Figure 1. The meter (white) is oriented at zero degrees approximately one diameter downstream of the pipe bend. The pipe bend is blue. PVC sections are dark grey, and the Vanstone flanges are light grey. The schedule 10 IPS pipe is white.

By Jeffrey T. Peery

Flowmeters are often installed without the required minimum length of straight pipe to fully develop flow. In such installations, meter accuracy may be adversely affected, but little is known about the magnitude of effect. For this reason, SeaMetrics (www.seametrics.com) performed experiments to quantify the accuracy of a magnetic flowmeter installed downstream of an elbow.

Methods
A six-inch magnetic flowmeter was installed downstream of an 89-degree, in-plane sweeping elbow with a radius of curvature equal to approximately one diameter (Figure 1). The meter was installed at zero, one, two, three, and four diameters in length downstream from the elbow. At each location the meter was rotated to zero, 90, 180, and 270 degrees about the pipe axis (Figure 2). At each angle the meter was calibrated at 35, 50, 500, and 900 GPM. Meter error was evaluated against the straight-pipe condition (where the flow field was assumed fully developed and steady).

Figure 2. The meter is positioned at 180 degrees about the pipe axis. The positive direction is clockwise.

The downstream pipe was initially composed of five sections, each one diameter in length. Each section was attached via flanges that allowed the meter to be rotated in place. Between calibrations, one section was removed from the downstream side of the meter and installed on the upstream side. In this way, the upstream length increased by one diameter between tests and the downstream length decreased by an equal amount.

Results
At zero diameters, accuracy was improved from approximately 2.5 percent of rate to 0.5 percent of rate by rotating the meter in place. At all distances, optimum performance was achieved at zero and 180 degrees (Figure 3). At 90 and 270 degrees, meter accuracy was at its worst and improved as the distance between meter and elbow increased (Figure 4). At zero and 180 degrees meter accuracy was at its best and was less sensitive to the distance between meter and elbow.

Recommendations
Accuracy was evaluated for the installed magmeter at various distances from an elbow and angles of rotation about its axis. Accuracy was optimized by rotating the meter to either zero or 180 degrees. Its performance at zero and one diameter may be approximated from Figure 3.

Figure 3. Each plot displays contours for the shift in output. The y-axis variable is the angle about the pipe axis, and the x-axis variable is flowrate. Large magnitudes are red, and low magnitudes are blue. Contours levels are labeled.

Installations at zero diameters from the pipe bend had different contour shapes and worse accuracy than other locations. This was expected because the elbow distorted the flow profile. Also, at zero diameters the meter was installed between ferrous and nonferrous pipes (PVC and iron). The difference in materials may have distorted the magnetic field and further affected meter output.

Rotating the flowmeter should have produced no change in output if its magnetic field and the flow profile were symmetrical. However, at a given flowrate the meter’s output did change when it was rotated. The cause of the change is related to the weight function [2], a description of meter sensitivity to flow at particular locations within the magnetic field. Typical magnetic flowmeters do not have uniform weight functions and are sensitive to the locations of flow distortions. By moving these distortions (i.e., rotating the meter) it is possible to optimize performance. It is unknown whether the optimum angles presented herein are specific to the particular installation.

Figure 4. An illustration of the shift in meter output at 180 degrees angle of rotation and at various distances from the elbow.

Each plot in Figure 3 was created from 12 points (three points of rotation about the axis per four flowrates). These points were curve-fit and triangulated to produce a smooth two-dimensional plot. Resolution of the plot is low, and small details may not exist. Data presented in Figure 3 should be considered an approximation.

References
1. R. C. Baker and J. Hemp and J. E. Heritage and M. V. Morris and J. C. Coulthard, “Installation Effects Electromagnetic and Ultrasonic Flowmeters,” Cranfield Institute of Technology Department of Fluid Engineering & Instrumentation, Teeside Polytechnic School of Information Engineering, 1988.
2. J. A. Shercliffe, “The Theory of Electromagnetic Flow-Measurement,” Cambridge University Press, London, 1962.

Jeffrey T. Peery is a mechanical engineer for SeaMetrics Inc., a designer and manufacturer of mechanical and electromagnetic liquid flowmeters. His work focuses on flowmeter research and development, design, and experimentation. Mr. Peery earned bachelor’s and master’s degrees in mechanical engineering from the University of Washington. He can be reached at jeffpeery@seametrics.com or 253 872-0284.

For More Information: www.seametrics.com