Effect of a 90° Elbow on the Accuracy of an Insertion Flowmeter, Results and Comparisons for 4 and 6 in. Diameter PVC Pipe

Abstract

Thermal energy consumption in buildings with chilled or hot water distribution systems is often monitored through the use of some type of flow metering device. These flowmeters can be fixed types, such as venturis or orifices, or insertion flowmeters which can be more easily installed and removed. The easy removal and reinstallation of the insertion type flowmeters makes them good choices for use in existing buildings or in retrofit projects. Besides the installation benefits, insertion flowmeters can also be installed while the pipe is in service or ''hot tapped". With any type flowmeter however, location in the pipe is a critical problem and deserves special consideration. Ideally, the meter should be inserted in existing pipe with a minimum of 10 to 15 diameters of straight pipe upstream of the meter location. This is rarely the case in existing piping distribution systems. It is much more common to be faced with only one or two candidate metering locations and these often are very short straight runs or will have elbows upstream and downstream of the proposed metering location. This paper reports on flow measurement error resulting from an insertion flowmeter installed downstream of a 90° elbow. The measurement errors were compared for tests conducted in 4.0 and 6.0 inch (0.1 and 0.15 meter) diameter PVC pipe. The insertion flowmeter was a nonmagnetic, tangential paddle wheel type. The flowmeter was located from 2 to 10 pipe diameters downstream fiom a 90° elbow with fluid velocities ranging from 1.0 to 10.0 ft/s (0.3 to 3.0 m/s). At each flowmeter location, the meter was rotated in 45° increments around the circumference of the pipe to quantify the effect of circumferential location on flow error. The flowmeters were tested at the energy metering calibration facility at the Texas A&M University Energy Systems Laboratory Riverside campus. Flowmeter output was compared to mass flow measurements obtained 6om precision load cells mounted beneath a 1342 ft^3 (38 m^3 ) weigh tank. All output is given in terms of percent error relative to the load cells. Final results are presented as a bction of flowmeter downstream location, circumferential rotation angle, and fluid velocity. Circumferential meter location was found to be a very important factor. The percent difference for the tested flow meters ranged 6om -23% to -5% in the 4.0 in. (0.1 m) pipe and 6om -33% to 1% in the 6.0 in. (0.15 m) pipe. The ''best" location for these flowmeters was at zero degrees rotation angle, regardless of pipe size or meter location relative to the upstream 90° elbow.