Applying Combined Tolerances of Two Scales - Balance Tolerance Question

[Pharmagal]

Hi:

My question may be a very basic, easy one to answer. Hope someone can help.

This is my scenario:
In manufacturing, 2 scales are used during filtration of liquid product from tank to a bag:
1) Floor scale: 0 to 1500 kg capacity, 1.4 kg tolerance, used to weigh a tank and tank with product. Gross weight of tank ? tare tank weight = amount of starting product to be filtered
2) Bench top scale: 60 kg capacity, 0.1 kg tolerance, used to weigh bag and bag with product. Gross weight of bag ? tare bag weight = amount of product filtered.

Combined tolerance of both scales = 1.5 kg.

I have situations where the amount of product filtered exceeded the amount of starting product available for filtration. Product #1 was over by 0.516 kg and product #2 was over by 1.761 kg. When manufacturing was questioned on how it is possible that more product was filtered than what was available for filtration, their response was that this can happen when 2 different balances are used. They concluded that product # 1 overage is well within the combined tolerance of the scales. For product #2, the overage is slightly more than the combined tolerance of both scales and manufacturing will investigate.

Since the gross weight of the tank is ~230 kg and the gross weight of the bag is ~ 18 kg, these amounts are on the low range of both scales. Shouldn?t manufacturing be applying a different tolerance other than the combined tolerance of 1.5 kg, which appears to be for the high range?

 

[dgriffith]

Might need some additional information. How is the spec applied to each scale--full scale out, % reading, etc.? Second, the scale tolerances may not combine linearly. If their calibration isn't correlated to each other, then they would be summed differently, I believe. Thirdly, what is the tolerance distribution and coverage factor--normal at k=2 or k=3, or another distribution type with coverage factor. You can't combine them until they are normalized, especially if you are not linear summing.

When you say gross weight, do you mean net product weight in the tank and bag?
Yes, the weights are on the low end but well over 5%, which is why I'd like to know how the spec's are applied.

 

[dwperron]

A problem you have may be in the calibration of these scales. On most quality scales there is more than a single "accuracy" specification, though some times the different accuracy components can be combined to achieve a worst case tolerance.
You might check into how your scales are being verified. There are several tests that should be performed, they will give you a better idea as to how the scales will perform in "real life" use. These tests may not have published specifications on many scales, but performing the tests may point out performance problems. These tests should include, along with the basic accuracy test:

Linearity - checking at least over the range you will use the scale
Repeatability - checking with a weight approximately the size of your target measurement, checking between 5 and 10 times
Eccentricity, or off center loading - setting a weight on the center of the platform then moving it to the corners of the platform, comparing the readings
Zero tracking - placing a weight on the scale, taring the scale, then repeatedly re-weighing the weight to see how the zero holds

Problems with these tests could cause a scale that passes basic accuracy calibration to give faulty results.

 

[Pharmagal]

Thank you dgriffith for the reply. It is my first time in the forum, so I appreciate it very much. I guess my question wasn't as simple as I thought. I'll try to find out the additional information from our Metrology group and report back. With regards to gross weight, it does mean net product weight + tank or bag weight.

 

[Pharmagal]

Thank you dwperron for your help as well. I'm trying to look into calibration as well so this is very helpful. So far, I have read manufacturing's monthly weight check procedure for the floor scale and daily weight check procedures for the bench top balance. The procedures are not written in enough detail for me to comprehend fully.

Simply described, for the 0-1500kg floor scale, they either use a 10 kg or 25 kg calibrated weight placed to the four corners of the scale. It's not clear if they are moving the weights from corner to corner or just adding weights to each corner. They are using 1.4 kg as the tolerance for the calibrated weights. I always thought that weight checks would have different tolerances depending on the calibrated weight that is chosen. Is it appropriate to use 1.4 kg tolerance for both the 10 kg and 25 kg calibrated weight.

For their 60kg benchtop, it looks like manufacturing only does single weight checks before use choosing a calibrated weight that is representative of the measurement to be made and applying a 0.1 kg tolerance to all calibrated weights. Should they be doing bracketing weight checks instead and applying a different tolerance to the bracketing calibrated weights? I'm more used to analytical balances in a laboratory setting and we would perform bracketing weight checks using different tolerances.

We have set up meeting with metrology and manufacturing, so thank you again for your input. I will bring it up to them.

 

[dgriffith]

Simply described, for the 0-1500kg floor scale, they either use a 10 kg or 25 kg calibrated weight placed to the four corners of the scale. It's not clear if they are moving the weights from corner to corner or just adding weights to each corner. They are using 1.4 kg as the tolerance for the calibrated weights. I always thought that weight checks would have different tolerances depending on the calibrated weight that is chosen. Is it appropriate to use 1.4 kg tolerance for both the 10 kg and 25 kg calibrated weight.

I think they're doing it this way: The 1.4kg tolerance is for the reading on the floor scale, not the check weight. If the tolerance is 1.4kg for the scale, then all weight readings at the corners should read the calibrated weight of the check weight plus or minus 1.4 kg. They are likely using only one check weight and moving it around.

For their 60kg benchtop, it looks like manufacturing only does single weight checks before use choosing a calibrated weight that is representative of the measurement to be made and applying a 0.1 kg tolerance to all calibrated weights. Should they be doing bracketing weight checks instead and applying a different tolerance to the bracketing calibrated weights? I'm more used to analytical balances in a laboratory setting and we would perform bracketing weight checks using different tolerances.

Not an expert here, but if the check weight is similar to the expected production weight, that is all that's needed to check zero offset. Again, the 0.1kg tolerance is for the bench scale, not the check weight. You could do a bracket to check for linearity, but then you have to haul out more than one.

 

[dwperron]

If they are checking a 1500 kg floor scale with a 10 or 25 kg check weight you have a big problem. You are most probably using it at well below the minimum recommended weight (typically 5% of the range, or 75 kg for your scale). And if they have a 1.4 kg tolerance on a 10 or 25 kg reading - I don't see how that scale could ever fail!

 

[BradM]

Do understand... It's easier for manufacturing to point fingers at "Calibration" than determine if something else is going on. What I am not seeing here is them trying to actually determine why there is variance in the process.

Hi:

My question may be a very basic, easy one to answer. Hope someone can help.

This is my scenario:
In manufacturing, 2 scales are used during filtration of liquid product from tank to a bag:
1) Floor scale: 0 to 1500 kg capacity, 1.4 kg tolerance, used to weigh a tank and tank with product. Gross weight of tank ? tare tank weight = amount of starting product to be filtered
2) Bench top scale: 60 kg capacity, 0.1 kg tolerance, used to weigh bag and bag with product. Gross weight of bag ? tare bag weight = amount of product filtered.

Combined tolerance of both scales = 1.5 kg.

I have situations where the amount of product filtered exceeded the amount of starting product available for filtration. Product #1 was over by 0.516 kg and product #2 was over by 1.761 kg. When manufacturing was questioned on how it is possible that more product was filtered than what was available for filtration, their response was that this can happen when 2 different balances are used. They concluded that product # 1 overage is well within the combined tolerance of the scales. For product #2, the overage is slightly more than the combined tolerance of both scales and manufacturing will investigate.

Since the gross weight of the tank is ~230 kg and the gross weight of the bag is ~ 18 kg, these amounts are on the low range of both scales. Shouldn?t manufacturing be applying a different tolerance other than the combined tolerance of 1.5 kg, which appears to be for the high range?

Set aside the calibration tolerances of the two devices for a second, and let's just talk resolution/sensitivity. On the 1500 KG balance, once it is tared for the tank, what kind of resolution is there? And they are measuring a process requirement of .5 kg? That scale is simply not designed and not capable to deliver those kinds of results.

You list two products. If there are two products, when does mfg. "combine" the two tolerances? Is one scale for one product, and the balance is for the other product?

Simply described, for the 0-1500kg floor scale, they either use a 10 kg or 25 kg calibrated weight placed to the four corners of the scale. It's not clear if they are moving the weights from corner to corner or just adding weights to each corner. They are using 1.4 kg as the tolerance for the calibrated weights. I always thought that weight checks would have different tolerances depending on the calibrated weight that is chosen. Is it appropriate to use 1.4 kg tolerance for both the 10 kg and 25 kg calibrated weight.

The scales need to be verified across the range of use. Depending on it's mfg., many have four load cells. One of the load cells can be bad, and if not checked at the use, it won't be detected. It's possible (not saying probable) that floor scale could have a very large error at 750 kg (relevant when you tare the scale), however, no one will ever detect the error.

For their 60kg benchtop, it looks like manufacturing only does single weight checks before use choosing a calibrated weight that is representative of the measurement to be made and applying a 0.1 kg tolerance to all calibrated weights. Should they be doing bracketing weight checks instead and applying a different tolerance to the bracketing calibrated weights? I'm more used to analytical balances in a laboratory setting and we would perform bracketing weight checks using different tolerances.

My opinion... this is certainly better than the first scenario. If you are using a balance at 60kg and you verify the accuracy at 60kg, then you can assure the balance would not affect your process. However, the calibration procedure (different than a routine performance check) should be across the range of the balance, with corner load and other checks being made.

So what I would do...
1. Determine what equipment is actually needed. What are they trying to weigh, and how discriminating do the results need to be? This will let you know what equipment you need.
2. See if the equipment you currently have match 1.
3. Once those two are acceptable, then you can focus on appropriate calibration of the equipment.

 

[Pharmagal]

If they are checking a 1500 kg floor scale with a 10 or 25 kg check weight you have a big problem. You are most probably using it at well below the minimum recommended weight (typically 5% of the range, or 75 kg for your scale). And if they have a 1.4 kg tolerance on a 10 or 25 kg reading - I don't see how that scale could ever fail!

Thanks for this reply. I also thought that 1.4 kg tolerance on low weight readings would mean no failures. Since then, I did get more details for the monthly weight checks on the floor scale. Looks like they just do the zero tracking check which you described earlier. This is what they do:

1) A tank or weights (that exceed 25% of the scale's calibrated range) are placed on the weighing platform and tared.

2) Starting with the right front corner, a standard weight of 10 or 25 kg is applied.

3) Continuing clockwise, another weight is applied to each corner (i.e. left front, left rear and right rear)

4) The weight indicated after each additional application of weight on the floor scale needs to meet the tolerance for the scale.

Is this any better? Doesn't your response of never failing still apply?

 

[Pharmagal]

Do understand... It's easier for manufacturing to point fingers at "Calibration" than determine if something else is going on. What I am not seeing here is them trying to actually determine why there is variance in the process.

This is the best point. Mfg. does not even bother to check to see what else may have contributed to the additional weight. Now I have to look for trends. For the product that was over by 1.761 kg, this was the 6th time that they manufactured the product and every single time, the amount filtered was more than what was available. This is the first time where it exceeded the tolerance of the scale. I've never seen a batch of this product where the amount filtered was less than what was available. It doesn't make sense since there are losses during filtration, i.e. sampling and priming.

You list two products. If there are two products, when does mfg. "combine" the two tolerances? Is one scale for one product, and the balance is for the other product?

The scales are actually used to weigh many different products where product is filtered from tank to bag. The floor scale is used to weigh the starting or remaining product in the tank to be filtered into the bag. The benchtop scale is used to weigh the amount of product filtered into the bag. Hence, they always combined the tolerances of the two scales.

So what I would do...
1. Determine what equipment is actually needed. What are they trying to weigh, and how discriminating do the results need to be? This will let you know what equipment you need.
2. See if the equipment you currently have match 1.
3. Once those two are acceptable, then you can focus on appropriate calibration of the equipment.

Thank you for all the points and questions that you posed. Much appreciated. I will keep them in mind as I question mfg. Grrr!