Modelling and Control of Dry and Wet Gas Compressors
Abstract
Performing gas compression subsea is an important component for future realisation ofcomplete subsea processing facilities. Gas compressors operating with raw well fluids, categorised as wet gas, enables an increase in hydrocarbon recovery and operational efficiency together with cost reduction for extended subsea processing. The wet gas, defined with a liquid fraction up to 5%, implies challenges for existing dry gas compressor technology. To accurately predict compressor performance with wet gas and enable efficient control, research leading to first principle modelling of wet gas compressors is important. The work in this thesis expands upon the specialisation project work in the autumn of 2016. The main objectives are to establish performance characteristics of a compressor, validate dry gas compressor models through simulation and expand the models towards simulation and control with emulated wet gas flow.Two compressor characteristics have been curve fitted to a Garett GT2252 turbochargercompressor map. The turbocharger is stationed in the Wet Gas Compression laboratory atthe Department of Mechanical and Industrial Engineering at NTNU. The map characteristicswere curve fitted with two different methods in MATLAB. The least squares non-linear fitfor a first principle characteristic function and with a traditional empirical 3rd-degree polynomial, commonly used in literature. The latter curve fit method produced the best fit in terms of mean absolute error. The two characteristics have been validated in simulation with two compressor models, the axial compressor model developed by Greitzer and the centrifugalcompressor model developed by Gravdahl.Through a controller tuning study, a non-linear gain scheduling controller has been developedfor the normal operating region of the compressor. Two 3rd-degree polynomials, calculatedoffline, are used for continuously updating each controller setting in the schedulingalgorithm. The performance results of the gain scheduling controller have proven to reducethe integral of absolute value of control error compared to a single linear controller forsetpoint and load disturbance changes. The controller has been subject to testing with modeluncertainties, by interchanging the first principle with the empirical compressor characteristic in the model.The first principle compressor characteristic has been extended with the application ofwet gas parameters. The inlet gas density, gas specific heat capacity and impeller frictionloss factor are empirically calculated with changing gas volume fraction of the gas flow.Gas volume fractions of 100% to 99.90% and corresponding gas mass fractions of 100% to54.67% are used to simulate a wet gas compressor using the gain scheduling controller. Thishas been performed to validate the controller robustness. The results of utilising the wet gas compressor characteristic in simulation have shown a decrease in achievable pressure ratio and an earlier saturation of the drive unit to maximum speed for the compressor.The author has contributed to the electrical installation on the Department Wet Gas Compression laboratory. Electrical installation schematics and an operation protocol have been produced to complete the laboratory documentation. The laboratory electrical infrastructure is tested, documented and verified for operation.