Written by: Scott A. Mackay
Why is the Manufacturing Engineering role in improving quality important?
Edward Deming tells us that the benefits of improved quality are: higher productivity, lower cost, better competitive position, happier people on the job, and more jobs through more sales. Quality must be the responsibility of everyone. However, as point five of Edward Deming's fourteen points states: "It is managements job to work continually on the system (product design, processes, tools, training, retraining) and provide their employees with the tools and skills needed to perform a quality job.
Management’s solution to process improvement and control in the manufacturing industry is generally implemented through the hiring of Manufacturing Engineers. The primary role of the Manufacturing Engineer is the resolution of process problems. He or she can make a large cost impact in most companies. A persistent Manufacturing Engineer can typically identify and facilitate resolution to problem with a savings of up to ten times their salary each year.
How good is good?
In his book "World Class Quality", Keki Bhote states that a process capability index (Cpk) of greater than 5.0 and a facility overall efficiency (FOE) of greater than 85 percent is where a company needs to be. At this performance level there is no significant waste or rework and very little defects to be recorded or re-inspections and re-tests to be performed. This is a very aggressive benchmark to achieve for most companies.
Bhote also talks about quality attitudes of companies as being on four levels of evolution. The first level he calls innocence, where quality control is a necessary evil and problems that are detected and largely corrected through rework. The second level is awakening, quality is discovered to be a cost and problem correction is pursued. The third level is commitment/implementation, where quality becomes an economic imperative and resources are committed to prevention of errors and process capabilities with Cpk of 1.3 of higher are strived for. The fourth level is world class, where quality is a super ordinate value and prevention is a way of life. Process capability Cpks are now 5.0 or better.
A company with world class quality will have little to no defects being created and consequently little to no rework, retesting, or schedule slippage since errors are improbable. The Manufacturing Engineer is empowered to implement and maintain the quality processes needed to be a world class company.
Where should you start?
The first step to improving quality and reducing cost is to identify which errors are costing the most money. Cost data must be collected for waste, rework, retest, and other areas of cost related quality deficiencies. Isikawa states that "the purpose of collecting data is that data will form the basis for action and decisions". A Pareto Analysis of the cost data, the ranking of cost problems by significance, will identify which problem area you should work on first.
Unfortunately, most companies do not collect cost data to a level where it can reveal how much is spent on doing a job the first time versus correcting the quality of the work that was performed. This cost accounting challenge is worthy of tackling, however it should also be noted that if great improvements in quality are being made, it should surely show on the company's financial bottom line, with or without the detail cost data. In other words, yes you will need a good cost accounting system as part of your company infrastructure, but don't wait for it to be installed, you have other work to be doing.
Note that the collection of cost data and the Pareto analysis will need to be performed over and over again as cost problems are solved. New problems will rise to significance and will need to be worked next. Continuous improvement progress should be reflected in the elimination of the top cost contributors and their replacement by the next most significant cost drivers.
Problem targeted, what next?
With the highest cost problems identified, the next step is to determine the causes for the problems. A brainstorming session with the people who perform the work under study should reveal most of the possible causes. The causes can be diagramed onto a "fishbone" or Ishikawa chart (named after its originator) for ease of study and communication with others. The individual causes should them be studied in more detail and data collected to further define the causes of variability that leads to defects being made.
How do you expose the cause?
Data will be required for making decisions about the process under study. This data will provide the performance assessment of the current process, help justify the solutions, and confirm the results of corrective action. Statistical process control (SPC) data must be collected and statistical analysis techniques must be applied to baseline, or characterize, the current process and continued until the corrective action is implemented such that no more defects are probable.
SPC charts should be established to measure the performance of all parameters that may exceed process goals. The SPC chart can also serve to detect trends and signal the need for corrective action adjustments before a significant amount of defects can occur. This alone can greatly reduce rework and retest costs.
Data is your friend.
Your statistical data is also valuable for communication with other people of different expertise’s. Data can "bridge" the communication gap that typically exists between the people across the many departments of a company. Data can also give you power; data with a clear trail to its attendant costs will usually get peoples attention toward the need to resolve a problem.
Data will also present problems without bias. Data is what it is. Instead of stating subjectively that there is a problem with the process, the data will present a numerically quantifiable story about the problem. But, remember that this is data about variability showing where defects are occurring, it does not tell you why the defects are occurring. Don't jump to conclusion that someone must be blamed for causing the variability. The focus must remain on finding solutions to the causes for the variability, which is more often in the process or design rather that the people.
What can you do to eliminate variability?
Variability can be either controlled to an acceptable level by improving the process, or the tools and materials can be changed such that the variability no longer has a negative effect. Either way, or both, may be required to solve a variability problem.
People generally like to solve problems, and armed with your variability data you should call together the key people who have influence over the process (the Stakeholders) to present your data and discuss the options. A brainstorming session will usually identify several good solutions for farther study and later narrowing down the most cost effective approach.
Juran's Quality Control Handbook gives facts that "40% of product failures are usually due to design deficiencies and the 80% of the changes made to a products design are typically related to correcting defects". These facts relate only to product quality costs. There is also variability caused by equipment, tools, materials, and human error. The product design process should be looked at first because corrective measures in these areas are far more effective than leaving variability to its natural course.
Why not just add more inspection and testing?
Adding more inspection and testing is never a good solution, it only adds to the cost of producing the product. You must strive to prevent defects in the first place. Phillip Crosby says "Checking, sorting, and evaluating only sorts what is already done. What has to happen is prevention. The defect that does not exist, cannot be missed" The secret of prevention is to look at the process, identify opportunities for defects, and to implement measures that will not let defects occur. Your approach must produce outputs that can not fail rather than can pass most of the time.
SPC is not a solution, it is a tool.
With the process managed using SPC and generation of defects eliminated, the problem would appear to be solved, Right? But, there is now the cost of continuing the SPC to be reckoned with. This is where many people will stop and not proceed on to complete the problem solving effort. The job is not completed until the paperwork is gone! The cost to maintain the SPC should be collected and used to justify changes such as automated process control improvements that will monitor and correct the process without the need to collect and analyze data. Also, the product design may need to be changed to eliminate the probability of future defects.
Be sure that you are working the cause of variability, and not masking another problem. Stopping a "leak" is not the same as shutting off the water. Elimination of the cause for errors should not result in continued sorting and reworking at an earlier step in the process. You may need to go back many steps in the process, outside your company if needed, to eliminate the causes with the most cost effectiveness. Pushing tighter specifications back onto a supplier's process that is not capable will result in increased cost for the suppliers part or services. Your goal is a process the cannot and does not produce defective products. The same techniques you use in your factory might need to be applied at your suppliers too.
Sometimes experimentation may be needed.
The pursuit of a product design that can not fail can be inhibited by lack of known performance data. The best and fastest computers can't help you establish your performance limits without good performance data. This data is not always available from the engineering text books and must be obtained through experimentation.
G. Taguchi has published procedures on designing of experiments that can be used for characterizing new designs to determine their performance limits. This same technique can also be used to determine the performance limits of a process. Taguchi's technique involves varying each product (or process) parameter and observing the results to identify which parameters have an effect and how much variation can be tolerated without a failure.
What about manufacturing defects?
With your product and processes designed to tolerances that can not produce failures, nothing else can cause a reject, right? The human element and random chance are still there. There will always be a screw or connection that didn't get secured, the component or connector that only fails when hot or cold, or the circuit that won't operate in humid atmosphere. What can you do with these common types of defects? A manufacturing screening process will identify these elusive defects far better than any inspection or testing process.
Manufacturing screening was first developed for use on high reliability aerospace products and because of its effectiveness, it is now finding its place throughout the manufacturing industry. A manufacturing screening process starts with screening of incoming materials, as traditionally done by most companies, but then adds screening of the products being produced.
A basic manufacturing screening process for manufactured products includes thermal shock, random vibration, and thermal stress (also known as burn in). The thermal shocking of all rigid soldered connections is intended to stimulate thermal expansion related defects. Mechanical assemblies and assemblies with electrical interconnections will be subjected to random vibration to stimulate defects such as loose hardware and poor or partially seated connections. Mechanical assemblies with moving elements and all electronic assemblies will be subjected to thermal stressing to stimulate defects such as friction problems with moving elements during temperature extremes or early failure of electronic components. Thermal stress will usually also include humidity stressing to stimulate defects sensitive to humidity such as voids in potting, coatings, or sealants.
Design defects should rarely be detected through a manufacturing screening process. However, it should be noted that many times an existing design must be changed to make it robust enough to survive the manufacturing screening process. The process cannot be effective unless the levels of stress can be set strong enough to stimulate and reveal the manufacturing defects being sought.
What if you just can't meet the customers needs?
The customer is always right and if someone else is meeting the customers needs then it is prudent for you to do so also. But, there are times when a change to customer specifications can eliminate a recurring defect problem and lower the cost of the product. This is when you must get your data organized and show the customer how everything that could possibly be done to your process has been implemented ad short of a breakthrough in process technology, the only solution to eliminate the recurring defects is for a change to the customer's specifications.
Getting changes at the customer level will usually be easier if a cost reduction in the product can be offered. Obviously you will be saving some cost by eliminating the defects, and unless you can prove you are loosing money producing the product, this savings may be what you might offer in exchange for the loosening of their specifications. In the bigger picture, reducing your customers cost should help to boost their competitiveness and in turn create the need for more product to support higher sales.
Conclusion
The Manufacturing Engineer sits in a position where he or she can become a "hero" of quality and productivity improvement. There are many tools and techniques that can be employed in the pursuit of quality problem resolution. To make the biggest impact, they must work on the biggest problems first.
Cost and performance data must drive the decision process. Changes will be required in the product design, the manufacturing process, and the management process. Changes might also be required at the supplier and customer levels.
Empowered by management, the Manufacturing Engineer's primary role is to use their skills to a maximum for systematically seeking out and negating the endless causes of defects by continuously improving upon the product and processes, measuring progress as they go.
Bibliography
World Class Quality
Keki R Bhote, 1991
American Management Association, New York
Guide to Quality
K. Ishikawa, 1985
Juse Press Ltd. Tokyo, Japan
Quality, Productivity, and Competitive
Position
W. Edwards Deming, 1982
Massachusetts Institute of Technology
Center for Advanced Engineering Study, Cambridge, MA
Juran's Quality Control Handbook
J. M. Juran, 1988
McGraw-Hill, New York
Quality Without Tears
Phillip B. Crosby, 1984
McGraw-Hill, New York
Introduction to Quality Engineering
G Taguchi, 1986
Kraus International Publications, White Plains, NY
Reprint by permission only
Questions or comments, send to Scott.A.Mackay@Mindspring.Com