SPC from Each Cavity? Small Plastic Parts using Injection Molding Machine

[Marcin]

We manufacture small plastic parts using injection-moulding machine. Sometimes there are 2, 3 or 4 cavities in the moulding-tool, so in the same time machine injects and moulds 2, 3 or 4 parts.
The question is if we need to measure (for Ppk, Cpk and Cmk) parts from each cavity or it is enough to measure only one (always the same of course) that represents current injection process.

------------------

 

[pdboilermaker]

For this application, it would make sense to collect data on each of the cavities, each cavity is slightly different from the time the tool is new. After the tool is put into use each cavity will "wear" slightly different over time. Good data samples will allow you to detect this wear gradually over time (possibly used as a predictive maintenance tool). Allows you to say "hey this tool needs to be fixed before it becomes a problem for our customer" rather than your customer saying "part "a" will not fit into part "b" anymore, it was good but now it's not, please answer this corrective action explaining why"

 

[Marcin]

Yes it's true that each cavity will "wear" slightly different over time. But the process will be the same for all cavities. If the process is bad for one, it is bad for others as well. Am I right? I forgot to mention that besides SPC measurements there are "Special Characteristic" measurements performed during the process for important dimensions (for each cavity). And this makes us sure that each cavity remains within required tolerance and during the life-time of the tool gradual wearing will be detected.

So now, is it necessary to measure samples from each cavity to follow the process?

Answering Marc Smith: automotive.

 

[Marc]

You gave 2 important criteria:

Automotive
Special Characterics

The probable answer is - yes. I say probable as that has been my experience, but your customer is the ultimate in determining requirements.

In so far as it goes, the equal wear theory is typically correct, but not always. In addition, an individual cavity could be damaged in a number of ways.

I would communicate with your customer and ask them what their expectations. In the QS/Automotive world I would expect your customer to require data from each cavity.

 

[Kevin Mader]

Marcin,

I agree with most of what is written above. Keep in mind that while the process is the same, the results for each cavity can be quite different.

Let me give you an example. We have parts being produced for us with a multicavity tools (2,4, 16, and even one 64 cavity tool). The one that comes to mind was a 4 cavity tool. Eventually over time, we began to receive 1 bad cavity, then 2, eventually 3. Our supplier did not use process controls (nor inspection for that matter). This lead to production delays. Not a desired state.

What had happened was that thin slides/inserts within the mold began to wear and crack over time. Each failed at different rates and times. If they had been running SPC on each cavity (even a modest 2 or 3 part sample for each) they would have detected tool wear before failure of one or more slides. This could have been detected with variable or attribute data. The thing to remember that although a single machine and process are being used, each cavity represents an individual population. Combining the cavities creates a bimodal distribution, which is often difficult interpret in detecting potential/actual issues (and incorrect).

My suggestion: measure each separately.

Regards,

Kevin

 

[Don Winton]

<font COLOR=RED><BLOCKQUOTE>What is Cmk?</BLOCKQUOTE></font>

This is a term used for machine capability. See:

http://Elsmar.com/ubb/Forum10/HTML/000067.html

Regards,

dWizard

[This message has been edited by Don Winton (edited 10 April 2000).]