Do’s and don’ts of SPC
Comments Off November 30, 2010 at 9:58 am by Jill Nelson
Eight factors to consider when implementing SPC software in your manufacturing plant
Statistical Process Control (SPC) is a technical tool for process control and improvement. It uses graphical charts to depict process characteristics such as the process centre, the process spread or the defect rate.
The idea of charting process characteristics was pioneered by Walter A. Shewhart in 1924. Shewhart’s charts spawned a huge improvement in industrial quality control and were widely promoted by the American Society for Quality (ASQ) and by consultants such as Dr. D. Edwards Deming, who introduced them to Japan in 1948. Today, SPC is a critical part of practices used to control and improve quality, not only in manufacturing but in all sorts of transactional endeavors. Many organizations have embraced SPC because they have been told that it is the right thing to do. Some are not sure why they are doing it. For this reason, here’s a checklist for your own organization.
Develop a passion for reducing variation in all processes. Variation is the enemy of quality because it masks accuracy and destroys consistency. Measuring and reducing variability are the keys to success. This is why companies such as Motorola pioneered programs like Six Sigma. These are all designed to give new language to the task at hand, which is to reduce variation as near to zero as possible. Critical in successful implementation is to infuse a passion into the organization to reduce variation and defect rates in all processes. Management must understand and champion this.
Get Educated. SPC is a big subject. You will need to understand that there are different types of variation, piece to piece, time to time, and occurrence rate variation. For each of these types there are specific charts that must be used. Using the wrong chart, such as a classic average range (Xbar R) chart for time to time variation, will not work. It will be pointless and lead you to wrong conclusions. In process industries, most processes exhibit autocorrelation, which just means that data values are not statistically independent but are related in time sequence. A Shewhart Control Chart cannot be used in these situations. Time series models and residual charts must be used instead. Another potential problem is non-normal data where special techniques such as power transformations are necessary.
Provide training. Use training to instill into people a deep appreciation for the purpose of SPC. Training should be participatory and allow people to experience first hand how variation is reduced. In our courses, we use the Statapult (Statistical Catapult) as a hands-on tool to involve people in appreciating the power of statistical thinking and to provide a laboratory for learning. Then we transition into the specific types of charts that people will be using. Finally, we show how software can automate the process and ensure people learn how to interpret and apply what they have learned.
Put SPC in the hands of the process operator. If SPC is not in the hands of the person who can actually make adjustments to the process, then what you have is not SPC but SPM (Statistical Process Monitoring). That is great information but not real time and not control. It is also important to empower the operators to make changes to the process, taking advantage of the signals that the charts are giving. In one radiator plant, operators discovered that variation was greatly reduced after a maintenance cycle. By increasing the frequency of maintenance, they were able to reduce variation by 50 per cent. That enabled them to use a thinner wall tubing, saving thousands of dollars a month in material cost.
Don’t make SPC another legalistic requirement. People must understand why you are doing it, or it will just become another burden. Application without understanding is always a waste of resources.
Don’t just buy software and stick it on the process. Again, if people do not understand what the charts are saying, they will be bewildered and conclude that it is just another way of checking up on them. That can lead to all kinds of aberrant behavior.
Don’t use it as a policing tool. It only increases fear and anxiety often leading to increased process variation. A police force approach to quality will not help you in the long run. People will always find ways to escape detection or feed you misinformation.
Don’t confuse system problems with sporadic ones. The point of the charts is to help differentiate between these. There is nothing that the local operator can do to solve a system problem. That is management’s job. Berating the operator for a system problem is the same as beating them over the head for that which they have no control over.
SPC is a powerful tool for process control and for continuous improvement. To be effective, people must understand the purpose and benefits of controlling and reducing variation. Training and empowering people to use it properly will always lead to process improvements and reduced costs. Added benefits are the sense of achievement that comes from making a positive difference to the quality and productivity of the organization. QC
Les Galicinski is chief consultant and president of Qualitran Professional Services Inc., a Barrie, ON, consulting and training organization. He has authored a number of texts, including “The SPC Book.”