Will

The most accurate way to test a belt scale (assuming it has a totalizer) is to measure the weight of a given sample of the product either before it goes onto the belt or collect it after the belt and weigh it. If your scale is accurate your totalizer should be very near the weight of the measured material. This may be difficult in an underground environment. Some mines use truck scales U/G if you have that luxury you are set.

Repeated test will give you your percentage of accuracy. Be sure to zero the scale out each time prior to the test while the conveyor is running.

We have on one ocassion brought a portable axle scale U/G to measure axle weights of truck for calculating loads in our trucks, this may be an option for you.

Some information from NISA in the following link may be helpful also.

http://www.nisa.org/belt.html

Good luck

Gary Blenkhorn ■

I agree with Gary. One problem with large chunks of material moving over the idlers is the error caused by impact on the weigh idler. Depending on the amount of fines associates with the material, this can give you a variable result as the fines tend to lessen the impact.

I my opinion, the only way to prove a belt weigher is doing its job is by a weighed load check. I have produced a booklet on the do's and don'ts of belt weighing. If you would like a copy let me know.

Regards

Jon Scarrott

jonlodestone@aol.com ■

Jon

I would like a copy of your booklet. Is it something that can be emailed or do you need to send it snail mail?

Regards, Gary ■

Will,

There are four areas that need to be addressed when checking the functionality of a conveyor belt scale. Accuracy, precision, repeatability and linearity.

Accuracy is the ability of the system to give a correct weight.

Precision is the ability to give the correct weight to the nearest increment required.

Repeatablity is the ability to give the correct weight over time.

Linearity is the ability to give the correct weight at varying rates of flow.

The most straight forward way to determine these attibutes is by performing some variations of the previous mentioned material tests.

The NISA link in Mr. Blenkhorns reply covers a lot of the advantages of material testing, however the rules regarding material tests have since been updated by NIST to require testing as varying feed rates.

You may want to also see:

http://ts.nist.gov/ts/htdocs/230/235...1-H44-05-Z.pdf

please refer to section N.2 ■

Hello Will,

Just install two beltscales in series and you'll be able to measure and monitor not only the variance between belt scales but also their relative bias.

Regards,

Jan W Merks ■

WJansen

I agree with the other correspondents about live load testing. We are very fortunate to have access to a closed circuit live load test system so the same load can be repeatedly static weighed and circulated.

However, I wanted to mentions something else we do which I think is interesting. Our system has a live zero footprint of the belt, in other words, it keeps a zero constant for each section of belt. When we do a zero check or re-acquire a new zero image of the belt, we have the opporunity to compare the difference between the old weight and the new weight of each section of belt. We usually use sections which represent about 1 second of belt running, so 3 to 5 metres. Then we calculate and display a real time Std Deviation number which I think to a large extent represents the real precision of the weighing system, at least at zero.

This is something you could consider doing. You might need our electronics to do that, but maybe we can work something out. ■

The most effective system to measure and monitor the precision for a belt scale is to put two identical belt scales in series. The more so because such systems make it possible to derive precision estimates for wet masses of all sort of materials in bulk. Jan W Merks ■

Jan W Merks suggestion to put two belt weighers in series is a luxury few can afford. I have been involved in this type of testing in a power station in Scotland and it does work. However most customers would reject the idea on cost. Maybe a university would consider this option, but in real life situations it will generally be rejected.

I am also interested in L. Ian Burrells' idea of a continuous reading of the belt position and knowing the local weight of the conveyor sections over known lengths of belt. I have been involved with this with a certain Swiss company, but only with weigh feeders, basically short belts. The problem is marking the postion of the belt and informing the electronics with say a smart marker embedded in the belt layers. Great, until a new section of belt is let in and throws the system out of kilter.

RI50 deals with this problem by specifying that the weighed load checks should be carried out over a number of complete revolutiuons, even if at the end of the test there is no material on the belt. This obviates the need for such sophisticated electronics.

Jon ■

I am also impressed with the L. Ian Burrells' idea of zero tracking. I can think of another advantage in knowing when a specific section of belting is over the scale, and that would be precisely knowing the starting point and ending point of small quantities of product. As the belting travels over the scale, the tare weight varies both heavy and light from the average. By continually adjusting for this variation, you should be able to see a little bit more of the true weight of the material.

My question to Mr. Burrell Do you use a marker on the belt to aquire belt postion? ■

Todd,

We have used an optical sensor with a paint mark, we have used something vulcanised into the belt with a prox, we have used a metal clip in the belt, the system also works without a mark, using a manual PB input occasionally to re-synchronise it - used for instance just during live load testing with loads smaller than the normal delivery. System is not upset if the mark disappears, it always keeps the conventional belt scale zero in the background in parallel. The systems combines three methods of zero/totaliser treatment, conventional, totaliser update once per rev only and zero image. ■