air-source and ground-source heat pumps. When both space heating and domestic hot water are provided centrally (concepts 2 & 4), an air-source and ground-source heat pump with the same capacity perform best, as each heat pump can operate at its most optimal time. However, when only space heating is produced centrally and domestic hot water decentrally (concepts 1, 3 & 3*), a system with only a ground-source heat pump performs best. In this configuration, the air-source heat pump remains underutilised during the summer months when no central heat demand is required, leading to a significant cost increase.

Two-pipe network shows the best techno-economic performance

The techno-economic comparison revealed significant performance differences across the various network configurations investigated. Figure 9 demonstrates the relationship between annual total costs and system efficiency, represented by the overall SCOP (calculated as the total heat supplied by the heat pumps divided by their total electricity demand), with the improved two-pipe network (concept 3*) achieving the best balance of lower total costs and favourable SCOP values.

Network heat losses in the different system configurations

Network heat losses varied substantially across configurations based on operating temperatures, as shown in Figure 10. The ambient-temperature networks (concept 1) and improved two-pipe systems (concept 3*) achieved the lowest heat losses at approximately 9% of the centrally produced heat, thanks to their lower operating temperatures and simplified piping configurations. In contrast, the intermittent-temperature networks (concept 2) and four-pipe systems (concept 4) exhibited the highest losses at 18-24% due to elevated operating temperatures and multiple supply lines.

Conclusions

The use of model predictive control as a smart controller in thermal networks offers many advantages.  This article focuses on two key benefits. First, MPC enables the development of innovative concepts when included in the design process. In this way, a three-pipe network (concept 3) was reduced into an improved two-pipe network (concept 3*), emerging as the most promising solution with the best balance of technical performance and economic viability. Additionally, including optimal control in a simulation-based sizing approach leads to a reduction of the central heat pump capacity by up to 69% and significant cost savings compared to conventional methods. Second, the MPC controller minimises a multi-criteria objective function that simultaneously maximises energy efficiency, minimises operational costs, and minimises CO₂ emissions, while maintaining thermal comfort. However, the smart controller aligns the cost minimisation of developers with the emission reduction goals. These findings provide valuable guidance for evaluating collective thermal energy systems and demonstrate that, fortunately, economically driven decisions align with environmental objectives, supporting the decision-making of policymakers, developers, and practitioners.

Author contact information

References

[1]. This article is based on: Bellinkx, V., Gijsbregts, S., Michiels, E. (sup.), Verleyen, L. (sup.), Rogmans, S. (sup.), Helsen, L. (sup.). Techno-Economic Analysis of Domestic Hot Water Production in Optimally Controlled and Simulation-Based Sized Collective Thermal Energy Systems. Master’s Thesis. KU Leuven, Belgium. 2025.