Sustaining Gains: Other Control Techniques
Statistical Process Control (SPC) has been one of the most favorite topics of discussion among quality professionals. A very large number of articles and books have been published on this subject. There is, however, one technique that finds very little or no reference at all in the material published so far on SPC and that is pre-control (PC). Perhaps this is so because it is too simple to understand and equally simple to implement. Pre-control is a technique that is easily understood by shop operators and it enables them to control the process so that defective parts are not produced. Although simple to understand, pre-control is a statistically robust technique. Unlike SPC, where we need 25 subgroups before we can draw control limits and conclusions, pre-control starts giving feedback about the process from the very beginning, making it highly responsive to the process signals. Pre-control was developed by Frank Satterwaite during the 1950s (Bhote 1988).
In pre-control, the tolerance is divided into three zones as shown in Figure 18.1. These zones are green, yellow, and red. The middle half is the green PC zone. LTL means lower tolerance limit and UTL is the upper tolerance limit. UPCL stands for upper pre-control limit and LPCL for lower pre-control limit.
If CPk of the process is 1.0, it means that the tolerance equals Six Sigma and the mean of the process coincides with the tolerance mean. Sigma is the standard deviation. In such cases, and assuming normal distribution, we can expect that 86 percent of the readings will be in the green (PC) zone and 7 percent in each of the yellow zones. Thus we can expect that one out of the 14 readings will be in the yellow zone. Hence, the chances of getting two consecutive readings in the yellow zone will be (1/14) × (1/14) or 1/196. This is the foundation of pre-control.
To qualify the set-up for pre-control, the following steps need to be observed:
- If five consecutive pieces are in the green zone, the set-up is ok to run.
- If there is one yellow, restart counting.
- If there are two consecutive yellows, adjust the process.
- If one reading is red, adjust the process.
- If five consecutive greens cannot be produced, investigate and improve the process capability.
The set-up cannot be qualified unless five pieces in a row are in the green zone. If we cannot qualify the set-up, then there is a clear signal that the process is not capable of producing parts within specifications. In such a case, efforts must be made to reduce process variation so that the capability index improves. This is the power of pre-control. It just does not allow an incapable process to run. If the operator makes an attempt to continue, he/she has to check all parts as the set-up does not get qualified.
Figure 18.1 Pre-control zones
Running: Check two consecutive pieces.
- If both are green, continue the process.
- If one is green and the other is yellow, still continue the process.
- If both are yellow, adjust the process.
- If any of the pieces is red, adjust the process. In such a case, parts produced from the last sampling must be inspected.
On average, six sample pairs between consecutive adjustments are recommended (Breyfogle III 1999). See Table 18.1.
Table 18.1 Suggested sampling intervals
|Average time between process adjustment||Sampling interval for pairs|
Every 80 minutes
Every 40 minutes
Every 20 minutes
Every 10 minutes
Impact of CPk
The graph in Figure 18.2 shows the chance of getting five greens in a row for various CPk values. The probability drops sharply below CPk of 1.5. For CPk of 1.5, this is 0.88. This drops to 0.48 for CPk of 1.0. Thus, for lower values of process capability, it becomes increasingly difficult to qualify the process with the rule of five greens in a row forcing corrective action in order to reduce variation. The number of pieces required to qualify the process is in a way an indicator of its capability. It is, therefore, essential that process capability of 1.5 or greater is achieved before implementing pre-control.
Figure 18.2 CPk vs probability of 5 greens in a row
Sometimes, colour marking can be provided on the gauging to make pre-control more operator-friendly. Pre-control may be used with caution for unilateral tolerances as shown in Figure 18.3 (Juran 1999). The probability calculation may be required for each application before implementing pre-control.
Figure 18.3 Pre-control for unilateral tolerances
Concluding Remarks on Pre-control
Although pre-control is very simple to use, it is not a complete substitute for control charts. The purpose of control charts is to monitor a process for the presence of assignable causes, if any. The process log is expected to be maintained with control charts, making it a useful tool to understand variation with time and relate it to various events. Pre-control, on the other hand, is a simple tool that helps to prevent production of defective parts. It does not necessarily require any charting by the operator, although charting can be used in pre-control.
Measurement System Re-analysis
Reducing variation and improving process capability is a major objective in most Six Sigma projects. In MSA, we compare the measurement variation (σm) with a total variation (σT) which includes measurement and process variation (σP). As the process variation is often reduced in Six Sigma projects, the ratio of measurement and process variation is affected. Thus, it is essential to reassess the measurement system adequacy for the reduced variation. The method of measurement systems analysis and the acceptance norms do not change.
Control Plans and Audits
A control plan is like a ‘wedge’ that can prevent the ‘wheel’ of improvement from rolling backwards (see Figure 18.4). It is a critical requirement for ensuring that the benefits of Six Sigma projects do not diminish over time due to lack of adequate control mechanism.
Figure 18.4 Control plan as a ‘wedge’
Transfer of Process Ownership
This is the time when the project team ‘hands over’ the improved and changed process to the process owner who is responsible for operating it regularly. The process owner ‘sign off’ is essential to establish an agreement on the improvements achieved and the robustness of change with reduced supervision.
Control plans can be of various forms, the most common being the documentation of the changes. These could include drawing, process documentation, purchase order data, calibration frequency, tools and fixtures designs, software source codes, control charts, sampling plans, acceptance criteria, etc., and physical changes such as fixtures, tools, technology, equipment, accessories, etc. Most Six Sigma professionals use a single or a set of documents to assure the controls required to maintain the improved process capability. A sample control plan format for a process is illustrated in Table 18.2.
Table 18.2 Control plan format
Some examples of control plans are listed here:
- Work instructions for a changed operation sequence
- Changes in process drawing
- Changes in process sheet
- Changes in engineering drawing and/or specifications
- Deletion of discontinued supplier
- Avoiding duplicate databases
- Periodic qualification of welders, crack testers, lab technicians, call center operators, inspectors, etc.
- Process audit of a precision grinding, pathology lab sampling process, hospital administration, call handling, software testing
- Supplier surveillance audits
- Regular review of control charts for stability.
In the language of quality system auditors “what is not documented never happens!” Thus, if we want the improve process to “happen” routinely, we must document the changes in the standard operating procedures (SOP) of the company. It is essential that the procedure or work instructions be integrated with the regular system documentation rather than separate ‘Six Sigma work instruction’. If the changed procedure is not included in the documented system of the company, it is unlikely to get audited in the quality system audits as a part of the ISO 9000 quality processes.
Training Plan Deployment
Training the process users for the improved process is an integral aspect of installing the control plan. Remember the time you bought a new mobile phone or a CD player? While you might be excited with the new technology, how much time did it take you to get used to it? How did you learn about its usage? We all can, at some point, get frustrated with the time and the efforts it takes to learn something new. The same is true for most Six Sigma projects. These projects achieve improvement by changing something. The change can be speed, feed, tool, fixture, software, procedure, tightening sequence, method, acceptance criteria, gauge, use of a new material or a part in assembly, schedule or order quantity of purchase order, working with a different supplier, etc. Any of these changes requires adequate training and qualification. In critical skills, a certification may be desirable.
Most Six Sigma projects do not achieve the Six Sigma level of performance. A typical improvement is of the order of 50 to 70 percent. However, the customers would expect a Six Sigma level of performance from a company once it announces its Six Sigma initiative. This implies that the sigma level for CTQs needs to be regularly assessed. Also, when customers are satisfied with the performance in some of the CTQs, the next important CTQ needs to be taken up for improvement. Given this dynamic environment, benchmarks keep on changing. In the theory of constraints, when we resolve a constraint, a new constraint surfaces! We must address this new constraint as a potential Six Sigma project.
Audits serve to ensure that improvement actions are effective. Audits can be for a system, process or product. Depending on the actions taken and documented, actions can be verified periodically for compliance. Quality system and process audits are usually a regular part of ISO 9000. However, Six Sigma or other improvements call for focused audits for specific actions taken in projects.
After the completion of each project, the belt and the team should document lessons learned so that others do not need to reinvent the wheel. Many companies implementing Six Sigma maintain a database of projects and related records. The lessons learned are useful for other similar process owners and plants to reduce improvement cycle times. We can call these ‘copying improvements’ in similar areas or functions.
Mistake-proofing or Poka-yoke
“Poka” means inadvertent errors and “yokeru” means to avoid. Poka-yoke means avoiding inadvertent errors or mistake-proofing. Shigeo Shingo, who developed the poka-yoke philosophy, calls it “source inspection”. He mentions that source inspection checks the factors that cause errors and not the resulting defects. These methods were formerly known as “fool proofing”. Shingo recognized that this could offend workers and, therefore, coined the word “poka-yoke”.
Many things can go wrong in the complex environment of a workplace. Some simple examples are:
- sending a letter to wrong address
- filing a document upside down
- forgetting to tighten the bolts fully
- inserting a compact disc upside down
People often ignore or do not follow instructions. Control plans need to address this fact and provide better controls than mere instructions. Physical and/or system controls are better in addition to instructions. Small innovations can make a difference. Some examples from daily life are
|Only instruction||More than instruction (Poka-yoke)|
|No smoking board||Smoke detector|
|Speed limit displayed||Speed breaker|
|Counting cash manually||Cash counters|
|Use seat belt||Red light flashes or car does not start if seat belt not used|
|Fill in all information||Validate, drop down box to ensure correct spelling and consistent format|
There can be different types of poka-yoke devices.
- Source inspection to detect errors at source as these cause defects. For example, an additional locator or pin that prevents a part from mounting upside down. Another known simple example: In the 3.1/2” floppy disc, there is a chamfer provided in one of the four corners to prevent the disc from being installed incorrectly.
- Hundred percent inspection for defects using inexpensive sensing devices such as a limit switch
- Immediate action to stop operations when error is detected. An example of this is an interlocking mechanism or circuit that stops the machine when temperature goes beyond a certain point.
The first technique of preventing defects is the most effective. Contrary to what many believe, poka-yoke does not require an automated factory to derive its full advantage. Poka-yoke requires a strong conviction that errors can be avoided or prevented. It also requires good communication between the production, process engineering, maintenance, quality and design teams. Moreover, it requires an open discussion on ideas from those who are close to the operation where defects are produced.
Many kinds of errors arise due to forgetfulness, misunderstanding, incorrect identification, untrained persons or amateurs.
Some of the commonly used poka-yoke ideas are
- different sizes of holes or pins to prevent assembly or locating incorrectly (Figure 18.5)
- error detection and alarms
- limit switches to detect correct or incorrect placement
Figure 18.5 Simple poka-yoke example
In today's world, computer usage has encompassed all spheres of life. Many poka-yoke ideas are used by software designers as well. Examples are: passwords for security, accepting only valid credit card number or other data, drop-down menus to ensure consistent spellings of countries, counter check of subtotals, etc.
Some poka-yoke hints (Shimbum 1987):
- Identify items by their characteristics such as weight, size or shape. For example, an assembly with a missing part will weigh less or one with a wrong part will weigh more or less. The idea can also be used in foundries or chemical companies to ensure correct chemical combination.
- Design the process in such a way that subsequent operations or steps cannot be performed if there is an error in the previous step(s). For example, a part which is not reamed (or finished) will not get located in the next step. A software example is: a payment will not be processed if credit card number is not valid or if any of the fields do not match.
- Use a counter. The simplest example is a cash counter. The method can also be used in assembly were an exact number of parts should be assembled. Counters can also be used to detect critical conditions such as pressure, temperature, etc. An effective application example is a pressure counter with a device that sets off an alarm when pressure falls below the specified limit, thus preventing insufficient torque to fasteners in assembly.
- Use a checklist. This is often used during surgical operations. At the end of an operation, the staff runs through the checklist of materials used and the steps followed.
Here are the eight principles of basic improvement using Poka-yoke (Shimbum 1987):
- Build quality into the process. Make it impossible for it to turn out defective pieces even if an error is made.
- We must assume that mistakes are not inevitable. If there is a will, there is a way.
- Stop doing things wrong and start doing them right—now! There should not be any “but's”.
- Don't think about excuses; think how to do it right.
- A 60 percent chance of success is good enough. Implement the idea now. It is rare that we can do things perfectly without trying. Refine the solution later.
- Everyone must work towards mistake-proofing to reduce defects.
- Work as a team. No one is as intelligent as the team collectively.
- If a defect occurs, do not request for more inspectors. Instead, find out the root cause and eliminate it.
Apart from SPC, there are many control techniques which can be useful in sustaining improvements. Some of these are as follows:
- Pre-control is a simple and, therefore, effective shop-floor technique to qualify set-ups and maintain control.
- Measurement system re-analysis is required to verify the suitability of the measurement system after variation is reduced.
- Transfer of process ownership is necessary to ensure smooth transition and continuing deployment of changes implemented in the project. Training plan deployment is an integral part of transfer of ownership.
- Poka-yoke or mistake-proofing is of great help to prevent going back to the old system.
- Audits and ongoing controls are necessary to raise alarm in case of non-compliance with the controls established through Six Sigma projects.
Checkpoints for Completion of Control Phase
- Control plan is put in place for sustaining improvements (short and long-term).
- New process steps, standards, and documentation are integrated into quality system documentation.
- Operating procedures are consistent. Knowledge gained on process is shared and institutionalized.
- Response plans are established, understood, and deployed.
- Ownership and knowledge is transferred to process owner and process team tasked with the responsibilities.
Project Closure Reviews
When the control phase is complete, a DMAIC project is completed. At this stage, it is necessary to review and evaluate the project for achievement of deliverables as per the project charter. In many companies, there is a practice of conducting a formal project closure reviews. This is done by the Six Sigma Steering Committee or its equivalent. Project closure reviews help to check
- whether the improvement achieved meets the target
- whether appropriate tools have been used to efficiently achieve the results
- whether the evaluation of financial benefits are realized as desired
- the difficulties that the team faced and to see if the organization is supporting the Six Sigma initiative
- to check the applicability of the improvement in other departments and/or plants
- to assess the lessons learnt from the project
- to Recognize the team for its effort and planning for celebration of the success
- to sharing the success story to motivate the team and others
- to redirect if the project is not closed in the review. A template for project reviews and tools matrix is provided in the CD. Filename is ‘Six Sigma Project Review Checklist.xls.’
Automotive Industry Action Group (AIAG) (2002). ‘Statistical Process Control Manual’.
Bhote, Keki R. (1988). World Class Quality: Design of Experiments Made Easier, More Cost-Effective Than SPC. New York, NY: American Management Association.
Breyfogle III, Forrest W. (1999). Implementing Six Sigma: Smarter Solutions Using Statistical Methods. New York, NY: John Wiley & Sons.
Juran, J. M. and Frank M. Gryna (1999). Juran's Quality Control Handbook. New York, NY: McGraw Hill.
Shimbun, Nikkan Kogyo (1987) (Ed.). Poka Yoke: Improving Product Quality by Preventing Defects. Norwalk: Productivity Press.
Urdhwareshe, Hemant (2001). The Power of Pre-Control. Symphony Quality and Productivity Journal. http://www.symphonytech.com/articles/precontrol.htm. Accessed on 22 December 2009.