Calibration of Load Cells - ISO 17025 & MSA (Measurement Systems Analysis)

[Dan M]

We are a chemicals manufacturer pursuing TS. We mix large volumes of raw materials in blending vessels, each raw material measured by weight via load cells. At present time, the load cells are not calibrated within the scope of our calibration contractor's ISO 17025 system. Our calibration contractor informed us that the current procedure they follow (which was specified by our organization over 10 years ago) will not be suitable for ISO 17025. Basically they're calibrating the load cells from 0 - 1,000 lbs., wherease the range of use may be up to 30,000 lbs.

It is practically impossible to calibrate these load cells through the range of use because (1) they can't place the amount of weights that would be necessary and (2) a build test would take weeks per blending vessel and we can't afford to shut down production for any extended period of time.

The calibration contractor did inform us that if we provided our Customer-specific calibration method in writing, which indicates calibrations areto be performed from 0-1000 lbs. he would can then generate an ISO 17025 Cert with a cavaet that the calibration was performed per Customer Specification. Does this seem acceptable?

Our TS auditor also informed us that we need to do MSA on these load cells and I would guess that the current calibration procedure would not be suitable for a proper MSA study to be performed.

I am not a calibration expert and unfortunately neither is anyone else in our organization. We've always relied on our contractors to be the experts.
Does anyone have any experience with how calibration of large-capacity load cells is handled? Certainly many TS organizations utilize truck scales - are they actually calibrating these through their range of use?

 

[Hershal]

Great question! Your calibration provider should be able to add load cells to the scope if not already there. 30,000 lbf cells are not uncommon. In concrete labs for example, load cells go to 500,000 lbf routinely, and I have done assessments that include cells to 850,000 lbf.

Typically the calibrations have been done using ASTM E 4 as the procedure, and comparison to a load cell calibrated under ASTM E 74.

The key difference here may be the use of a compression machine, and many are available on the market in lots of sizes. They can even be built using hydraulic pumps. The cal lab may be able to build or obtain one easier than your organization.

Truck scale cal houses often have sufficient weights to stack, but yes they do cal to the load cell upper end.

Hope this helps.

 

[simpleque]

As a TS and 17025 auditor/consultant I don't like what your calibration supplier is proposing to you. If I were your TS auditor I wouldn't accept that approach. My recommendation is to find another calibration source and someone who can help solve your problem. Hershal is right on with his comments.

For the MSA study, you'll want to reference the AIAG MSA manual to set up and perform an acceptable GR&R study. For that you get to pick the material and weight and select the operators to measure, so be smart about how you set up the GR&R study so you can get at least 2 operators and 5 trials (3 and 10 are ideal). Elsmar has GR&R templates you can use - they are easy to come by. Most templates can accommodate 2 operators and 5 trials, but if you can increase the sample the statistical validity is greater.

I am not a fan of GR&R performed during calibration. That doesn't meet the intent behind the AIAG MSA manual for the operators as whomever is performing the calibration is not the operator reading the scale/load cell in a production environment.

You can get around calibrating through the entire range of the load cell if there are typical operational ranges where the scale is used. Then you can justify calibrating in the operational range versus the full scale of the load cell. This might help some.

Hope this is helpful.

 

[BradM]

I would caution against not calibrating the load cells across the range of use. Now... do note this is "range of use"; not it's capability range.

If the use range is up to 30,000, then you really want to verify it up to that range. What can happen is if you have a vessel with three or four load cells on them, one of them can be bad. If you don't check at the use range (more than 1000 lbs) then what can occur is the other "good" load cells will compensate for the "bad" load cell; and you won't see the error.

A possibility is to have a spare set of load cells. Have one set installed, and send the other set out to a competent vendor for calibration. Then, when the other set is due for calibration, change the set out.

I know there are companies around here that verify weight scales for trucks and such that are 17025 compliant. That might be a fruitful option.

 

[Dan M]

I would caution against not calibrating the load cells across the range of use. Now... do note this is "range of use"; not it's capability range.

If the use range is up to 30,000, then you really want to verify it up to that range. What can happen is if you have a vessel with three or four load cells on them, one of them can be bad. If you don't check at the use range (more than 1000 lbs) then what can occur is the other "good" load cells will compensate for the "bad" load cell; and you won't see the error.

A possibility is to have a spare set of load cells. Have one set installed, and send the other set out to a competent vendor for calibration. Then, when the other set is due for calibration, change the set out.

I know there are companies around here that verify weight scales for trucks and such that are 17025 compliant. That might be a fruitful option.

Thank you Brad. I was under the impression however that calibrating the load cells offsite would not be sufficient since it would not represent the true operating conditions, i.e. not perfectly level surface, interference from piping, electrical. By no means am I an expert and this was just my assumption.

 

[Dan M]

I discussed this issue further with our scale calibration contractor and the feedback I received is that calibration of the load cells off site only confirms that the devices are functioning but doesn't necessarily confirm that they are producing an accurate reading of the actual weight in the blending vessel. We also concluded that it would not be practical to do a build test one-thousand pounds at a time up to 30,000 pounds and besides this method introduces quite a bit of error by the time we reach 30,000 lbs.

A suggestion I received from our consultant is to continue with the current method of calibration and ensure that the results are linear. His other suggestion is to obtain a calibrated flow meter as a tool in determining Gage R&R by using it to deliver a known weight of raw material in to the blending vessel by at least 2 operators over 5 production runs.

Does anyone have any thoughts about using this as a method for MSA?