# Determining AQL and Sample Plan - Batch size is from 1-500

1

#### 1winnyAK

I recently been assign a project on finding a correct sample plan.
My company does a 5% inspection once we get the part from our vendor from there we put it on the shelf. The problem I have is that we do not have much data at all besides the inspection of how many that was inspected when it comes in but nothing on when its on the shelf and use out on the the shop floor.

My boss thinks our AQL is 0.010, I differ on that but I could be wrong too. So if my AQL is 0.010 does that mean I accept lots with 0% defective only 100 of the time. In other words, more than 1 of "good" lots would be rejected. Lots would have to be down to about 0 defective before 100% would be accepted.
Is this correct?

Our batch size is from 1-500, it is very small. Also, what's the difference between Single Sample Plan and Double Sample Plan like what are the pro's and con's of using one or the other?

Last edited by a moderator:
S

#### sitapaty

I recently been assign a project on finding a correct sample plan.
My company does a 5% inspection once we get the part from our vendor from there we put it on the shelf. The problem I have is that we do not have much data at all besides the inspection of how many that was inspected when it comes in but nothing on when its on the shelf and use out on the the shop floor.

My boss thinks our AQL is 0.010, I differ on that but I could be wrong too. So if my AQL is 0.010 does that mean I accept lots with 0% defective only 100 of the time. In other words, more than 1 of "good" lots would be rejected. Lots would have to be down to about 0 defective before 100% would be accepted.
Is this correct?

Our batch size is from 1-500, it is very small. Also, what's the difference between Single Sample Plan and Double Sample Plan like what are the pro's and con's of using one or the other?

All these sampling schemes should be based on sound judgement of the criticality of characteristics you measure.There cannot be any hard and fast rule for fixing AQL.Statisticians say even in 100% inspection, 1% rejects escape.The only guaranty for quality is control of process through SPC.
Blindly following AQL sampling is like searching for a black cat in a dark room.
-Sitapaty

1

#### 1winnyAK

Okay if I understand correctly even if it is a 0.010 AQL there's always going to be a 1% rejection so that means in order for a lot to be accept it has to be 99%. Is that right.

1

#### 1winnyAK

So about SPC, If I have no data right now..I should start collecting data from the time there receive and than when they grab it off the shelf and collelct those too? Is that what u mean by SPC and determine my decision by that.

As far as deciding about sample size is it better to go off of that MIL-STD-105D chart?

C

#### CristianoP

My boss thinks our AQL is 0.010, I differ on that but I could be wrong too.
AQL is the proportion of defectives that you accept with a given probability. Usually, that probability is 95%.

Suppose your lot size is 500, then you inspect 25 samples (5%, you said). If you reject the lot when you find at least 1 defect, you have this plan:

grey: Pa (probability of acceptance)
orange: AOQ (average outgoing quality)
green: ATI (average total inspection)
abscissae: proportion of defectives

http://www.itl.nist.gov/div898/handbook/pmc/section2/pmc22.htm#Acceptable Quality Level

Your AQL (considering Pa = 95%) is 0.2%.
Let me say that your plan is very "strange".

Cristiano

Last edited by a moderator:
1

#### 1winnyAK

Christino,

Thank you! but how do I figure that out. like how you did it? ..does that also include in lot size 25 too that my aql is 0.2?
My AQL is 0.2 but i do not see that anywhere in that ANSI-Z1.4 Chart. So what do I use to determine my sample Size for each lot?

Last edited by a moderator:
C

#### CristianoP

Thank you! but how do I figure that out. like how you did it? ..does that also include in lot size 25 too that my aql is 0.2?
I don't understand, sorry; my English is really poor.

My AQL is 0.2 but i do not see that anywhere in that ANSI-Z1.4 Chart. So what do I use to determine my sample Size for each lot?
I can only suggest to use the math.
For the single sampling plan, the math is very easy. A good starting point:
http://www.itl.nist.gov/div898/handbook/pmc/section2/pmc232.htm

Cristiano

1

#### 1winnyAK

My AQL would be on my current process right? its not what I want it to be is that correct?

C

#### CristianoP

When you came up with AQL 0.2 is that for all lots that are from 1-500.
For example I sample size lot 25 my AQL be 0.02 as well.
No, I calculated 0.02% for lot size = 500, sample size = 25 and acceptance number = 0.
When, for example, the lot size is 40 and you take 2 samples (5% of 40), your AQL is 2.5%.

Cristiano