Gas turbine circumferential temperature distribution model for the combustion system fault detection

The gas turbine combustion system operates under harsh conditions, and any faults such as nozzle clogging, cracks, etc. can endanger the downstream components and cause catastrophic accidents. Thus, monitoring the combustion system status and detecting its performance degradation timely are crucial for gas turbine safety. The current predominant method for monitoring involves analyzing exhaust gas temperature (EGT) dispersion, utilizing thermocouples placed at the turbine outlet to indirectly assess combustion system performance. However, the EGT is affected not only by faults but also by various non-fault disturbances, such as operational conditions, atmospheric variations, compressor performance degradation, and uneven hot gas mixing and rotating. To improve the sensitivity of fault detection, it is necessary to investigate how various factors affect the EGT dispersion and differentiate between EGT changes due to interference and those due to combustion system faults. However, using actual operational data from gas turbines to analyze and reveal these impact mechanisms is challenging. Therefore, this paper develops a gas turbine model for combustion system fault detection that accurately reflects the uneven circumferential distribution of EGT. The model incorporates a multi-combustor burner, a multichannel turbine, and a module for simulating the mixing and rotating of the hot gas. Using this model, simulations are conducted to study the effects of non-fault interference factors, such as variations in ambient temperature, fuel flow changes, and compressor performance degradation, on EGT distribution. Additionally, simulations are conducted to analyze the effects of faults, including combustor cracks, uneven fuel distribution, and the coupling of interferences with faults, on EGT distribution. By comparing the normal and abnormal modes, the study reveals the different EGT responses due to non-fault interferences and combustion system faults. Based on these findings, a fault detection method for coupled conditions is proposed. The research results provide a basis for the fault detection of combustion system, enabling practitioners to reduce interference in fault detection and to increase the detection sensitivity.