01 August 2003
All steamed up
Cone differential pressure meters serve as metering standard.
By Kenneth Lloyd, Brian Guthrie, and Dr. R.J.W. Peters
The University of Iowa has a steam distribution system that includes over 158 flow measurement points (2 inches to 24 inches) throughout the university. These metering sites are used to bill the departments for the steam taken from the main campus power plant. The plant can produce 20 megawatts of electrical power and 400,000 pounds per hour of steam. Pressure-reducing stations and de-super heaters supply the steam for central heating for the entire university—sending steam to the distribution system at 150 pounds per square inch, gauge (psig) and 410°F and at 20 psig and 330°F. Orifice plates and differential pressure (DP) transmitters measure the heating steam flows at the power plant in each of the 12-inch to 14-inch diameter headers.
The existing distribution metering system consisted of vortex meters from the same manufacturer—insertion, wafer, and flanged varieties. The university questioned the accuracy and repeatability of the meters, which were installed in 1995; the meters were also a high-maintenance item.
In 1998 the university invited Control Application and Maintenance (CAM) to undertake a metering study with their staff, including an assessment of several meter sites to determine if the meters were installed correctly. They also wanted to investigate the data collection system and look at the computational system used to derive the mass flow and British thermal unit consumption. At that time it appeared the monthly totals created inconsistent line losses relative to the power plant output. An uncertainty existed between the site totals and the reading at the power plant. CAM and the university staff identified some installation and calibration problems at the power plant and throughout the facility.
At the vortex-meter manufacturer, technicians used a 6-inch square wind tunnel for testing and certifying their insertion-type vortex meter. With this wind tunnel, they established the shift in the calculated Strouhal number, which is relative to the frequency per foot per second for that particular vortex meter. In the wind tunnel they measured the atmospheric velocity with a Pitot tube.
CAM, during the tests at the manufacturer, concluded that the University of Iowa had not been accounting for the reduced cross-sectional area of the steam pipe, caused by the installation of the vortex meters in different sizes of pipes—ranging from 6 inches to 24 inches. Not correcting for cross-section area reduction meant the vortex meter, which strictly measured a localized velocity in feet per second, was giving incorrect and high velocity readings of volume flow rate.
Most issues CAM and the university staff identified were corrected, but at the completion of this work, the metering discrepancies still existed. CAM told the university the meters were having problems metering steam flow and something was changing in the meters to cause the inconsistent and high flow readings. The only way to verify this possible application problem was to test the existing meters in a flow-proving lab.
Ohio test data
The university did a search for a steam-flow test facility and found one in Ohio. Researchers prepared a testing schedule for the university meters and fabricated test piping in 2-, 4-, and 6-inch sizes to use with the meters at this test facility.
The Ohio test facility tested boiler steam drum pressure and relief valves and used a condenser and weighing system to determine the mass flows from the steam. The facility also used a small boiler to produce steam at a set pressure.
The steam flows through the pipe section and meter under test. Following the meter, technicians condense and weigh the steam to give the mass flow at the end of a test run. However, the boiler was not large enough to produce steam flows at the higher velocities desired for the 6-inch insertion meter test. Also, using the facility for extensive testing was cost prohibitive.
CAM helped evaluate the Ohio test data, which they data logged in ACCESS format with a series of 1-second readings taken over the 6-minute test runs. These readings were of flow velocity in feet per second, pressure, and temperature. They collected the average velocity, pressure, and temperature over the test period and used the macro from the university to calculate the mass-flow total over time. Then they compared this to the mass flow the test facility measured. CAM produced a spreadsheet to account for the change in cross-sectional areas, and this greatly improved the accuracy from the Ohio test data. But high flow-rate errors still existed.
CAM contended the tests were not valid because of the short run time, and in the case of CAM's data, the use of average velocity to equate the flow totals. Researchers should have continuously totaled and run test runs over a longer period to gain better test results. Due to the limitation of the tests, CAM suggested retesting with running total and longer runs with better defined test procedures to gain more consistent and accurate results. The university staff revised the procedures after consulting with CAM.
Developing a steam-flow test cell
Due to the inadequacy of the result from the Ohio and manufacturer's test facilities, CAM recommended the following:
- Test the meters over the complete flow range of the meter.
- Improve test procedures to include longer run times.
- Vary flow rate to test for repeatability.
- Total the individual readings in real time with the errors calculated and evaluated.
CAM did an Internet search and made inquiries with associations such as ISA, the National Institute of Standards and Technology (NIST), the American Society for Quality (ASQ), and the American Society of Mechanical Engineers (ASME). The university did not find a test site to fulfill these requirements using steam as the test medium. Consequently they decided that given the distance, difficulties, and cost of using the Ohio test cell, an adequate source of steam and building space for the project was available at the university.
CAM provided a test cell master meter recommendation and proposal. The test cell would use the master meter technique—a primary device against which the meters under test would be compared.
The power plant uses orifice plates and standard DP meters to measure the steam sent to distribution. It is known that the orifice is not ideal in steam applications, where condensate can occur. It will build up against the plate even when one provides drain holes. The orifice plates also lack turndown capabilities. However, differential devices have the advantage of no moving parts and are known to have long-term reliability in steam applications.
The master meter would require calibration to NIST standards and have at least a 10:1 turn-down ratio. The uncertainty of the meter must be better than 1% over the 10:1 turn-down range. The transmitter must also perform the mass-flow calculation. An orifice plate could not meet all these requirements. A cone-type meter combined with an intelligent DP transmitter met these requirements and (due to the fact that there is no obstruction on the bottom of the pipe wall) any liquid in the pipe passes through the meter and will have minimum effect on the meter-reading accuracy. We noted the cone had been used extensively in steam applications.
The test cell was very useful for evaluating meters for the custody transfer of steam. It was also an economical method; it reduced handling charges as well as the direct cost of undertaking external tests. The system uncertainty was a little higher at 2% than the initial design aim of 1%. The new evaluation matrix is potentially an effective tool for establishing the best meter for the application. The final conclusion: the best overall meter was the cone with the intelligent DP transmitter. The downside of the cone is the initial equipment cost, long lead-time for delivery, and higher installation costs when compared to finding a metering system to reuse existing insertion meter penetrations. P
Behind the byline
Kenneth Lloyd is a professor at the University of Iowa, Brian Guthrie is president and owner of Control Application and Maintenance, and Dr. R.J.W. Peters is a flow measurement technology manager at McCrometer in Hemet, Calif.