Fast-Tracking Friction Plate Validation Testing: BorgWarner Improves Efficiency with Machine Learning Methodology

This case describes the process BorgWarner Transmission Systems undertook when an increase in demand required that it create new production lines in another country. To minimize inventory costs, the company engaged with a graduate student team from the University of Michigan’s Tauber Institute for Global Operations for assistance with creating optimal operating procedures. The student team was tasked with developing a prediction tool to ensure a high probability that BorgWarner friction plate samples would pass pre-production verification testing.

The case provides students with an overview of the production and quality assurance processes for friction plate manufacture, as well as data classification models for machine learning. Students then develop a reliable forecasting methodology to systematically evaluate process parameters and maximize test pass rates for friction plates.

• Understand when standard linear statistical models are effective and when they are not.
• Determine criteria for evaluating different machine learning models.
• Understand how to utilize software tools to predict machine learning.