Sampling Plans - Guidance to Sampling a Process during Validation

[v9991]

Well, i have to validate a powder filling in hard capsules, so i took 10 full capsules every 15 minutes, but i just ask my boss why and he dosen't know why, that's why i ask you

I haven't seen a specific principle/formula being applied to justify the sampling-qty/plan proposed for the given process. It might be little simpler if the machine was dedicated to a given product. but, often the machine has to be setup for different products.


Verification of capsule from each station is part of machine setup. and usually samples from one-complete cycle is verified while operating the machine at optimum speed.

Now, the question is....""at what frequency the samples have to be verified"" ... which depends on the precision of the machine and operating parameters for that specific product.

for eg., if fill-weight or speed-rpm is lower/higher side (when compared to machine fill-weight/speed window), then we might be little more cautious; if fill-weight/rpm is in middle, then we might be little relaxed.

the common physical parameters would be evaluated; but at top-middle and bottom layers of hopper, would be evaluated for content uniformity..

Hope that helps.

 

[Roberticus]

and this is in the FDA forum. AQL sampling plans are NOT appropriate for process validation or product validation.

Steven and Barbara are correct that we need to undertand the severity of a failure, the probability of a failure (or the standard deviation of the process) how many failures are 'acceptable' and then the Alpha and Beta risks we are willing to take.

So, for the OP if you want specific advice you will need to give us a bit more information on your specific situation...this is not a cookie cutter one size fits all situation. And of course your FDA reviewers will have their own 'favorite' methods...

Hello, I've read several threads on this topic over the past hour or so, and downloaded two useful attachments for sample sizing , one previously posted by yourself Bev D. I'm planning to move forward tomorrow armed with these formulas, which are broken down far enough for me to confidently reproduce.to meet the needs of our Customer's pending process validations.

I'm curious about the "Type I and II", or "Alpha and Beta" risk definitions. Can I get a good lead on where to get better definition of these terms? Is it a Minitab thing? our Customer and their Customer process a risk-based algorithm in Minitab to determine valid statistical rationale, but I think the formulas from these attachments may be enough to get me where I need . . .

 

[Ajit Basrur]

NIST's Engineering Statistics Handbook is a good start - http://www.itl.nist.gov/div898/handbook/pmc/section2/pmc22.htm#Type

Type I Error (Producer's Risk): This is the probability, for a given (n,c) sampling plan, of rejecting a lot that has a defect level equal to the AQL. The producer suffers when this occurs, because a lot with acceptable quality was rejected. The symbol is commonly used for the Type I error and typical values for range from 0.2 to 0.01.

Type II Error (Consumer's Risk): This is the probability, for a given (n,c) sampling plan, of accepting a lot with a defect level equal to the LTPD (lot tolerance percent defective). The consumer suffers when this occurs, because a lot with unacceptable quality was accepted. The symbol is commonly used for the Type II error and typical values range from 0.2 to 0.01.