The need for an accurate belt weigher

Belt Weighers are used for applications which really test their accuracy. If belt weighers were just flow rate indicators, the accuracy issue would be simpler; however, since belt weighers also include a totaliser upon which all the errors are accumulated, the errors can seem larger than they really are. The following discussion explains how weighing accuracy error can accumulate with a significant effect on product reconciliation, into and out of a stockpile.

Weighers Used for Stockpile Management

Stockpile management is the ideal example of the need for accurate belt weighers for product reconciliation.

Firstly, there is a need for an understanding of real belt weigher accuracies, and then an informed use of the information when used to value assets such as stockpiles. There are many horror stories about “the coal that disappeared from the stockpile” or "how the train or ship could be not loaded because the coal that was supposed to be there did not turn out to be there". Some peoples' careers have been unfairly affected by this basic misunderstanding of how belt weigher errors work.

As with all measuring instruments, all belt weighers have two types of error

• systematic error ie. the error that’s always there; and
• random error ie. the inability of the equipment to give exactly the same answer when measuring the same thing a second time.

A typical stockpile problem is one which has an incoming belt weigher with a 0.25% accurate belt weigher, and one reclaim belt weigher also 0.25% accurate. For this example we will consider that the two belt weighers each have a systematic error of 0.25% each, and that one is reading 0.25% high, and the other 0.25% low. Over large amounts of material, the random errors will in theory have averaged to zero.

The problem with stockpiles is that belt weighing errors add and also accumulate.

The errors from the input and output weighers will combine, even though the difference (or subtraction) of the two results is used as the stockpile figure. Worse still, this same error accumulates after the weighed material has left the stockpile.

Stockpile errors add and accumulate

The error in the stockpile figure is the sum of the errors of the belt weighers used to calculate the figure. If two belt weighers are involved, each with real errors of 0.25%, then the error in the stockpile figure will be 0.5%. In terms of statistics, the random components of error (known as the variances) of the two weighers would add, but here we are speaking about a hypothetically known systemic error, and the random error component really should integrate to zero.

As material passes through the stockpile, the error we are considering remains 0.5% of the total amount of material which has passed through, it is not just 0.5% of the amount in the stockpile now. After, three million tonnes have been through a stockpile there may be theoretically 30,000 tonnes remaining. The problem is that 0.5% of three million tonnes is 15,000 tonnes. The truth is that the stockpile figure is the difference between the 'ins' and the 'out' plus or minus the sum of the errors. So the real stockpile could be anywhere between 15,000 and 45,000 tonnes.

If systematic errors can be determined, they can be adjusted to zero. The problem is that it is difficult to accurately determine the amount of coal or ore that is being moved because a reasonable proportion of it is water, and the amount of water changes over time.

Another factor in the equation is that the random errors in belt weighing don’t enter results at a high frequency, so that errors might average to zero over one day. Random errors in belt weighing are more properly called Influence Factors. Some of these have a period of one day, a week, or a year, as temperatures and seasons change.

Process Control Belt Weighers

The other major purpose of belt weighers is process control. In this role, the focus is on the indicated flow rate than in the accumulated tonnage figure.

The common belief is that a belt weigher with much lower accuracy is adequate for this role. Usually, a single idler belt weigher is chosen. However, systematic and random errors are still at work. The combined effect of these can lead to an apparent need for heavy maintenance, and errors much larger than expected.

Process Control Weighers and accumulated error

The process control belt weigher is at the other end of the application spectrum when compared to the ‘product reconciliation’ use. In the case of product reconciliation, final results consist of the accumulation of many measurements taken over days, weeks or even months. The result is accumulated in the totaliser. It includes all the systematic error of the belt weigher and in a sense, none of the random error. The process control application however, uses the instantaneous flow rate from the device. This output has all the random error and all the systematic error in it. As a result, and especially given the common choice of a single idler belt weigher for this role, the user has built into their process a +/-5% variability from day to day. This can hardly be helpful in controlling a process.

Belt weigher maintenance is an important issue to be considered in relation to initially inexpensive single idler belt weighers. This type of equipment is often thought to be in error (because it often is), so units are often re-calibrated. As calibration seems to change significantly, and the weigher had been in error beforehand, regular calibration is “seen” to be a beneficial routine. However, the regular calibration work is probably just moving around inside the random error of the system. The unit would be better left alone. A proper analysis of this situation would lead to the realisation that a higher quality belt weigher was a better idea, less expensive overall, and even more importantly, saves monetary losses from inaccurate weighing on an ongoing basis.