The term Statistical Process Control is a set of statistical tools that help control process parameters so that the outcome of the process is predictable and the process produces a minimum of erroneous, i.e. non-conforming outputs.
This comprehensive definition reflects the efforts of each manufacturer, regardless of product, to satisfy the customer as much as possible. To achieve this goal, techniques are used that monitor the output of the process and classify products of good quality, i.e. conforming, from products of lower to unacceptable quality, i.e. non-conforming products. Conforming products travel to the customer, while non-conforming products are reworked, at worst eliminated. The disadvantage of this approach is its reactivity. Quality control, as these techniques are collectively called, cannot reduce the occurrence of non-conforming products at the output. It can only capture them with some efficiency.
From the manufacturer's point of view, any nonconforming product is a loss. To create it, it consumed energy, human labor, materials and devalued the equipment used in its production. If it can process a nonconforming product, it will save at least part of the cost at the cost of another investment. In the case of an unrepairable product, however, he loses everything he put into production. Few people can afford such a waste today.
After the Second World War, Japan was in a similar situation of shortages of labor, raw materials, finances and time. A broken economy had to deal with every resource in order to get the most out of it. In this atmosphere, techniques developed in the 1920s and 1930s in the United States are beginning to be widely introduced into manufacturing plants. Their successful deployment in practice helped to verify their effectiveness and later became a widespread and common standard of a modern manufacturing company. These procedures are called collective statistical process control, or SPC for short. This name could be translated as Statistical Process Control. However, you may encounter the notion of Statistical Process Control, which is not entirely correct, as the notion of "control" is rather passive, which is contrary to the SPC approach.
The SPC approach is the opposite of the reactive approach. It does not try to control the quality of the output and thus sort the outputs into identical and non-compliant, but tries to set up the process so that the probability of non-compliant output is reduced to a minimum. It therefore interferes with the process itself, in a targeted and conscious manner so that there are as few failures as possible at the end of the process. This proactive approach can prevent waste by eliminating the mere occurrence of nonconforming products. As a result, the whole process is more efficient and can predict its behavior in the future. The SPC does not only focus on the process output but tries to control the inputs and parameters of the process.
The SPC is based on the generally applicable law on uncertainty. Each parameter set to a specific value naturally oscillates around that value according to the laws of statistical probability distribution. In other words, even if we set the parameter to a constant, point value, its real value in time will not be point, but interval. The values will oscillate around the average value. Interval width and average value are quantities that can be used to statistically estimate the occurrence of a particular parameter value. Using the mean and standard deviation, it is possible to construct a curve that can predict the probability that a particular value will occur in the future. The curve is called the Gaussian function, or the normal distribution curve.
It should be noted that the vast majority of events in nature take place according to the normal distribution, but not all. However, SPC can be used for events that do not follow a normal distribution. However, these are more complex and special techniques. At the outset, however, it is sufficient to clarify the principle of operation of the SPC on events that follow a normal probability distribution.
If the process parameter reaches values that are consistent with the normal distribution, we say that the parameter is under statistical control. The effects that cause it are just random. The parameter behaves as given by the applicable law of uncertainty. If a parameter is affected by a non-random external factor, the parameter values are no longer controlled by the normal distribution. The SPC can detect such effects and point out that something is wrong. Appropriate intervention in the process then eliminates the external, non-random impact and the process comes under control again, we can control it.
By applying statistical methods in practice, we can assess the state of the process based on its parameters and eliminate appropriate fluctuations in the process. A stable process provides outputs that can be estimated with some probability, and therefore we can predict the share of nonconforming products in total production in the future.
The basic tools of SPC include graphs that monitor the behavior of process parameters over time and can indicate the occurrence of a non-random impact. These graphs are called control charts. They were developed by Walter A. Shewhard in the 1920s. You can find more details on the Internet, but the important thing is that we can use control diagrams to effectively manage the process so that it is always under our control.
Control charts are not just passively plotted values. It is a set of procedures that, based on the nature of the non-random impact, define how to intervene in the process. It is understandable that interventions must be based on knowledge of the process, internal dependencies. Otherwise, they are just another, more or less random intervention on a trial-and-error basis. However, this is contrary to the SPC's philosophy, as such interventions do not stabilize the process but, on the contrary, disrupt it even more.
Another group of tools are the so-called competency indices and process performance indices. These describe the likelihood of a nonconforming product occurring. In fact, they talk about the likelihood that another product will not meet customer expectations. The customer usually transforms their ideas about the product into a specification, which is actually a set of product parameters and their values, usually expressed in the form of an interval. If the process is affected only by random influences, which we can ensure with the help of control diagrams, the output of the process will also have the character of a quantity behaving according to the normal distribution. By comparing the values of the parameters defined in the specification to a certain extent with the values of the standard deviation and the average of the process outputs, we can say with what probability we will not meet the customer's requirements. If these values are unacceptable, so we have a high error rate, it is necessary to return to the process and find the influences that are responsible for the situation. We return to the beginning, to the point where we tried to separate random influences from non-random ones. We must now identify some of the effects that we considered to be incidental, by deepening our knowledge of the process as non-incidental and eliminating their impact through appropriate corrective action. The result will be an output with less dispersion, and therefore less likely to occur.
Implementing the SPC in practice seems to be relatively simple. However, this is not the case. First of all, SPC requires knowing the dependence of output quality on process parameters. We therefore need to know what is related and what is not related to the quality of the output and to what extent a specific process parameter is responsible for the quality of the output. We also need to know what values it is necessary to set the process parameters so that the resulting output corresponds as much as possible to the customer's ideas. We need to know to what extent the measured value of the process parameter is affected by the measurement itself. We also need to identify effective remedies for cases where non-accidental effects occur in the process. We need to know how to get the process back to normal, so we need to know how to control it. All these requirements are provided within the organization by a trained team who masters the basics of statistics, but also knows the processes he wants to manage. They are people with practical experience who are not afraid of statistical methods and have the potential to interpret the results of the SPC.
The organization itself must undergo some change, because the SPC requires an honest approach from measuring the parameters, through their interpretation to the implementation of corrective measures. Any dishonesty or laxity will affect the quality of the SPC outputs. The worsening of the current situation can also be an extreme case.
SPC is also not a method that by its very nature works. The whole process must be constantly monitored and evaluated in order to eliminate external influences as much as possible so that the process output is as close as possible to the required value and with minimal deviations. Therefore, if you decide to implement an SPC, you need to be aware of the necessary investment in the resources you spend to create and maintain the SPC.
Undoubtedly, however, these investments are profitable in the long run. Whether it's a direct profit from the saved costs of processing rejects, the elimination of losses caused by scrapping rejects, or greater reliability of the manufacturer in the eyes of the customer. However, a side effect of SPC is also the professional growth of employees, who learn the details of the process, reveal its shortcomings and the potential for further improvement. The SPC also brings a sense of professional and professional growth to all who use its tools on a daily basis.
Training:
AIAG members and Odette - together with established industry partners and stakeholders have decided to release a new version of MMOG/LE, in order to support the following objectives for a more stable and predictable supply chain.
More
ISO / IEC 17025 is a standard that applies to laboratories in various industries and ensures that standards for laboratory testing and calibration are followed in practice.
More
Recently, we have begun to accumulate events that we are not used to in our latitudes. Apart from the pandemic, we have certainly all caught a tornado in Moravia and other extreme weather events. ISO 22301 specifies business continuity requirements and rules and helps companies recover quickly from unforeseen events. Its aim is to prepare companies and protect them in the event of such an exceptional unforeseen event.
More
What is the difference between PSB and PSCR? Why is PSB no longer enough and what has changed? What training should you undergo?
More
