Abstract
The objective of this investigation is to provide an overview of the development of a virtual methodology to evaluate the influence of shifting clutch friction coefficient on automatic transmission shift quality. The virtual tool also enabled evaluation of embedded transmission control algorithms relative to shifting clutch adaptive learning to compensate for unexpected deviations in friction coefficient, improve shift control, and shift quality. Siemens LMS Amesim was used to create the automatic transmission plant model and was co-simulated with the embedded controls software and calibration model created in Matlab/Simulink. Powertrain dynamometer testing at various power on upshifts were performed over a range of operating speeds and torques to collect model validation data and to estimate friction coefficients for all the shifting clutch packs to establish baseline performance. A variety of hypothetical clutch slip-friction profiles were then defined for each shifting clutch and sequential upshifts to investigate the compatibility with the transmission from a shift quality perspective in addition to determining how well the adaptive learn features worked with known friction coefficients. The model was then exercised over matrix of upshifts at speeds and torques similar to dyno testing with the baseline friction coefficients for correlation and then with the hypothetical friction coefficients to check control and behavior robustness. The results were post-processed and analyzed for shift quality using vibration dose value (VDV), while adaptive control was examined by looking at shift quality and clutch pressure learning over repeated shift events. It was found that the model developed correlated well with dynamometer measurements and was suitable for the investigation objectives. Further, it was found that decreasing clutch friction-speed profile (decreasing clutch friction-speed moving from slipping to sync) performed the best for the transmission being tested and analyzed.