Is Test Equipment Accuracy an Illusion?
Posted by Art Gatenby on Fri, Oct 24, 2008 @ 11:05 AM
On a regular basis I’m asked the question; “How accurate is this moisture balance?”
The questioner asks for the answer in terms of percent moisture. When I answer “it depends”, I always get the unspoken response. “Why did I ask this stupid idiot?”
The detailed answer is complicated because the questioners want to know accuracy in terms of a percentage (%). This is a relative rather than an absolute measurement term. In the case of a Loss on Drying (LOD) moisture instrument it depends on the size of the sample.
The detailed answer can be understood with this example of using the LOD method. The percent moisture (%) is calculated buy subtracting the weight of sample at the end of the test from the weight of sample at the beginning of the test. We call that the difference. The last part of the calculations is to divide this difference by the beginning weight..
Given a sample that has 23.5% moisture the following table shows the implications of sample size. In this application (LOD) an electronic moisture balance is normally used. Parts of this instrument specification are sensitivity and repeatability (often referred to as accuracy). In this example, we assume a sensitivity of 0.001 grams and a repeatability of ± .005 grams.
Effects on Percent % Accuracy of Changes in Sample Size
Using a Ten (10) gram sample the results are as follows: |
Beginning Weight Ending Weight Difference % Moisture |
In the Middle of Range 10.000g |
7.650g |
2.350g |
23.50% |
Start Wt. High – End Wt. Low 10.005g |
7.645g |
2.360g |
23.59% |
Start Wt. Low – end Wt. High 9.995g |
7.655g |
2.340g |
23.41% |
Difference between High & Low = 0.18% |
Using a five (5) gram sample the results are as follows: |
Beginning Weight Ending Weight Difference % Moisture |
In the Middle of Range 5.000g |
3.820g |
1.180g |
23.50% |
Start Wt. High – End Wt. Low 5.005g |
3.815g |
1.190g |
23.78% |
Start Wt. Low – end Wt. High 4.995g |
3.825g |
1.170g |
23.42% |
Difference between High & Low = 0.35% |
Using a One (1) gram sample the results are as follows: |
Beginning Weight Ending Weight Difference % Moisture |
In the Middle of Range 1.000g |
0.765g |
0.235g |
23.50% |
Start Wt. High – End Wt. Low 1.005g |
0.760g |
0.245g |
24.38% |
Start Wt. Low – end Wt. High 0.995g |
0.770g |
0.225g |
22.61% |
Difference between High & Low 1.17% |
From this illustration it should be clear that one of the illusions of test equipment accuracy is caused by the size of the sample. It this case the accuracy in percent moisture ranges 0.18% (plus or minus 0.09%) to 1.17% (plus or minus 0.56%) This shows that the accuracy of moisture balance (% Moisture Error)in this example can be improved six and one half (6.5) times (6.5) by changing simple size from one (10 gram to ten (10) grams.
There are other implications of sample size in the use and selection use and selection of test equipment. I am sure that these will be the subject of future ramblings.
I’ll be here again next week.
Art