Industrial refrigeration systems are known to consume approximately 17% of electrical energy, a figure that is projected to rise in the future. This high energy consumption contributes to global warming and environmental degradation since conventional sources of energy are typically utilized for electricity generation. Moreover, the energy-intensive nature of industrial refrigeration systems leads to increased costs for major food and beverage industries. Consequently, optimizing the energy efficiency of these systems becomes crucial.
In this research, the application of Digital Twin (DT) technologies was explored, which have demonstrated effectiveness in various areas such as supply chain streamlining and system optimization. By combining physical and virtual spaces, DT and big data analytics can facilitate energy performance evaluation and optimization. The literature review identified three categories of DT models: physics-based, empirical, and data-driven. Considering their accuracy and efficiency, empirical models were recommended for developing DT models, while data-driven models proved useful for performance prediction applications. It was recommended to establish empirical equations based on correlation analysis by adjusting higher degree terms for accuracy. Additionally, input-output parameters for the DT should be tailored to the specific application and equipment. The literature study showed the possible identification of energy performance deviations, their root causes, and potential optimizations, including equipment optimization, load sharing among parallel equipment, and optimization of condenser set points and defrosting time.
This thesis research focuses on three industrial refrigeration plants: the Verkade Plant, the LST Plant, and the GIST Plant. For the Verkade Plant, empirical models were developed and validated for the screw compressor, evaporator, and evaporative condenser. An algorithm for condenser optimization was proposed and tested, while deviations in evaporator performance were analyzed. Similar models were developed and validated for the LST and GIST Plants, enabling the prediction of equipment performance. The predicted results were compared to actual plant performance, and deviations were carefully examined. Furthermore, optimization techniques were applied to improve equipment efficiency.
The thesis research findings indicate that the empirical models for each equipment piece at the Verkade Plant achieved an accuracy within a 5% error range, suggesting their suitability for analyzing the other two plants. The proposed condenser optimization algorithm has the potential to annually save 7% of energy, resulting in savings of 32 MWh of electrical energy and 11 tonnes of CO2 emissions. The application of the proposed optimization techniques to the LST and GIST Plant resulted in a significant reduction in energy consumption. It was determined that these techniques can achieve savings of approximately 13% and 14% in total energy consumption, corresponding to 200 MWh and 170 MWh of electrical energy, as well as 70 tonnes and 60 tonnes of CO2 emissions, respectively. These energy savings contribute to the reduction of CO2 released into the atmosphere, aligning with the goals of the Paris Agreement. Consequently, this research offers valuable insights into mitigating global warming through the optimization of industrial refrigeration systems using DT technology.