Uncertainty (Top Down Method)! Auditor argues that this is not enough

[t.PoN]

In our lab, we use the replicate of internal quality control sample to build a control chart. The control chart include all analyst, different days, different times.

Our uncertainty calculation is based on control chart limits, because once you build the chart, you had all the variation from all measured including equipment uncertainty in that chart.

Now some auditor argues that this is not enough, and we should use the control chart as a indicator to repeatability and then add the uncertainty from equipment calibration certificates to build the uncertainty budget?

Is he right? Or am I right?

 

[Mikishots]

Re: Uncertainty (Top Down Method)!

Did he really say "should use the control chart" or did he say you must? Is he intending to raise a nonconformance if you don't do what he asks?

 

[t.PoN]

Re: Uncertainty (Top Down Method)!

he intends to raise nonconformity because I'm only using the control chart to estimate the uncertainty and I don't include the calibration certificates in the the uncertainty. but I refuse to accept the nonconformity because all variation in my tests are included in the control chart

 

[BizPro]

Re: Uncertainty (Top Down Method)!

When developing your uncertainty budget you need to include the uncertainty of the standards used. (I think this is what you were asking?) You also need to include the repeatability or test validation results. The test validation is performing the test per your approved test procedure, under the same conditions, at a minimum of 30 times. You need to include all factors that have affect on the test (ie: temperature) to determine your uncertainty of your test. This is the K=2 coverage factor to show that you are 95% confident in your test/calibration. This is industry standard for in-lab calibrations. Environment plays a BIG role in uncertainty and often it is not possible to achieve 95%...maybe only 94% or 92% (depending on where you are at).

I hope I answered your question.

 

[tacom]

Re: Uncertainty (Top Down Method)!

Dear t.Pon

if your bias is under control (if you manage to hold it unimportant level) the precision from control chart will be the major compenent of your measurement uncartainty.
for top down approach you don't have to use uncertainty component coming from calibration certificate . what you want to do is to find your bias and precision (within laboratory or intermediate ) and combine them . if your bias is negligible besides precison then precision will be the main component on your uncertainty.

tacom

 

[Hershal]

Based on the description, the auditor seems to be suggesting that you can use the control chart for your Type A contributions. If Type A is properly calculated, then that would be an acceptable approach.

However, as Bizpro pointed out, the Type B must be specifically considered. If not independently broken out, then the chart MAY actually cover the Type B, but there is insufficient evidence to determine that.

I would break out the Type B independently and resubmit to the auditor.

 

[Randy]

Regardless of what he said, if he went any farther than to determine effectiveness/conformity he's crossed the line. It's not his place to say should, would or anything else other than you have a way, it works and conforms to whatever the requirements are yes or no....That's it!

I don't care if you have 1 or 100 as long as you get what you're supposed to get and you can do something with whatever the information is

Tell the auditor to stifle himself unless he can show you the shall or where your method is not effective....And he better have objective evidence as to why

 

[tacom]

hi everybody
I think
Top down approach (emprical approach) based on intralaboratory method validation or verification studies doesn?t look uncertainty components as type ?A? or type ?B?. According to this approach you should find systematic error (bias or recovery) and precision (called intermediate or within laboratory precision).
İf the within laboratory precision includes all variation expected test method ( different days, different peoples, different glassware, different chemicals, diffrenent calibration, and so on) İt means that type B uncertanity is already included by within laboratory precision and there is no need to handle separetly.

tacom

 

[Bill McNeese]

t.PoN,

Your method is excellent. You are running a control on a regular basis using different operators, conditions, etc. As you said and has been said, you are accounting for all the sources of variation. This method gives the best accurate estimates of accuracy (from the X chart) and precision (from the mR chart) because you include all the variation.

Plus you are continually monitoring the test method for any out of control points. Far too often, we calibrate when there is no need to - the control chart will tell you when you need to consider it (among other possible reasons for out of control points).

To me, you are doing what is best for your process. And the uncertainty is already built it because of the way you are monitoring the test method.

 

[JAltmann]

It sounds as if you have not included the measuring equipment's uncertainty into your budgets. If that is the case the auditor is correct and it should be added. Simply taking measurements and plotting them on the control chart is really only showing you repeatability and possibly reproducibility (if using multiple operators).