obviously this issue falls aside from the classical pharmaceutical approach using drug potency and batch size to calculate limits, so in my opinion you are basically free to set your own and achievable limits. Some thoughts:
20 ppm sounds reasonable, although probably 10 ppm is more common. Nevertheless, one is as arbitrary as the other.
Is there any product specification that could be used to set or to back this limits? IMS is at least as volatile as IPA, so if only used for cleaning purposes, a small cross contamination would probably not pose a high risk. Perhaps the 70/30 is sold as 99,99% pure allowing for a higher carryover.
20 ppm is equivalent to a residue of 100 ml IMS in the 5000 l tank, or roughly 80 g IMS, equivalent to 1,7 mol or 37 l, in gaseous state. Be sure that all the IMS is eliminated by the rinses, even the evaporated IMS, which will not be detectable in the 60 l of the last rinse but will probably be present in the next batch when the tank is completely filled.
Assuming that the product is filtered, be sure to pass the rinses through the filter and the filling line. Any contamination there will be carried to the first bottles of the next batch.
The problem that I have is that we don’t have a defined process at the moment for this activity. Therefore, I’m trying to establish a process which I can actually validate.
What you’ve said makes good sense. I will be splitting this work into two seperate validations.
The first validation will concern itself with the actual tank rinsing / cleaning to remove the residue from the previous batch (20ppm).
The second stage of this work will be to get 20ppm at the filling machine , through the transfer pipeline, filters and filling machine itself.
again, and without deep knowledge of your equipment, only thoughts:
If you separate the two parts of the production train in two separate validations, get sure the acceptance criterion is applied to the sum of both contaminations. ppm is a concentration unit and the contaminations sum up.
Get sure the sprayball pattern covers the whole of the tank. Normal practice is to spray a riboflavin solution through the sprayball and observe the spray pattern on the tank walls with an UV lamp, but in your special case i wouldn’t recommend this practice.
Is the sampling point the lowest point of the tank outlet (the point the recirculating pump takes the liquid from)? If not you will get samples with different proportions of alcohol trapped in lower points of the tank.
Probably (IMHO) the residual contamination does not come from the tank walls but from the evaporated alcohol contained in the tank atmosphere. As stated in a post before, 20 ppm is equivalent to roughly 40 l of gaseous alcohol in the 5000-l-volume. Unless the tank is swept with a fine mist of water, this alcohol remains in the tank atmosphere and is only removed by solubilization in the 5x 60 l of water, possibly depending on contact time and rinsing water particle size, which would explain the spikes of alcohol found is successive rinsings. Try filling the tank completely with WFI. If you found 400 ppm (or 400 µl/l) in your 5x60=300 l rinse, in your tank were initially 400 x 300=120000 µl of alcohol, which in 5000 l are 120000µl/5000l= 24 µl/l= 24ppm, which is perfectly measurable. Executing this test, make sure the tank is slowly filled and thoroughly mixed, preferably with the cleaning system (recirculating pump and sprayball) working, in order to solubilize the atmospheric alcohol in the water and not simply replacing the alcohol loaden atmosphere with WFI.
That does make sense, and I hadn’t thought of that at all.
I will need to validate this idea, but I believe that this could be a significant contributing factor in the results.
The next step will be to plan some experiment time to prove or disprove this theory.
If the vapour is contributing to the results then I can’t see any practical way of avoiding this issue.