Verify continuous-weighing instrument accuracy
Information received from continuous weighing devices is indicative of what’s happening in a plant, and most particularly, how much material actually passes through process-production equipment.
The information supports many purposes, including inventory material-balance, production control, employee compensation and customer invoicing. At the end of a year of operation, the accumulative tonnage can seem monumental and the dollars involved considerable.
The "scale" of the effort thereby documented only underscores the need for verified instrumentation accuracy by means of material testing. Material testing involves weighing a material sample on a known accurate, static scale either before or after it has crossed the conveyor belt scale.
Material tests are the most accurate means of verifying the performance of a belt scale, weighfeeder or other continuous-weighing solution.
Preliminary considerations
Finding a way to capture a material sample or introduce one in the process can take a bit of creativity. Sometimes rotating a screw conveyor or chute has room for a small hopper placed under the discharge. Another common way of material testing is to empty the bin before the scale, partially filling it with pre-weighed material.
Make sure extraneous factors such as material tare weight, container variation, drivers’ schedules and behavior or vehicle fuel capacity don’t spoil data dependability. Make sure extraneous factors are eliminated or taken into consideration, before losing hours or days of work.
Sample size is a function of the test rates and durations planned. Ideally, the test should be under normal running conditions. If this isn’t possible, you must be well above the turndown of the scale, with a minimum of 50 percent of the scale’s rated capacity typically recommended.
In determining the minimum material sample size, factors include accuracy, test-scale and conveyor-scale resolution and belt condition. In general, the material sample should be as large as possible, with transport between scales often being a major constraint.
Minimizing the effects of belt-weight variation and error caused by material ramp up and down must be considered for continuous weighing. The National Institute of Weights and Measures recommends a material test for a certified scale be at least 10 minutes, with exceptions. Sometimes the material amounts needed to do that large a test can be challenging. In most cases a minimum of five minutes or three belt revolutions of continuously loaded belt is enough to provide an accurate material sample.
Test the scale
Before starting, calibrate the belt scale and verify static-scale operation, a step too often overlooked when material testing.
Before performing material tests, ensure the scale to be used is commissioned and at hand is a record of the last zero and span calibrations, needed to verify repeatability.
Note that the desired accuracy must represent at least two totalizer counts. For example, consider a scale that totalizes in tenths of a ton (0.1 tons per count) and uses a one-ton material sample. If the scale is just about to increment to the next count (1.1 tons), the totalizer will read one ton when you actually have 1.0999 tons. In this case, the scale will appear to be very accurate, but in fact the error is almost 10 percent.
So, for a 0.5 percent scale at least 400 totalizer counts is recommended and for a 0.25 percent scale at least 800 totalizer counts is recommended. The resolution of the test scale should be at least double that of the conveyor scale.
To correctly material test a conveyor-belt scale, run three weighed material samples. The first two determine if there is a repeatable error. Each material sample should be run across the conveyor belt scale continuously at a rate more than 50 percent of the scale’s rated capacity. If there is a repeatable error, the conveyor belt scale can be adjusted and a third test done to verify the correction. If there is a not-repeatable scale error, its source must be found and corrected.
Sample test
To run a percentage change material test:
- Run the belt or pipe empty.
- Perform a zero calibration.
- Put the integrator into RUN mode and ensure that the totalizer resolution is appropriate.
Resolution versus sample size is an important relationship. A resolution of 0.1 tons in a one-ton sample size is 10 percent. By comparison, the same resolution in a 25-ton sample is 0.4 percent. Generally, as a guide, the resolution should be in thousandths (0.001).
- Record the integrator total as the start value.
- Run material close to the normal flow rate or a minimum of 50 percent of design rate over the belt scale, weighfeeder or flowmeter for a minimum of five minutes.
- Stop the material feed and run the conveyor or pipe empty.
- Record the integrator total as the stop value.
- Subtract the start value from the stop value to determine the integrator total.
- Weigh the material sample if not already known.
A minimum of two material tests should be performed in order to observe the consistency of deviations, should they occur. If the tests are not consistent, investigate installation/calibration issues. If they are consistent, then a correction to the integrator should be made. Perform a final material test with the factor in place and verify that the system is within the required accuracy.
Verify span-calibration accuracies
There are two possible outcomes from a material test:
- If multiple material tests show a consistent deviation, then a span adjustment can be made.
- If the material test results are not consistent, there is likely an influence from installation.
Before performing further material tests, consider the following:
Idler alignment — using the instruction manual as a guide, confirm that the correct numbers of idlers are aligned both before and after the scale idlers. Scale-quality idlers should be used in high-accuracy applications.
Check belt tension by measuring the sag between the idlers. The sag should be more than 1 percent of the idler spacing, but less than 2%.
Scale location -— ensure that the scale is far enough away from any transition idlers, curves, pulleys, feed points, training idlers, or other vibration sources.
The conveyor structure must be suitable to support the load of the conveyor itself as well as the material it carries. If deflection in the structure occurs, supporting members must be added.
Load cell signal — when the scale has the normal loading on the belt, check the mV output of the cells compared to the dead load. If the difference is less than 3 mV, the load cell capacity may not be suitable.
Retention time — If the belt speed is high, then the material may not be on the scale long enough. The minimum required retention time is 0.25 seconds.
As the chart accompanying this article clearly shows inaccurate weighing is an expensive proposition.
Matt Morrissey is with the Industrial Automation Division of Siemens AG.
Siemens AG focuses on the areas of electrification, automation and digitalization. In 2014 it had more than 300,000 employees in more than 200 countries, generating revenues of nearly 72 billion euros.