Injection Molding Capability Studies for PPAP Submission

B

bfore

Guest
#1
Ok i'm back with some more questions, and maybe the final piece of the puzzle for my 1st ever PPAP submission!
First an overview, we are currently contracted to ship 400pcs weekly to our customer, (automotive component that is injection molded)
I had manufactured at on-set of this program a CMM fixture that also included an SPC port to monitor overall length of part (furthest point away from the fore/aft locator). The confusion is as follows, this is the only in-process check i have added to the fixture, the rest was designed to be certified by a CMM (part layouts)
How do i complete an SPC report for this?? I have read that i will require 4-5 values x 25x, do i just repeat the same measurment?? Can someone recommend an appropriate excel worksheet that i can submit the results in my PPAP submission?
Sorry these are real newb questions....but i am really newb!! Fortunately or unfortunately I am the chief cook and bottle washer...and have little quality experience!!
Thanks in advance! :)
 

bobdoering

Stop X-bar/R Madness!!
Trusted
#2
Re: Help with Capability (injection molding)

A couple questions: How many cavities are you dealing with? How many parts are run as the injection molding process comes to steady state (warm up-setup)? How many parts can you run after setup, but before any adjustment?
 
B

bfore

Guest
#3
Re: Help with Capability (injection molding)

This particular tooling is 1 cavity injection mold, generally the injection process is saved (machine parameters) and is very stable, however i'd say the first 10-15 piece are thrown away as start-up scrap. I hope this helps clarify.
 

bobdoering

Stop X-bar/R Madness!!
Trusted
#4
The first step is to ensure the gaging is suitable for the measurement. You want enough statistical resolution (ndc) to ensure the variation you detect is from the process, not the measurement system. ndc>10 is a good start.

I would take 100 consecutive pcs, and chart them as a run chart. If the overall length varies is randomly about a mean value, you may be on the right track. For automotive, they want a 300 pc run - to ensure the run is representative. If the process is very stable, you can use every other part. If you are collecting the data automatically, you can use all of the data (except warm-up - that is special cause variation you do not need to include).

Next step is to verify the distribution (I use Distribution Analyzer from variation.com). If the distribution is normal (or near normal), then you can calculate the capability (Pp. Ppk, etc.) from the data retrieved. (Distribution Analyzer will provide that calculation if you use it.)

If the process variation seems to be normal, then I would go with a X-MR chart with a single cavity process. I would not trust that over time a 5 pc sample will represent the long-term variation of the process (as is needed to make X-bar R chart control limit calculations of any value)

Does that help?
 
B

bfore

Guest
#5
I think it helps....
The "X-MR chart with a single cavity process" that you are referring to is available how?? Can i d/l from some of the stuff on this site?

Also, regarding Distribution Analyzer...are there any free (besides the 30 day trial) spreadsheets etc that accomplish this same task??

I have the data collected from only 100 pcs already, from our last production run, i guess i need to do some data entry now, but wanna tackle that with the least expensive option....

Thanks already for your help....I will get thru this eventually!!
 

bobdoering

Stop X-bar/R Madness!!
Trusted
#6
Here is one starting point for I MR (same as X MR) charting: I MR Chart

As far as something free similar to Distribution Analyzer, I am not aware of anything. You can pay more for Minitab, but Distribution Analyzer is a lot easier to use. It does quite a bit...more than just a spreadsheet.
 
J

Jjoed

Guest
#7
I'm a little late on this conversation, but I just thought I would throw my 2 cents in:). Could you do a process study and define what parameters in the molding process would yield good parts instead of checking the actual parts??Does this make any sense??
 

bobdoering

Stop X-bar/R Madness!!
Trusted
#8
I'm a little late on this conversation, but I just thought I would throw my 2 cents in:). Could you do a process study and define what parameters in the molding process would yield good parts instead of checking the actual parts??Does this make any sense??
You have to check parts as the output of the process parameters to see if they are optimized, so you don't get out of measuring parts completely. If your process is not capable, a DOE - similar to what you are suggesting - would certainly help. It can also be used as process improvement of a capable process. But, the capability study previously described is actually a very good starting point.
 
J

Jjoed

Guest
#9
Yes very true thanks!!. Question, after you have established the "process" by measuring the parts and making the adjustments to the molding/ machine parameters. Then you should be able to establish what machine parameters would yield accpetable parts and measure the machine outputs??

Sorry if I'm making more out of this thread than necessary
 

bobdoering

Stop X-bar/R Madness!!
Trusted
#10
Yes very true thanks!!. Question, after you have established the "process" by measuring the parts and making the adjustments to the molding/ machine parameters. Then you should be able to establish what machine parameters would yield acceptable parts and measure the machine outputs??
It certainly should allow you to determine the optimum range of controls for the machine. Of course, there are the other parameters that you have less control over, such as lot to lot variation of the raw material, environmental conditions that affect certain polymer systems, etc. That is why I stated the first step - always - is to develop the total variation equation. Whether charting or DOE, you need that solid foundation of process understanding.
 

Top