Desenvolvimento de modelagem matemática e observador experimental para cálculo em tempo real das características de funcionamento de compressores radiais de turbocompressores automotivos

Abstract

The reciprocating type internal combustion engines are the main power source in the transportation sector. Thus, recurrent demands for energy efficiency increase and emission reduction are imposed. The turbochargers are devices used aiming an engine with greater energy efficiency than a naturally aspirated engine with equivalent power output. The optimisation of the turbocharger matching and its control has been increasingly crucial in the internal combustion engine project and its electronic control systems parameters calibration processes, making use of tools like the control-oriented modelling. A control-oriented modelling-based turbocharger observer for engine test in dynamometer bench is proposed in this work. Three modelling approaches found in the state of art, namely J&K, Stricker and Polynomial modelling, and the thermodynamic models for compressor performance calculation where fitted and validated against steady state experimental data from a BIAGIO BBV 1000 tested at the LabTURBO - CTM UFMG turbocharger test bench. The J&K modelling approach were selected to develop the observer due to its low uncertainty levels needing around 0;0580ms to estimate ṁcorr, Πt-t and ηis in each operation point. It was observed prediction intervals of [-1;11%;0;89%] for Πt-t and [-4;06%;4;14%] for ηis of the fitted models. The validation inside the mapped region resulted in an ER of 0;20% with σER of 0;25% for Πt-t and ER of 1;54% with σER of 2;82% for his. Finally, the validation in lower angular speed resulted in ER of -0;28% with σER of 0;58% for Πt-t and ER of 15;53% with σER of 16;90% for his. The J&K was implemented in the test bench controller software. The resulting observer used compressor upstream temperature and pressure measuring and compressor mass flow rate and turbocharger angular speed measuring from the control. The observer also has a user interface for presenting the performance estimation during tests and experiment report creation for points of interest. An internal combustion engine operating curve was simulated in the turbocharger test bench to validate the proposed observer. The observer validation resulted in a ER of -1;11% with σER of 0;23% for ṁcorr prediction, ER of 0;70% with σER of 0;80% for Πt-t prediction and ER of 6:83% with σER of 17;73% for ηis prediction. Based on the uncertainties presented during validation, the observer algorithm was considered apt to estimate the compressor thermodynamic state during internal combustion engine tests.