Professional Documents
Culture Documents
022 Ultrasonic Meter Donde No Usarlos
022 Ultrasonic Meter Donde No Usarlos
After almost a decade of research and development, in 1986 British Gas (BG) licensed Daniel
Industries to manufacture gas ultrasonic meters according to their patented technology. The BG
experience was that, for pipeline applications, these new meters offered many advantages over existing
devices, such as higher turndown, no pressure drop, greater immunity to installation effects, and more.
A new era in gas flow measurement was born.
By 1990, however, measurement specialists whose realm of interest was farther upstream had
begun to take notice of this new technology. Those whose responsibility was fiscal measurement on the
production platforms of the North Sea were wondering if these advantages could be brought to bear in
environments much closer to where the gas was produced, where space for metering skids was tight, the
operating temperature was high, and the gas often contained entrained liquids. These measurement
engineers faced daunting problems in measurement on a daily basis. “Why does the measurement
equipment take up half the space on the platform?” “Why are we measuring gas flows today that are
20% higher than they were yesterday?” (“Yes, yes, I know the test separator isn’t working perfectly, but
that’s life offshore, isn’t it?”) Facing what seemed a never-ending series of issues, these engineers were
keen to try new technology which might relieve some of these burdens. It was in this environment that
Project UltraFlow was conceived.
Project UltraFlow was simply the first attempt to push the application of gas ultrasonic meters
beyond the place where experience and standards said they should be used. While their use was taking
off in pipeline transmission applications, where documents such as AGA-9, ISO/TR 12765:1997(E), and
BS-7965 suggested how the meters should be installed and used, there has never been a definitive work
which could be used to describe how to proceed if the application were not so pristine. While this paper
is no such definitive work, it will describe how these meters have been used in unconventional ways,
and will suggest other ways they might be used to advantage in the future.
1
First Attempts to Measure Wet Gas – Project UltraFlow
The first meeting of the Project UltraFlow Wet Gas Joint Industry Project (JIP) was held at the
National Engineering Laboratory (NEL) in the autumn of 1992. Concurrent to this project another
project was conducted to investigate installation effects on the meters (the “Dry Gas” project). The
original members of the group were British Gas, British Petroleum (BP), Shell ExPro, Phillips
Petroleum, Nederlandse Aardolie Maatschappij (NAM), Amoco, and Daniel Industries. NEL was used
as a contractor for much of the work, and the U.K. Department of Trade and Industry (DTI), Oil and Gas
Measurement Branch was an active participant in the most of the JIP’s activities.
The goal of the UltraFlow Wet Gas Project was to test a variant of the Daniel ultrasonic meter,
the only commercial multipath meter available at that time, under a variety of controlled conditions of
wet gas, first at NEL and subsequently in a field location. Specifically, the desire was to demonstrate an
accuracy of 1% of reading in a wet gas environment of up to 0.1% free liquid, typical of conditions
downstream of primary separators. An additional goal was that the meter would be designed to resist
corrosion, and that it would automatically recover from the liquid flooding which would inevitably
occur in these applications. Finally, it was intended that the results of the project would convince the
regulatory authorities that the meter was suitable for allocation applications in wet gas service.
The development and testing took place over three years, a period which could have been halved
if not for flow facility breakdowns. Reference 1 (Wilson, 1996) is the best source of information on the
history of the project and the testing discussed here. The meter selected was a six-inch, four-chord
device of the British Gas design, but using transducers which were smaller and were hardened for wet,
high-temperature service. A diagram of the arrangement of chord paths is shown in Figure 1.
Figure 1. Schematic of Multipath Meter used for UltraFlow Wet Gas Project.
The testing of the meter at NEL used their circulating wet gas flow facility in its early years of
operation. Very good control of the gas and liquid phases was possible, with Liquid Volume Fractions as
high as 5% and gas flow velocities of up to 20 m/sec. Its main shortcoming was the fact that the fluids
2
used were air and water, which guaranteed that the only flow regimes possible were stratified or annular.
Even at 20 m/sec, the gas was unable to carry liquid droplets in a dispersed manner for mist flow.
In contrast, the fluids flowing through the bypass test facility at Shell’s Bacton terminal were all
light hydrocarbons, i.e. natural gas and condensate. At low flow velocities, e.g. 5 m/sec, stratified flow
similar to that which was seen at NEL was observed. However at flow velocities of 10 m/sec and above,
the droplets of liquid were clearly being carried along at the same velocity at which the gas was moving,
thus mist or dispersed flow was in evidence. This can be perceived in Figure 2, where the over-reading
of the gas flow at NEL and Bacton is plotted against the Liquid Volume Fraction (LVF). Since the over-
reading is due to the presence of liquid in a gas meter, in mist flow conditions a 1% liquid loading yields
a 1% error in gas measurement. However, when the flow is stratified, the liquid moves at a speed which
is slower than the gas, thus the presence of liquid creates a gas measurement error which is typically
four to five times as great as the LVF. These distinct flow regimes can easily be discerned in Figure 2.
3
Figure 3. Recovery of UltraFlow meter after injection of 1% water.
Figure 4 is a chart that shows the meter’s accuracy at Bacton over a nine-month period during
1993-94, as well as notes describing the events which took place during that period.
190
180
170
160
130
120
110
100
0 0.5 1 1.5 2 2.5 3 3.5 4
LVF %
Figure 5. Gain Required by Liquid Loading Level at Bacton Mist Flow Tests.
Clearly what was shown by the gain curves was that the signal strength could be used as a
measure of attenuation through the droplets, and from there possibly an indicator of what liquid volume
fraction (LVF) was present in the meter. In addition to gain, other parameters were shown to have
similar response to the presence of droplets in the gas. One of these parameters, the sample standard
deviation of the delta-T, or transit time difference, is shown in Figure 6. Note the strong similarity to the
attenuation indicator of the previous figure. With no liquid droplets present, this timing “jitter” should
range from about 50 to 400 nanoseconds depending on flow velocity, whereas it becomes almost 2000
nanoseconds in the presence of liquid.
Thus the potential was demonstrated to design a meter which could measure the local conditions
in those places in the pipe interrogated by the ultrasound. In the case shown, the effect of gravity on the
liquid could clearly be observed.
5
Bacton Data - Std DLTT against Liquid Fraction
2000
1800
1600
1400
1200
Chd A
Chd B
1000 Chd C
Chd D
800
600
400
200
0
0 0.5 1 1.5 2 2.5 3 3.5 4
LVF %
Figure 6. Timing “Jitter” on Travel Time Difference at Bacton Mist Flow Tests.
6
- Accuracy. The accuracy of the meters, either in parallel or in series, was acceptable.
- Diagnostics. This was an early test of some common ultrasonic meter diagnostics, such
as the use of speed of sound (SOS) and chord profile measurements. Results of these tests
were discussed.
- Liquid Detection. Before there was significant experience with the meters, there were
indicators suggesting the presence of liquids. From the outset failures on the bottom
chords were common, and the standard deviation of the travel time – “jitter” as observed
in the Bacton tests – was much greater than one would expect in dry gas.
After three years of using the meters, Stobie (1998) had much more to say about the experience:
- Pipework. The meters were installed at a low point in the process piping on the platform,
which tended to make the area a trap for liquids. Coupled with the conviction that the
hydrotesting of the system had left residual water, low chord failures were a fact of daily
life for the meters.
- Hydrates. Since the meter skid was short, flow conditioners were installed upstream in an
attempt to insure accuracy. Instead they served as a pressure drop which encouraged the
formation of hydrates, plugging holes in the conditioner and damaging some transducers.
Operating the skid in the range of hydrate formation meant that heat treating and/or
insulation had to be retrofitted.
- Meter Recovery. Time and again the meter demonstrated the ability to recover from an
upset, whether due to liquids or to transducer failures.
- Noise. One of the early instances of an installed meter being brought to its knees by the
noise created by a pressure control valve downstream (<20D) of the Dawn meter system.
- Transducers. The transducers used on the two meters probably accounted for at least half
of the anguish experienced in maintaining the system. An early goal of the UltraFlow JIP
had been to develop a transducer which was small in diameter, highly resistant to effects
of the liquids encountered, able to survive and operate over a wide range of temperatures
and pressures, and more. In retrospect the author believes the goals were too ambitious,
especially reducing the transmitted acoustic power by a factor of four to achieve the
smaller diameter. The transducers which were eventually installed on the Dawn meters
were a variant of the standard Daniel transducer used in its standard meter. Once they
were installed, the meter’s internal diagnostics measured how well the improved
transducers had solved the problem. Chord Failures went from 100 occurrences per day
to fewer than 10, a normally expected value.
Remember to think wet gas at the design stage and avoid expensive corrective action
in the field.
If you have to go wet gas metering......read the Standards......but do not follow them
slavishly. Use your brain...its as good as anyone else’s.... especially those of the guys
who write the Standards!!
7
The bold print is due to Stobie.
After the impressive results which were obtained during the first UltraFlow JIP, there was strong
interest among the participants – as well as some who hadn’t participated but were aware of the results –
in carrying forward with the development of wet gas metering technology. In particular they wanted to
determine if a meter could be developed based on the Bacton results which would estimate both gas and
liquid flow rates. Though it took three years from the end of the first UltraFlow Project to begin, Project
UltraFlow II – also called the Bi-Phase Ultrasonic JIP - began its work in 1997 with the seven members
BP, British Gas (Advantica), Phillips, Conoco, Statoil, TotalFinaElf, and Daniel.
The first task undertaken by the Project was the duplication of the Bacton results. Though this
seemed like it should be a straightforward task, it was anything but. In the three intervening years the
Shell Bacton wet gas flow facility, which had actually been an arm of the Shell Amsterdam laboratory,
had been closed and the personnel retired or moved. In casting about for other possibilities, the BG
facility at Low Thornley seemed the most attractive, however the flow rates were so low that only a
special 4-inch, 2-path meter could be utilized there. Still looking for confirmation in a facility which
used fluids which were the same (or behaved the same) as those used at Bacton, a new wet gas facility at
CEESI in Nunn, Colorado was used. While some of the data sets acquired there were useful, the stability
8
of the high flows required was not adequate to gather a totally meaningful data set. Finally, there was a
return to NEL to try to acquire the kinds of measurements needed in mist flow to confirm the Bacton
phenomena.
Unfortunately, the totality of these data sets never completely confirmed what had been observed
at Bacton, nor allowed a physical model to be developed which could be described as a “meter”. This
was not the greatest problem faced by the project, however. More important was the fact that no solution
had been found for measuring the flow rates if the regime in the pipe were stratified flow rather than
mist flow. There were sufficient examples experienced at every wet gas flow facility where stratified
flow (or its cousin annular flow) was dominant at some set of conditions. So even if the promising
Bacton data could be modeled such that mist flow could be measured properly, one would only have a
complete solution to the wet gas problem if it could deal with the case where the liquids dropped out of
suspension in the gas and traveled as a separate “stream”.
(a) (b)
Figure 8. Schematic of Daniel JuniorSonic Meter (a) in Normal and Rotated Positions, and (b)
with Reflection from Stratified Liquid at Bottom.
In the spring of 2000 Daniel suggested a solution to the JIP for metering the flow rates of gas and
liquid in stratified flow regimes. The technique, first proposed by Klaus Zanker, involved the use of a
standard Daniel JuniorSonic two-path meter. By rotating the meter 45 degrees, one of the ultrasonic
paths is horizontal while the other is vertical. A cross-sectional schematic of the meter before and after
rotation is shown in Figure 8(a). Each path includes a bounce of the transmitted signal off the opposite
wall, and then back to a transducer which receives the energy. In the case of the vertical path, the energy
traverses the path from the top of the meter to the bottom and back to the top. However, if the reflecting
surface from which the energy bounces is not the pipe wall, but the liquid which is traveling along the
bottom of the pipe, the meter will measure the energy traveling along a shorter path, namely that shown
in Figure 8(b). Since the path is shorter by the average height of the liquid h (commonly known as the
liquid holdup), the travel time of signals in both up- and downstream directions will be shorter, which
will appear to the meter as if it is traveling through a higher-velocity medium, and has a corresponding
effect on the measured speed of sound. Through geometrical analysis Zanker showed that by measuring
the difference in speed of sound between the horizontal and vertical paths of the meter one could
estimate the liquid holdup in the pipe, and thereby the percent error of the gas reading due to blockage
caused by presence of the liquid. By simply correcting the reading of the horizontal chord path with this
information, a somewhat crude estimate of the true gas flow rate could be made.
9
Figure 9. Error in Gas Flow Rate After Correction for Liquid Holdup.
Measurements were carried out at NEL to test the method’s effectiveness. The results are shown
in Figures 9 and 10. The test shown was run at a pressure of 50 bar, with the reference gas flow rates
ranging from 2 to 15 meters per second, and LVF as high as 5%. Using a simple slip model which
required continuity of shear stress at the liquid-gas interface, Zanker estimated the liquid velocity, liquid
flow rate, and liquid volume fraction, as well as the true gas flow rate. The errors in his gas flow rate
estimate are shown in Figure 9, and range from –1% to +3% of reading. Figure 10 is his plot of Liquid
Volume Fraction, where his errors are within a +/- 1% (absolute) band. It should be noted that these
results were obtained without making obvious improvements to the method , such as changing profile
assumptions in the gas and liquid and in some way accounting for the curvature of the stratified surface.
10
Figure 10. Liquid Volume Fraction (LVF) and LVF Error.
Another benefit of this technique is the fact that the signal on the vertical chord path will provide
a means to detect the presence of liquid along the bottom of the meter as well as to measure its height. If
there is no liquid stream there, the return signals should be clean and easily detectable. However if the
signal that is received has been reflected from the moving surface of the liquid, the signal detection will
be highly erratic, and show a discrepancy in the measured SOS as measured in the stratified flow testing
performed at NEL.
Although no commercial meter has been introduced to take advantage of these results, it can only
be a matter of time before one appears.
The results shown here and more can be found in Reference 4.
12
Figure 11. Jade Ultrasonic Meter Section, Plus Bishop Auckland Pipework Adaptor.
Figure 12. Testing of Jade Ultrasonic Meter Run at Bishop Auckland Flow Facility.
13
The essential results of the flow tests are shown in Figures 13 and 14. The meter proved itself to
be a stable and consistent metering platform, demonstrating a high degree of repeatability.
In Figure 13, the offset indicated by Grimley (1997) was clearly present. The meter appeared to
over-read by approximately 2.5%. Following the usual correction to account for the centreline velocity
(which is applicable if the flow profile is fully developed), the meter under-read by 2.67 to 3.16%.That
the flow was disturbed and asymmetric was also observed from the A and B chord velocities which were
considerably different and variable. However the readings were relatively stable, and in many (non-
fiscal) measurements this might be considered adequate.
3
2
1
0
-1
-2
-3
-4
0 5 10 15 20 25
Nominal Flow velocity (m/sec)
As one can observe in Figure 14, following the installation of the Flow Conditioner, it was
apparent that the flow profile had become less disturbed and that the asymmetry had diminished. This
confirmed what had been demonstrated by Reader-Harris (1995) in the NEL Headers Program. While
the A and B chord velocities continued to show some offset, it was considerably reduced. With respect
to meter error, i.e., how well the meter agreed with the reference meters, the base reading now showed
an over-measurement of approximately 5.5%. However after once again taking the flow profile
correction into account, the error was almost negligible – better than the users had any right to expect.
14
With Flow Conditioner
6
5
4
3
2
1
0
-1
0 2000 4000 6000 8000 10000
Flow rate ACMH
Base Error Error after Re Correction
Figure 14. Jade Ultrasonic Meter Run Performance with Spearman Flow Conditioner Installed.
The conclusion which therefore follows is that by (1) being diligent in calibrating the entire
pipework surrounding the meter and by (2) conditioning the flow as well as possible using the NEL
(Spearman) Flow Conditioner, Stobie (2001) was able to achieve fiscal performance from a meter which
is not generally regarded as a fiscal meter. All this was accomplished in a meter run less than half as
long as others might recommend, which included a reducer, a double-S bend, and a closely coupled flow
conditioner, all within the 8D upstream of the meter.
Conclusions
From what has been shown here, the following general conclusions can be drawn.
Ultrasonic technology can measure wet gas flow, perhaps better than any other method. We
have shown how analyzing the ultrasonic signals which propagate along various chord paths through the
meter can yield estimates of the key measures of wet gas flow. This is true in dispersed (mist) flow as
well as in stratified or annular flow. Furthermore, with proper chord placement one can detect which of
these flow regimes is present, and even detect the transition between the two. The meter passes slugs of
liquid, and is thus self-clearing of liquids. Except when using flow conditioners, formation of hydrates
around the meters is unlikely.
15
Lab testing is no substitute for extended Real-World experience. The long-duration Dawn tests
carried out by Phillips has demonstrated problems which were never observed in flow laboratory testing,
such as
- Ultrasonic noise generated by a pressure control valve.
- Hydrate formation on flow conditioners when flow rates were kicked up.
- Liquid collection in the meter, which was at a low point in the pipework.
- Increased chord failures over time traceable to deteriorating transducers.
With sufficient effort, extremely compact meter runs are possible which appear to be operable
to fiscal standards. Both limited space and poor surrounding pipework can be overcome. Dawn was
an early indication that through clever piping one could use ultrasonic meters to measure large gas
flows in very little space. Jade then showed that this was possible to an even greater degree, and that
“non-fiscal” meters (e.g., JuniorSonic meter) could perform to fiscal standards, even when placed in the
midst of non-ideal pipework. The key was to use a performance flow conditioner, e.g. Spearman or
Zanker plate, and to flow calibrate the complete meter pipework.
Ultrasonic measurement should be the front-running technology in upstream measurement of
natural gas – however it is not clear that any manufacturers are actively pursuing it. We hope what
has been shown will help to change this situation.
References
1. Wilson, M., “The Ultraflow Wet Gas Development Project”, Seminar on the Measurement of Wet
Gas, National Engineering Laboratory, East Kilbride, Scotland, October 1996.
2. Stobie, G., and K. Zanker, “Ultrasonic Meter: In-Situ Skid Mounted Flow Testing”, North Sea Flow
Measurement Workshop, Lillehammer, Norway, October, 1995.
3. Stobie, G., “Ultrasonic Wet Gas Measurement – Dawn Gas Metering: A ‘Real World’ System”, North
Sea Flow Measurement Workshop, Gleneagles, Scotland, October, 1998.
4. Zanker,K., “An Ultrasonic Meter for Stratified Wet Gas Service”, North Sea Flow Measurement
Workshop, Kristiansand, Norway, October 2001.
5. Grimley, T.A., “Performance Testing of Ultrasonic Flow Meters”, North Sea Flow Measurement
workshop, Kristiansand, Norway, October 1997.
6. Reader-Harris, M., Sattary, J., and Woodhead, E., “The Use of Flow Conditioners to Improve Flow
Measurement Accuracy Downstream of Headers”, 3rd International Flow Symposium, San Antonio,
Texas,1995.
7. Stobie, G., Zanker, K., Brown, C., and W. Letton, “Flow Testing an Ultrasonic Meter Outside its
Performance Envelope - Another Real World System”, North Sea Flow Measurement Workshop,
Kristiansand, Norway, October, 2001.
16