A novel linear motor for a linear refrigeration compressor: modelling, measurement and sensor-less stroke detection

With the increasing global awareness of the environmental conservation, linear compressors have attracted growing attention with their applications in domestic and cryogenic refrigeration systems. A linear compressor is driven directly by a linear motor and the free-piston design allows piston stroke to be variable. An active control of stroke prevents piston-cylinder collision and enables efficient cooling capacity modulation. This thesis introduces the performance of a novel moving magnet type linear compressor/motor and investigates the approaches to sensor-less stroke detection. An experimental test facility incorporating the linear compressors into a vapour compression refrigeration system was introduced, in which piston displacement was measured with a displacement sensor. The piston stroke and offset were controlled with PID controllers implemented in LabVIEW. To investigate the characteristics of the moving magnet linear motor, a finite element analysis (FEA) model was built in ANSYS Maxwell 19.2. Simulations were validated through static force measurements. Force constant was given by the static shaft force against current. Saturation can be observed with the increase of current. A smaller saturation current was shown for a larger armature displacement. For the purpose of increasing cooling capacity of the linear compressor, operations with small axial clearance volumes were considered. Refrigeration performance using R1234yf as refrigerant with various clearance volumes and with an offset of 0 mm were experimentally compared. The cooling capacity for a pressure ratio of 2.5 and a stroke of 13 mm increases by 12% as the clearance decreases from 1.07 mm to 0.4 mm. Piston stroke detection without a displacement sensor reduces the cost and facilitates the stroke control especially in miniature linear compressors. An artificial neural network (ANN) based stroke detection was presented. Fast Fourier transform (FFT) analysis was performed on current and voltage signals to extract harmonic terms as inputs of the neural network model to predict the stroke. The ANN technique can achieve a good accuracy for most of the cases, but reliability remains a problem. A more reliable sensor-less stroke detection technique based on flux linkage variation using inductive coils was proposed. The technique requires resonant operation. A 1D (One-Dimensional) electromagnetic model and a 3D (Three-Dimensional) FEA model were built to compute the flux linkage variations. The open-circuit flux linkage in each core produced by NdFebB magnets varies linearly with the piston displacement. Flux linkage difference at two zero-crossing points of current was used to infer stroke. The proposed low-cost sensor-less stroke detection technique can achieve error of only 4%. The adoption of this novel technique is crucial to the commercialization of free-piston machines for high efficiency