Data driven analysis of district cooling substations for performance diagnosis

Abstract

District cooling (DC) is a crucial technology for providing cooling for purposes such as space cooling for indoor environment, industrial processes, food storage industry. To improve the efficiency of the district cooling system (DCS) periodic maintenance of the components in the system is required. This requires detection of faults in the system and addressing these faults. Previously, fault detection has been done in district heating (DH) system but not as much on district cooling (DC). In this work, a large scale data driven approach is implemented to identify the performance of district cooling substations and detect faults. The scope of this study includes all 180 substations from the city of Gothenburg, Sweden and the operational data from the primary side of the substation is analysed. Different signatures are analysed namely, energy signature, delta signature and re- turn temperature signature to identify faulty behaviour in the substations. From previous research and domain knowledge, the most suitable methods for the data analysis are chosen and applied. Out of 180 substations, customer categories for 41 buildings are identified and their delta T signatures are studied for performance diagnosis. The results show that the method implemented in this study can be used in the fault detection domain for DC substations on a large scale. A mix of domain knowl- edge and data driven tools are required to detect faults and analyse performance.