Flow Control: Technology information & new products for fluid handling engineers

May 2006

Back to List | Print Article

A Level Playing Field
Finding a True Measure of Measurement Error?

David W. Spitzer, P.E.

You would think that finding a common denominator by which to compare instrument performance would be straightforward, but it’s not. For example, an event occurred at a petrochemical plant whereby a compressor tripped as a result of high liquid level in a vessel upstream of the compressor. Investigation revealed that not only was the level transmitter not in service, but that an independent high-level switch had failed to operate. The compressor eventually tripped when the high-high level switches sensed that the level in the vessel was excessively high.

One of the recommendations in the report was to replace the high-high level switches with continuous level transmitters that utilized different measurement technologies. A subsequent report evaluated level technologies for the application and provided a comparative tabulation of the technologies with recommendations of which technologies should be used. A third set of documents contained the specifications for the proposed level instruments. My job was to validate this body of work.

The performance specifications in the second report were quite interesting. The stated accuracies were 0.075, 0.25, 1.0, and 0.25 percent for the four technologies evaluated. Further investigation revealed that each percentage was referenced to different parameters, such as a percentage of calibrated span, span, or a percentage of the maximum measurement range. These can represent different errors, depending upon technology and the actual instrument selected.

While it may seem that the percentages represent a valid comparison, it would make more sense to state the accuracies in terms of its measurement error in millimeters. However, these calculations can take a bit of work and analysis to perform. Ass- uming the measurement range is one meter and that the error associated with the first three instruments are percentages of full scale, their accuracies would be 0.75-10, 2.5, and 10 mm assuming the liquid density can vary by 1 percent from nominal while neglecting pressure/temperature effects (see table below). The error associated with the last instrument would be 12.5 mm, assuming it is expressed as a percentage of the maximum range that is five times the measurement span.

Tabulating these results makes it clearer that the stated performance percentages were not necessarily indicative of actual performance in this instance. Taking the time to evaluate instruments on an equal footing can help you achieve superior measurements that can make a difference in some applications.

David W. Spitzer, P.E., is a regular contributor to Flow Control. He has more than 25 years of experience in specifying, building, installing, startup, and troubleshooting process control instrumentation. He has developed and taught seminars for almost 20 years and is a member of ISA and belongs to ASME, MFC, and ISO TC30 committees. Mr. Spitzer has published a number of books concerning the application and use of fluid handling technology, including the popular The Consumer Guide to... series, which compares flowmeters by supplier. Mr. Spitzer is currently a principal in Spitzer and Boyes LLC, offering engineering, product development, marketing, and distribution consulting for manufacturing and automation companies. He can be reached at 845 623-1830.

www.spitzerandboyes.com

Back to List | Print Article

Get a FREE Subscription to Flow Control magazine | Subscribe to the Flow Control E-Newsletter

Site Developed & Maintained by the Pinnex Group, LLC

Privacy Policy