NSFMW 1992 Technical Papers
NSFMW 1992 Technical Papers
NSFMW 1992 Technical Papers
Paper
Paper Title Author Details Company Page
No
Opening Address: Flow
Measurement The Last 10 H Danielsen Statoil 3
Years
The Orifice Plate Discharge M J Reader-Harris,
1.1 Coefficient Equation Further J A Sattary and NEL 11
Work E Spearman
Orifice Metering Research A J E Gallagher Shell
1.2 38
Users Perspective R E Beaty Amoco
W T Lake Amoco
1.3 Optimal Flow Conditioner 70
J Reid NEL
Multiphase Flowmeter B C Millington and
2.1 NEL 114
Measurement Uncertainties T S Whitaker
A Multi-Capacitor Multiphase D Brown, J J den Boer and
2.2 Shell Research 127
Flowmeter for Slugging Flow G Washington
A Flow-Regime Independent E Dykesteen and
2.3 CMI 139
Multiphase Flowrate Meter Midttveit
Framo Multiphase Flowmeter A B Olsen and
2.4 Framo Engineering AS 155
Prototype Test B-H Torkildsen
Ultrasonic Flowmeter in a Gas
2.5 R M Watt NEL 187
Metering Station
D Thomassen and Institutt for Energiteknikk
Flow Conditions in a Gas
3.1 M Langsholt 189
Metering Station
R Sakariassen Statoil
CFD Analysis of Fluid Property
3.2 Effects in Coriolis Mass R M Watt NEL 203
Flowmeters
1992
Paper
Paper Title Author Details Company Page
No
Results of Investigations
Comparing Some of the
4.1 Recommendations Given for J F Cabrol and A Erdal Statoil K-Lab 205
Turbine Meters by ISO-9951 and
AGA-7
The AGA Transmission
J Stuart Pacific Gas and Electric
Measurement Committee and
J Savidge Gas Research Institute
4.2 the Revision of AGA Report No 8 225
S Beyerlain and University of Idaho
Compressibility Factors of
E Lemmon
Natural Gas
Density Metering Installation
4.3 J Gray Peak Measurement 242
Methods
A Lygre CMI
A New Multi-Phase Ultrasonic
5.1 R Sakariassen Statoil 283
Flowmeter for Gas
D Aldal Fluenta
Experience with Comparative
Testing and Calibration of
5.2 L Mandrup-Jesnen Force Institute 305
Coriolis and Turbine Meter
Offshore and in the Laboratory
Compact Large Bore Direct A J Mathews and
5.3 Schlumberger Industries 319
Mass Flowmeters C J Ayling
Pipe Elbow Effects on V-Cone S A Ifft and
5.4 Ketema Inc 335
Flowmeter E D Mikkelsen
Closing Address: Flow
Measurement The Next Ten J E Gallagher Shell 351
Years
I
Ie
I
I
I FI1JN MEASUREMENl' - THE IAST TEN YEARS
II
by
I
t' H B Danielsen
Statoil
I
,
I
I
Ie
.' OORTH SEA FI1JN MEASUREMENl' IDRKSHOP
26-29 october 1992
I
I
I
I NEL, East Kilbride, Glasgow
It
I
I,
Ie FLOW MEASUREMENTTHE LAST 10 YEARS
I H B Danielsen
I Summary
I' A review of developments within North Sea flow metering is given.Most of the issues
reviewed here are those which were brought up at the first Workshop ten years ago. In
This first Workshop took place in Stavanger in 1983 where the idea of a yearly metering
seminar ,alternating between Norway and UK,had been conceived at a more local seminar
the year before.
Let's have a look at the main issues of this program and use them as a basis for reviewing
what has happened from then and up till to-day.
,I Dr. Spencer
'e Among the names on the program you will probably notice Dr. Spencer.
No review of flow measurement for the last ten years can be said to be complete without
I commenting on the significant role played by this man.
'I' Anybody working with flowrneasurement in the 1980s would inevitably become aware of
Dr. Spencer: As an author of papers,as chairman of seminars and conferences,always
having a prominent position in whatever he took part.
I Also, he was one of the key persons in the process of establishing the North Sea flow
, Measurement Workshop
Dr. Spencer was pensioned from NEL some time late in the 1980s and then seemed to
drop out of sight for most of us.From what I understand,however,he has kept himself busy
by working for UN and for the EEC.
Hopefully,he is still in activity with the same energy and enthusiasm as he displayed at the
I, past Workshops.
It
I 1
:.
ISO 5167
eI
The first lectures.given by Dr. Spencer, was about the development of ISO flow I
'I,\
measurement standards.measurement uncertainty and comparisons between ISO 5167 and
AGA Report No.3.
ISO 5167 was relatively "new" at that point of time.It had been issued in 1980 to replace
ISO R-541 and had some significant improvements.such as the Stolz equation for
calculation of flow coefficients.ISO 5167 had become the standard to which all new
orifice metering stations for North Sea flow measurement now were designed .In the years
to come it was "retro-fitted" on those metering systems which had beeen originally
designed to ISO R-541.
't
In Americahowever.there seemed to be less enthusiasm for ISO 5167 and they still
seemed to put their trust in AGA Report No.3 with the "old" flow coefficient.formulas. I
In addition.a lot of the measurement engineers with a practical job were aware that ISO
5167 was not perfect. Its scope was so that it was not the complete document they would
have liked to have available to use as a basis for for design and operation of industrial
.",
orifice metering systems.
ISO was aware of this and had already.as early as 1977 initiated work with a separate
document,"Code of practice for ISO 5167".
I
Also ,right after the issue of ISO 5167 in 1980 a working group in ISO had already started
its work to revise it. I
On this background,a lot of metering engineers probably would have predicted the
following development within the next ten years from 1983 onwards: I
-The flowcoefficients of ISO and AGA would not be different.
,I
-ISO 5167 will be supplemented by a "Code of Practice Document" to cover the more
practical sides of orifice metering. .
To-day,however,there is still some way to go.The COP has not been issued and there is
J
still a conspicuous difference between the 1992 revision of ISO 5167 using the Stolz-
formula and the AGA 3 now using the new Reader-Harris-Gallagher formula.
I.
t
Metering Regulations
Metering regulations was the second subject at that first measurement workshop.The
I
lecture was given by Mr. 0glrend of NPD.
At the time of the workshop a draft of the NPD fiscal metering regulations had been
completed and had been sent out for comments to the industry.The regulations were put
into force in 1984.
I
II
2
I
I
Ie Although everybody were not always happy with everything that these regulations
.J implied,I think that they had a positive effect in terms of setting a standard to be met by
the industry .In addition they implied that NPD had to give their specific approval of a
metering system when all documentation had been provided and all the required tests had
As the years went on the industry got used to live with the reguiations.If the oil
I companies should initiate a revision,a relaxed regulation for metering of smaller streams
with less fiscal importance would probably have been on the top of their wishing list.
,tt The revised NPD fiscal metering regulations were put into force last year.
About the same time , work on other metering regulations started .As a new tax on gas
I, burnt as fuel or flare on offshore platforms on the Norwegian continenetal shelf was
imposed in 1989,regulations for measurement of this gas had to be made.The draft of
these regulations has now been completed and it is expected that they will be put in force
in a few months from the time of this workshop.
I
I Gas Turbine Meters
Two lectures ,titled " Gas Metering at high Reynolds Number" and "New Standard for Gas
I Turbine Meters" respectively,were given by Ioe Bonner.
..
'
Both of these lectures focused on gas turbine meters.
Although the first lecture also deals with orifice meters,one of the issues of the lecture is
a discussion about plotting the calibration curves of turbine meters as error against
I Reynoldsnumber instead of flowrate.
'I The second lecture was about an ISO standard for gas turbine meters .At that time, a draft
was being made by Working Group 15 of ISO TC 30, 90% of the draft was completed
and only the difficult last 10% remained to be made.Although not stated directlyJ think
I
,
the audience got a feeling that Mr. Bonner was very worried about the progress in the last
phase of drafting this standard .
. Sometime between now and then things must have stopped up as I can find no standard
for gas turbine meter in the 1992 ISO catalogue.
f, Liquid Densities
I
I
A lecture with the title "Calculation of Liquid Densities and their Mixtures" was given by
et
Risdon W. Hankinson from Phillips Petroleum Company's headquarters in
Bartlesville.Oklahoma, I
Because there seemed to be an increasing number of fiscal metering systems for
unstabilised oil,LPG and LNG in the North Sea area at that point of time.there was a
great interest in methods to calculate densities and volume correction factors for other
t
Iiquids than stabilised crude oils.
\1
Hankinson's lecture was partly about the the then new temperature correction factors of
the ASTMJIP/API Petroleum Measurement Tables.
These tables had been issued in 1980,replacing the very old table 6 of API 2450 ..
;1
This was a very big step forward as the old tables had been issued in 1940 based on a
limited number of crude oils produced before 1930.
I
After the workshop the Petroleum Measurement tables were further extended in 1984
and 1986 by new sections for compressibility factors.
The part of Hankinson's lecture that attracted most interest.however.was the part about
--,
the COSTALD compressed liquid density correlation. At that stage.in 1983,the Costald
correlation was being used in fiscal measurement,but not by many and only for "offline"
calculations.
At seminars later during the 80s,Phillips personell promoted the use of COSTALD and
outlined their plans to improve the correlation to be valid for temperatures near to the
critical temperatur and for higher pressures.
Up till to-day, COSTALD has gained increased acceptance and is being used as an
"online" density calculation method in the flowcomputers in several North Sea metering
systems.
The paper is missing in my folder but I think the main topic of the lecture was about
what was called at that time the " GR! method" for compressibility calculation.
Two years later,in 1985,this method was introduced as a standard,in AGA report no. 8.1t is
now the most commonly used calculation method for gas compressibility in the North Sea
gas metering.
I
I
On Line Gas Densitymeters
I densitometers.
The second lecture ,by Paul Wilcox ,was about practical experience with gas
"
densitymeters.
'I'. The lectures and the comments from the audience highlighted two problems:
I The first of these problems became a hot subject at one of the later workshops.There was
a lot of different opinions on which calibration and correction methods would give a
I compact provers.
Bill Pursley gave the first lecture.Although this lecture was about conventional provers,its
I, fmal remarks was about compact provers probably taking over a lot of the role of the
conventional prover on the future offshore platforms.
I Terry Noble gave a lecture on what was later named the Brooks Compact Prover.
This was followed up by Peter Jelffs who gave a lecture on compact provers in general
I and a short description of what MBR saw as desirable design features for this kind of
prover.
I To my knowledge there was only one compact prover in use in the North Sea activity in
1983.Phillips Petroleum Norway/Basic Resource Services were using their "Ballistic Flow
Prover" as a transportable calibration unit for provers in the Ekofisk Area.This had been
'I very sucessful and the prover is still in use for the same purpose at Ekofisk.
rt' As time went by,most of the companies offering prover calibration services in the North
Sea started to operate one or more transportable compact prover. But up till now the
compact provers have not been used on a large scale as permanently installed units.
II To my knowledge only 3 units have been installed for this purpose:One in each of the
Danish ,Dutch and Norwegian Sector.Reportedly,all these units operate satifactorily.
I 5
I
et
However, most people present at the first Workshop would probably have guessed a
higher number of compact provers in 1992. I
Significant Developments 1~S3-1992 I
Although not brought lip at the lust Flow Measurement Workshop.I cannot finish this
review of flow measurement for the last ten years without saying a few words about
I'
Coriolis mass meters.ultrasonic transit time gas flowmeters and multiphase metering.
"1
Coriolis Mass Meters
The existence of the Coriolis mass meter was probably known to most of us in 1983 but
I
not much attention was paid to it.One of the reasons for this may have been that .at that
stage, flowrates were big and mass meters were small.
A1; time went on, the streams to be metered have tended to get smaller while the
et,
massmeters have grown bigger. Also, an extensive evaluation activity were started to
evaluate the massmeters' operational characteristics.
To-day, a number of manufacturers offer Coriolis massmeters,the largest have a capacity
I~
in excess of a 6-inch liquid turbinemeter .Standards are being drafted by ISO and other
bodies and we have even got a Coriolis-based fiscal metering system for condensate at
Total's shore terminal in St. Fergus.
I
Being aware of that there is a lecture tiltled "The next 10 Years" at the end of this
Workshop, ] should not try to make any predictions about the future of the massmeters in
the North Sea.But I think that these meters will have a great potential for crude oil
I
measurement if it turns out that their calibration is so stable that they may be installed
without a permanent prover. I
Ultrasonic Gas Meters
At the time of the first Workshop, a development to use ultrason.ic transit time meters for
J
gas measurement had already started. Among the challenges they had at that time was to
make the ultrasonic transducers powerful enough to send their signal through gas and to I
develop methods to calculate flowrate from a number velocity readings along chords of a
cross section of flow.
To-day,ten years later, the manufacturers offer both multipath ultrasonic gas meters for
I
fiscal applications and flare gas meters with less accuracy but very wide flowranges.A
number of ultrason.ic flare meters have been installed in the North Sea .No fiscal meter is
in operation yet but we will see them in the very near future,
Multiphase Flowmeters
Even before the 1983 Workshop the North Sea oil companies were aware that some of the
fields found were too small to be developed with dedicated platforms,separation
equipment and conventional fiscal metering.
6
I
te New technology in terms of subsea wells,multiphase pumps and multiphase metering was
I needed .As a result of this a lot of the oil companies started or sponsored development
projects to provide the technology to measure muliphase flow.The objective was to get a
meter that could measure the flow of each of the components oil,water and gas .
'I Generally .not much of concrete information was given to outside world.At the
Offshore Northern Seas Conference in Stavanger in August 1992,however,a number of
I multiphase measurement devices were on display. Some impressions of state of the art
from the ONS were:
I Very large amounts of money had been spent during a number of years on the
development of each device.
I Although one got the impression that the meters were for sale,most of them had not yet
been out in the field.
t- The "readout" varied between the various meters, one could get one or more of actual
cubic meters ,standard cubic meters or mass of the full stream,each phase or each
component of the liquid phase.
I
The main impression were,however, that the oil companies involved now definitely had
I moved from the stage when they limited themselves to pay for development and tests in a
laboratory to a stage where they also would let the meters be installed in the field for
'I testing.
-,...
- '
I
'Ie
Ii
.'I
,I,
I,
It
I 7
I
Ie
I
I
I THE ORIFICE PlATE DISCHARGE COEFFICIENl' ~ION - FURl'HER ~RK
I by
I
II M J Reader-Harris, J A Sattary and E P Speannan
NEL
I
I
I Paper 1.1
..
I OORTH SEA F'I.C1fl MEASUREMENTw)RKSHOP
26-29 October 1992
I
I
I
NEL, Fast Kilbride, Glasgow
I
It
I
I
THE ORIFICE PLATE DISCHARGE COEFFICIENT EQUATION - FURTHER YORK
~ H J Reader-Harris, J A Sattary and E P Spearman
SUKKARY
I This paper describes the york undertaken to derive the revised orifice
plate discharge coefficient equation based on the final EEC/API database
including the data collected in 50 mm and 600 mm pipes. It consists of
several terms, each based on an understanding of the physics. An earlier
I version of this equation, based on a smaller database, vas accepted at a
meeting of EEC and API flov measurement experts in New Orleans in 1988, and
emphasis is placed on the two principal changes to the equation: improved
tapping terms for low Reynolds number have been calculated; and an
I additional term for small orifice diameter has been obtained, and its
physical basis in orifice edge roundness given.
NOTATION
~
A Function of orifice Reynolds number (see equations (6) and (7))
I d Orifice diameter
Quotient of the distance of the upstream tapping from the
I upstream face of the plate and the pipe diameter
Quotient of the distance of the downstream tapping from the
I L2 '
upstream face of the plate and the pipe diameter
Quotient of the distance of the downstream tapping from the
downstream face of the plate and the pipe diameter
I Quotient of the distance of the downstream tapping from the
upstream face of the plate and the dam height (equation (2))
I
-I 1
I
M2 ' Quotient of the distance of the downstream tapping from the
dovnstream face of the plate and the dam height (as in equation
(2 ) )
Diameter ratio
Friction factor
Shift in friction factor due to pipe roughness
1 IIfIRODUC'IION
Although the orifice plate is the recognized flowmeter for the measurement
of natural gas and light hydrocarbon liquids, the orifice discharge
coefficient equations in current use are based on data collected more than
50 years ago. Moreover, for 20 years the United States and Europe have
used different equations, a discrepancy with serious consequences for the
oil and gas industry since many companies are multinational. For more than
ten years data on orifice plate discharge coefficients have been collected
in Europe and the United States in order to provide a new database from
vhich an improved discharge coefficient equation could be obtained which
vould receive international acceptance.
In November 19BB a joint meeting of API (American Petroleum Institute) and
EEe flow measurement experts in New Orleans accepted the equation derived
by NEL. At that time the database contained 11 346 points, collected in
pipes whose diameters ranged from 50 to 250 mm (2 to 10 inch); 600 mm (24
inch) data were being collected but had not yet been included in the
database. 600 mm data have now been collected in gas and in water and
extend the database both in pipe diameter and in Reynolds number. The data
which were least well fitted by the equation presented at New Orleans were
the 50 rnrndata; so additional 50 mm data have been collected in water and
oil which provide additional information about discharge coefficients both
for small orifice diameters and for low Reynolds numbers. All the
additional data have nov been included in the database and the equation
refitted. This paper gives that revised equation and its derivation.
I
2 THE DATABASE I
The final database consists of 16376 points: the diameTer ratios range from
0.1-0.75, orifice Reynolds numbers from 1700 to 7 x 10 , and pipe diameters
I
2
til
I
I
from 50-600 mm. The data were collected in nine laboratories in four
~ fluids: water, air, natural gas and oil. The data points for which the
orifice diameter was less than 12.5mm (~ inch) and those for which the
differential pressure across the orifice plate is less than 600 Pa are very
I scattered and were excluded. The American data remain as in References 1
to 3; no additional American data have been collected. The complete EEC
data are tabulated in References 4 to 10; the data sets which have been
I accepted for analysis are indicated in Reference 11. A very small number
of points (0.5 per-cent of all the EEC points) was removed from the EEC
data as outliers; each of those removed was identified as an outlier within
I its own set of data by the Grubbs' extreme deviation outlier test; details
will be found in Reference 12.
I 3 TAPPING TERMS
The tapping terms are equal to the difference between the discharge
I coefficient using flange or 0 and 0/2 tappings and those using corner
tappings. They are expressed as the sum of an upstr~am and a downstream
tapping term. The upstream term is equal to the change in discharge
coefficient when the downstream tapping is fixed in the downstream corner
and the upstream tapping is moved from the upstream corner to another
~
position. The downstream term is equal to the change in discharge
coefficient when the upstream tapping is fixed in the upstream corner and
I In the database only the total tapping term, the sum of the upstream and
the downstream terms, is available. To divide the tapping term into two
parts so that each can be accurately fitted, measurements of the individual
I tapping terms collected outside the EEC and API projects were used to
indicate the form and approximate value of the upstream and downstream
terms; however the constants in the formulae were obtained to fit the
EEC/API database. The EEC collected data with several tapping systems; so
the total tapping term could be simply obtained (References 13 and 14).
~ Although the American data were collected with flange tappings alone small
t adjustments were made to the final tapping terms in order to obtain the
optimum fit to the database as a whole.
On examining the measured tapping terms it has been shown (References 13
I and 14) that for high5Reynolds number (orifice Reynolds number, Red'
greater than about 10 ) the tapping terms may be considered not to vary
with Re , but that for low Reynolds number the terms depend on Re. An
importa~t part of the work undertaken since the meeting in New Or~eans has
I been to provide more accurate low Reynolds number tapping terms. Since the
high Reynolds number tapping terms need to be determined first they are
described here first.
-I 3
I
Reynolds number data vere eKcluded. Details of the method by vhich the
values of the tDtal tapping terms vere calculated are given in References
13 and 14. eI
3.1.1 Upstream tera
To determine the correct form of the tapping terms, the upstream term,
I
h~ ,was determined first. The dependence of the upstream term on
~ ~~1-~ ) is veIL established (References 13 - 16); so here it is only
necessary to consider the dependence on L" the quotient of the distance of
I
the upstream tapping from the upstream face of the plate and the pipe
diameter. Several forms of equation vere tried, and their exponents and
constants determined, and the optimum one found to be the folloving: I
(1)
I
I
Tais eq~ation is physically realistic: it has the required dependence on
~ l(l-~), is equal to 0 for L,=O, does not become negative, tends rapidly
to a constant once Ll eKceeds 0.5, and has a continuous derivative.
Together with the downstream term it gives a very good fit to the total
tapping term data. It is plotted in Fig. 1 against many sets of
eKperimental neasurements of the upstream tapping term and the quality of
ea
the fit is good. References to the work of the nany eKperimenters who
collected the data in Fig. 1 are given in References 13 and 14.
3.L.2 Dovnstream term ,I
Many experinenters have measured the pressure profile downstream of the
orifice plate, and, although the data are more scattered than those
upstream of the plate, the pattern is clear: the pressure decreases
downstream of the plate till it reaches a minimum and then quite a short
I
distance dovnstream of the ninimum it begins to increase rapidly. The
orifice plate should not be used vith the downstream tapping in the region
of rapid pressure recovery. I
J
An important step in the determination of the dovnstream formula was the
vork of Teyssandier and Husain (Reference 17) who non-dimensionalised
downstream distances with the dam height rather than the pipe diameter.
Instead of vorking in terms of L2, the quotient of the distance of the
downstream tapping from the upstream face of the plate and the pipe
diameter, it is better to use H2, the quotient of the distance of the
downstream tapping from the upstream face of the plate and the dam height,
I
vhich is given by
I
(2)
1 - ~
L,' and H~' are defined identically except that in each case the distance
I
fron the downstream face of the orifice plate is used.
From consideration of data from many experimenters References 13 and 14 I
confirmed the advantage of non-dimensional ising with dam height by shoving
that the pressure min inurnoccurs for
I
4
II
I
I
-I H2 = 3.3.
I 6Cdown,min = -0.0101~
1.3
(4)
Several forms of equation for the complete downstream tapping term were
~ tried, and their exponents and constants determined, and the optimum one
found to be the following:
,
~
upstream corner from tappings at distances 0, 0/2, D/4 and D/8 upstream
was measured, where 0 is the pipe diameter, as well as the pressure drop
from the downstream corner to tappings at distances D/8 and D/4 downstream
of the downstream face of the orifice plate and to the downstream D/2
tapping. Whereas for high Red the tapping terms do not depend on Rerl, the
tapping terms at the Reynolds numbers obtained in oil are significantly
I different. Previous work by Johansen (Reference 18) had shown that the
upstream tapping term at the upstream pressure minimum decreases as Red
decreases. Similarly the downstream tapping term at the downstream
pressure minimum decreases in magnitude as Red decreases. Data from Witte
I and Schroder (quoted in References 19 and 20), who only measured the
upstream tapping term, agree with Johansen. So the equation presented at
New Orleans reflected these data on the reasonable assumption that the
I tapping terms, though decreasing in size as Red decreases, retain the same
shape as a function of distance from the plate, since data were not
available except at the pressure minima.
-I 5
analysis, but also provide revised tapping terms which correspond much
-I
better to the tapping term collapse found in the database as a whole than
the New Orleans tapping term did. Fig. 4 shows all the data for ~ = 0.74:
it can be seen that all the data collapse on to one another as Red
decreases. However, the amount of data makes it difficult to see that the
flange data collected in 50 mrn, 100 mm and 150 mm pipes collapse on to one
another at a higher Reynolds number than that at which the corner tapping
data collapse on to the other data. The collapse of the flange tapping
data on to one another can be seen clearly in Fig. 5 which shows the US
data for ~ ~ 0.74: the approximately constant values of the tapping terms
for high Re can also be seen. Figures 4 and 5 confirm the need for
tapping tergs which are functions of Reynolds number, but also show that
I
the simple dependence on Red used in the New Orleans equation is
insufficient.
I
The main features of the tapping term data collected in 50 mm pipe (in Figs
83-100 of reference 21) are as follows: the upstream tapping term for D and
I
for D/2 tappings decreases with decreasing Re as expected from the work of
Johansen and of ~itte and Schroder, although ehe NEL data decrease slightly
more slowly with Re~; the upstream tapping term for D/4 tappings remains
approximately constant; the upstream tapping term for DIS tappings
increases with decreasing Ren. The dependence of the downstream tapping
terms on Re depends on ~: for ~ > 0.7 they decrease in magnitude with
,
decreasing fien; otherwise they are constant. At the bottom of the Red
range the uncertainties in the data become large, especially for the ~
upstream D/2 tapping data.
It is interesting that downstream of the orifice plate the data in
Reference 21 are apparently inconsistent from those of Johansen: one
possibility is that it is only near the pressure minimum that the magnitude
t
of the downstream tapping term decreases with decreasing Rerl, through the
pressure minimum becoming closer to the orifice. For, wher~ the values of
the downstream tapping term deduced from the data in Reference 21 were
I
taken at the pressure minimum, that is for ~ > 0.7, they decrease in
magnitude like those of Johansen; elsewhere the two sets of data are not
directly comparable: the data in Reference 21 were not taken at the
pressure minimum, whereas Johansen's were.
I
It is important to see the pattern in the tapping term data: to do this it
is necessary to do an analysis of the uncertainty of these data. It is
I
then possible to analyse all the upstream tapping term4data ~imultaneously,
and, in particular, to verify that the dependence on ~ /(l-~ ) which
characterizes the data for high Red continues to apply for low Ren. Fig. 6
shows the change in discharge coefficient due to moving the upstream
pressure tapping from the upstream corner for the 4 values of &.f~r which
J
measurements were made. Yhere the data are multiplied by (l-~ )/~ they
fal140n t04a single curve for each value of L,. Data are only plotted if
(1-~ )u/(~ ~p ) < 0.05, where u is the uncertainty of the pressure
I
measurement aE that point and ~p is the pressure differential across the
orifice plate using corner tappi&gs. I
Various possible forms of the upstream tapping term, ~C ,for low Red data
were tried, and the best one found to be the following:uP
I
I
6
I
-I
I
(0.043 + (0.090 - aA)e-10L, - (0.133 _ aA)e-7L,)
'Ie 6CUp
(34
I (1 - bA)--"""4 '
1 - (3
(6)
I vhere
I
A = [2100(3)n
ReD
and a, b, and n are to be determined.
This equation has the same behaviour for L, = 1 as the equation accepted in
Nev Orleans, and is equal to 0 for L, = 0, but is significantly different
for intermediate values of L,. Vith this form of equation the best fit to
I the upstream tapping term data included in Fig. 6 vas obtained. Since the
product term is not included in the final formula for the tapping terms,
the fitted upstream term is adjusted to make allovance for it: folloving
the argument in Reference 13, 6C (1 + 36Cd IC), vhere C is the
"
I
discharge coefficient using cornMP tappings,oYR
I and b = 1.307.
..
'I,
Hovever, these values vere adjusted to give a better fit to the database:
the best fit of the complete database gave a larger value of a than the fit
to the upstream tapping term data: a compromise value vas obtained as
follovs: from the Figures in Reference 21 it appears that the data for L, =
1, those for L, = 0.25 and those for L, = 0.125 cross at Re = 13 000.
Since in equation (6) the three curves representing the thr~e values of L,
do not intersect at a single point, a further simplification is to consider
the intersection of the curve for L, = 0.167 (corresponding to flange
I tappings in 6-inch pipe) vith the curve for L, = 1: this occurs for a =
1.03. This constant is then rounded to 1. Equation (6) vith a = 1, b =
1.307, and n = 0.9 is then plotted in Fig. 6 for comparison vith the data.
This equation describes a change in the pressure profile upstream of the
I orifice in vhich, as Re decreases, the upstream tapping term at D
decreases but the gradi~nt of the tapping term near the corner increases.
II 7
I
I
tapping term (incorporating a small downstream effect) is therefore
I
where c is a constant and T is the pipe Reynolds number at which transition
to fully turbulent flow occurs. T varies, as would be expected, from one
set of data to another, but a reasonable estimate of the range of values
encountered in t~e database is 3000 - 4500, and T = 3700 has been used for
both the tapping term and the slope term. Vith this value for T c is
determined by fitting the data in Reference 10: using the difference
between flange and corner tappings only, c = 8.20; using the difference
between D and D(2 and corner tappings only, c = 7.88; using all t~e data, c
= 8.04. The agreement between the values of c obtained using flange and D
and DJ2 tappings is very good, and the downstream tapping term used in the
final equation is as follows:
I
4
bCdown = -0.031(K2' - 0.8H2,1.1){1 + 8 max(loglO(3700/ReD),0.0)}~1.3
SftALL DRrFICg DIAMETER TERM
(9)
I
An additional term for small orifice diameters has been added to the
equation accepted at New Orleans as a result of collecting the NEL 50 rnrn
data which include measurements of edge sharpness. The problem here is
J
that it is extremely difficult to obtain a sufficiently sharp edge where
the orifice diameter is small: Fig. 7, which includes averages of measured
edge radii from the plates used in the EEC tests, in which D was in the
I
range 50-600 mm, shows that for orifice diameter, d, less than 50 mm the
plates rarely meet the requirements of ISO 5167-1 (Reference 22). It is
clear that for d ~ 25 rnrn there will be large shifts in C. When the edge
radii themselves are plotted as in Fig. 8, it appears that the edge radius,
I
r , (in mm) increases as d decreases from 50 rom, whereas to meet the
s~andard it needs to decrease fairly rapidly. I
The change in discharge coefficient due to edge roundness, ~c rl' has
been measured by Hobbs (Reference 23) as a function of changer~Hnedge
radius. ~re' and can be expressed approximately as
I
(10)
t
8
-I
I
-I It seems reasonable
mm is given by
to suppose that the mean value of r , rem' for d < 50
rem =
e
I which on substituting
6Cround = 3.33(rem/d
from equation
- 0.0002),
(11) becomes
(12)
The slope term consists of two terms, an orifice Reynolds number term and a
(17)
= b,(106/Re )n,
I C
s
=
d
b,(106~/ReD)n,. (18)
-I 9
I
also a velocity profile term which can be derived using the fact that, for
a fixed Beynolds number, as the pipe roughness changes t~e change in the
discharge coefficient is approximately proportional to ~ ~, where AA is
eI
the change in the pipe friction factor and 1 ~ 4 (Reference 24). A simple
integral of this expression together with the orifice Reynolds number term
gives I
This term is adequate for high Reynolds number but for practical use it
(19)
I
requires three further enhancements. There is no data on the effect of
rough pipework on the discharge coefficient for lov Reynolds number, and a
better fit to the database is obtained by including an additional term I
proportional to A, on the basis that, as the tapping terms begin to change,
so may the dependence on friction factor:
Cs =
6
b, (10 ~/BeD)
n
1 +
1
(b, + b3A)~ A. (20)
I
A is an inconvenient variable with which to work, but for the pipes used in
I
--.
collecting the data in the EEC/API database a typical pipe roughness as a
function of Reo can be determined; so typical values of A as a function of
BeD can be calculated, and A can be approximated by a constant (which
leads to a term to be absorbed into Coo and a small term which is neglected)
together vith a reciprocal pover of ReO:
C
s
= b ,(106fl/ReD)nl + (b2 + b3A)~1(106/Reo)n,.
It is also necessary to make provision for transition from turbulent to
(21)
I
laminar flow since, except for very lov ~, the gradient of the discharge
coefficient as a function of a reciprocal power of ReD is very different
below a transition point in the range Re = 3000 - 4500 from that above it.
I
This change of gradient occurs because tRe velocity profile changes very
rapidly as the flow changes from turbulent to laminar, and vhen the
velocity profile term is extended so that it can be used below the fully
turbulent range the slope term becomes: .
I
C = bl(106~/ReD)nl + (b, + b3A)~1 maX[(106/Reo)n" c1 - C,(Reo/l06)}. (22)
s I
It remains to determine the constants and exponents in equations (17) and
(22). m, is equal to 8 in both the Stolz equation (Beference 22) and that
adopted in New Orleans (Reference 14) and using this high value gives a
good representation of the rapid decrease in C for high~. Previously ml
has been taken to be 2, but the optimum value ~f m1, in terms of the lovest
J
standard deviation of the data about the equation, lies betveen 1.2 and
1.3: the same value as the exponent of ~ in the downstream tapping term I
(equation (5 is used. nl = 0.7 and n, = 0.3 give the optimum fit to the
complete database. 1 is taken to be 3.5 rather than 4 because it gives a
better fit to the complete database: in terms of the dependence of the
effect of rough pipework on ~ an exponent of 3.5 is as acceptable as 4
I
(Reference 12). As stated in section 3.2 the mean ReD at vhich the flov
becomes fully turbulent vas taken to be 3700. This g1ves c1 in terms of
c,. c, is obtained by trying appropriate values in turn and obtaining the
best overall fit: c2 = 4800 gives an excellent overall fit. Given the
I
tapping terms in equations (7) and (9) and the small orifice diameter term
in equation (14) a least-squares fit of the complete database was
performed: on rounding the constants, the Coo and slope terms become
I
I
10
--I
I
-I C~ + Cs = 0.5934 + 0.0232~1.3 - 0.2010~8
+ 0.000515(106~/Reo)0.7
I H2 '
2L2'
1 - ~
I
A = (2100~)0.9
and
I ReO
-I 11
I
8515 data points for 0.19 < ~ < 0.67, Re > 30000 and d > 50 mm are
analysed the standard deviation of the p8ints about the equation is 0.208
per cent.
.-
1 CONCLUSIONS I
The orifice ~late discharge coefficient equation has been revised in the
light of the complete EEC/API database including the data collected in 50
mm and 600 rnrn pipes. There are tvo principal changes to the equation
accepted at New Orleans: using the additional data collected in 50 mm pipes
I
improved ta~~ing terms for low Reynolds number have been calculated; and an
additional term for small orifice diameter has been obtained, and its
physical basis in orifice edge roundness given. The revised orifice plate I
discharge coefficient equation is given as equation (24): the deviations of
the database from the equation have been tabulated and shown to be veIL
balanced as functions of ~, 0, ReD and pair of tappings used. I
A.CJafOVLEDGEIIENT
This paper is published by permission of the Chief Executive, National
Engineering Laboratory Executive Agency, and is Crovn copyright. The work
reported here was supported by DGXII of the Commission of the European
Communities and by the National Measurement System Policy Unit of the
Departnent of Trade and Industry.
REF EllEliCES I
1 ~ETSTONE, J. R., CLEVELAND, Y. G., BAUMGARTEN, G. P. and YOO, S.
Measurements of coefficients of discharge for concentric, flange-
tapped, square-edged orifice meters in water over a Reynolds number
I
range of 1000 - 2 700 000. NIST Technical Note TN-1264. American
2
Petroleum Institute, Yashington DC, 1988.
ERITTON, C. L., CALOYELL, S., and SEIDL, Y. Heasurements of
I
coefficients of discharge for concentric, flange-tapped, square-
edged orifice meters in vhite mineral oil over a low Reynolds number
range. American Petroleum Institute, Vashington DC, 1988. I
3 Coefficients of discharge for concentric, square-edged, flange
-tapped orifice meters: equation data set - supporting documentation
for floppy diskettes. American Petroleum Institute, Yashington DC,
1988.
J
HOBBS, J. H. Ex~erimental data for the determination of basic 100
I
mm orifice meter discharge coefficients (European programme).
Report EllR 10027, Commission of the European Communities, Brussels,
Belgium, 1985. I
5 HOBBS, J. H. and SATTARY, J. A. Experimental data for the
determination of 100 mm orifice meter discharge coefficients under
different installation conditions (European programme). Report EUR
I
1007~, Commission of the European Communities, Brussels, Belgium,
1986.
I
I
12
II'
I
I
- 6 HOBBS, J. H., SATTARY, J. A. and HAxVELL, A. D. Experimental data
for the determination of basic 250 mm orifice meter discharge
I
coefficients (European programme). Report EUR 10979, Commission of
the European Communities, Brussels, Belgium, 1987 .
7 HOBBS, J. H., SATTARY, J. A. and HAXVELL, A. D. Experimental data
for the determination of 250 mm orifice meter discharge coefficients
under different installation conditions (European programme).
Report EUR 10980, Commission of the European Communities, Brussels,
Belgium, 1987.
~
10
European programme. Progress Report No PRll: EUEC/17 (EECOOs) .
East Kilbride, Glasgow: National Engineering Laboratory Executive
Agency, 1992.
SATTARY, J. A., SPEARMAN, E. P. and READER-HARRIS, H. J.
Experimental data for the determination of basic 50 mm orifice meter
discharge coefficients - European programme. Progress Report No
PRI2: EUEC/17 (EECOOS). East Kilbride, Glasgow: National
I Engineering Laboratory Executive Agency, 1992.
11 SPEARHAN, E. P., SATTARY, J. A. and READER-HARRIS, H. J. The EEC
orifice plate project: index to the data tables. Progress Report No
I PRI3: EUEC/17 (EECOOs). East Kilbride, Glasgow: National
Engineering Laboratory Executive Agency, 1992.
-I 13
I
18 JOHANSEN, F. C. Flov through pipe orifices at lov Reynolds numbers.
Aeron. Res. Committee, Report and Hemo No 1252. London: HH
Stationery Office, 1930.
-
19
20
ENGEL, F. V. A. Durchflu~essung in Rohrleitungen.
ein Symposium in East Kilbride (Schottland).
Brennstoff-Warme-Kraft, 13, pp 125-133, 1961.
ENGEL, F. V. A. Nev interpretations of the discharge
Bericht fiber
(C. - Co )
For the ith point in a cell Per cent error, Pi = 1m le x 100 ,
I Cim
where Cim is the measured discharge coefficient of the ith point
I and C
(24)~e
is the corresponding discharge coefficient from equation
I N
"
I
Mean per cent error, ~ = ---------
N
I N
L (Po
1
_ ~)2
N - 1
I
..
I
,
1
Po2
1
'b
I Statistics for the entire population appear in the bottom right hand
cell.
I
I
I
--I 15
I
RESIDUALS
TAB
FROM EQUATION
L E 2
2142
0.242
- --I
0.065 0.027 0.099 0.040 0.001 -0.101 0.004
0.500 0.294 0.055 0.198 0.100 0.164 0.119 0.205
(0.4825
to
-
398
-69 -
285
-
109
-
392
-
526 1779
- I
0.5003) 0.300 0.061 0.221 0.107 0.163 0.156 0.205
0.570
-0.024
0.386
0.050
0.076
0.010
0.232
-0.049
0.109
0.059
0.258
-0.110
0.120
0.001
0.249
I
(0.5427 - - - - - - -
t()
0.5770)
348
0.386
72
0.091
992
0.233
136
0.120
1123
0.264
567
0.162
3238
0.249 I
J
-0.025 0.119 -0.007 -0.120 0.072 -0. 123 -0.015
0.660 0.287 0.102 0.236 0.128 0.190 0.151 0.224
(0.6481
t()
-
498
-
64
-
627
-
92
-
823
-
643 2747
-
0.6646) 0.287 0.156 0.236 0.175 0.203 0.195 0.225
I
0.023 0.114 0.105 0.097 -0.005 -0.238 0.005
0.750
(0.7239
t()
0.322
-
866
0.105
-
101
0.312
-
971
0.203
-
130
0.315
-
1478
0.359
-
336
0.326
3882
- I
0.7509) 0.322 0.155 0.329 0.225 0.315 0.430 0.326
0.031
0.403 0.310 0.383 0.000 0.353
I 50 -
728
-
1605
-
728
-0 -
3061
0.410 0.311 0.383 0.000 0.354
I 0.000
0.000
-0.045
0.210
0.000
0.000
0.000
0.000
-0.045
0.210
75 - - - - -
I 0
0.000
469
0.215
0
0.000
0
0.000
469
0.215
P 100
.-0.003
0.224
-
-0.002
0.254
-
0.061
0.323
-
0.000
0.000
-
0.013
0.266
-
1084 1932 933 0 3949
I 0.224 0.254 0.328 0.000 0.266
I 150
0.000
-
0
0.192
-
792
0.000
-
0
0.000
-
0
0.192
-
79
0.000 0.194 0.000 0.000 0.19
I 0.042 -0.027 0.006 0.056 0.011
0.237 0.210 0.266 0.305 0.250
- - - - -
I 250
1155 1841 1167 885 5048
..
0.241 0.212 0.266 0.310 0.251
I
I
--I 17
I
RESIDUALS
TAB L E 4
-0.003
0.250
-0.026
0.293
-0.082
0.000
-0.507
0.252
-0.320
0.193
-0.020
0.173
-0.009
0.186
0.064
0.128
0.008
0.205
0.001
I
0.570 0.965 0.361 0.238 0.196 0.272 0.225 0.249
(0.5427
to
-
18
-
59
-
502
-
1420
0.196
-
776
0.279
-
463
0.225
-
3238
0.249
I
0.5770) 1.066 0.480 0.238
0.660
(0.6481
-0.030
0.116
-
-0.296
0.391
-
-0.069
0.283
-
0.035
0.203
-
0.045
0.187
-
-0.087
0.180
-
-0.015
0.224
- J
to
0.6646)
5
0.108
35
0.486
466
0.291
1110
0.206
471
0.192
660
0.200
2747
0.225 I
0.005
0.750
(0.7239
0.000
0.000
-
0.353
0.443
-
-0.067
0.339
-
0.042
0.299
-
-0.018
0.310
-
-0.038
0.358
-
0.326
-
I
to 0 78 615 1668 1045 476 3882
0.7509) 0.000
0.086
0.565
0.027
0.345
-0.039
0.302
0.013
0.310
0.037
0.360
-0.043
0.326
0.001
I
Summary 0.593 0.392 0.257 0.225 0.260 0.239 0.269
by
ReD
-
470
-
675
-
4016
-
6089
-
3138
-
1988
-
16376
I
0.599 0.393 0.260 0.226 0.263 0.243 0.269
I
18
--I
.w-----, - - - - ,--- --,.-
0.055
0.050
o
x
~
i
I
"J
o "J
0.045 "Jx g !). !).
~ ~ X
X x
"J
0.040 o
~a x
"J ~
<0- X ~ o
<:a. 0.035 x
0
,...
"- "J.
<0-
0 ~ v
<:a. 0.030
I
<> x
o !).
~ "J
'"'a. 0.025 ~
u
:J ! Equation (1 )
"J
<I 0.020
13'1/(1-13'1)
0.015 !). 0.05 o.1
"J o.1 0.2
0.010 + 0.2 0.3
x 0.3 0.4
0.005 0 0.4 0.5
o 0.5 0.6
"J
0.000
0.0 o.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 .0 1 .1
L,
FIG 1 Upslream lapping lerm as a Funclion of L,
o.ooo~~------------------------------------------------~
--- Equo Li on (4)
-0.002
I
C
I
A
~
-0.004
..,
E I I
A
C
~
F I FI
0
'tl
U
<I -0.006 I
~
C C
A
A
B
-0.008
c
-0.010+-----.-----.-----.----.-----.-----.-----r-----r----.---~
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
~
. .
FIG 2 Downstream tapping term at the pressure mInImum
-~-----~----~----_&-
- - - - -,. - - - - -- - -~-
o.ooo~------------------------------------------------~-------
o ----- Equation (5)
t.0.25 - 0.35
-0.002 x 'J0.35 - 0.45
+ 0.45 - 0.55
x 0.55 - 0.65 0
0
-0.004 0 0.65 - 0.75
o 0 0.75 - 0.80
o
c -0.006 o
~ o o
o o o o
u x x o o
u X 0
<l -0.008 + o o o
'J X XI 0
.
('l X
-I
In. -0.010
x
x-__~~ __~~~O~~~~~D~~O 0
o x o
x
o
-0.012 o
+
-0.014
+
-0.016+------r----~----_.------._----._----_.----_,------,_--~
0.0 0.5 1 .0 1 .5 2.0 2.5 3.0 3.5 4.0 4.5
M2'
0.67 FLo.ng~: 50 mm
'"+ FLange: 75 mm
0.b6 0 FLange: 100 mm
u a FLangQ: 150 mm
..,c 0.65
'V
x
FLange: 250 mm
FLange: 600 mm
....
(l> <) Corner or Corner( GU)
.....u 0.61 lIE 0 and 0/2
u,
u,
(l> 0.63
0
u
(l> 0.62
(J\
L
0
L 0.61
u
.....lfJ
0 0.60
0.59
0.58
3.5 1.0 1.5 5.0 5.5 6.0 6.5 7.0 7.5 8.0
Log10 Red
0.60
0.59
3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0
Ll ;;; 1 .000 ; v ; -- --
0.
:I
,
Ll = 0.500~ [J --_.- Equation (6) (a=1,b=1 .307,n=O.9)
U
= 0.250: o ., ---_.
--
<l Ll
Ll ;;; o. 125 ; o ;
~~
0.05
0
v
ov 0
<>
o.oo+- -. .- -. .- -. .- ~
3.8 4.0 4.2 4.4 4.6 4.8 5.0 5.2
0.0008
0.0006
0.0004
0.0002
x
0.0000 , ,
0 50 100 150 200 250 300 350 400 450 500
d( mm )
0.05_
/'+
.+
A A
-, .~
<A. .~ +
00 G- /0 t. x
~.iOO x
FIG 8 r e as a Funct i on of d
I
t' J E Gallagher, Shell USAand R E Beaty, J\rooco USA
I
I
Paper 1.2
I
I
Ie IDRl'H SEA F'IIJfJ MEASUREMENT VDRKSHOP
I
I
I NEL, East Kilbride, Glasgow
I
II
II
I
-.
I
ORIFICE METERING RESEARCH - A USER'S PERSPECTIVE
I J. E. Gallagher, P.E. - Shell, USA
R. E. Beaty - Amoco, USA
I SUMMARY
I Several physical
natural gas
properties are required for fiscal calculations of
compressibility, absolute viscosity, isentropic
exponent, CSTAR for sonic nozzles and calorific value. A summary
I of the current
needs .
status is presented along with additional user
..I The orifice meter has a long history of use and experimentation.
Because of this, it sometimes may be perceived as outdated,
inaccurate and unreliable. In reality the opposite is true. To
illustrate this viewpoint, loss control performance is presented
for several pipeline systems transporting compressible fluids.
NOTATION
I Beta
CSTAR
diameter ratio, (Iit/Df)
critical flow factor for gas flowing through a
sonic nozzle
I orifice plate bore diameter at flow conditions
meter tube internal diameter at flowing
conditions
I D nominal pipe diameter
II
1 of 31
I
I
Orifice Metering Research - A User's Perspective eI
Ev velocity of approach factor
I
perfect isentropic exponent
~
L1 distance from flow disturbance to flowmeter in
nominal pipe diameters
I
L2 distance from piping disturbance to inlet of
L3
flow conditioner in nominal pipe diameters
distance from outlet of flow conditioner to
flowmeter in nominal pipe diameters
I
length of flow conditioner in nominal pipe diameters
fluid density at flowing temperature and pressure
fluid density at base conditions
I
1 XtiTAOOCCTION
I
The North American natural gas industry produces, transports, and
distributes approximately 700 billion cubic meters of gas each year
(25 tr illion standard cubic feet).
I series.
2 WORLDWIDE DISCHARGE COEFPICIENT EQUATION
I The greatest series of orifice coefficient of discharge data
completed prior to 1980 was conducted at Ohio State University
I (OSU) under the direction of Professor S. T. Beitler. All of these
experiments were conducted between 1932 and 1933 on water using
seven pipe diameter ranging from 25 to 350 mm (1 to 14 inches). It
is important to note that the experiments were conducted before the
I existence of any national or international orifice metering
standard. The OSU data base was adopted by Dr. Edgar Buckingham and
Mr. Howard Bean to derive mathematical equations to predict the
I flow coefficient for orifice meters. Pioneers like Buckingham and
Bean developed excellent equations based on the data at that time.
t' A joint committee from the American Gas Association (A.G.A.) the
American Petroleum Institute (API) and International Standards
Organization (ISO) was formed in the early 1970's to investigate
perceived problems with the OSU data base. Jean Stolz from France
I and Wayne Fling from the united States collaborated in the data
base assessment. They discovered upon analysis of the OSU data
that only 303 of the data points were technically defensible.
I In 1978, Jean Stolz derived an empirical orifice discharge
coefficient equation which physically linked near field pressure
tappings. This innovative equation was based on the OSU 303 data
I set for flange and radius tappings. The corner tap data was
obtained from witte's corner tap experiments conducted in the 30s.
In 1980, the Stolz equation was adopted by ISO 5167 replacing the
I Buckingham equation.
I and 250mm.
The EC and API/GPA data were collected using oil, water, natural
gas and air as the test media. The meter tubes used in the test
I were manufactured from commercial pipe. The criteria for the
experiments was a uniform fully developed velocity profile. When
the data base was combined into a regression data set, the US and
I EC experiments yielded highly compatible data.
It 3 of 31
I
I
orifice Metering Research - A User's Perspective
eI
Xn 1988, international cooperation between the North American and
I
EC flow measurement experts resulted in unanimous acceptance of the
equation form proposed by Reader-Harris of the United Kingdom's
National Engineering Laboratory with two amendments by Gallagher. I
The combined EC and API/GPA experimental pattern was well balanced
so that the data could be accurately evaluated for corner, flange
and radius tappings. The Reader-Harris/Gallagher (RG) equation was
regressed using this combined data set of 10,152 corner, flange and
I
radius tap points. At the instruction of API's Board of Directors,
this equation was balloted and subsequently adopted into the latest
revision of A.G.A. Report No.3 (ANSI 2530/API HPMS 14.3/GPA 8185)
I
published in 1990.
Since the regression of the RG equation, additional EC discharge I
coefficient data on oil, water and natural gas has been accumulated
on 50 and 600mm meter sizes. To date, none of the additional
flange tapped data has fallen outside the predicted uncertainty of
the RG equation. The RG equation could be updated to reflect the
additional data. However, no equation is expected to significantly
ea
improve the prediction results for flange tappings. I
3 FLOW CONDITIONS
All flowmeters are subject to the effects of velocity profile,
swirl and turbulence structure approaching the meter. The meter
I
calibration factors or empirical coefficients calculated from the
discharge coefficient equations are valid onlv if similarity exists
between the metering installation and the experimental data base.
I
These parameters should not be significantly different from those
at the time of meter calibration, or from those which existed in
the empirical coefficients of discharge data base. Technically I
this is termed the Law of Similarity.
Many piping configurations and fittings generate disturbances with
unknown characteristics. Even a simple elbow can generate very
different flow conditions from "ideal" or "fully developed" flow.
J
In reality, multiple piping configurations are assembled in series
generating complex problems for standard writing organizations and
I
flow metering engineers. The problem is to minimize the difference
between "real" and "fully developed" flow conditions on the
selected metering device to maintain a low uncertainty associated
with the fiscal application. For clarity, we will refer to this as
I
"pseudo-fully developed" flow.
A method to circumvent the influence of the fluid dynamics (swirl,
I
profile and turbulence) on the meter's performance is to install a
flow straightener in combination with straight lengths of pipe to
"isola te" the meter from upstream piping disturbances. Of course,
I
this isolation is never perfect. After all, the straightener's
objective is to produce a "pseudo-fully developed" flow.
I
4 of 31
II
I
I
.. Orifice Metering Research - A User's Perspective
Ie
this region. In contrast, the inner region (or pipe wall) has a
"shorter memory". consequently, the recovery from any disturbance
of the inner region will be much quicker than that for the outer
region. Also, we cannot forget the interdependence of the two
I regions.
3.2 Types of Plow conditioners
I Flow conditioners may be grouped into three general classes based
on their ability to correct the mean velocity profile, bulk swirl
I
I
orifice Metering Research - A User's Perspective
eI
The second class of straighteners is designed to generate an
axisymmetric velocity profile distribution by subjecting the flow
to a single or a series of perforated grids or plates. The profile
I
is redistributed by use of the blockage factor or porosity of the
flow conditioner. This class of straighteners includes the
Sprenkle, Zanker and Mitsubishi designs.
I
The third class of straighteners are designed to generate a
"pseudo-fully developed" velocity profile distribution through
porosity of the straightener and the generation of a turbulence
I
structure. The turbulence structure is generated by varying the
radial porosity distribution. This class of straighteners includes
the Sens and Teule, Bosch and Hebrard, K-Lab and Laws designs
I
(Figure 5).
The optimal flow conditioner has the following design objectives: I
elimination of swirl
production of axisymmetric,
mean velocity profile
production of pseudo-fully
pseudo-fully developed
developed turbulence
--I
structure
low pressure drop across the flow conditioner I
*
*
rigorous towards mechanical damage
tI
cyclic forces are no~ controlled by the conditioner for self-
maintenance of tlJe velocity profile with respect to the axial
position.
All flow conditioners
limitations -
have the following shared geometrical I
* A minimUlll distance between the upstream piping
elements and the inlet of the straightener to ensure I
that the straightener performance is optimized.
I 4 XNSTALLATXON RESEARCH
Historically, flow conditioners have defined an acceptable
uncertainty of +l> 0.50 percent due to piping elements. The
I original research of A.G.A. on orifice meter effects could not
discern effects below this level due to the limitation of the
research equipment in the 30s, 40s, 50s and 60s.
I with the birth of microchip technology, large steps towards
lowering the uncertainty are possible due to the advent of smart
transmitters, sophisticated flow computers, personal computers,
I Computational Fluid Dynamics (CFD), thermal anemometry (TA) probes
(i.e., hot wire, hot film, x-wire), Laser Doppler Velocimetry or
Anemometry (LOV/LOA), characterization of flow meters in real time,
I high pressure gas piston provers, ultrasonic flowmeters, coriolis
flowmeters, videoimagescopes, etcetera. These "new" tools are
providing significant advances in the refinement of existing
tt metering equipment as well as the birth of new technology.
I The optimum turbulence model does not currently exist for CFD
installation effects' applications. Hopefully, existing and
planned turbulence structure measurements and installation effects
I
II 7 of 31
I
I
orifice Metering Research - A User's Perspective eI
research will provide future scientists with the needed insight for
this development. Irrespective of this limitation, CFO technology
I
will still be utilized to maximum the experimental pattern
efficiency and to provide sensitivity analyses.
I
5 ORIFICE ME~ER XNS~ALLATION EFFECTS
The goal of current orifice research programs is to focus on the
effects of various installation conditions for natural gas
I
applications. since significant deviations from the Law of
similarity cause measurement errors, the main goal is to identify
and quantify the error associated with these flow disturbances. I
Present industry standards provide installation specifications for
I
..
pipe length requirements and flow conditioner location upstream of
orifice meters (A.G.A. 3/ANSI 2530/API MPMS 14.3 and ISO 5167).
Unfortunately, considerable disaqreement over straight length
requirements exist between these two highly respected standards.
CUrrent upstream effects research has focused on assembling
experimental data for evaluation of straight length requirements
stated in the respective standards.
In North America, the design practice is to minimize the upstream
I
piping and utilize A.G.A. tube bundles to provide "pseudo-fully
developed" flow. Typical North American installations consist of
90 degree elbows or headers upstream of the orifice meter. Tube
I
bundles are cOJlllllonly
used to eliminate swirl and distorted velocity
profiles.
upstream
In ltiesternEurope, the practice is to utilize long
lengths to generate "pseudo-fully developed" flow. I
Because of these design differences, the current research programs
do not fully complement each other in their direction.
I
5.1 Flow conditioner Location
Gas Research Institute's Meter Research Facility (MRF), loca.tedat
southwest Research :Institute (SWRI) in Texas, was constructed to
carry out definitive research in key flow metrology areas for the
rI
natural gas industry. In light of this charter, swRI is conducting
a series of experiments to address the user community concerns. I
The sliding vane technique is essential to the efficient and
effective research program at SwRI. This technique relocates the
tube bundle without venting the meter run or disconnecting flanges,
I
thereby saving considerable time and manpower without introducing
additional laboratory uncertainties. Results for a 100mm tube
clearly indicate the existence of a cross-over zone (Figure 6).
I
Comparison of the results for an Ll of 1000 and 450 indicate no
perceptible difference in the cross-over zone between the two
upstream lengths. These results are consistent with research data I
from several Western European flow laboratories. For example, the
European community (Ee) program conducted at Gasunie and NEL
I
8 of 31
--I
I
.. Orifice Metering Research - A User's Perspective
I
--I 9 of 31
I
Ori~ice Metering Research - A User's Perspective eI
possible.
I
5.2 praqaatic Solutions
The guiding principle of this section is to provide a systematic
compilation o~ instructions and basic rules for the location of
I
~lo",conditioners for nev and existing orifice metering facilities.
To assist the user community, the authors propose two methods for
I
determining the proper location of A.G.A. tube bundles - fixed
location method and correlation method. The Beta should be limited
to a ranqe of ~.20 to ~.60 for both approaches. I
For the first method, the authors recommend a fixed location for
I
..
the tube bundle as follows:
a minimum L2 length of 3D
For the second method, the authors propose a correlation which
predicts the cross-over zone for A.G.A. tube bundle (19 tube radial
design) flow conditioners. The correlation predicts the cross-over
I
ZODe as a function of Beta, EV, L1, and L3 for A.G.A. tube bundles
(Figures 8 and 9). In the authors' opinion, a minimum L2 length of
3D is applicable for all flow conditioners. This requirement has
I
been included in the correlation method.
Hany factors affect the solutions to flow conditioning problems -
economics, piping limitations, operating ..,indows,etcetera. The
I
options proposed by the authors provide pragmatic solutions to real
problems for operators of orifice metering facilities (Figure 10
and 11). The user selection is based on economic justification to
I
minimi2e the uncertainty associated with upstream conditions.
6 XN SXTU CALXBRATXON
6.1 xntroduction
I The first measurement standards were based on weight. Any
commodity to be traded could be judged against a known weight to
I evaluate its true worth. with solids the procedure is relatively
simple. However, liquids measured with this procedure require
additional information if the data obtained is to be reported in
traditional units. The additional information is the density of
I the fluid. To avoid measurement inequities, corrections are
required for acceleration due to local gravity and air buoyancy.
These correction converts observed weight to mass. It is obvious
I that this procedure is well suited for a laboratory environment.
verifying the accuracy of flowmeters in specific applications has
Ie
been one of the desires of the user community. Shop tests of
orifice meters and turbine meters with various upstream
configurations has been conducted for several years to aid in the
design of high volume metering facilities. The question posed by
I the user community is - "Can the operator ensure the parties
involved in the fiscal transfer that the measurement station is
adequately described by the tested design?".
I At this point, the scientific community followed three parallel
branches for high pressure/volume applications -
I * high pressure bell provers (> 33 bars)
*
I *
"bootstrapping" method
sonic or critical flow nozzles
II 11 of 31
I
I
Orifice Metering Research - A User's Perspective .-
situ calibration devices - sonic nozzles, master meters and piston I
provers. These tests, planned for 1993, will be conducted at GRI's
HRF facility.
One of the oldest laboratory volumetric gas measurement standards
I
is the bell prover. The early systems employed a low pressure bell
which was liDi ted to extremely low pressures. As measurement
technology progressed, the demand to measure gas at elevated
I
pressures increased.
The "bootstrapping" method requires a bell prover to prove a number
of highly accurate lower capacity meters. These parallel meters
I
are, in turn, used to prove larger capacity meters. A group of
these larger capacity meters are then used in parallel to prove an
orifice meter or turbine meter. The process works well in the
I
laboratory, but is less acceptable in a field environment.
master turbine Deter proven at low pressure and operated at
elevated pressures will exhibit K-factor shifts in the positive
direction. When air is used as the proving media for meters
normally used to measure natural gas, conversion errors occur.
Another major problem is one of logistics. Moving a large group of
A
--I
master Deters to a number of locations is difficult. Equally
difficult is the ability to duplicate actual operating conditions
in the laboratory. Since transporting master meters may result in
damage, time consuming validation cross testing may be required at
I
each test site. contaminants in a gas stream can damage or reduce
the accuracy of the master meters. As a result, field tests can be
biased and costly repair and recertification of the master Deter is I
required.
sonic nozzle technology remained dormant until Matz, Smith and
Stratford's design work in the early 60s. Real gas critical flow
I
factors, CSTAR, were developed by Johnson of NASA in 1965 allowing
the application of sonic nozzles for calibration purposes. Sonic
nozzles have been successfully applied as laboratory standards on
single component gas streams. Varying degrees of success has been
J
achieved on multiple component gas streams in field and laboratory
applications.
incorporates
A modification of the sonic nozzle, the Digicell,
eleven parallel sonic nozzles sized in a binary
I
progression and Dounted in a single housing.
For sonic nozzles, clean, dry natural gas is a prerequisite.
I
Solids and liguids swept along with the gas can damage the bore of
the sonic nozzle.
the nozzle throat.
Some contaminates (i.e., sulfur) can deposit in
Entrained or free water in the natural gas I
combine to form hydrates due to the drop in flowing pressure and
temperature. Additionally, hydrocarbon dewpoint concerns are real
for multiple component streams. compositional analysis from a gas
chromatograph in combination with an equation of state are normally
I
used to predict the physical properties of the flowing gas. The
sonic nozzle's mass flow equation requires an iterative solution to I
12 of 31
II
I
I
Orifice Metering Research - A User's Perspective
.. ----------------------------
determine the critical flow factor, CSTAR. The master meter's mass
I flow equation requires the calculation of density at both operating
and base conditions. Analytical errors and uncertainties related
to the physical properties' predictions increase the uncertainties
t'
North America. At this same time, the liquid small volume prover
(SVP)~ demonstrated the acceptability of double chronometry
interpolation techniques for turbine meters. Through the
pioneering efforts of Gasunie, Shell, Amoco, DSM, Ruhrgas and
I Ogasco, a modified SVP approach was developed for chemical, C02 and
natural gas applications.
The modified SVP technology launches a piston into the flowing
I stream and measures its progress through the prover using high
precision detector switches. The area between the detector
switches, known as the calibrated section, is traversed by the
I piston for calibration purposes. Most designs use free floating
pistons, although, ram assisted pistons work equally well.
critical requirement is to assure that the pressure disturbance
The
Ie TWo gas piston prover designs are currently available - single wall
and double wall. Both design are available in unidirectional or
I bidirectional options.
In the single wall design, the pipe wall acts as both the measuring
chamber and pressure containment vessel. As a result, correction
I for the change in the calibrated volume is required due to the
flowing temperature and pressure.
I
II 13 of 31
I
I
Orifice Metering Research - A User's perspective eI
The piston prover has been used as a primary standard to prove a
turbine meter or a master turbine meter installed in series with an
I
orifice meter. This arrangement ensures calibration under normal
operating conditions - velocity profile, instrumentation, etcetera.
Additionally, wet gas will not alter the performance of the piston I
prover as long as the flowing conditions are not lower than the
hydrocarbon dew point. Small amounts of fine solids will have no
effect unless the bore of the calibrated section or the piston
seals are damaged.
I
At this time, the high pressure gas piston prover is being
successfully applied by Gasunie, Shell USA, Amoco USA and Ruhrgas
I
for fiscal applications and/or laboratory flow standards. Gasunie
has replaced their "bootstrapping" method with the piston prover.
Shell USA has applied this technology to chemical and C02 systems I
--I
since 19S4 to identify out of tolerance metering facilities. Amoco
USA has operated an ogasco design for the calibration of small
turbine meters in a coal degasification project since 1990.
Clearly the piston prover offers the best opportunity for
successful field calibration of flowmeters on multiple component
natural gas streams. However, the measurement community needs;
additional research, an established gas piston prover design and
certification standard. The additional research needs will be
assessed with the GRI activities for 1993. In answer to the need
for standards, the API Committee on Gas Measurement (COGM) has
I
recently established a Working Group to address the global
conmunity input and concerns. I
7 PHYSICAL PROPERTIES
Several physical properties are required for fiscal calculations of I
natural gas compressibility, absolute viscosity, isentropic
exponent, CSTAR for sonic nozzles and calorific value.
7.1 Density J
Accurate values of flowing density, RHO..
" and base density, RlID.,
are required for accurate base volume calculation. The densities
I
may be obtained using two methods - direct measurement using
density meters or an acceptable equation of state.
I
Online density meters, for both flowing and base density, has
exhibited problems when a fluid is measured near a phase boundary
or passes throuqh the hydrocarbon dew point. Small amounts of
liquid dropping out of solution will cause density meters to
I
perform erratically. Calibration of gas density meters, not
conducive to field operations, should be performed in an ISO 9000
certified laboratory.
I
The use of equations of state to accurately predict the density of
I
14 of 31 II
I
I
'- Orifice Metering Research - A User's Perspective
I gases and liquids has met with varying degrees of success. When
the data used to derive the equation was representative of the
flowing fluid, the results have been acceptable. The density
I uncertainty increases as the operating composition, pressure or
temperature deviates from the data base used for the equation of
state.
I Attempts have been made since the turn of the century to develop an
equation of state that accurately predicts the physical properties
I of a variety of natural gas streams. Developers were hindered in
their endeavors by a lack of high quality data.
Research directed by Howard Bean of the United states National
I Bureau of Standards produced the first generally accepted
supercompressibility data in 1928 and 1929. The data was limited
to 4 Mpa (600 psia). The next significant body of work in the us
t' was published by Professor Samuel Beitler of Ohio state University
(OSU) in 1954. The Beitler work was extended and an equation of
state-was completed in 1962 by Mr. R. H. Zimmerman at OSU. The
Ie approximately 6 Mpa (900 psia). The data was obtained from open
literature as well as data supplied by the Groupe Europeen de
Recherches Gazieres (GERG). GERG continued to expand their high
I quality data base through 1989. This work demonstrated that the
equation of state used in Report No. 8 needed to be improved.
also showed the velocity of sound data obtained under GRI
It
I also utilized the new data collected and revised the generalized
equation of state for the 1992 Report No.8. Both the GERG and the
GRI equations are utilized in the revised report.
It 15 of 31
I
I
Ori~ice Metering Research - A User's perspective
.-
No. ~ is recommended
ranges.
for use within restricted compositional
The report also states that an acceptable database
water, heavy hydrocarbons,
for
and hydroqen sulfide is not available at
I
this time. The accuracy of the report is likely to be equivalent
to other existing generalized equations of state. I
International cooperation towards development of a single standard
for the prediction o~ compressibility
gas streams has been highly successful
of multiple component natural
due to strong leadership
I
from the western European and North American communities. This
effort has culminated
at the ISO Technical
A.G.A. Distribution
in a draft standard currently being balloted
Committee 193, A.G.A Transmission
committee and API COPM levels.
committee, I
Large errors usually
analysis arena.
occur in the gas sampling and compositional
In the United states, most analytical work is
1
perforned using a gas chromatograph.
been made in gas chromatography
the capillary
produce accurate extended analyses.
the need for highly
standards.
accurate multiple
significant improvements have
in the past decade. The advent of
column has increased the ability of the analyst to
The major problem remaining is
component chromatographic
Pure component calibration is not economically feasible
--I
on a routine basis. As this problem is solved, laboratory
analytical uncertainty will be reduced. As the laboratory problems
are resolved,
systems will
the use of sophisticated
be followed
online gas chromatographic
by sophisticated portable systems.
1
Systems now being marketed are capable of analysis through nonanes
plus.
I
7.2 Absolute Viscosity
The absolute
Reynolds number
fluid viscosity is required to calculate
for the new orifice coefficient
the pipe
of discharge
I
equation. Viscosities may be measured of computed from appropriate
equations
the accuracy
applications
of state. In some areas additional data could improve
of the predicted viscosity.
the absolute accuracy
For high Reynolds number
of the viscosity is not as
J
critical.
inaccuracies
application
The coefficient
in Reynolds
the accuracy
of discharge is not as effected by small
number. In low Reynolds
of the viscosity prediction
number
has a much
I
greater impact on the accuracy of the volume calculation.
I 7.3 CSTAR
I empirical coefficient
predictability
data bases or in situ calibrations,
and variations in the fluid properties,
uncertainties associated with the secondary devices.
and
II 17 of 31
I
I
Orifice ~etering Research - A User's Perspective
eI
The first loss control example is one of four polymer grade I
ethylene grids operated by Shell USA. The annual commodity value
for the grid presented is approximately 800 million USD (Figure
12). The results exhibited in the graph were achieved as a result
of; in situ calibration, enforcing a Beta range of 0.2 to 0.6,
I
maintaining an L3 of 10 to 12D and an L1 of at least 170. The in
situ calibration was accomplished using small volume prover (SVP)
technology in combination with master turbine meter techniques for
I
compressible fluids. The calibration factor was used for
analytical purposes only, not for fiscal purposes. The results of
the orifice meter calibrations were used to identify. facilities
which exhibited high bias errors. Upon investigation, the bias
I
associated with the orifice meters was always a result of physical
deviations from the standards, human errors or electronic errors. I
The second exanple is one of two carbon dioxide systems operated by
Shell USA (Figure 13). The annual throughput for the C02 system
presented is approximately 6 billion cubic meter of gas (200
billion standard cubic feet). The same techniques applied to the
ethylene qrid were utilized on the carbon dioxide system.
addition, the najority of the physical problems were identified
I:n
--I
through the use of videoimagescope technology.
The orifice meter has a long history of use and experimentation.
Because of this, it sometines may be perceived as outdated,
I
inaccurate and unreliable. I:nreality the opposite is true. The
orifice meter is a well established device with known weaknesses
and strengths. I:f applied with expertise, the long term I
performance is exceptional.
, FOTURE RESEARCH DIRECTION I
Additional research into the application of orifice meters and
metering standards developnent is justified by the current capital
investments. Maximization of current capital investments is an
efficient and effective approach for the petroleum, chemical and
II
natural gas industries.
continuation of current installation effects research for orifice
I
meters (and other flowmeters) will identify uncertainty
linitations. Assessment of new flow conditioner designs would
improve users' alternatives. Two new approaches (LaWS, K-Lab) to
I
flow conditioners has shown significant potential towards
elinination
pressure drop.
of upstream disturbances with minimal permanent
The benefits to the global community will be I
efficient and effective improvements of line integrity and loss
performance with Dinimal capital investments.
In situ calibration techniques for multiple component gas streams
I
should be assessed in light of achievable laboratory and/or field
uncertainty estimates. In particular, two areas have particular I
18 of 31
II
I
I
-.
I Orifice Metering Research - A User's Perspective
I smart transmitters,
inspection intervals
chromatographs, etc.), required physical
(i.e.,videoimagescopes, etc.), required
certification of field standards on specified intervals and
statistical footprinting of field devices and standards.
I Additional improvements in the prediction of physical and transport
properties for multiple component gas streams are a prerequisite
I for the 90s. CSTAR for sonic nozzles, viscosity and perfect
isentropic exponent calculations, extension of the compositional
limits for density or compressibility predictions are the most
1.0 ACXNOIILl!:DGEMEJn'S
I
The authors wish to thank the manaqement of Shell USA and Amoco USA
for their valuable support and dedication towards publications of I
--I
the results and opinions presented herein.
In addition, the authors wish to thank 'the Gas Research Institute
(GRI) , National Institute of Standards and Technology (NIST), API's
committee on PetroleUlllMeasurement (COPM), American Gas Association
(A.G .A.) Gas Processors Association (GPA), the European
Community's Dr. D. Gould, and many individuals for their commitment
to e:xcellence.
We also want to express our sincere respect to -past researchers
such as Buckinghan, Bean, Beitler, witte and Marchetti.
I
REPERXHCES I
Worldwide Discharge coerticient Equation
it Concentric,
16,000,000";
Flange-Tapped, Square-Edged orifice Meters in
Natural Gas Over the Reynolds Number Range of 25,000 to
NIST Technical Note 1270, september 1989,
'-Gaithersburg, MD, USA.
I EC Experimental Program
-.I 9.
Belgium.
Hobbs, J. M., "The EC orifice Plate Project, Part I:
Traceabilities
PR5:EUEC/17,
of Facilities Used and Calculation Methods
Employed"; Commission of the European communities,
1987, Brussels, Belgium.
Report
10. Hobbs, J. M., "The EUEC Orifice Plate Project, Part II:
I 12. Watanabe,
corner-Tapped
N., "coefficients
Orifice Meters";
of Discharge
National
for
Research
II 21 of 31
I
I
Orifice Metering Research - A User's perspective
Laboratory
Japan.
of Metrology, Private c01lllllunication,
1987,
--I
Equation pevelopment
I
J.3. Stol~, J., "A Universal Equation for the Calculation of
Discharge Coefficient of orifice Plates"; Proe. Flomeko 1978
_ Flow Measurement of Fluids, H. H. Dijstelbergen and E. A.
Spencer (Eds), North-Holland Publishing Co., Amsterdam
I
(J.978),pp 519-534.
~lIstallatioliResearcb
I
23. Mattingly, G.E., Yeb, T. T., "summaryReportofNIST'S
I
22 of 31 II
I
I
.. orifice Metering Research - A User's Perspective
it 26.
Gas 'Measurement School, USA, March, 1929.
Sattary, J. A., "Progress Report No. 10 on The Effect of
~nstallations and straighteners on the Discharge
I Coefficient of Orifice Plate Flowmeters", Department of
Trade and Industry, National Engineering Laboratory, East
Kilbride, Scotland, March, 1991.
I 26. Scott, J. L. et al, "The Effects of Flow Conditioners and
Tap Location on Orifice Flowmeter Performance", NIST
Technical Note 1352, U.S. Department of Commerce,
I Washington, DC, 1991.
28. Morrow, T. B., "Orifice Meter Installation Effects:
I Sliding Vane Test With a 1000 Meter Tube and a Tube
Bundle straightening Vane Downstream of a 90 Degree
Elbow", GRI Technical Memorandum No. MRF-UE-04, Gas
Ie 29.
Research Institute, chicago, IL, March, 1992.
Morrow, T. B., Park, J. T., "Orifice Meter Installation
Effects: Sliding Vane Test with a 170 Meter Tube and a
I Tube Bundle straightening Vane Located 130 upstream of an
Orifice Plate Downstream of a 90 Degree Elbow", GRI
Technical Memorandum No. MRF-UE-02, Gas Research
I 30.
Institute, Chicago, IL, March, 1992.
Morrow, T. B., Park, J. T., "Baseline Calibrations for
In situ calibration
I 31. Bellinga, H., Hoecks, C.P., van Laak, F. A. L., Orbons,
P.J., "piston Prover Used to Calibrate Gas Meters", oil
I & Gas Journal, August 19, 1985.
"I 23 of 31
I
orifice Metering Research - A User's Perspective
eI
32. Gallagher, J. E., "Shell Tests Indicate Ethylene Meters
Best Proved with Master Meter Method", oil & Gas Journal,
I
33.
December 15, 1986.
3B.
section Proceedings, 1989.
Beaty, R. E., "Field operation of a Portable Piston
I
Prover for Testing Gas Meters", American Gas Association,
operating section Proceedings, 1990. I
Pbysical Properties
39. American Gas Association, "PAR Research project NX-1.9;
A.G.A. Manual for the Determination of
II
supercompressibility Factors for Natural Gas", Arlington,
VA, 1969. I
40. American Gas Association Report No.8, "Compressibility
and supercompressibility for Natural Gas and Other
Hydrocarbon Gases", American Gas Association, Washington, I
O.C., 1985.
I
I
I
24 of 31
--I
I
..
Q
I Orifice Meter Piping Layout
I
I
1.<4
Flo~ ..
I ~
(g ~
fj I I I L3
I L4
I
L2 LC
I I
I L1
Figure 1
I
I
I Correlating Parameters
,.
... orO ........ 'R.tlo
Piping _peolf.lon_
L1 If_U~n.
I
L4 IOM (> - ID)
,,"Uw lID .. ...,.
I
L2 ept.lnulioM (> 3D)
U .... IoMk>ft.
FluId o,NatD
.'I
.,,. of cDilllUbenoe
R.rno1do Numb.r
wloolty profile
... 1I1.ngle
..... U.N ruoture
Figure 2
I
III 25 of 31
I
I
..
Q '.
lassification of Flow Conditioners I
TLib. Bundl..
1 Lo
I
Radial
H.agonal
Etoile
1
1
Lo
lolled I
ANoCAHoneycomb 1 lolled
MII.ubl.hl II 2 La
HI
I
2",,1<_ II II
Sprenkle II 15 HI
~
Laws III 2 Med
K-Lab
Sens T.ul.
Bo.en Hebrard
III
III
III
3
5
5
Med
VHI
VHI
'I
Figure 3 J
I A.GoA. Tube BundleJ I LAWS Conditioner J I
I
~
I
I
I
Figure 4 Figure 5 I
I
26 of 3~ II
I
't.
Q
I A.G.A. Tube Bundle
.'I Meter Tube Downstream of 90 Degree Elbow
it O.Of-I-----=~-------I
-0.2 - _.. - - - - - - . - - ...
-0.4 - - - - - - - - . - - - . - .. - - - - - - - -
- - . - - . - -
---------_.------------
I -0.6
-0.8 ._----------_ ... --------
-1.00 10 20 30 40 50 60 70 80 90 100
I
I A.G.A. Tube Bundle
Meter Tube Downstream of 90 Degree EI
I 1.0..-------------,
0.8
0.6 - -
0.4
0.2
.----------_.-------_._.
.- - - - -
- - - - . .- - -
0.01--_~~..o--------.:::::2----j
--. --- -
I ..
-0.2 - - - -....- - - - - - - . - - - - - - - - ..
..
.
I -0.8 - -A - - - - - - - - _ . - - - - - - - --
" Q
27 of 31
I
.r
Q I
I
'I.
r,
I
..
Figure 8
II
28 of 31
I
I
..
'
Q
I
I
Existing Facimio.
I
it I I J I
'I eoll L1ml1ltlQII RokK:lIo ConcUlbnor New Condltbn., k1 SIIu Calibratbn New FlowmelDr
'I. Figure 10
I
Ie New Facilities
I
I I I I
I Bnl L.ilnllalbn Nt. L2 & L3 ~na" Nt_ Cord.or. t In"uca~bn Nt_ Fw... ,
I,
I Figure 11
I'
ft 29 of 31
I
I
.-
Q '.
Ethylene Pipeline I
ual Loss Control Performance
'.
'.
2.00 .----------------,
B3 84 as 86 87 ee 69 90 91 92 93
I
Year
Figure 12
II
I
I
I by
I
W T Lake, Amoco Production Co
I and J Reid, NEL
t'
I
I Paper No 1.3
I
I
Ie NORTH SEA FLOW MEASUREMENT WORKSHOP
I
I
I NEL, East Kilbride, Glasgow
I
It
I
..
I C 0 N TEN T S
I 1 INTRODUCTION
.'I 2
4
CATS METERING DESIGN
TEST ARRANGEMENT
NBL TESTS
5 NBL TEST RESULTS
it 7 CONCLUSIONS
, 8
10
REFERENCES
LIST OF FIGURES
APPENDIX
I
I:
I
Ie
I
I
I
'I"
-I
I
'II 1 INTRODUCTION
The Central Area Transmission System (CATS) is a North Sea pipeline
I development scheduled for operation in April 1993. The pipeline is
400 kilometers long and 36 inches in diameter capable of
transporting up to 1.4 billion standard cubic feet of gas per day.
I The gas is delivered to Teesside, England to fuel a new combined
cycle heat and power station being constructed by Teesside Power
Limited. Amoco-operated gas fields (Everest and Lomond) will flow
I 300 million SCFD which is only 21 per cent of the total pipeline
capacity.
I This excess capacity along with the connectors that have been
built into the pipeline will allow other gas fields to tie
" into the line and act as a common transportation system for
I any newly developed gas fields.
Amoco is committed to accurate natural gas measurement
a-I
and operating procedures consistent with the Cullen Report,-
Amoco CATS project group has designed each offshore gas
metering system with three (3), single chambered type orifice
fittings sized so that any single meter may be serviced during
any expected flowing conditions. This design philosophy
incorporates an 18 inch header with 10 inch branch connections
(1.7 to 1 reduction in diameter) for each meter run.
I
I
configuration in this section leaving the Licensee with the
task of trying to interpret the standard, subject to
II
government bodies approval, when installing any header
arrangement. section 6.3 of the Standards recommends that
particular types of flow straightening devices can be used to
I
permit the installation of a flow measuring device downstream
of fittings not listed above. However it also specifies a
minimum overall length of 42 diameters shall be used for all
straightening devices unless the conditions stated in Section
I
6.4 are met. This generally discourages the use of
condi tioners as it invariably requires longer (rather than
shorter as expected) upstream run lengths or an expensive and
I
time consuming test to demonstrate compliance with
Section 6.4. I
After considerable research into recent developments in the
field of flow straighteners and conditioners for orifice
meters, Amoco proposed an optimum installation design for the
upstream meter tube length and straightener location as shown
I
in the meter layout in Fig. 1 (29 diameters (D) overall
upstream with the straightening device at 100 from the closest
disturbance) . The metering des ign has an upstream
configuration consisting of a 12-inch vertical (down) inlet
ew
into a IS-inch horizontal header with three 10-inch meter
tubes off branch type connections.
constructed
Each meter tube is
with two matched bore valves (100 total), a
J
flanged type flow straightening or conditioning device, a
straight section of 17-19 pipe diameters (depending on the
straightening device length) and a flange neck single
I.
chambered orifice fitting.
I
3 TEST PIPE ARRANGEMENT
I
I
.. were machined to best fit the pipe. The companion orifice
flange unions were machined and dowelled for near perfect
I alignment regardless of the arrangement of the conditioners or
pitot. Spacers (12 mm thick by 10.060 10) with standard '0-
ring' seals were placed between all flanges not holding the
I pitot or conditioner so that there was a smooth transition,
free of gaps or intrusions, throughout the entire pipe
section.
t' measured the test pipe and pitot carrier ring internal
diameters and the relative roughness of the pipe bore.
I
I
Flow Conditioners. The following four flow straighteners or
conditioners (all designed to be held between 10-inch 6001
..
ANSI Raised Face flanges) were tested.
a Conventional 19 tube 20 long Short Tube Bundle (Fig. 4).
I
b Zanker (10 long) constructed to ISO 5167 requirements
(Fig. 5).
I
c Laws Conditioning Plate (University of Salford) (Fig. 6).
A perforated plate flow conditioner with an open area or
porosity of about 51.5 per cent.
I
d K-Lab Mark 5 Conditioning Plate (Confidential). I
I,
pitot Tube. The single traverse assembly unit (Fig. 7),
designed by Gasunie for the EEC orifice discharge coefficient
tests, incorporated a constant blockage swirl angle and impact
res sure probe. pitot side pressure sensors were located 400
from the centre impact pressure sensor. The device was
.,
installed between modified orifice flange unions designed to
enable it to be centred in the pipe, without any gap, step or
offset from the pipe wall, and perpendicular to the flow.
I
Before each new installation the pitot ring was centred on the
outer diameter of the companion flanges using adjustable
matched 'tee blocks', such that no internal offset could be
detected visually or by touch.
I
4 NEL TESTS
I
The main objective of the tests was to determine the optimum
location of various flow conditioning and straightening I
devices within the upstream orifice meter tube and to verify
that the velocity profile and swirl components of the
installation were within the specified limits set out in ISO
5167, Paragraph 6.4. In order to establish the amount of tJ
swirl generated by the header configuration and the
effectiveness of the conditioners, flow profiles were measured
at 00 and 190 without any flow conditioner installed. After
,I
the profile tests were completed the discharge coefficient of
a nominal 0.6 diameter ratio orifice plate was measured with
the Laws and Tube bundle conditioners in the test line.
Selection of the various test configurations, by AmOCO, were
I
based on data from the conditioner plates designers, published
papers and Amoco's own research and design expertise. I
The face of the flange located 10D from the header to branch
connection was chosen as the datum from which the positions of
the flow conditioners were measured. Since the overall length
I
of the various flow conditioners varied from 0.120 to 2D the
distance between the conditioner and pitot was measured from
the face of the conditioner flange. The majority of the tests
were conducted with a flow conditioner (with its flange
upstream) installed at the datum pipe flange; exceptions to
this were the K-Lab device at 60 and the tube bundle at 130 as
I
I
.. mentioned below. For each test, the pitot device was moved to
a position representing a possible orifice plate location.
I pitot traverses were conducted in the vertical and horizontal
planes ~ with negative radius ratios corresponding to the
bottom or left-hand pipe wall (looking downstream)
I respectively. The swirl angle was measured by rotating the
pitot tube until the differential pressure, across the side
pressure sensors, was zero~ the angular rotation of the tube
I representing the swirl angle. The impact pressure was then
measured at that angle with the central pitot orifice since it
represented the peak impact pressure. Any offset (bias) of
I For the profile tests the mass flow through the test rig was
held constant during each traverse. The first flow
conditioner to be tested was the tube bundle and the flowrate
t' was set to give the maximum possible Reynolds number~ the K-
Lab and Zanker devices, having greater pressure losses, had to
be tested at lower flowrates, the Laws device was tested at
the same flowrate as the K-Lab. During the orifice plate
I tests the flow was varied b~tween the minimum and maximum rate
attainable.
I
I
.-
I
where V is pitot velocity, op and Pl are the pitot
differential and inlet pressures and y is the isentropic
component for air. I
The velocity ratio, VR, (point velocity, Ve, to centreline
velocity, VeL) was rationalised with respect to the volume
flow associated with the centreline velocity, ie I
I !
I
.,
No Flow Conditioner. Figs 8 and 9 shows that the velocity
profiles at both positions were significantly inverted.
maximum swirl angle at the 00 position was found to be 24
The I
degrees with 20 degrees of swirl remaining at the 190
position. The initial test (OD) demonstrated
conditions at the flow conditioner inlet.
flowing
The second test I
(190) approximates flowing conditions for a common North Sea
installation without a flow conditioner, that is, in general
accord with ISO 5167 design criteria of 30 diameters
downstream of a 2 to 1 header to branch connection.
I
Zanker Flow Conditioner. The velocity profile from the Zanker
was examined only at the 190 position (Fig. 10). Although it
t
produced a reasonable velocity profile it did not remove
enough of the swirl, a total of approximately five degree
remained in both the horizontal and vertical planes.
profile produced was a bit flat and somewhat asymmetric.
The
tI
K Lab FlOW Conditioner. The K-Lab flow conditioner was tested
with the pitot at 60 and 190 from the datum pipe flange (Figs
I
11 and 12). At 60 the device was installed in reverse with
it's datum face downstream allowing a full 60 between the end
of the device and the traverse plane. The measured profile
I
was flat compared to the theoretical profile and slightly more
than one degree of swirl remained. At 190, the remaining
swirl was similar but the flow profile was nearer the
theoretical prediction but slightly asymmetric in the vertical
I
plane such that the end points were outwith the five per cent
limit. I
B. Laws Flow Conditioner. The flow profile was examined at
three positions: 6, 9, and 19 diameters downstream of the
conditioner (Figs 13 to 15). Since the length of this device I
--I
was so short it was not reversed as was the K-Lab conditioner.
The profile at 60 was rather flat and slightly asymmetric with
I
.. a maximum swirl angle of about one degree. The asymmetry
caused the profile to be greater than five per cent above the
t' degree.
At 9D the flow profile deviated significantly from the
theoretical profile and the outer annular portion exceeded the
I mid 25 per cent to produce an inverted or collapsed profile.
Ie edge of the pipe began to increase but did not exceed two
degrees.
Pressure prop. The pressure drop for each device, recorded
I during the test, is shown in Fig. 20. The tube bundle
displayed the lowest loss of less than one velocity head and
the Laws was next with less than two.
I Coefficient of Discharge (Cd) Tests. Orifice discharge
coefficient tests, using a 0.597 beta ratio orifice plate in
I two positions, were conducted with the Laws and tube bundle
conditioners. The plate was manufactured to ISO 5167
specifications, and the edge sharpness and internal diameter
measurements were checked by the Metrology Section of NEL.
I Each test consisted of a Cd at five flow rat~s; the results,
compared with the NEL standard Cd equation ( ), are shown in
Figs 21 to 24.
The Laws conditioner was tested with the orifice plate 9 and
19 .diameters downstream (19D and 29D overall length). As
expected, the test Cd results, Figs 21 and 22, are nearer the
I
I
standard at the higher flow rates and larger Reynolds numbers.
The results at 90 are slightly better than those at 190.
.-
The tube bundle was tested with the plate in the 130 and 130-
Reversed positions downstream (230 and 240 overall). The 130
I
position, Fig. 23, produced the Cd results most near the NEL
standard (within 0.25 per cent). Overall, the 130 tube bundle
Cd was closer to the NEL prediction than the Laws at I90. The
I
Laws at 90 and tube bundle at 130 produced similar results at
the larger Reynolds numbers.
I
6 BIGH PRESSURE NATURAL GAS TEST
Note: For these tests, the pitot device was fixed at the 290
fla':lge
10hcation 5.0that the ~ength upstream of the conditioner
I
d ev~ce c anged (~ncreased) ~nstead of the overall length.
..
I
conditioners were more flat than the fully developed
theoretical flow profile. .
"I
I
When installing orifice meters downstream of headers, the
predicted orifice coefficients may be used with greater
confidence if a flow conditioner is installed at the proper
.-
location. I
The installation requirements for flow conditioners set
forth in ISO 5167 Section 6.3.1 exceed the actual
requirements for custody transfer meters with a diameter
ratio maximum of 0.6 when such orifice meters are
I
downstream of common headers with branch connections.
"I
I
I
I
I
II
I
I
.. APPENDIX
where U is the pipe maximum axial velocity, R the pipe radius and
u is the axial velocity at a point where the radial distance is
I r. This gives a good fit to experimental data, but there is no
accurate theoretical way of determining n. In Nikuradse's data
n ranged from 6.0 where the pipe Reynolds ~umber, Reo was 4000,
I to 10.0 where Reo was 2.0 x 10 or 3.2 x 10
The best method of determining n is to calculate it by fitting
data whose Reynolds number is similar to that in the installation
I being tested for acceptability. Data collected in air at NEL(5)
with Reo = 9 x 105 , pipe diameter = 102 mm and 1400 of straight
pipe upstream were available: fitting these data using a least-
I squares fit gave n a 9.9.
..
I
One problem with the velocity profile in equation (1) is that it
does not have a zero derivative on the pipe axis. To solve this
problem the following was tried:
a
1
(1 - ~) Ii , I
->c
(2)
u = R
I
U
1-b (~r' I
-5:C
R
I
where a and b are chosen so that the equation both is continuous
I and has a continuous derivative at r/R = c. This equation has
a zero derivative on the pipe axis and has a very similar
behaviour to equation (1) in the neighbourhood of the wall.
I However, when the NEL air data in Ref. 5 were fitted there was
almost no improvement in quality of fit from that obtained with
equation (1); moreover n was almost unchanged.
"I
I
Data, collected by British Gas, with a least 1000 of upstream
pipe in 259 mrn pipe at Reo = 1.4 x 107 and in 600 mm pipe at Reo
= 2.2 x 10 , were included in reports to the EEC,6,7 , but only
..
in graphical form. The data, in tabular form, for the 600 mm
pipe were obtained from British Gas.
I
Three sets of data were agailable, on ~wo planes at ReD = 2.2 x
10 , and on one plane at Reo = 8 x 10. Only the data at the
higher Reynolds number have been analysed, since those at the
I
lower have a maximum velocity two per cent higher than the
centre-line velocity. The exponent n in equation (1) was
obtained by using a least-squares fit to the data on each plane:
I
on the 45 degree plane n = 9.7; on the 30 degree plane n = 10.1.
This supports the use of the power law profile in equation (1)
with n = 9.9 as a good representation of what the velocity I
profile would be after a long length of pipe at the Reynolds
numbers encountered in both the air and gas tests. The 600 mm
data have been plotted in Fig. 25 for comparison with equation I
(1) with n = 9.9 and it can be seen that there is good agreement.
Gasunie have also collected data downstream of 800 of 600 mm
pipe'S) (including a full-bore ball valve SOD v-pstream of the
measuring point) for Reo from 2.5 x 106 to 5 x 10 and found that
ea
the profile can be described quite well with a power law and that
n appeared to be around 10. The value of n is a little larger
for higher Reo than for lower.
I
From the data analysed the power low profile in equation (1) with
n = 9.9 gives a good representation of what the velocity profile
I
would be after a very long straight length of pipe at the
Reynolds numbers encountered in both the air and the gas tests.
I
I
II
I
I
I
I
I
--I
I
.. REFERENCES
Ie
I
I
I
I
I
"I
I
LIST OF FIGURES II
1 Proposed Metering Station I
2 Drawing of Test pipe Arrangement
3 Test Pipe I
4 Tube Bundle
10
Profile at 190 - No Flow Straightener
12
K-Lab - Profile at 60
17
Tube Bundle - Profile at 9D
19
Tube Bundle - Profile at 13DR
..,...
OQ
.....
...,
B,fYATIIII
---
"nlll_ or
..,...oo ....__.._-_
- ....--.-
1.~-
.
.. .....
.--,------
.:;":=-.::=-~-
."
"":x ,, -----------
I::='-:""
."
rt
......
."
_----
--
:I
OQ --
-
-------.
1--- .
In
rt
...
~
-
o
:I
.......
---..,-
~"
.......
0.,.
".,.,.,.......
..............
,..
~,..~...".
....".,."",
..
............
....-.
" .....
l_
FRCFOStD ARA"~MfHf
_1M EWfRUf
_ lIAS !EfEIIINII.
- ---- ---- ----- -
J
a
...""
OQ
'" ~, ~.
..
1:1 :. ~
..;
'"...::s
c:
_._1..._ __.-
..--
OQ e... ELEYAT/ON
.... _._ 01 ._._.
...
0
........
; anr
_,_
__ n
_' __ 011 __
"1._."'
'-OOL'
_ ...._
_
'".....
II
.,.-....
....
." "'
:p .
,
.
o.n
..,...
"II
II
.~
....
:.-
OQ'"
::s
L='---~ L. ---~~IP.-~"!
. .... _-
-
II t -..,!,--. ',.,
9
II .- !
"'
::s or-
r+ ~.
~ ~ .....IC- 1
ptAN
~"'0I'1fff""""
J; .11""'''''
It 'J
~I SI_r,,,
*'UnIlIi
_ ........ 0 "
" 1$ IUIIf" DOl' C'OIWDI"3N ,,,., "
IS .vol IIfNtmUCCC :0..,.., 011
DlroOSID ID .. '1fI1ttJ ,. .. r
.RRANGEHENT I DET"L
OF TEST RIG SPOOLS
(IF,.,_ """""
, ........ l1tro\
l1li '"
C',."UIf1
'VI' .".".111M"
(I' ,,,. AHOCO CATS PROJECT.
-- ---
I
..
I
I
I
I
I
t'
I
I
I
I
..I
I
I
Fig. 3 Test Pipe
I
I
"
I
I
--I
I
I r----- ------
..
- --
I I
I
t'
I
I
I
I
..I
I
I
Fig. 4 Tube Bundle
I
I
"
I
I
-.
I
I
I ;--' .
I
I
t'
I
I
I
I
..I
I
I <'
"
I
-- -
I
..
I
I
I
I ;-
-.
I
it
I
I
I
I
Ie
I ,
I I
.1
I
il Fig. 6 Law Conditioner
-
I
It
I
I
..
I
I
I
I
I
t' ,o~~
~."'~'/_-
I
I
. /i' '9".
,
0' t" _~
.... ,
I
I
..I
I
I
I
F1g.
. 7 Pitat Traverse Unit
I
II
I
I
-.
I
Horizontal Plane
I
I.~
0 0
0 0 0
0 0 0
1.2
I 0
...., 1.0 r
0 0
.....
I co
c, o.e
.. .. .. .. -
30
20
~
>-
.....
....,
.. QJ
I U
0
.....
0.6 - 10 .....
CJ1
C
--I
0 co
QJ o.~
.....
>
.. -10 c,
....,
0.2 - .. .. .. .. -20 (f)
~
I
0.0 -30
-1 -0.5 o 0.5 1
I Radius ratio
I I.~
0 0
0
,
Vertical Plane
0
0
0
_0 0
I 1.2
0
0
..I
0
...., 1.0
..... - 30
co
c, o.e -a .. .. .. - 20
>-
.....
....,
U
0.6
.. - 10
QJ
.....
CJ1
0 C
..... 0 co
I >
QJ o.~
- .....
.. -10 c,
....,
I 0.2
.. .. .. ..- -20 (f)
~
0.0 -30
I -1 -0.5
Radius
o
ratio
0.5 1
I
I
-.
I
1.4
Horizontal Plane
I 1.2 ~
I 0
...., 1.0 f.o
0 0 0
0 0
0 0 0
0
...... 30
I m
L
>-
0.8
A
A
A
20
......
....,
A
Q)
I U
0
......
0.6 - - 10 ......
OJ
c
Q)
0 m
fI
0.4 I-
> ......
A
A
- -10 L
....,
A
0.2 I- A
- -20 ~
rJl
I 0.0
-I -0.5
I
o
-30
0.5 1
I Radius ratio
I 1.4
Vertical Plane
I
1.2
I 0 0 0 0 0 0
..I
0 0 0
0
...., 1.0
~
......
m
- 30
L 0.8 I-
- 20
>-
......
A
A
.-1
U
0.6
- 10
CIJ
,....,
OJ
0 c
,...., 0 m
I >
Q) 0.4 I-
A
- -10
......
L
0.2 A
~
..-1
I
A
~ -20 rJl
0.0
,
-30
o
I -I -0.5 0.5 1
Radius ratio
I
I
..
I 1.2
Horizontal Plane
I 1.0 _ - --0
--- --- - -
--- - - ---
I 0
......
..... O.B
~
<I
ttl /
I L
>-
..... 0.6
I
.. 2
0
......
.. .. Q)
.....
I U
0
..... 0.<1
.. 0
rn
C
ttl
>
Q)
.. .. .....
fI 0.2 .. -2
L
......
3:
.. Ul
I 0.0
-I -0.5 0 0.5 1
-4
Radius ratio
I
Vertical Plane
I
1.2
..I
,,
..... O.B 4
ttl ,
, '0
..
0
L
..
.. ..
>- 0.6 2
.....
..... Q)
.....
U
0
.....
.. Cl
C
0.4
.. 0 ttl
.. .. .. .....
Q)
I >
0.2 -2
L
.....
3:
I .. Ul
0.0 -<I
-I -D.5 0 0.5 1
I Radius ratio
I
I
..
I Horizontal Plane
1.2r---------,---~~~~~~~~_,--------_.
I --- --- ---
1.0
I --- _ _ _
-----
-O-_Q,
... ...t:>
,,
o
..... ,
..., O.B I
,, ' 4
I to
L I
I
o
...,
>-
.....
0.6 2
QJ
.....
I u
o
..... r---------------~----_+----~~--------------4 o
OJ
C
to
QJ
0.4
fI >
0.2 -2
.....
L
......
~
(fJ
I o . 0 L-
-1
-l...
-0.5 o
L ......L.
0.5
---.I -4
1
I Radius ratio
Vertical Plane
I 1.2
r----------r-,--~~~~~~~~--._--------_,
I
..
o
.....
..., 4
to
L o
...,
>- 0.6 - 2
I .....
u
o
QJ
.....
OJ
C
..... 0.4 r---------------~----_+----~.~--------------~
o to
I >
QJ
,& ,6 ..
- -2
.....
L
......
0.2~ ~
I 0.0 L- L-, ....L .l....- ...J -4
(f)
o
I
-1 -0.5 0.5 1
Radius ratio
I
I
Ie
I 1.2
Horizontal Plane
I 1.0
--- --- - -
--- ---
I 0
.,...,
..... O.B 4
ttl
I
0
c,
>- 2
.. .. .. ..
0.6
.....
.,..,
QJ
......
I U
0
...... .. .. 0
en
t=
co
.. .. .. ..
0.4
>
QJ
......
fI
c,
-2
.,...,
0.2 ~
CJ)
I 0.0
-\ -0.5 0 0.5 \
-4
Radius ratio
I Vertical Plane
1.2
I --- ---
\.0
I ~
--- --- -- --
0
.....,...,co
..
/
~ , ,\ 4
O.B /;,0
, \
0
c, I
0 2
>- 0.6
.. .. .. ..
.....
.,...,
QJ
......
I u
0
en
t=
......
QJ
0.4
.. .. .. .. 0 co
......
I >
0.2
-2
c,
.,...,
CJ)
~
I 0.0
-\ -0.5 0 0.5 \
-4
I Radius ratio
I
I
..
I 1.2
Horizontal Plane
I ,
'..0
0
....ro
M
O.B , ,\
\
'I
L
I ....>-
..
0.6 2
OJ
M .....
u
Cl
I 0
.....
OJ
0.4
0
C
ro
.....
>
fI
L
M
0.2 -2
~
(f)
I 0.0
-\ -0.5 0 0.5 \
-'I
Radius ratio
I Vertical Plane
1.2
I 1.0
--- ---
I 0
--- --- ---
....ro
M
,,
..I
O.B I 'I
I /
/
L
....>-
M
0.6 2
OJ
......
U
Cl
C
0
..... 0.4 0 ro
>
OJ
.....
I 0.2 -2
L
M
~
(f)
I 0.0
-I -0.5 0 0.5 1
-'I
Radius ratio
I
Fig 13 Laws Profile at 60
I
II
I
/
(
I
-.
I
I 1.2
Horizontal Plane
--- ---
I a
1.0
--- --- --
....,
M
I co
C-
O.B 4
....,>- 0.6 2
I M
U
a
OJ
......
OJ
c
...... 0 co
fI
0.4
>
OJ ......
C-
M
0.2 -2
~
I 0.0 -4
Ul
-\ -0.5 0 0.5 \
I Radius ratio
I 1.2
Vertical Plane
co
C-
>-
....,
.-<
O.B
0.6
d
4
2
0
OJ
......
U .. OJ
c
I >
a
......
OJ
0.4 .. .. .. .. 0 co
......
C-
M
I 0.2 -2
~
Ul
-4
I 0.0
-\ -0.5
Radius
0
ratio
0.5 1
---
I .....
0
...,
co
O.B 4
c,
~
I ...,
>-
.....
0.6 2
QJ
.....
u 01
I 0 C
..... 0.4 0 co
>
OJ
.....
.....c,
"
0.2 -2
~
[J)
I 0.0 -4
-\ -0.5 0 0.5 \
Radius ratio
I 1.2
Vertical Plane
I 0
..... O.B I
/
/
,, 4
,
..I
/
-.J I /
co I
c, I
>- 0.6 2
.....
.....
QJ
.....
U 01
0 C
..... 0.4 0 co
>
OJ .....
I 0.2 -2
.....c,
~
[J)
I 0.0
-\ -0.5 0 0.5 1
-4
Radius ratio
I
Fig 17 Tube bundle Profile at 130
I
II
I
I
Ie
I Horizontal Plane
1.2
I 1.0
--- --- -0- -
I 0
--- --- --- ---0-_
(t,,
...,
'M
O.B
/
, ,\ 4
~
I
<0 / \
L 0
/
...,
>0- 0.6 2
Q)
......
I
'M
U Cl
0 C
...... 0.4 0 ro
Q)
fI >
0.2 -2
......
L
M
3
Ul
I 0.0
-I -0.5 0 0.5 1
-4
ratio
I Radius
Vertical Plane
I 1.2
..I ...,
'M
O.B 4
co 0
L
...,>- 0.6 2
Q)
......
'M
U Cl
0 c
...... 0.4 0 ro
I >
Q)
0.2 -2
......
'M
L
I
UJ
0.0 -4
-1 -0.5 0 0.5 1
I Radius ratio
II
I
I
..
I \,2
Horizontal Plane
u 01
I >
0
.....
Q)
0.4
0
c:
ro
.....
c,
ft 0.2 -2
......
X
(J)
I 0.0
-I -0.5
Radius
0
ratio
0.5 1
-4
I \,2
Vertical Plane
I 0 ,6. ,
--- --- --- -- -a-
...... . O.B I , 4
..I
+' I ,
ro
c, I
I
,
>- 0.6 2
Q)
+'
...... .....
U 01
0 c:
..... 0.4 0 ro
>
Q) .....
I 0.2 -2
c,
......
X
(J)
I 0.0
-I -0.5 0 0.5 1
-4
Radius ratio
I
Fig 19 Tube bundle Profile at 190
I
II
I
- - - - - --- - - - - ,- - - - -,. -
7.------.------,-------.-------.------,-------.-------r------~
Law..s
lJ.
6 GI
B
B 0 Zanker
0 D
0 K-Lab
0 Tube bundle
5
a
......
~
rn
L
In
In
a
.-i
3
D
m
Q)
I 2
o ~ ~ ~ _L J_ ~ __ ~--L-----~------~
o.~ o.~ o.~ o.~ O.M O.M O.M 0.00 1.00
o test Cd
NEL Cd
+J
C
OJ
......
u 0.610
......
4-
4-
OJ
o
U
OJ o
o o
01
L o
ro
.r: 0.605 o
U
tn
......
o
0.600~------------~=_------------~--------------L-------------~
0.20 0.30 0.40 0.50 0.60
0 test Cd
-- NEl Cd
.4-'
C
Q)
.-1
U 0.610
..-1
.....
.....
Q)
0
a 0
0
U 0
Q)
OJ
L
co
.r: 0.605
U
1Il
.-1
0
0 test Cd
NEl Cd
~
C
OJ
rl
U 0.610
rl
....
....
OJ
0 0
U
OJ
OJ
L 0
ro
1:: 0.605
U
Ul
rl
0
(J test Cd
NEl Cd
~
C
Q)
......
u 0.610
--
......
Q)
0
(J (J
(J
(J
(J
U
Q)
01
L
CO
.c 0.605
u
Ul
......
0
Re = 2.2 x 107
D
D 30 deg plane
0 ~5 deg plane
EQuation (I) n 9.9
1.0
~OO
1;10
0
.....
+J
ro
L
()
>- 0.9
+J
.....
U o
0
......
QJ
>
O.B
0.7~------------~ ~ L- -J
-I -0.5 o 0.5 1
Radius ratio
Fig 25 Velocity profile - British Gas 600mm pipe
I
II Velocity Traverse (British Gas) - Laws
I U
I'oIiIioa 190 - Hori2lOlllaiPlue(Re 10z10A6)
,
I 1.2
I
6
I u
o.a
2
~
011
~ =
I 0.6
0
~
0.4
I Q.2
-2
"I 0 -4
-1.5 -I -0.5 0 0.5 I 1.5
RadiUl ratio
__ Trav __ '-(0-9) ........ Swirl
I'
I Velocity Traverse (British Gas)
POIition 190 - Vertical Plane (Re JOxlO ....6)
- Laws
1.4 s
I 1.2
6
'-I u
~
0.8
0.6
4
2
~
1
VI
0
I 0.4
0
I 0.2 -2
0 -4
I -1.5 -I
__ Trav
-0.5 0
Radiu ratio
-.-Swirl -e-
0.5
thear (01:19.9)
I 1.5
I '" -
It Figure 26
I
- .. - - - - -". - - - - -a - - - - - -
0.606 TEST
~
MEAN
<>
'C
0.605
I ISO
U
-
....~=
ill
....
,
CJ
i::
c;..;
U
~
Q
0.604
...til
~
-=
CJ
....Q
<II
II
0.603
}O.O83%
0.602 L-_L------1_---i._----L._---L..._--1-_---L-_--'--_-'--_.l--_'------'_---'-_---"
MEAN
~
-e
U
....
...=
Q,j
CJ
0.605 f-
,
~
ISO
....
t:: <>
Q,j
Q r-
0.604
U
Q,j
...~
~
t-
I
...-=
Q
CJ
G'l
0.603
}0.083%
t . I I I I I I I I
0.602
0.40 0.50 0.60 0.70 0.80 0.90 1.00 1.10
6
Pipe Reynolds No.1 (t0 )
I
by
I
I
,. B C Millington and T S Whitaker
NEL
I
I Paper 2.1
I
I
..I IDRTH SPA FIJ::M MEASUREMENl' mRKSHOP
26-29 October 1992
I
I
I NEL, East Kilbride, Glasgow
I
II
I
I
Ie
I MULTIPHASE FLOWMETER MEASUREMENT UNCERTAINTIES
I
I SUMMARY
I The paper describes the state of the art in multiphase flow metering from a theoretical point
of view, giving brief details of how the individual mass flowrates of each phase are derived.
tt
The limitations imposed by using such calculation procedures are then fully explained in
terms of metering uncertainties.
By referring to previous theoretical work, the level of uncertainty in the individual mass
I flowrate measurements for a typical multiphase flowmeter are shown to be in excess of 10
per cent for virtually all possible combinations of oil, water and gas phase fractions. Most
importantly though, it is shown that the individual mass flowrate uncertainties are very
I dependent upon the flow composition. It is therefore concluded that for any multiphase
flowmeter, uncertainty data must be qualified with a statement of the flow composition to
which it applies.
I
NOTATION
I
..
I
mg
mo
mw
mass jlowrate of gas
I {1
p.
liquid water content
liquid fraction of the multiphase mixture
I
II
I
I
I
eI
gas density at the metering conditions
I
oil density at the metering conditions
Most oil companies are therefore vigorously pursuing the development of multiphase
I
flowmeters which would be capable of replacing the test separators, and which can also be
located subsea. At the present time the authors are aware of 18 separate multiphase
flowmeter development projects worldwide.
I
In the haste to achieve an operational prototype multiphase meter, some of the fundamental
measurement difficulties have often been overlooked. It is the purpose of this paper to rectify
this situation, giving a balanced assessment of the likely measurement uncertainties involved
fI
in this type of metering, and identifying the major problems. I
This information will be of interest to both the developer and user alike, and it is hoped it
will provide a basis for realistic performance aims for first generation multiphase flowmeters. I
2 MULTIPHASE FLOWMETER DESIGNS I
Ideally a multiphase flowmeter would make 3 separate, but simultaneous measurements: one
of the oil flowrate, one of the water flowrate, and one of the gas flowrate. Each of the
separate measurements not being influenced by the fact that the other two phases are present.
I
This type of meter is not yet technically possible, and it is unlikely that such a meter will be
available within the next 10 years. I
II
2
I
I
..
I However, instrumentation is available which is capable of discriminating between two flow
components; water-in-oil monitors are a good example. By combining a number of indirect
measurements it is possible to determine the flow composition of 3-phase flows, and coupling
I this information with a measurement of the total volumetric flowrate it is then possible to
determine the individual mass flowrates of each phase. All of the present multiphase meters
under development use this same basic technique, although the measurements made vary
I from meter-to-meter.
An example of a typical system configuration is shown in fig. I. In this design the three key
I measurements are the water content in the liquid after sampling and removing any entrained
gas; the total oil/water/gas density; and the total flowrate. The procedure used to calculate
the oil, water and gas phase fractions and flowrates is given below. (For simplicity it is
I
,. assumed that pressure and temperature variations are accounted for)
Consider the flow to be gas-liquid with a liquid fraction of p., then the total density can be
written:
I (1)
Similarly, if the water content in the liquid is denoted by (3 then the expression for the liquid
I density is
I (2)
I Substituting the expression" for the liquid density into equation (1) gives an alternative
expression for the total density
..
I
(3)
where the coefficients of the oil, water and gas densities are the volumetric phase fractions,
therefore
I ,
a = 1 -p.
I aD = p.(l-{3)
aw = p.{3
I
I So if p. and {3 can be measured or calculated then the flow composition can be derived.
II
3
I
I
In the system shown in fig. I assume that the water content in the liquid is found by
.-
measuring the liquid density, then from equation (2) I
(:3= (4) I
where the oil and water densities are known from the well characteristics. I
Likewise from equation (1) the liquid fraction /l. can be found by re-arranging to give
I
(5)
I
where the liquid density and the total density are measured, and the gas density is known
from the well characteristics.
Having calculated the phase fractions using the values of /l. and {:3, the individual mass
flowrates of each phase can be found by obtaining a measurement of the total volumetric
--I
flowrate
I
I
I
There are many technical difficulties in achieving a working system of this type, but from
a theoretical point of view the above analysis is the basis common to all of the multiphase
flow meters under present development. J
3 MULTIPHASE FLOWMETER MEASUREMENT UNCERTAINTIES
I
3. I The key problems I
The technique outlined above is a perfect! y acceptable method of calculating the individual
flowrates of oil, water and gas, but there are fundamental limitations on the measurement
uncertainties achievable.
I
Unlike a single phase flow in which any measurements made relate directly to the liquid or
gas present, the measurements taken to define the flow composition of a two-phase or
I
mul tiphase flow are indirect, and relate to 2 or more phases. If the phase fraction of the
required flow component is low, then in relative terms the measurement uncertainties in the I
instruments used can have a major effect on the uncertainty in the mass flowrate calculation.
II
4
I
I
..
I To demonstrate this effect a simple analogy using distances to represent phase fractions can
be used. Consider two points A and C with a third point B lying on the line between A and
C, but located quite close to C. The distances AB and BC can therefore be thought of as the
I phase fractions of two flow components in a two-phase mixture: one with a high phase
fraction, the other with a relatively low phase fraction. The short distance BC can be
measured by two ways, either a single direct measurement, or by measuring AC and AB and
I subtracting AB from AC.
I (6)
I (7)
I Substituting for the absolute uncertainties in AC and AB, this equation reduces to
I (8)
I To give an indication of the potential difference in measurement uncertainty between the two
methods, set c = 100 mm with a = 95 mm and b = 5 mm. Then the absolute uncertainty
I so the uncertainty in the deduced value of BC would be 19 times greater than if a single
direct measurement were made. In percentage terms the indirect method of measuring BC
I would have a random uncertainty approaching 20 per cent, compared to the I per cent
associated with the direct measurement.
I The above analogy highlights quite clearly the limitations of taking multiple indirect
measurements to determine flow composition. The only way of overcoming these difficulties
is to have extremely low uncertainties on all such measurements, but because of the
I multiphase flow environment this is very difficult to achieve.
II There are two principal difficulties: the natural structure of the flow can vary considerably,
and measurements involving two of the flow components can be hampered by contamination
5
I
I
by the third phase. Liquid water content monitors are an excellent example of this because
eI
the measurement technologies presently used tend to give erroneous readings if there is gas
in the flow. I
However, it is the spatial and temporal variations in the flow structure which are technically
most challenging. For example, a common technique used in multiphase flowmeters is
gamma densitometry, in which the absorption of gamma rays gives a measure of the total
I
flow density. If the flow is well mixed then a reasonable average density measurement will
be obtained quite quickly, but if the flow is slugging it becomes much more difficult to
define an average density let alone measure it. In such circumstances the longer the averaging
I
period the better, but there is obviously a practical limitation if real time readings are
required. A balance has to be struck, which is inevitably to the detriment of the uncertainty I
in the total density measurement.
Finally, and perhaps most importantly, any instrument within a metering system that requires I
an input from another instrument in order to interpret its own measurement is clearly at risk
of having an increased uncertainty. An example of this is described in the next section.
~
3.2 An assessment of the uncertainties involved
Consider the multiphase metering system shown in fig.l, with the three key measurements I
being the water content in the liquid, the total flow density, and the total volumetric flowrate.
These measurements being made by a vibrating tube densitometer, a gamma densitometer and
a venturi, respectively.
I
To calculate the liquid water content, phase fractions and individual mass flowrates the
I
following must be known:
I The first thing to be noted is that the interpretation of the gamma densitometer measurements
requires knowledge of the liquid composition in order to assign the mass absorption
coefficients correctly. Likewise the venturi requires the total flow density measurement from
I the gamma densitometer to correctly convert the differential pressure measurement into a
measurement of volumetric flow. So if the water content measurement has a high random
uncertainty associated with it, then this will add additional uncertainty to the total density
I measurement and this in tum will be transferred through to the volumetric flowrate
measurement.
it This interdependence makes a rigorous derivation of the uncertainties in the oil, water and
gas mass flowrates lengthy, and consequently unsuitable for inclusion in this paper. The full
derivation can however be found in reference I.
I Using the equations derived in reference 1 a computer program was written to perform the
calculation of the uncertainties in the individual mass flowrates over a wide range of oil,
I water and gas phase fractions. Table 1 gives an example of the data produced, and from this
it is evident that the percentage uncertainty in the individual mass flowrates of oil varies
I considerably according to flow composition, and can be extremely high. (Note that this data
was produced using best estimates of the measurement uncertainties in each parameter listed
above, and these values are shown in Table 2)
I Tables showing the uncertainties for the water and gas mass flowrates are available in
reference 1, and they show similar trends to that of Table 1. Summarising all the uncertainty
There is a lot of information available from the full uncertainty analysis, particularly
I regarding the sensitivities of each measurement uncertainty. However, there are two points
of major importance:
II 7
I
I
b All things being equal, the flow composition has the greatest affect on the percentage
..
uncertainty in the individual mass flowrates. For credibility the uncertainty figures for any
multiphase flowmeter must be quoted at a specific flow composition. Ideally an upper and
I
lower limit (as above) should be given.
I
4 CONCLUDING COMMENTS
The main points arising from the study of multiphase flowmeter uncertainties are:
I
4. I Present designs are far from ideal, because they do not measure the individual mass
flowrates directly. Inferring the mass flowrates from a collection of less direct measurements
I
gives scope for large uncertainties to develop.
4.2 In such systems the interpretation of signals from some instruments often requires
I
precise measurements from other instruments. An uncertainty in one measurement can
therefore have considerable knock-on effects.
4.3 A full uncertainty analysis of a typical multiphase metering system has shown the
minimum uncertainty that can reasonably be expected is 10 per cent, and this occurs when
--I
the phase fraction approaches unity (ie nearly single phase).
4.4 As the phase fraction reduces, the percentage uncertainty in the individual mass flows I
increases and can be over 1()()per cent as the phase fraction approaches zero.
4.5 In the light of 4.3 and 4.4 it is essential that multiphase flowmeter uncertainty data I
is quoted along with the phase fraction at which it applies.
4.6 The complexity of specifying multiphase flowmeter uncertainties will inevitably lead I
to confusion, and it is therefore of paramount importance that a standardised method of
quoting the uncertainties is developed in the near future which is easy for manufacturers to
understand and comply with, yet conveys the maximum amount of information to the user.
NEL are actively pursuing suitable techniques at present.
J
I
ACKNOWLEDGEMENT
This paper is published by permission of the Chief Executive, NEL Executive Agency, and
I
is crown copyright.
I
REFERENCE
I
it
I
I
I
I
..
I
I
I
I
I
II 9
I
I
..
I
TABLE I
I
THE PERCENTAGE UNCERTAINTY IN THE OIL MASS FLOWRATE
MEASUREMENTS AS A FUNCTION OF FLOW COMPOSmON
I
(The water cut is the percentage water in the liquid) I
I
5
10 20
Gas fraction in the mu1tipha flow, (')
30 40 50 60 70 80
-------------------------------------------------------------
13.77
13.91
14.03
14.17
14.40 14.96 15.85
14.53 15.07 15.93
17.37 20.29 26.98 49.43
17.41 20.26
90
26.80 48.91
--I
10
15 14.15 14.39 14.73 15.25 16.0e 17.52 20.29 26.69 48.42
:10
25
14.47 14.70
14.90 15.1:1
15.03 15.53 16.33
15.43 15.91 16.67
17.71 20.39 26.63 47.98
17.99 20.58 26.64 47.59
18.39 20.87 26.74 47 .26
I
30 15.46 15.66 15.96 16.41 17.13
Water
35
40
45
16.16 16.35
17.04 17.21
18.13 18.30
16.63 17.05 17.73
17.47 17.86 18.50
18.53 18.90 19.48
18.92 21.29 26.95 47.01
19.62 21. 86 27.29 46.84
20.53 22.63 27.80 46.79
I
19.87 20.20 20.74 21.71 23.66 28.55 46.90
Cut
(tI)
SO
55
60
19.51 19.66
21.25 21.38
23.49 :13.61
21.58 21.88 22.36
23.78 24.05 24.48
23.24 25.04 29.60 47.22
25.27 26.89 31.10 47.86
I
65 26.44 26.54 26.69 26.92 27.30 27.99 29.44 33.24 48.99
70
75
80
30.44
36.13
44.77
30.53
36.20
44.83
30.65
36.31
44.91
30.85 31.18
36.47 36.74
45.04 45.25
31.77 33.02 36.39 50.91
37.23 38.29 41.16 54.17
45.65 46.49 48.84 60.00 J
85
90
95
59.32
88.63
59.36
88.66
177.04 177.05
59.42 59.52 59.68
88.70 88.76 88.87
177.07 177.11 177 .16
59.97 60.60 62.38 71.28
89.06 89.48
177.25 177.46
90.67 96.8e
178.05 181.22
I
I
I
I
10
I
--I
I
..
I
I
I
I TABLE 2
t'
Path length of gamma ray across the pipe 2.0
I Gamma densitometer calibration count rate 0.5
Ie
Liquid sample representivity 1.0
I
I
II
11
I
I
..
I
I
gas return I
separator
I
I
oil/water/gas sample
oil/water return
water
--I
monitor
I
---- ---- I
volumetric flow rate
mixture density
I
Fig.1 Schematic Diagram of a Multiphase Metering System J
I
I
I
I
I
II
I
I
..
I
I
A MULTI -cAPPCI'lDR MULTIPHASE Fl.O'IMETER FOR SUG:;:rn:; F'Ikl
I
I by
I
t' D Brown, J J den Boer and G Washington
Shell Research, The Netherlands
I
I
I Paper 2.2
I
Ie
I NJRTH SEA F'Ikl MEASUREMENl' \'l)RKSHOP
26-29 October 1992
I
I
I
NEL, East Kilbride, Glasgow
I
It
I
I
.. The North Sea Flow Measurement Workshop, Peebles, Scotland, October 1992
J.J. den
I D. Brown, Boer and G. Washington
Shell Research, b.v., Rijswijk, The Netherlands
I
SUMMARY
I A multi phase flowmeter suitable for use in individual flowlines has been developed and
tested in a full-scale multiphase flow loop. The instrument, which operates in the slug flow
I regime, uses an array of capacitive sensors in the flowline to measure directly holdups and
characteristic velocities. It is called, therefore, the multi-capacitor multi phase "flowmeter or
tt MCF. The measured parameters are used to derive the flowrates of the individual phases
with an accuracy over a wide range of flow conditions of better than + / - 10% of reading.
The present instrument is suitable for oil-external emulsions, with watercuts up to around
40%. The MCF system comprises a simple spool piece with certified local electronics, a
I remote signal conditioning unit and a control unit to calculate and display the results. It is
unique both in its simplicity and the use of the slug flow characteristics. It will provide the
operator with reliable flowrate data at a relatively low cost.
I After a brief introduction to the operating principle, the results of laboratory tests with this
new instrument are presented. The flowmeter is being commercialised by Kongsberg
I Offshore, a.s., Norway. The further development programme and planning resulting from
this cooperation will also be briefly described. This is aimed at producing by 1994 an
instrument with the functionality of a test separator at a fraction of the cost, making its
..I 1
1.1
INTRODUCTION
Flowmeter requirements
The need for a compact, reliable multiphase flowmeter is well established. Such a meter will
enable the cost effective development of both offshore and onshore satellite fields and the
possibility of optimising the operation of existing fields. Moreover, the availability of such a
--I - 1-
I
The North Sea Flow Measurement Workshop, Peebles, Scotland, October 1992
eI
1.2 Alternative measurement approaches
Many approaches to multi phase flow measurement are being investigated by the oil I
industry. These can be grouped into three categories:
Separation: separate
Homogenisation:
the phases and measure them individually before recombination.
create a homogeneous mixture and measure total flow and the fractions of
I
oil, water and gas to give the flowrates for the individual phases.
Leave as it is: change nothing to the flow and use intrinsic flow properties to derive
I
flowrates.
The MCF belongs to the last category. Separation operates best at high gas volume fractions
where the phases are already naturally separated in stratified or annular flow. Mixing is
I
easiest at low gas volume fractions where the natural flow regime is close to bubble or plug.
The MCF covers a large area between these two, operating in the intermittent flow regimes I
most often found or easily created in real well flowlines. No flow conditioning other than
sufficient horizontal upstream pipe length is required.
Fig.1 shows a multiphase flow map for horizontal flow with the various flow regimes
III
superimposed, and also lines of constant gas volume fraction at 50,90 and 99%. The
superficial velocity is the actual volumetric flowrate of one phase divided by the total cross-
sectional area of the pipe. The gas volume fraction (GVF) is the ratio of actual gas flowrate to
I
total flowrate.
I
2 OPERATING PRINCIPLE
2.1 Liquid and gas transport in slug flow I
To understand how gas and liquid flow rates can be measured with the MCF, an
understanding is needed ofthe way in which gas and liquid are transported in slug flow. A
good description is still the simple, well accepted model first introduced by Duckier and
I
Hubbard [1] . They describe slug flow in two parts, the slower- moving liquid film which
partly filis the pipe, and the faster-moving liquid slugs which totally fill the pipe but only
for a fraction of the time. As the slug moves down the pipe it picks up liquid at its front and
sheds it again at its rear. The slug itself is made up of both liquid and gas because gas is
II
trapped in the liquid at the front of the slug as it overtakes the slower-moving film. The
model states that the actual gas velocity is almost constant and equal to the slug front
I
velocity. Fig. 2 shows shows measured holdup as a function of time for various flow
conditions, clearly demonstrating
gas and liquid flowrates.
the dependence of both holdup and slug frequency on the I
The MCF gauges the liquid and gas flowrates in intermittent flow by measuring the cross-
sectional areas occupied by each phase and multiplying each area by the velocity of that I
phase. To do this, it continuously measures the cross-section of the pipe occupied by liquid,
the velocity of the liquid and the velocity of the slugs. Assuming that the slugs travel at the
same velocity as the gas, the flowrates are calculated from the product of the cross-sectional I
areas occupied by the liquid and gas and their respective velocities.
2.2 Measurementtechnique I
The measuremen ts are all made with a pair of plates inserted into the pipe in line with the
II
-2- I
I
.. The North Sea Flow Measurement Workshop, Peebles, Scotland, October 1992
flow. Etched onto these plates are a number of electrodes forming capacitors and it is the
I impedance variations in these capacitors that are used to produce the basic signals from
which the flowrates are calculated. The plates have a width of just 4 ern along the pipe and
typically contain one complete column of electrodes plus two further electrodes (Fig. 3).
I One signal is obtained from each capacitor and this is scaled such that the zero voltage level
represents the case when there is only gas between the capacitor electrodes and one volt
represents the case when there is only dry oil between the electrodes. Each capacitor is
I actually connected to an electronic circuit that produces two output signals representing the
real and imaginary parts of the admi ttance, one proportional to the conductance one
proportional to the susceptance. Of these two signals, only the susceptance signal is used
I and this is scaled to give the zero and one voltage levels for gas and oil. The actual signal
range is far higher than 0-1 volt as mixtures of oil and water have far higher susceptances
than oil alone.
I The way in which the signals are processed is described in su bsequent sections.
I signal for gas only and that for liquid only. However, the signal representing liquid only is a
priori unknown because it depends on the actual ratio of oil and water in the liquid. Finding
the correct value to represent liquid only is crucial to the operation of the MCF and is, often,
..I
The simplest way of determining the signal representing liquid only, known as the span
factor (SP), is to take the signal measured by a electrode when it ,is immersed in a
homogeneous and gas free mixture of the oil and water. In the simple slug model presented
previously, this situation occurs in the liquid film sometime between slugs. The slug, when
it passes, creates a homogeneous mixture of the liquid but it also mixes in some gas.
Sometime before the next slug this gas has time to separate out of the film, leaving only the
homogeneous liquid mixture. The point at which the gas has disappeared is easy to
recognise because this is the point at which the measured susceptance is a maximum. In its
I simplest form SP is, therefore, taken as the maximum signal over a defined measuring
period of a few slugs.
I electrodes some distance apart at the same height near the top of the pipe. These signals are
non-zero only during the passage of a slug. The signals are fed to an electronic circuit that
produces a number of measurements per slug from the variations seen during the slug
I passage. These measurements are then averaged. It takes between 10 and 50 slugs to make a
good measurement of the slug velocity depending on the flow conditions.
II
I -3-
The North Sea Flow Measurement Workshop, Peebles, Scotland, October 1992
I
Z.2.3 Liquid velocity
eI
The liquid velocity is measured continuously from the signals
same height near the bottom of the pipe. These signals are fed
from the two electrodes at the
to a cross-correlator to
I
produce a continuous measurement ofthe time delay between the two analogue signals and
hence, since the distance between the electrodes is known, the
2.2.4 Flowrates
velocity of the liquid.
I
The flowrates are calculated from the average liquid holdup and the average liquid and gas
velocities. The liquid flowrate is simply the product of the holdup and liquid velocity, the
I
gas flowrate is the product of the pipe cross-section filled with gas, i.e. (1 -liquid holdup),
and the slug velocity.
Z.2.S Watercut
I
The watercut is calculated from the span factor, SP, which is a measure of the dielectric
constant of the liquid. The dielectric constant of an oil/water mixture is a unique function of
I
the watercut while the oil remains the external phase, up to around 40'70 watercut.
3
3.1
METER DEVELOPMENT
Laboratory prototype
--I
After an initial short feasability study had clearly demonstrated that gas and liquid flow
rates could be deri ved from slug parameters, a 4" prototype meter was constucted in 1989
and Installed in the multi phase test loop at Shell Research, Rijswijk. This was used to further I
develop the sensor geometry and the signal processing to a stage where the target accuracy
of +1-1D% per phase could be achieved over a wide range of flow conditions. This
performance is discussed in more detail in the following section. The next step clearly had to I
be a field trial, but this required a commercial prototype certified for use in hazardous areas.
3.Z Commercial prototype I
A joint development project was agreed between Kongsberg Offshore, a.s. (KOS) of Norway
and Shell Research, co-sponsored by Norske Shell, leading to a number of 3" and 4"
commercial prototype meters by the middle of 1992. These meters became available on
schedule and have also been tested in the multiphase flow test loop. The MCF system
II
comprises a simple spool piece, with certified local electronics, a remote signal conditioning
unit and a control unit to calculate and display the results (Figs. 4 and 5). It is unique in its
I
simplicity and will provide the operator with reliable flowrate data at a potentially relatively
low cost. A field test is arranged for the period September-October at a suitable location in
Oman. Results of this trial, if available, will be presented, but at the time of writing the
I
equipment is still being installed.
I
4 PERFORMANCE
4.1 Test conditions I
Testing was performed in the laboratory multiphase test facility, which can simulate flow
behaviour in both 3" and 4" flowlines over a wide range of multiphase conditions using
gasoil, water and air as the component fluids. The flowrates used covered as large a range of
I
II
-4 -
I
I
.. The North Sea Flow Measurement Workshop, Peebles, Scotland, October 1992
liquid and gas superficial velocities as could be achieved with the nominally 100 mm
I diameter laboratory prototype and the 3" commercial prototype selected for the most
extensive testing. Gas superficial velocities ranged from 1 to 12 ml s and liquid superficial
velocities from 0.1 m/s to 3 m/s. These values reproduce the flow conditions in the majority
I of actual flowlines, and also give the approximate boundaries to the slug flow regime.
The meters were mounted in horizontal flowlines, with an upstream straight length of
I approximately 15 m for the 4" meter and of approximately 6 m for the 3" meter. This was
sufficient to develop intermittent flow, thereby enabling the meters to operate, but no
attempt has been made to determine the minimum straight length requirements. The
I technique does not depend on the slug flow being fully developed.
The sensor plates are constructed using the same production techniques and material as
multilayer printed circuit boards. In parallel with the performance tests a separate sand
I resistance test was performed on a sensor plate. Using high levels of sand to accelerate
testing, the plate was subjected to the equivalent of more than three years of operation at
t' sand levels of 5 gil (50 ppm). The conclusion was that sand erosion is not a problem for the
MCF probe. The MCF probe material will have a service lifetime of at least 5 years at sand
concentration levels of around 50 ppm, which is typical for those wells producing from
unconcolidated sand formations which are especially prone to generate higher levels of
I sand.
4.2 Laboratory prototype
I Figs. 6 and 7 show the areas on a flow map where the liquid and gas measurement errors
respectively are below 10%. The liquid flow errors are shown for a 10% watercut emulsion,
the gas flow errors for air and dry oil (0% watercut), Liquid velocity measurement becomes
I difficult at flows less than around 0.2 mls since the slug frequency is low, and between
slugs the film is almost standing still. Gas measurements are limited in the upper part of the
flow map by the transition to bubble flow, and in the low liquid/high gas corner by the lack
I of slugs, essential for obtaining gas velocity. Nevertheless, for both liquid and gas a large
range of conditions are contained within the + 1- 10% flow error borders.
I Repeatability of the measurement depends on the time constant used since, by its very
nature, multi phase flow contains a degree of randomness which must be averaged out. With
a time constant of 100 seconds, short by multi phase standards, the measurements were
I repeatable to better than + / - 5%.
II
I -5-
I
--I
The North Sea Flow Measurement Workshop, Peebles, Scotland, October 1992
/"
/"
/"
z-
u
I 0
c;;
>
"0 1.0 /"
5
I g-
o;
u
't
/" Slug
Q)
I 0-
:::J
en
0.1
it 1.0
Superficial gas velocity (m/s)
10 100
I
Mandhane flowpattern map for horizontal multi phase flow
I
I "Reconstructicrvrecoqnltion of multi phase flow patterns"
I Gas
(m/s)
liQuid
(m!!L
..I Top
Bottom
Top
1.0 2.1
2.6 1.5
Bottom
I
TOp
I Bonom
1.3 0."
I Top
Siratiliediwavy flOw
7.5 0.3
I Bottom
I Dr.no.101060 FIGS.1,2
I
eI
I
~
p
....~o
)r1'r1'
I
i'Jil
l\l2
i'Jil
~
I
~
~
I
~
~
I
--I
Fig. 3 MCF sensor plates
I
I
f Id electronicS
support for ,e
I
II
I
I
3- Of 4"'
Position of sensor
plates in the line
I
diameter
I
Fig.4 MCF mechanical design I
II
Dr.no.l01051
FIGS. 3, 4 I
I
..
I
I
I MCF
,MCF
Field I!!!!!!!!!!!!!!'!\'.!!!!!!!'!!!!!!!!!!!!!] ; Sig n a I . tm::::i
I Unit
Conditioning
.....Unit
I
t' MCFFU .!IJ!I!!!!!!!l!!!!!!!!IJ!I!!!!!!!!!!!!!!!jj .
'<,:;:,,' .:'
MCFdcu ':I!!!!!!!
I MCF
Control
I . Unit
I MCFFU
I
..
I MCFFU MCFSCU
.,-
.- "
, <
I
Complete four-well metering system
I
I
II
I oj D(.00.101070 FIG.5
I
--I
5 I
~
I
3
iz-
'e::;
0
a;
>
'@
'(3
1
I
'E .5
Q)
Cl.
::>
'" .3 I
--I
"t:l
::>
.6-
::::;
.1
2 3 5 10
Gas superficial velocity (rn/s)
~
5
I
.!!? 3
.s
z-
5
Q)
0
J
co
0
't
>
.5
I
Cl)
0-
::>
'"::>
-0
.0-
.3
I
::J
.1
1 2 3 5 10
Stratified
flow
I
Gas superficial velocity (mls)
Gas error,0% wate rcut I
Fig. 7 Gas Flowrate errors for KSEPL 4" prototype
I
II
.. Dr.nO.101071 FIGS. 6,7 I
I
-. Liquid 0%
I 3.0- .....................
,.... .-.;...i.-.
28 : 23 20: -14
.30
1
I ,8
.t
i. ,"
~ 35
I ::-
'u
.2
0>
1.0-
+3
.....~
,
i
:
I
-
> .~7
,
,,
I 0;
'u
:;:
0>
...............J~;~L
.... H1~~
i.
i -6
.-- ..~
9
~ .
,
_
~0.5
I : .
,.
:g
,
:J
.Q"
-' l .17'I
:.
. '
v
+4
'.+21
,'.,6
, ..
i+26
0.2 ................................... , . ______
1...
r+2
, ,
.,
,
.3 [19
:.-
_,_ ....
;.......
.:
..I ,.
+7 ,, 3 t-a "~
,,, :.
.,
.
+51 :
:
j+20 ,,
.4 4
~
~1
-!4 .1
1.0 .~.
.3'
..1 ;....
.2
......
-~
r
-s 1".
:
i -2
. .,
I 0.5 .
:.
.28
i~ .
,
.4
.
...__ __ ._-_. __ :
.
__ ................. ; --.-.
:: ,,
--j- -
:. ,'j
;
i .+28:
I
.3 : +5 , :,
.. -26
;'38
. , I
4
1-
l+5 "
I
II
I II 0r.no.101072 FIGS. 8, 9
I
..
I
I
A FUM-REX:;IME INDEPENDENT MULTIPHASE FI.GIRATE METER
I
I by
I
E Dykesteen and 0 Midttveit
t' Christian Michelsen Research
I
I
I Paper 2.3
I
Ie
I OORTH SEA FUM MEASUREMENI' IDRKSHOP
26-29 ~ 1992
I
I
I
NEL, East Kilbride, Glasgow
I
II
I
I
.. A FLOWREGIME INDEPENDENT MULTIPHASE FLOWRATE
I METER
I
E. Dykesteen and 0. Midttveit
I Christian Michelsen Research. Bergen. Norway.
I
I
t' SUMMARY:
I In a paper presented at the North Sea Flow Measurement Workshop in 1990. Christian
'-
I Michelsen Research (CMR) presented test results for a new concept for measuring
fractions of a multi phase flow. In the follow-up of this work. a new project to develop a
the phase
multiphase flowrate meter was initiated in June 1990. This has now resulted in a commercially
I available multi phase meter. the Fluenta MPFM 1900.
I The new flowrate meter is of a non-intrusive design, and mounted in a vertical upwards flow, it
I The paper discusses why installation of mixers for preconditioning of the multiphase flow
should be avoided, and it presents test results from extensive laboratory testing of the new
I multiphase flowrate meter. The first field test installation is already in progress.
I
I
II
I
2 I
THE FLUENTA MPFM 1900;
A NEW MUL TIPHASE FLOWRATE METER DEVELOPED BY CHRISTIAN
MICHELSEN RESEARCH
--I
In a paper presented at the North Sea Flow Measurement Workshop in 1990 [l), Christian
Michelsen Research (CMR) presented test results for a new concept for measuring the phase I
fractions of a well mixed multiphase flow. The measurement principle is shown in Figure I:
Based on the density of a well mixed multiphase flow as measured by a clamp-on gamma
densitometer, and the permittivity of the same flow as measured by a non-intrusive in-line
I
capacitance sensor, the fractions of oil, gas and water at the sensor location are determined.
This principle of measurement has been realised in the Fluenta MPFM 900 Multiphase Fraction
I
meter.
I
--I
I
I
\ MULTIPHASE
I I
"I
COMPOSITIONS
__ J
Figure I Measurement principle of the MPFM 900.
I
With support from BP Norway, Saga Petroleum, Elf Petroleum Norge and the Norwegian
Petroleum Directorate, we have now also developed a velocity measurement technique based on
I
cross correlation of signals from a multi-electrode capacitance sensor. Assuming that there is no
mter-phasia! velocity slip between oil, water and gas, this velocity measurement can be I
combined with the MPFM 900 in order to provide a multiphase flowrate meter suitable for no-
slip conditions. I
II
I
I
.. However. tests early in this project indicated that interphasial slip between liquid and gas is
3
I significant in a vertical multiphase flow. and is re-established rapidly after a mixing point. At
best. no-slip will be established only over a limited range of flow rates and compositions.
I thereby limiting the range of any no-slip measurement system.
I Therefore. the electrode layout and detector system of the capacitance sensor was optimised in
order to be able to discriminate between the velocity of small bubbles dispersed in the liquid.
I and the velocity of fast moving larger bubbles/slugs. By using this additional information. in
combination with knowledge of multiphase flow behaviour. the model based measurement
system directly determines. and accounts for. interphasial slip. The new flowrate measurement
I
,. system does therefore not require any upstream mixer to be installed.
The performance of the complete measurement system was put to a thorough test in the CMR
multiphase flow-loop during March '92. The results from these performance tests are very
I satisfactory. Over most of the tested range. measured flowrates of liquid and gas are found to
be well within 5% from the reference flowrate.
I As part of the test programme. two "production profiles" during approximately 3 hours were
produced through the multiphase flowrate meter, and accumulated volumes of oil. gas and
I water were measured. During such a test all random errors due to natural fluctuations in the
flow are filtered out and become insignificant. and the effects of any systematic errors are
I clearly shown. In Figure 2 it is shown how the liquid flow rate during one such test period has
..
been varied between 35 and 45 m3/h. while gas flow rate has been varied between 10 and 35
m3/h. The water cut has been varied in the range 15-22 %. The results of this "production test".
which could be compared to a test-separator run, are shown in Figure 3 and Figure 4.
I
I
I
I
I
II
I
I
4.-
45
I
I I
40
35
I I
I
30
25
I I
Flow ral9S (m3'h)
20
15
I
10 --
-
Ga,s flow ral9
o
1 101 201 301 401 501 IiOl
Measuremant no.
100
J
I
Aa::.. liquili
lIOIumo (m3)
IiO - Measured QJrI1Jlalive
liquid voIurre (rrC)
I
I
40 -- Reference QJrnJlative
liquid volurre (m3)
20
60 120 180
I
TIrre (rrinutes)
I
Figure 3 Measured accumulated liquid volume during a production test.
II
I
I
-. 5
I
80
I 70
I 60
50
I Aa:.gas
voIuma(rrG)40
- Measured OJmJlaliw
30
,. 20
10
60
--
120
Reference OJrrula!ive
gas volume (rrG)
180
I Tima (rrinutes)
I The discrepancies between measured and reference volumes of liquid and gas in the two
I production tests were found to be between 2% and 5.5%. During the period of these tests the
corresponding water-cuts were measured to be 14.7% and 19.7%, as compared to reference
Based on these very encouraging test results, F1uenta has already made an industrial version of
the meter available to the market. This has been made possible through the co-operation with
Conoeo and the Norwegian Research council (NTNF) within the KAPOF - programme.
I In the new instrument, analogue signals are converted to digital at the sensor head, and all
I signal transmission between the sensor and the control room unit is in digital form on a fibre-
optic link. Thereby a low power, but high speed signal transmission has been achieved. The
I sensor head has also been fitted with a pressure compensation system. With this design the
maximum pressure is no longer limited to the ceramic liner specifications. The measurement
system has been approved by Baseefa for use in hazardous areas. Also the user interface has
I been improved from the MPFM 900 version.
II The new meter, The MPFM 1900 Multiphase Flowrate meter, replaces the MPFM 900 fraction
meter on the market. Saga Petroleum has purchased the very first unit, and following a
I
6 I
successful factory acceptance test in the CMR laboratory in August '92, Saga is now
(September '92) in the process of installing the meter at Gullfaks B for offshore field testing.
eI
The results of the performance testing of this new flowrate meter, prior to the delivery to Saga, I
are presented laterin this paper.
I
NON-INTRUSIVE MEASUREMENT OF THE FLOWRATES OF OIL, WATER
AND GAS IN A NON-CONDITIONED MUL TWHASE FLOW IS A COMPLEX
I
MEASUREMENT PROBLEM
I
A variety of different flow patterns must be expected to occur in a non-conditioned multiphase
flow. By restricting installation of the multiphase meter to vertical upwards flow, the possible I
flow patterns are fewer, usually classified into four flow regimes exhibiting significantly
different features (see Figure 5): II
In bubble flow the gas is uniformly distributed as small gas bubbles dispersed in a continuous
liquid phase. It generally occurs when the gas flowrate is low compared to the liquid f1owrate.
I
The slug flow regime develops from a bubbly pattern when the gas flow rate increases to such I
an extent that it forces the dispersed bubbles to become closely packed and coalesce into larger
gas volumes. Stable slug flow is generated at high gas rates in pipes of long free vertical I
stretches.
Chum flow is observed as a transition regime from bubble to slug flow, and may also be
I
experienced at very high gas flowrates in the transition region from slug to annular flow.
II
In annular flow the liquid phase moves upwards partly as a wavy liquid film along the pipe
wall, and partly in the form of small droplets entrained in the gas core. Annular flow occurs I
when the flowrate of gas is significantly higher than that of the liquid.
I
I
I
I
II
I
I
.. 7
I
I
. . (
I ... '. \
I
I
,.
I Bubble Slug Churn Annular
I The natural grouping of gas and liquid into specific flow regimes at different flow conditions,
I clearly means that one cannot assume the gas to be uniformly distributed in the liquid across the
cross-sectional area of the pipe. The degree of in-homogeneity will increase at high gas flow
Ie
rates. The gas is only uniformly distributed in the liquid in bubble flow. For all other regimes
the gas tends to collect itself in the middle of the pipe as large bubbles in slug and churn flow,
and even more dominant in the form of a gas core in annular flow.
I
In a gas-liquid flow, the larger bubbles will rise in the fast moving liquid in the centre of the
I tube, other small bubbles will be near the wall and will consequently move more slowly. This
velocity slip is further enhanced by buoyancy due to the difference in density of gas and liquid
I so that the gas phase will be transported with a larger average velocity than the liquid phase.
There are a range of different mixers available, both inline static devices which draw the mixing
I
energy from the flow itself, as well as dynamic mixers which require energy input. One
advantage of using some type of dynamic mixer is that little or no pressure drop is created
I
across the mixer. However, in most cases a static mixer will be preferred, because no energy
input is required, and because they generally are rugged devices with no moving parts. Pressure ttl
drop across a static mixer will typically be in the order of 0.5 to 3 bar, depending on flowrate
and composition. I
A multiphase flowrate meter which depend on a mixer to provide suitable measurement
conditions will lend itself to the efficiency of available mixers. The range of such a flowmeter
I
will be limited by the range where sufficient degree of mixing can be guaranteed. The
uncertainty with respect to remaining slip at the flowmeter location, and/or in homogeneity of
I
the radial phase distribution, will add a significant contribution to the overall measurement
uncertainty. I
To our knowledge a thorough analysis of the performance of mixers for use with multiphase
meters has never been performed. It is however evident that even the most efficient mixer will fI
not be able to create no-slip conditions and homogeneous phase distribution over an un-limited
range of flowrates and compositions. Since the overall performance of many multiphase meters
I
will directly depend on the performance of available mixers, a comprehensive test program for
evaluation of available mixers over a wide range of flowrates and compositions should therefore
I
be initiated. Some parameters important for evaluation of the mixer performance in such a test
programme will be bubble size distribution, velocity profile, interphasial slip, axial and radial I
phase distribution, and pressure drop.
I
Awaiting the resuIts of such a test programme, and, until now, not having identified a mixer
with suitable performance parameters, we at CMR have chosen to accept the existence of slip,
and to develop methods to measure the degree ofinterphasial slip, rather than try to avoid it by
I
mixing.
II
I
I
.. In addition, we believe that the industry will see some advantages with a completely non-
9
I intrusive system with no pressure drop. In particular, some multiphase pipelines even require
pressure boosting in order to transport the flow from a remote satellite to a processing platform.
I Since increasing the line pressure just a few bar using a multiphase pump will be quite
expensive, it does not appear very cost efficient to consume pressure head just for the sake of
creating suitable measurement conditions for a multiphase rneter.:
I
The major drawbacks by having to rely on a multiphase mixer can therefore be summarised as
I follows:
I
,. it is not evident that there exist a multiphase mixer which will be able 10 create no-slip
conditions and homogeneous phase distribution over a sufficiently large range of
flowrates and compositions
the range of the multiphase meter will be limited to the range where a mixer is able to
I provide acceptable flow conditions
I any remaining slip, or in-homogeneous phase distribution, will directly contribute to the
overall measurement uncertainty
I
MEASUREMENT OF MULTIPHASE FLOWRATES UNDER PHASE SLIP
Ie CONDITIONS
I regime and bubble sizes. While small bubbles dispersed in the liquid phase will be transported
with close to no slip, larger bubbles or gas slugs will take on a slip velocity in the range from
I 0.5 to 2 rn/s relative to the liquid velocity. Therefore, in principle, the total velocity range must
be measured, and correct volume fractions of liquid and gas must be allocated to each pan of the
velocity range. In order to do this correctly, we have found it necessary to try to understand the
I behaviour of multiphase flow, and to build such knowledge into signal interpretation models in
the model-based measurement system.
I
.. Schematically the new multi phase flowrate meter is illustrated in Figure 6. Its operation will be
described in the following.
I
I
10
--I
I
I
I
~ P~
r---------------~
~::I
FLOWRATE
CALCULATIONS
I
Figure 6 Measurement
.
principle
. of the CMR multi phase flowrate meter.
I
I
Measurement of phase fractions
As in the MPFM 900, the multiphase
measurement
fraction routine is based on a non-intrusive capacitance
to determine the mean permittivity (dielectric constant) of the mixture and a gamma
I
meter measurement to determine the mean density. From these two independent measurements,
and knowing that the sum of all the fractions will always be equal to one, the individual fractions
of oil, gas and water at the sensor location are detenni ned. Laboratory tests have shown that this
tI
measurement system is capable of determining the composition of a multiphase flow to an I
uncertainty of better than 3% for each of the components [1]. The tests have been carried out in
a vertical upwards bubble/chum
uncertainty is experienced
flow in a gas fraction range from zero up to about 60%. A larger
at higher gas fractions where the impact from the in-homogeneous
I
phase distribution becomes more significant as previously explained.
I
Measurement of phase velocities
I
The phase velocities are determined using cross-correlation of signals provided by different
sequentially located surface plate capacitance electrodes. The gas phase in a multiphase flow I
generally comprises several velocity components.
II
I
I
..
11
Different sensor geometries or electrode lay-outs have shown to be sensitive to different bubble
sizes of the flow due to different spatial averaging of the variations detected. It is also possible
I to use different types of detectors being sensitive to different physical properties of the mixture.
In addition, further pre-conditioning of the signals may be carried out as they are recorded as
discrete time series. This further tuning of each specific technique includes determination of
I optimal values of parameters connected to the correlation computation such as logging
frequency, number of samples in time series, filtering schemes and averaging methods.
I
During the project correlation methods have been developed which determine two significant
I flow velocities of the flow. These are the velocity of the fast moving larger gas bubbles/slugs,
and the velocity of the small bubbles dispersed in the liquid (with velocity very close to the
I liquid velocity). Slip between water and oil is generally assumed very low for crude oils in a
vertical upward section, and is therefore neglected in the models.
I of the three phases. Since the gas phase moves with different velocities, the fraction routine has
been further developed so that in addition to the average gas fraction, the fraction of large gas
I bubbles, as well as the fraction of small gas bubbles dispersed in the liquid, are measured on-
line [2].
I The signal interpretation models use these fractions, and the two significant flow velocities, to
calculate the volumetric flow rates under phase-slip conditions. The interpretation based model
Ie thus determines the total gas flowrate as the sum of gas transported as large bubbles/slugs
(QGB), and gas transported as small bubbles uniformly distributed in the liquid phase (QGo):
I .0_,
I The mutual
.' .
order of magnitude'. of QGO and .QGB will vary depending on the actual
' :.
flow regime
appearing in the flow.
I
In bubble flow the gas is mainly transported as small dispersed gas bubbles. In such a condition
I the main contribution to the gas flowrate is from QGO'
I In churn flow there are larger bubbles moving faster than the surrounding highly aerated liquid
mixture. In such flow conditions there is also a volumetric contribution, QGB, from the faster
II
I
I
moving chum bubbles along with the flowrate provided by small dispersed gas bubbles in the
liquid. QGD'
12
..
In slug flow. an even larger portion of the gas will be carried by the large bubbles compared to
I
the condition of chum flow.
I
RESULTS FROM LABORATORY TESTS OF THE MPFM 1900 I
During August 1992. prior to delivery to Saga Petroleum. the performance
1!X)() Multiphase Flowrate Meter was put to a thorough test in the CMR multiphase flow-loop
of the new MPFM I
using diesel oil. The test programme included a wide range of flowrates and compositions.
I
Some volumetric flowrate results of the performance
Each point in the plots is the average of 10 independent measurements.
the vertical test section of the CMR loop was bubble and chum flow.
tests are shown in Figure 7 and Figure 8.
The flow condition in
--I
65
60
s-. I
55
.>. ';I.
V
50
. /
..
..
':" .
I
.V
/li
. V I
~ Gos~ 20'4
7-
J
!--
I /.
. ;,0 o Gc:n .. 3O'Y.
/. - Gos- 4O'l. r-
15 o Gos .. 50'10 r-
10
7 Gos .. 6O'l.
!-- I
/' I
5
-: i I
o
o 5 10 15 20 25 30 35
Reference liquid ncwrete (m3/h)
50 55 I
Figure 7 Liquid flowrate test results (m3/h). Nominal gas fractions: 2060%. Water cut I
20%. The dotted lines represent the 5% relative error.
I
I
..
I
I
.. 70
'/,
13
I 65
,0
.>:
60
55
.>:
I .>: ' '
50
45
?
J
, '
. ~ "t.
I 40
!I~35
/'
/-
, '
i'g3O
,
Gos-2O"'L f----
I a 25
~ 20
!Y."
~ o Gas .. 30'1.
Gas .. 40'10
-
-
15 /" , o Gas .. sen;.
-'--
I 10
5
/"
~.
GO$" 60'1. -
/
t' o
o 5 10 15 20 25 30 35
Reference single phose gas ftowrate (m3/h)
45 50 55 60 65 70
I Figure 8 Gas flowrate test results (m3/h). Nominal gas fractions: 20-60%. Water cut
20%. The dotted lines represent the 5% relative error.
I
I 25.0
20.0
I
I 15.0
10.0 I
'..~-~:-
...~-L .~ I .~
I
Ie 5.0
0.0 &
.g
~~il
-
_
,,~ '<\. \l
... Ill"':~
Id:.~,'8
__
~<'~-o,_ ~
,
If_,
.~,r.. "I
-
j
a
Hi
.'
~ ~
13 ,o
I ,5.0
3 .0 4 .0
'i'
50
:r- 0:]'1 ~!'lOO
60
.0 11 .0 12 .0
-10.0
I -15.0
-20,0
...
II
I I
I -25.0
~
I ~
I Figure 9 Deviations between measured and reference gas and liquid flowrates, relative to
actual total flowrate. Nominal gas fractions: 20-60%. Water cut 5 - 40%.
II
I
14
I
The results from these performance tests are very satisfactory. Generally. over most of the eI
tested range, measured flowrates of liquid and gas are found to be well within 10% from
reference single phase flowrates at the rig. Over a somewhat reduced range. the agreement I
between measured flowrates and reference flowrates is within S%. As we can see from the
flowrate deviation plot in Figure 9. the measurement uncertainty increases at low total
flowrates. Measured watercut is also within S% over most of the tested range.
I
The deviations in measured flowrates at low total flowrates are explained by the fact that the I
fraction measurement. which is used as input to the flowrate routine. still assumes a
homogenous mixture. Work is in progress to model in-homogenous phase distributions into the I
fraction measurement module. This way. simply an updated version of the system software. is
expected to improve tile system performance. I
A RANGE OF NEW MULTIPHASE METERS ARE NOW BECOMING
AVAILABLE TO THE MARKET. A TEST AND QUALIFICATION
PROGRAMME IS NEEDED.
--I
At the North Sea Flow Measurement Workshop in 1991 a discussion group compiled a list of I
1g different organisations currently working to develop a multiphase flowrate meter. At the
Offshore Northern Seas in August 1992. five out of these eighteen (including CMR/Fluenta)
announced meters available for the market.
I
The application of such. multiphase meters can drastically change a Satellite development
I
concept. Separate testlines and manifold systems at the satellite may be omitted. The same may
be the case for test separators. inlet separators and complex single phase measurement systems
at the processing platform. Such drastic changes to proven technology can of course not be
II
accepted unless the reliability of the new technology has been thoroughly qualified. In order to I
progress the maturity of such technology towards proven and accepted technology. it is
therefore now important that the oil companies involve in test and pilot installations of I
multiphase meters.
The step from testing in a friendly laboratory environment on a model oil, to tough field
I
applications on live crude is huge. In order to qualify this new technology for a wide application
in future field developments, it is therefore now of the utmost importance to gain field I
experience on non-critical installations. It may be a costly process both to install the meters and
to maintain a qualification test programme. however. the potential future cost savings should I
make such an investment profitable.
II
I
I
.. Field scenario studies suggest that subsea satellites is one of the most interesting applications
for multiphase meters in the future. The subsea challenge is addressed in several multiphase
15
I meter development projects, and the first subsea versions are likely to be available for field
testing at least by 1994. Provided that sufficient reliability of topsides meters has been verified,
I there will then exist a need for places where meters can be installed subsea for qualification.
t' REFERENCES
I [1) NSMW 1990, Field Experience with the eMI multi-phase fraction meter, by K.H.
Frantzen and E. Dykesteen
I [2] Model-based measurement ofmulti phase flow, Multiphase Transportation III, Present
Application & Future Trends, 20.-22. September, Reros, Norway.
I
I
Ie
I
I
I
I
I
II
I
I
I
FRANO HULTIPHASE FLOWl1ETER - PRO'T'OTYPE TEST
I
I by
I
A B Olsen and B-H Torkildsen
I Framo Engineering AS
t'
I Paper No 2.4
I
I
,
"
-. NORTH S!:~. FLOW nASURENEHT
26-29 October 1992
WORKSHOP
I
I NEL, East Kilbride, Glasgow
I
I
"I
,
I.
SUMMARY
A prototype test of the FRAMO Multiphase Row Meter has been performed under flow conditlons
typical for North Sea oil and gas production systems.
The prototype includes a combination of the FRAMO flow mixer, a multi-energy gamma meter
and a venturi meter built together wtth a barrier fluid arrangement and electronic processing
equipment in a vertical stack.
The results show that the multiphase flow meter can be used to measure volume fractions and
flow rates of oil, water and gas over the entire range of gas volume fractions 0 - 100% and water
cuts 0 - 100%.
Most of the test points measure liquid volumetric and total mass flow rates well wtthin +/- 10%
relative error.
CONTENTS
1.0 INTRODUCTION
2.0 MULTIPHASE METERING
4.1
4.2
4.3
Test rig arrangement
Test rig instrumentation
Test conditions
5.0 RESULTS
6.0 DISCUSSION
7.0 CONCLUSIONS
ACKNOWLEDGEMENTS
REFERENCES
1.0 INTRODUCTION
Multiphase metering has recently been subject to increasing interest due to "s potential to
enhance reservoir management and reduce capital expenditure and operating costs by
eliminating the need for test separators and dedicated test lines. These advantages are
particularly important for marginal fields and subsea satellite developments.
Several different methods can be used to distinguish between oil, waler and gas in a production
stream. One technique based on the attenuation of gamma rays at two energy levels was
developed by M"subishi Electric Corporation (MELCO) in 1984/85. A prototype gamma ray
compositional meter was built and extensively tested by MELCO, Petro-Canada Inc. (PCI) and
Alberta Research Council (ARC), including a field test in 1987 and a subsequent flow loop test of
a modified version in 1988, ref. 11/.
The meters capability of measuring volume fractions of oil, water and gas was demonstrated, and
the results were excellent for homogeneous flow. However, due to flow effects such as phase
slip and slugging, practical applications of the meter were lim"ed to conditlons with very low gas
contents.
A static flow mixer was developed and tested by Framo Engineering AS in 1988, as part of the
Subsea Multiphase Booster Station (SMUBS) programme for boosting of unprocessed well
stream. It was demonstrated that the flow mixer significantly improves booster performance at
multiphase conditions, and particularly at slugging conditions, use of the flow mixer is mandatory.
During the last five years this mixer has been extensively tested, and its performance has been
demonstrated in flow loops and during several field tests. Today, the mixer is an integrated part
of several multiphase booster concepts, including a version of the SMUBS which will be installed
for subsea duty at the Draugen field in 1993, and the POSEIDON pump which will be installed on
the Gullfaks B platform the same year.
In 1990 a joint industry project was established to qualify the combination of the FRAMO mixer,
the MELCO compositional meter and a standard venturi meter for measuring oil, water and gas
flow rates under flow conditions typical for North Sea oil and gas productions systems.
A successful test was performed in a high-rate oil, water and gas test facility. Tests with and
without the mixer showed that the mixer was able to eliminate upstream flow effects well enough
to obtain accurate and repeatable estimates of oil, water and gas flow rates. The test results was
Encouraged by the results we started the development of a subsea version of the tested metering
system. A functional prototype for dry testing was designed and manufactured with some of the
subsea parameters and equipment configuration aspects included. The subsea elements are
based on the available SMUBS technology. The subsea flow meter was presented in ref. 131.
This paper presents the test results from the prototype test of the FRAMO multiphase flow meter
performed at the FRAMO test facility at Fusa, Bergen this year.
Muttiphase flow metering wtthout first applying phase separation is difficutt due to the different
flow effects resulting from the interminable variety of phase distribution which can occur in such
flow.
The most important flow effects which can affect meter performance are:
These effects are closely connected to various flow regimes. However, since flow regime
description to some degree is arbitrary and the transitions between different flow regimes is a
gradual process, the impact on flow meter performance can not be sufficiently predicted.
The implications of this are twofolds. Primary sensor response will obviously be sensitive to the
flow effects, which not necessarily correlates with the quantities to be measured. Sensor
integration and interpretation are therefore difficult and critical.
However, more seriously is the lack of reproducability, and the need for in-line calibration at
various flow condhions, Since flow conditions will not only change from field to field, but also
during time for a given field, such calibration will neither be practical nor will it normally be
possible.
The flollY conditions will also depend on the actual physical location in the process relative to
risers, manifolds, bends, etc. Conveying the performance of a multiphase flow meter from one
system to another will therefore always be questionable.
Generally two approaches exist to these problems. One method is to measure the individual
phase fractions and related phase velocities. To establish the mass flow rates of the three
phases, oil, water and gas, five measurements are required; three velocities and two phase
Iractions (the third being reduced since the sum of the three phase fractions equals unity).
The 01her method, which is applied in the FRAMO multiphase flow meter, makes use of a flow
mixer and measures the individual phase fractions and the velocity of the mixture. In this case
Ihe required number of measurements are reduced from five to three.
The application of an effective flow mixer to eliminate or reduce the influence of flow effects, is
in this case essential in order to obtain accurate, repeatable and reproducable estimates of the
individual oil, water and gas flow rates.
Multiphase flow metering in general is further complicated by the fact that it is utmost difficult to
measure small components of a large system with high relative accuracy. This can be illustrated
by the production composed by 90% gas vofume fraction and 90% water cut. In this case the oil
volume fraction is only 1% to the total production. A typical uncertainty in the fraction
measurement for such conditions might be in the range 1-5%, resulting in 100-500% relative error
in the measured oil fraction, 11-56% relative error in the measured water fraction and 1.1-5.6%
relative error in the measured gas fraction. This fundamental problem will probably impede the
use of multiphase metering for fiscal applications yet for a long time.
3.0 MULTIPHASE FLOW METER - PROTOTYPE
The muttiphase flow meter prototype was originally designed to meet some of the requirements
to a subsea retrievable installation, refer to figures 3.0.1-3.0.2.
The technology forming the basis for a subsea flow meter is to a large extent developed for other
Framo products. All vital elements are maintained in a vertical stack-up configuration forming a
retrievable cartridge. The cartridge is when installed, located inside a receiver barrel which totally
protects H from the environments. The cartridge requires no orientation, and hence the crude
inlet and outlet enter and leave the barrel through ring volumes sealed axially by pressure
energized resilient seals.
running tool
lock-down mechanism
receiver barrel
cartridge seal system
have all reached a commercial level through the SMUBS project. In addtion, an integrated wet
mateable electriClsignal connector has been developed and tested by Framo.
The multiphase flow meter prototype includes some of these elements such as the receiver
barrel, cartridge seal system and the barrier fluid arrangement. The retrievable flow meter
cartridge consists of these elements:
Flow Mixer
Multi-energy Gamma-Meter
Venturi Meter
Cartridge Seal System
Barrier Fluid Arrangement
Electric/Signal Connector
MonHoring/Control System
The functional schematic of the flow mixer is shown in figure 3.1.1. The purpose of the mixing
unit is to provide always identical homogeneous flow condltions in the measuring section,
independent on upstream conditions.
Turbulent mixing is efficiently utilized in a turbulent shear layer, resutting in minimum pressure
loss. The feature of axial mixing incorporated in the unit makes Hpossible for efficient operation
also during intermittent and slug flow conditions.
The flow mixer is a purely static device comprising a tank into which the muttiphase flow is fed.
The most dense part of the fluid is drained from the bottom of the tank through an ejector, while
the least dense part is drained from the top and directed via a pipe back to the ejector, where H is
mixed with the dense part of the fluid, according to the ejection ratio.
Operation of the flow mixer can be described by grouping the various multiphase flow regimes
into Dispersed, Separated and Intermittent flows.
DISPERSED or distributed flow regimes such as bubbles in the liquid or droplets in the gas:
This flow is by ~s nature already well mixed. The same mixture will therefore be drained from
the top and the bottom of the tank, and mixed together in the ejector. Due to the very short
residence time, there is no phase separation occurring in the tank.
SEPARATED flow regimes, such as stratified or annular flow with low entrainment rates:
In this case each phase is continuously distributed in the axial direction, resulting in a steady feed
of both liquid and gas into the tank. Since the phases are already separated, the gravny force will
result in the formation of a liquid pool in the lower part of the tank wijh a body of gas above it.
The liquid flow from the pool creates a suction in the ejector. This draws gas from the top of the
tank via a pipe into the liquid flow. The resulting gas volume fraction is in accordance to the
ejection ratio. In the ejector where the gas and liquid meet, a strong shear layer is created.
Consequently an effective turbulent phase mixing takes place in the downstream section.
INTERM ITTENT flow regimes, such as slug flow and elongated bubble flow:
In this case the performance of the mixer is similar to the separated flow case, except that the
gas and liquid are not continuously fed into the tank. Instead, gas bubbles and liquid slugs are
entering the tank in a successive manner, causing the liquid level in the tank to vary. However,
the perforations in the interior pipe acts as an integral regulator, ensuring that there is always
both liquid and gas present in the tank. As the liquid level in the tank decrease and more gas
flows through the perforations, the gas volume fraction drawn from the mixer will increase and
consequently the liquid level will stabilize. If on the other hand, the liquid level increases, more
liquid will flow through the perforations. This liquid will also partly choke the gas flow. As a
result, the gas volume fraction drawn from the mixer will decrease, and consequently the
increase in liquid level will be reduced. This system is stable and the liquid level will always find
its equilibrium position. .
The multi-energy gamma meter provides the fractions of oil, water and gas in the flow, which can
be considered as volume fractions since the gamma meter is located immediately downstream
the flow mixer.
Calculation of the oil, water and gas fractions is based on the attenuation of different gamma
energy levels. The prototype meter consists of two gamma isotopes, Americium 241 and Barium
133 with collimators and two independent Nal (Til scintillation detectors of nuggedized design,
which can accept rapid temperature changes and sustain Shocks and vibrations.
The energy levels which can be used for determining the fractions are 18 keY and 60 keY from
the Am 241 source and 30 keY, 80 keY and 350 keY for the 8a133 source. The combination of
two different energy levels is sufficient to determine three fractions, Since the third fraction can be
deducted from continuity.
The use of a low energy level 18 keY or 30 keY is essential in order to distinguish between oil
and water. It is, however, important that the pipe wall is transparent to the low energy gamma
rays, since these absorb very quickly. A Boron Carbide cylinder is used as a gamma ray window
material. Independent tests have shown that this material is very transparent for low energy
gamma rays, and yet strong and hard enough to sustain the design pressure 01 345 bara and any
practical erosional load.
3.3 Venturi Meter
A venturi meter arrangement is used in combination wijh the gamma fraction meter to obtain the
flow rates of oil, water and gas. This is possible since the venturi meter is located immediately
downstream the flow mixer. Here the rnultiphase mixture can be treated as a single-phase fluid
with an equivalent mixture density, and the single-phase venturi relation can be applied. Defining
the equivalent mixture density as:
the relation between venturi differential pressure (DP) and total mass flow rate (rn-), can be
written as:
(3.3.2)
where;
v.; = GVF V.+(1-GVF) (3.3.3)
V. is the gas expansion factor, CG is a geometry constant and CF is the venturi flow coefficient.
The venturi differential pressure is the measured differential pressure corrected for the static
height between the pressure ports. Similar to the practice in single-phase measurements, the
venturi flow coefficient must be found by calibration. This is possible since the total mass flow
rate and the mixture density are known from the reference measurements.
It is an increase in dynamic pressure rather than the fluid velocity which is measured wijh a
venturi meter. The fluid velocity as calculated from the venturi meter is therefore dependent on
the mixture density obtained from the gamma meter. This is of great advantage when the flow
rates of the liquid components are sought, because an error in the liquid fraction will be partly
compensated by a resulting opposite error in the calculated fluid velocity.
The prototype venturi meter has a ~ ratio of 0.71, and is configurated with high precession quartz
crystal absolute pressure sensors, (Digiquartz 46K). These sensors are temperature
compensated. By using absolute pressure sensors, the measured venturi differential pressure is
not influenced by the density of the fluid in the wet legs. Included in the venturi meter is an
arrangement which allow a continuous or intermittent flushing of the wet legs. This way any
contamination in the wet legs is prevented, which otherwise could lead to a blockage of the
pressure ports. Flushing becomes particularly important for subsea and long tenm applications.
The seal system provides the sealing of the inlet and outlet ring volumes. Double pressure
energized lip seals above and below the ring volumes ensure proper sealing against the
environment.
In a subsea installation, the seals are set and pressure tested via the running tool during
installation and a pressure higher than wellhead pressure is maintained during operation of the
flow meter. This pressure is fed through the barrier fluid arrangement.
The barrier fluid arrangement provides for cleaning and protection of vital elements in the flow
meter cartridge. Important features are:
The electric/signal connector used for subsea installations provides for transmission of low
voltage power and signals to and from the cartridge and consist of two assemblies:
A female part integrated in the flow meter receiver barrel lower end
A male connector mounted at the lower end of the flow meter cartridge .
The connector requires no orientation and is made up simultaneously with the installation of the
flow meter cartridge. The connector allows supply of hydraulic fluid to the barrier fluid
arrangement. '.
Communication with the prototype flow meter is performed via the control system which is
located in and forms an integrated part of the flow meter cartridge.
The control system which applies transputer technology, conditions the signals from gamma
meter detector and the sensors and transmits them to the topside control unit (host computer) for
further processing. The topside control unlt will typically provide data communication, power
transmission to subsea flow meter and remote calibration of the multiphase flow meter.
The multiphase flow meter has been tested in a closed loop, where individual measurements of
single-phase oil, water and gas streams were compared with the multiphase flow meter
measurements on the combined oll-water-qas stream. The fluids used are Exsol D80, fresh water
and nitrogen gas. Schematic of the test rig arrangement is shown in figure 4.1.1.
Oil is taken from the oil outlets of four identical vertically installed three-phase separators and
routed to a horizontally installed two-phase oil-water separator for removal of any remaining
water. The oil from this two-phase separator is then routed through pumps, a single-phase oil
reference metering section and a remotely operated control valve before tt is combined with the
water and gas streams. Oil saturation pressure and temperature is measured at the oil exit from
the three-phase separators.
.,
Water is taken from the water outlets both from the four three-phase separators and the two-
phase oil-water separator and routed to a large low pressure water tank for removal of any
remaining oil. The water from this water tank is then routed through pumps, a single-phase water
reference metering section and a remotely operated control valve before ~ is combined w~h the
oil and gas streams. Any oil from the water tank is routed through a pump back to the three-
phase separators.
Gas is drawn from the top of the four three-phase separators and routed through a gas
compressor, a single-phase gas reference metering section and a remotely operated control
valve before ~ is combined w~h the oil and water streams. Strainers and scrubbers are located
both at the suction and the discharge side of the compressor.
The combined oil-water-gas stream is routed through a 26 m long 3" flow loop to ensure that any
three-phase flow regimes are fully developed. Part of the flow loop has been made of
transparent material, allowing flow regime visualization and registration. The stream is then
routed through the multiphase flow meter and back to the four three-phase separators. Pressure
is measured upstream and downstream the multiphase flow meter, so that the total pressure loss
through the multiphase flow meter can be calculated.
To enhance test rig operation, the oil and water pumps are ajustable by means of two
independently and remotely operated frequency converters, while the gas compressor is
equipped with a remotely operated flow control valve.
The oil and water streams can each be individually routed to an open tank with an accurate
known volume. This way the respective flow rates instruments can be checked, or when
necessary recalibrated. Samples of the single phase oil, water and gas streams are taken
regularly during testing and analysed in order to monitor separator efficiency.
As a reference to the rnuniphase flow meter the flow rates of oil, water and gas through it are
calculated based on measurements of the single-phase oil, water and gas streams and known
PVT correlations for the different phases. Minimum, maximum and standard deviations of the
single-phase measurements during the sampling time are obtained as well.
The single-phase measurements are corrected for any water in the oil caused by an incomplete
oil-water separation, and for the difference in the amount of gas dissolved in the oil and the water
between the multiphase flow meter station and the single-phase metering stations.
Dual instrumentations are used for critical measurements. The arrangement and specification of
the test rig instruments are shown in figures 4.2.1-4.2.5. For each instrument an operating range
has been defined to ensure an optimum overall accuracy.
All instruments has been factory calibrated with the actual fluids used in the test rig. PVT-data of
the fluids has been obtained from specific laboratorium analysis.
It was aimed to simulate flow conditions typical for North Sea oil and gas production systems,
and emphasis was given on creating realistic flow regimes. Most of the tests were run in the slug
flow regime, which is believed to represent the most demanding conditions.
Initially several twophase conditions was tested in order to verify the mixer performance and the
ability of the gamma fraction meter to distinguish between the different phases. The gamma
meter was calibrated on single-phase oil, water and gas.
The following four test series have been performed, and are described in this paper:
TWO-PHASE OIL and GAS
Total flow rate: 20-230 m31h
Gas volume fraction: 30-95%
Pressure: 9-14 bara
No. of test points: 155
TWO-PHASE OIL and WATER
Total flow rate: 20-135 m3/h
Water cut: 5-85%
Pressure: 3-11 bara
No. of test points: 44
5.0 RESULTS
Performance of the Multiphase Flow meter is evaluated by comparing measured and reference
quantities. "Measured" refers to the measurements obtained with the Multiphase Flow meter.
When the term "error" is used, lt is assumed that the reference quantity is correct, and the error
is then relative to this quantity.
Some of the results obtained by using t8 keV from the Americum 24t were hampered by drift in
the gamma meter, so this option could not be fully explored. All the results presented were
obtained by use of the Barium 133 source. The two-phase results were obtained from the 30 keV
energy level, while the three-phase results from the combination of the 30 and 350 keV energy
levels. The sampling Iime was always 1 minute.
The isolated performance of the venturi meter is expressed through the venturi flow coefficient as
defined in section 3.3, equations 3.3.1-3.3.3. It should be noted that any error in the measured
venturi differential pressure or in the reference flow rates will also affect the venturi flow
coefficient the way it is defined here.
The flow rate measurement results, as presented in this paper, are obtained by using a venturi
flolN coefficient equat to unity.
10
5.1 Two-phase Oil and Gas
Measured and reference gas volume fractions are compared in figure 5.1.1, and the
corresponding relative errors in the measured gas and oil volume fractions are shown in figures
5.1.2 and 5.1.3 respectively.
Rgures 5.1.5-5.1.7 compare measured and reference oil volumetric flow rates, gas volumetric
flow rates and total mass flow rates respectively, while figure 5.1.8 shows the corresponding
relative errors in the measured total mass flow rates.
Measured and reference gas volume fractions and total mass flow rates are compared in figures
5.2.1 and 5.2.2.
Measured and reference water cuts and total mass flow rates are compared in figures 5.3.1 and
5.3.2.
Measured and reference gas volume fractions, oil volume fractions and water cuts are compared
in figures 5.4.1-5.4.3 respectively.
Figures 5.4.5-5.4.8 compares measured and reference liquid, oil, water and gas volumetric flow
rates respectively, while measured and reference total mass flow rates are compared in figure
5.4.9. The relative error in the total mass flow rates is shown in figure 5.4.10.
6.0 DISCUSSION
Results from the two-phase oil and gas tests show that Ihe gas volume fraction is predicted with
good accuracy and repeatability over the whole range tested.
This has been possible since flow regime effects have been eliminated by the flow mixer.
Nevertheless, large relative errors are associated with small volume fractions.
Also the venturi meter shows good performance over the whole range tested, w~h an average
flow coefficient of 0.96. This shows that a simple equivalent single-phase venturi equation is
valid for a venturi meter located immediately downstream the flow mixer. All flow rate
measurements have been obtained by using a venturi flow coefficient equal to unity.
The oil volumetric and total mass flow rates correlate well with the reference values, while some
more scatter is observed in the gas volumetric flow rates. By replacing the applied venturi flow
coefficient of unity with the calibrated flow coefficient of 0.96, all measured flow rates would have
been reduced by 4%, resulting in additional improvement of the accuracy.
11
Observe that the relative errors in the total mass flow rates, as obtained by combining the volume
fraction measurements with the venturi measurements are significantly less than the relative error
in the volume fractions. This is a favourable feature of applying a venturi meter.
The two-phase water gas tests (100% WC) and the two-phase oil water tests (5 - 85% WC) show
similar performance to the two-phase oil gas tests.
Results from the three-phase oil water gas tests show that the gas volume fraction is predicted
with good accuracy over the whole range tested. The predictions of water cuts are more
scattered, however, some of the scattering can be explained by a slight drift in the gamma meter,
and the fact that the gamma meter needed a longer warm-up period to stabilise than initially
anticipated.
It is possible that these results could have been improved by an increased sampling time beyond
one minute.
The venturi meter shows acceptable performance with an average flow coefficient equal to 0.96
for the range up to 60% gas volume fraction.
The venturi flow coefficient appears to increase slightly for higher gas volume fractions. However,
ij should be kept in mind that the flow coefficient as defined here, is affected by the accuracy and
repeatability in the pressure sensors which, despite their high quality specilications, were no
better than 0.02 bar. This influence can be particularly significant at high gas volume fraction.
The average value of the venturi flow coefficient for all the three-phase test points is equal to 1.0.
All flow rate measurements have been obtained by using a venturi flow coeHicient equal to unijy
over the entire range. The liquid volumetric and total mass flow rates correlates well with the
reference values while more scatters are observed in the individual components' flow rates. 90%
of all the three-phase test points are measured within +/- 10% relative error in the total mass flow
rates.
7.0 CONCLUSIONS
The results presented show that the FRAMO Multiphase Flow Meter can be used for measuring
oil, water and gas volume fractions and flow rates over the entire range of gas volume fractions
from 0 - 100% and water cuts from 0 - 100%, and under flow conditions typical for the North Sea
oil and gas production systems.
The use of 30 keY and 350 keV energy levels from the Barium 133 source enabled accurate
estimates of gas volume fractions and reasonable estimates of water cuts.
The venturi meter shows good performance over the entire range tested, with an average flow
coefficient close to unity. This demonstrates that a simple equivalent single'phase venturi
equation holds for a venturi meter located immediately downstream the flow mixer.
For most of the test points, liquid volumetric and total mass flow rates are measured well within
+/- 10% relative error.
12
ACKNOWLEDGEMENTS
The authors express their gramude to the partners in the flow meter development project; Shell,
Norsk Hydro, BP, Conoco, Elf Aquitaine and Saga Petroleum for permission to publish this paper,
and to Frank Mohn Fusa AS for co-operation during testing.
REFERENCES
111 Rafa K., Tomoda T. and Ridley R. Flow Loop and Field Testing of a Gamma Ray
Compositional Meter, Paper 89-Pet-7 presented at the Energy-Sources Technology Conference
and Exhibition, Houston, Texas, Jan. 22-25, 1989.
121 Martin W., Woiceshyn G. E. and Torkildsen B. H. A Proven Oil-Water-Gas Row meter for
Subsea. The 23rd Annual OTC in Houston, Texas, May 6-9, 1991.
131 Olsen, A. B. and Torkildsen, B. H. Subsea MuHiphase Flow Meter System. UTC'92
30 March - 1 April 1992 in Bergen, Norway.
13
Tool Interfcce
Secls
Electronics
Connector
Jumper to X-mcs
Tree Control Pod
I I Unit
Inferior Pipe
Ejecfor
Threephose
Se~rators
(4 offl
IIGas
III 0;1
ml Woter
(RO
~G? Gp ~1
-
~
INSTRUMENT
OL
10
A.OW
INSTRUMENT
DESCAIPTON
..- ........
TKHNOI..OGY
NORMAl
RANGE
5-51
EXTENOED
RANGE
0.6-57
OPERATING
RANGE
0.6-35
-......
ACCURACY
0.0255
I.,
0
....'"
o 0
Hl
......... .,..
R..OW T&eHNOl.OGY
15-150 1.7-170 35-170 0.0750
o '''''''' ..,.,.
0
H2
(Ill"""
"""" """""
""""' ........
9.6-192 9.6-192 9.6-170
0.0432
0.4032
MICI'W) MOTION
esseee 200-1100 200-1100 700-1030 0.5000
ROO'~"" COAIOI..I$JIET'EI'I
rAAOSCIEHTlAC
"...,
OIGlOUAR1'l:3IK
Po ~AESS.TlWIISDUCEFI
0-69 069 5-20 0.0069
WATER RGfJ ~ p
EXTENDED
10 DESCRIPTION RANGE RANGE
OPERATING
RANGE ........... '"
ACCURACY
()
ow L
(m'lII)
FU:IW TECHNOLOGY
TUR81NfMETER 5-51 0.6-57 0.6-35 0.0255
OW
H1
.., ... FlOW TECMNClI..OOY
TURSINE"!'reFl 15-150 1.7-170 35-170 0.0750
..
H2 FLOW TECHNOLOGY
OW TURBlNEME'TEFI 15-150 1.7-170 1.7-170 0.0750
'"
TW """""""'NT
2....P RTD 0-150 0-150 5-40 0.1200
'''' TEMP. 1'RANSUITTEFI
~
G? p pGj'
,
J ~
Q L R.OW TECHNOI..OGY
TU .... IEMETEA 3.4-34 1.7-42.5 1.7-35 0.0340
G (m''''1
QGH1 FLOW T~Na.OGV
TUABNBoIETER 37382 8.5-425 35425 0.3820
H2
"'" ""' ...
QG "..,
V.eOH~"'fTEA
~ 20200 20,200 20-200 0.2-2.0
"AI'IOSCIENTIFIC
PG2 (b_.)
DIGIOUAR1'Z31K
"RESS. TJ'WoI5DUCEA
069 069 5-20 0.0069
TG ('<" yg"uo.
.... "'"
'''''''''""'
TAAHSMfTTiR
0-150 0-150 540 0.1200
FOUR THREE-PHASE
SEPARATORS
,.-- ,-
r-mr-
OIL OIL OIL E
- --
I-- r- l- t-
h
'-- - '-- -
~ OIL
~
ROSEMOUNT
Ps (bu"
3051 Pl.US
PRESS. 'TRANSMITTER
1-21.6 1-21.6 5-20 0.0155
ROSeMOUNT
TS 2440 RlD 0-150 0-150 5-40 0.1200
re. TEMP. TRANSMITTER
MULTIPHASE
FLOWMETER
<9
T
(9
T MULTIPHASE
EO
INSTRUMENT
Po
10
-,
INSTRUMENT
DESCRIPTION
ROSEMOUNT
3051 PLUS
PRESS. TRANSMITTER
ROSEMOUNT
NORMAL
RANGE
1-21.6
EXTENDED
RANGE
1-21.6
OPERATING
RANGE
5-20
ACCURACY
(NOfIIMAL
()
0.0155
~E1
TWO-PHASE OIL - GAS
~
L 100
c COIL .. 10-2(} m'/h
Z ~00lL .. 25-35
Q
I-
90
..0011,. .. 4050
m'/h
m'/h
0
-c 80
~
0011....
COOL..
55-65
7()'80
millh
m'/h
II: 00lL .. 8090 m'lh
u, 70
LU 0 0""- .. ~100 m'/h
:::;: 60 a .. 100-110 m',
=>
..J
0 50
>
'"o
-c 40
0
w
c:
=>
en
-c
W
:::;:
20
:20
10
0
0
REFERENCE GAS VOLUME FRACTION ('10)
FIGURE 5.1.1
v 0Ol." "'0-50 m'/h
0
...; 30 II> Qo.. .. S5-65m'lh " _ _
II: ~ 0OL" 7080 m'/h
~ LL. <0 001,. .. eo~om'/h
c: LU 20
c Qell" 510100 m'/h
.- , "'0
0 :::;:
c: => 10~~~a~~o2'~O~~',~c~m~~~ ..... -
c: ..J
W 0 ~ : ~ ..; ~ : ~
W >
> C/)
f= -c
o
-10
,..- h~r~\:\~:~~~~ ..
......r ~.,..,.."'.~ -,.~ ~ ~..-~ ~-, -
..J
o
W 0
c:: c:
W -20 l- ._-- ... ................... _ , - -
=> -20 ... -
'"
-c
w -40 l- . ... .;... ....;. ... ..,
:::;: .. .... 0
i
80 90 100
TWO-PHASE OIL - GAS
*
z
0
f=
(.)
z
a:
u..
a:
0 UJ
a: :::;
a: =>
UJ -'
UJ 0
> >
f= :!
0
-' 0
UJ
-20 - .... .. ; ...:.. -,.
a: UJ
a:
=> -30 ...................... .......... ....-
'":::;
UJ -40 ~ ... . ...... ( ......,... ...... -
-50~--;~~~~--~--~ .i ; __ ~ __ ~ i __ ~1__ ~ __ ~
o 10 20 30 40 50 60 70 80 90 100
REFERENCE GAS VOLUME FRACTION (%J
FIGURE 5.1.3
0.0 i i
0 10 20 30 40 50 60 70 80 90 100
REFERENCE GAS VOLUME FRACTION (%J
FIGURE 5.1.4
TWO-PHAS E OIL - GAS
(VENTURI FLOW COEFFICIENT EQUAL TO UNIT't')
120
c GVF .. 3040 %
::2
e- 110 ~ GVF .. 40-50 '"
.s
LU
100
v
~
GVF_50_60%
GVF_60_70%
<CI GVF .. 7080 %
~ 90
GVF _80.90%
c: eo o
0
LlJ ;;:
c: ...J
0 70
=>
Vl
LL
60
S2
LlJ c:
:::; ~ 50
LU
:::; 40
=>
...J 30
0
> 20
...J
0 10
0
0 10 20 30 40 50 60 70 80 90 100 110 120
c QVF .. 30-4.0""
A QVF .. 40-50 "" 8.
V GVF .. 50-60 %
10 GVF _6070%
....-.~~';~g.
<I aVF .. 70-80 '" D!<I
e aVF eo-so %
o GVF .. 90-95 %
.j6'., ..
I-~_~_....J .
....
; .~
. .
..
"
TWO-PHASE OIL - GAS
(VENTURI FLOW COEFFICIENT eauAl TO UNITY)
25.0 r--=-=-::::-:::-:::--1-....,.-....,.-....,.-"'"T'-""T"-""I:"'~,..
c GVF. 3O-~"lfo
co 22.5 ,6.
v
GVF .. 4().50'"
GVF .. 50-60 "
]> 20.0 ~ GVF .. 60-70 '"
W '4 GVF .. 70-80 '"
!;;: 17.5 o GVF .. SO-90 '"
::;:~ ,0.0
::;:
...J 7.5
5.0
~
2.5
50r-~c~a~~--_~1~O.~20:=m;.~h~r----r,.----r----",-----!r----r,,----;----,
'*
W 40 6 0cu._2S_35m'lh .. _~ : ~ _
I- v 0Ol." AQ50 m'fh
..:
0: 30 ~ o -5S.65m'lh .. . _.. . : _
i!: ;;: <4 0o .. -70-80m'lh . _ . . _ a
0: 0 20
0 ...J
u, ~ ~: ::::om;:Ih")'H~"~'r~~~~c~
0:
0:
W
W
'"'"::;:
-c
10
a _loO-110m\~
...... , ;..
i:t~~"Ifi~'~"tg::-
............... , , ., B ' H ", _
~ ...J C'
I- -10
..: ~ H'" H'''"" ."._
...J
W 0l- -20 HHHH'HHH
H' ..: ..
0:
e
( ..;
"." HH_
W
0: -30 I- "".-
::J
'"
..:
-40 ... ~
........
:.... ..... i .
H,'''H ,HH"."""'''H'''' 'H"",;""",_
W
::;: ;
-50 L.....-..J"--.....L_....I.._
....._..I-_~_L..--..I_....I.._.J :
o 10 20 30 40 50 60 70 80 90 100
REFERENCE GAS VOLUME FRACTION (%)
FIGURE 5.1.8
.J
TWO-PHASE WATER - GAS
~
1..- 100
D Q.T9l" 10-20 m~J
z .4 a"T~"25-35 m" . .. .~
. , . ....
Q 90
I- V a..TlIt".w.SO m' .;
0 to O.T9I" 5565 m' o ..~..
80
<C
cr:
<CI Q.la" 70-80 m'l
<> 0
'.a
u, 70 .. 8595 m' . ~. . ... .;.... . .. ! ... .......t;,
-.~... s .
w
:::E 60
:::>
-'
0 50 "C' c
> o
Cf)
40
<C
C>
0 30 . J
W
cr:
:::> 20
Cf)
<C .......
W 10 (
:::E
0
0 60 70 80 90 100
REFERENCE GAS VOLUME FRACTION (%)
FIGURE 5.2.1
~
en 25.0 GVF .. 2030 %
,:. GVF .. 3040 %
til 22.5
e
GVF .. 4050 %
ClJu. 15.0
;::'jCIJ ......... .. ..... .. ......
12.5 ; , ;
:::E~
:::E 10.0
-'
7.5
I-
0
I- 5.0 ........ ; ,
2.5 .... ; .
0.0
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0
.,
t,
TWO-PHASE OIL - WATER
100
o 0a.. .. 1C~
..20 m'lh
A 00.. .. 25-35 m'/I'I
90 .!, ..
~. .
;,
" .
'V 001." 4Q..50 m'/h
~
~ 11000.." 55-65 m'ni
80
f-- <I 001." 70-80 m'lh
::> o 0k" 85-95 m'/h
0 70
ex: o 0Ol. .. 100-110 m'l
w 60 v< .
f--
W 40 ...~..c..
ex: :ACI>
::> ..
en 30 ... :1 ... :
..: ,v
w
::;: 20
10 ....... ; ..
... :
30 40 50 60 70 80 90 100
REFERENCE WATER CUT (%)
FIGURE 5,3.1
A WC_l0_20%
.!!! 35 v WC .. 20-30 % .
'"
c ,.. WC_3C).~'"
W
f--
30
o
we .. 4.0-50"-
WC_50-80% ...:. ..., ; ..
..: o WC_BO-70%
0
w
ex: 25 we.7UO'" --.
ex: 3: ... we.so-ss".
::> 0
en -' 20r------!
..: u.
, ..
-j ....
w en
::;: en ....................
1)
..: 15 ...... : :
::;:
-' o
s0 10 .......""
<I 01 ~
0 .
f-- ................... .
5 . , ; ;
O~--~--~--~--~----~
o
__~ __~ __
5 10 15 20 25 30 35
..J
40
REFERENCE TOTAL MASS FLOW RATE (kg/s)
FIGURE 5.3.2
.'
..
THREE-PHASE OIL - WATER - GAS
~ 100~-C--W-C-_--~-'-0-~--~----~--~----r---~--~~--~--~
z 90 to. WC.1D20% ....:..
Q v WC.20-30",,"
I-
U 80 ""we .. 30-40 %
'4 we. 4050 ""
a:
u, <> we .. 50-60 ""
70
UJ
:;;
::l
-..J
: 50
UJ
40
o
o 30 ...,..
UJ
a:: 20
::l
UJ
UJ 10
:;;
100
REFERENCE GAS VOLUME FRACTION (%)
FIGURE5.4.1
;;e 100
c we_
~ 510%
A WC.1Q.20%.
Z 90
Q '7 WC.2030 ""
l- ~ WC_30_40 %.
U SO <I we _ 40-50 %.
a: 0> we _ 50-eO "'"
u, 70
o WC_60-70 %
LJJ
:; 50
::l
-..J
50 ....................... _;. .....
0
>
-..J 40
0
0 30 ...: ; .. . ...... l' ..
UJ
a::
::l 20
UJ
10 ..: : ,..
UJ
:;;
0
0 10 20 30 40 50 60 70 SO 90 100
REFERENCE OIL VOLUME FRACTION (%)
FIGURE 5.4.2
.,
UJ
a: 30
:l
CI)
..: 20
UJ
::;:
10
0
10
0 10 20 30 40 50 60 70 80 90 100
REFERENCE WATER CUT (%)
FIGURE 5.4.3
1.8 6 we ...10-20%
WC_2030%
......................... . -
V
I-
Z 1.5 we .. 30-40 ""
...... -
~
U
u::
u,
1.4
0
WC_40_50%
0 we .. 50-60""
we .. 60-70%
l..~
,. .....;...0 ....
UJ
0
U
;:
0
1.2 I---,---~--!
o.
Y.~.9
:
..9.-9-",
:
..'-
........ -
"1" ...-
-'
u, 0.8 r" H .. _.H ~H._
a: 0.5 ...... ....................... . ..................; -
:l
I-
Z
UJ
0.4 r ...... -
> 0.2 ..: .i- :.. ... , :... ... j ! j ! _
0.0 L.._.I....,_.I....,_-'-_-'-_-'-_; ....... ;
_-'-_-'-_....I
_ .......
o 10 20 30 40 50 50 70 80 90 100
REFERENCE OIL VOLUME FRACTION (%)
FIGURE 5.4.4
THREE-PHASE OIL - WATER - GAS
(VENTURI FLOWCOEFF/CIENT eOUAL TO UNITY)
120
D GYF_30_40
~
g 110 ." GYF.4Q.50
v GYF.5060
UJ 100 GYF.8070
... ~ ).. .
. .:.. . :
I- &10
0:;: 80
UJg
a:u.. 70
:::>0 ...... .....
(fJ- 60 ' "','
...:a:
UJI- ..... ; .
::;:UJ 50
::;:
:::> 40
...J
0 30
>
Q 20 .: - ,.. . , , .
:::>
0 10
~ o~-L--~~~ __~~ __~~~ __~~~
o 10 20 30 40 50 60 70 eo 90 100 110 120
REFERENCE LIQUID VOLUMETRIC FLOW RATE (m'th)
FIGURE 5.4.5
~-
:c- 100
c GVF _3D-AO
g 90
Ii. GVF _ 040-50
v GYF_50_60
... ~.
..
LU I> GVF.6o..70
I- 80
."
<{ <CI GYF_70_S0
a: GYF_BO_85
..
:;: 70 (10
--:. '
.. ..... ~ ) .
0 0
UJ ...J 60 ~ t'\J~ .... ,
.
a: u..
:::>
(fJ s:1 50 ..:-.. . :
..: ( )
...:a: v:
UJI-
::;:w 40 .....~.~
.. ~ .~ ..... ~ ... ~ .... j .
::;: . l>' .
:::>
...J 30 . ... ..!.. ........ , .. ", : " ..
0
> 20
...J
0 10 .... ; ......... .;.... "';' .; ............. ..; ...... . ...;..... ''':;'0'
0
0 10 20 30 40 50 60 70 80 90 100
REFERENCE OIL VOLUMETRIC FLOW RATE (m'th)
FIGURE 5.4.6
'.
THREE-PHASE OIL - WATER - GAS
(VENTURI FLOW COEFFICIENT EQUAL TO UNITY)
? 50
o GVF _30-.0
'k
w
45 .4 GVF_~50
V GYF _50-60
..... 40 1:10 GVF _60-70
... ~
.
<
a: 35
4 GVF _70.80
... .. ;
Co GVF .. 80-85
~
cO 30
w-'LL
a: 25
::>0
cn- 20 ................... !
<a:
W .....
15
::;;~
::> 10
-'
0
> 5 ......... :
a: .., ......,
w 0 0> . "'~A
....... ~ i ~. ...... ;
.....
-5
~
-10
0 5 10 15 20 25 30 35 40 45 50
180
V GVF .. 50-SO
.....!
"','
~ GVF .. 6().70
160 ..~..
<III GVF .. 7()..80
00 GVF .. eo-es
140~ ~-, .; ..
120
..:...
100
:.. , ..! .
"',
80
~~1:Io~.G~1~rr .
60
40
"'i
..n .
:4
.
<III
. .... -: " .;
.'". (.
20 ...
;
:.,i ;. , : : J
O~~--~--~--~~--~
o
__~__~~__
20 40 60 80 100 120 140 160 180
..J
200
"
.'
THREE-PHASE OIL - WATER - GAS
(VENTURI FLOW COEFFICIENT EQUAL TO UNITY)
!!!
30.0
'"
~ c GVF_3O-~
LU 27.5 .6 GVF _ 040-50 .............. ;;0
v GYF_5D-60
~ 25.0
a: ~ GVF_60_70
<I GVF ...70-80
3: 22.5
e GVF_80_SS
0
...J 20.0
u..
en 17.5
en
:;; 15.0
...J
-c 12.5
>-
0 10.0
>-
Cl 7.5
LU
a: 5.0
=>
CfJ
2.5
LU
:;; 0.0
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0
REFERENCE TOTAL MASS FLOW RATE (kg!s)
FIGURE 5.4.9
::i"
~
50.0
o we_ 5-10% , !
- " -
LU A WC_10_20% ....... .
>- 40.0 - -, ......
v WC_20_30"t.
a: 30.0
~ we 30-40 '% " -
.....
z 3: e we 4050 %
0
cr: ...J 20.0 c we .. 5060 % ...~
.... ..~. ..... ,..... -
o WC .. SO70%
0 u..
cr: en 10.0
a: en
LU
:;;
LU 0.0
> -c -'
i= >- 10.
...J
~ v
LU 20. I- .~ :.. ..:4 -:. _
a: Cl
LU
a: -30. I- . 4 A : .. .9
=>
CfJ
-40. I- l .: .... ....:... ....:. ...... -
LU
:;; -50.
i i ;
a 10 20 30 40 so 60 70 80 90 100
I
..
I
I ULTRASOOIC FU:W1ETER 'WET' GAS TESTS AT NEL
I
by
I
I R M Watt
NEL
t'
I
I Paper 2.5
I
I
Ie OORTH SEA FIJ:::1N MEASUREMENl' WJRKSHOP
26-29 October 1992
I
I
I
I NEL, East Kilbride, Glasgow
I
fj
I
I
I
ULTRASONIC FLOWMETER "WET"
GAS TESTS AT NEL
R M Watt, Flow Centre, NEL
I lntroductlon
As part of lhe "Ultraflow" Joint Industry Project to develop a
drop). Inside the separator vessel is a meshpad demister (Knit-
mesh Ltd) of a type developed specially for similar applications
offshore.
"wet" gas flowmeter(l), the British Gas/Daniel mulLipaLh
"I
I
I Close-up view of test meter with gas/liquid
separator in background General view of facility showing injection
head and controls trolley
turbine meter and the test meter (its axial position is actually o Liquid types: Water, Glycol, Water/Glycol
I variable from between 10 and 50 pipe diameters upstream of the
tesrmeter), Itis fitted with 200 spray nozzles (Bete:impingement
o Injection types: "Fine-mist" spray, Free liquid along
bottom of pipe
type, PI series) which inject the liquid in fine-mist form (droplets
I range.
At the end of the test-section, in order to protect the compressor
Refs.
I 1
I
..
I
I
I FU1fI cnIDITlOOS IN A GAS ME."l'ERIN; STATlOO
I by
I
,. D Tharassen arrl MLangsholt, Insitutt for Energiteknikk
arrl R Sakariassen, Statoil
I
I
I Paper 3.1
I
I
I
NEL,East Kilbride, Glasgow
I
II
I
I
FLOW CONDITIONS IN A GAS METERING STATION
~
I
Dag Thomassen and Morten Langsholt
I Reidar Sakariassen
I
SUMMARY
~ This paper focuses on the ability for computational models to predict
the decay of asymmetries in axial velocity profiles in long straight
pipes. In the design of gas metering stations this is an important
I parameter since it influences the required upstream straight lengths.
Asymmetries are typically generated by bends or by sharp-edged tees. In
this presentation we consider asymmetries with the latter origin,
I exemplified by the geometry of a gas metering station. Experimental
data were retrieved from a scale model of the geometry. Numerical
simulations have been performed using both the k-E model of turbulence
I present state of development. This is not to say that the k-E model
attains the desired accuracy - it gives too low decay rate for
asymmetries generated by a sharp-edged tee.
~ 1. INTRODUCTION
-I
actual discharge coefficient for orifice meters in metering stations of
different geometrical design. Provided a high degree of accuracy can be
achieved in this kind of simulations, we have established a useful tool
I
for detailed flow studies of the effect of piping layout in new gas
metering stations as well as for modification of old metering stations.
A question that is often raised is: Does the metering station comply
with requirements given in IS05167? One way of answering this question
I
is to check the geometry against the required straight lengths given in
the standard. This standard does not mention the manifold T disturbance
explicitly, so one is left to interpret the standard at this point. In
I
our mind, the layout does satisfy the standard.
Altogether this means that the upstream geometry to the manifold of the
I
gas metering station, i.e., primarily the S-bend, is of little
influence to the conditions in the meter-runs. Further, since the
manifold and the meter-runs all lie in one plane, the horizontal plane, I
one will not experience swirl of solid body type in the meter-runs.
This was observed in the simulation work, but will not be presented
here. I
Since the flow situation is very similar in all the meter-runs we will
limit the detailed results to be presented for meter-run B only. In
these high pressure simulations we monitored the solution in cross- I
-I
sections 6.8 0, 32.3 D and 50 0 into the meter-runs. Figure 3 shows the
non-dimensional axial velocity profiles along the horizontal diameter
in these 3 cross-sections. Figure 3 also includes the simulated fully
2
I
I
developed turbulent profile, using the k-E model of turbulence. The
profiles are made dimensionless using the maximum axial velocity
appearing along the same respective horizontal diameter.
I 2.3. Consequences
The results showed that the velocity profiles did not fulfill the
I requirements of ISO-5167. The TandemlPhoenics tool was, during the
verification phase, tested against experimental data for a geometry
similar to the metering station manifold. This test showed reasonable
agreement between measurements and simulations, but it must be added
~ that the measurements suffered from lack of accuracy. Therefore, when
we faced the simulation results, we did not have complete confidence in
them. Therefore, it was decided to build a Plexiglas scale model of the
I metering station and measure the profiles.
The Plexiglas model was built in IFE's atmospheric air rig in the scale
1:2.85. Figure 4 shows a photography of the header and the meter-runs
I of the scale model. Atmospheric air was sucked through the meter
station from a 30 D long pipe entering the scale model header. At the
inlet to this pipe we mounted a perforated plate to avoid external
3.3. Consequences
Through these scale model experiments we have experienced that in the
I
position of the orifice plate the velocity profile has a maximum
deviation from the fully developed profile of approximately 4%. This is
below the upper limit allowed for according to the requirements in ISO-
5167 regarding velocity profile conditions.
I
Compared to the simulation results for the high pressure gas metering
station, there is a distinct difference between measurements and
I
simulations at the position 6.8 D from the header as well as in the
32.3 D position. In the simulations the asymmetry in the axial profile
tends to have a slower decay rate than found in the experiments. It
must, however, be added that while the experiments were performed at a
I
Reynolds number of 9.104, the high pressure natural gas simulations
represented a Reynolds number of 9.106. The effect of increasing the
Reynolds number is to reduce the decay rate for anomalies like swirl ~
and asymmetries.
Therefore to achieve a proper evaluation of the Tandem simulator also
the scale model experiments should be simulated. The results from this
I
action are the topic of the next chapter.
I
4. THE SCALE MODEL - MEASUREMENTS VS. SIMULATIONS
The Tandem simulator was set up to simulate the flow in the Plexiglas I
scale model. Main geometrical data and operational properties for the
simulations were in agreement with the specifications for the scale
model experiments given in the previous chapter. The entire geometry
was divided into 5 sub-geometries. The only modification relative to
I
the gas metering station of Figure 2 is that the straight pipe and the
S-hend were exchanged with a straight pipe, see also Figure 4. The
manifold was operated with meter-runs A, Band D open for flow, while
I
outlet C was closed.
4 -I
I
~ 4.1. Tandem simulation of the scale model.
I 6.80 from the header we observe that both the simulated and the
measured profiles are asymmetric, but this characteristic is far more
pronounced for the former. This demonstrates that the simulations
I were not able to reproduce the measured decay rate for the asymmetry
between 3.80 and 6.80.
I This fact and the question it raised to the validity of the simulation
results put forward the idea of building a Plexiglas scale model of the
gas metering station. Measurements on this scale model showed that
there were discrepancies between the high Reynolds number natural gas
simulation results and the low Reynolds-number scale model
~ measurements.
I high Reynolds number case was 10%. The only conclusion we shall draw
from this observation is that Reynolds number effects are present and
must be considered when the scale model measurements are evaluated.
-I 5
I
Wa~~ roughness. All simulations are performed with hydraulically smooth
walls. Due to prior eKperience this is a reasonable choice both for the
Plexiglas tubes and for the process pipes in the gas metering station.
However, the value used for the wall roughness parameter is itself a
model assumption that influences the decay rate. A sensitivity test was
therefore made. The main conc~usion with respect to the decay rate in
the meter-runs remained unchanged even with relatively rough walls.
I
This eKcludes inappropriate setting of the roughness parameter to be
the source of error for the deviation between measurements and
simulations.
I
The turbu1ence modaL. Comparison of simulated and measured axial
velocity profiles 3.8D from the header showed good agreement. This
means that the entrance conditions for the velocity profile in meter-
I
run B is essentially correct. In other words, the manifold geometry
seems not to be the source for the disagreement between simulations and
measurements. We know from earlier verification tests(l) (2) that the I
Tandem/Phoenics system is able to predict the development of an
asymmetric velocity profile, when this is generated by a curved bend.
One major difference in the flow conditions downstream of a curved bend
relative to that downstream of a sharp-edged tee (which is the
I
equivalence of the gas metering station manifold) is the much higher
turbulence intensity associated with the latter.
6
II
I
I
.. 5.2. Results
Simulations with the Fluent code(5) have been performed using both the
I k-E turbulence model and the full Reynolds-stress turbulence model. Both
models are optional in the latest version of the program. The geometry
is described in curvilinear coordinates and the grid density is
approximately as in the Tandem set-up.
I X-E turbulence model. Figure 9 shows the axial velocity profiles
resulting from the Fluent simulations plotted together with the
I measured velocity profiles. The Tandem/Phoenics counterpart to these
results are found in Figure 7. We can see that the two sets of plots
(Fluent versus Phoenics) compare relatively well (notice the difference
I in y-scale). The most striking difference between the two sets occurs
for position 3.80 where the Fluent results deviates considerably more
from the measurements in the 'lower end' of the curve than do the
Phoenics results. This tendency lasts also for 6.80. At 32.30 both sets
I of results have a maximum deviation from the measurements in the radial
position O.BO. Due to the Fluent results the deviation from the fully
developed profile reached a maximum of 9% in this area.
it The Reynolds stress turbulence model (RSTM). The results from the
application of the RSTM of Fluent are plotted together with the
measured velocity profiles in Figure 10. 'We observe that the solution
I has changed considerably due to the switch-over from the k-e to the
RSTM. Unfortunately the RSTM did not improve the performance of the
simulations. On the contrary, both in the positions 3.BO and 32.30 from
I the header the RSTM results show a significant increase in the
deviation from the measurements compared to the k-e model. In the mid-
position, 6.BO from the header, the results compare to the measurements
I approximately as the k-e model does. Because of the poor comparison with
the experimental data we have found no reason to include profiles
showing the deviation from the fully developed profile for the RSTM.
I 6. CONCLUSIONS
The sharp edged tees connecting the meter-runs to the header in the
7
I
I
-I
plate, 32.3 D from the header. In this position the profile is
nearly symmetric, but flatter than the fully developed profile.
2
on Flow Metering, NEL, East Kilbride, October 1990.
..I
I
I
I
I
II Figure 2. The gas metering station interpreted for Tandem
I 9
I
I D.9
"
ro
I
E
OJ
:3
z- DB
I
'13
"
0
0;
>
'x
ro
0.7
/
"c -----a-- 6.8 0
I 0
.~
c
~
E
0.6 ~32.30
-X-50.0D
~
I Z
0.5 --Fultydev.
t' 0.4
o 0.1 0.2 0.3 0.4 0.5 0.6 0.7
II 10
I
I
Ie
I K
0.9
'"
E
I ~
~
O.B
I ~
'">
K
'"
1ij
0.7
--0- 3.BO
c:
I .2 0.6
'"
c:
"E
--c--
-32.30
6.8 0
~ --Fullydev.
e 0.5
I
0
,.
z
0.'
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.B 0.9
I z
25
15
Ie i
"
'0
"0-
S!
5
I
5
">
~
"
'0
15
:2
I g
c
0
E
25 ~3.80
-a-- 6.80
35
~ -32.30
I "
0
45
I
r-
I
I,
I
55
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 O.B 0.9
11
I
I
.. .- - - ....
I 0.9 .- ...
O'
X ......
- .. , ...... - .
I ~
~
i?>
0.8
o'
'8
I Q;
>
c;;
'x
0.7
~3.80sim.
... 3.80 meas.
'"c:
c;;
I .2
'"
c:
0.6
---0---
. ,. ..
6.80 sim .
6.80 meas.
"E
Vc: . . -32.3Dsim .
I 0
Z
0.5
.. --
'
.... - . - 32.30 meas.
t' 0.4
0 0.1 0.2 0.3 0.4 0.5 0.6
I
35
I 25
Ie ~
.l!1
eC-
Il
15
I s.,
~
5
"
~
15
I ~
E
,g
25 --0-- 3.8 0
6.80
c: --0--
g 35
I i
0 45
~32.3D
I 55
0 0.1 0.2 0.3 0.4 0.5 0.6
12
I
I
--I 0.9
. e- - .
.... : ..
..
K
. #1 :', .. - - ' - .
I E
~
2-
0.8
0.7
~
8
I a;
>
~
.,
><
0.6
... ~3.eOsim
.. ...
.. -. 3.80 meas.
0.5
1ii ....
I c:
.~
.,
c:
E
0.4
. - .. ....' ' --+-
......
6.80 sim.
6.80 mess.
I
0.3
0 .. e- 32.30 meas.
Z
0.2
t' 0.1
0 0.1 0.2 0.3 0.4 0.5
I ......
.,. _ .,.
. ..::::.,::: .'
0.9 .
I K
~ 0.8
~
Ie
0.7
~
~
...
0.6
>
1
I .,
1ii
e
0.5
I ~
'9
e
0
Z
0.3 .. ,.
~32.30sim.
6.80 meas.
0.2
.. e- 32.30 meas.
I 0.1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
II
13
I
I
.. CFD ANALYSIS OF FLUID PROPERTY EFFECTS
I
I
by
I
I R 11 Watt
NEL
t'
I
Paper No 3.2
I
I
I
..I NORTH SEA FLOw f!EASUF.E11ENT
26-29 October 1992
YORKSHOF
I
I
NEL. East Kilbride. Glasgow
I
I
II
I
I
..
I
I
I
TO BE DISTRIBUTED AT THE ~ORKSEO?
I
I
t'
I
I
I
I
Ie
I
I
I
I
I
..
I
I
Ie
I
I
RESULTS OF INVEST1CAT1OOS CGlPARIN:; SCME OF THE REXXM1ENDAT1OOS
I by
I
J F cabrol and A Erclal
it Statoil K-lab
I
I
I Paper 4.1
I
Ie
I OORl'H SEA F'IiJfJ MEASUREMENT Vl:lRKSHOP
26-29 October 1992
I
I
I NEL, East Kilbride, Glasgow
I
II
I
I K-Lab 92
Ie
I
I RESULTS OF INVESTIGATIONS COMPARING SOME OF THE RECOMMENDATIONS
GIVEN FOR TURBINE METERS BY ISO-9951 AND AGA-7
I
It SUMMARY
I KEY WORDS
Laboratory investigations, Turbine Metering, Standard Requirements
I
I
II page 1
I
K-Lab 92
I
RESULTATER AV UNDERS0KELSER SOM SAMMENL1GNER MOEN AV ANBEFAL1NGENE
I TURB1NMALER STANDARDENE IS0-9951 OG AGA-7.
ell
SAMMENDRAG I
Artikkelen unders~ker noen av de anbefalingene fra det nye inter-
nasjonale standardutkastet, 150/0IS-9951, for gassturbinmalere
mht. tetthetsmaling, tappepunkt for trykkmaling, kalibrerings- og
installasjonsbetingelser, hvor de avviker betydelig fra AGA-7
I
rapporten.
--I
I
RESULTATS D' ETUDES COMPARANT QUELQUES UNES DES RECOMMANDATIONS
DE IS0-9951 ET DE AGA-7 POUR DES COMPTEURS A TURBINE
I
Cette publication etudie les recommandations de la nouvelle
ebauche de norme internationale IS0/01S-995l sur les compteurs
I
a turbine, en ce qui concerne la me sure de la densite,
la localisation des prises de pression, et les conditions
d'etalonnage et d'installation, la ou elles different de fa~on
notable du rapport AGA-7.
J
Les debits volumiques mesures par plusieurs compteur a turbine ont
ete compares en gaz nature 1 aux debits massiques mesures par des
I
tuyeres soniques qui sont les compteurs de reference a K-Lab.
La masse volumique, necessaire pour transformer Ie debit de
massique de reference en debit volumique, est soit calculee au
I
niveau du rotor (confer ISO/OIS-995l) soit mesuree en aval du
compteur (confer AGA-7). Cette etude souligne l'importance de
la determination de la masse volumique et son influence sur
I
la precision des debits massiques correspondants.
1. INTRODUCTION
IS0/DlS-9951, see reference n02) are necessary in order to allow
turbine meters in being fully accepted for fiscal service,
especially in the North Sea, were operating conditions are far
away from usual calibration conditions. The requirements
prevailing today are those reported in the Transmission
Measurement Committee Report N7 (ccAGA-7in short, see ref. nOl).
This laboratory investigation compares the recommendations
circulated in 1990 by the International Standardization
Organization for gas turbine metering (IS0/D1S-9951) to those
issued in 1985 by the American Gas Association (AGA-7).
reference flowmeters and statement o.funcertainty).
Tests on 6 different types of turbine meters were performed during
the period October 90 -August 92.
2. DESCRIPTION OF THE TESTS
sections describing the various tests undertaken. The common point
of all meter installations is that they comply with both AGA-7 and
IS0/0IS-995l.
The flow-straightening vanes used are built according to the AGA-7
The reference flowmeters are K-Lab's bank of toroIdal throat sonic
nozzles, designed according to IS0-9300 and individually primary
calibrated in K-Lab's gravimetric calibration rig using natural
gas.
II The reference mass flowrates measured using the sonic nozzles are
evaluated to have an uncertainty of 0.3%, calculated with two
standard deviations.
page 3
K-Lab 92 I
2.3 Other instrumentation used during the tests
The two gas density transducers used during testing (a low density
transducer from a to 60 kg/m', and a high density transducer from
--I
o to 400 kg/m') provide a continuous measurement of gas density.
The transducer senSing element consists of a thin metal cylinder
which is activitated so that it vibrates in a hoop mode at its
natural frequency.
I
The mass of gas, passing over the inner and outer surfaces of the
cylinder and vibrating with the cylinder, decreases the natural
I
frequency of vibration. Frequency and gas density are related.
page 4
II
I
I K-Lab 92
Ie 3. DENSITY MEASUREMENTS
I unmetered by-pass.
In practice (reflected here by the AGA-7 report), a length of
5 pipe diameters (5D) downstream of the meter body, is often
I considered to be as close as possible, without perturbating the
flow profile significantly.
I
K-Lab 92
I
Even if there is a risk that the measured density in such a
by-pass loop does not represent the density at conditions
eI
required, this arrangement is often preferred in metering service
because maintenance, checks and verifications (such as vacuum
test) are more easy to perform. Further, protection of the
I
measuring element is more easy to design and control than with
inserted densitometers (see ref. 4).
This very configuration is used in K-Lab's test section to get the
mass flow rate evaluation when using turbine meters.
I
The densitometers were installed in a 6" test section, within 5D
(5 pipe diameters) downstream of turbine meter A.
I
turbine meter A was installed 15 internal diameters of straight
pipe (ISD) downstream of 2 twisted bends. A flow conditioner
specially deSigned by K-Lab was installed upstream of the meter. I
The meter proved during several tests to give very repeatable and
reproducible results, which is important for the present tests.
The tests were done in February with a gas temperature of 38C in
I
the test section, which is about 30C above the average ambiant
temperature during the year. The sample gas flow has an initial
temperature which is approximately the line temperature.
Ey exchanging heat with the environment the gas temperature of the
sample will change. A difference between the temperature inside
the densitometer compared to the line temperature might occur,
--I
even if the densitometer is installed into a pocket protuding in
the pipe.
Therefore a cold ambiant temperature cools the sample gas, which
should ideally be maintained at the same temperature level than
I
the flow.
As the same gas is continuously circulating around the calibration I
loop, the gas composition could be measured accurately with an in
line gas chromatograph and the density, calculated with the AGA-8
equation (1985 version) could be used as reference in this test. I
A densitometer was installed in a specially manufactured pipe
spool. Both the densitometer and the inlet tubing were well
insulated within a box mounted around the instrument. The length
of pipe between sample take off and densitometer was minimized and
J
kept as close as possible to the surface of the meter run pipe.
There was no heater in the box. I
All these precautions were taken because the temperature
difference between sample gas flow inside the densitometer
and main gas flow depends on the type of insulation, heating
I
or cooling, piping geometry, and sample flowrate as well as
on the difference between ambient and run temperature.
I
As previously said, the piping geometry and the insulation were
optimized for these tests. No supplementary heating or cooling was
provided in the test section because a temperature equilibrium is
difficult to achieve, due to ambient temperature variations and
I
various exposure to sunlight.
I
page 6 II
I
I K-Lab 92
page 7
I
K-Lab 92 I
~
Another 5" turbine meter of the same type was installed (see ins-
tallation sketch nC4). The tests consisted to vary the sample gas
flow through the densitometers by closing more or less the needle
valve (see figure n02), from 10% to 100% of the maximum sample I
flowrate.
Testing was performed at 35 bar and 37C. The measured density
was obtained with two positions of the temperature sensors used
I
for temperature correction (Twa 11' and Tups t ream ; see figure n02).
Only the results obtained with T IJ will be reported here.
I
(However, the same procedure couYa Oe done with Tups t ream ).
The temperature of the insulated box was maintained at 2B C.
The tests were runned at 30% and 70% of the maximum flowrate of
I
the turbine meter.
As said before, generally, the density finish to be adequated with
I
the calculated density, whatever the flowrate, when the density is
stabilized sufficiently long time. After start up of a cold meter
run, it would take hours before the density was stabilized without ~
the proper sample f10wrate (with it some minutes were sufficient).
During these tests, we always waited for density stabilization.
I
3.3 Discussion of the results
3.3.a Densitometer correction (see figure 3) I
At 100 bar the results show a difference between measured and
computed density of 0.5% at 966 acmh (actual cubic meter per hour)
and 0.9% at 45 acmh.
I
This deviation is more than expected and is larger at the lower
flowrates. It shows again how difficult it is to measure density
I
without a proper procedure. Moreover, it is especially difficult
at low flowrates were the gas cools down faster.
For turbine metering, it means that this is not a proper method of
measuring the volume reference flow. J
At 20 bar the results were expected to be worse than at 100 bar
because the gas transports less gas and heat than at 100 bar.
I
Nevertheless the difference between measured and computed density
was between 0 and 0.2%, which is within the uncertainty bands of
both the densitometer and the AGA-B equation.
I
At this pressure, a low-density densitometer was used, but it
seems difficult to impute the improved results to it because both
high-density and low-density densitometer were of the same type.
I
The most likely explanation is that the f10wrate through the inlet
tubing of the densitometer is higher at low pressure.
I
Indeed, if the flowrate has a higher value through the
densitometer, the gas has no time to cool down so much, and the
densitometer should provide a more accurate density.
I
~
page 8
I
I K-Lab 92
V(PlOO/P20) = 2.4
I
,. where PIOO is the gas density at 100 bars and P20 at 20 bar.
The calculation use the general equation given for the rotameters.
Therefore, at lower pressure, the actual sample flowrate through
the rotameter and the densitometer becomes higher (even if the
I level in % is the same); this results in a decrease of transporta-
tion time between tapping and densitometer; hence the risk of
temperature decrease is minimized.
I 5uch densitometer-correction effect is not taken care of in the
AGA-7 reports, but is implicit in the 150 recommendation because
..
I
is to be selected in all cases, when using the temperature sensor
associated to the densitometer (in the wall).
3.4 Conclusions
I In AGA-7, the practical procedure recommended for density
measurement should be improved by taking care of densitometer
I correction and of sample flowrate.
Replacing measurement by calculation would be another solution,
which is implicitely allowed by the new 150/D15-995l requirements.
I
I
II page 9
I
X-Lab 92 I
I
% OF
I MAX
FLOW-
Protor P
100D
DIFF Protor P100D DIFF
it
I 4.3 Discussion of the results
Ie
of the gas through the inlet stator of the turbine meter.
4.4 Conclusions
I Both requirements will lead to install pressure tappings at the
right location, namely at the turbine wheel of the meter.
I
K-Lab 92 I
S. CALIBRATION CONDITIONS eI
5.1 Standards requirements I
ISO claims in section 8.2.2 that
The prefered calibration is one which is carried out at
I
conditions as close as possible to the conditions under
which the meter is to operate.
I
AGA specifies in section 5.6.1 that
In the flow measurement of natural gas, the accuracy of a gas
turbine meter as indicated by the meter output is generally
I
specified as within l.O' of the true volume over a certain
specified range and pressure range using air as the calibration
flow medium. For accuracy better than l.O' .. meters should I
--I
be calibrated under conditions near the meter's intended
operating condition.
This last specification means that calibration under operating flow
conditions should only be necessary when the required accuracy
should be better than within the error band, generally specified as
within tIt of the true volume.
I
,. Then, and even if the ISO recommendations are somewhat vague by not
specifying precise installation requirements, it fix a maximum meter
error of 4/3 % for a perturbation to be generally acceptable.
I calibration purposes.
Nevertheless, this last requirement might not be stringent enough.
This issue will be discussed in the following.
II page 13
I
K-Lab 92 I
Unfortunately, the ISO/OIS-9951 contains an informative annex E,
eI
which gives other informations concerning perturbations effect,
that we shall question through the following statements I
Namely section E.4.2 determines piping configurations, with two
elbows not in the same plane and 5 pipe diameters downstream,
I
as representative for Low Level Perturbations.
The paper ref. n06 establishes for several turbine meters a swirling
flow up to 5 with such configurations and an average overmetering
of nearly 1% in natural gas at high pressure.
I
This cannot be deemed as low level perturbations.
This review shows again that one has to be very careful when
generally speaking about installation effects.
I
The 150/015-9951 recommendations are deemed more appropriate on this
matter, apart from the informative sections in appendix which might
be discussed.
J
Nevertheless, the AGA-7 requirements are more spread and certainly I
fits better metering needs for pratical requirements.
I
I
I
I
page 14
II
I
I K-Lab 92
~ 7. CONCLUSIONS
The overall conclusions to be drawn from the tests and the analysis
I can be summarized as follows :
I AGA-7 I50/0IS-9951
REQUIREMENTS
I CONCEPT PRACTICE CONCEPT PRACTICE
DENSITY MEASUREMENTS + + ++ +
~
I PRESSURE TAPPINGS ++ + ++ ++
I INSTALLATION CONDIT. - ++ ++ -
-I page 15
K-Lab 92 I
8. ACIrnOWLEDGEMENT eI
This study on turbine metering requirements has been possible thanks
to the cooperation of the following manufacturers : Instromet, Faure
Herman, Equimeter, Daniel, Elster and Hydril.
I
The assistance of our colleagues from K-Lab, all along the tests and
in their analysis is gratefully recognized.
I
9. REFERENCES I
1. Measurements of Gas by Turbine Meters , " AGA-7 "
Turbine Meter Task Group, American Gas Association
Transmission Measurement Committee Report N7 (1985)
I
2. Measurement of Gas Flow in Closed Conduits - Turbine Meters
Draft International Standard " ISO/OIS-9951 " I
--I
International Organization For Standardization (1990)
3. Petroleum Measurement Manual -Part VII- Density - section 2
~he Institute of Petroleum, London (1983)
4. Installation Details For Gas Densitometers
Reidar Sakariassen
North Sea Flow Measurement Workshop, Bergen, Norway (Oct. 1991)
Ie FLOW OIRECTION
I
I - ,---
"'lf~Ll ".~ ;
9
'It
0::
I ~
L-
""LEi OUilE
rr..
All
'~OT 4Q'I' '1l a'll
I
/
I I
I
(
I
I
I
1\ I
I
I I
I '\
f- 250mm 6S0mm
II
I
I
DENSITOMETER CORRECTION
0.'
0
o
( SERIAL NR. 45623 ) I
0.2
I
01-----~~----------_=~~-~~------------------------------------------
."
-0.2
Y
.,/ ..............
~.--
..-g--------
A '___ 6 ~
I
e~
n
-0 ....
-o.S
E a 100 bar
Ita
e.
o
o
-0.11
-.+-----.------,----.------.-----,----.------.----------.----0-----.
20 bar I
o 100 ~OO !!IOO 400 500 BOO 700 800 gOO 1000
"...
MA[N VOLUME FLOWRATE (ACMH)
K-LAB 1992 I
I
I
II
I
6" T[ST St(TlO"
I
nov OII/EeTIQN
---
I
-
1000
:11
l~OO.".,. l""Hi SPODL PJEtE 16", 1500
~\ eeo- .. lONG ... -ox 16"" ''::.?O =<TJ I~~ 'S"'!AlL ~[::; C'I!\
-: ',6~"'''' _C~G \~"S'r~"'!"'l S~DO~If: !')O()'" ~iJl .. ,;~ 0"': !!~P!RA:UQ! <;[ ... SO~ ' ..D D["S!'O'"'!!P ,..5-~;.!:"
I
-.1 I
'i , ......
, ... ~r.,, --
I i
I I
6"
FLOW
TEST SEC TiOrJ
DIRECTION
I .-._='------
500 1000
I 'O)~
/
HH I
!1I (2)
iH ~
tJ) (4)
i-=-1 f-----1 Hf---1 f---if--f
(5) (5) (7)
I ,1)
(2;
om
1.300mm
LOrlG
LOC;..TEC'
SPOOL PIECE .e-.
AT 21G'Torr. FROM THE OUTLET rUNGE
1=00- FLANGED, R1 J )
LurlG SPOOL PIECE (6", 1500- rLAt~GEu. RTJ) wITH 19 ItlTERNAL STRAJGHTENING VANES
(3) 540mm L(;rJG SPuOl F-IECE 16'.1500 RTJ uPSTREAM, 600 FLANGED DOWNSTREAM. RTJ)
"
I
(4)
!s)
'E-)
i7)
.'.",0,
450mm
STATIC
110Qmm iouc
THERM(i"N[Ll
1000m,r.
hem
cone 6"
PRESSuRE
METER t. WTH 600
TRAtl$MlTTER
W[LDOl.[T
l (Itl'~ SPOOL
L'':'rlG '~FOOL
LO(t..T[D
SPOOL PIECE (6" lS00,RTJ
FTJ FLANGES AND THE HQNN[YW[LL
AT THE ROTOR.
DOWNSTREAM,600- RTJ uPSTREAM)
LOCATED AT ':OOmm FROM THE INLET FLANGE.
PIECE
FIEe[ 16", 1500-, Fi TJI (wITH
,6',1:'00-.;:;1))
DErj$IT i SOLARTROrl POCK[
SMART
WtTH k-LAS'S
T ItI OPTlGt~)
SFECIAL
I
I
I DENSTIY EFFECf ON TURBINE METER CALIBRATIONRESULTS
Ie 1.8
2
( Naturcl gas at :!il Oeg C and .35 bar)
( !URSINE METER A' )
I 1.8
I.' _._.---8. __.__.__ .-_.-_.---<1
1.2
0---'-'-
I '"
0,0
0.0 <>--.--.--.--.-- ~.~.=
..
==:.=.= .
t>:
0
0.0 )( . ...~..
I
t>: 0.2 x- .. :1(
t>:
0
'"
Z -0.2
0
-0.4
~
I >
Oil
'"
-O.B
-o.e
-I
(CWHXlTTID: = of m=n.m _ of 1Iw maIr) 0 Comp~d.n.;~.A-8)
o ~~red I?ensity. (Twf?~
I
-1.2
( DOTTED : 7w. of rTD<hun _ of 1Iw maIr ) x .(~qA::.~1)
~.~.r:'.~.~ry
~.c?~p..~.~~~.
-1.
-I.B
x ~.~~~~!:~~ ..(]:~~!!J..
..?~0.~!!y.
-r.e
-2
0 10 :'0 2040 so so 70 80 00 I "0 I2D
SAMPLE GAZ FLOW RATE ( ~ of the maximum)
T~ of !he nsuatod to< = 28 c>: K-LABl992
I
I
--I
I
I THE NlA TRANs-ITSSIOO ME'ASUREMENl' CXMIITI'EE AND THE REVISIOO
NlA REPORT No 8 CXMPRESSIBILITY FJ\C'IORS OF NATURAL GAS
OF
I
I by
it J Stuart, Pacific Gas and Electric, J Savidge, Gas Research Institute and
S Beyerlein and E Lamon, University of Idaho
I
I
I
I Paper 4.2
Ie
I
I NJRTH SPA FI.lM MEASUREMENl'
26-29 ~ 1992
IDRKSHOP
I
I
I NEL, East Kilbride, Glasgow
II
I
I
..
I The AG.A Transmission Measurement Committee and the Revision of
AG.A Report No.8 on Compressibility Factors of Natural Gas
I John Stuart
Chairman, Transmission Measurement Committee
I Pacific Gas and Electric
Jeff Savidge
Gas Research Institute
I Steven Beyerlein and Eric Lemmon
University of Idaho
I
"
I
SUMMARY
In the United States and Canada, there are several organizations that are directly
involved in improving the understanding and practice of natural gas measurement.
One of these organizations, the AG.A Transmission Measurement Committee, is
described in this paper. This paper also discusses the results of a recently completed
I project of that committee, the 1992 revision to AG.A Report No.8 on
Compressibility Factors of Natural Gas. As part of this discussion, two new
compressibility calculation methods will be compared to each other and to the
NX-19 method for a reference Ekofisk Gas as well as two gas compositions found in
I the North Sea.
1. FLOW MEASUREMENT ORGANIZATIONS IN TIlE U.S.
I The following is a list of the major U.S. organizations directly involved with the
research, testing, technical recommendations, standards, and regulatory matters of
II
I
.'
.-
2. THE AMERICAN GAS ASSOCIATION I
The American Gas Association (AG.A) is a national trade association with a
membership from 250 natural gas distribution and transmission companies located
throughout the United States and Canada, as well as overseas. I
The AG.A staff consists of approximately 180 people located in Arlington,
Virginia.just across the Potomac River from Washington, D.C. Another 180 people
work in the AG.A Laboratories, Cleveland, Ohio, testing and certifying gas
.
I
appliances. AG.A's annual budget is approximately $50 million, supported by
member dues, testing fees, conference registrations, publication sales and other
sources.
I
The AG.A committees are organized into four sections:
I
--I
1. Legal Section
2. Marketing Section
3. Financial and Administrative Section
4. Operating and Engineering Section
Each section has several committees made up of member company experts. These
committees typically meet two or three times a year to discuss mutual problems,
share solutions, produce recommended practices and/or standards, and formulate
gas industry policies. AG.A's government relations and communications staff then
strive to effectively communicate these policies, etc. to legislators, regulators, and
industry. For example, AG.A spends about $15 million per year on national
I
advertising, describing the benefits and advantages of natural gas to our customers.
The AG.A Operating and Engineering Section consists of about 700 technical
I
experts representing their individual companies, on 17 different committees.
Ct!mmittee Members
38
I
Automation and Control
28
tI
Compressor
Corrosion Control 29
Customer Service and Utilization 40
Distribution, Construction and Maintenance
Distribution Engineering
Distribution Measurement
Environmental Matters
Fleet Management
Gas Control
63
40
35
46
44
35
I
Materials Management 30
31
Pipeline
Plastic Materials
Safety and Occupational Health
68
37
I
Supplemental Gas 37
Transmission Measurement
Underground Storage
49
40 I
TOTAL = 690
I
II
I
I
..
I Committee activities and projects are aimed at addressing a set of key industry
issues: saftey, regulations/standards, environmental, communication, Gas
Engineering ana Operating Practices Series of books, technology, third party
damage, quality and productivity improvement, and research identification.
I
3. mE A.G.A TRANSMISSION MEASUREMENT COMMI n EE (TMC)
I The TMC consists of 20 members from transmission and distribution companies,
and 26 associated members from manufacturing, research, and educational
institutions.
I The scope of projects covered by the TMC include the procedures and practices for
installing, operanng, testing and maintaining metering and associated equipment,
I including volume and pressure control equipment which is used in the production,
gathering and transmission of natural and substitute gas from the source to the
outlet of a transmission line ~ate station. Also included in the scope of TMC
t' activities are the scientific pnnciples, applications and usage of all types of
volumetric, weight, and energy measurement devices associated with the metering
equipment specified above.
I The current three year plan for the TMC includes the following projects:
I
II
I
I
..
4.1 Background
During the period 1981-1984, the AG.A through its Transmission Measurement
I
Committee and the GRI sponsored development of an equation of state to provide
tile gas industry with state-of-the-art predictions of compressibility factors for
natural gas metering applications. Initial work used data ranging in pressure up to
I
approximately 6 MPa obtained from the literature and provided by the GERG.
However the GERG data bank was extended considerably in the period 19851990 ..
The new data showed that the original equation developed in the period 19811984 I
needed improvement. In addition, velocity of sound data obtained under GRI
sponsorship during 1985-1989 showed calculations for rich gases were not
sufficiently accurate for critical flow applications, These data were included in a
new thermodynamic property correlation for natural gas mixtures. The resulting
I
equation of state is referred to as the Detail Characterization Method and is
documented in AG.A Report No.8. I
--I
4.2 Natural Gas Characterization Methods
Two highly accurate models for computing compressibility factors in gas
measurement applications are presented in AG.A Report No.8. One model
applies a detailed knowledge of natural gas composition to compute the
compressibility factor (i.e. usin~ standard composition information from a
chromatographic analysis). This is the Detail Characterization Method and can be
applied over a wide temperature, pressure, and composition regime. A second
model applies an aggregate or gross knowledge of natural gas composition to
compute the compressibility factor. This model is the Gross Characterization I
Method which was developed under sponsorship from the Groupe Europeen de
Recherches Gazieres (GERG). The method can be applied within the custody
transfer region which extends from 265 to 335 K at pressures less than 12 MFa.
Nei ther model is recommended in the liquid phase or within 5K and 0.2 MPa of the
I
critical point.
The Gross Characterization Method was developed by GERG and modified for
I
implementation in the U.S. These modifications have to do with the specification of
reference conditions for metering and for determination of heating value. The
Detail Characterization Method was developed using a three ste!? procedure using
compressibility factor data obtained from the literature and provided by GERG. J
(1) First, an equation of state for key pure components was developed using
compressibility factor data for methane, ethane, nitrogen, hydrogen, and
I
carbon dioxide along with velocity of sound data for methane. The equation
of state terms were chosen using a procedure which minimizes the number of
terms required for a given accuracy, I
(2) Second, compressibility factor data for key binary mixtures were used to
(3)
determine binary interaction parameters for key binary component pairs.
Third. the GERG compressibility factor data for 84 natural gas mixtures were
I
used to evaluate the accuracy of the equation of state for natural gas
compressibility factors. In addition, velocity of sound data measured by NlST
for four natural gas mixtures were used to evaluate tile equation of state for
I
velocity of sound predictions.
The Detail Characterization Method and the Gross Characterization Method have I
been incorporated into efficient computer programs for computing the
II
I
I
.. compressibility factor, Z, the mass density, o, and the supercompressibility factor,
I Fpv. The programs have been designed for the following purposes: (1) efficient
implementation on flow computers, (2) as a guide for the development of
application programs in the gas industry, (3) for computational verification; and (4)
I for utility purposes such as tabulating Z, p, or Fpv for particular gas mixture
compositions.
During the last two years the Gas Research Institute in coordination with Gasunie,
Ruhrgas, and Gaz de France have sponsored highly accurate measurements of
I natural gas mixtures at the Nationallnstitue for Standards and Technology (NIST),
Texas A&M, Van der Waals Laboratory, and Ruhrgas. PVT data for the five
natural gas mixtures shown in Figure 1 were obtained in this research. All mixtures
I were gravimetrically prepared and chromatographically verified by NIST and then
sent to each of the participating laboratories. Density measurements were taken
over temperatures from 225 to 350K and at pressures up to 70 MPa. These
t' "reference data" represent the state-of-the-art in PVT measurements for natural gas
mixtures. Intercomparison of the data shows an average agreement between the
experimental measurements from the four laboratories of O.035%. These PVT
reference data were acquired after finalization of the Detail Characterization
I Method and the Gross Characterization Method. As such, these data provide an
independent verification of both characterization methods.
I Figure 2 compares PVT reference data for Ekofisk gas with density predictions from
the Detail Characterization Method as welI as the Gross Characterization Method
at 275K and 300K. Density deviations are calculated as,
I compares density predictions from the NX-19 Method against the Detail
Characterization Method for Ekofisk Gas. Density deviations are calculated as,
I Gas within 0.05% at pressures less than 12 MPa, while the NX-19 Method is in error
as much as 2%. As the concentration of heavier hydrocarbons increases, the
differences in predicted density between all three methods becomes more
pronounced. Molar composition of natural gas from two North Sea fields is given in
I Figure 5. Fi~re 6 compares density predictions from the Gross Characterization
Method against the Detail Charactenzation Method for the Statfjord Gas. Figure 7
compares density predictions from the NX-19 Method against the Detail
I Characterization Method for Statfjord Gas. Figure 8 compares density predictions
from the Gross Characterization Method against the Detail Characterization
Method for Veslefrikk Gas. And Figure 9 compares density predictions from the
II
I
I
..
NX-19 Method against the Detail Characterization Method for Veslefrikk Gas.
Figures 6-9 illustrate that the Detail Characterization Model should be favored in I
predicting compressibility factors for rich gas mixtures such as those found in the
North Sea. Unfortunately, little reference quality experimental data are available
for evaluating the accuracy of compressibility factor predictions for rich gas
mixtures,
I
4.4 Recommendation to Gas Industry Users
I
In the United States the Gross Characterization Method will be implemented
primarily for transmission/distribution system applications. This method will only
be used for compressibility factor calculations. All derived physical properties will
be calculated with the Detail Characterization Method. The Detail .
I
Characterization Method is applicable to transmission/distribution conditions and is
expected to be applicable to a broad range of production/processing conditions.
Research is underway to investigate the data and modeling needs for heavier gas
I
constituents.
REFERENCES ~
STARUNG, K..E. and SAVIDGE, J.L, "Compressibility Factors of Natural Gas and
Other Related Hydrocarbon Gases", AG.A Transmission Measurement
Committee Report No.8, 1992.
I
STARUNG, KE., FITZ, C.W., CHEN, Y.c., and RONDON, E., University of
Oklahoma, and JACOBSEN, R.T, BEYERLEIN, S.w., CLARKE, W.P. and
I
LEMMON, E., University of Idaho, "GRI High Accuracy Natural Gas Equation of
State", ISO TC193/SCl/WG7 Report, 1991.
JAESCHKE, M., Ruhrgas AG., and HUMPHREYS, AE., British Gas plc.,GERG
I
Technical Monograph 5, 1991.
I Figure 4. Deviation in Predicted Density Between the NX-19 Model and the Detail
Characterization Method for Ekofisk Gas.
Figure 5. Percent Molar Composition of Natural Gas from the Statfjord and
I Veslefrikk Fields.
Figure 1. Percent molar composition ofPVT Reference data for Natural Gas Mixtures.
0.0 '~x:x"x;P u u u v
I
0 X X
X x xDx X 0 +
0
~
I I I I I I I I I I I
~.1
I 0.0 2.0
GROSSYETHOD
4.0 6.0
EKOFISK(RG21
B.O 10.0
275 K
12.0
0.1
I i-
_00000 000 a 0 a 000
a
a
x
0.0 .~ ,., n .....,
t'
'x x x ~x x
o o
~
~.1L- __~
I __~I __~I __~I __~I __-L
I __ -LI __~ I __~ I __~ I__~~~ I
I ..Q
.....o
.....a:s
0.0 2.0 4.0
DETAILYETHOD
6.0
0.1r---~~~~~~~--------~~~~~~--------~~
B.O
EKOFISK(RG2)
10.0 12.0
300 K
I o
>
Q) -
0.0 _~"'- 0 0 0 ;
I ~.1L- __ L-
I
t-
"XO~
I
__ L-__L-
I
oX
I 0.0
GROSSYETHOD
2.0
Ie
0
Xo 0 0
t-
+ a+ 0
o 0 000 o
00 ,., ,., ,., a
o w
x <e o 'b
I .
-
~.1L- __ L-~
I
-- ..xo~
I
0
Pressure (MPa)
I + Te:ns A&W Pycnometer Data x Tezu AAcWBurnett Data
II
I
I
Temperature (K)
.-
250 275 300 325
I
350
r-~~~-+~-L-L~~~~~~~~~~~12
I
1600
I
1400
I
,.,
ci
I
I
1200
tIJ
-..
....rn= 1000
I
...
Q
I
0...
I
2
-
oI I
I
600
I
~
I
II
400 I
I
200
I
I
-10 15 4-0 65 90 11
5 140 165
Temperature (F) I
Figure 3. Deviation in Predicted Density between the Gross Characterization
Method and the Detail Characterization Method for Ekofisk Gas.
II
I
I
.. Temperature (K)
I 250 275 300 325 350
12
I 1600 11
I .,
,
c;;
, 10
I 1400 d,
I .,
9
1200 ,
t'
0
8
I
-....
ctj ., -
ctj
I
-
~
rn
Q)
1000
-
I
?'
d
I
7 ~
6
-
~
(1)
'rn"'
'rn"'
;:j
800
;:j
I rn
Q)
()'
., 5
fIl
(1)
~'"' I
'"'
~
I 600
4
...
Ie
J
i'~ 3
400
I ~':> ,
/
2
I 200 ..0'
1
I ..0,\
I 0
-10 15 40 65 90 115 140 165
0
Temperature (F)
I Figure 4, Deviation in Predicted Densi~ Between the NX-19 Model and the Detail
Characterization Method for kofisk Gas.
II
I
I
eI
STATFJORD
FIELD
VESLEFRIKK
FIELD
I
CI4 Methane 73.21 66.5 I
N2 Nitrogen 0.65 1.22
Figure 5. Percent molar composition of Natural Gas from the Statfjord and
Veslefrikk Fields.
I
I
I
J
I
I
I
I
I
II
I
I
..
I 250 275
Temperature
300
(K)
325
I
I 1600 11
I 1400
10
I 9
tI 1200
8
I -
.....~m
-
1000
c,
I Q)
'"'
I ~
sn
Q)
800
I '"'
p..,
Ie
I 400
I 2
200
I 1
I O~~~~nT~Tn~~~~~~~O
-10 15 40 65 90 115 140 165
I Temperature (F)
I
I
-.
1200
.... 1000
cc
rT.l
-
I
--I
I=l.
I
I
600
I
J
400
I
200
I
I
-10 15 40 65 90 115 140 165
I
Temperature (F)
I
Figure 7. Deviation in Predicted Density between the NX19 Model and the Detail
Characterization Method for Statfjord Gas. II
I
I
.. Temperature (K)
I
I
1600
I' 11
I 1400
10
I 9
t' 1200
8
I -....
~
rn 1000
c,
I ...
CI)
I ~
rn
...
CI)
800
p..
I 600
4
Ie 3
I 400
I 2
200
__
-----0.05
I 1
I o~~~~~~~~~~~~~~o
-10 1 5
TeDlperature (F)
I
II Figure 8. Deviation in Predicted Density between the Gross Characterization
Method and the Detail Characterization Method for Veslefrikk Gas.
I
I
TeDaperature (K)
.-
250 275 300 325 350
I
~~-L~~+-~~-L~~~~-L~~T-r12
I
1600
I
,
1400
I
I
-
.....t'Il
1200
'j
--I
--tJl
P..
1000
- I
I
600
I
J
400 I
I
200
I
I
15 40 65 90 115 140 165
Tennperature (F) I
Figure 9. Deviation in Predicted Density between the NX-19 Model and the Detail
Characterization Method for Veslefrikk Gas. II
I
I
..
I
I DENSITY ME'l'ERIN:; INSTALlATIOO METlmS
I
I by
I J Gray
peak Measurement Limited, Sarasota
it
I
I
Paper 4.3
I
I
Ie roRTH SFA F'f.JJfI ME'AStJRElolENl WJRKSOOP
I
I
I NEL, East Kilbride, Glasgow
I
II
I
I
--I Jim Gray
DENSITY METERING INSTALLATION METHODS
I
SUMMARY
I The paper concentrates on density meters which utilize the well
established technique of a vibrating element to continuously
determine the density of a fluid. A review of 2 primary types
I of element and 3 methods of installation are used to highlight
the benefits of each type and method together with some of the
problem areas. The intention of the paper is to help alleviate
I problems in new metering systems and provide guidelines on
trouble shooting existing measurement difficulties.
Ie
This causes the spool to vibrate and this movement is
detected by the pick up coil and the resulting signal is
amplified and supplied back to the drive coil. The spool
is therefore maintained in oscillation by this feedback
I circuit.
The spool vibrates in a hoop mode and this is shown in this
I section through the spool. Obviously this is very much
magnified for clarity and the actual movement is very small
indeed. This vibration is of the same type that you get if
you rub your finger around the rim of a wine glass.If the
I wine glass is full it will give a different note from that
it gives when empty. This differing frequency of vibration
also occurs in the density meter and the spool vibrational
I frequency varies with the density of the fluid surrounding
it.
I
II - 1 -
I
I
As can be seen from FIGURE 2, the advantage of this
approach is the fluid is present on both sides of the
vibrating element (which is called the spool). This means
that there is no differential pressure across the thin wall
--I
of the spool and therefore the spool is not stressed by
increasing pressure. The body of the instrument is merely
a pressure vessel in which the spool is mounted and this I
makes the instrument suitable for operation at high
pressures. As mentioned previously, the mode of vibration
of a spool is circumferential hoop mode. This is shown
diagrammatically on the left hand side of FIGURE 3. The
I
vibration is always mechanically balanced so that there is
no reaction on the point where the spool is attached to the
body assembly.
I
If we look at a different mode of vibration as illustrated
on the right of FIGURE 3, we can see the second type of I
--I
vibrating element. Here we have a longer tube that is
clamped rigidly at each end. The tube is caused to
vibrate in a transverse mode, i.e. the centre of the tube
is deflected from side to side. This causes minimal
shearing of the fluid and an instrument based on this
principle is thus unaffected by the viscosity of the fluid
passing through the tube. All the fluid in the tube is
forced to take part in the vibration and the measurement is
then one of the bulk or average density of the instantaneous
sample. This means that non-homogeneous fluids such as
slurries can be measured with this technique. By sealing
I
the outside of the tube from the process we can magnetically
drive the tube without worrying about the corrosion
resistance of the magnetic materials as they need not be in
I
contact with the process fluid. Thus we can use a 316
stainless steel tube with magnetic armatures fixed to the
outside of the vibrating tube to give us a magnetically
driven density meter with the corrosion resistance of 316
I
J
stainless steel. One disadvantage of using this method is
that the vibration is no longer dynamically balanced; there
is a net reaction on the clamps at each end as the tube is
deflected from its rest position.
To look at the practical implications of a density meter I
using a tube in transverse vibration, as we have just
discussed, we must firstly provide a massive clamp at each
end of the vibrating tube section to define these points as
nodus points of vibration. This limits the energy transfer
I
from the vibrating tube to the holding structure by ensuring
there is no movement at the coupling points. This is shown
diagrammatically in the top illustration of FIGURE 4, where
I
we have a stiff frame welded on to the tube.
One disadvantage with this meter is with the central tube I
held rigidly when the temperature of the fluid passing
through the vibrating tube var ies a stress will be generated
in the vibrating tube as the clamping structure remains at
ambient temperature.
I
- 2 - II
I
I
--I A method of compensating for this effect is to make the
frame a part of the fluid path through the instrument. This
is shown here where the fluid flows through the instrument
in one continuous path. The top and bottom tubes are made
with a thicker wall than the central vibrating element,
The design brief for the new transducer was to make a HIGH
I ACCURACY WITH LONG TERM STABILITY meter. We explored the
performance of a whole range of possible ways of making a
II - 3 -
I
I
L.3 INSTALLATION METHODS
.-
The 3 basic installation options of density measurement
currently used worldwide today are:- 'IN-LINE' 'OFF-LINE'
I
and 'ON-LINE' as shown in FIGURE s. The 3 titles are taken
from the IP Petroleum Measurement Manual Part VII Density
section 2 continuous Density Measurement and broadly defined
I
as follows:-
Density Meter, IN-LINE - A density meter in which the I
transducer is located
directly within the main
line or vessel and measures
continuously. No sampling
I
system is required.
Having def ined the methods we can now consider the key
aspects and examples of each method. All three methods are
I
used on gas applications, generally only IN-LINE and OFF-
LINE are used for liquid applications. I
GAS APPLICATIONS
IN-LINE GAS measurement should always be used when the tI
highest accuracy of measurement is the prime factor. The
Direct Insertion Density Meter is still probably the most
accurate gas density measurement installation available
I
today as it measures true In-Line density with a high
degree of immunity to gas borne dirt and moisture and no
potentia 1 of pressure or temperature gradients; factors
I
which are often overlooked when the user is making an
assessItlentof an installation's desired accuracy.
examples of this are to be found in the rapidly expanding
Good
- 4 - II
I
I
--I OFF LINE
reasons:
1
GAS measurement is normally used for
Ie
ON-LINE
In-Line and Off-Line methods, whilst the insertion Density
Meter will always be the ultimate in overall accuracy terms.
The Pocket Density Meter has the same transducer
I calibration accuracy capability. On applications where the
temperature changes of the product in the main line are
relatively small and fluctuations do not occur
I instantaneously, then this accuracy can be reflected in the
overall installation performance. An ideal application for
this method would be a natural gas pipeline where the change
in temperature of the gas is only influenced by ambient
I temperature.
FIGURE 6 is an overview of the most common configurations
I of installations used for gas applications.
III - 5 -
I
I
"Gl" (FIGURE 7) is a typical OFF-LINE fuel gas. We have
started with one of the most difficult system applications.
This is used where the gas composition can be anything from
Hydrogen to C6 plus heavy ends. In the "REAL WORLD" it will
often be dirty, corrosive (sour) and "wet". Gas
--I
applications normally use the short cylinder (spool) type
Selection of material is important as Ni-span c
element.
is not suited to sour gas with H2S present. I
To a first order, this .type of element does not work on "wet
gas". However, "wet gas" should be better defined as gas
with liquid droplets. A vibrating cylinder element on gas
I
service will not work if liquid droplets are present on the
element. This is recognisable in the field as a very
erratic output caused by the liquid droplet rolling up and
down the element.
I
Two methods have been used to reduce this problem. A
combination of cyclone and coalescing filters together with
I
--I
a heat tracing technique, usually in the form of an
electrical self regulating system as steam tracing is often
not available.
It is very difficult to achieve a totally successful design
on this type of application from "best estimate composition
data often from a design process engineer for a platform yet
to be built. However, many successful systems have been
custom designed and used mainly where the measurement
engineer has been able to obtain real composition data on I
an established platform or plant. Heat tracing, where the
product is maintained at a temperature above the lowest dew
point value is the most successful of the two methods. Some
systems built by analyzer companies, with limited experience
I
on density measurement, appear to work satisfactory due to
the removal of the heavy ends as well as the dirt and water.
The result is a non-representative clean dry light ends only
I
sample.
Accurate quick response, low thermal mass, temperature
thermowells are a critical component for this type of J
application to correct to reference or line conditions. In
many cases, we have to design and build our own, due to the
low volume throughput dictated by the conditioning system.
I
Attention to any pressure reduction is also needed as this
can create more liquid formation. Short well lagged impulse
pipe work increases the potential performance of this
I
measurement as well.
The combined cooperation and experience of the user and the
supplier is the key to this application.
I
I
I
- 6 -
II
I
I
--I "G2" (FIGURE 8) is a typical IN-LINE density meter with a
retractor mechanism for removal under line conditions. As
mentioned previously this is the most accurate method of
measurement of gas density. The basic design has been
available for many years, however. A number of developments
have occurred more recently. In the "REAL WORLD" it could
I be said that there is no such thing as a totally clean fluid
on a platform or in a pipeline, therefore any direct
insertion density meter must have some protection from dirt
I pulling vacuum for more than half an hour as oil within the
pump can back stream into the density meter.
The paper, "Experimental Evaluation Of Densitometers In The
I Presence Of Condensation Or "Wet Gas"" by Dr S Kostic, Dr
T M Svartas and G staurland from the Rogaland Research
I
Institute presented at the 8th North Sea Flow Measurement
Workshop in 1990, identified that the direct insertion
density meter recovered significantly faster than the pocket
density meter after an injection of "wet gas". In general,
most gas density meters are subjected to occasional liquid
carryover. If the gas is continuously wet then only "G1"
should be considered.
II - 7 -
I
I
Direct insertion density meters with their unique inherent
accuracy can be used, particulary as most modern fiscal
metering stations now use two transducers with back up PTZ
.-
calculation to qualify the transducer's status.
I
Lube oil mist down-stream of a compressor on Natural gas
pipelines can cause problems which are difficult to identify
without PTZ back up calculation. I
Unlike other liquid carryover which is easily identified
by erratic performance, lube oil mist can form a very fine
deposit on the element not visible to the eye. Dual
I
density meters have been seen to be more than 2% off
specification but still within 0.2% agreement. Where ever
possible on new metering installations, it is best to avoid
locations immediately down-stream of compressors. Hopefully
I
in the future there will be a filter which can totally
remove this. A dimension of how far down-stream this type
of mist becomes relatively harmless droplets should be
I
identified.
When used with a retractor mechanism another option is
available to the user to improve performance should there
be an excessive frequency and volume of liquid carryover.
As we have already mentioned, the latest sampling technique
--I
is similar to what occurs in the chimney when wind passes
over the tip. This draws the sample from the base of the
probe. The instrument therefore suffers no loss of response
time if the inlet is positioned in the pipe stub away from
I
the contamination. Furthermore, in extreme cases, heat
tracing can be applied to the pipe stub to ensure the carry
over stays in the vapour phase. I
The final comment for this type of installation is
applicable to all installations of density meter. When it
humanly possible, ensure the density meter is kept off-line
I
or isolated until 24 hours after start up. Flow computers
etc can be given "fall back" values to get the system
running. More damage is caused to density transducers in ~
this time frame than the rest of the instrument's life time.
I When the line size is below lOOmm (4") we recommend the use
of lOOmm (4") equal tee with eccentric reducers to suit the
actual line size. Problems have occurred with turbines etc
it when concentric reducers are used due to the pipe work
"trough" collecting dirt/liquid and eventually causing slug
flow, when there is a significant change in flow rate. From
I
I
A range of wall thickness on the pockets selectable on the
basis of maximum design/operating pressure, ensures thermal
mass of the pocket is kept to a minimum, enabling the
quickest possible response to a change in the main pipeline
temperature.
--I
When using a class 900 lb pocket, a 5 degree centigrade
change in temperature could typically take approximately 20
minutes before equilibrium between the mainline and the
I
measuring element is restored. This aspect was more
extensively covered by Mr Reidar Sakariassen from Statoil
in his paper Installation Details For Gas Densitometers at
I
the 9th North Sea Flow Measurement Workshop.
Finally on the construction The most important feature of I
any Pocket Density Meter is an integral PT100. Based on
years of experience in IN-LINE and OFF-LINE density and flow
measurement we have proven that a custom built integral I
PTI00 unit is a mandatory requirement for any accurate form
of density meter installation. It ensures that there is no
temperature gradient error between the precise point of
density and temperature measurement within the transducer.
..
Furthermore, on installations operating at extreme
temperatures, it allows the user to monitor, correct and/or
alarm on any potential temperature differentials between the
I
point of density measurement and the main line. Often where
the user is using the density meter as a component of a mass
flow meter ing system, errors due to temperature differential
I
can cause significant offset in the overall system accuracy.
II - 11 -
I
I
TABLE 1. Differences in pressure and temperature
a change in liquid density of 0.03 per cent.
that will each cause --I
STABILIZED
Density
CRUDE OIL
O.850g/mI
I
Temperature coefficient
Pressure coefficient
O.0007g/mIoC
O.00007g/mI/bar I.
Therefore
Maximum temperature difference
Maximum pressure difference
O.4C
4 bar I
*LIQUID BUTANE AT OC
I
Density
Temperature coefficient
Pressure coefficient
O.580g/mI
O.OOllg/mIoC
0.00025g/mI/bar
Irt,
Therefore
Maximum Temperature difference 0.16C
1.2 bar
I:
.'I
Maximum Pressure difference
*LlQUID PROPANE AT OC
Density 0.520g/mI
Temperature coefficient 0.OOI5g/mIoC
Pressure coefficient
Therefore
Maximum Temperature difference
O.0003g/mI/bar
0.1OC
I
Maximum Pressure difference 1.0 bar
GASOLINE J
Density
Temperature coefficient
0.660g/mI
0.00075g/mIoC
I
Pressure coefficient
Therefore
O.00019g/ml/bar
I
Maximum Temperature difference 0.26C
Maximum Pressure difference 1.58 bar
I
.. NOTE: The above values are specific to the conditions quoted and
change dramatically around the critical region.
I
I
. 12 -
II
I
I
-. From this table it can be seen that a 1 degree C difference
in temperature between the point of flow measurement will
I generate 0.3% of reading error on propane and almost ~
of reading error on butane, making an IN-LINE density meter
essential for these 2 liquids if 0.1% of reading is to be
II - 13 -
I
pulsation can be the same as the operating frequency
or a harmonic resonance of the vibrating element. This
can cause an unstable output and under extreme
conditions an offset in the performance. 180 degrees
of pipe bends will normally eradicate this.
With dual density meters on a typical fiscal metering
station the same effect can occur between the 2 density I
meters if they are operated close together in series.
This is not normally a problem as the conventional
installation method is to operate them in parallel on
identical pipe work configurations to avoid different
I
thermal gradients and maintain operation if one unit
is removed.
On a few occasions the parallel installations can show
I
C)
a small bias. The installation of a small volume
I
..
header appears to resolve this effect.
.,
has a large thermal expansion coefficient. Due to the type
of vibrating element this method should not be used when the
viscosity of the liquid exceeds 20 centipoise or the
location of the measurement is in a pigged line.
II - 15 -
I
I
"L3" FIGURE 19 shows the pitot tube scoop method. The
response time and thermal and pressure gradient will change
.a.
with flow rate and condition of product. Therefore this
method is only suitable when the span of the flow rate is
known to generate sufficient differential pressure. Each I
application will require specific design based on product
composition, line size and flow rate. Under low flow
conditions, stratification of density and "vapour locks"
can occur in the by-pass pipe work.
"L4" FIGURE 20 shows the use of a main pipeline restriction
to generate a flow around the by-pass pipe work. The
operating characteristics are similar to the pi tot tube
scoop method. Addi tional care is needed to ensure the
potential downstream gas bubbles do not adversely effect
any other measurement devices. An advantage of this method
I
can be the ability to fine tune the system on site by use
of a partially closed valve as the restriction in the main
pipeline. Downstream flow rate reduction must also be
I
reviewed when installing this method of installation onto
an existing process plant or pipe line.
"LS" FIGURE 21 is a pipe work configuration which has been
successfully used on 50 to 100mm (2" to 4") diameter lines.
Dependant on the flow rate and product condition etc the
ratio of pipe diameters of the two lines can be varied.
--I,
With the J valves shown, flow rate, back pressure and
isolation for maintenance can be achieved. Using the "pipe
splitter" shown, this installation has been particularly I
'I.
successful on applications where the J previous OFF-LINE
installations can have problems in achieving a
representative by-pass sample of a non homogeneous liquid.
I SUMMARY ('
I REFERENCES
"I 3.
Dr T M Svartas and G Staurland from the Rogaland
Research Institute presented at the 8th North Sea Flow
Measurement Workshop in 1990.
Installation Details For Gas Densitometers at the 9th
North Sea Flow' Measurement Workshop, by Mr Reidar
Sakariassen from Statoil.
~ "REAL WORLD" at many previous workshops by Mr Brian
4.
- 17 -
I
=Sarasola
PRINCIPLE OF OPERATION
AMPLIFIER
- ~~
",
/"
---
- -- ~,
... -- .... ,,
,
,,\
I /
I I
II
I
\
,\,
I I
\ \ I I
\\
,,
/,
", "
.... - --
PICK-UP COIL
HOOP MODE
VIBRATION
MAGNETIC CYLINDER
_ .8 _
- .. ...- afigure
~
1
~,--- .&..
- - - .. --- - ... -, ... - '
FD700
BODY
FLUID FLO..;;IJ~ .~
COIL
Figure 2
.~Sarasota
PRIME VIBRATION PRINCIPLES
,,. ~- --
."- .... '-
I '
':
I I
,,
I I
,,
I \
,' - - ----- --
, .:.:::::<
CLAMP CLAMP
~""~I
\~" ~,,;~I TRANSVERSE MODE
'~=:" OF VIBRATION
HOOP MODE
VIBRATION
Figure 3
_a,~ ~ ~ a_
- ,- - - - -tr -.- - .' Figure 4
- ---
- - -
FD800 SERIES
-
I
-
VURA TJNG TUII:
-
FD810 - 850 MODELS /
MANIFOLD
.-- r--
..
-I-
I,-
-f'--I-
,....
I- I-
I- 1-
"'" '"L-
--
2 VIIIRA TJNG TUJI(S
l-
____t-
I-
,.. ,..
i-
'"
r-I- -
I-
'-
..
-
'-- FD860 MODEL L-
=5arasota Figure 5
I I
--
IUICIIr
I
I
SAMPLE FLO'.'
I
ra.tDt
i~~~~
: :
I
I I
II
II I n:- ......
,,, I
I
fLO'.' I fLO."
80-----1
-a-- ~_~~~~-----a-
-'-----r----'-----_=Sarasota
Figure 6
Gl
._----- _. __ :.::-==----
.._- -=:-:-:: ..
G3 G8
G2
G6
__
._...
_....
_~~~---!L..-
_L ...... ,,_ ........
- ..-.- .. ..1 .... . ..- .--...
-. ..l
L~-~
.._... .....__ G4
GT
:J~~~"~~-,:.=
-:.8:"':::::"-_: .::____=_ -::~:_~: __ .. .._~.:
..._:=._.._. --.lI__
bJ=~(-,
5
0
ISOLATING VALVE
/"'- - mESSURE
// TRANSHITTER
CYCLONE ROTAHETfR
FILTER
PRESSURE
RE(jIJLATOR
- - - - - -'- - - - '.
OFF-LINE Gas Filter System
- #
~
I
.. =Sarasota
I
I
I
I
I
"I
I
I'
"I
a.I
I
I
Figure 8
I IN-LINE Direct Insertion Density Meter
With High Pressure Retractor
II
I
I
= Sarasota
Figure 9 I
I
I
I
mzz?V27W2Z2V?Zl2ZaZ22Z72ZZZ22ZZViZZV?Z22amvmmmzadWZzvzmzmzmzz
I
I
I
~
I
Ffow-- __
I
I
High sensitivity PTIOO
thermowell
I
Flow Path Perfo/mance of an
I
IN-LINE Direct Insertion
Density Meter II
I
I
--I =Sarasota
I
1<
I
I
It
I
I
I'
I
Ie FlDW ntROUOH ow _
.-FLOW
I
I
I Figure 10
I,
II
I
I
=Sarasota
I
I
I
l"NPT HOLE
IHliO' FLANGEI
&. TAPPED.
I
1
--I'
I
I
I
FLOW THROUGH OM -. - ~
I
:::J FLOW
I
I
Figure 11
OH. LAGGING
\
\
.A
\\ }..
!t---------- --------...,j-!
5d
- - ----- -
,
-
--~'-
\-!-- ------'
(Otf'lECTIONS
10 - ZOv
1- --- ,
,
1
GAS
, ,
Z OUTPUT
] 0'11 EARTH)
Z PIATIN~
X R5ISANC[ I
Y THERMOMETER
W IIF FInED) rnl I 111m /
-- f)-" ,,
\
Figure 15
-'- - ..
- ,-----,---- ,-- - Figure 16
-::L.IsOia
TYPICAL METHODS OF LIQUID INSTALLATION
Aa recommended by the I.P. Petroleum Measurement Manual Part VII - Density
- -
TIMIBIN
MET(R
VALVES
ROil rlOIi -
'OFF-LINE' 'IN-LINE'
=Sarasota
Figure 17
L3 J2 L4
---~Jt\J mt - L-~
L9
L7
E:-:l
_N __
'< -"('-- ft;!/ r;8~-'\"-
-- -~_~--~~------ -----
_~ ~ ~ a_
-------~
.---- FULL BORE ISOLATING VALVES
FLOW ~
,----_.
FLOW ~
_a ~ ~ a_
-,-----, .. -- - ,- - ----
;;Sarasota
- Figure 20
/DENSITY METER
1-----1 / /
---j--r-- ___=T];~/[=L __ --ln~-~-~-:-~-:------~-
,-- --------
1 -U
\- FULL BORE ISOLATING VALVES
__ 1,-,--1 _ __ _ _ ____ _ __ JL _
------------
FLOW ~
RESTRICTION DEVICE
~- -'- ~- .&
I
I
=Sarasota
I
I
I
I
,.
I
I
I
I
'-I
FLOW TIOUGH 011 _
"'f1.0W
I
Figure 22
I
IN-LINE Direct Insertion Liquid Density Meter
I With Welded Flange
I
It
I
I
Figure 23
=Sarasota .-
I
I
I.
I
t I
I
\ .
--I
\ \ :
I
\UJ I
I
~
I
Flow
I
I
OFF-LINE density meter I
utilising the differential pressure
of main-stream bends
I
tI
I
I
I
A NEil !':ULTI -BEA!-i ULTRASONIC FLOillCETER FOR GAS
I
I by
I
I A Lygre, CHI, R Sakariassen, Statoil
and D AIdaI, Fluenta
"I
I Paper No 5.1
I
I
..
I
HORTH SEA FLOil NEASUREHENT
26-29 October 1992
wORKSHOP
I
I
I NEL, East Kilbride, Glasgow
I
,.
I
10'" North Sea Flow Measurement Wotbhop
CONTENTS Page
1 INTRODUCTION 1
2 MEASUREMENT PRINCIPLE 2
3 SYSTEM DESCRIP'IlON 4
6
SPECIAL FEA 1URES
PROVING
6.1
6.2
6.3
Zero calibration
Flow calibration
Self diagnostics
14
15
15
16
16
6.4 On-site zero calibration 18
7. CONCLUSIONS 18
8. ACKNOWLEDGEMENTS 19
REFERENCES
Offshore metering Slations based on orifice plates are bulky and require much space. Accordingly, the platfonn costs
are considerable, and can be significantly lowered if the size of the metering station is reduced.
Ultrasonic lJallsit time multi-path meters will allow compact metering stations to be constructed due to increased
now metercapacity and reduced iDstallatiml lengths. In addition these meters offer improved Dow meter performance
and potmtial for simpler maintenance procedures.
A new mlllti-patIl flow meter for gas bas been developed by Christian Michelsen Research in Norway. The Dow
meter has mdergooe testing on nacuraI gas at K-Lab, Norway. The Dow meter (FMU 7(0) will be manufactured by
Flllenta AS.
The FMU 700 features new technical solutions such as titanium housed u1trasooic transducers, automatic gain
control, DI1-line measurement of transit time delay in cables and electronics, and software pulse detection.
The deviations between the K-Iab sonic-nozzle reference mass metering system and the FMU 700 are less than 0.8%
at gas velocities between 1 and 8 m/s. These results were achieved with IOD straight inlet pipe downstream of a
90" bend Tests wele also petformed with the FMU 700 installed only SD downstream a 90 bend At gas velocities
between 2 and 8 m/s the me&'lurement uncertainty is not changed despite the reduction in upstream straight pipe
ICIIgth from 100 to SO. The observed deviations were independent of the test pressure varying from SS to 100 bar.
The test results show that it is quite feasible to build a very compact and light metering Slation compared to
conventional solutiorls and at the same time comply with the requirements set to fiscal metering stations.
The foDowing procedure is suggested for proving of a multi-path meter in a fiscal metering station :
- On-site zero calibration of individual transducer pairs which are removed from the pipe line and insta11ed
in a zero calibration cell
10'" Nonh Sea Flow Measurement Wmbhop 1
1 INTRODUCTION
Offshore meu:ring of II8IUrlIl gas bas been, SlId stilJ is, based 00 the cxifice plate. This technology is proven and well
known both 10 the opeIlIIDI'S and the authorities. However, metering stalions based 011 orifice plales are buIky and
require much space. Accordingly, the platform COSIS are considezable, and can be significantly lowered if the size
of the metering sralion is reduced [1.2].
Swoil, as 8 major gas producec, bas been coocaued with reducing the COSIS of offshore metering of gas [1]. Based
on the general devdopnent within electronics SlId sensor technology dming die SO's, it became clear that multi-path
ultrasonic flow meters represented 8 realistic alternative 10 oriIx:e plates, Specirlcally, multi-path meters would allow
compact metering SI8lions 10 be constructed due 10 increased flow meier capacity and reduced installation lengths.
In addition Ihesc meters offered !he advantages of :
low uncertainty,
no moving parts,
no pressure loss,
rapid response,
potential for omitting flow calibration,
self-cbecldng possibilities,
reduced maintenance.
Thus, it was recognized that introduction of ultrasonic meters would both reduce costs and improve flow meter
performance. In 8 study carried out by SWOil, savings of 100-150 mill.NOK were estimated if offshore metering
stations were based on ultrasonic rather than orifice meters [1].
On this basis Swoil and 0Iristian Michelsen Research (CMR) launched a project in 1988 with the objective of
developing a 12" multi-path meier. The projecl was funded by Statoil. The design phase was successfully concluded
by the end of 1990 and it was decided 10 build and lest a 12" prototype fiscaJ metering system (FMU 700). The
prototype and testing phase of die development was funded by Swoil and Fluenta, a subsidiary of CMR. The FMU-
project was carried out jointly by Statoil, CMR, Fluenta and Kongsberg Offshore (!COS).
CMR has been active within ultrasonic flow meter technology for more than a decade [3,4], and proposed as early
as in 1981 to develop a multi-path gas flow meter.
These projects served as a technology basis for the FMU-projecl where CMR developed the ultrasonic transducers,
the geometrical arrangement of !he sensors 011 the spool piece, the hazardous and safe area e1ectronics, the flow
computer solution and software as well as the signal processing technique.
The projecl was coordinated by Fluenta which also will manufacture !he FMU 700 flow meter.
KOS developed design solutions for metering stations based on multi-path meters and provided the secondaJy spool
piece carrying temperature, pressure and density sensors.
The meier was tested on II8IUrlIl gas from mid November 91 10 early February 92 at K-l.ab which is a high-pressure
flow calibratioo facility located at KArst0, Norway. K-l.ab is 8 joint venture between Total and StalDil and is
operated by SIalOil
In this paper we describe the results of the FMU-projecl focusing on the flow meter concept, a discussion of some
of the test results and fmaJly a presentation of procedures for proving of multi-path ultrasonic gas flow meters.
z ~UREMENT
Basic rormulas
InFigme 1 a siJIgIe-paIh uItraSOOic flow metez is ilIusIraIed with two uluasooic transducers facing each other at an
oblique angle to the pipe axis. The individual upsIre8IIl (t,,) and doWJlSlJTam(~J transit times are given by [6]
(2)
and
(3)
where
v = Axial flow velocity averaged along a chord D which is the projection of L in 8 plane perpendicular to
the pipe axis., see Fig. I,
c = Speed of sound in !he fluid averaged along the chord D,
L = The portion of the intertransducer center line lying in the flowing fluid,
9 = Angle between the intertransc1ucer center line and a line parallel to the pipe axis,
It, = Downstream transit time, from transducer 1 to 2,
"" = Upstream 1raIISit time, from transducer 2 to 1.
We observe thai both the flow velocity v and the speed of sound c in the fluid are measured. Thus, the 1I8Ilsit time
flow meter also povides information on 8 physical property of the fluid.
In practice the transducers are often set back, i.e. the actual distance between the transducers is lsrger than L as e.g.
sho'oVll in Fig. 1. Acc:crdingly the measured transit times also incorporate the tranSit time in the cavity in front of
the tmnsdUerS. However, it is easy to implement 8 procedure in the flow canputer which allows the measured
transit time to be ceerected for the unwanted time delay in the uansducec cavity. For low Mach nwnber flows this
practical problem can also be solved as described in [7].
By measuring along five different acoustic paths across the pipe, the gas volume flow can be measured accurately
even when the flow profile is disIorted. FiJ!UIe 2 illustrates the positioning of the ten ultrasonlc transducers in the
FMU 700 u1tmscnic JIBSfIow rnetez. The measured velocities v represent averages along the parallel chords shown
in Fig. 2, i.e. the acoustic transit time technique in facts integrates the velocity profJle along the parallel chords. The
volwne flow is given by
-;
r
(4)
q .. I D(y)v(y)dy
-r
where
The multi-path mew measures v along a limited num~ of chords and the inlegra\ in Eq.(4) can be approximated
by
(5)
where
WI .. Weight faclOlS depending on the numerical inlegration technique applied in Eq. (5).
The geometrical configuration of the ultrasonic transducers. or the position of the parallel chords in Fig. 2. therefore
depends on the numerical inlegration technique which is applied.
Transducer 2
Flow v
Transducer 1
Figure 1 lliUSU'8tion of the principle of a single-path ultrasonic transit time flow meier.
TRANSDUCER POSITIONING
End view
A
A 0.2.4
AA
Figure 2 Transducer positioning in the five-path FMU 700 ultrasonic gas-flow meier.
1C1" Nm1h Sea Flow Measurement WOIbhop 4
3 SYSTEM DF:SClUPTION
The flow meier masures Wllumc Dow, now velocities and speed of sound avenged along parallel cItonIs, see
Figs. 1-2. Mass flow can be computed provided the density is made available to the flow computer.
The FMU 70() fivc-palh ulll'aSODic gas flow meier consists of, see rig. 3 :
A cabiDel coolainiDg a CODpuIer and an electronics unit,
Two sigDaI cables and one qJIical cable,
lntrinsica1ly safe eie(:tronics,
10 1itaniwn1lotJsed ulII'aSODic uansducers,
A flanged spool piece.
A secondary spool cmying tempenllUre. pressure and density sensors. can be installed downstream of !he flow
meter. 1f required, the signals from the secondary sensors can be received and converted to physical values by tile
flow computer.
The flow canputer can store all measured data and diagnostic parameters and transfer !he intormation OIl digital
fonnat Ie an exttmal computer via a series communication link (RS232).
SpeciCicatiollS
Tbe design temperature mnge for the pipe work. e.g. -46 to + 105 "C, will normally exceed the operational
tempe:nd:tUe range. However the flow meter spool will comply with the requirements set for the pipe wodt. But as
yet, the flow meter is not designed to operate over the entire pipe wlrl design temperature range.
At K-Lab tile flow meier was tested down to 20 bar and no change of the flow meier performance were observed.
This indicates !hat the pressure mnge can be extended below SO bar.
1D5tallatioll requirements
The Dow meter was designed to operate wi!h 10D of Slraight pipe upstream of the meter spool and 3D straigbt pipe
downstream of the meter spool The tota1 installation length amounts to appro 16D. The total instal1ation length does
IIOt change if the 3D downstream spool is equipped wi!h a tbermowell or an intrusive densitaneter, i.e. if tile
downstream spool acts as a secondary spool
For a bi-directional mstallatioo, the total insta1lation length will be 23D without a secondary spool and 26D with
a secondary spool.
The tests at KLab indicate that downstream of a 90" bend, tile upsIream length may be reduced to SO. In !his case
the installation lengths reduce to II D for an installation with a fixed flow direction. For a bi-directional installation
the total installation lengths may reduce to 13D(no secondary spool) and 16D(including secondary spool)
'.
Flow computer
HI'" North Sea Row Measmemenl Wmkshop
The Dow c:ampula' is based OIl an iDdusIrial PC widl keyboard and color graphics screen, The computer COIlIl'OIs
the entire IIICIISIIIaIIeDtprocess in real time IKXOIding ID a pre-sct measurement procedure stored in file during the
5
configmaliQII of the meier. Jnstructions to the hazardous area electronics from the flow computer is cransmined via
an optical cable ID ensure fast and reliable transmission of the control paramClCl'S. The sensor signa1s are ttansmined
via two cables between the conII'Ol room and the Dow meter.
The operator can Qll\y get access ID the flow compuler by specifying the c:mect password. and the Dow computer
prognun can Qll\y be baited by specifying the correct password. In a practical measurement situation the keyboard
can be removed or locked to increase the security. Funhez, the flow computer operation will be made independent
of the bard disk by SlDring all programs in ROM. If the hard disk fails Ibis will not influence the meier performance.
The c:ampuler iniliales the transmissiQII of ultrasonic puJses and then reads a iep s e ntalioo of die received pulse inID
the computer's memory in real time. In a mcasmernent cycle each of the 10 InIiISduccrs act once as a ttansmitter
and once as a receiver and 10 pulse n:presentations are reconIed during die cycle. Based on die 10 recorded pulses,
10 transit times are computed n:presenling a single sample of die volume flow. When volume flow samples have
been acquired over a user specified time inICI'val, 10 mean times-of-flight arc computed in software from die
individual ttansit limes recorded during die interval.
From die 10 calculated times-of-flight the flow velocities and speed of sound along each of the five acoustic paths
are calculated. The volume flow is calculated by integrating the flow velocities across me pipe profile. The mean
flow velocity in me pipe, mass flow, total volume and mass are men calculated along with statistical data. Readings
are displayed on die computer screen and are sent in digital format ID e.g. the computer in me fiscal measurement
station.
There is no additional microprocessor in the system except for me standard processor installed in me PC.
1
HAZARDOUS AREA: 1
SAFE AREA
1
1
1
1
: Computer
1
1
1
1
Intrinsically safe
electronics
Test procedue
During !be tesIs, !be gas density in the test sectioo was calculated based 011 measured pressure, temperature and gas
ccmpositiou (AGA 8). The mass flow was compultld from !be FMU-metered volwne flow aDd !be calculatec1 density.
The FMU 700 was coofigwecllll average !be volwne flow. flow velocities and the speed of soond for each path over
a paied of 10see. The number of samples in a IOsec paiod is approximately 110. 1be flow meter readings were
c:ontiJwoosJy IIIIIISIIlitled III !be K-Lab c:anputer 011 digital format.
1be reference mass flew rate for a single comparison test was defined as a 300sec avemge n:ading of tile 1lOZ2Ies,
and the FMU 100000n:adings were averaged over tile same time intecval. During each run tile flow ronditioos were
kept as stable as pessible.
At each Y'ebcity, at a given pressure, 5 or usually 3 cooserotive nms were made.
The temperature during the IeSIS reported here varied between 36.7 and 37.9 "C. The velocity range was 0.4 10 8
mls which is the muimwn "iIlue in a 12" pipe at K-Lab. and tile pressure was set 10 55, 70 or 100 bar. 1be flow
meta was tested 100 and 5D downstream of a 90 bend. The 30 1000gseoondary spool was installed downstream
of the meter spool "flange to flange", carrying a pressure sensor and a thermowelL
The uncer1ainty of the K-lab reference mass metering system is estimated by X-lab to be 03%.
The demtiOlL'l between abe K1ab sonic-nozzle reference mass metering system and tIleFMU 700 are less than 0.8%
at gas velocities between 1 and 8 mls. see Figure 4. These results were achieved with 100 straight inlet pipe
downstream of a 9()0 bend. TesIS were also performed with the FMU 700 installed only SD doWDStream a 9(JO bend.
At gas vekx:ities between 2 and 8 m/s the measurement uncertainty is not changed despite the reduction in upstream
straight pipe length frun 100 to SD. From Figure 4 we can also see that the measurement uncertainty is independent
of )IJeSSlIre c:hanges. In Fag. 5 the individual 300sec readings are plotted III give an impression of the repeatability
(2a-UDCertainty) of tile meta mder test. As can be observed, the repeatability is satisfactory. see below ..
The smic oozzIe readings were also converltld III volume flow and compared with the volume flow measured by
the FMU meta. The observed Variability of the FMU and nozzle readings were quite similar.
It shooid be ooltld that abe test results referred III above. were achieved without using any calibration 0" meter factor
in the FMU 700 meter. 1be FMU 700 flow meter was zero.caJibraled independently of the
K1ab refereoce system and then installed at K-lab. see Section 6.
l-<J UDcertaiDt,
The 2-17 uncertaiJIty is defined here as 2a/(average reading) where a is the sample standard deviation of N flow
me&er n:adings recorded in B given time interval where tile flow rate is constant
An im)lOl'18Jlt pope<'1y of the FMU 700 flow meter is tile stability of the flow meter, see Figure S. Dming the tests
at K-lab it was demonstrated that the observed repeatability is comparable to a good turbine meter, which is
recognized as a very stable 8IId repeatable flow meter.
Table I displays 3 different estimates for tile 2auncenainty of the FMU 700 during the tests at Klab during a
period when the flow rate was particularly stable. It is important III be aware of that the ultrasonic flow meter
measures the twb11lent fluctlJalions of the flow and twbuIence will contribute to the 2auncenainty klgetha with
the CIttibution fiom the finlte resoIution of tile transit time measurement, The :la-uncertainty win decrease when
'.
I
die time avenging interval ina
111'"NorIh Sea Flow Measurement Wmbbop 7
es ......, .... die average reading gets c\o&a" and clc!su ID die true mean. and this
is also obsaval from Table I'. In Figure 6 some of die daIa in Table 1 me ploUed.
Table 1 2G-UllCalllinty for die FMU 700 based 00 150 mosec:utive flow meter ~gs. The 150 ~gs rqxesent
eitIa IOsec averages, l00sec moving average of IOsec readings or 300sec moving average of IOsec
readings During die lSOOsec time interval die flow conditions were kept as SIBbie as possible.
100. 100 BaIG. 2.S'II>Sonic nozzles. 0394 0560% 0.177'11> 0.102'11>
100. 100 BaIG. 6.2S'II>Sonic nozzles. 0.986 00512% 0.123'11> 0.046'11>
100. 100 BaIG. 13.75% Sonic nozzles. 2.168 O.2l!I% 0.074'11> 0.041%
100. 100 BaIG. 27.S'II>Sonic nozzles 4321 0237'11> 0.062'11> 0.034'11>
100. 100 BaIG. SO'lI> Sonic nozzles. 7.736 0204'11> 0.061'11> 0.023'11>
2216
4.400
7.890
0.458'11>
033S'II>
0218'11>
0.186'11>
0.207'11>
0.107'11>
0.065'11>
0.051%
0.074%
0.061 'II>
0.026'11>
0.016'11>
SO. 100 BaIG. 2.s'll> Sonic nozzles. 0384 0.773'11> 0.206'11> 0.102'11>
SO. 100 B81G. 62S'II> Sonic nozzles. 0.981 0.762'11> 0.234'11> 0.132'11>
SO. 100 B81G. 13.75'11>Sonic nozzles. 2.168 00597'11> 0.172'11> 0.083'11>
SO. 100 B81G. 2705'11>Sonic nozzles. 4313 0.487'11> 0.093'11> 0.043'11>
SO. 100 B81G. SO'lI> Sonic nozzles. 7.723 0.445'11> 0.152'11> 0.101 'II>
, Aa:ordingly. a comparison of die 2a-lDICC:rtainty between various flow meters is only meaningful if the
averaging intervals are sirniIar f(l' the various meters.
10'" North Sea Flow Measurement Workshop 8
FMU700 ~
(J.S
0.5 Natural Gas
K-LAB 19!12
(J.O
0.0
-0.5
-0.5
z;
0
1.0
~
-
<:
;.-
1.::1
Q
-1.0
1.5
-1.5
~
2.0
2.0 -0-- 100,100 Dar, 37 DegC
---f:r-- 100. 70 Bar, 37 DegC
3.0
3.0
4 5 6 7 8
0 2 3
FLOW VELOCITY (m/s)
Figu e 4 Calibration lest results at K-tab for FMU 700. ) 00 and 5D measurements.
FMU700
(J.S
(J.5 Natural Gas
K-LAIl19!12
(J.O
(J.O
o
0.5
(J.S
--
;Z
0
E- 1.0
...: I.(J
>
IOl
Q .1.5
I.S
tl!
3.0
3.(J
(J.()() 1.00 2.00 3.00 4.00 5.00 6.00 7.00 8.00
'.
-
~
;.-
-0-- 100. !flO B
!Z... 0.10
---0--- 100. '10B.
<:
f- ---1:1- 100. S5 SU
0.08
'"
~
U
;Z
;J 0.06
...e
0.04
0.02
0.00 6 7 8
0 2 345
FLOW VELOCITY (m/s)
Figure 6 Estimates of the 2CJ.uncenainty of the FMU 700 during the tests at Klab. The :2CJuncenainty
displayed here is based on a 300sec moving average of the IOsec average flow meter readings. At
K.Lab each comparison between the FMU 700 and the reference nozzles was based on an average
reading over 300sec.
The results show that it is quite feasible to build a very compact and light metering station compared to conventional
solutions and at the same time comply wi!h the requirements set to fiscal metering stations.
The flexibility of the flow meter may be utilized by e.g. calibration laboratOries to monitor the flow conditions as
shown in the following. In view of the good repeatability of !he flow meter. it should also be fully possible to use
it as a reference flow meter in calibration loops.
During continuous metering. measured data can be stored in a me by giving an appropriate command. Similarly.
the system can be set to measure only a single path, fOl' test, uouble shooting or calibration pwposes. Stored data
can be used to euminc earlier series of measurements by fetching data from file and displaying them on the screen.
This enables the operator to scan rapidly through the data to interesting areas of the measurement series.
Figure 7 is an example of such a stored time series from the tests at KIab. where an interesting part bas been ploUed
showing mean flow velocity and velocities aloog each of the five acoustic paths around an abrupt change of the flow
velocity. In Figure 8 the same incident is shown for the speed of sound along each of the five acoustic paths. The
mean flow velocity and the speed of sowId along one of the acoustic paths are ploued on top of one another in
Figure 9 to show the simultaneOUS change in boih measured values. This event also illusttates the flow meter
response the ability to resolve rapid changes of the Dow I1IIe.
In Figure 10 another part of the time series is p10ued showing the turbulent fluctuation in mean Dow velocity and
velocities aloog each of the five acoustic paths at a constant flow rate. Notice how the flow velocities along !he
acoustic palhs closest to the pipe wall. ()"9 and 45. display the highest turbulent fluctuation and how the flucwations
appear to inversely correlate. The same is the case for the mid-paths, 18 and 3-6. but with less turbulent fluctuation.
The center pa!h is less influenced by turbulence while the calculated mean now velocity is almost constaDL Ano!her
.'
intereSting 00servaIi0n is dial die IIJJbuIent f1uctua1ion along two paths 011die same side of the center path. r. ex.
0-9 and 1-8, also illversely c::orreIaIe while die turbulmt fluctuation aIoog twO asymmelric piths on differmt sides
Clf abe center JIBIII, e.g. paths ()'9 and 3-6. wrrelale. This c:orreIaIion and illvetSe correlation elJect between die
twbWent f1uctuatiOllS aloIlg diffamt acoustic piths is easia seen in Figure 11. It shows an extnlCl Clf die time series
from Fiame 10. with the flOlV velocity along the five paths plotted as a prof'lle. with time as parameter.
InFiame 12 the same pari or abe time series as in Figure 10 is plotted showing the fluctuation of the speed of sound
alollg each oi the five acoustic paths. No rapid fluctuations of the speed of sound are observed. The slow overall
f111C11la1ioooi the speed oi sound is probably due 10 the regulation sySlall 011the centrifugal compressor ciIt:uiating
Ibe gas arDIIIICI abe loop at K-lab.
DuriIIg test metering. the received uJtrasonic pulse representations can be sacred in file. Due to the brge amount of
data tbat a single uluasonic pulse .epresents. only short time series can be stored. The stored uluasonic pulses can
be examined later by a sepamte program which e.g. can produce various plots of single pulses as shown in Fig. 13
and time series of computed timeHlffJight as shown in Fig. 14. Individual pulses and transit times can also be
valuable tools for analyzing various flow phenomena.
Analyses as described above can provide information on the meter perfonnance and the flow conditions.
--36
-Mean
--45
--()'9
2.0
J.5~-.-,........ .........,....-.-........,......
2000 2100 2200
..............
,...,............,....-.-.-+ .........
2300 2400
...,....-t-"""""""''''''''''''''''''''I-r-''''''''''''-+''''''''''''''''.-i
2500
TIME (sec)
2600 2700 2800 2900 3000
Figure 7 Tests a1 K-Iab. Showing mean flow velocity and velocities along each of the five acoustic paths
around an abrupt change in flOlV velocity.
i ,
413.0
~
~
E 412.8
~
>-
!: 412.6
U
0 -09
..l 412.4
~
... 412.2 ...... 1-8 ..
!:l --2-7
~
o 412.0 "" .. "" 3-6
til --4-5
411.8
411.6
2000 2100 2200 2300 2400 2500 2600 2700 2800 2900 3000
TIME (sec)
FigureS Tests at K-Iab. Showing speed of sound along each of the five acoustic paths around the abrupt
change of the flow velocity displayed in Fig_7_
~
..
~
U 3.5
4.0
413.0
~
~
412.8 E
~
412.6 Eo-
>-
-
U
o 412.4 0
..l
~
... 3.0
412_2 !:l
...
~
::
02.5 - Mean flow velocity
z
;:J
412.0
0
Ii -- ~9.Sound velocity til
2.0 411.8
Figure 9 Tests at K-Iab. The mean flow velocity and the speed of sound along one of the acoustic paths are
ploued on top of one another to show the simultaneous change of both measured values,
10'" North Sea Flow Measurement Worbbop 12
3.7
1000 1100
- .. - - 2.7
1200
------ \-8
1300
--
1400
36 -
1500
TIME (sec)
Mean --
1600
4-5
1700
--
1800
0-9
1900
Figure l() jests at KIab. Showing !he turbulent fluctuation in mean flow velocity and velocities along each
of the five acoustic paths at a constant flow rate. Notice !he ccnelation and inverse correlation
effect between the turbulent fluctuations along different acoustic paths.
"0-9 11
Figure 11 Tests 81Klab. Showing an extract of the time series from Figure 10. with !he flow velocity along
the five paths plotted as a profile. with time as parameter. Notice the correlation and invene
correlation effect between the turbulent fluctualioos along different acoustic paths.
10" Nonh Sea Flow Measurement Workshop 13
413.05
~
.!!?
,5 413.00
~
U 412.95
0
..l
~ 412.90
Q
~ 412.85
0
en --09 1.8 -- 27 ..... 36 -- 45
412.80
412.75
1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
1000
TIME (sec)
Figure 12 Tests at K-Iab. Showing the fluctuations of the speed of sound along each of the five acoustic
paths.
m
p
1
22.
'92
'60
'28
I)'
/
"""-
\\
u 9.
I 1\
0
6 f \
32
0
700
j
TIME MEASUREMENT
I
750 800
Time
850
[mt c r c se c I
900
'"'" 950
Pelh no: D, Seen no: 10
1000
Figure 13 Tests at K-Iab. Received uluasonic pulse representation showing peak amplitude and negative
going zero.crossing time for each period of the received pulse.
10'" North Sea Flow Measurement Wmb""" 14
m 2.32 2.32
e
:2.0 J 2.03
1.7<4 1 . 7""1
m
1. <45
1." 5
e 1. 16
1. Hi
r
0 0.87 0.87
O. SB 0.58
e
c O. 29 0.29
O. DO O. DO
Figure 14 Test.s at K-lab. Time series and disnibution of computed times-of-flight f()f a single path at 8 mls.
s SPECIAL FEATURES
In the following we will highlight some of the technical acltievements which are of particular importance 10 the
users. PaniculaTly the FMU 700 features :
Titanium bousecl ultrasonic transducers which eliminates the risk of gas penetration into the transducer and
subseqllCnt tJIIIlSduc failure.
Automatic pin control keeping the amplitude of the received ulu-asonic pulses at aconstant level. This makes
it possible for tile flow meter to operate over a wide temperature. pressure and flow velocity range without any
manual adjustment of the electronics, If automatic gain control is lacking, the flow meter will cease to opezate
if e.g. the p-eSSUl'e changes.
On.Une measuremenl of trllll5lt time delay in cables and electronics is implemented and accounted f()f in the
transit lime measurement. In most ultrasOnic flow meters fixed values for tIlese time delays must be implemented
in the Row computet and drift in the ll'allSit time delay will not be accounted for.
Software pulv detection eliminating electronics for pulse detection and making it possible to implement a signal
processing tecltnique wbicb is able to recognize pulses modulated by turbulence effects, pressure and temperature
changes. Accordingly the flow meter reliability is improved.
The gcomearical amngement of the ultrasonic transducers and the COII'tSpOIIdingintegration technique, see Eq. (5).
was developed as result of extensive numerical simulations where various methods were investigated. Based on set
of 34 flow profiles an adaptive and robust integration method was developed particularly designed to integrate
asymmetric flow profiles.
The flow computer concept, based on an indusnial PC. has several advantages since this 5Olutiooresults in :
'.
'. J
A less complex Dow _
)(,.. Not1h Sea Flow Measumnent Wmbhop
elecuonics,
Easy aa:css III flow m~ rawdaIa which makes seMce and 1eSIing simple.
Simplc proc:ccIuIa for upgrading flow canputer software. c.g. diagnostic software,
Usc of SI8IIdard equipment whicb is under continllOUS technological improvement,
15
The teclmical solutions implemented in the FMU 700 flow _. thus lepi sent 8 general improvement of the
u1uasonic Dow _ technology and will beIp to improve the flow m~ n:liability,
6 PROVING
An important aspect of the introduction of multi-path ultrasonic flow metecs for fiscal or allocation measurement
of gas. is the proving procedure for the meter, In the following we will discuss some poSsibilities and propose 8
practical solution as to bow proving can be carried out.
The following procedure is suggested for proving of 8 multi-path meter in 8 fiscal metering station:
In addition to the above items it is fully possible to increase the redundancy by installing two meters in series. Since
the flow meters are non intrusive they can be installed close to eacb other. potentially "flange to flange".
In the following we will describe the proposed procedure with reference to the experimental experience gained
during the testing of the FMU 700.
The basic mC8SUlCd parameter is the acoustic uansit time in the flowing gas. see Eq.(2). However. the measured
transit times also contain the transit times in the acoustic transducers and the accompanying electronics. Thus. the
measured transit time must be corrected for these time delays. It is of particular importance to enSUIC that the flow
meter Jeading is zero wben the gas it at rest, i.e. the transit times in the gas in each diJection along the same path
must be identical at zero flow. It is easy to measure the transit time diffen:nce accurately. But, the absolute time
delay is more diffICult to measure, Accordingly the transit time difference at zero flow is meBSUICd precisely and
the less 8CC1D'8le measurement of the absolute time delays are adjusted to secure zero measured flow in 8 gas at rest,
When the time delays and the transit time differences have been measUJed they are stored in the flow computer.
During flow m~ openuion. the measured transit times are corrected prior to calculation of the flow velocity by
Eq.(2).
The FMU 700 was zero calibJated by pressurizing the spool piece using nitrogen. Before pressurizing the spool
piece. the distances between the acoustic transducels were measured with an uncertainty of O.04mm. To eDSUICstable
conditions. i.e. no tbemJaI flows. the pressurized spool piece was kept in 8 temper8lUJe bath during calibration. The
transit times were then measured over the pressure and temper8lU1C range in question. Based on the transit times.
the uansit time difference comctions were calculated Using the mC8SUlCdtransducer disIances. com:cted far thermal
expansioo. the time delays and the speed of sound in the gas were calculated. For control JlUiPOSCS the measured
speed of sound was compared to calculated values using the llJPAC-18bles[5]. The deviations between the measured
and the calculated values were less than 0.5 mIs. i.e. 0.15%. The measured "zero calibration times" were
subsequently stored in the configuration file of the flow computer.
At pJeSCDt wc are working on B new method for zero calibration which may represent 8 significant
and simplification of the above proceduJe. The new method will soon to be tested at CMR.
improvement
.'
The zero caJiIraIioo ensures !bat die inherent offset in \be measured transit times can be IIIXXlWItedfor when
calculating \be velocity using Eq.(2). This offset is indepeDdent of \be flow ~Iocity. Due \0 the simple reIatioIlship
betw-. abc Dow velocity and \be transit times gi~ by Eq.(2). where calibration CXlIlSI8nIS are IlOl required, it
shouleillOl be DeC: "y to carry out a flow calibralion. When die uansit times are measured c:ateC1Iy, Eq.(2) will
provide die flow velocity without abc use of additiona1 calibratioo CODSIaDIS.
As describecI ~ abc FMU 700 W&'I zero calibrated 011 nitrogen prior to traDspClIIIIIiOll and installation at K-Lab.
The IICOIIStiC tnmsdng:rs were DOt removed from the spool piece after:zero calibration. The meter was zero ca1ibrated
at 31 'C (oormal opeIllIing tempendUre at KLab) and then traJJSp<Kted \0 K-Lab. During the transportation the meter
was el<JlOliCd\0 tempemtures arooncI O"C and tbeD installed at KLab and warmed lIP \0 37 'C &pin. During the
test period \be meter was exposed \0 tempezature cycling between oormal qJClIIIiDg tempezature and ambient
tempendUre (0 'C) ooseveml occasjnns. The observed deviations sbown i rig. 4 were obtained without any
adjustment of die now meier after zero c:alibratioo at CMR. The meter was simply installed and the tests begun, It
is of oonsidemble JIlICtical importance \0 notice that the zero calibration was carried out without the long signal
cables (approx. 130m) which were used at K-Lab. This is possible because the FMU 700 features on-line
measurement of abc transit time delay in signal cables, safe area e1ectrmics and pans of the intrinsic:ally safe
electronics.
During abc tesI periocI we did not observe any detectable drift in the zero calibration. This verifies that it is not
necessary 10 carry ()UI a flow calibration if the meter is propecly zero calibrated.
However, for fiscal purposes it is likely 10 assume that a flow calibration will be required,
Ultrasonic flow meters offer \be possibility 10 monitor the fl()w meter performance \0 a certain extenL This can be
ulilized both as an indicator of when proving is necessary and to reduce the wOlk IoacI connected \0 inspection and
malntenance, If an abnonnal situation occurs a message will be written \0 the system log in the flow computer 811d/or
a warning can be given 10 the operator depending on the nature ()f the detected error.
In \be FMU 700 meter the following meier performance parameters can be monitored continuously :
Comparisons ()f the measured speed of sound along the five acoustic paths,
Transducer failwe,
Transit time enor.
Frequency mooulati()n ()f pulse,
Failme \0 recognize received pulse,
Measured venus calculated speed of sound,
Standard deviation of the speed of sound measurements.
Standard deviation of the flow measuremenL
BasecI on the above diagnostic paramelerS ultrasonic flow meters can detect meier malfunctioning and give the
operaIDr a warning in contrast 10 e.g, an orifice meter where it is nearly impossible \0 detect a change in the meter
penormanc:e.
ComparisclJlS or souDd speeds
If the PJOputies of the acoustic transducen or the electrmics change with time this may lead 10 a drift in the zero
cal.ibratioo, and the flow meter uncertainty is likely \0 increase. II is particularly important to awiel drift in the
measured vansit time difference in Eq.(2). Drift in the absolute transit times is far less important
In a multi-path meter the speed of SDUDdin the gas is measured for each path. If the pipeline is propecIy insnl8\C'd,
the temperatwe difference between the area close 10 the pipe wall and the central portion of the pipe will be very
small (if any). CcJnsequently the speed of sound will be constant across the pipe section and the measured speed of
sound aloog die paIbs sbould be equal. F~ very low flow velocities and large temperature difference between die
gas and die surroundings, a small temperaWre gmdient may be presenL This was in fact observed at K-Lab.
During die K-Lab tests, die difference belweeII die maximum and die minimum values of die mEwed speed of
sounds were typically less !ban 0.07 m/s. (0.01%). even during abrupt cbaDges of die Dow velocity. see FIgS. 8 and
12. This difference is due 10 die measW'CIIIeDt1DICCltaintyof die transit times (DOl the transit time differences) ~
inhomogeneities in die flow.
If !his difference "ceeds O.lm/s. i.e. a change of O.04m/s. !his may represent a drift in die transit time difference
measurement of lOOns. At Sm/s a lOOns drift corresponds to a shift in die c:alibration of die meter of around 1%.
Thus. tbeR is a pocential f~ delec:ting a change in die calibration of the meter by continuously monitoring the
difference of die measured sound speeds.
However, die speed of sound is popw1ionallO die sum of die transit times and DOlthe transit time difference.
Accordingly. changes in the measured speed of sound can also occur even if the transit time difference is unaffected.
At CMR more wodc will be undertaken 10 establislt a procedure for using the speed of sound difference as a
diagnostic parameter.
Based on measured temperature and pressure and a specifIed gas composition. the speed of sound can be calculated.
Comparing the measured and calculated speed of sound, can be used as a rough. but independent, check of the transit
time measurements.
Transducer railure
If one of the acoustic transducers fail and is unable 10 emit ~ receive an acoustic pulse. !his will be detected
intmediately and a proper warning will be given. If a transducer pair drops out, the flow meier is still able 10
measure die flow by using the transducer pairs in operation to estimate the velocity along the path which has
dropped OUL
Nonphysical transit times. i.e. transit times which cannot occur based on the known distance between the acoustic
transducers and known upper and lowec limits for the speed of sound. will nevec be measured due 10 the time gating
Pulw recognitioo
The percentage of pulses recognized by the flow computer is monitored continuously and will be stored in the
system log. Normally 100 % of die pulses are recognized by the flow computer. During transducer malfunctioning.
heavy turbulence. electric or acoustic noise. die percentage of pulses accepted may be low and a warning will be
given.
Every accepted pulse is checked for "period error". i.e. if for some reason a pulse period is missing. If a "period
error" is found the pulse is rejected.
The frequency content of every recognized pulse is checked and Slrict limits are set for the allowed variation of the
pulse frequency around die known frequency of die emitted pulse in ordec 10 ensure a high-quality transit time
measuremenL If these limits are violated die pulse is rejected and the Dumber of rejected pulses is monitored and
will be stored in die system log. If the Dumber of rejected pulses is high. a proper warning will be given.
FilteriDg
When the transit times have been ooniputed a filtering algorithm cbel:ks the measured set of tmnsit times for outliers.
This filteriD& is based on the lIIClmIured transit time distributions. Tnnsit limes lying oulSide the allowed spreading
are discarded from !be data set which is used when cak:uJating the flow velocity.
StaDdard deviatioa
The standaJd derialioo of the measured now velocity and die measured speed of sound are calculaled for each path,
see Fiame 6. 1be standaJd deviation can be used as a meier performance parametez as well as an iDdical of die
stability of the flow.
Ideal1y, it should be possible to carry out zero.calibnltion of the Row meter on-line, i.c. when the meta" is installed
in Che pipe line. This implies !hat the Row must be bypassed and it must be possible to keep the gas in the spool
piece at absolute rest at a constant temperature. In a practical measurement situation this is not easily obIained.
However. "inline" zero calibration may be a future possibility.
By removing !be meta" from the pipe line and insta1ling the spool in a zero calibration facility. This requires the
now to be bypassed in addition to Ihe mechanical WOJk needed 10 remove die meta" spool.
By removing a single transducer pair and die corresponding elec1ronics and carry out zero calibration of roly
one transducer pair at a time in a special zero calibration facility. In this case it is not necessary to bypass Ihe
flow and Ihe metec will be able to operate at a slightly reduced uncertainty.
The latter method is by far Ihe best melhod from an operational point of view. The acoustic transducers can be
removed from the pressurized pipe line either by insta1ling pennanent ball valves at each transducer port or by using
an extraCtor tool which can be moved from one transducer poll to another. BoIh techniques are being considered
for the FMU 700.
The zero calibration facility will be a pressure cell which should be immersed in a temperature bath and pressurized
with nitrogen. ldea11y the cell should resemblance the meter spool as much as possible. This is imPortant in order
to ensure !hat the SOIDIddiffraction effects in the Row meier and in the calibration cell are as similar as possible.
This may be of importance for the measurement of the transit time delay. After zero calibration the transducers and
the electronics are reinstalled, and the transit time delays in the Row computer are changed if necessary.
In order for such a procedure to w!Xk it is important that the distance between the transduceIs is unaffected by the
dismounting and reinsta1Iation of the transducers. At CMR the distance between the transducers was measured before
the sensas were removed and after they had been reinstalled. The distance did not change mere than O.03mm ,and
this is we]) below !be acceptable limit, i.e. O.lmm.
Since the FMU 700 measures transit time delay in signal cables, safe area electronics and the inUinsically safe
electronics on-line, this zero calIbration method is particularly atlJaCtive for Chis meter.
7 CONCLUSIONS
A new multi-path ultrasonic flow meter for gas has been developed and tested on natura1 gas.
The development bas contnbuted to major technical acbieventents wilhin the ultrasonic gas now meter teChnology
S1lCh as titanium housed ultrasonic transducers, automatic gain control. software pulse detection and on-line
measurement of transit time delay in electronics and signal cables. These achievements represent a significant
improvement of the now meter reliability.
..
The _ results show that it is quite felWble 10 build a verJ CClIIIpact and light metering station compared 10
c:onventiolla1 solutions and at the same time axnply widl the requirements set to fisca1 metering stations.
8 ACKNOWLEDGEMENTS
The authors woold like 10 express their \banks to Statoil and Flueota Cor the opportunity 10pub1isb the project results
and for making the FMU-project felWble by providing the necessary financial support.
The authors acknowledge the high-qua1ity WOlle carried out by the rest of the CMR-project team ronsisIing of Reidar
~.IW Lunde. Aile 1obannessen. Magne Vestrbeim. Gunnar Wedvich. Hans IngebrigUen. Stig Heggstad and Roald
Toftevaag.
Further we woold like 10 thank Framo Services and Framo Engineering for doing an excellent job on bod! design
and fabrication of the spool piece and Kongs1lezg OffsIDe for providing the secoodaJy instrumentation.
Fma11y we appreciated the service and the endlusiasm shown by die K-Lab staff dwing the test period.
REFERENCES
[I] Sakariassen R. :"Development of a new gas metering system". Paper presented at "Gas Transport
Symposium". Haugesund-Norway.1anuary 30-31. 1989. Norwegian Petroleum Society.
[2] Hannisdal :"Metering study 10 reduce topside space and weight". Paper presented at the "91h Nord! Sea
Flow Measurement Workshop. 1991" in Bergen. Norway. October 22-24. 1991. Norwegian Society of
Chartered Engineers.
[3] Mylvaganam K.S. :"Ultrasonic flowmeters measure flare gas in the Nonh Sea". Oil&Gas Journal, 17
October. 1988.
[4] Folkestad T. and Mylvaganam K.S. :"Acoustic measurements detect sand in the Nord! Sea flow lines".
OiI&Gas 1oumaJ, 27 August. 1990.
[5] IUPAC :"Niuogen Intemational Thennodynamic Tables of die Fluid State - 6". International Union of Pure
and Applied Chemistry. Chemical data series no. 20. Pergamon Press. 1979.
[6] McCartney M.L .Courtney P.M. and livengood R.D. :"A corrected ray dleory for acoustic velocimeuy".
1ouma1 of Acoustical Society of America, Vol. 65. No. 1.1anuary 1979.
[7] Daniel Industries :"Multipadl ullrasonic fIowmetec for gas CUSIOdy InIIISfcc". Sales brochure. Daniel
Industries, Scotland.
I
II
I
I
I EXPERIEN::E WITH a:MPARATIVE TESTIN; AND CALIBRATICN OF CORIOLIS
AND TURBINE METER OFFSHORE AND IN THE IAOORATORY
I
by
I
"I L Mandrup-Jensen
Force Institute
I
I
Paper 5.2
I
~
I
N:>RTII SEA F'IJ:M MEASUREMENl' OORKSHOP
I 26-29 ~ 1992
I
I
I NEL, East Kilbride, Glasgow
"I
I 1
I
Table of contents
I
, 1. Introduction..............................................................................
page
I 2.2 1 Turbinemeter.............................
2.2.2 Coriolis mass flowmeter ...........
5
6
I
I
I
I
" "... '''Ih'1i'"',...''''....,..... p. ~,. 92-09-."
"Experience with comparatiYe testing and calibratioD of Coriolis and turbine metfr off-sbore aod in. the laboratory
l.. I:ntroduction
I
I
I
I
I
til
=
1992 _ Nonh Sea no,., Meas~ Wod:abop ~bb Hydro 2628 moocr
I
I p. 3
I
"Experieoce with comparative testing and cab"bratioD or Coriolis and turbine meter off-shore and in the laboratory
FORCE
INST:TUTcS
Mcuo1oo' DivisiaJ
"
I
Turbine meters are used to measure the amount of condensate and
water. The two meters are connected in parallell (tk1, tk2 and
tv1, tv2 , see the figure) and each cover a different flow range.
An orifice meter (or) is used to measure the gas.
I
gas
I mixture
af water,
I
gas 09
condensate weir ~ test separator
..
I "choke" valves
-,
I ILIII
IIIII
I
I Figure 2.1. 1 Schematic diagram of the test separator on Tyra West
"I 199'2 North Su Flov.- Mcasuremcnl Worbhop feeble. Hydro 26 -28 october
p. 4
"Exper-ience with comparative lestin~ und culibratien of Coriolis and turbine meter off-shore and in the Iaboratory
p. 5
"Experieoce with comparative testing and calibration of Coriolis and tuibine meter off..more and in the laboratory
FORCE
I INSTITUTES
Melrokv DMllial
I and the flare gas metering system. Some selected data for the
turbine meter is:
, Model, sin
Size
Accuracy
Model F/0.75"/ 30,
3
3/4 ", 0.8 - 8 m /hour
Linearity
Repeatability,
S/N: 50242
, spec. 0.25 %
spec. 0.02 to 0.05 %
The turbine meter has last been calibrated (with water) 22.11.88
I by Hydril,
factor:
with following calibration
K = 433.24 pulses/litre
results Average
& linearity ~ 0.29 %
K-
I 1)
2)
GROSS VOL.
NET. VOL.
Qt.,g
Qt.,D
[m3/h]
[m3/h]
Vt.,g
Vt.,D
[m3]
[m3]
2) MASS [kg/h] M t.,m [kg]
I Re.1)
Qt.,m
I where f = pulses
K = 433.24
per. secund from the turbine meter
pulses/liter, dr= integr./summ. time
;
I where Ct.l/C = temperature/pressure corrections factor
dr pe" = Integra t"Ion / summa t"Ion tIme
" "
I Re. 3) Mass is the total mass: flow rate (Qt.,m)and the accumulated
mass (Mt.,m),the parameters are calcuated as follows:
I Qt.,m= Qt.,g " density ; Mt.,m= l: Qt.,g " 640 " dr
"I
!
p. 6
"uperietKe with comparatiwt" testing and calibration of Coriolis and turbine meter otr~ore and in the laboratory
The signal from the sensor of the mass flowmeter is send via
I
cabel to the transmitter, from which the frequency (and
milliampere-) signal is transmitted to the Digital Rate
Totalizer, DRT (and manual terminal, 268) in the control room.
I
A shematic
2.2.2.1.
diagram of the connections can be seen in figure I
I
<- <- flow
~
mass trans-
flow 220 V m;tter ~
sensor
tocess/~tc _____
____ =_______
ORT
286
I
Figur 2.2.2.1 : Electrical connections of the mass flowmeter
I
Some selected data for the mass flowmeter is:
Manufacturer:
Sensor
Micro
Model
Motion
08065 S , sIn : ~36894
II
Size
Transmitter
:
DN 15
in EX
mm , Qmax = 136 kg/minute(8~60 kg/hour)
box in proces/plant I
DRT : Model FM8-3
Model, sin
Size
:
Model F/O.75"1 30,
3
SIN: 50242
3/4 ", 0.8 - 8 m /hour I
Accuracy Spec. 0.2 % of rate zero stability
where zero stability = 0.84 kg/hour
I
The mass flowmeter is latest calibrated 91-11-16 (with water) by
Rosemount, with following results :
I
Flow calibration factor = 20.5205.13
Density calibration factor = 12550~42754.44
I
1992. M_~n1
NMh Sea FlClOltl Worbbop Ptebla Hydro 2628 october
--I
I
p. 7
"Experience with comparative test:ing and calibration of Coriulis 8Pd turbine meter off-shore and in the laboratory
FORCE
I INSTITUTES
Mctro1ocY Divitkn
I 3.1 Method
Me Mt,m . (Dc/640)
I Fe-t =
Mt,m . (Dc/640)
. 100 (%) [3.1.1]
'-I
Me
Mt,k = the turbine meter's display of total mass.
Dc = density (kg/ltr), measured by the mass flowmeter
other relevant parameters have also been noted, e.g. the process
temperature and the process pressure. Also the temperature and
I equation [3.1.2].
I
p, 8
"Experience with c{)mparative te;ting and calibration or Coriolis and turbine meter off--shore and io the laboratory
[3.1.3]
I
The uncertainties of Me og De is estimated in the
following: I
The display of the mass flowmeter (Me) is "trigged" (manually
read) at the same time as the shift in the display of the turbine
meter (Mt). which is updated with every 4 kg. The display of the
I
mass flowmeter is updated with every 1 kg. On this basis the
uncertainty of Me og Mt
estimated to be:
on a 95 % confidence level. is
I
(0.52 + 0.52)1/2 ~ 0.7 kg [3.1.4]
~
( 12+ 1 2)1/2 ~ 1.4 kg [3.1.5]
[3.1.7]
I
As the value of De is approximately 650 kg/m3 following values
for the uncertainty of Fe_t as a function of Me can be given:
fI
I
Mc [kg] 100 500 1000 2000 00
--I
I p, 9
I
"Experieuce with comparative testing and calibration of Coriolis and turbine meter off-shore aod iII the laboratory
FORCE
INSTITUTES
MctroJocy 0iviJia:1
3.2 Results
I The test / calibrations of the coriolis meter versus the turbine
meter has been performed simultaneously with the platforms normal
I "6 hours test" on the following wells : TW BOB, TW B12 og TW C02.
The table given in Figure 3.2.1 gives an extraction of the
results.
I Me [kg]
--
date well flow Dev. denSi~ Xm , S, n Omin - Qrnax
[kg/h] [X] [kg/ ] X X [kg/h] u( Fe' t) [X]
[mld/y] I.d.
I 1/26/92
1/26/92
1/26/92
TW
TW
TW
812
812
812
288.0
300.0
302.0
8.98
9.95
9.89
666.8
667.0
666.5
666.9
9.49
s = 0.50
(n = 4)
288 . 320 = 100 . 200
=1.60.8
1/26/92 TW 812 302.1 9.13
"
1/26/92 TW 812 970.9 1.75 665.9
1/26/92 TW 812 1,012.5 1.37 665.8 1.39 971 . 1083 = 300 1000
1/26/92 TW 812 1,062.0 1.27 665.9 s = 0.25 = 0.5 0.2
1/26/92 TW 812 1,082.9 1.18 665.9 (n = 4)
.73 665.3
I 1/25/92 TW 88 1,584.0
1/25/92 TW 88 1,664.5 1.10 663.5
1/25/92 TW 88 1,783.6 .82 665.3
1/25/92 TW 88 1,815.0 .81 664.0
1/25/92 TW 88 1,835.6 .84 665.0
1/25/92 TW 88 1,842.9 .90 665.6
-
1/25/92 TW 88 , ,864.4 .89 665.3
1/25/92 TW 88 1,876.9 .93 664.0
1/26/92 TW 812 1,877.0 .83 665.7
1/25/92 TW 88 1,886.9 .88 663.5 0.88 1584 2180 = 10001500
1/25/92 TW 88 1,929.4 .80 663.9 s = 0.18 =0.20.1
I 1/26/92
1/26/92
1/26/92
1/26/92
TW
TW
TW
TW
812
812
812
812
1,938.2
1,940.0
1,947.9
1,979.0
1.23
.77
1.09
.96
666.2
668.3
665.9
665.7
(n = 22)
I
1/26/92
1/26/92 TW 812 2,038.7 .69 665.8
1/25/92 TW 88 2,042.0 .73 664.3
1/26/92 TW 812 2,042.1 1.31 666.1
1/26/92 TW 812 2,180.0 .55 665.9
'-I -
1/27/92 TW CO2 3,050.8 .59 645.0
1/27/92 TW CO2 3,054.2 .48 644.8
1/27/92 TW CO2 3,099.3 .39 645.4
1/27/92 TW CO2 3,105.0 .41 645.5
1/27/92 TW CO2 3,108.0 .69 644.1 0.53 3051 . 3369 = 15003000
1/27/92 TW CO2 3,143.6 .55 645.6 s = 0.12 =0.1'0.1
1/27/92 TW CO2 3,154.3 .76 643.8 (n = 11)
1/27/92 TW CO2 3,246.8 .54 644.7
1/27/92 TW CO2 3,290.0 .58 645.0
1/27/92 TW CO2 3,322.5 .35 646.1
1/27/92 TW CO2 3,369.2 .54 645.2
I Fl.gure 3.2.1 : Extraction of the results from the cal. on Tyra west.
I - flow
- Dev. :
Mean flow, mass divided by measured time
Deviation between the Coriolis- and the turbine
I - density
-XJn
meter (Fc-t)
density measured by the coriolis meter
mean value of the deviation for measurements
within a flow range
I -
-
s
Qmin/Qmax:
standard deviation for n measurements
min. and max. values of flow for n measurements
- Mc nominel amount measured for one test.
"
- u(Fc-t) Uncertainty on Fc-t
I
1992. North Sea F""'" Meu~n1 Worbhop - }>eeble. Hydro26lB O<:tob:t
p. 10
"Expeeieece with eemparative testing and calibratioD of Coriolis and turbiDe meter orr-shore and in the laboratory
FORCE
IN~~!
The values listed in Figure 3.2.1 are shown grafically in Figure
3.2.2 which represents an "error curve". I
I
12 Erro r Ie-11'U'\ 7. Ilensi-t .
~W1 kg! m3 675
I
.
10 670
8
~ .... .~~
O'ao (}
665 I
\
Co
2
\
-,
~ rO
=CO
0 ~
660
6S5
--I
o o. ~ &45
-2
o 100J 200J
610
1000
I
Ilov , (kg/hourI
p. 11
"Experience with comparative testing and calibration or Coriolis and turbine meter off-shore and io the laboratory
FORCE
I INSTITUTES
I 4. Calibration in laboratory
I 4.1 Method
The plan was that the metering section consisting of the turbine
I meter and the mass flowmeter, was to be calibrated against a
weighing system.
"
I
But as the turbine meter was damaged when it arrived to the
laboratory (probably during the transportation off-shore to the
laboratory), calibration of this meter was not possible in the
laboratory.
where: Fe-y
=
=
100
I
I
"
p, ~2
"Experieoce with compasanve testing and calibratioD of Coriolis and turbine meter off-shore and iD the laboratory
'.2 Itesults
The mean value of the error in the flow range of 223 to appro
II
2600 kg/hour is -0.42 % with an unlinearity of t 0.11 \, which
at least regarding the unlinearity is within the specifications.
I
Comparing these results with the off-shore calibration results
(see figure 3.2.~ and figure 3.2.2) it can be seen that it was I
the turbine meter that was in error in the low flow range. This
is in accordance with theory and practise of the calibration
curve of a "normal" turbine meter. I
Besides the above calibrations the mass flowmeter has been
calibrated using pulsating flow. The flow was turned up and down I
every 10 seconds a total of ten times during a calibration. 5
repeat measurements have been performed, in which the mean value
of the error was measured to be -0.48 % with a standard deviation I
--I
of 0.02 % Therefore no significant difference from the
calibration with non-pulsating flow.
I p. 13
II "Experieace with comparative testing and calibration of Coriolis and turbine meter off.more and in the laboratory
FORCE
INSTITUTES
I
s. Conclusion
I The Coriolis meter has been tested off-shore against a turbine
meter, and acceptable results were obtained. The turbine meter
I and the Coriolis meter compare well down to 1000 kg/hour. In the
flow range under 1000 kg/hour the deviation between the meters
grow as the error of the turbine meter grows. This is to be
I expected from a normal turbine meter.
I
,.
a weighing system and there the meter has shown good repea-
tability but an off-set of approximately -0.42 % . The Coriolis
meter was found not to be significantly sensitive to pulsations.
Benefits:
I Good measurement uncertainties in a large measuring
range.
Small sensitivity towards pulsations
I Measures the density and temperature of the fluid
continuously
will not be damaged by particles in contrast to turbine
I meter
-.I Non-benefits:
Safety risk in the case of fatigue fracture, as the fluid
then will flow uncontrolled (in contrast to turbine meter)
Sensitive to air in the fluid
possible sensitivity to vibrations in the fundament
"I 1992 North Su Fk- M_~ Worbbop Peeblcl Hydro 2ti28 octob:,
I
III
I
I
I CCMPJ\CT LARGE OORE DIRECl' MASS FlCklMETERS
I
by
I
, A J Matthews and C L Ayling
SChlUlllbergerIndustries
I
I
I
Paper 5.3
I
-.I
NJRTH SEA F'IJJiJ MEASUREMENl' mRKSOOP
I 26-29 ~ 1992
I
I
I NEL, Fast Kilbride, Glasgow
I
I
I
COMPACT LARGE BORE DIRECT MASS FLOW METERS
I
I A. J. Matthews and C. L. Ayling
I
, SUMMARY
I Although suitable for the measurement of hydrocarbons, few large bore direct
mass flow meters are available, and these are generally limited in their
I application by their large size.
This paper presents a compact direct mass flow meter that uses a novel
I concept to sense the flow. A resonating tuning fork is used which enables the
size to be kept similar to that of a turbine meter for bore sizes of 4 inch and
above. The meter provides outputs of mass flow, density and temperature. Its
I naturally rugged design offers very high pressure containment.
'-I collaboration with Statoil. Tests results from Norsk Hydro, Porsgrunn are
presented showing repeatability of 0.05 % and linearity of 0.25 % for a 4
inch bore meter at flow rates up to 350 tonnes/hr.
I
I
I
I
II
I Page 1
I
CONTENTS II
1.0 Introduction
I
2.0 Meter Design
I
2.1 Theory of operation
I
2.2 Pick-up and drive mechanism
I
I
I
I
I
tI
Page 2
I
I
It 1.0 INTRODUCTION
I Direct mass flow meters are rapidly being accepted as a good method of
measuring mass flow rate to custody transfer standards. Meter sizes are
I available from a variety of manufacturers in sizes from 3 mm bore up to 6
inch. All of these meters work on the principle of passing the fluid to be
measured through the inside of one or two resonating tubes. These tubes can
I be configured in straight or bent configuration dependent on the
manufacturer. Only one meter is currently available in 6 inch bore and this
has a shipping weight of 636 Kg and is 1m x 0.3m x 2m in size. This generic
I physical size, which creates difficulties both for the manufacturer and the
user, seems to be the major limitation on these meters.
I
, For the Schlumberger Industries single straight tube design the problem was
even more severe. In order to achieve the required pressure ratings
combined with suitable mass flow sensitivity, a 4 inch meter would be 3
meters in length and weigh 300 Kg and a 6 inch meter would be 4.5 meters in
length and weigh 1100 Kg. This was believed to be impractical and so an
I connected with passing the fluid through a resonating tube, this meter turns
the problem inside out and inserts the resonating sensor into the flow. This
means that the pressure rating in not limited by the sensor itself and the mass
I flow sensitivity can be adjusted as desired.
The meter takes the form of a stretched tuning fork either cast or wire eroded
'-I
from a solid billet of stainless steel. The configuration is shown in fig 2.0.1.
I
I
I
I Fig 2.0.1 Mass Meter Tine Configuration
It
I Page 3
I
2.1 THEORY OF OPERATION
The operation of this meter is similar in theory to that of the tube type Coriolis
-I
meters, although the mathematics is a little more complicated. The tube wall
defines a volume of fluid that acts on the resonant tines. The resonant
frequency of the tines is dependent on the combined resonant mass of the I
tines and the fluid surrounding them. Thus the resonant frequency is
dependant on the fluid density, which enables an accurate density output
(O.S Kg/m3) to be computed. As fluid flow occurs Coriolis forces are I
generated which distort the tines' resonant mode shape. The magnitude of
this distortion is proportional to the mass flow rate of the fluid flowing past the
tines. This distortion can be detected as a phase shift in detectors mounted at I
each end of the tines.
I
It should be remembered that the tine movement is only micrometers in
amplitude, and hence cannot be seen. All diagrams of the tine mode shapes
are grossly magnified.
ta
I
I
Resonant shape
down the height
01 the tine is the
I
first cantilever
mode.
I
"I
I
View down the bore 01 the meier.
The tines are showing their resonant
shape (grossly exaggerated) in mode (l.3).
I
I
, Cutaway \li8'IN from lJ'Ider'neath lhe lines
The nree are shOWingtheir r~t
shape (!1QSsty exaggereled) in mode (1.8).
I
Fig 2.1.2 View of Resonance from Underneath Tines.
I
\\1 The resonant mode chosen is a combination of the first tuning fork mode
shape with the third longitudinal mode shape. This can be denoted as mode
(1,3) and was found to be the most suitable mode due to a combination of
I physical characteristics. The discussion of such effects is beyond the scope
of this paper.
It
I
I
I
I
I
,
Cutaway 3-dmenslonal View of the Massrnasler04(l()
showing the meters resonant sensors.
I PageS
I
2.2 PICK-UP AND DRIVE MECHANISM .-
The mechanism for maintaining the tuning fork in resonance is similar to other
Coriolis meters in principle: a single drive transducer translates electrical
I
energy into tine movement, and two pick-up transducers detect the
movement, developing electrical signals which are fed back into the drive I
circuit.
I
I
Pick-up Pick-up
I
,
I
I
Fig 2.2.1 Pick-up and Drive Arrangement I
I
Since the pick-ups are displaced up- and down-stream of each other, a
4
phase diffence is detected between the two Signals during flow. This is the
prime measurement for the mass flow calculation.
I
I
I
I
I
~
Page 6
I
I
II 2.3 MECHANICAL CONFIGURATION
I The meter has an extremely high pressure containment ability as can be seen
from the wall thickness in figure 2.1.1. The one inch thick wall is thinned only
I over very small areas, visible in figures 2.2.1 and 2.2.2. The wall thickness in
the base of each area is still Smm.
I
TINE SHAPE
I
The shape of the resonating tines has been
I chosen with a view to optimising Coriolis
performance whilst avoiding cavitation, minimising
I
Ie
I
I
"
I Page 7
I
3.0 METER PERFORMANCE III
The meter produces outputs of phase shift, resonant frequency and line
I
temperature in identical fashion to most other Coriolis mass flow meters. Also
the factors that affect the meter's performance are very similar to those that
affect traditional tube type Coriolis meters.
I
Accordingly the output can be compensated
systematic effects using a flow computer.
for all the characterised I
I
3.1 EFFECTS ON METER PERFORMANCE (THEORY)
I
EFFECT OF TEMPERATURE
Page 8
--I
I
It EFFECT OF FLUID VELOCITY OF SOUND (VOS )
I The VOS of a fluid causes the resonant weight to be heavier than the static
mass of a volume of fluid. Hence Coriolis meters always 'overread'. However
since all meters are calibrated on real fluids the effect is automatically
I compensated for the calibration fluid. Subsequent use on a different fluid
produces VOS errors.
I
VOS (mls) Water Calibration VOS (m/s) Kergsene Caljbration
I 2000 ,...----,-----,
'--.....
2000 ,...----,,...-----,
<; "'--.
I '-,
~----===- ~~
, 1500
1000-
~::::
- . -,_~'--:---
-, ,
. , '
...- - .. '-,
~
--,-
,
-
I 50050l::0,-----:-10:'::0=-O
I
----:-!15OO
I
50050l::O,----::!IO'=OO,-----:-!15OO
I Density (kg/m3)
I
Fig. 3.1.1 Graphs of constant Massflow VOS offset.
I
The meter could conceivably be sensitive to some types of flow profile within
the pipeline. However no such effect has yet been identified and the
sensitivity could be negligible. Through symmetry, swirl is not expected to
I create any systematic offset. Initial installations could use the same flow
conditioning that is used for turbine meters, although tests may oneday prove
that this level of conditioning is unnecessary.
I
I
I
,
I Page 9
I
3.2 TEST RESULTS
II
ln coltaboratlon with Statoil, a development program of tests has been
planned and started. The results displayed herein are from the initial I
prototype, tested at Norsk Hydro, Porsgrunn, Norway on 3-4 October 1991.
I
NOTE ON FIG. 3.2.1
On the first day of testing, an attempt was made to find the flow range of the
I
meter. Linearity was lost above 350 tlh due to the onset of cavitation. At low
flow a zero error caused the percentage error to veer positive. I
Developments both in streamlining to minimise cavitation and optimising zero
stability are continuing at Schlumberger. I
NOTE ON FIG. 3.2.2
On the second day of testing, the repeatability was assessed at three flow
rates. The results at 170Uh and 300Uh were within the O.05% specification of
--I
the volumetric flow rig at Norsk Hydro.
Page 10
--I
I
K2Iibr2:j~gsb2vis for
Porscrunr: gjennc rnst rom ningsmalere
Autc rna ,Ise ~i:"";':;5a'lde:;ngen
I
I~~jcn
I - .. . ..
Massfiaw
P;-otot',pp
..... _.n. !.._, "
ISer.em
Scbl,
001
1
"",bprgpr
':,';..
4"
I ~.!~!'!cmf1c:'!
40-400 m3~
,
I '{!i.'s..e
~::-err:SIl'lC;;"
I~~~cr.
I
, .:.:Jrl
'"",, i
" ~ad()f
D,m J
B~
o leF~;1~~~~
~. " , opl
o
1 !12194.8IJ 1 I 11-97
I
,
I
80. 00 ! SS28. 00 ~H 22.0 15921.30 I i 5923.10 I I 11-03
; 47.00 : 6004.00 'H za n i 5996.1Q I I6!l!RSll 1 i , a. 23
; ;
21. 00 2012. OO'H 23. 0 1 2O!j9, 34 I I 2028.. 20 o 94
, 1
'''';: "':U~~.L"lCfr:-.a;,
,;,::!-e~l:;:-.EI: ':lO:'.. THloenlur1u)r:ekslOflSi.aklor: l-OJXXI05. T, cc:
I
;:;.: ':;eierG::seffi.c.:e~: "',;~a..;:;:'Iet velum: ~ -:.. Noyald.ighet '1RSkeslfom: :
;<'crr.;!f1 e!:.N 1o.aIibrecing mOl hutm.\lnormal.
Gj,snlns metertaktor
o :.!eleli;;~I:r
I iJ ~.le~
1.000
\, 1'1
\ 1 ill Iii / I
-~~--~--~--~--~--~--r~-l
I a. 500
\1
':-'''-
1
1
1
1
1 1 I~..-r'
1
I
,
."- ~
I
0.000
0.. 0 so. a lCD.O rso, a 200. a 2SO.. 0 300. a 350. a 0400. c
n"2vF5
I"
'-I
lCO
. "-'''- ..;" OMGJORT VOLUMETRISi\ TI!.. ..MASSE.
"-I;
I
: i
I
I
I
, Fig.3.2.1 Test Results: Day 1.
I Page 11
I
& HYDRO
Fcrsgrunn
K2Iibr2r:n';SG2Vis fer
sje nn 0 rnstrcr n ;ngs;nale r02
001
4"
I ~\31<!omJ~C:,?
40-400 m3/h I
I
306. on i12098. 00 IH '22. 5 112090.20 I 112118.60 I
297.00112085.00!
29S.. on
, '
'12079.00 ! H
H 122.2 h2070.80
~2. 5 h 2064. 20
h2100.90
it2093. 20 j
l
I
299. on il2094.00 i H ~5 112079.20 I 112107.40 I
0.200 c- --~
r-~
---:~~-
I
--:---~--
I 1
I
0.000
1
~--~---~--~~---+---I--
I
1
I
,."
1
1...........---/1
1 1 1
I
I I I I I I
I
-0.200
0.0 so. a 100.0 1SO. a 200.0 250.0 300.0 350. C:
O~,avFS
l:ll
!DO
OMGJORT VOLUMETRISK TIL M~SS.L
~
:;
I
I
I
I
I
Fig.3.2.2 Test Results: Day 2.
Page 12
--I
I
I
JJ1HYDRO
Pcrsgrunn
A;.;:o~a ~iS2 rir:9 5 avce Iir:g en
K.alibreringsbev:s tor
9ienncmstrnm ni ngsma Ieie
I,~~" E 2048
I I . Mossflow
I 5"n('nr I \:.~,(1'::~If:'ce
40 - 400
ro 0 ~~e
.:::~. _,-::~'; ~-:--:~~r , 001
.~,~,>f:i~,~,~,,~;
m 3/h
I ,0i~r-"
I",""",m
:::'::'1 ~'"''
vc!um
LJ' 'c
r.,
;~y::" ~~m
Konic;:er:
'~""
0 ...
.1
Oml.
!""",..
~:r:I_m~IR(I'''. vc!um n
0 ';~Ue""rll
I&'IH nl~
P.'::SH :I.H.u-
I.UCf
Opm
[Jp.1
:3
Fc;i
:-:.
o . 0
I I,
,
S2. 00
54. 00
S3.
I 6054.
I 6049.
oo! 6045. 00
00 I H
80 1H
H
124.0
124.0
124.2
6044.52
6040.66
603S. 631
1
I
I 60Q.
I 6041.73
1 6038. 13
49 I
1
1
-0. 03 I
0.021
0.04 !
2Il2. 00 h2046. 00 1H 124.0 112027.90 ! I 112024.60 I -0. 03 1
I
, :
,
i
':-::
I
I
i
I
~im:Jnormal.
I
1
,,
i
NO'ic"l~nel:
,
I,
!
i
O.05~.
I
I
I
I
I
'femperalUfkOrrellspnsWdor.
I
1
i
I,
1~_c...'"'OC5.T
1
,
I
1
i::lC;
I
:
I
I
!
,
I
I
i
!
,
1
I
:
Gi,sn.::s ;nelerlakl:lf
I o ~elerl.klcr
K1 ~ lei
0.050
0.040 - -
1
-1-
1
- .,. - -
1
r-
1
-1-
1
- .,. - -
1
r -
1
-1-
1
-.,. - -
a. 020 - -k!- - ...J. - - L __ 1__ ...J. __ L __ 1__ ..1 __
I 0.000 -
1 'I
~~-I--I-----
1 1 1 1
I -0.040
40.0
,
SO.0 80.0
'1"
100.0 120.0 14.0.0 160.0 180.0 200.0
" 22O.C:
n....
avF5
'-I
I ::r:
100
I
I
I
I
, Fig.3.2.3 Test Results: Day 2 cont.
I Page 13
I
3.3 SECONDARY OUTPUTS
The meter can be calibrated to compute density from the resonant time
period, in a similar fashion to other Schlumberger 'Solartron' densitometers. A
I
3 liquid calibration with temperature calibration should achieve accuracies of
O.5kg/m3 Knowledge of the generic characteristics of vas. flow and viscous
effects will then permit accurate density measurement of the full 4" flow
stream. As previously mentioned, there is no pressure effect on this meter.
I
Knowing the mass rate and line density enables the volume rate to be
derived. I
A PT100 embedded into a tine gives an accurate measurement of
Temperature for density referral calculations. Hence referred density, I
standard volume, and nett mass can easily be computed.
Despite the radical mechanical differences between this new meter and the I
standard Coriolis meter, the information output is identical. Hence the meter
interfaces directly to a Schlumberger flow computer in which secondary
calculations can be performed. ~
I
I
I
I
~
I
I
I
I
I
tI
Page 14
I
I
II 4.0 CONCLUSIONS
I This meter places the sensing resonator within the fluid, instead of
surrounding the fluid with a resonating tube. Thus it overcomes the problems
I of scale by turning the problem inside out. Large bore coriolis meters of only 3
diameters length will soon be commercially available.
I Initial test results show that this design of meter can achieve similar
performance to that of existing mass meters, yet have the advantages of
compact size, rugged design and high pressure capabilities.
I Work is planned to continue in optimising the design of a 4 inch meter and to
produce 6 and 8 inch versions during 1993. Initial applications are expected
I
, to be on refined single phase products, LPGs and gases.
I
I
I
I
~
I
I
I
I
I
,
I Page 15
I
I
I
I PIPE El.EO'J EFFEX:1'S 00 THE v-coe F'LGlMETER
I by
I
, S A Ifft and E D Mikkelsen
Ketana Inc
I
I
I Paper 5.4
I
~
I
I
I
NEL, East Kilbride, Glasgow
I
II
I
I
It
I PIPE ELBOW EFFECTS ON THE V-CONE FLOWMETER
I
I SUMMARY
The paper discusses the performance of the McCrometer "V-Cone" differential pressure meter in
"I
studying instaIlation effects on several flowmeter technologies, including a typical orifice plate
flowmeter. McCrometer, along with an independent test facility, has conducted tests on the
V-Cone using similar configurations. AIl tests on the orifice plate and the V-Cone use water as
the test fluid in 50 mm stainless steel pipe.
In matching configurations and Reynolds number ranges, the V-Cone is less susceptible to
I NOTATION
I K Meter factor
I k Isentropic exponent
II I
I
I
~ Beta ratio
I
AP Differential pressure psf
Each flowmeter technology requires a certain distance of straight, undisturbed pipe before and
after the primary element. Depending on the flowmeter requirements, these distances are often I
difficult or impossible to achieve. For instance, a flowmeter may require more piping than is
available in an existing system where flow measurement was not necessary before. A new
application may have limited area, such as on oil platforms or vehicles. Perhaps the distance is
I
available but the cost of providing straight pipe before and after the primary element is limiting.
In systems with 25 mm line sizes, installation requirements may only require a meter or two.
This cost may not be significant. However, in systems where line sizes reach over 700 mm, the
I
cost of providing 35 D of straight run before a meter may be prohibitive.
The design engineer can usually compromise in some way to accommodate the limitations of the
I
flowmeter. However accuracy or permanent pressure loss sacrifices. Accuracy decreases if the
piping requirements are ignored. Permanent pressure loss increases if a flow conditioner is
employed upstream of the meter. Either solution is less than ideal tI
With these problems in mind, the National Institute of Standards and Technology (N.I.S.T.) in
Gaithersburg, Maryland, decided to form a consortium consisting of sponsor members from
I
industry and other government programs, both domestic and foreign. N.I.S.T. thus began the
"Industry-Government Consortium Research Program on Flowmeter Installation Effects." The
goal of this group was to study the flow patterns of water after common installation problems
I
suc::has single elbows, double elbows out-of-plane, tees, and reducers. N.I.S.T. then installed
flowmeters after these disturbances. Industry members donated two flowmeters to represent a
typical differential pressure orifice plate flowmeter and a typical turbine flowmeter.
I
Characteristic:: curves were plotted to show the flowmeter's variance at different positions relative
to the elbows or other disturbances. This might allow end users of orifice plates or turbine
meters to characterize their meter's performance according to the installation.
I
One hope of the consortium was that industry members with proprietary meters would run tests I
parallel to the tests run at N.I.S.T. Ketema/ McCrometer Division, a flowmeter manufacturer
2
II
I
I
It
I and consortium member, in Hemet, California decided to run installation effects tests on the
V-Cone differential pressure flowmeter. This is a patented design, differential pressure
producing device using the same basic principles of flow measurement as the orifice plate. The
I overall goal of these tests was to define installation requirements for the V-Cone downstream of
common disturbances. McCrometer converted an existing test lab to emulate N.I.S.T. test
conditions. Configurations tested to date have been the single 900 elbow and the double 900
I elbows out-of-plane .
The geometry of the V-Cone suggests a radically different approach to differential pressure
flowmetering, see Figure 1. As with other differential pressure devices, the flow constricts to
create high and low velocity areas, which creates a differential pressure signal. However, the
I V-Cone's constriction is not a concentric opening through the center of the pipe. The V-Cone
creates an annular opening, forcing the fluid to flow around a cone suspended in the center of the
pipe.
I Equations for the V-Cone are slightly different from an orifice plate or venturi. The Beta ratio,
/3, is the ratio between the square root of the open area in the pipe and the square root of the
I open area at the meter's constriction. The V-Cone's Beta ratio is:
(1)
I
The standard equation for differential pressure flowmeters is:
~ (2)
Q=Kyfi
I The k factor for the V-Cone is:
I (3)
I For compressible flow, McCrometer applies the standard equation for the adiabatic expansion
factor, Y:
I
I
It 3
I
I
[1-~4][b][l-~J~[ 1-(1-~)]
!
2
I
y= (4)
I
[1- 1- is) t ][ 1- (1- !':) ]
~4 (
I
Note: The adiabatic expansion factor applied only if Y> 0.96. Otherwise a characteristic
expansion factor must be derived for the meter based on calibration data in a compressible
fluid.
I
2. TEST PARAMETERS
I
The test parameters for the V-Cone tests were set to follow the test parameters established by ~
N .I.S. T. tests on the orifice plate.
The McCrometer static gravimetric flow calibration stand can test 12 mm to 100 mm nominal
diameter flowmeters. Figure 2 shows a schematic diagram of the testing apparatus. Figure 3
I
shows a scaled diagram of the test section.
The closed system recirculates water constantly from a 2200 liter storage tank. An electric pump
I
draws the flow from the tank through a 100 mm PVC pipe. From the pump, the water enters an
upstream header. The 250 mm by 1200 mm chamber incorporates straightening vanes and a
dampening screen to lessen pulsations from the pump. A recirculating by-pass line of 50 mm
I
PVC pipe also helps to reduce 'pulsations. The water leaves the header horizontally through a 50
mm PVC ball valve, used to ease startup vibrations. I
The water passes through 50 D of straight 50 mm PVC pipe before entering the single elbow or
the double elbows out-of-plane. The elbows are all 90long radius (centerline curvature=1.5 D.)
Flow then passes through the 200 D of horizontal test section. Test section piping is schedule ~
40, stainless steel with an approximate wall roughness of 3 J.l"l. After passing the test section,
the water turns vertical, passing a PVC ball valve. This valve is used for flow regulation I
purposes. The diverter section follows.
A pneumatic system diverts the water to either a receiving tank, open directly to the storage I
tank, or to a collection tank. The collection tank weighs the collected water over a measured
time. An optical sensor on the diverter triggers a timer to measure the precise time of the
collection period. A mercury thermometer measures the fluid temperature. I
In the test section, the meter was leveled prior to testing. Differential pressure taps on the meter
face the "inside" of the elbows. A "smart" differential pressure transmitter measured the
I
differential pressure created by the meter. The 4 to 20 rnA signal from the transmitter was
measured with a multi meter. A computer collected 200 data points over the test period through
an IEEE-488 bus.
I
4
II
I
I
II
I
Prior to testing, the transmitter was calibrated using a pneumatic dead weight tester. The
I "smart" capability of the transmitter allowed the full scale differential pressure of the transmitter
to be scaled to the full scale created by each meter at the desired maximum flowrate.
The objective of the first set of McCrometer tests was to detect the effect of upstream elbows,
I
,. both a single 90 elbow and double 90e1bows out-of-plane, on a V-Cone meter. The double
elbows were close coupled.
Accuracy for the V-Cone primary element is 0.5%. During the evaluation, a deviation outside
of 0.5% was considered to be an effect of the elbow configuration.
The McCrometer tests included three 50 mm V-Cone flowmeters with Beta ratios of 0.363,
I 0.650, and 0.750. Beta ratios for V-Cones represent the same area ratio that standard orifice
plate Beta ratios represent. These meters represent the typical range of Beta ratios for V-Cone
applications. End connections were standard ANSI flanges (150 pound, raised face, slip-on.)
I The test fluid was water at approximately 20C.
I Beta= 0.363
Beta= 0.650
6 to 37
8 to 31
11,000 to 65,000
14,000 to 51,000
Beta= 0.750 15 to 65 25,000 to 110,000
~
The meter was first placed a maximum distance from the elbows. The data taken at this point
I was the baseline data for the particular meter. In this position the meter was 190 D away from
the elbows. Each meter was then moved in intervals closer to the elbows. Six different
positions relative to the elbows were tested. The positions were approximately 190, 23, 9, 2, and
I o D away from the elbows.
At each position, each meter was tested at five flowrates covering the range stated above. At
I each flowrate a repeat point was taken for verification. Thus for each position, a total of ten test
points were taken. These ten points were then averaged and plotted on figures 4 and 5.
I Figure 4 shows the change in the meter's coefficient of discharge, Cd, versus the distance of the
meter from the single elbow in nominal pipe diameters. The change in the meter's Cd
represents the percentage deviation from the baseline data taken at 190 D. These points
I represent an average Cd of the Reynolds number range.
II
5
I
I
II
Two dashed error bars show the stated accuracy of the meter. These bars are at O.S%. Points
outside these bars are considered results of the elbow upstream. On Figure 4, the V-Cone with 13 I
= 0.750 showed a deviation at 0 D of +0.622% from the baseline data. All other points fell
within the 0.5% of the meter. The maximum effect of the single 90 elbow on the three
V-Cones during the McCrometer testing was 0.122%. This was the largest deviation from I
baseline data with the V-Cone.
Figure 5 uses slightly different X and Y scales to show the effect of double elbows out-of-plane I
on the V-Cone. Only one point falls outside the accuracy bars. This point is at 100 D with the
V-Cone at ll= 0.650. The deviation was +0.504%. The maximum effect of the double 90
elbows out-of-plane on the three V-Cones during the McCrometer testing was 0.004%.
I
3.2 Independent test results with the V-Cone
I
SIREP in England performed an evaluation of installation effects on the V-Cone. SIREP is an ~
international instrument users' association. The international industry members ofSIREP
approached McCrorneter with the offer to evaluate the V-Cone according to the specifications of
the meter. SIRA is the instrument testing branch of SIREP and was responsible for the
evaluation process. Installation effects tests were among the variety of tests SIRA was to
I
perform.
SIRA tested both the single 90 elbows and double 90 elbows out-of-plane before the V-Cone.
I
The double elbows were close coupled. SIRA was to test the single elbow in two configurations,
once with the taps in the same plane as the elbow, another with the taps perpendicular to the
plane of the elbow.
I
McCrometer provided a standard 50 mm V-Cone for the tests. End connections were standard
ANSI flanges (150 pound, raised face, slip-on.) The test fluid was kerosene at 30C (density =
I
80 I .4 kg/m", viscosity = I. 73 cSt.)
The meter was placed at two positions relative to the elbow configurations, at 2 and 10 D
I
downstream. Baseline data was taken from a straight line test with no elbows. At each position,
the meter was tested at five flowrates. Three points were taken at each point. I
SIRA results concur with McCrometer results on both the single and double elbow tests. On
request, McCrometer will provide a copy of the SIREP evaluation report E 1705 S 92. I
I
6 II
I
I
II
I 3.3 NIST results with a typical orifice plate flowmeter
Dr. George Mattingly and Dr. T.T. Yeh of the Fluid Flow group of the N.I.S.T. in Gaithersburg,
I Maryland performed installation effects on a typical orifice plate flowmeter. This was part of a
government-industry consortium to study such effects.
I Both a single 90 elbow and double 90 elbows out-of-plane were tested. The double elbows
were close coupled.
The N.I.S.T. tests included three orifice plates in a 50 mm line. The stated accuracy of the
I meters was taken to be 0.5%. The Beta ratios tested were 0.363, 0.500, and 0.750. Flange
connections were weld-neck ANSI flanges. The test fluid was water. Flow criteria for these
tests were the same as the McCrometer tests.
I
, The positions of the orifice plate to the elbows were similar to the McCrometer tests.
Figure 6 shows the effects of the single elbow on a typical orifice plate. The scales for the X
and Y axis match those of figure 4, single elbow effects on the V-Cone. The orifice plate
showed significant effects from the single elbow. The maximum effect of the elbow (at 3 D
I with (3= .750) was approximately -4.5%.
Figure 7 shows double elbow effects on the orifice plate. Again, the scales of Figure 9 match
I those of figure 5, double elbow effects on the V-Cone. The orifice plate showed slightly less
effect from the double elbows than the single elbow. The maximum effect of the elbows (at 3 D
with (3= 0.363) was approximately +2.6%.
I
I 4. CONCLUSIONS
In matching piping configurations and Reynolds number ranges, the V-Cone demonstrated less
susceptibility to elbows upstream than a typical orifice plate flowmeter.
~ V-Cones showed some effect from the elbows, up to 0.122% in one test. Orifice plates,
I however, showed extreme effects. This was not unexpected according to existing international
orifice metering standards, both ISO-5167 and ANSIIAPI-2530. With a Beta ratio = 0.750
orifice plate, ISO-5167 recommends 70 D upstream for double elbows out-of-plane.
I ANSIIAPI-2530 recommends 35 D for the same installation.
McCrometer's goal was to identify installation requirements for the V-Cone. These first tests
I were not conclusive for those purposes. These tests do quantify the effects of elbows upstream
of the V-Cone. For any V-Cone with a Beta ratio between 0.363 and 0.750, the maximum effect
of either a single elbow or double elbows out-of-plane would be approximately 0.12%.
I More research is necessary to describe the V-Cone's total performance. The geometry of the
meter does not easily lend itself to comparison with other meters. Past studies have noted the
I flow pattern through a V-Cone primary element. Fluid traveling in the center of the pipe is
forced by the cone to the wall of the pipe and through the annular constriction. This mixing of
II 7
I
I
..
the low and high velocity areas of the flow creates a pronounced "flattening" of the flow profile
directly upstream of the meter. This characteristic of the V-Cone is the most probable cause of I
the V-Cone's consistent performance in less than ideal flow situations.
I
REFERENCES
I
I
I
I
I
8
II
I
I
II
I The CONE
I
Differential Pressure
I
Flowmeter
I [}{] Ib
I
,
I
I
I
I
~
I
I
I
I
I
It Figure 1
I
NUMBER TITLE
Figure 2 SMALL STAND OPERATIONAL SCHEMATIC
S.A.I. 09-27-92
rILE NAME DRAWN CHECKED APPROVED DATE
TRIGGER
COMPUTER
I
I
TIMER
r-- 1 /
P MULTIMETER
'-r- DIVERTtR 1-------'
~oz:; PSIG f IZ BIT RESIlLUTIDN
DODD
II
r-~ I
: 4-ZOMA (~DJ:;r. of Nlndlng) OZ T "-
W
c::l
<[
W DP
I TRANSMnTR re"~f""enc:ed -to
AMETEK RK certIfied D'JT
T \ illOGII TANl(1
J: /
~
<[
w
0::
HI LO DODD
~
CI1 V-CONE FUJII 1I!IU1!II\ :toz pounds
a.
::>
------ SIGNAL
T
PUMP
FLOIJ
Me" - Fi"'e ~ - M~uMBJW- - - ..
1--...,...1----.I-D-'\-AC-A-D\-E-LB-O-'W-2-.D'W-G---I -, J 1
[1'JCS@ffi[1'J@17@ffisAl.11108-27-92
i;LETLBD~PIN~DNF'!mRA_ -
FILE NAME DRAWN CHECKED APPROVED DATE
REV DATE 8. BY'
.367 BETA
- -a--
.650 BETA
c::
- I
Q.l
Ol
c::
co
s:
I
C,,) -2.0-
I
-
-4. O+-----,---,---.-----r,--.--T.,----~.---.,--.------,--.--.---1
o 8 16
Diameters from the Elbow, Z
24
"I
I
I
I
Fiqure 4
I
411
I
I
I
I
I
DOUBLE ELBOW EFFECTS ON V-CONE
I tested at Ketema/ McCrometer Div.
3.0.,.----------------------. .---------,
I
, -
--+-
.367 BETA
-B--'
.650 BETA
I~1.5- ~--.
.750 BETA
u
I~
Q)
I~
~
It) 0.0 m'"
~:";:"
X ---_.--- _-"
s-.- ..
....
'-I o
-
-1.54---".----.....---..--,..,..,-..--.....---r----r-..---r,-r--...,--r--,""T,--I
W 100 150
Diameters from the Elbows, Z
I
I
I Figure 5
I
II
I
I
til
I
I
I
SINGLE ELBOW EFFECTS ON ORIFICE PLATE
tested at NIST I
0'01===::~~~~~~~
--+-
.363 BETA
I
-t --- ,-/8--
- ......,- -El--
13---------------------------
------ -
_ ...-,-_----xc---i .500BETA
......
.500 BETA
\
\,
~ 1.5- \ .750 BETA
J3.!:
\ \
:El
<,l"-l.
I~
ttl
<, .....
,..... ,..
-Fr!-
-- ..-----
x-
f'
~
0.0 ......
.,.,,~ -- -- -... '*
...-
~ r.-,
--
.~... .. .....
x-_--_
~
\ ,I \
\ /
\I
I -1.5 :t., , , I I I I I I I I
o 50 100 150
I Diameters from the Elbows, Z
I
I
I Figure 7
It
I
,
,
@o
4IIt -----------------------------------------------------------------
FLOW MEASUREMENT - THE NEXT TEN YEARS
"To Accomplish Great Thinqs, We Must Not only Act But Also Dream,
Not only Plan But Also Believe." - Anatole France
1. GLOBAL OUTLOOK
-----------------------------------------------------------------
1 of 29
-----------------------------------------------------------------
'It
The cultural differences, the lack of a free market
infrastructure, the legal system and the political risks
are significant issues. Protection of investment, free
prof it distribution in hard currency, operating
management and other issues frequently call for joint
ventures to minimize risk associated with a single
project. Expansion of these markets will occur only when
experience meets expectations over time.
Progress will emerge slowly. Patience is needed as much
as capital, if not more. The culture shock of moving to
a free market system is enormous for both sides of the
equation.
At today's oil prices, the investments required over the
next decade is a formidable challenge. The sums needed
are not likely to be met out of cash flows alone. Also,
increasing production in the regions mentioned above will
involve leading edge western technology combined with
excellent management skills for increasingly complex
projects .
East Asia is also a major player on the energy scene.
Its decisions and choices affect the global community.
Obviously, the presence of Japan within the region is the
primary factor in this development.
As energy consumers, the USA and western Europe remain at
the heart of the world energy drama. While they continue
to exert great influence, many of the levers of power are
now found in the Middle East, South America, Northern
Africa, East Asia and, in the next century, the CIS.
competition among the sources of energy will grow as a
result of environmental legislation, market forces and
capital investments (Figure 2). The future energy
situation can be stated simply - Energy demand will
continue to grow steadily with the economies of the
world. Alternative energy sources will be developed as
a result of economics and the environment.
other things, that the science underlying these
environmental beliefs needs to be sound and
-----------------------------------------------------------------
2 of 29
4It -----------------------------------------------------------------
comprehensive.
Due to these global pressures, the future for flow
measurement and quality assurance is bright.
2 UPSTREAM SECTOR
Crude oil
-----------------------------------------------------------------
3 of 29
-----------------------------------------------------------------
activity will level off and gradually migrate to a mature
region.
In the crude oil supply picture there exists an
interesting dichotomy. Greater than anticipated growth
in key national economies - USA, Japan, and Germany -
will boost petroleum demand and cause a supplier driven
pricing scenario. The International Energy Agency (lEA)
estimates a lack of spare production capacity in the
global oil market in the near term. As a result, crude
oil prices will rise sharply with the economic growth of
these key national economies.
Ifaturlll Gas
The international gas industry has undergone an extremely
dynamic evolution over the two decades. This evolution
was spawned by energy policies oriented towards supply
diversification, crude oil substitution, environmental
concerns, and in the USA - regulatory restructuring of
the transmission segment.
The worldwide demand for natural gas is projected grow
rigorously over the next decade. Conversion of power
generation plants, home heating systems, mass
transportation vehicles and fleet vehicles are projected
as a result of environmental regulations and alternative
energy pricing structures.
Natural gas is believed to be more environmentally
fr iendly than crude oil. Whether or not the science
behind this belief is sound, there seems little doubt
that gas as a source of energy will become increasingly
acceptable and, where economics permit, preferable to
other fuels.
2. DOWNS~REAKSEC~OR
In recent years, refining capacity for the key
industrialized nations has remained at a constant level.
Since 1984, USA refining capacity has declined slightly
due to environmental legislation - the Clean Air Act
(CM), and the Clean Water Act (CWA), and the global
decline of sweet crude oil. As product demand increases
and additional refineries permanently close their doors,
capacity utilization rates, now at 87%, will increase to
90-95%.
Refiners can't or won't add capacity, due to the capital
-----------------------------------------------------------------
" of 29
4IIt -----------------------------------------------------------------
investments and the resulting cash flow crunch associated
with compliance to the CAA and CWA. Capital investments
required by the CAA alone is currently estimated to be
approximately 40-50 billion USD by 1995. This cash flow
drain will not increase capacity and will consume most if
not all short term profits. Joint ventures, toll
processing and a migration by some refineries to niche
markets will be necessary to survive in the North
American market.
3. TRANSPORTATION SECTOR
-----------------------------------------------------------------
5 ot 29
-----------------------------------------------------------------
The transportation industry will undergo significant
expansion, contraction and regulatory changes over the
next ten years.
Current and future customers still want the same things
that Yesterday's customers wanted, namely, a quality
transportation sector which provides reliability and
security of supply. But they may have uncomfortable
perceptions regarding supply reliability, and/or pipeline
capacity reliability.
The transportation business will be accelerating towards
computerized transfers of commodities, pipeline
movements, tariff billings, pipeline scheduling and
supply contracts/agreements. One analogy is that of the
international banking industry and its dependence on
computerization to conduct daily business.
In highly populated nations, the need for high
discrimination line integrity systems will be fierce.
Densely populated regions should not be subjected to the
uncertainty of eighty year old systems not being
monitored for dynamic leak detection. Regulations
requiring internal inspection and possibly hydrotesting
on a set time interval are probable if the leak detection
technology has not met the needs of the general public.
One thing is for sure, the oil, natural gas, refined
products and chemical industry isn't going anywhere
without pipelines. But even more than that, the industry
must demonstrate to the customers that we can get the
commodity to the right place at the right time and the
least cost. Dr. Edward Deming would probably agree that
the challenge awaits those who wish to pursue progress
and viability.
oil Systems
The oil transportation sector will undergo the most
significant pains (Figure 3). As domestic production in
the USA continues to accelerate its decline, the grid
demographics will be redefined towards imported crude
movements. Older onshore domestic crude oil pipeline
-----------------------------------------------------------------
6 of 29
-----------------------------------------------------------------
grids will be eliminated and replaced with independent
operators or truck gathering alternatives. Offshore
pipelines will grow only when the pricing structure meets
the capital investments required for deep water
operations. New batched crude oil pipeline systems and
modifications to existing domestic pipelines will be
justified to satisfy the appetite for imported crude
oils.
The refiners' sensitivity to crude oil quality assurance
(QA) will be amplified due to refinery design,
environmental regulations, market demands, product
specifications and economics. The transportation
industry will need to address these customer concerns
with clear vision, a creative atmosphere and mutual
agreement on acceptable performance levels.
Natural Gas
-----------------------------------------------------------------
7 of 29
-----------------------------------------------------------------
market will
growth.
be regionalized with respect to economic
Checmical systems
with the cycles associated with sophisticated chemicals,
existing pipeline systems will grow at a rate directly
with the global economy (Figure 6). Again, the market
will be regionalized with respect to economic growth.
all senior pipeline management personnel (Figure 6). The
issues of increased capital investment, operating and
maintenance costs, training costs, additional personnel
and ISO 9000 certification are real concerns in today's
business climate.
However, QA will also reap benefits. By applying
statistical quality control (SQC) to equipment
performance, the maintenance costs and risks will be
minimized. By investing in equipment which has a record
of quality performance, minimum operating and maintenance
costs will be achieved. This plateau does not occur at
the lowest capital investment level. Other returns are
the minimization of store stock inventory for equipment
which hardly ever breaks, standardization of equipment,
and training costs.
-----------------------------------------------------------------
8 of 29
-----------------------------------------------------------------
5. FLOW MEASUREMENT TECHNOLOGY
-----------------------------------------------------------------
9 of 29
----------------------------------------------------------------- ~
Installation Effects
The problems associated with installation effects of
flowmeters have been with us a long time. Pragmatic
solutions have evaded our industry due to our limited
insights into the "real" flow field.
Experiments over the last ten years has shed light on
increasing our insights. In the USA, Canada and EC,
scientists have recently conducted installation effects
research on turbine, orifice and ultrasonic flowmeters.
This data measured the meter's performance as a result of
interation with the near term flow field. with this
insight, postulates have been proposed and are planned to
be validated or revised within the year.
with the application of microchip technology into smart
transmitters, sophisticated flow computers, personal
computers, Computational Fluid Dynamics (CFD), thermal
anemometry (TA) probes (i.e., hot wire, hot film, x-
wire), Laser Doppler Velocimetry or Anemometry (LDV/LDA),
characterization of flow meters in real time, high
pressure gas piston provers, ultrasonic flowmeters,
coriolis flowmeters, videoimagescopes, etcetera, large
steps toward lowering the flow measurement uncertainty is
possible. These "new" tools are providing significant
advances in the refinement of existing metering equipment
as well as the birth of new technology.
The advent of LDV/LDA technology has provided a tool to
perform three dimensional flow field measurements. This
technology is capable of measuring three non-orthogonal
velocity components simultaneously, resolving from those
three independent orthogonal velocity components, and
then computing
averaged
the mean velocity vector, the time
Reynolds stress tensor, and other
-----------------------------------------------------------------
10 of 2g
-----------------------------------------------------------------
research is attempting to provide a thorough
understanding of the complex flow field. The LDV/LDA, or
TA probe are tools which provide us with the needed
insight to the microscopic flow field. Studies at NIST
Gaithersburg (Mattingly and Yeh) , Texas A&M (Morrison et.
al.), NIST Boulder (Brennan, et. al.), SwRI (Morrow, Park
et. al.), NOVA Husky (Karnik, et. al.), CERT (Gajan et.
al.), NEL (Reader-Harris, et. al.), Gasunie, K-Lab
(Wilcox et. al.) and others have recently measured mean
velocity profiles and turbulence structure associated
with upstream flow conditioning effects.
Flow Conditioners
-----------------------------------------------------------------
11 of 29
-----------------------------------------------------------------
disturbances with unknown characteristics. Even a simple
elbow can generate very different flow conditions from
"ideal" or "fully developed" flow. In reality, multiple
piping configurations are assembled in series generating
complex problems for standard writing organizations and
flow metering engineers. The problem is to minimize the
difference between "real" and "fully developed" flow
conditions on the selected metering device thus
maintaining the low uncertainty required for fiscal
applications. For clarity, we will refer to this as
"pseudo-fully developed" flow.
A method to circumvent the influence of the fluid
dynamics (swirl, profile and turbulence) on the meter's
performance is to install a flow conditioner in
combination with straight lengths of pipe to "isolate"
the meter from upstream piping disturbances.' Of course,
this isolation is never perfect. After all, the
conditioner's objective is to produce a "pseudo-fully
developed" flow.
In general, upstream piping elements may be grouped into
the following categories -
*
*
those that distort the mean velocity
profile but produce little swirl
those tha.tboth distort and generate bulk
swirl
Flow conditioners may be grouped into three general
classes based on their ability to correct the mean
velocity profile, bulk swirl and turbulence structure
(Figure 9).
The first class of conditioners is designed to primarily
counteract swirl by splitting up the flow into a number
of parallel conduits. This class of conditioners
includes A.G.A. radial tube bundles, A.G.A. hexagonal
tube bundles, ISO 5167 tube bundles, AMCA's honeycomb and
the Etoile.
The second class of conditioners is designed to generate
an axisymmetric velocity profile distribution by
subjecting the flow to a single or a series of perforated
grids or plates. The profile is redistributed by use of
the blockage factor or porosity of the flow conditioner.
This class of conditoners includes the Sprenkle, Zanker
and Mitsubishi designs.
-----------------------------------------------------------------
12 of 29
-----------------------------------------------------------------
The third class of conditioners are designed to generate
a "pseudo-fully developed" velocity profile distribution
through porosity of the conditioner and the generation of
a turbulence structure. The turbulence structure is
generated by varying the radial porosity distribution.
This class of conditioners includes the Sens and Teule,
Bosch and Hebrard, K-Lab and Laws designs
this research.
Smart Flowmeters
-----------------------------------------------------------------
13 or 29
-----------------------------------------------------------------
conditioners.
If additional tests confirm this performance, the
installation of flow conditioners upstream of ultrasonic
flowmeters should lower the uncertainty by minimizing the
installation effects on its' performance. rn other
words, it is better to have a flat, repeatable velocity
profile, for ultrasonic flowmeters rather than a fully
developed one.
For compressible fluids, tests conducted by Gasunie and
Daniel rndustries indicate the viability of this
technology to natural gas applications. To date,
additional development effort is needed in the electronic
packages associated with these devices. The future looks
promising.
Coriolis Flowmeters
-----------------------------------------------------------------
14 of 29
-----------------------------------------------------------------
meters and turbine meters with various upstream
configurations have been conducted for several years to
aid in the design of high volume metering facilities.
The question posed by the user community is - "Can the
operator ensure the parties involved in the fiscal
transfer that the measurement station is adequately
described by the tested design?".
In response to the North American community's request,
the Gas Research Institute (GRI) has initiated funding
for assessment of in situ calibration devices - sonic
nozzles, master meters and piston provers. These tests,
planned for 1993, will be conducted at GRI's MRF
facility.
In situ calibration of flowmeters now appears feasible
-----------------------------------------------------------------
15 of 29
-----------------------------------------------------------------
Potentially,
opportunity
the
for
piston
successful
prover
field
offers the
calibration
flowmeters on multiple component natural gas streams.
However, the measurement community needs; additiona 1
best
of
research, an established gas piston prover design, and
certification standard. The additional research needs
will be assessed with the GRI activities for 1993. In
answer to the need for standards, the API committee on
Gas Measurement (COGM) has recently established a Working
Group to address the global community input and concerns.
7 SECONDARY DEV:ICES.
-----------------------------------------------------------------
16 of 29
4IIt -----------------------------------------------------------------
Field Calibration standards
with the advent of smart transmitters, field calibration
standards have been required to improve. Using SQC and
corrective software, field standards will improve in the
next five years due to market forces. After all, if the
transmitter is good for 0.05% of reading, why can't the
field standard be at least twice as good?
other Instrumentation
other secondary instrumentation will be required for
efficient and effective operations.
8. OPERATIONAL ENHANCEMENTS
Flowmeter Applications
-----------------------------------------------------------------
17 of 29
r
-----------------------------------------------------------------
to match with other devices. Twill be possible to
calibrate and inspect flowmetering devices on process
units prior to plant turnarounds.
The life of the orifice meter has been extended asa
direct result of "smart" transmitter technology and new
flow conditioner designs. This meter will continue to be
preferred on certain fluids because of its lower
sensitivity to density determination errors.
transmitters, chromatographs, etc.), required physical
inspection intervals (i.e.,videoimagescopes, etc.),
required certification of field standards on specified
intervals, and statistical footprinting of field devices
and standards.
Line Xnteqrity systems
The detection of leaks in pipelines presents a number of
technically challenging problems. Compounding the
technical difficulty of leak detection is the
environmental and safety concerns. The leak detection
response rate must be sufficiently fast and of a high
resolution.
As a practical matter, pipelines cannot afford to install
CTMs every 200 kilometers. Therefore, the number of
monitoring points should be minimized by the resolution
and response of the device.
Changes in temperature between monitoring points can
result in significant changes in fluid and pipe volumes.
Also, operating conditions such as batched operations,
elevation differences, and dynamic transients pose unique
problems.
Leadership
-----------------------------------------------------------------
18 of 29
-----------------------------------------------------------------
requirements. Leadership should be perceived as competent,
credible, and sharing a unified direction toward a stated
purpose by all customers.
Technology transfer and optimization of individual skills
should be an ongoing effort by all members of the Team.
*
prevention
Conformance
contractual
to the requirements
obligations, federal
(tariff
&
&
state
regulations, measurement standards, and customer
established requirements)
* Document measurement problems or nonconformances
(troubleslip system)
Maintain 5 year evergreen plan covering performance
*
-----------------------------------------------------------------
19 of 29
-----------------------------------------------------------------
commonly referred to as the "Ten Most Wanted". Manpower and
fiscal resources are then directed towards identifying the
"root cause" for their performance. The list is "rolling" in
nature. When a targeted loss unit's problems are defined and
corrected, it is removed from the list and replaced by the
ne~t loss unit in line.
The SPLC's Team is challenged by the breadth of operations
which covers crude oils, refined products, LPGs, chemicals and
C02 measurement applications. Measurement is a technically
demanding, complex, state of the art field with significant
impact on the profitability of any transportation system.
Technology, research and testing of hardware, instrumentation
and flow standards are ongoing efforts due to this state of
the art field. Innovation, creativity and realistic
assessment capabilities are required traits of any quality
measurement organization to adapt new technology with an
acceptable level of risk. This blend was the key to
establishing new levels of performance in SPLC while
maintaining low operating costs.
transportation systems and storage facilities within the next
five years.
Innovation and creativity by the SPLC Team resulted in
computerized automation of monthly loss control reports
thereby eliminating redundant effort throughout the
-organization. The report package is layered to provide
management with overview information, and measurement
specialists with two levels of detailed information. This
approach places the required information in the necessary
format for the appropriate individual.
certification of Technicians
with the complexity of future technology, the need will occur
for certification of measurement and laboratory technicians. tit
-----------------------------------------------------------------
20 of 29
'.
, -----------------------------------------------------------------
Eventually the ISO 9000 philosophy will provide market
distinction and enhanced profitability for companies with
visions.
Application of Quality concepts
-----------------------------------------------------------------
21 of 29
-----------------------------------------------------------------
9. SUMMARY
We have attempted to provide a vision of the measurement
community's direction for the next ten years. The
technological advances will be fast and fierce. Companies
should be positipned to take advantage of these
opportunities through a strategy, structure, systems and
people. The elements to success in the next ten years will
be flexibility, proper corporate cultures, innovation and
creativity, maximization of current capital investments,
and prudent capital investments.
-----------------------------------------------------------------
22 of 29
..
4IIt -----------------------------------------------------------------
"'_9_9_.2_rv_>C>_r_th __ S_-_-__ VV_>C>_r_k_S_h_>C>_P
__
Q
USA Sources of Energy
60.0%
50.0%
40.0%
30.0%
20.0%
10.0%
0.0%
Clear Vision
People
Strategy
Structure
Systems
-----------------------------------------------------------------
23 of 29
----------------
-------- ---- ---------------- ---------- -----------
ICRUDE OIL SYSTEM.
-
1"1
"L
1
.....
~
--
INATURAL GAS SYSTEM I
[.j.-----:> -=:
. ~
---------- 24 of 29
..
4IIJ -----------------------------------------------------------------
IREFINED PRODUCTS' SYSTEM I
ICHEMICAL SYSTEMI
.=
-~.-
I
-----------------------------------------------------------------
25 of 29
.
-----------------------------------------------------------------
-mstallaticn Effects
-Flow Conditioners
-Srnart Flowmeters
-Ultrasonic Flowmeters
-Coriolis Flowmeters
Flo-w- ..
2
,,
-.-
,
- - - - - --I
-----------------------------------------------------------------
26 of 29
..
-----------------------------------------------------------------
Q
Classification of Flow Conditioners
Typo Class Head Ratio Cost \
Tube Bundles
Radial I 1 Lo !
Hexagonal I 1 Lo ,
i
Etoile I 1 I Mod
AMCA Honeycomb I 1 , Mad ,
Mltsublshl II 2 I Lo
Hi
Zanker II 6 i
Sprenkle II 15 Hi i,
I
-~-
_.__ .~ __l ____
" L
SECONDARY DEVICES
Future Direction
-Srnart Transmitters
-Onlina QA Analyzers
-Portable QA Analyzers
-Portable Densitometers
-----------------------------------------------------------------
27 of 29
.
----------------------------------------------------------------- 1IIt
OPERATIONAL ENHANCEMENTS
Future Direction
-Flowmeter Applications
-Certltlcatlcn of Technicians
-Allocation Metering
.,_9_9_2
__ IV_~_,._~_h_:s_-_-__ VV_~_r_k_-_h_C>_P
__
Q
ECMTS PROJECT
DATA FLOW CHART
-----------------------------------------------------------------
28 of 29
....
4IIJ -----------------------------------------------------------------
Ethylene Pipeline
Annual Loss Control Performance
+
g
~ ~ ~ ~ U M
Year
~ 00 91 n ro
7_9_9_2
__ IV_~_r_t_h
__ .s_-_-__ ""__ ~_r_k_-_h_~_P__
Q
METER FACTOR CONTROL CHART
LACT METER Noncompensated : Meter 002
.. -
Qj
~
1. It' , :J
u. '00 " ~
::
~
I.OUO
1.1\0lIl
... Qj
a.
E
Qj
U. ~o" I-
:E 1._' ."
~
..
Qj
>
,- , o~
a.
1_0000 L,:-,:-,:-,:-:-, :-,::-,-=-, -=-, :":" -:Coo :7" 7:,,7:,.7:,,7:,.":':,,-::,,-::.. :-:,,0-:,,:-:,,;--_..J ~o 0
-+- TEMPERATURE
-----------------------------------------------------------------
29 of 29