28 replies to this topic

[Lelouch_rayne]

Hai guys,

 

I have another concern on the new version of BRC. In clause 6.4.3 Reference equipment, it says that that "The uncertainty of calibration shall be considered when equipment is used to assess critical limits." I tried making one for our CCP's, I used the data from our calibration certificate which says, The estimated uncertainty of the measured temperature is equal to ±1.4 C. So, i assumed, that was the BRC clause is talking about. Meaning, the recorded value from actual monitoring has a correction factor of ±1.4 C. The true value might be +1.4 or -1.4.

 

But then again, our TSD Head said, That it was wrong, i should be using the correction value used during calibration indicated in the certificate not the uncertainty measurement. I mean, how come?..Is she right? or i'm just dumb...

 

Thanks,

 

Joan

[Charles.C]

Hi Joan,

 

So what was the value of the "correction value during calibration" ?

 

IMO probably Yes and No.

 

I assume BRC mean calibration error. This is usually either a positive or negative value for which the user designates an acceptable limit.

IMEX the +/- data reflects the accuracy of the calibration result.

 

See this document -

 

[attachment=7044:measurement good practice guide.pdf

Attached Files

•  measurement good practice guide.pdf   428.26KB   318 downloads

[Lelouch_rayne]

Hai charles,

 

I already read that pdf, I just want someone to confirm that i'm doing the right thing.

But i still don't get it. Should i used the uncertainty or the correction factor? 

 

I badly need a proof so that i can defend it in front of my boss.

 

Thanks,

 

Joan

[Charles.C]

Hi Joan,

 

did you see the Yes ?

 

I think yr "Head" is correct. That's why i asked you for the correction value. I was curious as to the value of the reference being checked and the error reported.

 

For example if a thermometer it might be checked at 100degC, or 0 deg C, or any other value with a reported error for each level.

 

PS - IMEX there are typically 3 data on the certificate -

 

(1) the value of the standard reference level , eg 100 degC

(2) the value shown by the instrument under test, eg 101.0 degC

(3) the estimated accuracy (uncertainty) for the result in (2) eg +/- XdegC

 

The calibration error in above would be 1.0degC, ie the instrument is reading 1.0degC high.

 

But yr quoted +/- 1.4 degC looked a bit odd so i wondered if yr certificate is presented in a different way ??

[Charles.C]

Hi Joan,

 

Have a look at this link which should make it clear -

 

http://www.teltherm....ertificate.html

[Lelouch_rayne]

Thanks for the link, I get your point but our certificate is presented in an odd way, So i think thats the reason i'm confused. Please try to look at this sample. The calibrator calibrate the equipment using different points. But our expected reading should be between 139 - 143 Celsius. Unlike your sample, Only one value of correction factor is given, but in our certificate is a lot. Should I get average?

 

Thanks,

Joan

 

 

Attached Files

•  cert..jpeg   215.84KB   3 downloads

[Charles.C]

Hi Joan,

 

Thanks for the pic.

 

The certificate looks usual style to me with the instrument checked over the temp. range 28 – 150 degC. (IMEX the points are selected by the customer).

Traceability looks a bit weird but maybe normal for Philippines. The +/-  value looks maybe a bit high but then the temperature goes quite high also.

 

Unfortunately only the last data point (150degC) closely relates to yr application. Do you have any previous calibration data to compare ?

If not, it may depend on what you are doing with the thermometer, ie how important the function, but if you trust the lab then the correction data (-0.3) at 150 point would be OK for me unless producing gold bricks. The only sure way to know more is to re-calibrate which I doubt possible.

 

Hope that helps.

[IzzyP]

Hi Mary

 

the way I have interpreted the clause is as follows.

 

Lets say you have a pasteuriser , and that the uncertainty of measurement of the calibrated probe is +/- 1.4 degrees , then if legal definition of pasteurisation for divert 71.7, then you need to set the divert temperature at 73.1 (minimum)

 

Hope this helps

[trubertq]

I agree with Izzy... 

[Charles.C]

Hi TBQ/Izzy

 

I guess it may depend on what the OP is doing.

[Lelouch_rayne]

hi Charles,

 

I finally get your point, so the easiest way is, I should just pick the range (let's say our CCP temp is between 139 - 143, therefore,150 Celsius range with a correction value of -0.3 should I be using) which is closer to our expected value. Am i right?

 

Thanks,

 

Joan

[Lelouch_rayne]

hi Izzy,

 

I'm talking about the monitoring. For instance, during production, the operator would monitor the reading of temperature. During recording he/she should apply the correction factor given by the calibration certificate. I'm confused if what value should i be using. Please see attached picture at the thread. Can u enlighten me a bit more.

 

Thanks,

 

Joan

[Charles.C]

Hi Joan,

 

It looks like my contributions may be confusing you. Sorry about that.

 

maybe better to ask your TSD Head to select their preferred continuation. No chance of (personal) error for yr specific process.

[Tony-C]

Hi Joan,

 

Maybe I have got this wrong but looking at your certificate I conclude this:

 

At your operating temperature you have a Correction of -0.3 °C so in the first instance you should take this into consideration when deciding on minimum temperature settings. Secondly you have an uncertainty of +/- 0.49 °C. Since we are concerned with a minimum process temperature your process temperature setting should be adjusted by 0.3 + 0.49 (0.79) so a minimum divert setting of 140.79 °C if your minimum process temperature is 140.0 °C but I always like to have a bit more of a 'safety' margin.

 

BRC Interpretation Guideline 6.4.3
Where equipment is used to measure or monitor a critical limit then it is important that the tolerance (or uncertainty) of the calibration is considered. For example, if a thermometer is required to measure a critical limit of 72°C, but the calibration shows an uncertainty of 0.5°C, then it must be noted that a thermometer reading of exactly 72°C could in fact represent a true temperature of anywhere between 71.5 and 72.5°C (i.e. the temperature could be lower than the critical limit). In this situation, to guarantee a minimum temperature of 72°C it would be necessary for the reading on the thermometer to always be 72.5°C or higher.

 

Kind regards,

 

Tony
 

[Charles.C]

Hi Tony,

 

Thanks and yes, I agree yr calculation but (I think) the OP appeared to be being instructed to ignore the second factor. Reason ( specific application ?) unknown. Maybe no reason / unaware of the BRC text / no Guidance.

 

PS - the exact interpretation of the Certificate "Uncertainty" factor seems sometimes variable but yr post is aligned to the link i previously offered so here's hoping.

 

(For example -

 

What is meaning of "uncertainty“ term on the calibration certificate?
This term expresses the calibration uncertainty. It is not to measurement error or uncertainty of the device! The quality of the calibration depends on several independent things that are have different probabilities of occurrence (repeatability, uncertainty of standards, homogeneity of the temperature in the chambers, the resolution of the measuring instruments, the parasitic temperature dependences etc.). Using statistical methods are quantified these various effects and shall be included in resulting uncertainty, that is mentioned on the calibration certificate.

 

[Lelouch_rayne]

Hi Tony,

 

That's a bit confusing,.. Can I just used one factor, it's a little bit redundant to use both factor. Since your uncertainty may be based on your correction factor used during calibration. Correct me if I'm wrong 

 

Best Regards,

 

Joan

 

[Tony-C]

 

Hi Tony,

 

That's a bit confusing,.. Can I just used one factor, it's a little bit redundant to use both factor. Since your uncertainty may be based on your correction factor used during calibration. Correct me if I'm wrong 

 

Best Regards,

 

Joan

 

 

 

Hi Joan,

 

I can see why people find it confusing

 

There are two factors here, the adjustment on the reading from the probe to the absolute correct temperature during the calibration - 0.3 and then the uncertainty of the result +/- 0.49. Since the minimum process temperature is required you add both to ensure that you are hitting your required minimum process temperature.

 

Kind regards,

 

Tony

[Charles.C]

Hi Joan,

 

Just to be sure. I assume yr query in post 16 referred to the calculation of the critical limit and not the question you posed in yr post 12. Correct ?

 

I also assume that the thermometer being discussed is a permanently fixed unit attached to some kind of process equipment as compared to a thermometer stored in the lab and used for a variety of purposes / different equipments. Correct ? 

[Charles.C]

This is an addendum to my previous post -

 

After some thought I’m inclined to the conclusion that Critical Limits as nominally stated assume  calibrated instruments, ie are to be used with "calibrated data".

 

Referring to the thermometer in this thread, both the factors (-0.3 and +/- 0.49) are related to the thermometer, so that an indicated value of 140.6 degC is best expressed as a (calibrated) result something like –

 

The “true value” is 140.3 +/- 0.49 degC with 95% confidence

 

In a process scenario as applicable to BRC 6.4.3,  it is regarded as necessary to assume a worst case scenario related to the (permanent)  factor 0.49 which is compensated for by increasing the theoretical critical limit of, say, 140degC to a minimum of 140.49degC. (as per Izzy/Tony).

 

Assuming the use of calibrated data  the Critical Limit would then be a minimum of 140.49 degC

 

However in practice for a dedicated thermometer/machine, IMEX machine operators prefer to act as per the thermometer’s indicated value which would necessitate an action limit of 140.79 degC (as per Tony’s formula). (digital offsets / re-settable analog units would also work if available)

 

Again, if the thermometer is in general random use, IMEX a calibration correction factor as appropriate to the application (cf. -0.3) is simply, directly noted on the instrument. For some usages the uncertainty factor IMEX is undoubtedly simply (wrongly ?) ignored.

 

So IMO the preferred action regarding factor -0.3 depends on the situation. Either way, it’s used.

Plus one certain thing from BRC’s POV (6.4.3) is that the uncertainty factor +/- 0.49 must be used.

 

So IMO the answer to post 16 is regretfully No.

 

PS - texts suggest that the quoted uncertainty factor is derived within a lab's highly controlled environment and that values in  practical situations can be substantially higher.

 

PPS - this is not a simple subject.

[Lelouch_rayne]

Hi charles,

 

Post 12 and 16 are related. The point why i'm asking what factor should i be used, its because the operator who monitor the temperature (in a timely basis) would immediately apply the correction factor (every time he records the results) to consider the uncertainty of measurement.

 

And yes, the thermometer is permanently fixed.

 

Thanks,

 

Joan 

[Lelouch_rayne]

Hi Charles,

 

Your post is something.  I'm still trying to comprehend it. And God!!, I think I'm surrounded by highly intellectual human being that i can't follow anymore the clash of minds. 

 

So, back to the topic, now i remember something. I think the certificate i posted is incomplete compared to other certificates. There are some certificates which indicate the correction factor after and before the calibration. For instance during calibration, the actual measurement is 139 C.  But your expected value should be 140 c.

 

Meaning you have a correction value of 1 Celsius.

 

Sometimes, there are calibrator who correct this error by eliminating the correction value after calibration. So that, the next time you measured, the recorded value would be certain. But you should still apply the measurement of uncertainty which is +/- 0.49 C. 

 

I also have this kind of certificate with a column indicating the correction value (before adjust and after adjust). Which make my job easier. 

 

So, thanks a lot for all the info, my problem right now is how to convince my boss about this issue.  

 

PS. Sorry if my statement is a little bit confusing.

 

Best regards,

 

Joan 

[Lelouch_rayne]

And based on Tony's post on:

 

BRC Interpretation Guideline 6.4.3

 

I Think what i did on my previous report is in line with the interpretation, I only consider the uncertainty of measurement considering that the calibrator adjusted the device to its expected value and all correction factors had been zero out.

 

So, every time we check the device the true value would be between 140 C +/-0.49. So its either 139.51 or 140.49.

 

Best Regards,

 

Joan

[Tony-C]

Hi Joan,

 

If the calibrator adjusted (corrected) the device then I would expect to see the adjusted readings on the certificate.

 

So I'm still of the view as per post 14 that you apply both the correction factor and uncertainty value

 

Kind regards,

 

Tony

[Charles.C]

Hi Joan,

 

Thks yr several posts,

 

Yes, I agree that calibration certificates vary in their style of presentation. Some of them require a detective.

 

And yes, as you state, some customers request the external lab to reset their thermometer (if it has the capability) so as to pre-calibrate it. This can give a zero correction factor for the user at a specific temperature. Other customers prefer to avoid tampering by unknown people and choose to implement  the correction factor  themself.

 

As per Tony's comment, any zero-ing  should be evident on the Certificate which in posted example seems not.

 

Just for interest and at the risk of more confusion here is a UKAS detailed illustration of the calculation difference of  “uncertainty factor” for 2 thermometers which are respectively categorized as either “reference” or “working units”, see pgs 10 – 13. I deduce your instrument would be as the second category although the type of calibration process in actual use may not be obvious on the certificate. AFAIK, there are a number of different procedures in commercial use ranging from temperature-controlled baths to electronic black boxes. (The conclusion is the U number.)

 

 use of thermometers in UKAS accredited labs.pdf   133.31KB   75 downloads

 

There seems to be no comment regarding “uncertainty factor” in BRC6 so my suspicion is that BRC have found that some Companies have been simply ignoring it. (After all some "quality"  thermometers [see UKAS above] only have a result around 0.05-0.1 degC at the ice point which would IMO be rather negligible for many routine purposes).

 

IMEX Process / QA people are trying to avoid all the above “confusion” and simply add the uncertainty factor in the Certificate to the theoretical critical limit as per Izzy/Tony and as (now) required by BRC 6.4.3.

 

Re yr comment -

So, every time we check the device the true value would be between 140 C +/-0.49. So its either 139.51 or 140.49

 

Actually 95% of the (assumed “many”) results are predicted to be in the range 139.51 - 140.49 degC.  5% are predicted to not be.

(Crudely, can say that there is a 95% probability that the true value lies between 139.51 and 140.49degC although classical statisticians may protest)

 

Probability can be tricky, and cunning.

Edited by Charles.C, 22 November 2015 - 06:36 AM.
amended

[Tony-C]

There seems to be no comment regarding “uncertainty factor” in BRC6 so my suspicion is that BRC have found that some Companies have been simply ignoring it. (After all some "quality"  thermometers [see UKAS above] only have a result around 0.05-0.1 degC at the ice point which would IMO be rather negligible for many routine purposes).

 

 

Hi Charles,

 

BRC Food Issue 6 and Ice Point seem a bit irrelevant?

 

Calibration should be in the practical temperature ranges of the equipment but particularly the critical temperatures (CCPs).

 

Kind regards,

 

Tony