|
| | | Supreme Being
Group: Forum Members
Last Login: Today @ 9:37:52 AM
Posts: 111, Visits: 673 |
| | I would like to know what everybody is doing as far as BOD sample duplicates. We are required by the great state of Texas to run a BOD duplicate per 10 samples. Currently we satisfy this requirement by the current method: If we run five dilutions of a sample, we will duplicate one (or more) of the dilutions and call this a duplicate: for example we would use 1mL, 3mL, 3mL dup., 5mL, 5mL dup., 10mL, 15mL. We don't include the duplicates in the average for the BOD calculation, and if the RPD is out of range we throw that number out. This seems incorrect for several reasons but I want to stay focused: I think a true BOD duplicate would involve taking the sample from the original container and putting the sample and sample dup through the same process and running the same dilutions on the sample and sample dup (1,3,5,10,15 for the sample, average the result for the BOD value, do the same for the duplicate). This is the logic used for every other analysis we do. I'm not sure how we arrived at the current BOD duplicate ideology. Any thoughts? FYI we do analyze GGa samples in duplicate and our standard deviation for the results is steady around 5% over the last few years. |
| |
| | | Supreme Being
Group: Forum Members
Last Login: Today @ 1:37:46 AM
Posts: 231, Visits: 927 |
| I agree with you, Stressed O., that the current process is bogus. The average of the 3-mL and 5-mL dilutions could be quite different than the average of all dilutions because of the nature of the sample, and have nothing to do with analytical precision. But more importantly, by definition (e.g., Standard Methods 1020B.7, 20the Ed.) a duplicate is analyzing a "...second portion of sample...processed in the same way as the [first]... (emphasis added). For the purposes of this discussion, that would mean putting both aliquots through identical dilutions.
It sounds like you withdraw one "double-size" aliquot from your sampling container, split it, and analyze the two sub-aliquots. In my opinion, withdrawing the aliquot from the sampling container is an analytical step as opposed to a sampling step. You might want to consider withdrawing each aliquot separately so that the imprecision of doing that step will be included in the final results for both aliquots. One might think this is trivial, but in a lab I assessed a few years back, BOD results had suddenly doubled without there having been any change in the waste treatment process. The plant manager suspected intentional manipulation of data and called me in to investigate. To make a long story short, it turned out that the compositor collection container had a spigot on the bottom for withdrawal of sub-samples. Normally the analyst would vigorously shake the carboy and immediately withdraw the sub-sample. Coincident with the BOD (and TSS) results doubling, he had injured his shoulder and could barely lift the carboy, let alone shake it. And as one might expect, all the high BOD/TSS material was on the bottom of the carboy. My point? Withdrawal of the sub-aliquot is an important step that can have considerable influence on within-batch analytical precision and needs to be included as part of the duplicated process.
Perry |
| |
|
|