Apr 02 2007

The Re-qualified(?) Mold

Published by at 2:41 pm under Uncategorized

Ever been told to Re-Qualify a running mold only because you’ve (1) moved to another machine or (2) Moved it to another of you facilities?  Kinda dumb.  But here’s what to do.

Let’s look at a piece of fiction to learn a lesson:  You work for INTERGALLACTIC TOOL AND MOLDING PARTNERS.  Your company has three facilities.  The one your work in is HQ.  Your sales force goes out and solicit jobs.  When the job is awarded it is transferred to a ‘project manager’ who is really a mid-level manager (your boss) who Lord’s over tooling and process engineers.  When the tool is finished the process engineers run the mold at this facility, qualify it and (should the customer want) run it for a period of months to assure the processing and costing remain stable.  After that period of time (unless the customer thinks differently) the mold is shipped to one of your production facilities who run at a lower cost.  The production facilities specialize only in molding and routine mold maintenance.  Major maintenance, Engineering Changes, tooling replacement, and blah-blah are the exclusive aegis of HQ.

All of your injection molding is run under the concept of decoupled molding.  Its procedures are almost viewed as Holy Writ and have been raised to the status of a cult.  Your quality people invest most of their time assuring the repeatability of the process before the mold goes into production.  You ship a small sample of parts for functional testing after you’ve achieved a stable process.  Production only begins when you receive a written Released For Production letter from your customer signed by the appropriately authority, based on your submitted samples. So far life is good.

With successful production it has come time to move this particular mold into your normal production facility.  Then the Bomb drops.  Your customer pulls a quality audit and finds several dimensions are not within the stated specifications although the part has never changed from the initial submitted samples and was Released For Production in writing.  You curse the ground the sales force walks on for not requiring an engineering change to make all the specifications line up with functionally acceptable parts before production began.  You go up the food chain to the project manager, who talks to sales, who talks to the customer, who (kicking and screaming) agrees the parts are good (because of the release letter) and the specifications need to be modified.  You receive assurances that they will ‘get to them as soon as they can’.  You go back to your cube a happy guy.

You are making arrangement to send the mold to your production facility and the bomb that initially dropped now explodes:  “Wait!” Says your customer’s QC people, “We know it’s qualified at HQ (and blah-blah) but we want it re-qualified at the new production source.  Otherwise we’ll be getting dimensionally different parts!”  Here’s where the folks who view quality as a religion but know nothing about injection molding, go astray.

The overall purpose of Quality Assurance is to minimize part variability.  If we were to go into the world of theoretical mathematics the formula for part variation would be a function of six variables:
1.    Specifications
2.    Design
3.    Tooling
4.    The specified plastic
5.    The molding machine
6.    The process used to mold the part
In a perfect world each of the component’s contribution to overall part variability would be equal to zero and when combined together the overall variation would be zero.  In the real world however there is variation, so the job is to have each component’s contribution to be minimal or it falls within a tolerance than (by definition) doesn’t contribute to variation.  Let’s look at each of them with this in mind.

1. Specifications have to be reality based.  What you specify has to be workable and manufacturable.  The more complex you make the part and its specifications; the higher the probability it will fail regardless of its variation.

2. Anyone can design anything.  All parts specifically require achievable tolerances.  Unless you are designing parts for micro-molding, it is a fool’s folly to live in the world of tenths of millimeters.  There are industry standards that can be referred to for the expected variation of a molded part.  You’ll get what the industry standards state, no matter what you specify.

3. With numeric cutting, CAD designs, and flow simulations tooling can be made with surprising precision to produce reasonably accurate parts.  Molds wear both from fatigue and abrasive materials.  With this in mind, good maintenance can interdict this as a contributor to part variation and the mold will operate repeatably for millions of closures.  When talking qualifications or re-qualifications it is important to remember that the mold is made out of metal.  Unless it is exposed to an abusive environment where it will chemically/dimensionally change (rust); it has the same dimensions whether it is in Boston or Beijing.

4. The specified plastic naturally has variation.  However by employing decoupled molding it is possible to overcome the effects of varying melt temperatures and flow characteristics minimizing if not eliminating the variation cause by the plastic.  It must be said however, that qualifying a mold using virgin plastic then going into production with a colored plastic, or switching grades of plastic, or the actual type of plastic will show variation from using the original virgin material.  But once the ‘new’ plastic has been processed properly, its inherent variability will also be minimal.

5. Poorly maintained or worn molding machines introduce their own variables simply because they cannot provide the precision repeatably required by the machine settings.  There are tests and information available to see if your machine is repeatable (robust) enough to produce parts with the least variation for precision molding.  Robust machines are repeatable.  They show little variation that will enter into the equation of part variability.

6. The premise of  the de-coupled molding process is to show the material a repeatable set of conditions so overwhelming that the material and the mold “see” the same thing every time.  Obviously the underpinning of decoupled molding is that the machine you use is Robust. The folks who make plastic food containers we all have in our refrigerators make and qualify their molds in the USA.  Their production facilities are all over the world.  The mold and the qualified process documentation commutes from manufacturing site to manufacturing site knowing a container made in Hong Kong or Hungry will fit its mating cover made in Japan or Jakarta.

Now let’s get back to the Bomb exploding.  This is a two handed argument.

On ONE HAND if the new or old manufacturing site doesn’t have robust machines, they are constantly diddling with the process conditions to maintain their quality. Because they keep adjusting the equipment they never really have a qualified process.  If they shipped the mold anywhere else, whatever work they did originally is useless.  Therefore the argument of re-qualifying at a new manufacturing site is completely valid.

On THE OTHER HAND if the current manufacturing site has nothing but robust machines, they do the qualification, ‘lock’ the process conditions in, and walk away as the machine spits out good parts with a minimum of variation.  You’ve also been polite enough not to mention you’ve run this mold and shipped off qualified parts having been run in several different machines.   When they ship the mold to another site that also has robust machines; at the very worst they have to translate the conditions to allow the mold to run in a different size or type of machine, the same way you did at HQ.  However from the point of view of the plastic and the mold, if nothing has changed, whatever variability was seen before will simply be identical in the new facility.  Therefore in a practical sense the argument for re-qualifying at the new manufacturing site is a waste of time, money, and talent.

Probably the toughest argument to a customer is to not talk about their mold or part but to talk about your company’s ability to do robust processing at all of your manufacturing sites.  My best, but laughable, analogy to clients is baking cookies:  No matter where I am I can follow the recipe repeatably and exactly.  My only problem is that when I set the oven at the appropriate temperature (the process).  The only way I’ll get same flavor, ‘doneness’, moist texture and all the other things people like about baked cookies is that the cookie dough sees the same temperature air circulation and whatever else (robust) ovens do; regardless of where it is baked.  If that doesn’t happen, the cookies never come out the same.

Cookies aside, if your customer insisted on a re-qualification, I’ve never been against folks paying for something this silly.  I am however, strongly against having to pay for it as the cost of moving the mold. Here’s an interesting way to avoid the time, effort and cost of re-qualifying a mold that doesn’t need re-qualifying:

It is called ‘The Crown Jewels Test’.
Take the parts you molded at the previous facility and the new and label them randomly.  What is important is that you have a representative sample of both parts, not necessarily a 50:50 mix. Go to your customer and say “Of these (for this example) twenty-five parts there are UP TO a dozen parts made at the old facility in this inspection sample, but there MAY be only one.  If you are correct in requiring a re-qualification; your inspectors should be able to separate these sample parts into the two groups, based on your historical inspection records with 100% accuracy.  If you’re wrong, it will be impossible”. Within the bounds of the molded part specifications and the robustness of your processes; it is a sucker’s bet.  My advice is to never make this bet until you already know the outcome before you bet the expense of a re-qualification.

Do you have Robust Machines?  There are two easy tests you can run.  One does nothing more than look at the timing and pressure between the averages of ten air shots and ten injected shots.  All the credit for this test goes to RJG Associates.  This is an easy test but only tells you if your machine is robust or not.

The late Denis Hunkar developed a classification system for machines based on the precision (lack of variation) of several molding parameters.  You measure each parameter and the one with the widest variation tells you the class of the machine and if it is robust or not.  Hunkar’s system points to the deficiencies.  Usually with a few minor repairs you can upgrade a non-robust machine to a robust one.  Both tests are part of a spreadsheet available for free at this link.
They are excerpted from the book QUALIFICATIONS STARTUPS AND TRYOUTS available from the bookstore at wjtassociates.com

*   *   *   *
As with all my articles you might agree with me on the first reading (the initial qualification) then when you are at your desk in the Cube Farm decide on the second reading your wussy management would never go for it (resulting in the required unnecessary but politically correct, ‘the-customer-(even if they are morons)-can-do-no-wrong’, profit eroding re-qualification).  OR you can print it out, roll up the paper into a tight tube and go swat the salespeople on the head for not standing up for you.
Alternatively you can think this is all a pile of Horse Purge and simply enjoy the laugh then delete this article.  The choice is yours.

No responses yet

Trackback URI | Comments RSS

Leave a Reply