You have got me on my high horse now.
IMHO the demands for SPC in QS are antedeluvian and refer to process that are controllable and are given to immediate detection.
Most SPC is on the part and not on the process. I experimented with finding parameters of injection molding that can be charted but in most cases this was not possible.
If you are using closed loop machines then they corrct themselves. The only real parameters that are worth checking are total cycle time and plastification time.
As plastic shrinks to measure dimensions immediatley means extrapolation of the shrink and thus measurement of the "wrong" dimension.
I have persuaded customers that this is irrelevant to the spirit of the Standard.
Weight is a parameter that could be used but it needs to be changed according to the density of each lot of material.
For PPAP I submit a PPK number based on the measurements of samples 24 hours after production. This is acceptable.
I hope that this helps you for further information please return.

We are an injection moulding and metal pressing company supplying 1st automotive suppliers. As a result of a recent QS9000 audit i have been asked to re-look at the entire SPC strategy as to its applicability in supporting process controls (mainly process parameter control, First off, patrol inspection etc) and reviewing what is sensible industry practice. I am currently concentrating mainly on new product introduction - What are other similar organisations doing ?

I agree with Howard. I have tried several different methods for applying SPC to the injection molding process. For the same reasons Howard cited I too have reverted to dimensional measurement and/or testing to satisfy the requirement.

"For PPAP I submit a PPK number based on the measurements of samples 24 hours after production. This is acceptable."

This is the same criteria that I have used and have found it to be adequate, in that most materials haved "cured" within this time frame.

Ford requires on-going SPC to monitor the dimensional and test results and this method satisfys them when noted in the control plan.

Thanks Howard. I did read it. By the way guys can anyone guide me on doing SPC for Plastic Injection moulding?

Thanks Howard. I did read it. By the way guys can anyone guide me on doing SPC for Plastic Injection moulding?
First what and why do you want to do?

As we say SPC on dimensions is not useful.

Please give more info

Thanks

Well Howard,

I'm thinking of doing it on moulding defects. We are currently using a material called Bakelite. And there are some defects such as excessive flashing, short shot, burnt part etc. These defects appear at the moulding stage itself.

Further down the line, there are secondary processes such as deflashing/deburring, screw hole taping, clip hole drilling, etc. Now I would like to apply the SPC on this line of processes. Lets say:-
1. Molding
2. Runner cut
3. Deflashing
4. Screw Hole tapping
5. Packaging

I'm also not looking at the dimensions. Right now its more focused at the appearance of the part itself. Dimensions are usually ok. The Outgoing QC in my company confirms that dimensions are ok (except in the case of short shot of course).But usually what fails is the appearance. Eg. The part may have flow marks, Deflashing notdone properly(i.e flashing still visible) etc.

I'm sorry I didn't explain further in detail in my original thread. As I am new to this field. :o .
I hope this provides further explanation. Thanks

Now I would like to apply the SPC on this line of processes. Lets say:-
1. Molding
2. Runner cut
3. Deflashing
4. Screw Hole tapping
5. Packaging


Nice list. They are all dependent variables. Now, find the independent variables that control these variables. Perhaps a fishbone diagram with each of these items as an output would be a good start.

Well I will begin the fishbone diagram ASAP. The problem is that currently all the processes are already in the Masspro stage. I would have to start with a current running one. I was actually thinking for a new product. That was what I thought will be easier. But will try to fit in the analysis into my work. Unfortunately I'm the only engineer in the whole company, so right now I'm overloaded with work. But will try to do it ASAP.:thanx:

In general I agree that the visual defects are the biggest problem and dimensions are usually stable. This is one of my objections.

In my mind to find the contributing variables is likely to be very difficult and controlling them even more difficult.
I do not believe that you will find one factor that affects problems such as :

excessive flashing
short shot
burnt part

These will be combinations of temperatures, mold cleanliness, viscosity etc etc.
In some cases the machine parameters are so accurate that control charts will be valueless. If you have close loop machines again they are not helpful.

I think that you will find that there are so many different variables that you will not succeed to control them
I am sorry to be so pessimistic but rather than use SPC ensure that the equipment is well maintained.
I do not think that you will be able to get early warning which is the point of SPC as in many of these cases they are one off events.

In my mind to find the contributing variables is likely to be very difficult and controlling them even more difficult.


I agree. That is one of the fundamental difficulties with plastics injection molding - especially capturing lot to lot issues. But, that really does not preclude the attempt to do so.

I do not believe that you will find one factor that affects problems such as :
excessive flashing
short shot
burnt part
These will be combinations of temperatures, mold cleanliness, viscosity etc etc.

Good opportunity for a DOE to sort out the variables that make the most difference. The frustration of charting the wrong or insignificant variables will likely lead one to a conclusion that SPC does not work. But, it may not be SPC that is the problem.

In some cases the machine parameters are so accurate that control charts will be valueless. If you have close loop machines again they are not helpful.

Yes, those parameters will likely not be the ones to be charted.

I think that you will find that there are so many different variables that you will not succeed to control them.

Not all of them, but perhaps some of them - and perhaps the most significant.

I am sorry to be so pessimistic but rather than use SPC ensure that the equipment is well maintained.

I would look at both.

I do not think that you will be able to get early warning which is the point of SPC as in many of these cases they are one off events.

I would not give up so easily. I had a situation where I was injection molding Torlon. The screw kept getting bound up in the barrel because the material would break down. When it got bad I had to do very bad things to get the screw out. By charting the heaters, I was able to determine that the heaters would start cycling slower - and when it got to a limit, I would purge. By using that control, I saved the machine from a lot of abuse.

Sure, it is only one example...but it is an example!

I'm glad to see this thread started again. We've just started a SPC program, tracking 4 variables: weight, cycle time, peak pressure, fill time. I'm relatively new to injection molding, but I'm still on the fence that the latter 3 variables are meaningful to track. Cycle time and fill time don't seem like real outputs of the process, just settings. Again, I'm still learning about the molding process. Other parties (engineer, technicians) in the organization came up with the variables that they thought were appropriate.

Charting these items, we will get values literally off the charts!
I know the drastic changes on the chart can be linked to process changes techs have made on the machines. The charts are just telling us what has happened, instead of what will happen in future.

So far, it's been a challenge, and we've yet to see value. But we're hanging in there!

I'm glad to see this thread started again. We've just started a SPC program, tracking 4 variables: weight, cycle time, peak pressure, fill time. I'm relatively new to injection molding, but I'm still on the fence that the latter 3 variables are meaningful to track. Cycle time and fill time don't seem like real outputs of the process, just settings.

I think you sense some of the issues here. First step - find some variables. Second step - sort them into dependent and independent. I take a stab at the ones you have - assuming typical process:

weight - dependent
cycle time - independent
peak pressure - independent
fill time - independent

By the way, "settings" can be roughly considered "independent" if that helps the sorting - but not all independent variables are settings!

Are your settings constant - or do they in reality vary due to changes in material, humidity, etc.? If the are set, and do not vary, there is little to chart. You might as well chart the on/off switch!

So, then, what variables (as in varying!) that are affecting the dependent variable - weight? And, is it the only output (dependent variable) that changes?

I know the drastic changes on the chart can be linked to process changes techs have made on the machines. The charts are just telling us what has happened, instead of what will happen in future.


No we are getting somewhere! Why are changes techs making process changes on the machines? How do they know what to change? How do they verify those changes? Look at those variables! Are the techs becoming the process? That is a real danger in many of these processes. It is an issue we keep trying to get people to eliminate in machining by using correct charting methodology.

By the way...watch out for nasty measurements. Weight and physical data (such as tensile test) have very risky measurements. Do not forget your GR&R (ouch - destructive gr&r for tensile!) to ensure that you are measuring the process and not the variation from measurement and gage error!!

I

I would not give up so easily. I had a situation where I was injection molding Torlon. The screw kept getting bound up in the barrel because the material would break down. When it got bad I had to do very bad things to get the screw out. By charting the heaters, I was able to determine that the heaters would start cycling slower - and when it got to a limit, I would purge. By using that control, I saved the machine from a lot of abuse.

Sure, it is only one example...but it is an example!

I am pessimistic so that I can be pleasantly surprised!!!!!


Your example is of process problems rather than the result of the process an as such
Is it relevant????

Your example is of process problems rather than the result of the process an as such
Is it relevant????

Not to anyone else's situation - and probably not the clearest example. But, it was a process parameter that generated a signal that required operator attention from data. I will give it credit.

I guess we will see if we can come up with better examples. That would be cool.

Perhaps for plastic injection molding we should be benchmarking chaos theory? :tg: