Jimo A. Arasi
Phillips Petroleum Co.
Tananger, Norway
Laboratory tests on several commercially available orifice-plate meters for use in pulsating flow indicate that none yields acceptable accuracy.
These tests suggested, however, that if the objective of monitoring pulsating flow is to indicate or quantify pulsation magnitudes for comparisons, then at least two instruments are acceptable.
Use of such meters, particularly in low flow rate gathering systems, can be a viable alternative to attempting to reduce the intensity (amplitude and frequency) of pulsation by expensive installation and maintenance of chokes and bottles.
Phillips Petroleum Co. set out to find a meter that would be sensitive enough to measure pulsating hydrocarbon flows with acceptable accuracy using the orifice plate.
Several orifice measurement systems were simultaneously investigated at the Southwest Research Institute, San Antonio (SwRI).
DYNAMIC EQUIPMENT
Much has been written about the measurement of pulsating hydrocarbon flows using orifice plates. Several studies have occurred in the laboratories and field locations.
These tests have indicated that error exists when head meters (orifice or venturis, for example) are used to measure pulsating flow.1-5
Dynamic equipment, such as the reciprocating engines and control valves usually cause the pulsations. Certain static configurations such as the p (Pi)-shaped orifice meter station also experience induced pulsating flows as a result of turbulence.
A Pi-shape orifice meter station has closures at both ends through which the meter run can be cleaned.
Flow pulsations introduce errors in the measurement. These errors easily translate into settlement errors when purchases and sales of gases and liquids are made.
Several investigators have advanced methods to reduce the errors. Some methods involve reduction or elimination of the pulsation at the source. These methods include using the chokes and bottles which work satisfactorily with dynamic sources.
PULSATING FLOW
Pulsating flow is a continuous mass flow variation with a repeating wave form, usually containing components greater than 1 hz. Conventional differential-pressure instruments cannot adequately sense and record time-varying differential pressure whether pulsating or transient.
Typical pulsation sources are reciprocating and centrifugal compressors.
Reciprocating compressors create pulsations with frequencies which are a function of the revolutions and its multiples (harmonics). The amplitude of individual pulsation harmonic is controlled by its resonant amplification in the attached piping system.
Accordingly, the residual pulsations in the system are controlled by the total plant layout. These residual pulsations are in the range of 0.11.0% of line pressure.
For a line pressure of 300 psig, the residual pulsations fall in the range of 0.3-3.0 psig (about 8-80 in. of water at 68 F.). For a 100-in. orifice meter, these conditions could mean a pulsative intensity of 8-80%.
In the case of centrifugal compressors, pulsations occur at the blade-passing frequency which can be as great as 3,000 hz. However, these pulsations are attenuated faster than those produced by the reciprocating compressors.
In addition, flow-induced turbulence at junctions, restrictions, and discontinuities is another common source of pulsations.
Efforts to deal with the pulsating flow divide into four main categories:
- Isolation, attenuation, or elimination of the pulsations in the observed flows. Passive filters, such as volume tanks and in-line restrictions, have been used to reduce the pressure fluctuations to some tolerably low levels.
It is widely accepted that the practical solution to the pulsating flow problem is to dampen the fluctuations, Because in many cases pulsation also causes unacceptable pipeline vibrations, pulsation dampners (volume and choke arrangements) are usually part of the initial piping design.
- Pulsation quantification. Methods of quantifying pulsation, such as "pulsation number" or "threshold of pulsation," identify pulsation magnitudes. It is impossible to measure pulsating flows over these magnitudes.
The potential quantifiers include the Hodgson and Strouhal numbers.
The Hodgson number relates the volume and pipe length between the meter and the source of pulsations to the errors caused by the pulsations. The Hodgson number does not correlate well with the observed errors.
The Strouhal number, which historically relates to the shedding of Von Karman vortices, has not been successfully correlated with unsteady flow measurements. Besides, mathematical functions of the Strouhal and Hodgson numbers have been correlated with the pulsation errors, but practical results have vet to emerge.
- Pulsation factor. The concept of the pulsation factor, which is very much like the super-compressibility factor or the Reynolds number factor, has been advanced but as yet has seen little success.
- Pulsation flow meter. Other investigators such as Phillips have attempted to find a high-frequency response differential-pressure transducer that would measure the pulsating flow with acceptable accuracy.
In general, the investigation involves installing the potential and reference meters in a flow line, with approximately same pulsating or steady flow rates through them.
By comparing the flow rates at the potential and the reference meters, the investigator can examine the errors caused by pulsations and determine accuracy.
Before the test is described, it should be pointed out that there are other errors which relate to pulsating flows. These are gauge line errors and inertial errors.
These errors are not addressed in this investigation. Several other investigators have published the effect of these errors on the pulsating flow measurement.
Investigators use the square root error (SRE) to characterize pulsating flows. Suffice it to say that the error is always positive when measurements are made at the flanges of the orifice plate.
"True" flow rate is less than the measured or indicated flow; "true" flow is the flow without pulsation error. It can also be demonstrated mathematically that the true flow is less than the indicated low.
The accompanying box with equations offers a mathematical explanation of this phenomenon.
TEST CONFIGURATION
The objective of Phillips' test was to compare several meters over different test conditions of pulsating flow to determine which was accurate and usable. Instruments included in the investigation were the following:
- Rosemount differential pressure transmitter
- Validyne differential and static pressure transmitters
- Applied Automation Inc. (AAI) flow computer with internal differential and static pressure transducers
- Daniel Square Root Error Indicator (SREI)
- Barton mercury chart recorder.
The tests were conducted at the Southern Gas Association (SGA) gas flow facility located at the Southwest Research Institute (SwRI) in San Antonio (Fig. 1).
The facility is a closed loop in which a positive displacement (PD) compressor circulated the flow. The loop's pressure can be raised independently of the flow rate or the compressor load. The maximum rated working pressure of the test facility is 190 psig.
The reference meter run was protected at both ends by acoustic filters. These filters isolated the pulsations in the rest of the flow-measurement system. One of the filters had a cut-off frequency of approximately 2 hz. It consisted of two vessels (A and B) with a choke connecting them. The filter protected the upstream end of the reference meter run.
The reference meter run consisted of a parallel combination of 6 in. and 4 in. orifice meter runs located indoors. For the test, the 6 in. run was not used and was isolated.
The 4 in. run with a 2.818-in. orifice plate was used. The second filter, Vessels C and D, and a connecting choke completed the isolation of the reference section from any pulsations in the rest of the flow facility.
The meter run satisfied the requirements of AGA Report No. 3. The run could be isolated by two butterfly valves in the event of emergencies.
The pulsator consisted of a set of stationary and rotating 0-pitch blades. The rotating-blade assembly was arranged so that it could block more or less of the flow cross section. The rotational speed was controlled from outside the pipe.
The pulsation frequency range was 2-100 hz. The pulsation amplitude depended on flow rate and test section acoustics. It could be varied to higher or lower levels.
The test meter run was a Schedule 40, 300 lb ANSI orifice with two pairs of flange taps. The AAI flow-computer transducers and SwRI dynamic transducer were connected to a pair of taps.
The Rosemount and Validyne transmitters were connected to the second pair of taps. In addition, a pair of block valves and flexible pressure tubing connected this pair of taps to the Daniel SREI transducer and the mercury chart recorder (Fig. 1).
The Daniel SREI recorder and the chart recorder were on line continuously. The recorder was used only during a few test cases.
The test meter run pressures were measured upstream and downstream of the orifice for monitoring the pulsations. The temperature for each meter run was measured with a type T thermocouple inserted directly into the gas stream. Each thermocouple had an electronic 0 junction and analog readout.
The instrument outputs were connected to a data-acquisition microcomputer. Table 1 shows the hook-up configurations.
The data collection system was programmed to sample each channel at a suitable frequency and to average the data over 16 samples. On selected channels the amplitude of the dynamic variation was determined.
CALIBRATIONS
All the transducers were calibrated before their installations, The static-pressure transmitters - including SwRI's Validyne, the test Validyne static transmitter, AAI pressure transmitter-were zeroed and set at full scale against a deadweight tester. Zero and span were repeatable within 0.1% of full scale.
The differential-pressure transducers were all calibrated with a pneumatic deadweight tester with calibration weights in inches of water. Zero and span were repeatable with is 0.1% of full scale.
The thermocouples were calibrated in a water bath. Accuracy was 1 F. within the normal test ranges and over an extended period of time. The barometric transducer was compared to a laboratory-type measuring barometer with ambient temperature compensation.
AAI's flow computer and Daniel SREI were also calibrated. The dynamic-pressure transducers that were installed upstream of the orifice had a stable sensitivity to pressure fluctuations so that the amplifier setting was used on the calibration setting for then.
The reference orifice meter run was thoroughly cleaned. The run was measured and certified at 4.036 in. ID.
The microprocessor data-acquisition program for the input data from the transducers was scaled. Initially, each channel was fed with 0 input. The input was increased to full scale or a large known value.
The data-acquisition system read the upper value. With the zero and the upper values identified to the program, the digital sensitivity was determined.
TEST PROCEDURE
A flow rate was established through the loop with the test gas being nitrogen.
Data from the reference and meter were recorded digitally. This occurred at first with no pulsations and then with a low level of pulsation at a selected frequency.
The frequency selection procedure is described presently. At each selected frequency the flow was stabilized before the data were collected. The frequency was within the range 2-80 hz.
The pulsator position was varied in steps of approximately 25% (i.e., 25, 50, 75, 100%). In addition to the automatically collected data, the SRE was read from the Daniel SREI.
AAI's flow computer calculated flow rate and quantity with its own hardware inputs and internal software.
During each test, 16 samples of each data signal were taken and averaged to yield a single data value. For samples of differential pressure signals from the reference and test orifices, the square root values were also obtained and averaged.
For the differential-pressure signals, both the average of the data samples and the average of the square root of each sample were saved. The averages of each static pressure and temperature were also saved.
The AAI flow computer was started and carefully timed while the test was under way and the corresponding dynamic data were being collected. At the end of the timed period, the flow computer totalizer was stopped and a printed report requested.
FREQUENCY SELECTION
The frequencies at which technicians conducted the tests were selected to represent certain dynamic flow conditions at the test orifice connections. With the pulsator at the middle position and a moderate flow, the frequency was varied fairly rapidly from about 2 hz to approximately 80 hz.
During this frequency sweep exercise, a spectrum analyzer monitored the static pressure and differential pressures. The differential-pressure spectrum showed peak responses at 16.8, 28.0, and 65.0 hz. The minimum responses occurred at 20.8 and 36. 8 hz.
The peak responses in the differential-pressure spectrum indicated maximum flow modulations at the orifice connections. In these cases the orifice upstream and downstream static pressures are out of phase; upstream static pressure is high and the downstream static pressure is simultaneously low, thus the difference in modulation is high.
The frequencies corresponding to the peak responses were selected for the test. The minimum responses in the differential pressure showed the situation in which the orifice upstream and downstream static pressures are in phase; upstream and downstream static pressures increase or decrease together; thus the difference in modulation is low.
The frequencies corresponding to the minimum responses were selected for the test. Additional frequencies, 8 hz and 6.4 hz, for example, were also selected.
As a matter of information, Phillips could have pursued the frequency spectrum analysis further and developed transfer functions for the system using the techniques of non-linear control system engineering.
We did not derive any transfer functions from the test results because such derivations were irrelevant to the objective of the test.
Table 2 shows the series of flow conditions.
TEST RESULTS
Information contained on a test printout appears in Table 3.
The density of nitrogen was calculated. Key parameters such as the beta ratio, pipe Reynolds number, and the square root error - which was calculated from the differential-pressure signals-are also included.
The corresponding SRE as shown by the Daniel SREI occurs at the bottom of Table 3.
AAI's flow computer printouts presented the total volume measured during a 10-min test period. Table 4 shows information from a sample AAI computer printout.
It should be pointed out that the calculation of the SRE from the values of the stored averages of the squared root of the differential pressure follows from an accepted method of SwRI.
The calculation in the equation is shown in the accompanying box. Accordingly, a computation procedure which uses the average value of differential pressure and takes up the square root will contain the SRE in the calculated flow rates. The SRE is not contained in any of the calculated flow rates to which the transducer responds adequately with the dynamic signals.
The result also shows the total error which consists of the difference between the reference flow at the test condition and the initial reference flow at the corresponding nonpulsating condition and of the SRE measured by the transducer.
Because of calibration errors, offsets, or bias, an instrument may indicate a difference from the reference meter that is not due to pulsations. If this relationship to the reference flow changes during a test series, then a pulsation effect is indicated.
Thus the total error is the change in indicated flow as a result of pulsation including the SRE effect as measured in comparison to the undisturbed reference flow.
Observations specific to the transmitters and derived from the entire test results appear in Table 5.
GENERAL CONCLUSIONS
If the objective of monitoring the pulsating flow is to indicate or quantify pulsation magnitudes for comparisons, then the Daniel SREI or a dynamic transducer such as the Validyne will suffice. These instruments can be used for instantaneous or trend assessments.
If the objective is to measure a pulsating flow accurately, however, then none of the meters investigated in this study are recommended. Particularly in the cases of pulsating custody transfers it will be difficult to prove such a pulsating flow meter. It will be more appropriate to eliminate or isolate the pulsation.
If this cannot be achieved, then an acceptable minimum accuracy should be specified in the contract. The capability of a meter to respond fast and dynamically to track pulsation is necessary but not sufficient.
It should be possible to calibrate and prove such a meter or metering system with respect to an industrially acceptable or traceable system.
ACKNOWLEDGMENT
Assistance for this article was received from Messrs. Robert J. McKee, Phil J. Kruger, SWRI, and Barry Tate, AAI.
REFERENCES
- Gregg, D., "Orifice-meter measurement errors caused by gas-system pulsation can be controlled," OGJ, Oct. 16, 1989.
- Miller, R.W., Flow Measurement Engineering Handbook, McGraw-Hill Book Co., 1983.
- Mottran, R.C., and Mohammed, W.A., "High Frequency Pulsation Effects on Orifice Meter Accuracy," International Conference on Advances in Flow Measurement Techniques, Sept. 9-11, 1981.
- Oppenheim, A. K., Chilton, E. G., "Pulsating Flow Measurement-A Literature Survey," Transactions of ASME, February 1955, pp. 231-248.
- Sparks, C. R., "Recent Research Results on Pulsative Flow Measurement Using Orifice Meters," AGA Distribution/ Transmission Conference, May 1985.
Copyright 1992 Oil & Gas Journal. All Rights Reserved.
