| EngD trainee | Anand Thamban |
| Project | Fault detection and diagnosis of the low 鈭員 syndrome in cooling coils of chilled water systems |
| University supervisor | dr.prof. ir. Wim Zeiler |
| Company supervisor | ir. Alet van den Brink |
| Name of company | Kropman Installatietechniek |
| Period of project | September 2020 鈥� August 2023 |
Public summary
With the rising global temperatures around the world due to climate change, there is an increasing demand for cooling in indoor spaces to ensure that occupants can live and work in a comfortable environment. To ensure this, the chilled water systems in buildings are operated more frequently than usual to comply with the comfort requirements. With the prolonged and more frequent use of the equipment, the chance of faults occurring in the system increase. One such phenomenon which affects the performance of the chilled water system in buildings/distribution plants is the low 螖T syndrome. The main characteristics of the low 螖T syndrome are a reduced return water temperature and an increased mass flow rate through the cooling coil. These characteristics occur when certain faults are present in the system, which lead to a reduced cooling output. The consequences of the low 螖T syndrome are an increased energy consumption and/or inability to meet the cooling requirements leading to discomfort. To avoid both issues, a fault detection and diagnosis tool has been developed to detect the low 螖T syndrome swiftly. Bayesian networks are used to diagnose the various faults which can cause the low 螖T syndrome.
The tool is developed as a larger continuous monitoring tool to detect and diagnose faults in the HVAC system, with more focus on the cooling, heating and heat recovery system. The special focus of this project is on the development of an algorithm to detect the low 螖T syndrome. The tool is designed as an online webpage which can be hosted locally or on the cloud. It consists of multiple sub-pages for main alarms, diagnostic Bayesian network analysis and machine learning analysis. The tool is intended for use by multiple types of end users including HVAC experts, facility and building managers and machine learning experts, hosting special features for each of the end users. The tool provides a simple and easy to understand alarm system for when the low 鈭員 syndrome has been detected (red indicator for faulty conditions and green indicator for normal conditions), with multiple to-do actions and interactable graphs to assist the user in the final decision-making process. The architecture of the tool is shown in the figure below.
Funded by NWO project TransAct
