Cpk definition for EBM and SBM bottles,

[Bev D]

Not to mention that if you sample from every cavity (as you probably should) your Ppk will be unsalted by the systemic difference between cavities. Then your Cpk can be really messed up depending on how much systemic variation exists between cavities and how you subgroup. And a single sample from only one setup & material batch is another statistical blunder.

Cpk and Ppk are just abominations of mathematical formulas masquerading as ‘statistics’. Too many people just blindly dump some data into a formula they don’t understand via some software program and then think it has some meaning…it doesn’t. Quality engineering is way more complicated than the internet PhDs would like you to think.

My advice would be to discuss the critical characteristics with your Customer and try to get them to forget about the non-critical ones. Present run chart data that distinguishes between cavities and if possible over 2-3 different material lots. (Google “multi-Vari” it will help you). This can show that you are not making any out of spec parts and that the use of the “Normal” distribution to predict OoS parts when none are observed violates statistical requirements for this to be useful…

 

[Bev D]

SO we can give you better advice if we understand your design relationship with your Customer.

Are you designing and selling ‘off the shelf’ bottles? In other words, you design the bottles and put them for sale ‘as is’ claiming only that they meet the published specifications. (Like commodity semiconductors, paint, screws, etc.)
OR
Did you design the bottles to meet a Customer’s specific use requirements? You own the drawing but your bottle must meet the Customer’s requirements for use, function and dimensions etc.
OR
Did the Customer give you the design/drawing to build the bottles to their exact specifications?

The answer matters to the best advice

 

[mcpe72]

our company design the bottles, but they have to meet specific customer specifications, depending on the client they request different things, we work for each company from R&D ( im in ) to production that is in each plant dedicated to only them.

 

[Bev D]

Are these single or multi use bottles?

 

[mcpe72]

for instance, Windex, its only for that customer, made only by us.

 

[Bev D]

Do you make the squirt mechanism or just the bottle?

 

[mcpe72]

just the bottle

 

[Bev D]

OK so the first question you need to ask yourself is if your specifications are correct - if you have a part that is at the specification limit but not out will the bottle still meet that Customers requirement as intended? If so ask your Customer what you should do with parts that are beyond the 1.33 (or whatever) line but are still in spec. Will they accept those parts or will you need to scrap them? That is a critical element to understanding why the Customer is insisting on Cpk - often a customer won’t believe your specs were properly engineered, that they are too loose to accommodate non-disciplined manufacturing or they truly believe that there are parts (never observed but still possible) that lie beyond 3 SDs from the mean. This lack of knowledge and belief leads Customers to blindly apply a Cpk requirement. In essence: are they just looking for tightened specs or are they looking for an imaginary guardband?

CAn you substitute a Cpk value for reporting your performance data in the form of a control chart (probably a 3D chart to deal with your non-homogeneity) or at least a multi-vari run chart with spec limits included to show that you are stable and predictable within the spec limits?

 

[mcpe72]

it is a qualification to start producing and validate the industrial tools, so they can start production, so its a technicality because the pilot tool was approved and weight, one of the non normal data its more a mistake from the operator more than the machine itself. for the body they vary so much that i think its silly, and we sholuld focus in the volume which gives us an idea or the goal which to maintain the right amount of content.

 

[Bev D]

Then you still need to talk to your Customer about what is critical and show them the data you have. Of course it’s silly. But that doesn’t maatter - you need to convince your customer that your DESIGN is capable.