R
Rio BP Simbolon
Yes, this is good writing. Thanks for your sharing
SPC and Injection Moulding
- Thread starter Howard Atkins
- Start date
leftoverture
Involved In Discussions
Reviving an older thread. It has been many years and I am interested in knowing what people have found since this thread originally started. What have your experiences with SPC in the injection molding industry been lately? My company has just begun the SPC journey, and I am very interested in any benchmarks.
Thanks,
Tim
Thanks,
Tim
ezekieltemple
Starting to get Involved
{{{crickets chirpin}}}
I am interested in any updates in commonly accepted practices for implementing SPC in injection molding (actually we deal with rubber, but data on plastics is welcome as well)...
Can anyone bring us up to date...?
I am interested in any updates in commonly accepted practices for implementing SPC in injection molding (actually we deal with rubber, but data on plastics is welcome as well)...
Can anyone bring us up to date...?
leftoverture
Involved In Discussions
Hi Ezekiel,
Great name. Love it!
As for SPC, since I posted in 2017 we have made quite a bit of progress but maybe not as much as I hoped. We do X-bar and R charts on many higher volume jobs, but most of our production seems to be in the lower to medium run size. For these we use individuals charts (run charts) and we set guardband limits on those. Our guardband limits typically are a percentage of the total tolerance and serve to help us react before we go out of tolerance.
One thing we found, with the lower volume jobs, is that material lot can be an assignable cause. So, we run for maybe 36 hours, then do not run again for another 3 months. When we run again in 3 months it is a different lot of material. This played havoc with calculated control limits, which is why we switched many of these jobs to calculated guardband limits. This seems to work quite well for us.
I would be happy to stay in touch with you can compare notes and to provide any input I can.
Thanks,
Tim
Great name. Love it!
As for SPC, since I posted in 2017 we have made quite a bit of progress but maybe not as much as I hoped. We do X-bar and R charts on many higher volume jobs, but most of our production seems to be in the lower to medium run size. For these we use individuals charts (run charts) and we set guardband limits on those. Our guardband limits typically are a percentage of the total tolerance and serve to help us react before we go out of tolerance.
One thing we found, with the lower volume jobs, is that material lot can be an assignable cause. So, we run for maybe 36 hours, then do not run again for another 3 months. When we run again in 3 months it is a different lot of material. This played havoc with calculated control limits, which is why we switched many of these jobs to calculated guardband limits. This seems to work quite well for us.
I would be happy to stay in touch with you can compare notes and to provide any input I can.
Thanks,
Tim
Johnny Quality
Quite Involved in Discussions
Tim,
Seems you've already read the paper.
What's your experience with SPC in injection moulding now? Has it improved your organization?
Seems you've already read the paper.
What's your experience with SPC in injection moulding now? Has it improved your organization?
leftoverture
Involved In Discussions
Well, I had forgotten about this thread. If you go back to the second response in this thread, from Razors Edge, I would say this is spot on and not much has changed over the years. Just like I said above, material lot is an assignable cause and it will wreak havoc on efforts to be in "statistical control" when using traditional X-bar and R chart methods. This is especially true for part weights. Dimensions are maybe a bit less sensitive to material lot variations. I discovered this back in 1988 and not much has changed since then, even with better machinery.
The trouble is, when people say "SPC" a lot of folks instantly think "X-bar and R charts". Statistical process control can take many forms, and the run charts with guardband limits that we use certainly fall in the category of SPC methodology. So do p-charts, Pareto charts, and many other forms of data collection and analysis. So we are an SPC company, even though our primary methodology is not X-bar & R charting.
Trying to get customers who are not as familiar with injection molding to understand our methods is not always easy, but it is necessary. As "the molding experts" we need to train our customers. In the quality profession we often talk about "best practices" but we don't talk often enough about "worst practices". Trying to get your suppliers to blindly follow a given SPC method such as X-bar & R charting without understanding the uniqueness of their process is a "worst practice".
The trouble is, when people say "SPC" a lot of folks instantly think "X-bar and R charts". Statistical process control can take many forms, and the run charts with guardband limits that we use certainly fall in the category of SPC methodology. So do p-charts, Pareto charts, and many other forms of data collection and analysis. So we are an SPC company, even though our primary methodology is not X-bar & R charting.
Trying to get customers who are not as familiar with injection molding to understand our methods is not always easy, but it is necessary. As "the molding experts" we need to train our customers. In the quality profession we often talk about "best practices" but we don't talk often enough about "worst practices". Trying to get your suppliers to blindly follow a given SPC method such as X-bar & R charting without understanding the uniqueness of their process is a "worst practice".
It goes back to meaningful signals to control the process. Once I was injection molding Torlon, and the most important data I needed was the frequency the heaters cycled. Once the heaters slowed - stopped - cycling, I knew I had to purge or the screw would lock up and need burnt or pounded out. At that time the screw was generating more heat from friction than the screw needed from the heaters. Yikes! It was done with a chart recorder, but essentially the same thing without sampling. I always look for data that helps make decisions, then try to apply statistics - sampling, charting, etc., to improve the decision making process. Pointing to a dimension and saying "chart this" probably will frustrate more than be successful.
Similar threads
