295 14 Six Sigma Problem Solving Jonathon L. Andell Many consultants and references advocate Six Sigma as a means to rectify quality problems in a manufacturing environment. This application is indeed valid, yielding impressive financial results, as we shall discuss. However, there is a variety of other situations wherein Six Sigma problem-solving methodologies can help an organi- zation, such as the following: • Identifying and eliminating the causes of nagging problems throughout a business — the application most commonly described in articles and brochures • Developing manufactured and service products with significant competi- tive edges — the realm called Design for Six Sigma (DFSS) • Planning and implementing management initiatives, including Six Sigma itself — setting up Six Sigma to match the requirements of each specific business As one might expect, achieving such divergent objectives depends on applying somewhat different tools. After all, the list starts with tactical issues dealing with things, and progresses toward strategic issues of people and organizations. In order to accommodate such diverse objectives, Six Sigma problem solving encompasses a variety of approaches. Most organizations have individuals with excellent backgrounds in Six Sigma problem solving, even if they call it by another name. Furthermore, many managers have seen literature and attended seminars on how it works. However, it is common- place for the state of problem solving at large to lag significantly behind what an organization’s best people contribute. The challenge, therefore, is to make excellent problem-solving teams less of an exception and more the rule. As Table 14.1 shows, quite a balancing act is involved in bringing this about. This chapter endeavors to provide managers with some guidelines for striking such a balance. However, there are limitations inherent in such a discussion: • No single chapter can provide enough detail to make the reader into an expert problem solver. (For that matter, nobody can become an expert simply by reading. It’s like golf, sooner or later you have to put down the books and pick up the clubs.) SL3003Ch14Frame Page 295 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC 296 The Manufacturing Handbook of Best Practices • A detailed description of all problem-solving tools also is beyond the scope of a single chapter. Fortunately, the chapters of this handbook address the more powerful tools. This chapter serves partly as an overview for when and where each chapter’s contribution might apply within the big picture. • Emphasis remains on tactical problem solving, the first of the three broad problem-solving applications described above. The object of this chapter is to enable managers to support Six Sigma problem solving within their organizations. The direct implication is that somebody other than managers will lead the teams, specifically the practitioners, experts, and masters described in Chapter 2, “Benefiting from Six Sigma Quality.” Managers generally provide a combination of guidance and support, as we will discuss. Numerous anecdotes are used, some to describe traditional businesses, others to illustrate how a Six Sigma organization functions. The distinction between a tradi- tional and a Six Sigma organization is not black and white. In some cases, both kinds of anecdotes emanate from within the same firm. The reader might wish to reflect on how both kinds of examples apply to his or her business. The chapter starts by linking problem solving to financial performance, by estimat- ing organizational resources tied up fixing defects. Next, a few established methodol- ogies are compared against the define–measure–analyze–improve–control (DMAIC) approach associated with Six Sigma problem solving, followed by a review of how the other chapters of this handbook fit into the overall picture of problem solving. The chapter ends with a return to the discussion of roles that was started in Chapter 2, this time considering how the roles apply to successful problem solving. TABLE 14.1 The Six Sigma Balancing Act Patience Urgency • Allow the process to work • Attendance at meetings • Accept realistic scope • Complete assigned action items Containment Correction • Protect the customer • Identify the root cause • Temporarily higher expenses • Eliminate the problem for good Executive Hands-Off Executive Hands-On • Analytical tools • Infrastructure & reward system • Challenge by implementation • Strategic project selection • Resource allocations Flexibility Rigor • Deal with team dynamics • No shortcuts • Act on findings • Diversity on team Autonomy Accountability • “Worker bees” on teams • Participation not optional • Trust team’s intent & skill • Zero tolerance for obstruction • Share information • Provide guidelines & objectives SL3003Ch14Frame Page 296 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC Six Sigma Problem Solving 297 14.1 PRODUCT, PROCESS, AND MONEY A manufactured product is a physical object, with tangible properties that enable you to test its conformance to customer requirements. When a product contains one or more defects, it is called defective. Presumably, defects are not deliberate. They ensue from flaws in the processes that create the product. A variety of process problems can lead to defects in manufactured products: • Design errors • Defects in the materials • Defects in the manufacturing process • Errors in the processes that support the factory floor Problem-solving teams identify which process, and which aspect thereof, is responsible for the defects. They then identify and implement remedies, with the intent of preventing the defects from happening again. Later we discuss how this is done. First, however, managers will benefit from understanding the costs of fixing defective products once they occur. 14.1.1 D EFECTS PER U NIT (DPU) Consider a product. It could be a manufactured product such as a hammer, a service product such as tax preparation, or something in between, such as automobile repair. Suppose we are able to contain every defect, meaning that the delivered product contains zero defects (though this final supposition is most unrealistic, we beg the reader’s indulgence). Over time, we produce an average of one defect per unit of deliverable product, or one DPU. Whether this is a good or a bad number depends on the complexity of the product: if a unit were one jumbo jet, one DPU would be an excellent number indeed; 1 DPU would be horrendous if a unit was a single carpet tack. Figure 14.1 shows how 100 defects might be distributed among a sample of 100 units. This typically is modeled using the Poisson distribution: Y TP ≅ e − DPU (14.1) In Equation 14.1, Y TP is called throughput yield . It is the probability that a given unit is nondefective. In Figure 14.1, DPU = 1.0, which corresponds to a value of Y TP ≅ 37%; thus, 37 of the units contain zero defects.* 14.1.2 T HROUGHPUT Y IELD ( Y TP ), K , AND R So how does this relate to managing a business? It comes down to how much it costs the business to fix defective product. Some have called the rework process * Over time, a process averaging 1 DPU should average approximately 37% defect-free units. However, any single sample is likely to vary somewhat from the expected value. SL3003Ch14Frame Page 297 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC 298 The Manufacturing Handbook of Best Practices “the hidden factory,” because rework usually is mixed in with first-pass product.* Because the two product streams are mingled, computing the magnitude of the hidden factory is difficult, especially using traditional cost accounting. Fortunately, we can use Y TP to estimate this magnitude, based on the following: R ≅ 1 + K ⋅ (1 – Y TP ) (14.2) In Equation 14.2, R represents the amount of resources required to produce and rework a product, including the 100% necessary to do everything just once. From Equation 14.1 we can tell that if DPU is low, then Y TP is nearly 1. From Equation 14.2, we can see that if Y TP approaches 1, then R does, too. In other words, low defect rates enable us to run our process very close to its “entitlement” level of R = 100%. However, as defect rates rise and Y TP falls, we must add extra resources to handle the rework caused by the (1 – Y TP ) units that contain one or more defects. The coefficient K quantifies the extra resources. To understand K, consider Figure 14.2, representing a ten-step process. Two defect scenarios are shown. In one, a defect is detected at step 3 and reworked at step 2. For this defect, the value of K is one step repeated out of a total of ten, or K = 1/10 = 0.1. However, we also show a defect detected at step 10 and reworked at step 1. What is not shown for the rework at step 1 is whether the product can be returned immediately to step 10, or whether it must pass through the entire process all over again. The answer depends as much on the type of defect as on the type of product. FIGURE 14.1 How 1 DPU might appear in 100 units. * One exception occurred on a certain automotive assembly line in Europe, where a full 1/3 of the factory floor was designated for fixing defects. 2021300200 0011001010 2112010102 0001010122 1201012022 3012021111 0110101240 2002113002 1213012221 0022210210 SL3003Ch14Frame Page 298 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC Six Sigma Problem Solving 299 Here the value of K can be anything from 0.1 to 0.9. In fact, K can be any value greater than zero, in light of other resource requirements: • Product disassembly • Problem diagnosis • Reviews, paperwork, and administrative support • Redundant inspections • Rework that fails to rectify the problem • Queues • Inventory: tracking, adjustments, expediting • Delayed shipments • Escaping defects A Six Sigma problem-solving team may be able to estimate an average value of K . However, it takes a lot of work to do so. Also, process changes that reduce defect rates are likely to alter the value of K . As a rule of thumb, consider using a value of K ≅ 0.5. Though this tends to be on the low side of reality, the following discussion will show its impact. 14.1.3 A N E XAMPLE C ALCULATION Consider a process with DPU ≅ 2.3. Based on Equation 14.1, the resulting Y TP ≅ 0.1, meaning that only 10% of product starts completes all steps of production defect- free. Using the default value of K = 0.5, we can use Equation 14.2 to estimate that R ≅ 1 + 0.5 ⋅ (1 – 0.1) = 1 + (0.5 ⋅ 0.9) = 1.45 Thus, rework consumes an estimated 45% more resources — floor space, capital equipment, personnel, etc. — than it should take to do the job right the first time. Putting it another way, approximately 31% of the process’s resources are consumed fixing defects. Suppose this team was able to reduce defects by 75% — an accomplishment that is fairly routine in Six Sigma problem solving. Table 14.2 shows the before and after numbers. Note that the reduction in the hidden factory is 42%, which is less than the reduction in defects. FIGURE 14.2 Various rework scenarios. 1 2 4 5 6 7 8 9 Is rework complete after step 2 (short dotted line)? Or must the entire process be repeated (long dotted line)? 10 3 SL3003Ch14Frame Page 299 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC 300 The Manufacturing Handbook of Best Practices Consider the ramifications of DPU and hidden factory. • DPU provides ease of measurement and process information. • Hidden factory estimates the financial impact of waste due to defects. This indicates why Six Sigma seeks eventually to achieve even lower defect levels and how such improvements relate to financial performance. 14.1.4 E SCAPING D EFECTS Recall that we started this discussion by presuming that all defects could be detected and contained. In reality, that seldom is the case. A rule of thumb is that one stage of visual inspection detects 85 to 90% of all defects.* Let us apply this to the process described in Table 14.2, presuming that the 2.3 DPU represent 87.5% of all defects, detected using a single visual inspection stage: DPU Actual ≅ 2.30 ÷ 0.875 = 2.63 DPU Delivered ≅ 2.63 – 2.30 = 0.33 ( Y TP ) Delivered ≅ e –0.33 = 72% 1 – ( Y TP ) Delivered ≅ 28% Thus, approximately 28% of the delivered product contains at least one defect. If customer complaint data show a lower rate, the business may have to contend TABLE 14.2 Impact of 75% Reduction in DPU Before 6 σ After 6 σ Defects Detected & Reworked Defects per unit (DPU) 2.30 0.58 Throughput yield ( Y TP ) 0.10 0.56 R value 1.45 1.22 % Hidden factory 31% 18% Hidden factory reduction 42% Escaping Defects Total DPU (detected + estimated escaping) 2.63 0.66 Escaping DPU 0.33 0.08 Field Y TP 72% 92% Shipped units defective 28% 8% * Automated inspection systems have become popular lately. However, the reader is cautioned: though their speed is indisputable, many have fared poorly in tests of accuracy. SL3003Ch14Frame Page 300 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC Six Sigma Problem Solving 301 with customers who are silently dissatisfied. The second column shows how reducing defects by 75% cuts delivered defectives to 8%. One can reduce defects by adding subsequent inspections, each of which should detect roughly 85 to 90% of the remaining defects. In this case, we include in our estimates the cost of inspection resources. A brief exercise in these numbers shows why quality cannot be “inspected in” as anything but a temporary containment measure. 14.1.5 F INAL C OMMENTS ON DEFECTS AND MONEY The primary mission of Six Sigma problem solving is to eliminate defects. However, the activity includes gathering defect data, which provide an estimate of the financial impact of the team’s efforts. When we compare escaping defects with customer complaint data, we begin to understand how quality may be affecting more than just profits. As a temporary measure, we can institute more inspections. However, the object is to eliminate defects. Now that we have considered the financial ramifications of defects, let us proceed to the means by which defects are prevented from recurring. 14.2 BASICS OF PROBLEM SOLVING The literature abounds with descriptions of MAIC and DMAIC as models of Six Sigma problem solving. In truth, these are variations on themes that have been around for decades, starting with the granddaddy of them all: Shewhart’s and Dem- ing’s plan–do–study–act (PDSA). The effectiveness Six Sigma problem solving is based on the same principles that make many other team-based, problem-solving approaches effective. 14.2.1 BASIC PROBLEM SOLVING Consider briefly the overall activities in Six Sigma problem solving, similar perhaps to Figure 14.3. This summary does not describe any single methodology, but rather describes common aspects of the more effective approaches. Table 14.3 summarizes the activities and why they are important. Traditional problem solving is character- ized by the tendency to omit or abbreviate steps. In such environments, problems tend to hide and reappear at inconvenient times. In Figure 14.3, each row represents a community within a business, and the sequence of activities proceeds from left to right. The white box naming each activity encompasses the typical participants in that aspect of the problem-solving process. Finally, the crosshatched boxes represent groups that may be called upon periodically during a given activity. Note the distinction between Upper Management and Middle Management. Middle management tends to be closer to immediate process supervision, so they participate more than top management. Also note that Team is separate from Oper- ators, because one operator usually represents numerous peers in team activities. SL3003Ch14Frame Page 301 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC 302 The Manufacturing Handbook of Best Practices Finally, note that the stages of DMAIC appear across the top of Figure 14.3, but without distinct boundaries. Accomplished problem solvers recognize that hard boundaries simply don’t exist. FIGURE 14.3 Effective problem solving in manufacturing. TABLE 14.3 Steps in Effective Problem Solving Step Purpose(s) Signs of Success Project kick-off • Common understanding • Focus on process to fix instead of “Rules of Engagement” Ⅲ Objectives Ⅲ Scope Deliverables & requirements • Understand customers & needs • Objective metrics • Fix the right problem Describe “as is”• Qualitative process description • Quantified measures • Objective performance data • Cost of poor quality Root cause • Fix the right things • Consensus on “Vital Few” problem causes Remedies • Implement the right fixes • Consensus on “Vital Few” interventions Implement changes • Test drive revisions • Process improves as hoped Implement control • Make improvements permanent • Self-sustaining at improved levels Reap benefits • Reward contributors • Wait lists to join teams • Spread the message • Project ideas proliferate Middle Management Team Operators Customer(s) Upper Management Project Kick-Off Describe “As Is” Implement Changes NO YES Success ? RemediesRoot Cause Implement Control Reap Benefits Time → Define Measure Analyze Improve Control Deliverables & Requirements SL3003Ch14Frame Page 302 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC Six Sigma Problem Solving 303 14.2.2 COMPARISON OF METHODOLOGIES Between published literature, Internet sites, and consultants’ offerings, the apparent variety of problem-solving methodologies can be downright intimidating. One way to classify myriad materials might be to use the following categories: • Tools: techniques and activities used to achieve specific outcomes, such as gathering information or making decisions • Methodologies: frameworks in which sequences of tools are selected and applied to achieve broader objectives, such as project outcomes • Infrastructure: organizational interventions to enhance the business’s abil- ities to benefit from methodologies and tools The above list proceeds from the tactical to the strategic. That is, individuals can understand and apply some tools rather quickly, whereas infrastructure requires investing time and effort in both personal and organizational growth. The above categories can be used to create a rough classification of the chapters of this handbook, shown in Table 14.4. As the table indicates, there is considerable overlap among the classifications. At the methodology level, three approaches to problem solving are currently being used extensively: DMAIC (Six Sigma), lean manufacturing (kaizen), and Ford’s eight- discipline team-oriented problem solving (also called TOPS or 8D). Ultimately, all three adhere to the precepts of Figure 14.3, along with the PDSA philosophy. TABLE 14.4 Six Sigma Context of Handbook Chapters Infrastructure Methodologies Tools Six Sigma Management (Chapt. 2) Design of Experiments (DOE) (Chapt. 3) Supply Chain Management (Chapts. 16 and 17) Measurement System Analysis (MSA) (Chapt. 9) Integrated Product & Process Development (Chapt. 5) Process Analysis (Chapt. 10) Agile Enterprise (Chapt. 1) Design for Six Sigma (DFSS) (Chapt. 4) ISO 9001 (Chapt. 6) Design for Manufacture & Assembly (DFMA/DFSS) (Chapt. 4) ISO 14001 (Chapt. 7) Theory of Inventive Problem Solving (TRIZ) (Chapt. 19) Theory of Constraints (TOC) (Chapt. 18) Lean Manufacturing (Chapt. 8) Quality Function Deployment (QFD) (Chapt. 11) Six Sigma Problem Solving (Chapt. 14) Robust Design (Chapt. 13) Manufacturing Controls Integration (Chapt. 12) Statistical Quality/Process Control (SPC) (Chapt. 15) SL3003Ch14Frame Page 303 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC 304 The Manufacturing Handbook of Best Practices Table 14.5 provides a rough comparison of the steps associated with the approaches, with a brief summary of each step’s purpose. Table 14.6 provides some guidelines on the strength of the tools in specific problem-solving situations. Here is a brief description of how the three methods work: TABLE 14.5 Comparison of Problem-Solving Approaches PDSA 8D (TOPS) Lean (Kaizen) 6σ (DMIAC) Purpose Plan Form team Recognize Tie quality to strategy Define Prioritize projects & resources Describe problem Define actual performance Measure Finalize project scope Understand “as is” • Requirements • Procedures • Performance Define desired performance Contain symptoms Do ID & verify root causes Gather & analyze data Analyze Understand process behaviors • Key input variables • Sources of variation ID root causes Choose & verify corrective actions Remove root causes Improve Finalize what to change Study Implement permanent corrections Change procedures to sustain gains Control Sustain gains Act Prevent recurrence Standardize Standardize Become accustomed to new procedures Integrate Propagate improvements Celebrate Recognize & encourage success TABLE 14.6 Applicability of Problem-Solving Approaches Application Ford 8D (TOPS) Lean (KaiZen) 6σ (DMAIC) Manufacturing quality Strong Strong Strong Lean manufacturing Moderate Strong Moderate Transactional Moderate Moderate Strong Design Moderate Moderate Strong Infrasturcture Weak Weak Moderate SL3003Ch14Frame Page 304 Tuesday, November 6, 2001 6:04 PM © 2002 by CRC Press LLC [...]... two kinds of projects ** Note that this is an application of Six Sigma problem solving in a business where nothing is manufactured © 2002 by CRC Press LLC SL3003Ch14Frame Page 310 Tuesday, November 6, 2001 6:04 PM 310 The Manufacturing Handbook of Best Practices 14. 4.2 BALANCING CONTAINMENT AND CORRECTION When an organization targets a significant problem for correction, there often is a flurry of activity... by CRC Press LLC SL3003Ch14Frame Page 312 Tuesday, November 6, 2001 6:04 PM 312 The Manufacturing Handbook of Best Practices Flexibility applies to the variety of tools that can be selected at a given time and to how an effective facilitator responds to the dynamics of her or his diverse team Here, managers simply need to resist temptation to provide too much help, as with the particulate control team... across a number of projects Some of this was discussed in conjunction with Figure 14. 3, as well as Tables 14. 3 and 14. 5 Table 14. 9 attempts to bring together the various individuals’ roles in this context Here, the five steps of DMAIC are complimented by three more • Recognize is an outcome of establishing the organizational infrastructure • • from Chapter 2 It refers to the identification of high priority... Press LLC SL3003Ch14Frame Page 316 Tuesday, November 6, 2001 6:04 PM 316 The Manufacturing Handbook of Best Practices liaisons between the team and management They balance technical understanding of the process in question with appreciation for change management issues 14. 7.3 MIDDLE MANAGEMENT As used here, middle management refers to people responsible for the day-to-day operation of the processes... your reward” is not applied selectively in Six Sigma organizations; cost tracking based on DPU makes the task easier than ever © 2002 by CRC Press LLC SL3003Ch14Frame Page 314 Tuesday, November 6, 2001 6:04 PM 314 The Manufacturing Handbook of Best Practices team members, exposed to empowerment and the problem-solving process for the first time, have said things such as, “Finally, somebody is listening... CRC Press LLC SL3003Ch14Frame Page 306 Tuesday, November 6, 2001 6:04 PM 306 The Manufacturing Handbook of Best Practices Affinity diagrams Brainstorming Check sheets Conditional probability analyses Descriptive statistics Design of Experiments (DoE) (Chapt 3) Failure modes & effects analysis (FMEA) Fish-bone (cause & effect) diagram Flow charts/S.I.P.O.C Force field analysis Hypothesis testing Interrelationship... CRC Press LLC SL3003Ch14Frame Page 308 Tuesday, November 6, 2001 6:04 PM 308 The Manufacturing Handbook of Best Practices Ideas Affinity diagrams Brainstorming Check sheets Conditional probability analyses Design of Experiments (DoE) (Chapt 3) Descriptive statistics Failure modes & effects analysis (FMEA) Fish-bone (cause & effect) diagram Flow charts/SIPOC Force field anlaysis Hypothesis testing Interrelationship... down the findings and add a dash of practicality The emphasis is on how executive management balances the issues in Table 14. 1, in order to derive maximum organizational benefit from problem-solving teams 14. 4.1 BALANCING PATIENCE AND URGENCY At times we are inundated with unsolicited offers of rapid weight loss, quick college degrees, speedy prosperity, and so on The offers prey upon people’s desire... Containment must be identified as no more than one aspect of true problem solving, and people must be held accountable to achieve the latter Nobody can do this for top management To quote Juran, the task is “nondelegable.” 14. 4.3 BALANCING “HANDS ON” VS “HANDS OFF” The most traditional of organizations condition their managers to be providers of answers Much of this relates to Taylorism described in Chapter... successful: participants, obligations, and so on If there is disagreement, the parties work to resolve issues before the team is affected When the team is convened, the executive, champion, and coordinator attend briefly to thank the participants and attest to the importance of the project © 2002 by CRC Press LLC SL3003Ch14Frame Page 315 Tuesday, November 6, 2001 6:04 PM Six Sigma Problem Solving 315 TABLE 14. 9 . LLC 304 The Manufacturing Handbook of Best Practices Table 14. 5 provides a rough comparison of the steps associated with the approaches, with a brief summary of each step’s purpose. Table 14. 6 provides. 6:04 PM © 2002 by CRC Press LLC 302 The Manufacturing Handbook of Best Practices Finally, note that the stages of DMAIC appear across the top of Figure 14. 3, but without distinct boundaries. Accomplished. LLC 310 The Manufacturing Handbook of Best Practices 14. 4.2 BALANCING CONTAINMENT AND CORRECTION When an organization targets a significant problem for correction, there often is a flurry of activity