Valuable Information

as you begin the Lean transformation

Six Sigma – Hands‐on Part 8

CASE STUDY

The following is a real-life example of how DRIVE was able to apply the “Hands-On” Six Sigma approach in order to help a customer solve a longtime crippling problem. The client is a US manufacturer of custom thermoplastic profile extrusion that engaged DRIVE to apply our brand of statistical problem solving.

Investment: $20K

Product A Savings: $198,773 annualized

Product B Savings: $296,843 annualized

Project ROI (annualized first year): 2478%

We gathered the supplied data and created a DECISION TREE to show the location of the largest problem for this manufacturing site. The data lead us to “surface finish on product A being manufactured on line 25.” Once the location and specific problem were narrowed down, we began a High Level Process Map of the line. We also created a Physical Flow Layout of the line to show all series and parallel paths.

We had to ensure that we could measure the defect (Poor Surface Finish) being addressed. We gathered samples of the product and immediately noticed that there was no common understanding of “good vs. bad” within the team. The team consisted of quality, maintenance and operations leaders. We used the attribute to variable transform to convert the subjective measurement into a more objective measurement. We gathered 8 samples that ranged from perfect to unacceptable. We had to pull the perfect sample from the Quality Lab, since we could not run a new one on the line. We ranked the samples from 1 (perfect) to 8 (unacceptable) with 4 being the “Can’t send to the Customer” sample. We conducted a Gage R&R on samples using this new method. Once we passed the Gage R&R, we could compare the process to our measurement system.

The process was running at a 3 (out of 8) on average. This was a completely unacceptable average because it clearly showed that there was an unstable process.

Using the same measurement criteria, we created a Contrast Matrix to look for contrast in the following areas:

  • Location to Location within the same product
  • Product to Product within the same line
  • Line to Line within the same time
  • Time to Time with the same product and within the same line

The data showed us that we had defects on the centers of the product at all times on the same line. We did not understand the “center of the product” contrast, so we could not leverage it. Using the Contrast Matrix, we decided to gather data for the Line to Line contrast. Our selection of the other line (Line 15) was based on the same material type being run there without surface issues. We planned an experiment to run the same material, extrusion die, and forming tank on Line 15. Armed with the planned experiment, the team moved the product to the new line. It quickly achieved a surface finish of 1 to 2 on the product versus the previous average of 3. It also only took 40 minutes to setup (down from 2 hours) and ran a total of 24 hours without burning the die (which previously burned every 2 hours). The biggest issue that day became running out of customer orders for the product. Therefore, we had to stop in order to avoid the deadly waste of overproduction. Once the problem was solved, we were faced with VALIDATING the change and UNDERSTANDING the differences between the lines. All we knew at that point was that the root cause lived in the extruder, since it was the only portion of the process that had been swapped. The improvement was validated by returning the material, die, and forming tank to the old line. This resulted in the surface finish returning to a level 3 per our new measurement method. Normally, we would expect the team to repeat this process for a total of 6 times (randomly) to have a 95% confidence in the change. However, due to costs and insufficient customer orders, we made the decision not to repeat the experiment.

During our discussion of the leverage phase, we concluded that we could leverage the application of Measurement System Analysis across all products within the plant. The team also implemented the Attribute to Variable transform on another product that had a Subjective Measurement. There were two other projects (Product B and C) running during the same timeframe related to setup and measurement. This seemed to be a trend at this site, so we concluded that there must be a systemic issue with the release of new products to the production area. The uncontrolled variables per product upon release made it impossible to repeat part quality at each setup. This would lead to long setup times and large quantities of scrap. This systemic issue had cost the company millions in lost business, wasted labor, and scrap. The next step was to overhaul the design value stream for this site (which flows from concept to launch).

The team also leveraged what they had learned and improved the output on another product by 50% while attaining a quality that could not be matched by their competitors. This allowed the team to increase the sales from this particular customer, making them the sole supplier for this product. By improving the quality to such a high level, this company created a barrier to market for their competitors.

This is the final issue of our six-sigma series. We at DRIVE have enjoyed sharing the DMAIC-L process with our readers. If your company would like to solve long-standing problems, we offer foundational, intermediate and advanced level problem solving including practitioner certifications. For a no-obligation introduction meeting, please contact Paul Eakle at paul.eakle@driveinc.com or 865-323-3491.

Looking for More?

Archives: