ºÚÁϸ£ÀûÍø

Information

Public Summary

In the built environment, most energy is used for comfort. The demand for cooling will increase sharply as a result of global warming, better thermal insulation, and the heat island effect. It is therefore increasingly important that cooling installations function optimally. Currently, there are many chilled water installations that suffer from the so-called low delta-T syndrome. The return water temperature from the installations is lower than predetermined and the difference with the supply temperature is smaller, low delta-T. This affects the efficiency of the chillers and/or heat pumps as well as the energy consumption of the pumps. The result is an energy consumption that is 20-40% higher for cooling. It is important to be able to detect and analyze the low-dT syndrome properly. The cooling coil in the Air Handling Unit, being the most common consumer of cooling, and a prominent location where the low delta-T syndrome has been observed, was chosen as the focus of this project. After a literature research to determine possible faults a building energy simulation program, EnergyPlus, was used to simulate the impact of low delta-T syndrome faults at the cooling coil. Next an experimental studies introducing some of the low delta-T syndrome faults was carried out in a real small-sized office building. Together, they revealed that reduced leaving air temperature setpoint and stuck valve fault cause the largest impact. Building management systems can be equipped with fault detection and diagnosis module for continuous monitoring of the performance of installations, and continuous commissioning (Cx). A software module has been developed that can use the data from a building management system to determine the low-dT syndrome and identify possible faults. This would ensure that the energy consumption of the cooling installations remains as low as possible. Within the project, the first prototype of such a module was built using a combination of 4S3F methodology based on Diagnostic Bayesian Network for fault diagnosis and Machine Learning based regression models.

Funded by: Topsector Energie TKI Urban Energy