I'm working on the SPC/measurement section of the local SkillsUSA aka VICA precision machining competition. The students entering this contest are high school students, and they probably haven't had any real training in SPC. (I'm taking a machine tool technology associates degree program and it doesn't include SPC....)
I'm operating on the assumption that the level of knowledge to be tested is that of a machinist who doesn't freak out when he is told by a QA guy that "the process is out of control", or "the pareto chart suggests that the following machining operations need to be better controlled".
Since measurement is combined with SPC, I expect to include some samples to measure and add to a chart and then answer some SPC related questions about the measurements and the chart.
Others are putting together test sections on print reading, lathe turning and milling tasks, etc.
Also of note: the national contest lists as the 2nd of 4 hints:
Teach students statistical process control (SPC).
So, the topic of SPC can't be covered in depth and it would be useful if it taught some key SPC concepts to the astute contestant.
Do you have suggestions for questions? Or a pointer to a quiz or two online?
Send to friendSign In to get started!
Comments
guest 9/12/2001
I see that my "discreet" email address did not take in my original post. So, I will try it again, raubitz7@aol.com.
guest 9/12/2001
For the past few years, I have used an accept on zero sampling plan. This eliminates one of the questions but you must still pick the AQL Level.
My theory is that anytime a defective part is found, the sample fails and some one must determine the criticality of the failure. Since I have the most experience in looking at parts, I am the one to review the defect. I can then direct further efforts as needed.
I have extracted the accept on zero plan from Mil-std-105 and could provide a copy if requested.
Sincerely, Ed Young eyoung@intronicspower.com
guest 9/12/2001
QUALITY CONTROL, by Dr. Dale H. Besterfield, P.E.; Prentice-Hall, 1979; has a very good explanation in Chapter 6.
Beginning on page 174 "The acceptable quality level (AQL) is the most important part of the standard, because the AQL and the sample-size letter index the sampling plans. AQL is defined as the maximum percent defective (or the maximum number of defects per hundred units) that, for purposes of sampling inspection, can be considered satisfactory as a process average. The phrase 'can be considered satisfactory' is interpreted as a producer's risk, [greek letter] alpha, equal to 0.05; actually, [greek letter] alpha varies from 0.01 to 0.10.
When the standard is used for percent defective plans, the AQL's range from 0.010% to 10.0%. For defect-per-unit plans, there are additional AQL's so AQL's are possible from 0.010 defect per 100 units to 1000 defects per 100 units. The AQL's are in a geometric progression, each being approximately 1.585 times the preceding one."
Picking up the thread on Sample Size on page 176 Dr. Besterfield writes: "The sample size is determined by the lot size and the inspection level. ... Three general levels of inspection levels (I, II, III) are given in Table 6-5. The different levels of inspection provide approximately the same protection to the producer, but different protections to the consumer. Inspection level II is the norm, with level I providing 1/2 the amount of inspection and level III providing about twice the amount of inspection. Thus, level III gives a steeper OC curve and consequently more discrimination and increased inspection cost. ... Four additional special levels (S-1, S-2, S-3, and S-4) are given in Table 6-5 and may be used where relatively small sample sizes are necessary and large sampling risks can or most be tolerated."
With a previous manufacturing employer our use of AQL sampling plans was used to reduce our risk and "beat the supplier over the head until they saw the light."
Normally, a supplier started out at level II (normal) inspection requirements for X number of lots. If the incoming product was of sufficient quality over the X number of lots the supplier was then placed on level I (reduced) inspection requirements. The supplier then stayed at level I requirements until one of two scenarios occured. Scenario One: A lot has a number of defects greater than the Accept level but less than the Reject level. The supplier automatically goes back to level II (normal) inspection requirements until X number of consecutive good lots in reached. Scenario Two: A supplier on level I (reduced) inspection has a lot that exceeds the Reject level requirements. The supplier is placed on level III (tightened) until X number of consecutive good lots are received. After X number of good lots are received then the supplier is placed on level II (normal) inspection. The supplier stays at level II until their ability is again demonstrated, i.e., X number of acceptable lots, at which point they are then allowed to be placed on level I (reduced) inspection. This method proved to be very effective when developing new suppliers who had not become, or where in the process of becoming, ISO certified.
With a major portion of quarterly (in our case) vendor ratings (and contracts) depending on acceptable product lots, the suppliers soon "saw the light" as far as getting their in-house quality systems working right as their AQL level and performance was noted on their Vendor Rating Report Card.
For a copy of the exact procedure, as written by me for the previous employer, contact me discreetly at <raubitz7@aol.com>
guest 9/12/2001
Unless you are required by a military contract to use MIL-STD 105, that was obsoleted in 1995 and replaced by ANSI/ASQ Z1.4, I would not. If you have the need to use a sampling plan then C=0 "Zero Acceptance Number Sampling Plans by Nicholas Squeglia is a better choice although not the best. Ideally you should be requesting test or SPC data from your suppliers. In direct response to your question the use of the AQL levels and catagory is defined within the standard. Basically the AQL is a level of how much risk of accepting nonconforming product you are willing to assume. I no longer have a copy of it otherwise I would point you to the specific area where it can be found.
chrisco1 4/9/2002
Greatings!
I'm trying to set up a ansi z1.4 sampling plan to accomodate 3 sigma not 6 sigma.
What can you share with me on this?
I'm limited to two inspectors and large lot sizes.
Thanks.
Chris.
donwheeler 9/5/2001
Deming observed that the main purpose of Mil-Std 105 was to beat the vendor over the head. You cannot improve the quality in the process stream using this approach. Neither can you successfully filter out the bad stuff. About the only place that this procedure will help is in trying to determine which batches have already been screened and which batches are raw, unscreened, run-of-the-mill bad stuff from your supplier. I taught these techniques for years, but have repented of this error in judgment. The only appropriate levels of inspection are (1) all or (2) none. Anything else is just playing roulette with the product.
guest 9/12/2001
I have always believed that the only appropriate levels of inspection are all or none. So why does ASQ still have Acceptance Sampling contained in their body of knowledge for their CQE exam?
guest 9/12/2001
It is impossible to inspect the quality of a product in once it has been completed. The idea that we either select to inspect all or none is ludicrous in a competitive society.
If you select to inspect "all" then how much do you inspect. If the product has a possible 200 mechanical charateristics do you inspect all 200? What about it's electrical characteristics. there are a myriad that we could select.
To inspect "none" is probably closer to sane if the processes used are arduously and correctly monitored and audited. Even with stable processes the selection of a small sample would be inspected for critical parameters to assure that the wear on the machinery is still within tolerance and that the process is still stable.
As Quality Assurance professionals I find it surprising to see comments on sampling such as the ones noted. Sampling is only as good as the processes and controls which produce the product. If the process is not within it's designed control limits then two things are true: 1). You are not being competitive with the rest of the world and 2). You better inspect every product or be prepared to go out of business.
dave555 9/13/2001
You have established the reason that AQL and similar standards are used: however, from your title, I would have expected you to go well beyond this into details as to why a given AQL level might be used for certain types of features. Have I missed the intent of your message?