SPC for Plastic Injection Molded parts

J

JPDaniel

Dear all,

As you may know that plastic components keep shrinking right from the moment the part is ejected from the mold and stabilizes after many hours. Can you please explain how SPC is practiced for plastic parts for critical dimensions ?

Thanks and Regards,
Daniel
 
M

mlee97

Personally, I would perform studies to create a predictor equation and plot the results in SPC. ( something like " Tfactor(time in hours) * tp factor(part temp) * (cross section diameter)." That would give me a pretty good idea of the final outcome much earlier in the process. One could simply plot the factors, but the relationship gets lost on the shop floor operators.
 

Bev D

Heretical Statistician
Leader
Super Moderator
I create a prediction equation that relates the 'fresh' hot dimension to the 'cool' stable final dimension that the custoemr will see. then I set specifications for the hot dimension that will yield in spec cool dimensions. this is a specification exercise of course.

but it sets up the SPC. I monitor the 'hot' results on a standard control chart (OK except for the fact that most injection molding has very little variation from piece to peice out of the same cavity and press so I have to use a different rules than standard shewhart charts). I track the 'shrink' by plotting the hot measurement - the cold measurement on a few parts every batch or soem time frequency for longer runs.
 
M

MikeSeymourAtl

Much like mlee97, I performed a detailed shrink study. I had the resources to be able to check several sample sets after 1 hour, 2 hours, 1 day, 2 days, etc, until they stopped shrinking. (Polypro, 9 days!). That gave me a detailed "shrink curve". Now I can use that curve for production monitoring, process improvement, tool improvements, all that.
For production monitoring, I have an "egg timer"; we pull the parts from the mold, and then let them sit for 60 minutes. That's about as long as we want to wait for results. At that point in time, the parts have experienced about 30% of the shrink. More importantly, I recognize, from my shrink study, that if the Quality Tech measures at 60 +/-5 minutes from mold time, we will get consistent measurements. Any less than an hour, and the results will be too volatile.
To make a short story short, pick a time frame, and stick with it. Even if you have to put egg timers next to the parts. Works like a charm for us.
 
J

JPDaniel

Thanks for the very valuable comments provided by all of you. Developing a shrink rate curve and obtaining an equation is the majority response. After reading Howard's article I wonder if how strong would be the correlation between the part weight and the dimensions. Since most modern machines have inbuilt SPC monitoring scheme for m/c parameter monitoring, a mechanism to sort suspect parts automatically would be possible, if there is a feed back from the machine. Thanks and Best Regads, Daniel
 
A

adamsjm

I have studied both methods and after much molding experience, I have come to the following conclusions.

Dimensional stability is usually well between USL and LSL for well designed parts (with some specific exceptions.) The shrink factor can be taken into effect when parts are measured on a consistent basis form the time of molding. Part weight is difficult to implement over a long run of parts. The specific gravity of the molding compound varies too widely. [Compounders usually give a 10% range.] Visual control of shorts, flash, and gas burn almost always proceeds dimensional and mass calculation control charts.
[I had one part which was in a state of dimensional control with a Ppk of 12 which looked like Swiss cheese.]

Just saying.
 

Proud Liberal

Quite Involved in Discussions
Also, do NOT use Xbar and R charts. Plastic injection molding is an autocorrelated process. As such, subgroup estimates of standard deviation will be artificially low resulting in control limits that are too close together.

Individual / Moving Range charts are more appropriate for your process.
 
Top Bottom