Calibration Methods
Simulation of Live Load for (Belt Weigher) Calibration.
There is debate over the best method of simulating live load, and many people strong views on this matter. In general order of perceived increasing quality, methods of simulating live load are:
- Electronic simulation
- Static mass applied on the weigh frame under the belt
- Rolling or dragging mass on the conveyor belt over the weigh frame
- Live slugs of material which are themselves weighed on other static weighing systems such as bins or weigh bridges Read More
On a good weigh frame, any of these live load simulation methods will give good results. On an inadequate weigh frame; even the best methods of live load simulation will not save the situation.
Live Loads of Material
In the case of weights and measures applications, only live loads of material are accepted for calibration purposes. This is detailed in a new document recently released by the National Measurement Institute called, NMI V7 (National Measurement Institute, an Australian Government instrumentality), a document based on the OIML (International Organisation for Legal Metrology, of which Australia is a member) document R50-1 and R50-2. The method stipulated in OIML R50 and hence NMI V7 uses a total of 10 test loads at three different flow rates, thus addresses issues of randomness (repeatability) and linearity. This is clearly the safest and most thorough method. However, it does require that there be a convenient reference scale (Control Instrument) available, and a suitable method of handling the large amounts of material involved. It is important to note that live loads or slugs of material should only be considered as a way of simulating real live load conditions because test conditions are often quite different to real operating conditions.
Electronic Simulation
The least reliable method of simulating live load is the electronic method. This is because it is based on a load cell calibration done elsewhere and then, on site, a reliable millivolt signal must be generated. This is equivalent to placing a known mass on the load cells in the weigh frame. It is not easy to produce a reliable mV signal within 1% (i.e. to the nearest 100 microvolt), often the inability to generate a reliable mV signal on site has caused this method to fail.
Additionally, the weigh frame has more parts than just the load cell. When real weight is applied, the load cell deflects and load is distributed to other parts such as alignment rods, or to large flexures. Belt stiffness may cause less of the belt weight to be experienced. The mV simulation misses all of these effects.
Another complication is that mV simulation does not recognise the inclination of a belt, making it necessary to take into account the cosine of the angle of inclination of the weigher when using electronic simulation. The actual angle of inclination needs to be measured and accounted for by means of a calculation.
These various sources of error combine to make electronic simulation notoriously inaccurate. Typical errors (from hearsay) in the industry are from 5% to 20%. This method is not recommended for any serious belt weighing application as it is as far from best practice as it is possible to be. The method of calibration checking defined by the Weights and Measures NMI V7 should be considered ‘Worlds Best Practice’.
Static Mass for Calibration
Static mass is the most popular method used for the calibration of belt weighers. The success of static mass as a calibration reference depends upon the stability and design of the weigh frame.
If the weigh frame is highly influenced by belt tension because the weigh length is too short, the weigh frame is too easily deflected, or belt tension is too high and variable, then the randomness of the system is very high. When static mass is used in this situation the calibration may well incorporate a large component of systemic error.
Another influence factor which has a very negative affect on belt weigher stability is idler friction which only affects pivoted weigh frame designs. Some pivoted systems may have as much as 2% to 3% of idler friction in their output signal. Idler friction is not a constant and changes with:
- levels and age of lubrication
- temperature
- seal drag and wear and
- changes depending upon belt condition (a large part of idler friction comes from indentation resistance which is more a characteristic of the belt than of the roller)
Weigh frames whose weight is supported fully by load cells, or by other measuring elements which all have the same sensitivity are known as the “fully suspended” type and are immune to the idler friction influence factor.
Static mass cannot fully simulate real material because it does not take into account some of the influence factors coming from belt tension and deflection that occur when load is applied through the belt. Systematic error can be relatively large due to these factors, when static mass is used to simulate live load for calibration purposes.
An experienced belt weigher manufacturer would design a weigh frame so that influence factors including belt affects are attenuated to acceptable levels. With the resulting suitably designed weigh frame, static mass can adequately simulate live load.
Calibration Chains
This is perhaps the most hotly debated means of belt weigher calibration. Many people believe that ‘calibration chains’ and sometimes ‘calibration trains’ are the definitive means of belt weigher calibration.
The calibration chain is a distributed mass which lies on the belt over the weigh frame and simulates a particular belt loading usually expressed in kg/m. The chain may consist of a series of rollers, be a series of masses on wheels, or be only a sliding mass which is tethered above the weigh frame. A ‘calibration train’ is a series of cars which run on rollers to which masses may be applied. The ‘train’ has the advantage that it allows a range of kg/m loadings to be simulated which permits linearity testing.
Experience in the field suggests that calibration chains produce no better results than static mass would in a similar situation. Errors in the order of 7% have been known to occur between belt weighers which have been calibrated with chains and true weight.
The calibration chain does not simulate belt tension. This is also the main criticism of static mass, so both methods share this common and significant failing.
Unfortunately the history of the application of calibration chains has tended toward the purchase of relatively light duty, lower cost weigh frames teamed with very expensive calibration chain installations. There has been a belief that it almost doesn’t matter what the weigh frame is like, it can be calibrated with a calibration chain, which is incorrectly believed to exactly simulate live load. It is true that a calibration chain may well be a better means of calibrating a high deflection weigh frame, however, this same weigh frame will be highly unstable and its high random component of accuracy will ensure that it also holds a large component of systematic error. This scenario tends to disqualify the application from being really good for product reconciliation or for process control.
The CST World's Best Practice Approach to Calibration
CST has long realised the importance of a suitable high quality (more expensive) weigh frame, which resists deflection, and use lower cost, lower maintenance, static calibration masses. Some customers insist on calibration chains, and we have supplied chains and static masses together on multi idler, fully suspended weigh frames. We have seen both static mass and the calibration chain agree within 0.25% on these occasions.
It is important to realise that those most critical factor impacting on the success of any calibration method is the strength and design of the weigh frame, which, if designed to be 'fit for purpose' for the particular application, will eliminate one of the most significant factors affecting calibration and system stability - weigh frame deflection. Poor quality weigh frame design, using steel not strong enough for the application, will lead to significant frame deflection in use, and when testing. This lack of quality cannot be 'calibrated out'. Using calibration adjustments to 'fix' poor quality weighing equipment has the unfortunate result for site management of introducing large, unknown, system errors. The loss of revenue, or error in process control can be significant. Click here to read about the impressive stability of a CST belt weigher used for trade purposes in Portland, Oregon. Call CST to enquire about a free or comprehensive Site Audit on your weighers if you have concerns that either your weighing equipment or calibration practices used on your weighers may be introducing accuracy errors in your site processes. (Free Audit Service)