You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267602920

Determination of Volume Fractions in a Two-Phase Flows From Sound Speed


Measurement

Conference Paper · August 2012


DOI: 10.1115/NCAD2012-1381

CITATIONS READS

2 1,860

3 authors, including:

Anirban Chaudhuri Dipen N. Sinha


Los Alamos National Laboratory Los Alamos National Laboratory
31 PUBLICATIONS   496 CITATIONS    149 PUBLICATIONS   626 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A frequency-domain technique for noninvasive, ultrasonic fluid detection in submerged sealed canisters View project

Noninvasive characterization of multiphase oil-water-gas flows View project

All content following this page was uploaded by Dipen N. Sinha on 08 March 2014.

The user has requested enhancement of the downloaded file.


An Algorithm for Determining
Volume Fractions in Two-Phase
Anirban Chaudhuri1 Liquid Flows by Measuring
Postdoctoral Research Associate
e-mail: anirban@lanl.gov Sound Speed
Curtis F. Osterhoudt2 This paper presents a method of determining the volume fractions of two liquid compo-
Scientist
nents in a two-phase flow by measuring the speed of sound through the composite fluid
e-mail: cfo@lanl.gov
and the instantaneous temperature. Two separate algorithms are developed, based on
earlier modeling work by Urick (Urick, 1947, “A Sound Velocity Method for Determining
Dipen N. Sinha the Compressibility of Finely Divided Substances,” J. Appl. Phys., 18(11), pp. 983–987)
Scientist and Laboratory Fellow
and Kuster and Toksöz (Kuster and Toksöz, 1974, “Velocity and Attenuation of Seismic
e-mail: sinha@lanl.gov
Waves in Two-Phase Media: Part 1. Theoretical Formulations,” Geophysics, 39(5), pp.
587–606). The main difference between these two models is the representation of the
Sensors and Electrochemical Devices (MPA-11),
composite density as a function of the individual densities; the former uses a linear rule-
Los Alamos National Laboratory,
of-mixtures approach, while the latter uses a nonlinear fractional formulation. Both
Los Alamos, NM 87545
approaches lead to a quadratic equation, the root of which yields the volume fraction (/)
of one component, subject to the condition 0  /  1. We present results of a study with
mixtures of crude oil and process water, and a comparison of our results with a Coriolis
meter. The liquid densities and sound speeds are calibrated at various temperatures for
each fluid component, and the coefficients are used in the final algorithm. Numerical
studies of sensitivity of the calculated volume fraction to temperature changes are also
presented. [DOI: 10.1115/1.4007265]

1 Introduction constant by means of a capacitance probe [4,13]. However, capac-


itance probes are subject to coating by paraffins that render them
Accurate measurement of the composition of oil-water mixtures
inaccurate within a short period of time; these type of sensors also
and emulsions within the process environment is one of the chal-
have problems when the medium becomes electrically conductive
lenging problems in the oil and gas extraction and petrochemical
(e.g., water continuous). Another method involves the laboratory
industries. It is important for reasons that include monitoring of
analysis of samples drawn from the pipelines [14], but this is labor
the inventory and quality of petroleum products, ensuring the
intensive, requires a substantial amount of time to complete, and
desired operating regime of the process plant (which may depend
does not allow the continuous monitoring of the liquid–liquid
on spatial and temporal distribution of phases in the mixed state)
system.
or identifying the environmental hazards such as discharge of oily
Ultrasonic techniques have also been applied to measure com-
water into the environment [1–3]. As the oil industry has begun to
position in two-phase liquid systems. For example, the ultrasonic
develop smaller and more marginal reservoirs, the demand for
measurement technique has been widely used for characterizing
compact and more cost-effective solutions for production and
heterogeneous mixtures in industrial processes. Its main advan-
monitoring of these wells has also increased. As an example, a
tages lie in the ability to (i) rapidly and noninvasively analyze
detailed discussion of the needs for identification of the oil-water
fluid mixtures, (ii) penetrate typically opaque highly concentrated
phase content in the primary separation systems, common within
mixtures, and (iii) not pose risks to human health (unlike X-ray or
the oil and gas extraction industry, is given in Ref. [4].
c-ray densitometry). There have been different studies investigat-
There are a number of techniques available to measure the
ing the use of ultrasonic time-of-flight to make composition meas-
phase content of oil-water mixtures. These vary from relatively
urements. A time average model, which assumed that measuring
simple off-line methods of sampling and establishing the
the travel time of sound through the dispersion and comparing it
volumetric content by gravity or centrifugal separation to more
with the travel times through pure phases enables one to calculate
sophisticated on-line and in situ methods relying on the interac-
the volume fraction of each phase in the dispersion, was used by
tion of particular “sensing fields” (such as ultrasonic [5,6], optical
Bonnet and Tavlarides [15] to estimate of the phase fraction of the
[7], electrical [8,9], microwave [10], X-ray [11], c radiation [12],
dispersed phase with reasonable accuracy. Correction factors were
etc.) with matter, which allow the user to deduce the phase content
later introduced [16,17] to improve the accuracy of the calculation
indirectly. Multiphase meters should preferably have nonintrusive
by considering refraction and reflection phenomena of sound on
sensors for several reasons, including the elimination of pressure
the droplet–water interface; this was verified at low volume
drop over the instrument, minimizing the impact on the flow, pos-
fractions (1%) of the dispersed phase [6]. There have also been
sibility of a clamp-on device, and the absence of detector corro-
studies to determine the concentration of solids in a three-phase
sion. One of the most common methods of measuring oil-water
slurry reactor by measuring the velocity of ultrasound in slurries
ratios in commercial use is the measurement of the dielectric
[18]. Regression based methods have been used in the food and
chemical industries too; Krause et al. [19] recently presented a
1
Corresponding author. multivariate regression method for simultaneous detection of
2
Present address: University of Alaska, Anchorage, AK. sugar and ethanol concentrations in aqueous solutions via
Contributed by Fluids Engineering Division of ASME for publication in the JOUR-
NAL OF FLUIDS ENGINEERING. Manuscript received October 4, 2011; final manuscript
temperature-dependent ultrasonic velocity. The inherent compro-
received July 19, 2012; published online September 24, 2012. Assoc. Editor: Mark mise in the choice of the frequency of the ultrasound between
R. Duignan. maximizing spatial resolution and ensuring adequate beam

Journal of Fluids Engineering Copyright V


C 2012 by ASME OCTOBER 2012, Vol. 134 / 101301-1

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 11/29/2013 Terms of Use: http://asme.org/terms


penetration, along with limitations of ultrasonic wave propagation and
as a nonintrusive diagnostic technique, have also been discussed
in literature [20,21]. q q ð1  /Þ þ qo ð2 þ /Þ ð1  /Þ þ rð2 þ /Þ
¼ w ¼ (5)
In this paper, an algorithm to calculate the volume fractions by qw qw ð1 þ 2/Þ þ 2qo ð1  /Þ ð1 þ 2/Þ þ 2rð1  /Þ
measuring sound speed in a two-phase liquid flow is developed.
The composite density is represented by two different formula- respectively, where r ¼ qo =qw .
tions, both of which have been previously shown [5] to represent In both models, the fluid compressibility is obtained from a lin-
properties of mixtures very closely. Note that the current exercise ear rule-of-mixtures approach for homogeneous liquids as follows
is in reverse compared to earlier studies; while others have [5,25,27]:
focused on calculating the density of a binary fluid mixture using
different (linear and nonlinear) mixing laws, our efforts were to j ¼ /jo þ ð1  /Þjw (6)
find the volume fractions while assuming that the mixing laws are
valid. Although the examples presented here pertain to oil-water where / is the oil volume fraction ð0  /  1Þ
mixtures, the derivation is applicable to any binary liquid mixture From Eq. (1), we can write
where the individual components have distinct sound speeds
within the temperature range of interest. Nondimensional parame- 1 1
ters are defined to simplify the analyses and allow better interpre- c2o ¼ ; c2w ¼ (7)
jo qo jw qw
tation of experimental data. The algorithm was applied to detect
fluid mixture ratios at two separate oil fields with different oil vol-
ume fractions and API gravities. The results have been compared Using the above expressions, we get
with readings from industry standard metering techniques.
jo jo qo qw c2w 1 o
¼  ¼  ¼ (8)
jw jw qw qo c2o r rw
2 Modeling Two-Phase Flows
The two models used for calculating individual volume fraction where o ¼ c2 =c2o and w ¼ c2 =c2w . Note that from these definitions,
(referred to as the “oil-cut” in oil-water mixtures) / in this paper r, o, and w are all positive quantities. Rewriting Eq. (6) and apply-
are those proposed by Urick [22] and Kuster–Toksöz [23]. Both ing Eq. (8),
models assume that (i) the wavelength of the acoustic waves is
much longer than the characteristic size of the dispersed particles/ j jo  o
phase, and (ii) multiple scattering effects of the sound field are ¼ / þ ð1  /Þ ¼ 1  / 1  (9)
jw jw rw
negligible. The calculation of the velocity of sound c in such a
homogeneous, nonscattering mixture of two fluids is given by By multiplying Eqs. (9) and (4), we obtain, for the Urick model
Ref. [24]
jq h  o i
pffiffiffiffiffiffi ¼ 1/ 1  ½1  /ð1  rÞ (10)
c ¼ 1= qj (1) jw qw rw

where q is the density of the composite medium and j is the adia- Similarly, by multiplying Eqs. (9) and (5), we obtain, for the
batic compressibility. Equation 1 also relates the sound speed, Kuster–Toksöz model
density, and compressibility of the two individual liquid compo-
h   
nents, oil and water. Throughout the rest of this paper, the oil and jq o i ð1  /Þ þ rð2 þ /Þ
water phases are represented with the subscripts o and w respec- ¼ 1/ 1  (11)
tively. For the tests analyzed in this paper, the crude oil and pro- jw qw rw ð1 þ 2/Þ þ 2rð1  /Þ
cess water mixtures are homogenized and the droplet sizes are
much smaller than the wavelengths (1 mm) used to interrogate The left-hand side is equal to 1=w for both Eqs. (10) and (11), and
the composite fluid; hence, both assumptions stated above are ap- they each can now be simplified to a quadratic equation of the
plicable [25]. form
The difference between these two models is the computation of
the composite fluid density q as a function of the two component A/2 þ B/ þ C ¼ 0 (12)
densities qo and qw [26]. The Urick model applies a simple linear
combination based on the definition of volumetric oil-cut3 as where the coefficients A, B, and C are listed in Table 1 in terms of
follows: the nondimensional parameters r, o, and w. This nondimensional
formulation also allows us to choose arbitrary units for c and r, as
q ¼ /qo þ ð1  /Þqw (2) long as they are the same for the two phases. Solution of the quad-
ratic equation yields two roots of the form
while the Kuster–Toksöz model uses
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
qw  q q  qo B 6 B2  4AC
¼/ w (3) /¼ (13)
qw þ 2q qw þ 2qo 2A

and the root corresponding to 0  /  1 is used.


Note that both models satisfy the conditions q ¼ qo for / ¼ 1 and
q ¼ qw for / ¼ 0. By simple algebraic manipulations, Eq. (2) and
Eq. (3) can be modified to
q q Table 1 Coefficients of quadratic equation
¼ / o þ ð1  /Þ ¼ 1  /ð1  rÞ (4)
qw qw
Coefficient Urick model Kuster–Toksöz model

3
From the definition of oil-cut, / ¼ Vo =ðVo þ Vw Þ, where it is assumed that the A ðo  rwÞð1  rÞ ðo  rwÞð1  rÞ
two components, oil and water in this case, are immiscible, and the total mixed vol- B 2rw  o  r2 w 2rw  o þ r 2 w þ 2rð1  o  rÞ
ume is the sum of the individual volumes Vo and Vw . Applying conservation of mass, C rð1  wÞ rð1 þ 2rÞð1  wÞ
qðVo þ Vw Þ ¼ qo Vo þ qw Vw ) q ¼ qo / þ qw ð1  /Þ.

101301-2 / Vol. 134, OCTOBER 2012 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 11/29/2013 Terms of Use: http://asme.org/terms


3 Fluid Characterization (DSA 5000 M) was then used for the calibration; the published
accuracies of this equipment are 0.000005 g/cm3 and 0.5 m/s for
Due to the strong dependence of both the density and sound
density and sound speed measurements, respectively. The process
speed on temperature ðTÞ, the two components of the liquid mix-
water samples obtained from the two sites have very different
ture have to be calibrated over a range of temperatures. For our
acoustic properties, mainly due to the presence of varying
tests, samples of the component fluids (one of crude oil and one of
amounts of salts and chemicals that are characteristic of the rock
process water) from each oil field were obtained and the corre-
substrate in an individual oil field [28]. For all four samples we
sponding densities and sound speeds at different temperatures
obtained, third order polynomials were sufficient to get a good
were fitted with Nth-order polynomials of the form
representation of the sound speeds and densities over the range of
X
N X
N temperatures of interest.
co ¼ aoi T i ; cw ¼ awi T i ; The nondimensional parameters used to calculate the coeffi-
i¼0 i¼0 cients (Table 1) of the quadratic Eq. (12) are plotted in Fig. 2,
X
N X
N along with the calculated values of volume fraction as a function
qo ¼ boi T i ; qw ¼ bwi T i of temperature. We observed that for this particular pair of fluid
i¼0 i¼0 samples, jo =jw > 1 (compressibility of the crude oil samples was
greater than the process water samples) and r < 1 (density of
where aoi and awi are the calibration coefficients for sound speeds crude oil sample is lower than the corresponding process water
in crude oil and water respectively, while boi and bwi are the cali- sample from the same oil field) at all measured temperatures; this
bration coefficients for crude oil and water densities, respectively. observation will be used in Sec. 4 to determine the properties of
Tests were conducted in the laboratory to measure sound speed the roots of the quadratic equation in Eq. (12) having the coeffi-
and density in the individual liquid phases at different tempera- cients listed in Table 1. The plots of / for two different values of
tures, the results of which are plotted in Fig. 1 for samples from c illustrate the importance of good temperature characterization of
two different oil fields. The crude oil produced was different at the samples and the complex nature of the problem at hand since
each field, both in API gravity as well as sound speed. To get the there are an infinite number of combinations of / and temperature
pure individual phases, a sample of the liquid mixture was drawn (T) that can lead to the same sound speed in the mixture.
from the oil pipeline and then separated into its constituents using Plots in Figs. 3 and 4 show the variation in volume fraction /
a centrifuge. An Anton Paar density and sound velocity meter as a function of the two independent variables c and T for the two
fluid sample pairs in our study. These calculations were carried
out using the Kuster–Toksöz model; the values using the Urick
model were almost the same. Keeping temperature constant, oil
volume fraction (/) decreases as the measured sound speed (c)
increases, as shown in Figs. 3(a) and 3(b). This is expected since
the speed of sound in pure crude oil is lower than that in the pure
process water, as observed in the calibrations (Fig. 1). The magni-
tude of the slopes of the curves are also greater at the lower tem-
peratures since the acoustic properties (sound speed) of the
individual components (crude oil and process water, in this case)
are more similar in this regime; in other words, at lower tempera-
tures, smaller changes in measured sound speed can lead to larger
changes in calculated volumetric composition. The greater slope
is expected since the volume fraction (/) changes by 100% within
a shorter range of sound speeds at lower temperatures, and vice

Fig. 2 Change in nondimensional parameters r 5 qo =qw ,


jo =jw , and / as a function of temperature. Solid lines are for
samples from oil field 1, while samples from field 2 are repre-
Fig. 1 Measured sound speeds and densities of the constitu- sented by dashed lines. The values of / were calculated for two
ent crude oil and processed water components with varying different sound speeds, 1450 m/s and 1500 m/s. The values of r
temperature in two different sets of samples (from oil fields 1 change from 0.883 to 0.867 for oil field 1, and from 0.787 to
and 2) 0.773 for oil field 2.

Journal of Fluids Engineering OCTOBER 2012, Vol. 134 / 101301-3

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 11/29/2013 Terms of Use: http://asme.org/terms


Fig. 3 Variation in calculated oil-cut as a function of measured
sound speed (using Kuster–Toksöz model), at different temper- Fig. 4 Variation in calculated oil-cut as a function of measured
ature levels that are characteristic of the respective field temperature (using Kuster–Toksöz model), at different sound
speeds

versa; for example, in Fig. 3(a), / changes from 1 to 0 while c


changes from 1320 m/s to 1541 m/s at 40  C, while the same range and
is covered while c changes from 1217 m/s to 1561 m/s at 70  C.
The trends for change in / as a function of temperature for jo qw c2w o
a particular measured sound speed, shown in Figs. 4(a) and 4(b), ¼  1 ¼)  1 ¼) ðo  wrÞ  0 (16)
jw qo c2o wr
however, vary greatly and a general conclusion cannot be reached.
One notable feature in Figs. 4(a) and 4(b) is the presence of a
stationary point in the lines representing c ¼ 1500 m/s, which From Eqs. (15) and (16), we can conclude that for both the
suggests that the same fluid composition (i.e., same value of /) Urick and Kuster–Toksöz models
can result in the same characteristic sound speed c at two different
temperatures. A ¼ ðo  wrÞð1  rÞ  0 (17)

Equation (14) also implies that


4 Analysis of Quadratic Roots
 2  2
In order to analyze the roots of the quadratic equation in c c
o¼  1 and w ¼ 1
Eq. (12), we start by assuming that for a certain temperature range co cw (18)
T 2 ½Tlow ; Thigh , the following relations apply: ¼) ð1  oÞ  0 and ð1  wÞ  0
qw  qo and cw  c  co (14)
The values of volume fraction / are plotted in Fig. 5 for a range
where c is the sound velocity measured in the composite two- of expected values of c/co and c/cw , as per Eq. (18), and show the
phase fluid. Using this, we can write typical variation in the dependent variable. This plot clearly shows
that the variation in / is nonlinear and is not as simple as the usual
rule-of-mixtures approach for mixture density (Fig. 2). Note that
qo
r¼  1 ¼) ð1  rÞ  0 (15) the density ratio r was kept constant for simplicity while getting
qw to know the nature of the variation; this, however, will not happen

101301-4 / Vol. 134, OCTOBER 2012 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 11/29/2013 Terms of Use: http://asme.org/terms


o=w ¼ c2w =c2o  1
¼) o=w  r  1r
¼) w½rð1  rÞ  ðo=w  r Þ  w½rð1  rÞ  ð1  rÞ
¼) B  wð1  rÞ2
¼) B  0

for the Urick model. By a similar approach, the same inequality


can be proved for the Kuster–Toksöz model too. Hence, in both
cases, the sum of the roots satisfies the inequality

B
 0
A

from which it can be concluded that the roots of the quadratic


equation are always positive.
Note that for the Urick model,
Fig. 5 Calculated values of / for a range of values of c. r 5 0:8
was selected for this plot.
B 2rw  o  r 2 w
 ¼
2A ðo  rwÞð1  rÞ
ðo=w  1Þr
in real cases since r is also a function to temperature T (Fig. 2). ¼1þ (20)
ðo=w  rÞð1  rÞ
Fluid characterization results (shown in Sec. 3) show that a mean
value of r ¼ 0:8 is representative of the samples used in this
study. While w ¼ 1 corresponds to / ¼ 0 (all water), o ¼ 1 corre-
sponds to / ¼ 1 (all oil); this can be proved by substituting in the
quadratic coefficients and calculating roots.
For the Urick model, if o ¼ 1, i.e., c ¼ co , then

A ¼ ð1  wrÞð1  rÞ

B ¼ 2rw  o  r 2 w

C ¼ ð1  wÞr

¼) B2  4AC ¼ ð1  2r þ r 2 wÞ2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
¼)  B þ B2  4AC ¼ 2ð1  rÞð1  rwÞ
¼) / ¼ 1

Also, when w ¼ 1, i.e., c ¼ cw , then

C¼0
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
B þ B2  4AC
¼) /¼ ¼0
2A

The variation of / is nonlinear with sections of higher slope


near the extremities (i.e., o ¼ 1 and w ¼ 1), which implies
that the variation in calculated value of the fluid volume fraction
(/) due to small changes in measured sound speed (c) will be
higher in these regions too. This is also verified in the next subsec-
tion on the sensitivity of derived volume fraction to measured
parameters.
Equation 18, along with the fact that 0  r  1, implies that

C0 (19)

for both models. Equations (17) and (19) tell us that

C
0
A
Fig. 6 Comparison between different methods of oil-water
fraction determination at two oil fields, where “new model”
from which it can be concluded that both roots of the quadratic refers to the algorithm developed in this paper starting from the
Eq. (12) are of the same sign for the set of assumptions in Kuster–Toksöz model. Note that the y-axes are in terms of per-
Eq. (14). Note that centage water-cut.

Journal of Fluids Engineering OCTOBER 2012, Vol. 134 / 101301-5

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 11/29/2013 Terms of Use: http://asme.org/terms


which is independent of the instantaneous measured sound speed entire test) of water fraction using the new model derived in this
c; hence, the roots of Eq. (12) are dependent only on the value of paper were within 1% of the mean values obtained from the other
the discriminant at different values of c. standard methods.

5 Results 5.2 Sensitivity Studies. A numerical study of the sensitivity


5.1 Field Tests. Long duration tests with flowing liquid mix- of the oil-cut ð/Þ as a function of the measured temperature and
ture from oil wells were carried out at the two test sites and com- sound speed was also carried out for both sets of samples. This
pared with results from industrially recognized techniques (using was important since it yielded a good estimate of the effect of any
a Coriolis density meter and manual sampling); the results are measurement error in the independent variables ðc; TÞ on /.
summarized in Fig. 6. The Coriolis meters had published density Figures 7(a) and 8(a) show the change in oil-cut value as a func-
measurement accuracy of 60:0002 g/cc, which was then used to- tion of the measured sound speed c at different measured tempera-
gether with Eq. (2) to find oil volume fraction. The sound speed ture levels. To interpret the plots, let us assume that the measured
through the composite liquid mixture in the pipeline was calcu- temperature is 70  C at oil field 1 and the calculated sound speed
lated by transmitting linear frequency chirps generated by acoustic is 1500 m/s. The corresponding values of oil volume fraction /
transducers placed on the outer wall of a steel pipe and then apply- and sensitivity @/=@c from Fig. 7(a) are 11% and 0.34%/m/s,
ing a cross correlation algorithm to derive the time of flight. The respectively, and imply that a 1 m/s error in measuring sound
compositions of the fluids at the two test sites were different from speed will lead to 0.34% error in oil-cut calculation.
each other; the fluid mixture at oil field 1 contained water predom- A similar estimate can be derived for temperature errors
inantly, while the oil content at oil field 2 averaged nearly 20% by too, assuming that sound speed has been measured accurately
volume. The average flow rates through the pipe were also differ- (Figs. 7(b) and 8(b)). For example, if the measured sound speed at
ent at the two locations; the first test oil field had mean flow rates oil field 2 is 1450 m/s and the temperature recorded is 20  C, then
of 1000 barrels per day (0.4 m/s through a 77.9 mm diameter an error of 1  C in measuring temperature will lead to almost 1%
pipe), while the second oil field averaged 6000 barrels per day error in calculated oil-cut (Fig. 8(b)).
(2.4 m/s through a 77.9 mm diameter pipe). It was observed that the sensitivity magnitudes are higher for
As per industry methodology, the water-cut, or percentage the composite fluid mixture at oil field 2, which operates at lower
volume content of water, was calculated during the tests using cal- temperatures; since the sound speeds of the individual fluid com-
ibration coefficients from Sec. 3. It was observed, from Figs. 6(a) ponents are much closer to each other at lower temperatures than
and 6(b), that the mean values (calculated over the duration of the at higher temperatures (Fig. 1) an error in measuring the

Fig. 7 Sensitivity of oil-cut calculation on changes in inde- Fig. 8 Sensitivity of oil-cut calculation to changes in inde-
pendent measurements for samples from oil field 1 pendent measurements for samples from oil field 2

101301-6 / Vol. 134, OCTOBER 2012 Transactions of the ASME

Downloaded From: http://fluidsengineering.asmedigitalcollection.asme.org/ on 11/29/2013 Terms of Use: http://asme.org/terms


composite fluid property (sound speed or temperature) leads to [4] Jaworski, A. J., and Dyakowski, T., 2005, “Measurements of Oil–Water Separa-
greater error in determining the volumetric composition. tion Dynamics in Primary Separation Systems Using Distributed Capacitance,”
Flow Meas. Instrum., 16(2–3), pp. 113–127.
[5] Meng, G., Jaworski, A. J., and White, N. M., 2006, “Composition Measure-
6 Conclusions ments of Crude Oil and Process Water Emulsions Using Thick-Film Ultrasonic
Transducers,” Chem. Eng. Process., 45(5), pp. 383–391.
This paper presents two algorithms, based on models developed [6] Tsouris, C., and Tavlarides, L. L., 1993, “Volume Fraction Measurements of
earlier by Urick [22] and Kuster–Toksöz [23], to calculate Water in Oil by an Ultrasonic Technique,” Ind. Eng. Chem. Res., 32(5), pp.
998–1002.
individual volume fractions in a two-phase fluid mixture by meas- [7] Jana, A. K., Mandal, T. K., Chakrabarti, D. P., Das, G., and Das, P. K., 2007,
uring sound speed and temperature in the composite fluid, and the “An Optical Probe for Liquid–Liquid Two-Phase Flows,” Meas. Sci. Tech.,
algorithms have been applied to two different mixtures of crude 18(5), pp. 1563–1575.
oil and processed water. The results of applying the algorithms in [8] Garcı́a-Golding, F., Giallorenzo, M., Moreno, N., and Chang, V., 1995, “Sensor
for Determining the Water Content of Oil-in-Water Emulsion by Specific Ad-
real time to samples from two different oil fields have been pre- mittance Measurement,” Sensor. Actuator. A: Phys., 47(1–3), pp. 337–341.
sented, along with comparisons with current industry accepted [9] Johansen, G. A., Frøystein, T., Hjertakery, B. T., and Olsen, Ø., 1996, “A Dual
techniques; the average values using the new algorithms are close Sensor Flow Imaging Tomographic System,” Meas. Sci. Tech., 7(3), pp.
to other metering techniques (within 1%). Some notable conclu- 297–307.
[10] Yang, Y., Scott, B., and Cregger, B., 1990, “The Design, Development and
sions of this study are: Field Testing of a Water-Cut Meter Based on a Microwave Technique,” 65th
Annual Technology Conference of the Society of Petroleum Engineers (SPE),
(i) The calculations for volume fraction (/) are a function of New Orleans, LA, Sept. 23–16, pp. 775–782.
the density and sound speed in each component, calibrated [11] Heindel, T. J., 2011, “A Review of X-ray Flow Visualization With Applications
for a range of operating temperatures. to Multiphase Flows,” ASME J. Fluids Eng., 133(7), p. 074001.
(ii) Both models used to represent the composite density of [12] Abro, E., and Johansen, G. A., 1999, “Improved Void Fraction Determination
by Means of Multibeam Gamma-Ray Attenuation Measurements,” Flow Meas.
the fluid mixture result in a quadratic equation in / with Instrum., 10(2), pp. 99–108.
nondimensional coefficients. [13] Meng, G., Jaworski, A. J., and Kimber, J. C. S., 2006, “A Multi-Electrode Ca-
(iii) Some properties of the roots can be determined based on pacitance Probe for Phase Detection in Oil-Water Separation Processes: Design,
the values of the nondimensional quantities, subject to cer- Modelling and Validation,” Meas. Sci. Tech., 17(4), pp. 881–894.
[14] Yang, M., 2011, “Measurement of Oil in Produced in Water,” Produced Water:
tain assumptions for the physical properties of the fluid Environmental Risks and Advances in Mitigation Technologies, Springer, New
components (r < 1 and jo =jw > 1). York, pp. 57–88.
(iv) A numerical study of the sensitivity of the result to errors [15] Bonnet, J. C., and Tavlarides, L. L., 1987, “Ultrasonic Technique for
in measured temperature and sound speed shows that Dispersed-Phase Holdup Measurements,” Ind. Eng. Chem. Res., 26(4), pp.
811–815.
higher errors occur when the properties (sound speed) of [16] Yi, J., and Tavlarides, L. L., 1990, “Model for Hold-Up Measurements in Liq-
the two component fluids are similar to each other within uid Dispersions Using an Ultrasonic Technique,” Ind. Eng. Chem. Res., 29(3),
the temperature range of interest. pp. 475–482.
[17] Tsouris, C., and Tavlarides, L. L., 1990, “Comments on ‘Model for Hold-Up
Measurements in Liquid Dispersion Using an Ultrasonic Technique’,” Ind. Eng.
Chem. Res., 29(10), pp. 2170–2172.
Acknowledgment [18] Soong, Y., Gamwo, I. K., Harke, F. W., Blackwell, A. G., Schehl, R. R., and
This work was supported by Chevron U.S.A. Inc. Zarochak, M. F., 1997, “Measurement of Solids Concentration by an Ultrasonic
Transmission Technique,” Chem. Eng. Tech., 20(1), pp. 47–52.
[19] Krause, D., Schöck, T., Hussein, M. A., and Becker, T., 2011, “Ultrasonic Char-
acterization of Aqueous Solutions With Varying Sugar and Ethanol Content
Nomenclature Using Multivariate Regression Methods,” J. Chemometrics, 25(4), pp.
/¼ volume fraction of crude oil 216–223.
[20] Atkinson, C. M., and Kytömaa, H. K., 1992, “Acoustic Wave Speed and
j¼ compressibility of two-phase liquid mixture Attenuation in Suspensions,” Int. J. Multiphase Flow, 18(4), pp. 577–592.
q¼ density of composite two-phase liquid mixture [21] Atkinson, C. M., and Kytömaa, H. K., 1993, “Acoustic Properties of Solid-Liquid
c¼ speed of sound through two-phase liquid mixture Mixtures and the Limits of Ultrasound Diagnostics—I: Experiments,” ASME
V¼ volume J. Fluid. Eng., 115(4), pp. 665–675.
[22] Urick, R. J., 1947, “A Sound Velocity Method for Determining the Compressi-
API ¼ American Petroleum Institute bility of Finely Divided Substances,” J. Appl. Phys., 18(11), pp. 983–987.
[23] Kuster, G. T., and Toksöz, M. N., 1974, “Velocity and Attenuation of Seismic
Waves in Two-Phase Media: Part 1. Theoretical Formulations,” Geophysics,
39(5), pp. 587–606.
References [24] McClements, D. J., and Povey, M. J. W., 1989, “Scattering of Ultrasound by
[1] Whitaker, T. S., 1996, “Multiphase Flow Measurement: Current and Future Emulsions,” J. Phys. D: Appl. Phys., 22(1), pp. 38–47.
Developments [for Offshore Industry Use],” Institution of Electrical Engineers [25] McClements, D. J., 1991, “Ultrasonic Characterisation of Emulsions and Sus-
(IEE) Colloquium on Advances in Sensors for Fluid Flow Measurement, Lon- pensions,” Adv. Colloid Interfac., 37(1-2), pp. 33–72.
don, UK, IEE, pp. 1/1–111. [26] Caleap, M., Drinkwater, B. W., and Wilcox, P. D., 2012, “Effective Dynamic
[2] Thorn, R., Johansen, G. A., and Hammer, E. A., 1999, “Three-Phase Flow Mea- Constitutive Parameters of Acoustic Metamaterials With Random Micro-
surement in the Offshore Oil Industry: Is There a Place for Process structure,” New J. Phys., 14(3), p. 033014.
Tomography?,” First World Congress on Industrial Process Tomography, Bux- [27] Ament, W. S., 1953 “Sound Propagation in Gross Mixtures,” J. Acoust. Soc.
ton, Greater Manchester, April 14–17, pp. 228–235. Am., 25(4), pp. 638–641.
[3] Jaworski, A. J., and Meng, G., 2009, “On-Line Measurement of Separation [28] Veil, J. A., Puder, M. G., Elcock, D., and Redweik, R. J., 2004, “A White Paper
Dynamics in Primary Gas/Oil/Water Separators: Challenges and Technical Describing Produced Water From Production of Crude Oil, Natural Gas, and
Solutions—A Review,” J. Petrol. Sci. Eng., 68(1–2), pp. 47–59. Coal Bed Methane,” Argonne National Laboratory, Technical Report.

Journal of Fluids Engineering OCTOBER 2012, Vol. 134 / 101301-7

DownloadedViewFrom:
publicationhttp://fluidsengineering.asmedigitalcollection.asme.org/
stats on 11/29/2013 Terms of Use: http://asme.org/terms

You might also like