It has been emphasized numerous times the importance to improve the products’ quality directly at the production process rather than through the inspection of finalized products. With Process Capability Analysis and Control Charts, we attempt to design and improve the process to obtain high quality from the manufacturing process. Acceptance Sampling is used as a measure to detect defective products once they have been made. As a supplier, rejected parts in acceptance sampling means that we may have to rework the defective parts, probably inspect rejected lots, pay warranties, lose future sales, and expend money to regain reputation.
The Process Capability Analysis deals with the quality of conformance. It evaluates the variability and performance of a process in reference to the specifications. The Process Control aid to identify assignable causes of variation. If there are not assignable causes of variation, the process is in statistical control. A process can be in four states based on its capability and statistical control. Each state can be identified by answering the questions “Is the process capable?”, and “Is the process in control? In the first state, the process is capable and is in statistical control. In this state, control charts are useful to identify patterns and trends that could give early warnings to prevent the process from going out of control. In the second state, the process is in control, but no capable. In this case, the quality characteristic should be centered in its target value (usually the middle of the specification band). If the process is still not capable, it should be changed through the design of experiments or other methods. The specifications should also be evaluated. As last resort, we may consider developing or acquiring new technology for the process with less variability. In the third state, the process is out of control but is capable. Even when the number of defects is very small, the control chart can be used to detect assignable causes that may affect the capability in the near future. In the fourth state, the process is not capable, and out of control. The assignable causes will be reported very quickly by the control charts, and they should be eliminated. If the process is still not capable, similar actions to the second state should be taken. One of the main concerns doing the Process Capability Analysis is finding the frequency distribution of the quality characteristic of interest. This can be done with a Histogram, a Probability Plot, and studding the skewness and kurtosis of the variable. If the frequency distribution is normal, many conclusions about the process can be quickly made. For example, from the standard normal distribution, the process mean, and standard deviation we can estimate the process fraction nonconforming. Another important indicator of Process Capability is the Process Capability Ratio (PCR). This indicator gave us information about the relation between the specification limit interval and the spread of the process usually defined by 6σ. PCR greater than one means that the process uses less than 100% of the tolerance band. However, as PCR says nothing about the process centering (how close is the process mean to the center of the specification interval), PCRk and PCRkm were developed. PCRk = PCR when the process is centered, and approach to 0 as the process mean moves from the center of tolerance band to the tolerance limits. The Process Fallout in PPM is tabulated for PCR and PCRk if the process variable has a normal distribution. On the other hand, Control Charts provide information about the process capability. The ẋ and R Chart allow us to estimate the mean and standard deviation of the process, and consequently the PPM and the PCR indicators. When evaluating the process capability using PCR, if there is a minimum PCR requirement, the sampling variation in estimating σ should be taken into account. This may require a higher PCR and a specific sample size. The new level of PCR will be determined by a PCR High that we would like to accept with probability 1 – α and a PCR Low that we would like to reject with probability 1 – β. Control charts are a widely used technique of statistical process control used to detect the occurrence of assignable causes before many nonconforming units are manufactured. It can be used to control quality characteristics that can be measured on a numerical scale, or classified as conforming and non-conforming. If the quality characteristic is expressed on a numerical scale, we talk about variable control charts, in the other case we use attribute control charts.
When selecting the control chart it is necessary to determine which process characteristic to control, wherein the process should the chart be implemented, what type of control chart to use, the sample size and frequency. There are economic models that can help you to identify the sample size and frequency. These economic models take into account the cost of taking a sample, the cost of finding an assignable cause, the cost of investigating the false alarms, and the cost of operating in out of control state. Talk to a manufacturing engineer for more information on these topics.
