Compensação de deriva térmica de encoder magnético analógico via software aplicada a compressor hermético para análise de diagrama P-V

Abstract

The construction of experimental PV diagrams is fundamental for the reliability analysis of hermetic compressors, since it enables the quantification of the work required by the process the identification of heat transfer losses, and the diagnosis of inefficiencies in the suction and discharge valves. Proper instrumentation of this system requires accurate measurements of the cylinder pressure and the shaft angular position. In this context, magnetic encoders are widely used due to their high resolution and contactless operation. Although magnetic encoders are subject to different sources of error, such as misalignment, electrical noise, and nonlinearities, as well as limitations in repeatability and reproducibility, this work focuses exclusively on thermal drift, as it is an error that can become predominant in applications subjected to large temperature variations, compromising the thermodynamic fidelity of the experimental P V diagrams. These errors arise from temperature-dependent variations in the sensor response, including changes in the signal amplitude and offset, which distort the calculated angle. This work presents a semi-empirical compensation strategy to mitigate the effects of thermal drift in analog magnetic encoders via software. The approach is based on the generation of reference curves under controlled temperature conditions, using a bivariate polynomial regression model of the shaft angle and temperature. The results demonstrate how this thermal phenomenon significantly affects the encoder measurements, particularly at elevated temperatures, leading to severe distortions in the PV diagrams. The application of the proposed method proved to be essential for measurement reliability, since thermal drift can shift the power calculation by approximately 4.25%. This level of error is critical, as it compromises the accuracy of loss segregation and energy efficiency analyses.