Here are some of the frequently asked questions (FAQ) we have been asked. We will add to this over time…
Can someone share how they have done gage R&R studies for destructive testing?
The only approach I’m aware of is to take samples with as close to identical properties (manufactured in the same lot, right after each other, with no other difference in people or machines), in order to minimize the variation between the parts. If possible, take one part and separate it into smaller pieces to perform the testing. Then you would treat these parts as “repeats” in the Gage R&R, even though technically they are not repeats (already destroyed). The problem is that variation in repeatability will be combined with any variation in these “repeat” parts, so it’s not a perfect Gage R&R, but should give you a good idea whether the measurement system is adequate.
STATISTICAL PROCESS CONTROL
I have a process with 10 samples per subgroup, and 20 subgroups total. When I only plot the average of the subgroups on an Individuals chart (ignoring the individual data points completely), the data values fall within the control limits. However, when I run the data in an X-bar and R chart (using the 10 samples per subgroup), the data points are out of control?
The control limits will be different. The Individuals chart calculates the standard deviation for the control limits based upon the overall moving range average (average difference from current data point to previous data point), divided by a constant, d2. In the X-bar and R charts, the control limits are calculated based on the variation within the subgroup, which is the overall range average (ranges of each subgroup), multiplied by A2.
|Control Limits for X-bar and R chart||Control Limits for Individuals chart|
Why aren’t specification (spec) limits displayed on control charts?
Control charts are intended to monitor the random variation in a process, which should not be affected by the specification limits (tolerance). Displaying spec limits can drive tampering and adjustments to a process in order to achieve the limits, which causes the process to go “out of control.” If the process is not capable of achieving the limits, then changes to the process mean or a reduction in standard deviation should be addressed. Control charts are measures of process stability only.
If you’d like to learn more about control charts, download our Control Charts PowerPoint course >>>
STANDARD OPERATING PROCEDUCURES (SOP)
How would you assess a Standard Operating Procedures? Is there any LSS tool or technique?
The main approach I would use is called “gemba,” where you go to the work area that represents the SOP, and review it with the workers. Observe how they follow (or don’t follow) the process, and seek to understand why it isn’t being followed. Is the SOP correct? Did they get trained on how it works? Do they have the right tools or access to the right systems? Are they given enough time to complete the steps?
If there are disagreements on how the process should be documented, then I would get them together to decide how to come to an agreement (maybe part off a kaizen event), or agree to test out their approach through experimentation, where the best approach is selected, based on the results of trying out everyone’s technique.
Finally, I would look for ways to mistake proof the process so that it’s easy to follow without always referencing the document. Also, look for ways to implement 5s principles of workplace organization (signs, labels, color coding, visual controls, etc)