Aluminum Sand Casting Operation Achieves World Class Results

The Challenge

An aluminum sand casting operation in East Tennessee was struggling to produce enough quality parts to meet its Tier 1 automotive customer’s demand.

Due to previous problems, their customer had placed them on the equivalent of “probation”, where every part was inspected three times (once before leaving the plant, once by a third party before being delivered to the assembly plant, and once after arrival at the assembly plant). Even worse, all the inspection costs were borne by the company on probation.

The overall scrap rate internally to casting plant was running as high as 50%. World class performance for the aluminum sand casting process is 10% to 12% scrap.

The Strategy

The management team at the sand casting plant chose to use the techniques of MSLLC to improve their performance and get off of “probation”.

Two major strategies were employed to improve the operation’s first pass yield:  
1) Measurement System Analysis  
2) Manufacturing  
Process Experimentation

Because most of the inspection process was visual, a blind sampling study using known “accepted” and known “rejected” parts was conducted.

After creating a repeatable and reliable measurement system, the team went to work on the manufacturing process.

The Measurement System Analysis

The measurement system study showed dramatic problems with the inspection process. Each of the 50 parts used in the study were inspected 5 different times by different inspectors on different shifts. Every part passed at least one inspection and every part failed at least one inspection. The inspection process was not much better than flipping a coin. Using process design and calibrating all the inspectors with definitions of reasons to reject a part, the repeatability of the measurement system was acceptable.

The Experiment

In the manufacturing process, operators, supervisors and inspectors were asked what changes could be made to reduce the scrap rate. Over 250 ideas were generated that spanned the entire process. Management selected 15 factors to test using a 32 treatment combination (or recipe) design to learn what parts of the process were creating the scrap. Below are a few of the ideas.

The Results 
The results were amazing. The measurement system improvement created a 15% reduction in scrap, simply by ensuring that the parts were accurately categorized as good or bad. An additional 27% reduction in scrap was achieved by experimenting with the manufacturing process. Three of the 15 factors tested had a significant impact on the scrap rate. The largest effect was a 17% reduction using molding sequence A instead of molding sequence B. All operators were trained in sequence A. Overall, the plant achieved a 7% scrap rate and was taken off probation.