Review of A Practical Guide for Manufacturing Process Improvement, Part 2
In my last post, I continued this series on a practical way to achieve manufacturing process improvement. I showed you how to conduct a customer dissatisfaction analysis and then explained that improvement can’t really be successful unless it is perceived by the customer in the form of an improved product. I explained that it is imperative that your improvement team visit your customer location so that a complete list of customer dissatisfactions can be compiled. I completed the post by presenting the basics of control charting and explained that success is not based upon the number of control charts, but rather the effectiveness of the charts being used.
In Part 3, I will discuss another key to success, which is how measurement systems should be established and implemented. I will then begin discussing process potential and process capability studies as a way of continuing the improvement initiative. Due to limited space, I will not provide details of key calculations, but I will provide references to other sources of information on these subjects.
Measurement system evaluation
One of the keys to successful improvement initiatives is to assure that all measurements are both accurate and repeatable. A measurement system is the combination of equipment used to make measurements and procedures used that could impact the measurement taken. In typical measurement systems, we generally include the measurement instrument and the people making the measurements.
One of the keys to a successful measurement system evaluation (MSE) is that we duplicate all the conditions we intend to bring under statistical control. Making decisions based upon measurements must be done with high levels of confidence that the data collected reflects the true process performance. It is for this reason that the measurement system being used must be evaluated correctly.
The first step in any study is to determine the details. There are four questions that must be answered to create the study details:
- Who should be included in the study? Normally, the current operators who now use the gage being studied should be part of this study.
- What is the sampling plan to be used during the study? Typically 10–15 samples will be sufficient for the initial study.
- How many times should each sample be measured? I always recommend that each sample should be measured at least twice, but three times is more thorough.
- Is it necessary to measure some sort of standard during this measurement system analysis? My answer is yes, absolutely, simply because it provides a level of reliability that your measurement system is operating correctly.
Collecting the data
Once we have completed the design of the study, it is now time to begin collecting the data. Each sample should fall within the expected operating range of the control characteristic being studied. The key is to duplicate the actual running conditions during the study. In addition, make sure that there is a repeatable procedure for making the measurements in place. Finally, it is important that each operator makes independent measurements so as not to be influenced by other operator’s results.
The results of the MSE must be reviewed for things like tolerance analysis (i.e., accuracy, precision, operator variation, and total error compared to both the tolerance and process spread width). In addition, the MSE must be documented, which will include the results of the initial study, any work done to improve the measurement system, and any drawings that were created as a result of the study.
Process potential and capability studies
The next step in the improvement process that I recommend is the completion of process potential and process capability studies. What’s the difference between the two studies? A process potential study typically consists of 30 individual samples measured by one operator on one shift, using product made from a single production run with no adjustments made to the equipment during the study. After the data is collected, the team should calculate two capability indices, namely Cp and Cpk, as well as making sure the data is normally distributed. We hope that the data is normally distributed (or close to normal), that the Cpk is greater than 1.33, and that Cp and Cpk are very close in value. If these conditions are not met, then more work must be done to increase the probability of success for the upcoming process capability study.
The process capability study represents data collected over an extended period of time, which is typically at least seven days with a minimum of 25 subgroups or 100 individual samples. The reason for this extended study is that we want to observe and measure normal variations associated with all of the input variables of the process. During collection of this study data, I can’t emphasize how important it is to record the details of all events which could affect the distribution of the data. When the data collection is complete, there are several key questions that must be answered:
- Is the data normally distributed? If the data is not normally distributed, we have more work to do.
- Is the process in statistical control? If not, then special cause variation is probably present. If it is, it must be removed.
- Does the Cp equal the Cpk, or are they at least close in value? If these two measures are not close in value, then the process is probably not centered. Action must then be taken to center the process.
As always, it is imperative that all work is not only documented, but that a time stamp is known for future teams.
Coming in the next post
In my next post, I will complete this series on practical process improvement by discussing the best way to implement control charts. I will also provide a more detailed explanation of Cp and Cpk.
Until next time,
Don't miss out!
Stay on top of the latest business acumen by subscribing to the Manufacturing Breakthrough blog.