Improvement of Piston Ring Quality: A Case Study
Student Name: H. S. Nataraja
The management
of the organization the student was working for, was facing a serious quality
problem due to high scrap rates in most of their products. In this work, a
systematic step by step approach was adopted for solving this quality
management problem using various statistical quality and process control
techniques. First the most delicate and costly ring was selected for this case
study and a Pareto analysis revealed axial thickness as the most problematic
quality characteristic. Next it was checked whether the processes affecting
axial thickness were in statistical control or not by means of control charts.
Two such processes were identified which were not in statistical control. A
simple recommendation solved the problem with one of the processes. For the
other process, which was not in statistical control, an in-depth study led to
an Ishikawa Diagram, diagnosing the various factors affecting the axial
thickness. Then these factors were classified into control and noise factors
and a full factorial experiment with replication was designed and carried out
to analyze the variances of the averages and signal to noise ratios of the
experimental results. These ANOVA's and multiple comparison $t$-tests
suggested certain optimum levels for the control factors. A confirmatory
experiment with these optimal settings yielded highly improved process
capability indices. The underlying assumptions of all the applied statistical
techniques were appropriately validated. This research helped the organization
drastically reduce its scrap rates resulting in a dramatic increase in profit.