Dear Rekhawar

Can you send me some details of your application so I can assess how much error you should expect from simple dead weight calibration. ■

Mr Rekhawar,

You could consider using normal link chains, enough strands so kg/m is similar to you loading, these would drag on the belt while belt runs. This method could provide some assurance about live load performance of your scale.

The fact that you seem to need 0.4% adjustment after 7 days and then straight away repeatability is only 0.12% says that the weigh frame is not good enough to infer live weight performance from dead weight calibration. It could be out by 5% or worse.

Really, you need a better weigh frame, long enough say 4 idler, and stiff enough so it is stable. Then you can have some confidence in live load / dead load relationship.

Please send me the GA drawing of conveyor, I will let you know what spec of scale you need in this location to be able to dead weight cal for live load performance. ■

Check the side alignment and height of carrier rolls in relation to the rolls preceding and after the scale. Compare that with the scale manufacturers recommendation. Any descrepancies can cause erronous readings. I have exprienced this type of error when the empty belt was lifting off the scale. Particular attention must be made for idlers systems at a belt scale.

Single idler scales are not as accurate as multiple idler scales on a deck frame.

Hope this helps.

Gary Blenkhorn ■

Sir,

There are a number of ways to check the calibration of a conveyor belt scale. In the US, the accepted method for proving a belt scale is by a material test. After proving the scale by the material test, a simulated load test is factored to the material test.

In this way, the scale can be checked to see if it is holding calibration. The simulated load tests that are most common are calibration test chains, static weights applied to the scale carriage, and an electronic method of elevating the load sensor signal. All of these simulated loading methods have advantages and disadvantages. The calibration test chain can compensate for some of the characteristics of the belt tension, but can be hard to handle and must be applied consistently. The static weights are easier to apply, but dont compensate for changes in belt tension. I have seen static weights applied to conveyors with poor idlers and calibrate fine, but when material is run, the weights are completely erratic. The last method is the electronic calibration that is very simple to use, but does not apply a load to the carriage or see any changes in belt tension. The electronic load method is best suited for scales installed as close to manufacturers reccomendations. Any of these methods can be used to double check a scales performance, but all of them should be factored to match a material test. The bias between the simulated load tests and the material test can be as high as 5% different, but can be used to check repeatability. In our experience, We usually see the test chain calibiration to bias low, the test weights to bias high, and the electronic load test to be an indication of how well the system was installed when checked against the material test. ■

Mr. Rekhawar,

I have worked on a hundreds of belt scales that can not be (material) drop tested. I need to mention the errors that can be associated with a belt scale, before I can suggest the methods of testing.

A belt scale must be stable. The calibration of the system should not require frequent readjustment. On legal for trade systems in service in the US, the errors found by the mentioned simulated load tests must hold within .25% and if the error is greater than .75% the system can not be adjusted and a material test is required.

A belt scale must be repeatable. The error found by material testing, must be consistent. After the scale has be adjusted, the scale is not considered acceptable unless subsequent test prove the scale is within tolerance and repeat tests are also within a tolerence band. The repeat tests must be within .25% of each other. As an example, (+.12%, -.06%, -.14%) would not be accepted as the scale repeatability would be .26%, but (-.20%, -.15%, +.01%) would be accepted with a repeatablity of .21%.

A new requirment that is being added next year is a test for linearity. The linearity is checked by running material tests at different flow rates. The requirments for the linearity test have not been formalized so I can't give you specific examples, but as I understand the proposed test, the scale is material tested at a low, a normal and a maximum feed rate and all three tests must be within the .25% tolerance.

The simulated load tests I mentioned in my previous post, are used to verify the scale is holding calibration, and in cases where a material test can not be performed they are used to set the scale as close as possible to a true weight. Without a way to material test, all a simulated load test can do is prove the scale is performing consistently.

A material test must be performed to verify the weight registered is correct. The reference weights used to check the belt scale are derived from a scale that is traceable to a known standard, and the accuracy of the reference scale must be at least twice that of the belt scale.

The material test will determine the bias between calibrating with a simulated load and what the true calibration should be. After finding this bias, the calibration value of the roller chain, static test weight or electronic test is corrected to be as close as possible to track with the true weight.

Being that a material test is not possible, there are a few suggestions that may help in determining this bias. The heat rate of a boiler based on the inventory of the coal being used, or the recovery of a finished product as compared to the raw product. I have seen stop belt samples where a fixed feed rate is checked by stopping the process and physically weighing the product over a set distance of the belt. None of these methods are as precise as a material test, but they can be used to double check the results of the simulated loading tests. If you were going to pick one method of simulated load testing, the test chain is probably the most representative.

So in closing, the short answer is a belt scale can not be verified without running a material test. The simulated testing methods can be used to keep the scale consistent.

I hope my reply has not created more questions than answers, ■