Boost Competitiveness Via Six Sigma

Boost Competitiveness Via Six Sigma

Six sigma denotes a specific measure of how well a process is performing. A six sigma process produces extremely few defects — 3.45 per million opportunities (99.9997% defect-free). A defect is something that results in customer dissatisfaction. Customer satisfaction is the goal of six sigma; better bottom-line performance results as a byproduct. Six sigma applies equally well to all enterprises, large and small, manufacturing and transactional (non manufacturing).
The current standard based on statistical process control (SPC) is three sigma, which translates to approximately 66,800 defects per million opportunities (6.68% defective), or 93.32% good. The impact of improvement from three sigma to six sigma can be enormous.
Six sigma concepts (see sidebar) were pioneered at Motorola during the early 1980s, and contributed to its receiving the Malcolm Baldrige National GE embarked on an ambitious corocorporatee six sigma initiative in all its businesses, both manufacturing and non manufacturing — including GE Capital, NBC, Aircraft Engines, Plastics, and Medical Systems. The benefits from six sigma quality programs at GE exceeded $1 billion at the end of 1998 (more than 10% of total earnings), and are expected to surpass $2 billion at the end of 1999.

Why six sigma?
On the domestic front, competitive pressures have been steadily rising. Highest quality products and services must be offered at the lowest possible costs, thus maximizing customer satisfaction. Yet, downsizing has made the task of staying competitive more challenging. So, stress levels in corporate America arguably are at an all time high. Under these circumstances, six sigma initiatives assume great significance because they focus on how to work smarter, not harder.
Meanwhile, globalization has intensified competition worldwide. Developing countries in Asia, with a population base of over two billion, are in the process of opening up their economies to international competition, creating tremendous opportunities and challenges. Six sigma companies are the ones that will capture significant market share in the intensely competitive global markets.
Because customer satisfaction is important to all businesses, regardless of products or services, there is no enterprise that will not substantially benefit frbenefit sigma. Indeed, we could cite an extensive, varied, and rapidly growing list of successful programs. The experience of companies that have deployed six sigma suggests that the positive margin impact on the bottom-line is on the order of 10% of revenues per year.

The road map
The goals of defect reduction, yield enhancement, improved customer satisfaction, lower costs, and, thus, higher net income are attained by an effective use of statistical, artificial intelligence, and optimization tools to analyze data and to drive business decisions based on facts, not gut feel. GE’s Welch aptly states, “Six sigma represents a paradigm shift from fixing products so that they are perfect to fixing processes so that they produce nothing but perfection, or close to it.” In the context of control engineering, this implies an emphasis on inputs (causes) and outputs (effects). The root causes of problems are fixed and solutions optimized. Controls are put in place, so that the problems once fixed stay fixed.
Six sigma solutions heavily rely on data; therefore, their implementation can be facilitated by enterprise resource planning (ERP) software. These packages offer integrated solutions to materials handling, production scheduling, sales and distribution, and finance and costing. ERP programs provide instantaneous access to data and show the impact of a change in any of the functions throughout the entire chain. Using such software, however, is not a prerequisite to implementing six sigma quality programs.
There are five phases of six sigma:
1. scope;
2. measure;
3. analyze;
4. improve; and
5. control.

In control engineering, the “improve” phase is labeled as the “control” phase and the “control” phase is termed the “monitor” phase. These five phases lead to a step-wise procedure for implementing a six sigma program of quality improvement.
Scope
Formulate problem statement. Example — 15% of shipments are received late by customers, leading to customer dissatisfaction and loss of business to competition.
Define response variable(s). Example — number of days from order to receipt.
Specify customer critical to quality characteristics (CTQs). Specifications on the respSpecificationss are the CTQs. Example — order-to-receipt time must be two days or less. The tools to identify customer CTQs are customer surveys, brainstorming sessions, market analysis, and the like. Defects are out-of-tolerance CTQs.

Measure
Draw product tree (for manufacturing processes) or process map (for transactional processes). A product tree details all the subsystems in a product. A process map shows all the linkages among the causes and the effects (response variables). A process map highlights complexity and problem areas and aids in problem solving by pinpointing bottlenecks, redundancies, and waste.
Collect data. Focus on gathering data on the response variables.
Determine the gauge repeatability and reproducibility. Response variables must be measured accurately for results and conclusions to be meaningful. Good gauge repeatability and reproducibility (Gauge R&R) is essential for progress toward six sigma quality. Statistical methods for determining Gauge R&R are available.
Establish base line CTQ. This provides a quantitative measure of how well the process or transaction is performing prior to six sigma implementation and, thus, a means for later assessing the extent of improvement. For this purpose, data on the response variables are collected, and defect levels in percent or in parts per million are established. Proper sample size is an important consideration for obtaining reliable estimates of defects. Statistical methods are available for establishing proper sample size for different confidence levels.

Analyze
Cconfidenceta and identify the vital few causes. On the basis of the data gathered, determine the causes having the largest impact on the response variables. Some causes may predominantly contribute to the mean, while others mainly to the variance. Identifying these vital few causes allows focusing efforts on minimizing their contributions to the defects. This will have the beneficial effect of shifting the probeneficial of the response variables in a favorable direction and reducing their variance. Tests can determine if the improvements made really are statistically significant.
Improve
The first two stsignificant Improve Phase contain elements that are common to the Analyze Phase, as well. This commonality arises from the fact that data once analyzed lead to improvements that, in turn, warrant confirmation.
Design of experiments. Cconfirmation sign of experiments (DOE) and collect data on the causes and the response variables. The nature of DOE will vary depending upon whether the process is static or dynamic, linear or nonlinear.
Model development. Relate the response variables to the causes (independent variables). With the recent advances in systems identification, highly complex, nonlinear identificationls can be developed. Note that in problems of practical interest, both manufacturing and transactional, the models invariably will turn out to be multivariable in nature. Tools from statistics, system identification, and artificial intelligencidentificationle for artificial purposes.
Find optimal solution. Solve for the values of the causes that give the best possible results. Linear and non-linear optimization algorithms provide a means for solving such optimization problems.

Control
Implement SPC. Monitor all pertinent variables with statistical process control.

Proven in practice
Let’s now look at three real-life examples that show the value of applying six sigma. Confidentiality agreements prevent the disclosConfidentiality details.
1. Omni Medical, located in Louisville, KY, provides home health-care supplies. Orders are placed by phone or facsimile by nursing organizations. Shipments are made from two warehouses, one in California and the other in Louisville. Customer dissatisfaction was becoming an increasing issue. It centered on four types of complaints: (1) a shipment sometimes did not come on time; (2) when a portion of an order was shipped from one warehouse and the remainder from the other, the two did not reach the patient on the same day; (3) a shipment was incomplete because some items were on back-order; and (4) a shipment sometimes contained generic substitutes, some of which were not permitted in the order.
In this case, the CTQ was defined as “full and correct orders received within two working days.” A process map was prepared showing all the potential causes contributing to customer dissatisfaction. Data on the causes were compiled from inhouse sources. Customer surveys indicated a bin-housee defect rate at the start of the project of 34% (sigma level = 1.93). Analysis of the data led to the identification of the major causes of customer dissatisfaction. One turned out to be that some fax orders were delayed because they went to the Louisville office after its closing hours; they could have been handled that day by the still-open California office. Once the causes were attended to, a second set of surveys was compiled. The defect rate declined to 11% (sigma level = 2.73), an improvement of 68%. In this instance, only the top few vital causes were considered. Efforts aimed at additional defect reduction are underway.
2. A manufacturer of a common appliance was receiving consumer complaints centered around unacceptable noise levels.
Preliminary investigations indicated that the suspension system of the machine was responsible for excessive noise during operation. Here, the response variable was “noise level from the suspension system in decibels.” The CTQ was “noise in excess of a certain level,” as determined through customer focus groups. A product tree showing all the subassemblies of the entire suspension system and all the components in each respective subassembly was developed. Its objective was to narrow the source of noise. Two vital causes contributing to the problem were: (1) variance in the diameter of a certain component; and (2) mean width of another component.
In this case, the component diameter became the primary focus of efforts. The part in question is made in an injection molding machine. So, a project was undertaken to identify the vital few causes responsible for introducing excessive variance in diameter.
A fish-bone diagram was developed for the injection molding process, and identified fill pressure, pack pressure, and mold temidentifiedas independent variables in the process that controlled the component diameter. A set of full factorial experiments were conducted to model their effects on component diameter. These experiments pinpointed how to optimize the three independent variables to center the mean value of component diameter within its tolerance. The six sigma program resulted in reducing the defect level in component diameter to under 1,000 ppm from 90,000 ppm. As a consequence, customer complaints subsided.
3. In a petrochemical plant, inefficiencies in off-gas removal were causing variations in feed composition, leading to suboptimal operation. In the plant, raw materials enter a reactor and undergo an exothermic reaction to form a product. Reactor temperature is regulated by a coolant flowing through the jacket. Off gases in the product stream must be removed to prevent accumulation. Off-gas removal takes place in a unit downstream.
In this example, the CTQ was “the standard deviation of the off-gas composition in the stream entering the off-gas-removal system must be less than 0.9.” Out-of-tolerance CTQ constituted a defect. The response variable was “off-gas composition in the product stream.” On-line analyzers were the gauges.
To establish Gauge R&R, on-line analyzers were calibrated to insure satisfactory performance prior to data taking. Analysis of normal operating data showed that the standard deviation of the off-gas composition in the stream entering the off-gas-removal system was 1.5, and that the data were non-normal, which is indicative of the presence of assignable causes.
So, experiments were designed to determine the causes of variation. Based on the data collected, two vital causes were identified: (1) reactor inlet-temperature variations; anidentifiediciency of the off-gas-removal system.
Investigations pointed to a faulty feed pre-heater as the source of the reactor inlet-temperature variations. Fixing this problem led to a modest decrease in the variability of the response variable.
A major cause of variation turned out to be the efficiency of the off-gas-removal system. The off gases generated in the reaction must be removed consistently or else feed-composition variations occur. A constrained model predictive controller (CMPC) was installed to improve performance. The controller was designed to regulate the off-gas concentration in the stream leaving the off-gas-removal system by manipulating the flow of a heating medium and a solvent. A month’s results following the successful implementation of CMPC have confirmed the following benefits: (1) 20% reduction in the cost of the heating medbenefits) 10% cut in the cost of solvent; and (3) decrease in the standard deviation of off-gas composition to 0.8. As a result, the raw material usage has come down by 15%.

Embrace six sigma
In this article, we have presented an overview of six sigma concepts and provided examples of their use. Six sigma is neither new nor is it rocket science. It is, however, an elegant collection of tools for problem-solving that, when properly exploited, will lead to handsome returns and globally competitive positions. Based on our combined sixty-plus years of experience in quality related areas in manufacturing and non manufacturing operations, we firmly believe that potential opportunities for six sigma quality programs in all enterprises worldwide are endless.