SIX SIGMA

What Is Six Sigma?
Six Sigma is a quality discipline that focuses on product and service excellence to create a culture that demands perfection. Its key goal is to achieve excellence by focusing on customer needs and reducing defects in processes, products, and services.
Six Sigma holds that there is a direct relationship between products and customer satisfaction: the fewer the defects, the happier the customer. Six Sigma stops variations in quality at the earliest possible point by attacking variation during design of products and processes.
Sigma is more than just a letter in the Greek alphabet. In this context, Sigma is a statistical measure that tells how much a product, service, or process varies from perfection. Based on defects per million opportunities (Table 1), it holds that the higher the sigma value, the better the quality. To put Table 1 in perspective, a measure of One Sigma would be equivalent to 170 misspelled words on this page; a Six Sigma level would equal one misspelled word in an entire library

The Six Sigma Process
There are four key steps in the Six Sigma process:
• Measure. Identify the key internal processes that influence the “critical to quality”characteristic.

• This phase ends when you can measure or count the defects that affect quality.
  • Analyze. Understand the root cause driving defects. Statistical tools are used to identify the key variables that are likely to drive process variation the most.
  • Improve. Confirm the key variables and then quantify the effects of these variables on the critical to quality characteristics.
  • Control. Ensure that the modified process now enables the key variable to stay within the maximum acceptable ranges
  
  
  How We Applied Six Sigma To Membrane Manufacturing
  Membrane quality is determined by two criteria:
  • Flux (how much water the membrane lets through during a given amount of time) and
  • How much salt (or other impurities) is removed from the water.
  If a membrane has a high flux variation (i.e., it either lets too much or too little water through), then the membrane modules that contain it has a high flux variation too. If a membrane module does not meet customer specifications for flux variation, it is not shipped to the customer. As the products are rejected internally it becomes harder to satisfy demand in a timely manner. This affects customer loyalty and ultimately, our bottom line.
  Membrane material is manufactured in large flat sheet rolls. Samples from each are tested for performance parameters. Rolls that do not perform to standard are discarded, creating both a waste problem and delays in delivery to customers.
  To respond to the quality issue, a Six Sigma “black belt” and his team applied Six Sigma methodologies to reduce the variation in flux in membranes and increase the yield of membranes to meet customer demand.
  As we began the Six Sigma project, we knew that:
  • A number of variables affect the quality of the finished membrane products.
  • Reducing large deviations from the desired recipe value would reduce scrap cost and improve yield.
  • Controlling less severe process variation will produce better performing products for the customer with the additional aim of improving the correlation between membrane properties and final spiral wound element properties. So our project focused on reducing the flux variation and the scrap that it generated due to large deviations from product specifications.
  We used Principal Component Analysis (PCA) to determine the most important process variables that affect flux.
  There are two ways to find out how variables affect operations: designed experiments and undesigned experiments. Designed experiments tend to test variables at extreme ranges of operation and are not always a realistic way to assess process variables. Knowing the constraints of designed experiments, the team chose to analyze data from the operation as it occurred.
  The team constructed a detailed process map of the membrane manufacturing process, including the preparation and introduction of all raw materials and the testing/grading/dispositioning of each finished roll of membrane material. The exercise of writing an as-is process map created a forum for good communication and the chance to more critically analyze individual operations. One of the results of the process map was that the team discovered that plant operators were not necessarily performing process operations the way the jobs were designed.
  Once the study was completed, members of the team used this as an opportunity for behavior-based changes in operating discipline as well as process changes. Starting with a process flowchart, team members identified steps in the process most likely to contribute to variable flux based on their experiences with the process. Those areas of concern were then compared to a Cause and Effect matrix. As stated before, the customer has two requirements: flux and salt rejection; both properties were included in the cause and effect matrix.
  Three factors were ultimately identified as negatively affecting flux and salt rejection. Only one, however, was severe enough to warrant process operating changes.
  
  
  
  
  
  
  Principal Component Analysis
  Using Microsoft Access™, data were pulled together from a variety of sources:
  • The computer log that recorded machine parameters and operating conditions;
  • Manual run sheets used to record hand collected samples (which had to be entered into a spreadsheet for the purpose of this study); and
  • The VAX database, which contained several tables recording the flux and salt passage of the tested product rolls.
  Operational computers gathered data from the operation such as temperature and other external conditions for three weeks. They then analyzed the data using Principal Component Analysis. The Six Sigma team performed a statistical analysis of data logs over several weeks of production to identify the variables that affected membrane flux most significantly.
  The team compared variables that changed and how performance changed when compared to those variables. They concluded that the presence or absence of a certain process chemical, and even the solution integrity, were the prime factor(s) in membrane performance.
  
  
  Drawing Conclusions
  Using Six Sigma reasoning tools, the team found that the concentration of a raw material varied because of interruptions in the process. The interruptions occurred because sometimes the container would be empty and would have to be replaced with a full one. Occasionally the empty container would go unnoticed for some time. To reduce the variation, a small, inexpensive reservoir was added to feed the chemical while the containers were changed. A level transmitter with an alarm was installed on the additional reservoir to alert the operators of a nearly empty container.
  Other times, a second chemical was improperly mixed,which also led to problems with flux. The Six Sigma team explained to plant operators the importance of proper ly mixing the solution.
  
  
  Sustaining Productivity Gains
  To sustain the gains from the project, the Six Sigma team made some changes to help manufacturing operators identify problems before they occur. Before the Six Sigma project, the measurements had been displayed in tabular form, in rows of numbers that were difficult to read. After the project, the measurements were displayed in the form of a Microsoft Excel™ charts that illustrate trends in the numbers and therefore processes. The improvements were significant. For standard surface water membranes with a target flux of 45 gal/ft2/day (gfd), the standard deviation of the process for the 8-month period preceding the improvements was 4.46 gfd and 14.5% of the product was out of specification, a Sigma level of 2.56. After implementing the Six Sigma improvements, the standard deviation was reduced to 1.83 gallons per day, with only 2.2% of the product out of specification, which is a sigma level of 3.51, almost one Sigma better than the initial performance