Hello everyone,
I have a small problem - and I am not a statistician, which in itself is another problem! The RSS suggested you may be able to help.
I need to determine the number of samples I need (or should take).
A filling line is filling bottles. It is a batch process, the batch size is 90,000, the product being filled is a true solution and is fed entirely from a singe tank throught the batch.
I need to demonstrate that the batch is homogenious -i.e. that one of the components in the mixture being filled is at the same concentration throughout the batch) I have been told to take samples at start, end and at two other points within the batch and send all samples to the analytical laboratories.
Rather than having to demonstrate that the samples are representative of the batch (which probably involves sampling some 400 bottles per sample period?), I think it is sufficient to be able to be able to demonstrate that there is no significant difference between the four sets of samples. I hope that this will allow a smaller number of samples to be taken and thus not swamp the laboratories.
How many samples would I need to take at a time?
The machine fills 6 bottles at a time (it has 6 filling heads) - but all the heads are fed from a common manifold and so I am not looking to be able to demonstrate that all the filling heads are equivalent. At present all samples to be taken are to be placed in a single tray and not segregated.
I realise that I have probably not provided all the information needed, but hope that I have given the basis details sufficiently.
Any help appreciated, Ideally a sample number and a two liner justifying the number - however even just a general pointer as to the techniques I should consider looking at would be a great help.
Many thanks
Trefor Jones ([log in to unmask])
|