Discussion on the representativeness of analytical sampling
Thread Content
Dear all: I have a question regarding intermittent production systems – once the product tank has received a certain amount, it starts to be sealed and no further feeding takes place. An analysis of the product in the sampling tank is conducted; once the analysis shows it is satisfactory, filling of the IBCs begins. One of the IBCs is set aside as a sample, but no analysis is performed on it. Is there anything wrong with this analysis sampling pattern? Today, the external audit experts said that according to GB-T 6678-2003, samples must be taken from the IBC tanks for analysis. So, if we fill 100 IBC tanks in one batch, how many samples need to be taken? If the analysis shows that the samples are not up to standard, then all 100 IBC tanks will have to be reprocessed, and it will also be impossible to arrange vehicles for loading them.
The expert explained it clearly. Do it according to the specifications. So, in addition to the outer tank of the sampling bucket, simply add buckets for the first tank, the middle tank, and the last tank, and that’s it.
If there are guidelines, follow them; if not, consult some examples on sampling in analytical chemistry.
It’s not said that way; what’s discussed here is how the sampling unit is determined. I think it essentially comes down to the sample needing to meet the requirements for analysis and also to be representative. We took samples from the tank for analysis; I consider them representative, with one sampling unit per sample. We can also take samples from the IBCs, but when filling them, the delivery vehicles wait in the factory at the same time. If the samples taken from the IBCs are not satisfactory, it means the delivery vehicles have to return to the shipping company, and it’s also troublesome to reprocess all those IBCs; it’s not a practical approach for the factory. So what we do now is take samples from the tanks for analysis; samples are kept when filling the IBC tanks, but no analysis is conducted on them. Is this in compliance with the requirements of GB/T 6678?
It is generally related to the uniformity of the sample； If it is divided into 100 IBC tanks, based on sampling units, at least samples from 14 tanks need to be taken. If you want to take samples from the tank, make sure that the samples taken from the tank are representative and uniform, and that each barrel is also clean and free from contamination; this will allow for direct sampling from the tank.