Abstract

To diagnose and correct a recurring problem, do not assume that the solution is known. Instead, systematically identify and solve the problem. This method has been applied to chronic gear nonconformities at the Central Foundry Division of General Motors Corporation. First, the gear porosity team used Pareto charts to define porosity as the problem. Then they established a measuring system based on diameter and depth of nonconformance. To avoid preconceptions, the team generated clues from concentration diagrams, a videotape, a metallurgical analysis, and a Multi-Vari study. This revealed clues like the tendency for porosity to be on certain sides of a casting. The clues led to a list of possible causes that were tested by experiment. A full factorial design with a 95% confidence level suggested primary causes. Next, the clue and experiment steps were reiterated until there was clear evidence of a primary cause. Corrective action was taken, which will be validated by a three-month pilot run. This approach is more diagnostic and less theoretical than traditional problem solving.