19 replies to this topic

[michela78]

How we can determine the frequency calibration of the equipments? How to perform a risk assesment?

Many thanks in advance for your precious collaborations

[johnmcmaster]

There is a difference between external calibration and internal verification.

We annually calibrate metal detectors, scales, weighheads (we are a bakery) and are verified weekly and monthly.

In my risk assessment I put it that the thermometers would be either working or not working as a guideline rather than a sliding scale of say being more than 2 degrees centrigrade out. This is done monthly after 5 years of weekly checks. The records showed that the risk of the probe being found to be out at the weekly check was low. I have found it easier and cheaper to buy a new master probe with calibration certificate each year (just before BRC audit) than to send off to get an old one calibrated.

The staff will be using the equipment and would not be able to tell if the probes are working & reading properly, other than on or off.

The scales are checked weekly and the weighheads 3 monthly as these are not a food safety risk of they are not functioning.

The metal detectors are checked at start up, hourly & product changeover and at the end of the run with 3mm ferrous, 3mm non ferrous and 4mm stainless steel test pieces.

Hope this is of help

[tsmith7858]

Calibration should be determined by the history of the tools/machines being calibrated along with the potential for failure and the risks associated with the failure. Unfortunately that means it is going to be different for everyone and everything.

Typically you would start at a higher frequency and reduce after enough conclusive evidence that the equipment is not going out of calibration.
Some manufacturers may recommend a calibration frequency.
If you conduct in house verifications, you may not need as many formal calibrations.

Bottom line is that it is up to you to determine the criteria. A few questions may be:

• How often do you want to put youself (or company) at risk?
• What are the costs if something is out of calibration (recall back to last test can be expensive if you only test once a year)?
• What are the costs of the calibrations (if they cost more than the resulting failure they might not be worth it)?
• Who is at risk if you find a failure and what are their risks (injury, illness, death)?
• Are there any regulations where you produce or ship your product?

[Charles.C]

Dear michela,

How we can determine the frequency calibration of the equipments? How to perform a risk assesment?

I hv to agree with tsmith. It very much depends on what equipment you are talking about, what you are using it for etc, etc.

@johnmcmaster -

In my risk assessment I put it that the thermometers would be either working or not working as a guideline

Surely this is meaningless without defining/measuring "working" ??
Again, it obviously depends on what yr temperature measuremnt is related to but I am surprised that yr BRC auditors are readily satisfied over what i wud hv guessed are far less exhaustive records than usual in their experience. On the other hand, perhaps this is (somehow) standard practice in the bakery area.
Personally, I hv found that it is quite difficult to maintain thermocouples to +/-1degC at a level of 100degC (a typical requirement) without monthly checking and not-infrequent adjustment of the zero. (this is perhaps not too surprising when you read the typical specifications on thermocouple instruments although I guess it may also depend on yr willingness to invest ). Moreover for the thermocouples used to control a CCP type cooking step, IMEX, yr procedure (if I understand it correctly) wud not pass the BRC auditors without unfavourable comment, regardless of its risk based logic.

If one has unlimited financial reserves, It is certainly convenient to outsource routine calibration of weighing equipment but I hv sadly never experienced that degree of good fortune. Carrying 20kg checkweights around can be quite boring after a while.

Rgds / Charles.C

[cdelmundo]

you can apply process capability studies to determine frequency of calibration.

[Tony-C]

In my risk assessment I put it that the thermometers would be either working or not working as a guideline rather than a sliding scale of say being more than 2 degrees centrigrade out.

I have to agree with Charles C on this. It depends on your application and safety margins but I would also say 2degC is a generous margin. I would normally work to +/- 0.5degC and build this into my process so the working specification would have at least a 0.5degC safety margin. I have seen many thermometers that were "working" which were giving a false reading.

Regards,

Tony

[Charles.C]

Dear Michela ,

I did a little more googling on yr general query and obtained some (possibly) useful further information for a specific case but which offers a semi-generic applicability.
The 1st link below details a (matrix based) risk assessment process for calibration of acoustic measuring equipment. It quite well illustrates the general factors which might be involved IMO. The attached worked example utilises another (linked) document to set appropriate calibration intervals but the link is not working so you will hv to access from my uploaded document. The latter is apparently often quoted as a source for (various) types of equipment calibration intervals. From a quick look, the periods vary from 3 months to (maybe) 5 years.
http://www.npl.co.uk...ustic-equipment

 LAB23_1.pdf   60.3KB   947 downloads

The above procedure is what one might call a “high-tech” approach. It is impressively rigorous (I suppose) but for many routine operations / equipments I’m sure people tend to use “judgement” calls based on their specific situation. It’s rather difficult to say much more without knowing yr particular application. Nonetheless, these more empirical procedures may sometimes be backed up by a mini-calibration which “enhances” the risk assessment. As an example see the (partially extracted) multi-stage procedure on calibration of a balance below where the additional 2nd stage check routine obviously reduces the risk of an incorrect result. This weight control technique is quite often used IMEX, particularly in manual-based manufacturing.

Calibration of the balance, no matter what the use is, is the same: place a
calibrated, certified (i.e. to a traceable standard) wieght on the balance
adjust the balance if necessary. That is the Calibration. Then you need to
ensure (verification) it meets you predetermined specifications. So place a
calibrated, certified (i.e. to a traceable standard) wieght on the balance
range you are weighing in to ensure linearity. Generally this is performed
annually, but depending on the use of the balance, for example weighing at
the extremes of the balance, weighing in aggressive chemical atmospheres
(i.e. increased uncertainties) etc may require more frequent calibration.
Your acceptance criteria will be dictated by the process or analytical
method you are weighing the substance for. It does not matter really where
you perform the calibration but it would be best to perform this within your
own working environment, this is because the uncertainties associated with a
controlled room will not be representative of your actual process
uncertainties.

Then everyday, or before every weighing you would perform a performance
verification. This would involve taking (at least) 2 calibrated, certified
traceable weights that bracket the weight you are going to measure. This
will check the balance is in sound working order. Some will argue that this
is not necessary as the balance is validated so why do you need to check it?
but to perform this daily check takes a minute or so and ensures all is ok,
but again it is your risk assessment.

http://www.pharmweb....3/msg00064.html

Rgds / Charles.C

[Charles.C]

Dear michela,

Since the thread interested me, I hv done a little more searching. The links below may be of some value to people who are interested in this topic. I have put the tougher material first (section1) with the, perhaps, more immediately useful stuff following in section2.

Section1
This topic, when treated rigorously, overlaps into metrology and (IMO) requires quite heavy statistical ability to progress very far. If you wish to see what I mean, just google “estimation calibration intervals., eg here is one extract -

The process of determining calibration intervals is a complex mathematical and statistical process requiring accurate and sufficient data taken during the calibration process.

 Calibration_0.ppt   189KB   758 downloads

This link is another fairly basic (!!) introduction
http://www.qualitydi...ement-equipment

To demonstrate a practical usage of the heavy stuff, this is a link to a comprehensive procedure for pipette calibration which includes an interestingly simple graph for predicting calibration intervals utilising the MBTF (mean time before failure) characteristic. This table may well be generic but my (lack of) statistical ability prevents a yes/no.
http://www.labautope...cy_for_pipettes

section 2

In a more friendly way, one of the other "techniques" I noticed includes a route nicely defined as “engineering intuition” (EI) which I suspect is more useful to readers of this forum.

Some justification for the EI idea also occurs in this quote from another forum -

the calibration frequency you set (barring industry/regulatory/customer requirements) is yours to choose. You will need to assess risk. Basically how risky is it should this item be out of tolerance? The lower the risk, the longer you can go between intervals.

Another description of the idea is introduced (see option No.1) in this extract –

 calibration_1.png   471.64KB   51 downloads

More officially, the principle is apparently also included within the BS EN 30012-1 standard as described in this calibration procedure for industrial weighing systems.
 calibration_3.PDF   1.83MB   660 downloads
(see pg13)

I do believe this semi-intuitive starting procedure will satisfy the typical (food) auditor expectations if backed up by an appropriate risk assessment (ie data ).

A table of (I presume) EI derived “starting” intervals is in this quite readable (except the statistics part) short book from an instrument supplier (see Ch3) with examples of interval shortening / lengthening procedures. However many intervals seemed rather long for (audit) comfort to me and I noticed in another forum that most people recommended setting intervals for maximum 1 yr unless the item was very infrequently used. (Maybe johnmcmaster's earlier logic will pass in some circumstances if one is brave enough. )
 calibration_2.PDF   3.29MB   525 downloads

Rgds / Charles.C

[Charles.C]

Dear Michela,

I hope all this stuff is of some use, somehow.

Yr original post was –

How we can determine the frequency calibration of the equipments? How to perform a risk assesment?

For the second question, I assume you prefer not to do heavy statistical analysis. If so, it is surely easier to start off with a chosen interval as discussed previously. Other typical “risk” parameters hv been well discussed in previous posts. One “intuitive” factor (already suggested earlier) which I used myself so as to include some HACCP type risk perspective is to list equipment as either critical (ie safety related, eg a cooking thermometer) or otherwise (eg weight). This allows you to group equipment / calibration intervals, etc if you wish. It also suggests a (risk related) corrective action where an out-of–tolerance occurs which will apply if you hv to do a SOP. IMEX, for most non-expensive units, disposal is the most usual corrective action. Conversely, for expensive items like digital weighing machines in wet environments, it often pays to invest in the well-recognised brands unless economically impossible.

The general answer to the first question is more or less covered in the preceding posts. A specific answer obviously depends on the actual equipment.

I hv added three more documents to the previous. First is a UN sponsored book for small/medium production enterprises on measurement / calibration which includes some statistics but remained readable to me. Well done IMO. Includes a table of “initial” calibration intervals for starting purposes.for a range of equipment types plus the typical adjustment procedures as already given elsewhere. I think many people simply stick to their first choices as long as they “work”.

 calibration_4_UNIDO.pdf   292.25KB   411 downloads

The 2nd attachment below is a practical discussion on the implementation of calibration methods / intervals to a gas probe sniffer. Seemed like a good illustration of actual usage.

 TN_148_Calibration_Interval.pdf   32.13KB   424 downloads

Thirdly, I extracted the 3 tables of calibration intervals previously presented and put them together to assist comparison. The large differences in opinion tend to reinforce the option to select yr own then validate.

 calibration_tables.doc   132.5KB   567 downloads

Rgds / Charles.C

[Redangel]

Hi!

Who's should be in-charge of calibration, QA or Engineering/ Maintenance?

 

Regards,

Redangel

[Andy_Yellows]

Personally think it should be QA if it isn't a difficult job. Calibration records form part of our quality records

[Charles.C]

Personally think it should be QA if it isn't a difficult job. Calibration records form part of our quality records

 

Hi Andy,

 

I take it that you are not part of the QA team ?  Production maybe ?

[Andy_Yellows]

Charles, what an extraordinary accusation! 

 

As it happens I am one half of the QA team at my place and when I started as assistant I was just told it was part of my job. Now I run the department I still believe this should be down to myself and my no.2 just so I know it gets done and the readings taken by monitoring equipment are legit. Obviously you should be able to trust production staff to do things for themselves but does anyone ever REALLY do what they say they're doing 100% of the time?

 

Andy

[Charles.C]

Charles, what an extraordinary accusation! 

 

As it happens I am one half of the QA team at my place and when I started as assistant I was just told it was part of my job. Now I run the department I still believe this should be down to myself and my no.2 just so I know it gets done and the readings taken by monitoring equipment are legit. Obviously you should be able to trust production staff to do things for themselves but does anyone ever REALLY do what they say they're doing 100% of the time?

 

Andy

 

Statisticians ?

[Andy_Yellows]

Statisticians ?

The world's second biggest bunch of liars behind the board at my football club!

[Scampi]

We do the calibration ourselves. Major devices get calibrated by an external vendor 1/year (laser thermometer etc) Our margin on the thermometre is +/- 1F

[mlcf]

How do we do risk assessment for equipment calibration? is there a decision matrix just like in ccps?  Our equipment are temperature controllers, RTD sensors, weighing scale, temp dial gauges, pressure gauges

[Charles.C]

How do we do risk assessment for equipment calibration? is there a decision matrix just like in ccps?  Our equipment are temperature controllers, RTD sensors, weighing scale, temp dial gauges, pressure gauges

The Calibration is the Risk Assessment.

 

If you really need a RA see parallel thread/post linked below or Posts 7,8 above-

 

https://www.ifsqn.co...ng/#entry183126

[EmeraldGirl]

Personally think it should be QA if it isn't a difficult job. Calibration records form part of our quality records

Yes, it should definitely be QA in charge of the calibration program. The person PERFORMING the calibration should be someone competent in the metrology activities involved (they could be a member of another department). QA's role is providing non-production oversight of the process (since production has an inherent conflict of interest). 
EmeraldGirl (who may or may not have had QA oversight responsibility for Philips Healthcare global service equipment calibration program for 2 years)

Edited by EmeraldGirl, 13 April 2022 - 02:18 AM.

[Charles.C]

Yes, it should definitely be QA in charge of the calibration program. The person PERFORMING the calibration should be someone competent in the metrology activities involved (they could be a member of another department). QA's role is providing non-production oversight of the process (since production has an inherent conflict of interest). 
EmeraldGirl (who may or may not have had QA oversight responsibility for Philips Healthcare global service equipment calibration program for 2 years)

Hi EG,

 

Re ^^^^(red) -

 

Welcome to the Forum !