X-R chart - Suitable with method? Tubes and hoses - ID

[Bahgia Minhat]

Hi to all,

First of all, this site is cool. Its like I dun have to go for training .

Straight to the problem.

The company where I'm working is a manufacturer of tubes and hoses. One of the product is the wire harness tubing for insulation and we have obtained QS9000. The size of the inner diameter is from 3mm to 32mm (child part under wire harness) with thickness of 0.5mm and 1mm standard.

Characteristic and method of measurement
We take measurement of the Inner diameter (ID) by cutting the piece of each end of extruded tube and cut it perpendicularly to form a rectangular shape (as we call it the circumference). We use digital caliper (2 decimal point). The measurement is based on concept of measuring the circumference of a cylinder (mathematically, pi x radius...). The tolerance are mostly non-central type (eg; +0.6mm/-0.4mm)

Sampling
3 samples per hour.

Problems encountered.
The graph comes out terrible (super excessive!!).
This is due the the ID size is controlled manually by operator (air blower - to shape the tube's ID)
A minor change in ID causing excessive change in circumference ([0.01mm(ID)/2]^2 x 3.142(pi))
Operator use gaging in determining and setting the ID while inspector use caliper to measure the circumference.
What to do?

quantity output
Some part (different ID size) is produced at very minimal quantity (less than 50 or 2 hrs production.
This is not enough for the spc program?

New sheet?
Some part is produced at large quantity (1 week production non-stop). After 25 readings (25 hrs sample taken) should we open a new sheet (start new sampling and new monitoring). My auditor told me that we only need to monitor only 25 readings only. After that, we don't have to do X-R chart unless we start other production.

MSA
I've tried doing MSA with 2 different size (3mm and 32mm) with the caliper. It comes out that the ndc for 3mm size is too small (<5 - cannot detect variance in sample - instrument is not capable or suitable).

(****, I hate this!!) continue.....

FMEA, Control Plan, PPAP....
Under PPAP, or FMEA, or Control plan, we put - to control the ID is using caliper (See problem MSA, sampling, bla.. bla.. bla... surely this will contradicts each other with actual doings (inspection)).

Other type of SPC
I've tried using p, np chart (since operator use gage), but it also come out pretty bad. My auditor also said probably the X-R chart is not suitable since the production of the tube is continuous (extruded from machine is continuous, unlike injection or stamping)

Questions (again)

What is the suitable control (SPC) for parts like that? (sampling, method of measurement, instrument)

If we consider different size under child part/class how about the FMEA, Control plan, MSA, APQP... (how to make words to show we know what to do.. he he)

Untill now I always alter the data to show the process is in control (Hey! What else to do?) and I hate it!

Please do give advice....

 

[Caster]

Some first ideas to start this off

Welcome to the Cove!

The company where I'm working is a manufacturer of tubes and hoses. One of the product is the wire harness tubing for insulation…..



1) The graph comes out terrible (super excessive!!).


It seems your Operators use Go No Go s and your inspectors use Calipers?

Some simple things you may already have considered.

• Is the tubing material flexible? Can it be distorting when measured by the calipers (pinched?)
• Are the calipers used correctly? Are they being read correctly?
• Are the Go No Go gages used correctly? Do people force them to fit?

I suggest you go see these things for yourself, and not rely on what people say.

2) Small output and SPC (less than 50)


Search for short run SPC on the Cove and with Google. There are really simple methods that will work.


3) New sheet? Some part is produced at large quantity (1 week production non-stop). After 25 readings (25 hrs sample taken) should we open a new sheet (start new sampling and new monitoring). My auditor told me that we only need to monitor only 25 readings only. After that, we don't have to do X-R chart unless we start other production.


This seems really odd. I would expect that you would have a control chart with locked control limits and that you would plot data on it. The only reason I can see to stop using the chart would be if your process can show excellent capability.

4) MSA I've tried doing MSA with 2 different size (3mm and 32mm) with the caliper. It comes out that the ndc for 3mm size is too small (<5 - cannot detect variance in sample - instrument is not capable or suitable). The tolerance are mostly non-central type (eg; +0.6mm/-0.4mm).

To solve ndc you need a caliper that reads to 0.01 mm or you need to look at the variation in the sample you measured. If your parts for the MSA study are very similar in size, you will see low ndc as well.. Please look at MSA v3 page 46 and 75, it helped me with a similar problem

5) FMEA, Control Plan, PPAP....Under PPAP, or FMEA, or Control plan, we put - to control the ID is using caliper (See problem MSA, sampling, bla.. bla.. bla... surely this will contradicts each other with actual doings (inspection)).


You can have both the operator Go No Go and the inspector caliper checks listed separately on the CP. This shows you care enough to check it twice!



6) What is the suitable control (SPC) for parts like that? (sampling, method of measurement, instrument)



If you measure 3 parts per hour, an X bar and R chart should work just fine.



You need to investigate the poor results. It sounds like a measurement problem, not an SPC problem. The charts are working ! They are telling you to investigate.



7) If we consider different size under child part/class how about the FMEA, Control plan, MSA, APQP... (how to make words to show we know what to do.. he he)


I think you are asking about family documents. They are well accepted. In this case you would not need one set of documents for each part number, you can group them as a family of similar characteristics. Many people do this and it is accepted by auditors and automotive companies.



8) Until now I always alter the data to show the process is in control (Hey! What else to do?) and I hate it!



Hmmm…changing data is a personal choice. Another approach is to document the issue, begin corrective action and issue a waiver.


I suspect the parts are just fine. If the customer has no problems, you just need time to resolve the measurement problems.

Good luck!