dcsimg
 

5 Critical Tools for Your Lean Deployment

As your customers demand more and organizations fight to stay competitive, the battle to create more value with fewer resources is ever present. How can you leverage Lean tools to thrive in your business?

Save Your Seat

Analyzing Data to Ensure Easy Access to Cool Beans

In my work I get to learn about how companies use Minitab products to improve quality, and it's always a treat to learn how a company whose products I enjoy puts Minitab to use. So it was a real kick to read a Quality Progress article (free with registration at asq.org) about Starbucks. 

The article relates how, when voice-of-the-customer data revealed a need to fine-tune how the Starbucks’ one-pound packages of coffee were being sealed, the company corrected the issue.

Air oxidizes coffee and affects flavor, so protecting beans with an airtight seal was critical. But the package has to be easy for people like me to repeatedly open without tearing the bag’s inner liner. That’s where there was room for improvement. Initial screening experiments revealed that it was easy to obtain an airtight seal, but not so easy to obtain an airtight seal while keeping the package easy to open.

The challenge was to find process conditions which would seal strongly enough to be airtight but not so strongly that the coffee inside was difficult to access. Starbucks' reliability engineer needed to identify which parts of the bag-sealing equipment affected these key quality characteristics, and find a way to adjust the process so that both requirements were met. Initial research identified three key variables that could affect the strength of the seal: plate gap, plastic viscosity, and clamping pressure.

To find process conditions that would meet specifications for both Leakage and Tear, Starbucks used Minitab Statistical Software's DOE (design of experiments) tools. Many different experimental designs are available, but the design a researcher selects needs to compliment the types of variables in the study and the goals of the experiment. The engineer needed a design that could model each response as a function of the three experimental variables. 

After discussing the challenge with Lou Johnson, a technical training specialist at Minitab, the engineer decided that a response surface design was well suited to this process problem. Response surface designs let you fit models that can more accurately predict responses at any set of input variable conditions, and she anticipated needing these more accurate models to find the best response.

She used Minitab to devise a central composite response surface design, which required 19 runs using a variety of combinations of the process variables. After collecting the average tear and leakage responses of 20 samples for each run, she analyzed the results in Minitab, reviewing each potential term to determine if it added value to the model’s predictions or not. After removing the insignificant terms, she arrived at final models for the Tear and Water responses as a function of the experimental variables.

The models helped identify which inputs needed to be most tightly controlled to keep the response stable over time. Plate Gap and Pressure had the strongest effect on both responses. Viscosity only affected the Tear response, and lower Viscosity minimized Tear, so it was best to operate at the minimum value of Viscosity.

To simultaneously optimize both Tear and Leakage, Burrows created an easy-to-understand overlaid contour plot of Pressure and Plate Gap, with Viscosity fixed at the low level. Minitab’s contour plots make it easy to visualize the effects of experimental variables and determine optimal settings. The plot revealed the Pressure and Plate Gap variable space that met both sets of specifications. 

Contour plot of tear and water

These contours represent the prediction for the average response from the process rather than the response for every sample sealed in the process. The unshaded region identifies the process conditions that met both the Tear and Leakage specifications. Since the goal was to maximize the number of individual bags that met both specifications, settings near the center of the acceptable process variable space were selected.

To confirm the effectiveness of the new process settings, 60 sample bags were produced. All of the bags were airtight, but the real proving ground would be the tear-test results. When Burrows looked at the data, only 3 of the 60 bags exhibited any tearing—and those tears were at very minor levels.

Based on the results of the verification run, the new process settings were implemented. Changing the sealing process parameters soon yielded impressive results. After just two months of operation, defect levels for the airtight seal were still at the benchmark of 0%, but tear levels had dropped to less than a 10th of their benchmark levels—and when it comes to high-quality coffee, that’s a tasty statistic.

This project was discussed in an article that appeared in the March 2011 issue of Quality Progress.

 

Comments

Name: Portella • Saturday, September 13, 2014

amazing


blog comments powered by Disqus