https://www.mdu.se/

Gas turbine performance is very responsive to ambient and operational conditions. If the engine is not operating atits optimum conditions, there will be high energy consumption and environmental pollution. Hence, a precise simulationmodel of a gas turbine is needed for performance evaluation and fault detection and diagnostics. This paper presents a twinshaft industrial gas turbine modeling and validation. To develop the simulation model component maps are important,however they are property of the manufacturers and classified documents. In this case, known the compressor pressureratio, speed, and flow rate, the missing design parameters, namely turbines inlet temperatures and pressure ratios werepredicted using GasTurb simulation software. Once the design parameters are developed, the nearest compressor andturbine maps were selected from GasTurb map collection. Beta lines were introduced on each map so that the exactcorresponding value can be picked for a given two parameters of a given map. After the completion of components model,a simulation model was developed in Matlab environment. The equations governing the operation of individual componentwere solved using iteration method. The simulation model has modular nature; it can be modified easily when a change isrequired. The parameters that the model can predict include terminal temperature and pressure, flow rate, specific fuelconsumption, thermal efficiency and heat ratio. To demonstrate the validity of the developed model, the performance ofGE LM2500 twin shaft gas turbine operating in a gas oil industry at Resak PETRONAS platform in Malaysia waspredicted and compared with operational data. The results showed that an average of 5, 3.8 and 3.7 % discrepancies forcompressor discharge temperature and pressure, and fuel flow rate, respectively. This comparison of results showed goodagreement between the measured and predicted parameters. Thus, the developed model can be helpful in performanceevaluation of twin shaft gas turbines and generation of data for training and validation of a fault detection and diagnosticmodel.