Fault detection and diagnosis through tool based expert rule methodology

Abstract: This thesis aims to develop a set of expert rules for fault detection and diagnosis,specifically for the Korsningen HVAC system, which has different sections, includingheat pumps, ventilation, supply/return hot and cold, chill beams, and radiators.The analysis focuses on identifying the most important sensors and other componentsthat provide valuable information to compress the data and improve systemreliability, reduce downtime, and increase efficiency. The FDD methodology was alsoimplemented, including determining areas for improvement, developing hypotheses,data classification, hypotheses testing, issuing solutions or maintenance suggestions,and monitoring system performance. All this was possible by reading and analyzingdata through a software created in collaboration with EQUA ab. This software hasthe capacity to read, graphically display and analyze commissioned data collectedfrom a building’s HVAC system. Programmed expert rules in the software will returnto the user visual guidance for checked, not checked, violated, and not violatedstates so that heavy data analysis becomes accessible for the manager. Eleven expertrules are proposed to diagnose potential faults and, this way, finding solutionswith greater precision. The Rule set is divided into six groups; checking comfort andthermal dynamics, ventilation systems, outer temperature measurement, Cooling andheating side of the HVAC, and special components. As a result of this study, the proposedexpert rule set accurately contributed to the data analysis, helping to diagnosepossible threats to the most important sections of the HVAC system and leading tosuggestions that can improve thermal comfort, system lifespan, energy performanceand more.