Professional Documents
Culture Documents
.
SPE 46236
E!!!! m
o.
Society of PetroleumEngineers
studied for over twenty years. 16-18 Measured impedance compute steam quality, x, assuming the vapor and liquid -
(conductance or capacitance) of a two-phase fluid depends phases flow through the sensor at equal velocities.
upon the sensor geometry and the liquid volume fraction.
In its simplest form, a capacitance transducer consists of r.. -1-1
two parallel conducting plates separated by a dielectric (6)
L (vL) l-~J
medium such as water. The capacitance is proportional to the
area of the plates, A, and inversely proportional to the distance
between them, d. The constant of proportionality for the area The annular flow geometry of the DSQS device tends to
of the plates is given by the product of the dielectric constant, promote mixing of the vapor and liquid phases. However, if
K, and the dielectric constant of a vacuum, &o.’9
the vapor and liquid velocities are not equal, then a slip
velocity factor must be included in equation (6). Introduction
C = K@/d (1) of a slip factor, however, adds an unknown variable that must
be determined by yet another measurement. Therefore, it is
For a given transducer geometry, the change in capacitance important that the steam flow conditions entering the DSQS
is directly proportional to the change in the dielectric constant device be as homogeneousas possible.
of the medium between the plates. If the medium is a gas-
Iiquid mixture, then dielectric constant, K, is determined from Field Tests
the liquid volume Ilaction, 1, the dielectric constant of the gas The purpose of the field tests was to evaluate the performance
phase, KG, and the dielectric constant of the liquid phase, KL. of a commercially available DSQS device installed at an
injection well for an extended range of steam conditions. A
K=(l-k)KG+~KL (2) steam separator vessel, installed directly downstream of the
device, was the primary means of validating the DSQS quality
Originally, the DSQS device was designed to measure the readings.
capacitance of deionized water vapor and liquid. Dionized Equipment and Procedure. In recent years, the DSQS
water is a nonconducting liquid (i.e., infinite electrical device has been modified to improve field performance and
resistance). However, oil field water contains dissolved allow easier installation. Key components have been upgraded
solids that make the liquid phase conductive. Consequently, to ensure reliable operation at steam pressures up to 1000 psig
the commercially available DSQS device measures (545 “F). The electrode length has been reduced from 18“ to
impedance, which is the vector resultant of capacitive 12” and the housing ID reduced from 2“ to 1.5”. A schematic
reactance, XC,and resistance, R. of a standard 2“ nominal DSQS device for injector installation
is shown in Figure 2. This “off-the-shelf’ configuration was
Z,o,a,
= (x, + R*)”2 (3) used throughout our testing.
A simplified schematic of a steam separator is shown in
The liquid phase of the steam has a much lower impedance Figure 3. The liquid and vapor phases are separated in the
than the vapor phase. As the volume fraction of the vapor vessel and a fluid level is maintained with a level indicator
phase increases or decreases, so does the measured and control valve. The volume flow rate of each phase is
impedance. measured using orifice plates or vortex meters. The same two
Each dielectric sensor system (mechanical components and separators were used throughout our testing.
electrical cable) has its own characteristic “background” The test setup, shown in Figure 4, was designed to take
impedance. Calibration of each sensor allows for the steam output from a 22 MMBtu/hr portable gas-fired
subtraction of “background” impedance from total impedance generator. The total steam output from the generator was
to give the impedance of the two-phase steam. alternately directed through one of two side-branching tees
and the split streams were injected into two nearby injectors.
z = Z,o,al
- Z,cnwr (4) Generator steam quality was varied by adjusting the fhel and
steam
feedwater rates.
The liquid volume fraction, A, is then determined from the The steam flow rates to each injector were controlled using
steam impedance, Z,tC,~,the steam pressure, P, the water or wellhead chokes. Thermocouples were installed upstream and
liquid phase resistivity, pW,and the impedance of the vapor downstream of each choke to monitor steam temperatures
phase, Z,. (and saturation pressures). Critical flow was achieved at each
1 = K (pW* P)’c Log(-Z,~Z,,,J (4) choke to maintain stable flow conditions during testing. A
DSQS device and separator vessel were installed downstream
The empirical constant, K, takes into account the geometry of each choke to measure quality (and rate) at each injector.
and mechanical components of the sensor. The liquid volume Data were collected for at least 30 minutes (under stable
fraction and specific volumes, Vv and VL, are then used to conditions) before changing to the next test case.
b
Data Analysis. The separator vapor and liquid flow rates vertically) to Phase II through IV (sensors positioned
were analyzed to determine the steam flow rates and qualities horizontal). The DSQS devices, that were positioned so that
exiting each branch of the tee. The separator data were the steam flowed vertically up through the electrode-housing
adjusted to pressure conditions upstream of the choke to annulus, consistently measured lower qualities than the
correct for liquid flashing as a result of the large pressure separator. This effect, which was especially pronounced at
drop. Isenthalpic throttling across each choke was assumed to low steam rates and qualities seen at Sensor 1, was reduced
obtain steam qualities at upstream pressure conditions. The significantly once the DSQS devices were installed
total adjusted liquid and vapor rates exiting the tee were then horizontally.
compared to the generator outlet data to ensure that the steam
mass flow rate and thermal energy were balanced for the Discussion
system. There are various reasons for the differences between the
The separator steam quality measurements were used as the DSQS and separator quality measurements. First, uncertainty
primary means of evaluating the performance of each DSQS is inherent in any measurement process and needs to be
device. Liquid and vapor volume flow rate measurements, Qv considered when comparing DSQS and separator steam
and Q~, at each separator were also used to estimate the qualities. Second, the DSQS device is more sensitive to inlet
superficial vapor and liquid velocities and the liquid volume flow conditions than the separator. Third, the DSQS device is
fraction entering each DSQS device. sensitive to changes in water resistivity, whereas the separator
is not.
v ~v = Q,/(Ac * 3600) (7) Measurement Uncertainty. As a rule of thumb, all metering
devices should be reviewed carefidly for inherent errors
associated with the measurement process and to determine the
v~L= Q,/(Ac * 3600) (8) impact of simplifying assumptions applied to governing
principles or equations on overall reliability and accuracy.
Sensitivity studies were conducted to estimate the relative
L.5Cp =(1 -XJVJ[( 1-X3=P)VL+
X,epv” (9) error contributions of various measurement parameters on
overall measurement uncertainty for a two-phase separator
The superficial velocities and liquid volume fraction and a DSQS device. A PC Windows application called
provide additional information for evaluating DSQS UncertaintyAnalyzer_ greatly facilitated the process.
performance at different inlet flow conditions. Heuristic estimates of error limits and confidence IeveIs were
Test Results. Test data are summarized in Tables 1-4. The entered into the program and weighted error contributions
tests were conducted in different Phases (1 through IV) over were computed.20
several months. Different pairs of DSQS devices, designated Steam Separator. The two-phase separator is considered to
Sensor 1 and Sensor 2, were used during each phase of be the most reliable method for measuring both steam rate and
testing. Sensor 1 and Sensor 2 data depict the DSQS qualities quality. It measures the liquid and vapor rates separately
measured at injectors 1 and 2, respectively. using well-established single-phase flow technology. The total
Inlet steam conditions for Sensor 1 ranged tlom 150 to 380 steam mass flow rate, m$ta~,is determined from the vapor and
psia, 75 to 650 b/d ewe, and 20 to 10OOA quality. Inlet steam liquid flow rates, Qv and Q~ and the phase specific volumes,
conditions for sensor 2 ranged from 180 to 450 psia, 315 to Vv and v~. The steam quality is determined by dividing the
900 bld ewe, and 20 to 90% quality. The vapor velocity and vapor mass flow rate by the total mass flow rate.
liquid volume fraction entering Sensor 1 ranged from 30 to
100 ftkec and 0.001 to 0.01, respectively. The vapor velocity m,,,,~ = Q\, f Vv+ Q. I V, (lo)
and liquid volume fraction entering Sensor 2 ranged from 40
to 150 fthec and 0.001 to 0.04, respectively. Qv /Vv (11)
x=
Figures 5 through 8 show comparisons of DSQS and Qv/vv+Q~/vL
separator quality measurements. With few except ions, the
DSQS qualities were within +/- 10’XO (quality percentage units)
A rigorous uncertainty analysis would of course include
of separator qualities for the entire range of inlet flow
meter factors, pulsed signal output from the vortex meters,
conditions. The DSQS and separator measurements were even
and steam pressure. However, for the purposes of this study,
closer for qualities above 40’?40.At lower qualities, the DSQS
uncertainty analysis of separator quality measurements were
measurements were consistently lower, but for the most part
conducted using estimated error limits, and associated
were still within -1O% quality units of separator
confidence levels, for the volume flow rates and specific
measurements.
volumes given in equation (11).
Additionally, the effects of meter orientation can be seen
The overall uncertainty in separator quality measurement
from a comparison of Phase I data (sensors positioned
was estimated to be 1.5°/0(quality percentage units). The error
4 FIELD TESTING OF DIELECTRIC STEAM QUALITY SENSOR SPE 46236 ‘
limits, based on a 95% confidence level, is +1- 3.0%. The quality exiting the generator. As generator quality decreases
percent contribution of each input variable to overall the TDS concentration in the liquid phase decreases and,
uncertainty is shown in Figure 9. As expected, the liquid correspondingly, pW increases. Since feedwater TDS is
volume rate contributes the largest source of error because of typically uniform for a given field, actual values of p. can be
periodic fill-up and dumping associated with fluid level measured for generator qualities ranging fkom 20°/0 to 90°/0.
control within the separator vessel. The value of pWused by the DSQS device would then depend
DSQS Device. The DSQS device measures impedance and upon the quality coming out of the steam generator plant.
converts it to liquid volume fraction using equation (5). In The DSQS data plotted in Figure 4 through 7 were based on
turn, the liquid volume fraction is converted to steam quality pWvalues for three generator qualities (30Y0, 50’% and 75?40).
using equation (6). Therefore, both equations were used in the Figure 12 shows the effect of using a single, nominal value of
estimation of DSQS quality measurement uncertainty. Error pW,based on 75% generator quality. The largest effect can be
limits, and associated confidence levels, were estimated for seen when the actual generator quality is much lower than that
each variable. used for the nominal value of pW. Since, the nominal
The overall uncertainty in DSQS quality measurement was resistivity is much lower than the actual liquid phase
estimated to be 1.7°/0 (quality percentage units). The error resistivity, the DSQS device will tend to underestimate the
limits, based on a 95?J0confidence level, is +/- 3.4%. The amount of liquid present. Consequently, the DSQS quality
percent contribution of each variable to overall uncertainty is readings will be higher than expected.
shown in Figure 10. The liquid phase resistivity contributed Low DSQS quality measurements, caused by inlet flow
the largest source of error. In most applications, frequent regime effects, tend to be offset by high quality
adjustment of pWto account for generator output variations measurements, caused by using low water resistivities. This
may not be practical. Therefore, error limits for pWwere can be seen in Figure 12, for inlet qualities below 30°/0. When
estimated assuming a nominal value of pW and generator the actual (higher) water resistivities are used, the DSQS
quality variation of -1-1-5%. qualities are much lower than separator qualities. When the
Inlet Flow Conditions. The DSQS device measures changes nominal (lower) water resistivity is used, the DSQS qualities
in impedance resulting from changes in the liquid volume track separator qualities more closely. It is not safe, however,
fraction. However, impedance electrodes are sensitive to the to assume that the errors occurring from these effects will
different flow regimes or patterns that can occur during gas- always offset one another.
liquid flow. lb Typical two-phase flow regimes in horizontal
pipes are shown in Figure 11. The occurrence of a particular Installation Guidelines
flow regime depends on the amount of vapor and liquid As with any metering device, special consideration must be
present in the pipe, their velocities, and their densities. given to installation. Often, unsatisfactory meter performance
Both DSQS Sensors tended to indicate low steam qualities can be attributable to problems associated with initial
when a combination of low inlet vapor velocities and high installation of the device.
liquid volume fractions. At these conditions, intermittent Meter Orientation. As previously discussed, installing the
slugging was apparent at Sensor 2. Stratified and stratified- DSQS device in a vertical orientation can result in erroneously
wavy flow conditions were also likely at Sensors 1 and 2 at low quality readings. The primary reason is believed to be
low vapor velocities and high liquid volume fractions. electrode “shorting” caused by liquid hold-up or
When a slug of liquid passes through the DSQS device, the accumulation.
measured impedance will reflect 100’%liquid even if the slug The preferred installation of a DSQS device at an injector is
does not fill the entire device. This effective “shorting” of the shown in Figure 13. The device should be positioned
impedance device results in a much lower estimated quality horizontally in the steam line, downstream of a wellhead
than is actually present. This “shorting” effect also occurs choke or other flow control device.
when the flow entering the DSQS device is stratified, Inlet Flow Conditions. Ideally, the flow entering the DSQS
provided the liquid level is high enough to bridge the gap device should be in the annular-mist regime. High vapor
between the bottom of the electrode and the housing. velocities and low liquid volume fractions observed
For slug flow, this problem can be minimized by increasing downstream of a wellhead choke often result in annular flow.
the data sampling frequency and reducing the electrode The DSQS device should be located at least three to four feet
length. For semi-annular and stratitied flow, increasing the downstream of the choke to ensure that stable flow conditions
gap width between the electrode and the housing would have been establish.
further reduce the problem. In other installations, where flow chokes are not used, the
Water Resistivity. The resistivity of the liquid phase must be vapor velocity can be increased by reducing the diameter of
known to convert the DSQS impedance measurements to the steam piping directly upstream of the DSQS device. A
steam quality. Liquid phase resistivity, pW,is a fimction of the flow straightener or mixing device can be installed upstream
total dissolved solids (TDS) in the feedwater and the steam of the DSQS to improve inlet flow conditions.
b
Sumlicial Swwf!cial Liwid Scwrator DSQS Superficial Suonticial Liquid Separator DSQS
Test Steam sfe.am Vwmr Liquid Voiume stem Ste& Q.dity Test Steam steam Vapor iiquid Volume Steam Ste3m Quality
C33e Pressure Rate WAY Vel&ity Fraction Qoali~ Quality Difference Case Pre2sure R3tc Velocity Vekxity Fraction Quality Quality oifferemx
(ysia) (b/d WC) (I-t&cc) (ftkcc) (%) (e&) (%) (p3ia) (b/d w.) (tvsec) (RKec) (%) (%) (%)
Generator Quality = 33*A to 3.s”/. 40 302,8 746.8 88.9 1.68 0.019 395 32.2 -7.3
41 210,4 2540 675 0,35 0,045 61,8 59,3 .2,5 41 295,1 7461 85,7 1.75 0,020 37,1 29,2 -7,8
42 169,0 1243 49,1 011 0,602 744 69,5 4,9 42 312,2 795,0 36.8 t .?S 0.021 37,3 292 -8,1
43 {70,7 1262 45,6 014 0,663 688 669 -1,9 43 305,2 744,9 90,6 1.65 0,018 40.7 33.4 .7.2
u 157,2 85,4 33,2 010 0,603 68,3 604 -7,9 44 298,4 7?9,7 ffi,3 1.63 0,018 39,2 32.8 4,4
4s 169,2 1378 47,9 017 0004 655 665 1,0 45 295.5 740,4 85.9 1.72 0.020 37.6 30.5 -7.1
46 179,9 131,5 438 0,17 0W4 834 629 -0,5 46 313,2 803,5 87.2 1,88 0.021 37.2 28.7 4,5
47 170,0 1337 43,7 0,18 0,004 62,0 624 0,3
48 1574 2!30 34,0 0,66 0,602 72,2 67,5 4,7
Superficial Supdcid Liquid Separator DSQS Superficial Supcrftcial Liquid Separator DSQS
Test Steam stem Vap.3r Liquid Volume steam Qu21ity Test Ste3m Steam Vac-3r Lrauid Volume Steam Steam Qu.aiiw
Case Pressure R3m Velocity Velocity Fracdon Quahy Qudlity DWference C23e Pressure R2te Velocity Vel&ity Fraction Quality QUZMY Differmce
(ps!a) (bid CWJ (ftiscc) (fhc) (! ’0) (%) (%) (psia) (b/d we) (ws..j (tv.wj (%) (%) (%)
-......=.
I sole 4D
-r
rnase IV I ests - sensor,4
-.” -
. *
ANSI 600#
co
Insulated
Flange
Si3acer.
Connecting
f E Rod
Electrode
(2” ID)
(1S” length)
Vapor - Phase
Tempereture
Sensor
Two - Phaee
Steam
2’ Vortex Meter
m- +/y CkQS
separator Device I
Level V-easel Ste&m \
Control Choke Tee Anchor
c
/ ,nd
v
Isolation Tee
Valve
+
v
To injector 2
100 100
/ ,
90 , 90 ~ Sensor 1
❑
4 -——,
. Sensor 2 . Sensor 2 ,’ ,’
80 80 /’ . ~
—.
——-—. —- - _ DSQS = SQT /
,
70 —=——...—-== .. —-—
— 70 ——. ._& ---
q0 q
“/ , /’-, ,’
‘-
~ 60 ~ 60
.=
z! 50 ; 50
e , a
40 40
8J s
~ 30 g 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80 90 100 O 10 20 30 40 50 60 70 80 90 100
Separator Quality (Y.)
Separator Quality (%)
Figure 5 DSQS vs Separator (Phase I Tests) Figure 6 DSQS vs Separator (Phase H Tests)
—.- —. —.. —
.
.
100 100 ,
90 90 ❑ Sensor 1 ——
. Sensor 2 . Sensor2 ,,
80 80 _DSQS = SQT
-. —......- ,’
70 , q 70
0
30 8 30
,
, 20
20 /
10
.!
. 10
7’-
“..
P’
0 0
O 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100
Separator Quality (%) Separator Quality (“/0)
Figure7 DSQSvs Separator (Phase III Tests) Figure8 DSQSvs Separator (Phase IV Tests)
O 5 10 15 20 25 30 35 05 10 15 20 25 30 35 40
I i
P.
Zv
- .. ....... .=_- z
VL
Vv
% Error Contribution P
I 1
0/0Error Contribution
Figure9 Error Contribution for Separator Quality
Figure 10 Error Contribution for DSQS Quality
/
100 ——.--., .——
Generator Quahty
❑ ,“
90
0 Nominal Quality = 75% ‘z DSQS Device
80 — ~ .
— DSQS = SQT
.. . ..—— .-.— ,< , /
.-——-<,-7-.. ‘
/’
———— .. —— --.— /
hf. ‘
“%0-2. “
~CA
/ /’
—— ..-=—
,
# .—. ..
(d Steam Choke
20 —~=oo-$~:—”
/
,
,
‘“ ‘-————
1To Injector
100
90
80 .--.. —
/
.. —- ..—...—
0 m’
.—
---
~ ,6 ————
[Y ‘
KPD
20 . ....- . . ... ..—. — ..
10 .L..:.d-–. ...-—––.-–.-– -—--–- -
/’m
ov~’ ) 1 1 1 t 1 I I
O 10 20 30 40 50 60 70 80 90 100
Separator Quality (Y.)