Part 3: Pareto Analysis and Check Sheets S

EM 8771 • January 2002 $2.50 PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY

Part 3: Pareto Analysis and Check Sheets S. Leavengood and J. Reeb

Part 1 in this series introduced the reader to Statistical Process Control, and Part 2 provided an overview of how and why SPC works. Part 3 begins the step-by-step process of building the practical skills necessary for hands-on implementation of SPC. This report discusses Pareto analysis, a tool we can use to help decide how and where to begin using SPC. We also discuss check sheets, which are data collection tools that may be used in Pareto analysis. Part 4 discusses flowcharts. Future publications in the series will discuss case histories of wood products firms using SPC, providing real-world evidence of the benefits of SPC and examining pitfalls and successful approaches. Where to begin an SPC program? Most manufacturing processes are sufficiently complex that at first glance it may seem impossible to decide where to begin using SPC techniques. SPC programs that attempt to monitor too many process variables are quickly overwhelmed by the time and labor required to collect, analyze, plot, and interpret the data. In such cases, SPC seems too time consuming and expen- sive to be of any benefit. The life expectancy of SPC in a company depends heavily on the results of the first few projects undertaken. With this kind of pressure, how do you decide where to begin? Obviously, we cannot measure everything. We must focus initially on the most important quality problems to get the “biggest bang for the buck.” This is especially true in the early stages of an SPC program when personnel are likely to be skeptical of SPC and hesitant to make the necessary changes.

Scott Leavengood, Extension wood products, Washington County; and James E. Reeb, Extension forest products manufacturing specialist; Oregon State University. STATISTICAL PROCESS CONTROL

Prioritizing quality problems for the company is a good first step. Then, determine which projects will have the highest return on investment and therefore should be the initial focus of quality improvement programs. Pareto analysis enables us to do all this. Pareto analysis Pareto1 (pronounced “pah-RAY-toe”) analysis uses the Pareto principle, also called the 80:20 rule, to analyze and display data. Quality expert J.M. Juran applied the principle to quality control and found that 80 percent of problems stem from 20 percent of the possible causes. The numbers 80 and 20 are not meant to be abso- lutes. The main point, as Juran stated, is that we should focus on the “vital few” problems (those in the 20-percent category) rather than on the “trivial many” to make the most significant improve- ments in product quality. Pareto charts are the graphical tool used in Pareto analysis. A Pareto chart is a bar chart that displays the relative importance of problems in a format that is very easy to interpret. The most important problem (for example, the one highest in cost, frequency, or some other measurement) is represented by the tallest bar, the next most important problem is represented by the next tallest bar, and so on. A check sheet is a useful tool for collecting data for Pareto charts. Check sheets Check sheets are relatively simple forms used to collect data. They include a list of nonconformities2 and a tally of nonconformities. Check sheets should also include the name of the project for which data is being collected, the shift when the items were produced, the names of persons collecting the data, dates of data collection and of production (if known), and the location of data collection (e.g., in house or at a customer’s).

1 Vilfredo Pareto was a 19th-century Italian economist who studied the distribution of income in Italy. He found that about 20 percent of the population controlled about 80 percent of the wealth. 2 A nonconforming product is one that fails to meet one or more specifications, and a nonconformity is a specific type of failure. A nonconforming product may be termed defective if it contains one or more defects that render it unfit or unsafe for use. Confusion of these terms has resulted in misunderstandings in product liability lawsuits. As a result, many companies have adjusted their internal terminology and now use the terms “nonconforming” and “nonconformity” in favor of “defect” and “defective.” 2 PARETO ANALYSIS AND CHECK SHEETS

Check sheets aren’t mandatory to construct Pareto charts. How- ever, because check sheets require you to standardize your list and definitions of nonconformities, they provide several benefits. First, people often do not agree on the major categories of nonconformities. Therefore, developing a list of common nonconformities (i.e., quality problems) is not as easy as it sounds. A good way to develop this list is to brainstorm with production personnel, management, QC personnel, and, most important, your Do not customers. underestimate... Second, people often do not agree on precisely what constitutes “nonconforming.” In other words, how bad does it have to be to the importance of get thrown in the scrap or rework pile? developing a standard Last, different people often will put a given item in different list of nonconformities categories. For example, one person may call an item with torn and precise definitions grain a machining defect, another might call it fuzzy grain, and for each. another may call it reaction wood. Without standard terminology and definitions, it becomes very difficult to conduct a Pareto analysis. To get an idea of the effect on your company of lack of standard- ized terminology and definitions for nonconformities, try a simple experiment. Select several items at random and ask different people to examine them and record nonconformities item by item. One experiment at a secondary wood products manufacturer involved five quality inspectors. The inspectors did not agree on the number of items that should be rejected due to quality problems (the scrap/rework rate varied from 34 to 49 percent) nor did they agree on the reasons for rejecting the products. Had we looked only at data collected by inspectors 1, 2, and 3, we would have concluded that torn grain and blue stain were the biggest quality problems. Had we looked only at data collected by inspectors 4 and 5, we would have concluded that dents (handling damage) and reaction wood were the biggest quality problems. Do not underestimate the importance of developing a standard list of nonconformities and precise definitions for each. The following demonstrates how to construct and interpret check sheets and Pareto charts.

Example The Quality Improvement Team at a manufacturer of wood compo- nents visited a customer and examined items in the scrap and rework bins. After looking at each item and talking with the customer, the team agreed on categories of nonconformities and developed precise definitions for each category. They created a 3 STATISTICAL PROCESS CONTROL

check sheet, then inspected each item and tallied the number of occurrences (frequency) for each cause of nonconformity. Figure 1 presents the results.

Project Quality Improvement Project Name QIT Shift All Location Customer A Dates January 2002 Rel. Cum Reason Freq. Freq. (%) Freq Size out of specification 194 Loose knots 18 Raised grain 4 Dents 3 Stain/rot 31 Fuzzy grain 105 Splits 11 Machine tear-out 61 Burn marks 44 Oil/grease marks 2 Total 473 Figure 1.—A sample check sheet.

Nonconformities were sorted from highest to lowest frequency, and the relative frequency for each was determined (Figure 2). For example, “size out-of-specification” was 194 out of 473 nonconformities, and so the relative frequency for size-out-of specification was: 194/473 = 0.41 = 41%

An optional final step is to calculate cumulative relative frequency. Cumulative relative frequency helps the user to readily see the combined effect of the “vital few” problems. For example, you could see that the top three quality problems were responsible for nearly 80 percent of the problems overall. To calculate cumulative relative frequency, add the relative frequency for each category of nonconformity to the sum of all preceding relative frequencies. For example, there were 194 occurrences of size out-of-specification or 41 percent (relative frequency) of the total. There were 105 occurrences of fuzzy grain. Fuzzy grain was therefore responsible for 22 percent of the total. Size out-of-specification and fuzzy grain combined (cumulative relative frequency) were responsible for 63 percent of the total. Size out-of-specification, fuzzy grain, and machine tear-out combined were responsible for 76 percent of the

4 PARETO ANALYSIS AND CHECK SHEETS

total. The cumulative relative frequency for the least frequent category (oil/ grease marks, in this example) should be 100 percent, however it is slightly less due to rounding. Figure 2 shows the check sheet with the nonconformities arranged in descending order of frequency and with relative frequency and cumulative relative frequency calculated. m. Rel. q. (%) Project Quality Improvement Project Name QIT Shift All Location Customer A Dates January 2002 Rel. Cum. Rel. Reason Freq. Freq. (%) Freq. (%) Size out of specification 194 41 41 Fuzzy grain 105 22 63 Machine tear-out 61 13 76 Burn marks 44 9 85 Stain/rot 31 7 92 Loose knots 18 4 96 Splits 11 2 98 Raised grain 4 0.8 98.8 Dents 3 0.6 99.4 Oil/grease marks 2 0.4 99.8 Total 473 99.8 Figure 2.—A sample check sheet showing nonconformities in descending order as well as relative frequency and cumulative relative frequency.

Figure 3 (page 6) is the Pareto chart for the data in Figure 2. The left vertical axis indicates the number (frequency) of each type of nonconformity. Always plot nonconformities in descending order of frequency, with the most frequent at the left vertical axis. The right axis indicates cumulative frequency. The Pareto chart makes it easy to see that size out-of-specification, fuzzy grain, and machine tear-out are the major nonconformities. Quality improvement that focuses on these items will give the “biggest bang for the buck.” Frequency, however, is not the only important consideration. Certain types of nonconformities, even if infrequent, may be very costly to scrap or rework. Therefore, the Pareto analysis should take into account both cost and frequency. Though scrap and rework often involve very different costs, it’s possible to calculate an average scrap and rework cost based on the

5 STATISTICAL PROCESS CONTROL

percentage of product in each category of nonconformity. For The Pareto analysis... example, let’s say we estimate that 10 percent of material with size out-of-specification must be scrapped, but the remaining 90 per- should take into cent can be reworked to produce a usable product. Further, let’s say account both cost that scrapping the product represents a loss of approximately $20 and frequency. per item, and reworking costs approximately $11 per item. There- fore, our estimate of the average scrap and rework cost for size out-of-specification is: (scrap cost) x (% scrap) + (rework cost) x (% rework) = scrap & rework cost ($20) x (10%) + ($11) x (90%) = $12

To account for frequency as well as scrap and rework costs, multiply relative frequency by cost to obtain relative cost. For example, we already determined that approximately 41 percent of nonconformities were size out-of-specification. Therefore, the relative cost due to size out-of-specification is: 0.41 x $12 = $4.92

Pareto chart

200 100

160 80

120 60

Frequency 80 40

40 20 Cumulative frequency (%)

0 0

Stain/rot

Machine tear-out

Fuzzy grain

Loose knots

Burn marks

Splits

Raised grain

Oil/grease marks

Dents

Size out-of-spec Cause of nonconformity Figure 3.—Pareto chart for the data in Figure 2.

6 PARETO ANALYSIS AND CHECK SHEETS

Rel. Cost Rel. Freq. Cum. Rel. Nonconformity ($) (%) Freq. (%) Size out-of-spec. 4.92 38 38 Machine tear-out 2.34 18 56 Fuzzy grain 1.76 13 69 Stain/rot 1.75 13 82 Loose knots 1.00 8 90 Burn marks 0.72 6 96 Splits 0.32 2 98 Dents 0.09 0.7 98.7 Raised grain 0.06 0.5 99.2 Oil/grease marks 0.03 0.2 99.4 Total 12.99 99.4 Table 1.—Nonconformities and relative costs.

Table 1 shows the relative costs, and Figure 4 shows the corre- sponding Pareto chart. We can see that size out-of-specification is the primary nonconformity from the standpoint of frequency (Figure 3) as well as relative cost to scrap or rework (Figure 4). Therefore, to get the

Pareto chart

5 100

4 80

3 60

2 40

Relative cost

1 20 Cumulative frequency (%)

0 0

Stain/rot

Machine tear-out

Fuzzy grain

Loose knots

Burn marks

Splits

Raised grain

Oil/grease marks

Dents

Size out-of-spec Cause of nonconformity Figure 4.—Pareto chart for the data in Table 1.

7 STATISTICAL PROCESS CONTROL

“biggest bang for the buck,” it would be wise to begin the SPC program by focusing on problems that lead to size out-of- specification. Conclusions We now know the primary nonconformities and therefore where to focus initial efforts of an SPC program. We do not yet know, however, the specific processing steps that lead to a given nonconformity—that is, where and how the problem arises—and therefore we do not yet know where or what to monitor. To help us discover the specific steps in the process that lead to a given nonconformity, it is helpful to develop a flowchart for the process. Flowcharts are the subject of the next report in this series. For further information Brassard, M. and D. Ritter. 1994. The Memory Jogger II: A Pocket Guide of Tools for Continuous Improvement & Effective Plan- ning (Methuen, MA: Goal/QPC). 164 pp. http://www.goalqpc. com/ Grant, E.L. and R.S. Leavenworth. 1988. Statistical Quality Con- trol, 6th ed. (New York: McGraw Hill). 714 pp. Ishikawa, K. 1982. Guide to Quality Control (Tokyo, Japan: Asian Productivity Organization). 225 pp. Montgomery, D.C. 1996. Introduction to Statistical Quality Con- trol, 3rd ed. (New York: John Wiley & Sons). 677 pp. Walton, M. 1986. The Deming Management Method (New York: Putnam Publishing Group). 262 pp.

8 PARETO ANALYSIS AND CHECK SHEETS

PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY ABOUT THIS SERIES

This publication is part of a series, Performance Excellence in the Wood Products Industry. The various publications address topics under the headings of wood technology, marketing and business management, production management, quality and process control, and operations research. For a complete list of titles in print, contact OSU Extension & Station Communications (address below) or visit the OSU Wood Products Extension Web site at http://wood.orst.edu

Ordering information To order additional copies of this publication, send the complete title and series number, along with a check or money order for $2.50 payable to OSU, to: Publication Orders Extension & Station Communications Oregon State University 422 Kerr Administration Corvallis, OR 97331-2119 Fax: 541-737-0817

We offer a 25-percent discount on orders of 100 or more copies of a single title. You can view our Publications and Videos catalog and many Extension publications on the Web at http://eesc.orst.edu

9 STATISTICAL PROCESS CONTROL

10 PARETO ANALYSIS AND CHECK SHEETS

11 STATISTICAL PROCESS CONTROL

© 2002 Oregon State University This publication was produced and distributed in furtherance of the Acts of Congress of May 8 and June 30, 1914. Extension work is a cooperative program of Oregon State University, the U.S. Department of Agriculture, and Oregon counties.

Oregon State University Extension Service offers educational programs, activities, and materials—without discrimination based on race, color, religion, sex, sexual orientation, national origin, age, marital status, disability, or disabled veteran or Vietnam-era veteran status. Oregon State University Extension Service is an Equal Opportunity Employer. Published January 2002.

12 EM 8772 • January 2002 $2.50 PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY

Part 4: Flowcharts S. Leavengood and J. Reeb

Part 1 in this series introduced the reader to Statistical Process Control, and Part 2 provided an overview of how and why SPC works. Part 3 began the step-by-step process of building the practical skills necessary for hands-on implementation of SPC. It discussed Pareto analysis, a tool to help decide where to focus initial efforts. Part 4 discusses flowcharts. Part 5 in the series will continue building implementation skills by discussing cause-and-effect diagrams. Future publications in the series will discuss case histories of wood products firms using SPC, providing real- world evidence of the benefits of SPC and examining pitfalls and successful approaches. What’s the next step in implementing SPC? After achieving top management’s commitment to using SPC, the next step in beginning an SPC program is to determine where to focus initial efforts to get the “biggest bang for the buck.” In Part 3, we presented Pareto analysis as a tool to locate the primary causes of nonconformities and therefore where to focus initial efforts. Now we need to know which specific activities in the process cause the nonconformity and which quality characteristic(s) to monitor. An example will help to clarify the above discussion and the objective of this report. The Pareto analysis conducted in Part 3 of this series revealed “size out-of- specification” as the major nonconformity, from the standpoint of both frequency and relative cost to scrap or rework. We now need to know: • The specific step or steps in the process (e.g., dry kilns, rip and chop, moulding) responsible for causing size out-of-specification • The quality characteristic (e.g., moisture content, width, thickness, motor amps, or proportion of nonconforming parts) to measure

Cause-and-effect diagrams are commonly used to identify specific activities responsible for causing nonconformities. However, we have chosen to discuss flowcharts first, postponing a discussion of cause-and effect diagrams until Part 5 in

Scott Leavengood, Extension wood products, Washington County; and James E. Reeb, Extension forest products manufacturing specialist; Oregon State University. STATISTICAL PROCESS CONTROL

this series. Our choice is based on the fact that flowcharts have Flowcharts can been found to be valuable tools for initiating discussion during reveal… cause-and-effect analysis and for ensuring that everyone under- non-value-added stands and agrees on what really happens—rather than what’s activities such as supposed to happen—in the manufacturing process. inspection, rework, redundant steps, and Flowcharts bottlenecks. Flowcharts graphically represent the steps in creating a product or service. The process of creating a chart is often beneficial because personnel may be unaware of all the “nitty-gritty” details involved in producing the product. Also, people often are surprised to learn of the differences between the ideal process flow and what actually occurs in the mill. This is particularly true when the team developing the chart includes representatives of all departments of the plant, not just production personnel. In addition to understanding processing steps, flowcharts provide other benefits. If detail is sufficient, flowcharts can help to reveal non-value-added activities such as inspection, rework, redundant steps, movement, unnecessary processing loops, and bottlenecks. From the standpoint of SPC, flowcharts also help to reveal the stages in the process where data may be collected. Flowcharts are also excellent tools for training new hires. Brassard and Ritter (1994) list six steps to flowchart development. 1. Determine the start and stop points the chart will cover. 2. List the major steps (inputs, decisions made, activities, inspection, delays, and outputs) in the process. 3. Put the steps in the proper order. 4. Draw the flowchart. 5. Test the flowchart for accuracy and completeness. 6. Look for opportunities to improve the process (i.e., reduce non- value-added activities). Developing a flowchart: An example We will demonstrate flowchart development using a secondary wood products manufacturer as an example.

Background XYZ Forest Products Inc. produces wooden handles for push brooms. Their customers produce finished brooms by adding a rubber grip to the top of the handle, inserting a threaded metal ferrule to the bottom of the handle, and attaching the broom head.

2 FLOWCHARTS

Last year, business began to fall off for XYZ; orders dropped 40 percent in just 6 months. Several customers stated that the competition’s quality was better. A few customers had begun asking XYZ to provide documentation of process performance— namely histograms, control charts, and process capability indices (see Part 2 in this series for an overview of these subjects). There- fore, XYZ was inspired to use SPC. Because customers reported several different quality problems (fuzzy grain, size out-of-spec., warp, etc.), XYZ personnel did not know precisely how and where to start their quality improvement program. They conducted the Pareto analysis, as presented in Part 3 in this series, to help them decide where to focus initially. Size out-of-specification was found to be the primary quality problem. Following the Pareto analysis, the general manager of XYZ con- vened a team of personnel from engineering, sales, production, quality control, and management to develop a flowchart for their process. We will summarize their activities using the six steps described above. Creating the flowchart

Step 1. Determine the start and stop points that the chart will cover. Because XYZ had never developed a flowchart for the process, the team decided to chart the process from start to finish. The start point was green lumber receiving, and the stop point was finished product storage. The team agreed to create a macro-flowchart; that is, a chart showing only the general flow of the process with minimal detail. The team decided that once they’d created a cause- and-effect diagram for the problem, and had determined the specific steps in the process most likely responsible for the problem, they would then create a flowchart with a narrower focus and more detail.

Steps 2 and 3. List the major steps in the process, and put the steps in the proper order. The team brainstormed (see Brassard and Ritter for a discussion of brainstorming) to develop the steps involved in the process. Then, they put the steps in the proper sequence. (Brassard and Ritter list steps 2 and 3 separately because, in a group setting, people usually name the activities most familiar to them, which

3 STATISTICAL PROCESS CONTROL

generally leads to a list of steps that is out of sequence). In our It is imperative… example, the team identified these steps. • Receive rough green lumber; tally. to list what actually • Sticker lumber. happens during • Move stickered lumber to green storage. production versus the • Move lumber to dry kilns. ideal for the process. • Kiln dry lumber. • Unsticker, tally, and stack dry lumber. • Move lumber to dry storage. • Move lumber to planer. • Unload and plane lumber. • Crosscut surfaced lumber. • Rip lumber to handle blank widths. • Tally handle blanks. • Shape broom handles from blanks. • Inspect handles with go/no-go gauge; tally and scrap no-go. • Load and move good handles from shaper to taperer. • Taper ferrule end. • Round grip end of handles. • Inspect handles for appearance; tally and send nonconforming to scrap and rework. • Load and move handles to sander. • Sand handles. • Load and move handles to packaging. • Package. • Move packaged handles to finished product storage.

Note: It is imperative to list what actually happens during production versus the ideal for the process. For example, if lumber leaving the planer goes to storage, as opposed to going directly to the crosscut saws as listed above, this should be specified.

Step 4. Draw the flowchart. Symbols are used in flowcharting to identify different categories of activity. For example, ovals may be used to indicate inputs/ outputs, and boxes indicate a processing step (Figure 1). It is important to maintain a consistent level of detail in the flowchart. Brassard and Ritter suggest the amount of detail to include in a flowchart. Macro-level flowcharts show key action steps but no decision boxes. Intermediate-level flowcharts show action and decision points, and micro-level flowcharts show intricate details.

4 FLOWCHARTS

Each step in the process should be labeled. Arrows should be used to indicate the flow of steps. To make the chart easier to read, Inputs it is helpful when using yes/no decision boxes to have the “yes” and boxes branch down and the “no” boxes branch to the left. This outputs will, of course, depend on the amount of space available. For future reference, names of team members, the date, and the pur- Processing pose for creating the chart should be included (Figure 2, page 6).

Step 5. Test the flowchart for accuracy and completeness. The team should make certain that symbols are used correctly, Decision process steps are identified clearly, and that process loops are closed (that is, every path flows to a logical end). Also, if the chart contains any process boxes with more than one output arrow, the Storage team may wish to consider adding a decision diamond. As a final check, someone outside the team should be asked to verify the chart’s accuracy and completeness. Delay Step 6. Look for opportunities to improve the process (reduce non-value-added activities). Data entry This is where the team seeks opportunities to optimize the process. An ideal process flowchart should be made and compared to the actual process flowchart. The team should then examine the non-value-added activities, which might include the following. • Unnecessary redundancy. (Two machines performing the same Movement operation might be necessary redundancy if they increase throughput without creating bottlenecks; multiple inspection points for the same quality characteristic are often unnecessary Inspection redundancy.) • Inspection Figure 1.—Flowchart symbols. • Delay • Many movements (for example, movement to a staging area, then to storage, then to another holding area, and then to production).

Montgomery suggests several ways to eliminate non-value- added activities. • Rearrange the sequence of worksteps. • Rearrange the physical location of the operator in the system. • Change work methods. • Change the type of equipment used in the process. • Redesign forms and documents for more efficient use. • Improve operator training.

5 STATISTICAL PROCESS CONTROL

• Improve supervision. • Identify more clearly the function of the process to all employees (flowcharts are good visual aids for explaining the process to employees). • Eliminate unnecessary steps. • Consolidate process steps.

A macro-level flowchart (Figure 2) lacks the necessary detail to identify non-value-added activities. Once XYZ team members have constructed a cause-and-effect diagram for the defect category, they will know the step(s) in the process for which they need a more detailed flowchart. Consider, for example, that the team determines shaping through sanding as the processing steps that deserve a closer look for size out-of-specification troubles. Their flowchart for this part of the process may look like the charts in Figures 3 and 4.

Macro-flowchart Green Sticker Storage Kiln dry XYZ, Inc. lumber 12/17/01 Team members S. Johnson Storage Unsticker Dried Plane lumber B. Jones and stack T. Williams B. Simonsen E. Fredricks Handle Rip Shape W. Harold Crosscut blanks Purpose Address customer concerns re: size Finished out-of-spec. handles Sand Round Taper

Package Storage

Figure 2.—Sample macro-flowchart.

6 FLOWCHARTS

From Handle Load blanks Micro-flowchart ripsaws blanks on pallet XYZ, Inc. 01/4/02 Move Move Move pallet to pallet pallet to Team members shaper 1 to shaper 3 shaper 2 S. Johnson Load blanks B. Jones Load blanks Load blanks into shaper 1 T. Williams into shaper 2 into shaper 3 B. Simonsen E. Fredricks W. Harold Shape Shape Shape Purpose Address customer concerns re: size Load handles Load handles on pallet Load handles on pallet out-of-spec. on pallet Focus on shaping through sanding

Delay Load handles Delay into taperer

Check Inspect. shape with go/no-go gauge. Shape OK? No?

Yes?

Tally Taper

Scrap Continued on page 8

To chipper

Figure 3.—Sample micro-flowchart, part 1.

7 STATISTICAL PROCESS CONTROL

Potential areas for improve- From page 7 ment are revealed in Figure 3. Notice the delay at the taper machine. Three shapers feed one taper machine which Round appears to lead to a bottle- neck. More detailed data (downtime, throughput, costs, Load handles etc.) would need to be col- on pallet lected to determine a solution. Move pallet Another area to examine is to sander dept. the two inspection points, one before the taper machine and Inspect. the other before the sander. Free from nonconformities? Handles are inspected for conformance to size specifications at the infeed to the taper machine and are checked for No? Yes? appearance at the infeed of the Inspect. sander. The team might Reworkable? address numerous questions, including: Load handles into sander 1. Are both inspection points No? Yes? necessary? Could the product be inspected for Sand both size and appearance before the taper machine? 2. Could appearance be Tally Tally checked earlier in the Finished handles process? It probably isn’t cost effective to check for Scrap Rework conformance to appearance specifications after signifi- Tally cant value has been added to the product. To To 3. If there is a problem with chipper patchline Load handles conformance to size specifi- on pallet cations before the taper machine, can it be deter- Move pallet mined which of the shapers Figure 4.—Sample micro-flow chart, to packaging part 2. is the likely source of the problem? Are size data fed back to the operators? 8 FLOWCHARTS

4. Can the handles be checked with calipers instead of go/no-go gauges? Much more information is obtained using measurement data than go/no-go information. For example, a go/no-gauge might reveal that handles are “small” after they go out of specification. Charting data obtained with calipers, on the other hand, would enable the operator to detect trends and make corrections before the product went out-of-spec. Let’s examine one more potential area for improvement. Notice all the movements in Figure 3. This company probably has a fleet of forklifts. Product is loaded on pallets, moved, and unloaded many times. How might throughput increase if the process flow were improved by, for example, using just in time (JIT) or lean manufacturing techniques such as work cells, which are groups of machines dedicated to producing a particular product or part. That question can be addressed by creating another type of flowchart known as a value stream map. These maps track the flow of value and information from customer order all the way back to first-tier suppliers. Value stream maps add a dimension—time— that flowcharts don’t cover. By tracking process cycle times, equipment uptimes, and inventories, companies can estimate the amount of time they spend doing things the customer would not be willing to pay for (movement, queues, delays due to large batches, problems related to the scheduling system, rework, etc.) versus time spent altering the product in ways the customer will pay for (generally, those are process cycle times). The current value stream map is used to redesign the process to reduce non-value-added time (thus eliminating waste) and reduce customer lead time. A detailed discussion of value stream mapping is beyond the scope of this report. For more information, see Rother and Shook. Conclusion We now have graphical representations of the steps involved in creating the product. In the process of creating the chart, we have had the opportunity to increase company personnel’s understanding of “how we do things around here” and perhaps also to stream- line the process and reduce non-value-added steps. We now also have a valuable tool for initiating discussion during cause-and- effect analysis, the next step in beginning an SPC program.

9 STATISTICAL PROCESS CONTROL

For further information Brassard, M. and D. Ritter. 1994. The Memory Jogger II: A Pocket Guide of Tools for Continuous Improvement & Effective Plan- ning (Methuen, MA: Goal/QPC). 164 pp. http://www.goalqpc. com/ Grant, E.L. and R.S. Leavenworth. 1988. Statistical Quality Con- trol, 6th ed. (New York: McGraw Hill). 714 pp. Ishikawa, K. 1982. Guide to Quality Control (Tokyo, Japan: Asian Productivity Organization). 225 pp. Montgomery, D.C. 1996. Introduction to Statistical Quality Con- trol, 3rd ed. (New York: John Wiley & Sons). 677 pp. Rother, M. and J. Shook. 1999. Learning to See: Value Stream Mapping to Add Value and Eliminate Muda, v. 1.2 (Brookline, MA: The Lean Enterprise Institute). 102 pp. http://www.lean.org Walton, M. 1986. The Deming Management Method (New York: Putnam Publishing Group). 262 pp.

10 FLOWCHARTS

PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY ABOUT THIS SERIES

We offer a 25-percent discount on orders of 100 or more copies of a single title. You can view our Publications and Videos catalog and many Extension publications on the Web at http://eesc.orst.edu

11 STATISTICAL PROCESS CONTROL

12 PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY

PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY

EM 8984-E • August 2009 PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY

Part 5: Cause-and-Effect Diagrams Scott Leavengood and James E. Reeb PERFORMANCE EXCELLENCE OurIN focusTHE WOODfor the first PRODUCTS four publications INDUSTR in thisY series has been on introducing you to Statistical Process Control (SPC)—what it is, how and why it works, and then discussing some hands-on tools for determining where to focus initial efforts to use SPC in your company. Experience has shown that SPC is most effective when focused on a few key areas as opposed to the shotgun approach of measuring anything and everything. With that in mind, we presented check sheets and Pareto charts (Part 3) in the context of project selection. These tools help reveal the most frequent and costly quality problems. Flowcharts (Part 4) help to build consensus on the actual steps involved in a process, which in turn helps define precisely where quality problems might be occurring and what quality characteristics to monitor to help solve the problems. In Part 5, we now turn our attention to cause-and-effect diagrams (CE diagrams). CE diagrams are designed to help quality improvement teams identify the root causes of problems. In Part 6, we will continue this concept of root cause analysis with a brief introduction to a more advanced set of statistical tools: Design of Ex- periments. It is important, however, that we do not lose sight of our primary goal: improving quality and in so doing, improving customer satisfaction and the profitability of the company.

We’ve identified the problem; now how can we solve it? In previous publications in this series, we have identified the overarching quality problem we need to focus on and developed a flowchart identifying the specific steps in the process where problems may occur. We now need to narrow our focus so that we know what is causing the problem—and therefore how it can be solved. Continuing our example from Parts 3 and 4, we determined that “size out of specification” for wooden handles was the most frequent and costly quality problem. The flowchart showed that part size/shape was inspected with a “go/no-go” gauge at the infeed to a machine that tapers the handles. The results of go/no-go inspection are either that the shape is acceptable (“go”), in which case the parts were loaded into the tapering machine, or that the shape is not acceptable (“no go”), in which case the parts are scrapped. However, customers are still indicating that the sizes of the handles are not meeting their specifications.

Scott Leavengood, director, Oregon Wood Innovation Center, Oregon State University; and James E. Reeb, Extension forester, Lincoln County, Oregon State University. Statistical Process Control

In short, our prior efforts have helped us identify what the problem is and where it might be occurring in the process. We still do not know, however, what to do to solve the problem because we do not know what might be causing the problem. Once we identify and confirm a solution, we can take steps to closely monitor the situation such that the solution is maintained over time.

Cause-and-Effect Diagrams A cause-and-effect (CE) diagram is a graphical tool for organizing and dis- playing interrelationships of various theories of the root cause of a problem. CE diagrams are also commonly referred to as fishbone diagrams (due to their re- semblance to a fish skeleton) or as Ishikawa diagrams in honor of their inventor, Kaoru Ishikawa, a Japanese quality expert. Like flowcharts, CE diagrams are typically constructed as a team effort; and as with many team efforts, the process is often more important than the end product. When a team is brought together to study potential causes of a problem, each member of the team is able to share their expertise and experience with the problem. The team approach enables clarification of potential causes and can assist with building consensus for most likely causes. By empowering the team to identify the root cause and its solution, the team gains ownership of the process and is far more motivated to implement and maintain the solution over the long term. Perhaps most importantly, using a team to develop a CE diagram can help to avoid the all-too-common challenge of pet theories. Pet theories might arise when someone asserts that he or she already knows the cause of a problem. The person(s) presenting this theory may well be right, and if they are in a position of authority, chances are their theory will be the one that gets tested! There are risks, however, in simply tackling the pet theory. If the theory is in fact wrong, time and resources may be wasted, and even if the theory is correct, future team efforts will be stifled, since team members may feel their input to problems is neither needed nor valued. Further, the theory may be only partially correct: It might address a symptom or secondary cause rather than the actual root cause. CE diagrams, instead, bring the team together to identify and solve core problems. Brassard and Ritter (1994) list two common formats for CE diagrams: • Dispersion analysis: The diagram is structured according to major cause categories such as machines, methods, materials, operators, and environments. • Process classification: The diagram is structured according to the steps involved in the production process such as incoming inspection, ripping, sanding, moulding, etc. We will discuss the developing a CE diagram via an example.

Developing a cause-and-effect diagram XYZ Forest Products Inc. produces wooden handles for push brooms. Com- pany representatives visited a customer facility and examined the contents of the scrap and rework bins. Through the use of a check sheet and a Pareto chart, they

2 Cause-and-Effect Diagrams were able to identify “size out of specification” as the most frequent and costly quality problem. A flowchart helped build team consensus on the actual (vs. ideal) steps involved in the manufacturing process and enabled the team to identify points in the process where the problems might occur, as well as where measure- ments were currently being taken. To be able to address this problem, the team members must now identify the root cause and then determine and test potential solutions. For the long term, they will need a plan to ensure that their solution to the problem becomes standard operating procedure. CE diagrams are often developed via a brainstorming exercise. Brainstorming can be either a structured or unstructured process. In a structured process, each member of the team takes a turn in presenting an idea. In unstructured brainstorming, people simply present ideas as they come. Either approach may be used, however the advantage of the structured approach is that it elicits ideas from everyone—including more shy members of the team. The following steps are taken to develop a CE diagram: 1. Clearly define the problem (effect): Ensure the problem is clearly stated and understood by everyone. In the example here, it would be good to ensure that everyone understands specifically what “size out of specification” means. In this case, the team might create a definition such as, “The diameter of the broom handle measured at the bottom tip is either too large or too small to meet our customers’ specifications of ± x inches.” The bottom line for CE diagrams is that there is only one clearly defined effect being examined. The process focuses primarily on the causes—of which there will likely be far more than one. 2. Decide on format: The team should determine if the dispersion analysis or process classification (described above) is most appropriate for the situation. Either approach is acceptable. The primary concern is which format works best for the group and the problem being explored. For our purposes, we will focus on the dispersion analysis approach. 3. Draw a blank CE diagram: The diagram should look like Figure 1. The effect or problem being studied is entered in the box on the right-hand side. The main backbone is then drawn, followed by angled lines for the various cause categories. In this case, we have entered the common dispersion analysis categories of machine, methods, materials, operator, and environment. 4. Brainstorm causes: The team can now begin brainstorming potential causes of the problem. It is typical for causes to come in rapid-fire fashion unrelated to categories on the diagram. The meeting facilitator will have to enter the causes in the appropriate place on the diagram. If ideas are slow in coming, however, the facilitator might address each of the categories one at a time with questions such as, “Could our machinery be leading to handle size being outside the specifications?” 5. “Go for the root” (cause): As the team discusses some of the causes, it will become apparent that there are underlying causes for some items. For example,

3 Statistical Process Control

under materials, someone might mention wood moisture content (MC). Within this item, there could be a problem of MC variation within a wood species as well as differences between species. There may also be MC variation due to mixing purchased materials (dried by a vendor) with material dried in-house. In addition, MC could be explored further with regards to the other categories such as incoming inspection failing to check MC (an issue involving both operators and methods) and/or extended storage of the material in areas without temperature and humidity control (related to environment). The basic idea is to ensure that causes are explored in enough depth such that the fundamental or root cause(s) is identified.

Of course, at some point, the process will come to a natural conclusion. This can happen either when the team has exhausted all possibilities, or some consensus is reached that the root cause has been identified. The completed CE diagram might look like the one in Figure 2. Due to space limitations, many of the items listed here are quite cryptic. When working on a flipchart or whiteboard, a team would want to use more detail in describing potential causes. As discussed in Step 5 above, notice that some causes appear in multiple categories. For example, causes related to moisture appear in “materials,” “methods,” “environment,” and “operator.” This is to be expected, since the issues themselves are multidisciplinary. Moisture content of wood, for example, is a material property that is influenced by the environment, and proper control requires the right methods as implemented by the operator. Also notice the secondary branches. For example, under operator, “size checks” is listed, with potential causes including “frequency” (i.e., the operator checks the part size but not often enough) and “skipping” (i.e., the operator doesn’t do the checks at all.)

Figure 1. Blank cause-and-effect diagram

4 Cause-and-Effect Diagrams

Conclusion Now that the team has completed the diagram, how do they know which cause is the root cause? As stated above, the process is as important as the end product. It is not the diagram per se that tells the team what the root cause might be, but rather the discussion while constructing the diagram that will help lead the team to a cause or two worthy of further exploration. In this case, the fact that “moisture content” appeared in so many places on the diagram might lead us to speculate that the team spent a fair amount of time discussing this issue. That fact, combined with a basic knowledge of wood (i.e., wood shrinks and swells with changes in moisture content) might lead the team to decide to collect data and/or conduct an experiment to verify one or more of the items on the diagram. For example, the team might decide to gather baseline data—measure the moisture content within species and between species and construct a histogram. They could then conduct an experiment to examine the impact of changes in moisture-check methods on moisture content variability and verify the effect of these changes by constructing additional histograms. If the changes appear to work, they would then need to ensure that the changes become standard practice (and of course, are followed). If the changes do not seem to work, however, the team might then move to the next most likely cause. In that regard, it should be noted here that merely reaching consensus on the cause of a problem certainly doesn’t guarantee accuracy. In fact, the team’s decision on the root cause might be wrong. In some situations, more advanced statistical tools may be needed to identify causes and conduct and interpret the results of experiments. Design of experiments (DOE) is a set of statistical methods and tools for ensuring the efficient and effective conduct of experiments. Our next publication in this series will present a brief overview of DOE. Using DOE, however, requires more advanced statistics than are within the scope of this series. We will

Machine Methods Materials variation between machine maint. quality of or within species bad bearings setup proced. knives moisture content knife grinding damaged mixing purchased w/in- knives moisture checks house dull sanding Size out-of- frequency specification size checks skipping incoming moisture checks moisture content setup material storage skipping conditions Operator Environment

Figure 2. Completed cause-and-effect diagram

5 Statistical Process Control

merely introduce DOE to give you some familiarity with the topic and to help you decide if you want to pursue formal training in the subject.

For more information

Brassard, M. and D. Ritter. 1994. The Memory Jogger II: A Pocket Guide of Tools for Continuous Improvement & Effective Planning (Methuen, MA: Goal/QPC). http://www. goalqpc.com

Ishikawa, K. 1982. Guide to Quality Control (Tokyo, Japan: Asian Productivity Organization).

PERFORMANCEPERFORMANCE EXCELLENCEEXCELLENCE ININ THE THE WOOD WOOD PRODUCTSPRODUCTS INDUSTRYINDUSTRY ABOUT THIS SERIES

This publication is part of a series, Performance Excellence in the Wood Products Industry. The various publications address topics under the headings of wood PERFORMANCE EXCELLENCE technology, marketingIN THE and WOOD business PRODUCTS management, INDUSTR Y production management, quality and process control, and operations research. For a complete list of titles in print, visit the OSU Extension Service catalog at http://extension.oregonstate. edu/catalog

PERFORMANCE EXCELLENCE © 2009 Oregon State UniversityIN THE WOOD PRODUCTS INDUSTRY This publication was produced and distributed in furtherance of the Acts of Congress of May 8 and June 30, 1914. Extension work is a cooperative program of Oregon State University, the U.S. Department of Agriculture, and Oregon counties. Oregon State University Extension Service offers educational programs, activities, and materials—without discrimination based on race, color, religion, sex, sexual orientation, national origin, age, marital status, disability, or disabled veteran or Vietnam-era veteran status. Oregon State University Extension Service is an Equal Opportunity Employer. Published August 2009.PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY 6

PERFORMANCE EXCELLENCE IN THE WOOD PRODUCTS INDUSTRY