Flowmeters typically contain multiple components that introduce error into the flow measurement system. A simple flow measurement system may be comprised of a primary flow element and a transmitter that processes signals from the primary flow element. Sometimes, the primary flow element and transmitter are physically integrated together as one piece, such as in potable water meters. More complicated flow measurement systems may include multiple components such as a flow computer or other electronic components that compensate for process pressure, process temperature, or other parameters.
It should not be forgotten that flow measurement systems are “systems” that measure flow. As an example, consider a hypothetical primary flow element that exhibits no error while the transmitter exhibits 5 percent accuracy. In this exaggerated example, the accuracy of the flow measurement system will be 5 percent. Assuming that the flow measurement error is that of the primary flow element only is an error of omission. Users should constantly be on guard to identify this type of error.
In most flowmeters, the primary flow element and transmitter are integrated electronically. For example, the wetted primary flow elements of Coriolis mass flowmeters, thermal flowmeters, and magnetic flowmeters are virtually useless without transmitters that contain their respective flow measurement algorithms and drivers. Therefore, flowmeter performance typically includes the combination of a primary flow element and a transmitter. Further, the performance of most flowmeters is predicated on the calibrated output that is usually the pulse/frequency output of the transmitter.
However, most process control applications of flowmeters involve the use of an analog output such as 4-20 mA to represent 0-100 percent of the desired flow rate. The analog signal is typically generated using circuits that convert the pulse/frequency signal (or its source) to an analog signal. This conversion introduces a measurement error that is constant throughout the signal range, so it can usually be expressed as a percent of full scale. The error introduced is typically between 0.03 and 0.10 percent of full scale depending upon the quality of the converter. To obtain the measurement accuracy of the analog output, this error is mathematically added to the accuracy of the flowmeter.
The analog output error may seem small, but at low flow rates, this error can become significant and actually dominate measurement accuracy. For example, consider a vortex shedding flowmeter that can operate from 10 to 100 units per minute with 0.75 percent of rate accuracy but has an analog output accuracy of 0.10 percent of full scale. At 10 units per minute, the pulse/frequency output has an accuracy of 0.75 percent of rate, whereas the analog output contributes an additional (0.1*100/10) or 1.00 percent rate error, so the measurement accuracy of the analog output is 1.75 percent of rate.
Most suppliers calibrate the pulse/frequency output. They typically state its accuracy as the performance of the flowmeter. The accuracy of the analog output conversion is often buried in the specifications in the fine print. Sometimes, it is not published and must be requested from the supplier. Sometimes the information is forthcoming, but often suppliers do not understand the question and try to state the analog output resolution (say 1 part in 4096, or 0.02 percent) as the analog output accuracy. After further investigation, many suppliers will admit that they do not know the analog output accuracy — even though most of their customers may use that output exclusively for their flow measurements. They also provide further enlightenment when they say that “no one ever asked for this before”.
The burden of obtaining the best flow measurement possible in a given application does not lie with the supplier — it lies with the user. Do not forget the fine points that may lurk in the details and the errors of omission that may be available for the asking.
This article originally appeared in Flow Control magazine (September 2004) at www.flowcontrolnetwork.com.