In the flow control industry, we hear a great deal about accuracy, repeatability and other means of establishing performance for a flowmeter or a valve, or other flow control device. Unfortunately, most of us are at somewhat of a loss to decode much of the information they are receiving, and so they have the unfortunate experience of either “underbuying” or “overbuying” their flow control products.
Accuracy, according to the ISA Dictionary of Measurement and Control, is “The ratio of the error to the full-scale output or the ratio of the error to the output, as specified, expressed in percent. …(the) degree of conformity of an indicated value to a recognized accepted standard value or ideal value …. The degree to which an indicated value matches the actual value of a measured variable….” So, the accuracy of a measurement is how well it matches the actual value being measured.
Accuracy as Shorthand
Accuracy, or rather, “error,” is measured many ways. Since the accuracy of a measurement is rarely linear, or equal across the entire range of values being measured, accuracy is usually given in a form of shorthand. A flowmeter can be said to be accurate to ±1 percent of full scale. Or, another device can be said to be accurate to ±1 percent of indicated flow rate. Sometimes, manufacturers’ literature simply says 1 percent accuracy, and neglects to specify the “error band.” This is often misleading, sometimes intentionally so. A flowmeter that is accurate to ±1 percent of indicated flow rate is generally much more accurate than a flowmeter that is accurate to ±1 percent of full scale. Let’s say that the flowmeters both are set up for the same full scale: 100 GPM. At 10 GPM, the flowmeter whose accuracy is given in “percent of indicated flow rate” will have an error of ±1 percent of 10, or 0.1 GPM. The flowmeter whose accuracy is given in “percent of full scale” will have an error of ±1 percent of 100, or 1 GPM. Typically, the lower the actual flow rate, the more error a flowmeter whose performance is given in “percent of full scale” will have.
There are other ways to fudge and nudge accuracy specifications, also. For example, a velocity-based flowmeter may have a single measuring range, 0 to 33 feet per second velocity. Regardless of what the output is scaled to, if this flowmeter is specified to be accurate to “one percent of full scale” it is the measuring scale, not the output scale that the accuracy refers to. So, in a situation where the output is scaled to 100 GPM maximum, but that flow rate is equivalent to a velocity of 3.3 feet per second (1 meter per second), the error is still ±1 percent of 33 feet per second (10 meters per second) or 0.33 feet per second. If 100 GPM is equal to 33 feet per second, 10 GPM would be equal to 3.3 feet per second, and 1 GPM would be equal to the entire error band, ±0.33 feet per second. Even at maximum output, this flowmeter is much more inaccurate than a flowmeter whose accuracy is specified in “percent of indicated rate.”
The good news is that in most process control cases, absolute accuracy in a flowmeter is not a critical requirement. In some cases it is, but in the vast majority of cases, it is repeatability that is the critical issue, not accuracy.
The Repeatability Factor
Repeatability, again according to the ISA Dictionary, is “The ability of a transducer to reproduce output readings when the same measurand value is applied to it consecutively under the same conditions, and in the same direction.” If you notice, this is much more important to controlling a process than absolute accuracy is. Repeatability is about how well the flowmeter tracks process flow changes. This means that you can use a flowmeter that is repeatable, but not very accurate, to control a process, simply by calibrating out the inaccuracy. A flowmeter that is accurate to ±10 percent of indicated flow rate, but is repeatable to ±1 percent can easily be used in control, while it may not be valuable in custody transfer applications.
The most important thing to remember is that it is the accuracy of the complete system that matters. For example, many wastewater treatment gas chlorination systems are controlled by magnetic flowmeters with accuracies ranging from 0.5 percent to 1 percent of indicated flow rate. Yet, the standard for the accuracy of the indicating rotameter and the dispensing valve in the chlorinator itself is ±4 percent of full scale. Selecting the magnetic flowmeter with the highest accuracy over one with less accuracy, and lower price, is meaningless because the system error is controlled by the inaccuracy of the dispenser itself. The same is true in many process control situations.
So what? Use accuracy and repeatability for what they are: measures of error and consistency in measurement. If you need very little absolute error in your measurement, go with the highest accuracy flowmeter you can find that will fit the application. If you need consistent readings, go with a highly repeatable flowmeter that may not be the most accurate device you can buy. Always make sure you understand the manufacturer’s specifications, though, so you can be sure you are buying apples instead of onions.
Walt Boyes is a is a principal in Spitzer and Boyes LLC, offering engineering, expert witness, development, marketing, and distribution consulting for manufacturing and automation companies.