Digital twin-based services for large-scale heat pump systems

Limiting climate change requires a rapid and sustained reduction in greenhouse gas emissions from energy systems. Expanding the use of low-to-zero-carbon electricity sources for heating has the potential to reduce emissions from buildings and industry. Large-scale heat pumps are an attractive technology for heat production, given that they enable system integration, renewable energy source utilization, and excess heat recovery. In contrast to household heat pumps, large-scale heat pumps have higher reliability requirements and increased complexity. To ensure the expected operation of a growing number of large-scale heat pumps in future heating systems, companies responsible for their operation and maintenance, will benefit from dedicated services for maintaining their reliability. Digital technologies such as digital twins hold the potential to outperform conventional surveillance and control services for industrial applications. However, the use of digital twins in the heat pump industry remains limited.

The present thesis investigated the use of digital twin-based services for ensuring the reliable and efficient operation of large-scale heat pumps. The work was divided into four parts. The first part characterized common challenges in the operation of large-scale heat pumps through information gathered from a literature review and a survey-based study. The most frequent faults described in large-scale heat pumps were fouling of the evaporator and source heat exchanger, along with refrigerant leakage. The collected information also indicated that operational parameters that can be applied for fault identification were monitored but not used for the prevention of faults. This underscored the potential for developing monitoring and predictive maintenance services that leverage existing surveillance in large-scale heat pumps.

The second part focused on the development and implementation of digital twin-assisted online monitoring for a large-scale heat pump system prone to fouling. Data retrieved remotely from the system enabled to develop and validate heat pump thermodynamic simulation models. An online calibration method was applied for the recurrent adjustment of model parameters related to fouling, after design parameters in a model were initially calibrated. The results demonstrated that the online calibration decreased simulation errors when compared to the initial calibration alone, regardless of the simulation model used. Moreover, the online calibration provided an estimation of fouling effects in the evaporator and the effectiveness of fouling mitigation procedures.

In the third part, the simulation models calibrated online and the large-scale heat pump system from the second part were applied for the evaluation of frameworks for set point optimization and operation scheduling in that system. The optimization of the intermediate pressure set point in the heat pump resulted in a performance increase of up to 3 % compared to the existing set point. The operation scheduling framework, as a replacement for the conventional operation of the system, led to a decrease in operational costs up to 5 %. The improvements related to the set point optimization and operation scheduling frameworks decreased for nonzero levels of fouling, which was characterized through the online calibration models.

In the fourth part, three predictive maintenance frameworks based on digital twins were developed for three case studies, namely the system from the second part, a different large-scale heat pump system affected by fouling, and a large-scale chiller. The first and second frameworks enabled to define a cost-optimal schedule for mitigating fouling and for predicting its growth over time, respectively. In the third framework, the simulation of the faulty operation of the chiller enabled to identify faults like refrigerant leakage, condenser fouling and reduced flow rate in the sink stream. Overall, the thesis highlighted the potential for improving the energy efficiency and reliability of large-scale heat pump systems via digital twin-based services. Future areas of development were derived from this research exploring further opportunities for heat pump digitalization.