In general, practically all organizations have the need to ensure that their products or services are free of defects, to avoid costs of recovery, reprocessing, customer returns, etc., without considering damage to the image for the company's brand.. The reader will imagine that inspecting all bolts manufactured by a company that manufactures bolts at the rate of 100,000 units per day would have a prohibitive cost. The way to do this is through the CEP or Statistical Process Control, which ensures that the maximum number of products or services produced have been made without defects, and without this meaning controlling 100% of the product units manufactured.
In this way, statistical process control or CEP is not applied to control manufactured products, but to prevent the appearance of defects during production.
In reality, the sampling required by CEP is not expensive, but it is a preventive control so that defective products are not manufactured. Statistical Process Control is a lean tool that provides the scientific basis for defining optimal controls that replace the need to have to inspect all products, thus minimizing control costs while providing the guarantee that practically the 100% of the delivered products are free of defects.
And here it is necessary to make a scope. In the case of tangible products, if our bolt manufacturer does not apply tools such as statistical process controls or some kind of error-proof devices (Poka-Yoke) in its production, some bolt units are likely to be defective. However, you will always have a second chance to control 100% of the manufactured products, to detect and prevent the delivery of a defective product to the customer. At least it will have saved the image of the company. However, in the case of service companies, this second opportunity does not exist because the service is delivered as it is produced. There are no second options.If the service delivered is poor, customer dissatisfaction will be immediate, and damage to the company's brand image will also be. On the other hand, if the defect in the service turns out to be dangerous for the user, the prevention of the error is the only available option, and the statistical process control can help to avoid any defect during the production, and delivery, of a service. For example, in a radiotherapy treatment, if the patients were not placed well in the machine, it would irradiate and destroy healthy tissues instead of destroying tumor tissues.
The question is Quality or Productivity? And obviously the answer must be "both." In other posts we have discussed that any excess of quality or productivity will harm the results. Companies must be productive, but they must also offer their customers the highest quality products that allow them the resources they have. Seen from the perspective of statistical process control, and since it is not possible to observe all the bolt units manufactured to determine whether they are defective or not, the solution is usually to sample the observations and measure the number of errors that there are in each of the samples collected. And about it infer the results.
Variability occurs in any process. By variability we understand those unavoidable changes or errors that modify the process (whether small or almost imperceptible) and that subsequently affect the product that is produced or the service that is offered. Errors due to common (random) causes always occur in any process. Therefore, it is necessary to first quantify this variability associated with common causes to determine if the process is acceptable. If it were not, it will be necessary to introduce an improvement in the process that tends to reduce said variability. Once the variability of the process is in a range defined as acceptable, we can say that the process is "in statistical control."A process is under statistical control or in a "stable" condition when its variability is within the previously defined acceptance range.
In order to control the stability of the process, upper and lower specification control limits (LSE and LIE) are established (process average: X ± 3 σ). If the process lost its statistical control, it would be due to an excessive variation produced by an “assignable cause” (not random). Therefore, it will be necessary to identify this assignable cause and once it has been done, take corrective action that re-aligns the process, taking it towards its original variability.
Like any method, statistical process control is designed to search for statistical bases that prevent errors without increasing the costs derived from quality control. In fact, it is possible to generate large cost reductions when you properly understand how the process varies and what are the factors that cause the process to escape statistical control. If these causes are correctly understood, it is even possible to reduce the number of controls, and still increase the productivity of the process and, therefore, increase the company's profits.
