In every process, there exists a certain amount of variation. Variation in a process cannot be eliminated, but it can be measured, monitored, reduced and controlled.
By utilizing process controls, taking measurements and using reliable, well-maintained equipment, variation in a process can have less effect on the quality of the output. The process can be capable of producing acceptable product on a consistent basis. We can maintain Process Capability.
What is Process Capability?
Process Capability (Cp) is a statistical measurement of a process’s ability to produce parts within specified limits on a consistent basis. To determine how our process is operating, we can calculate Cp (Process Capability), Cpk (Process Capability Index), or Pp (Preliminary Process Capability) and Ppk (Preliminary Process Capability Index), depending on the state of the process and the method of determining the standard deviation or sigma value.
The Cp and Cpk calculations use sample deviation or deviation mean within rational subgroups. The Pp and Ppk calculations use standard deviation based on studied data (whole population). The Cp and Cpk indices are used to evaluate existing, established processes in statistical control. The Pp and Ppk indices are used to evaluate a new process or one that is not in statistical control.
Process capability indices Cp and Cpk evaluate the output of a process in comparison to the specification limits determined by the target value and the tolerance range. Cp tells you if your process is capable of making parts within specifications and Cpk tells you if your process is centered between the specification limits. When engineers are designing parts, they must consider the capability of the machine or process selected to produce the part.
The Cp and Cpk indices are only as good as the data used. Accurate process capability studies are dependent upon three basic assumptions regarding the data:
There are no special causes of variation in the process and it is in a state of statistical control. Any special causes must be discovered and resolved.
The data fits a Normal distribution, exhibiting a bell shaped curve and can be calculated to plus or minus three sigma. There are cases when the data does not fit a normal distribution.
The sample data is representative of the population. The data should be randomly collected from a large production run. Many companies require at least 25 to preferably 50 sample measurements be collected.
Why Measure Process Capability?
In manufacturing and many other types of businesses, reduction of waste and providing a quality product are imperative if they are to survive and thrive in today’s marketplace. Waste exists in many forms in a process.
Monitoring process capability allows the manufacturing process performance to be evaluated and adjusted as needed to assure products meet the design or customer’s requirements.
Steps of Process Capability Analysis
We must know what type of data we are dealing with : Discrete and Continuous; likewise measuring process capability depends upon the data types.
Check the data type whether it is Continuous or Discrete.
If it is discrete data then apply the Capability Analysis (Binomial).
If it is continuous data then, check the process stability.
Process Stability can be checked by the I-MR control chart.
If the process is not stable, then we cannot calculate the process capability, we need to fix or adjust the data as stable.
If it is stable, then we check the process normality.
If the data is normal and stable, we can calculate the Capability for normal data.
If the data is not normal, first need to make the data normal, if the data cannot be normalized then use the Box-Cox transformation.
Once data is fixed to normal, calculate the process capability.
Partial Credit: Quality One & Grey Campus