Flow Meas. Instrum.

Vol 2 April 1991 115

3-D laser Doppler anemometry measurements

of the axisymmetric flow field near an orifice
plate
R. E. DeOTTE, Jr, G. L. M O R R I S O N , D. L. P A N A K a n d G. H. N A I L

Measurements of the axisymmetric velocity field in a standard 50.8 mm (2.0 inch)

orifice meter were made using a 3-D laser Doppler anemometer. Calculations
were performed using the instantaneous velocity measurements to yield the
time-averaged mean velocity components, the products of the fluctuating
velocities (which provide the entire Reynolds stress tensor if the density is
assumed constant), the correlation coefficients and the turbulence kinetic energy
(TKE). For an air flow with a Reynolds number of 54 700, data indicate the point
of reattachment, the downstream primary recirculation zone, the downstream
secondary recirculation zone and the upstream recirculation zone. It is further
evident that the axial location of maximum velocity (the vena contracta) differs
from that for minimum wall pressure.
Keywords: 3-D laser Doppler anemometry, orifice meters, velocity measurement

Nomenclature Introduction
A Pipe cross-sectional area Orifice meters are used extensively in the metering of
D Orifice pipe (tube) diameter natural gas and it is estimated that the 1% uncertainty
d Orifice diameter in measurement may account for tens of millions of
r Radial distance from the pipe centreline dollars in mischarges annually. To provide a better
R Pipe radius, £)/2 understanding of the flow behaviour near orifices and
Re Reynolds number, pUm=D/l~ to contribute to a database for verification of computa-
puivj Reynolds stresstensor, Tip for compressible flow
rms Root mean square of a fluctuating quantity, tional fluid dynamics (CFD) models, a 3-D laser Dop-
[(V')211/2 pler anemometer was used to .measure the instantan-
u Local mean axial velocity eous velocity fields of the axisymmetric flow in a
u' rms of the axial velocity fluctuations standard 50.8 mm (2.0 inch) ID orifice meter. The
u~ Time average of axial-axial velocity fluctuations presentation of the means, variances and covariances
(normal Reynolds stress) of the orthogonal velocity components provide a uni-
u-~ Time average of axial-radial velocity fluctuations que set of data that yield insight into the fluid me-
(shear Reynolds stress) chanics associated with orifice meters. The reported
uw Time average of axial-azimuthal velocity turbulence kinetic energies (TKE) will be useful to
fluctuations (shear Reynolds stress)
numerical modellers seeking to develop effective tur-
U Maximum centreline axial velocity (vena contracta)
Um Average uniform pipe velocity bulence closure techniques. As a clearer understand-
u Local mean radial velocity ing of the flow field evolves because of measurements
v-~ Time average of radial-radial velocity fluctuations such as these, a better perception of the appropriate-
(normal Reynolds stress) ness of the pressure measurements and their relation-
u' rms of the radial velocity fluctuations ship to the flow field can be formed. 1,2,3
v--~ Time average of radial-azimuthal velocity
fluctuations (shear Reynolds stress)
ww Time average of azimuthal-azimuthal velocity
fluctuations (normal Reynolds stress) Literature review
w' rms of the azimuthal velocity fluctuations Much of the previous work on orifice meters has
(azimuthal turbulence intensity)
x Downstream distance from the downstream surface concentrated on discharge coefficients and the rela-
of the orifice plate tionship of wall pressures to flow rates 4. Other studies
/3 Orifice plate beta ratio, d/D have investigated parameters such as plate dimen-
p Density sions 5, deflection and eccentricity 6 and pipe anoma-
# Absolute viscosity lies 7. Mittal & Sharma 8 and Patel & Sheikholeslami 9
have focused upon the improvement of numerical
The authorsare in the MechanicalEngineeringDepartment,Texas models, whereas the studies of Miller & Cullen 1° and
A&M University,CollegeStation,TX77843-3123,USA Hall et a111 have concentrated on the equations

0955-5986/91/020115-09 © 1991 Butterworth-Heinemann Ltd

 116 R. E. D e O t t e et al - A x i s y m m e t r i c f l o w field near an orifice plate

describing metering performance. The use of numer- Through both flow visualization and velocity measure-
ical modelling techniques in conjunction with experi- ments, they were able to identify the upstream prim-
mental studies to provide verification data and insight ary, the downstream primary and the downstream
into the flow fields provide significant promise for secondary recirculation zones. As expected, the three
understanding the operation and design of orifice normal stresses achieved local maxima along the shear
meters. Mittal & Sharma 8 modelled flow through thick layer, as did the kinetic energy and u-v shear stress.
orifice plates and predicted upstream and downstream The effect of radial acceleration through the orifice is
recirculation zones. Patel and Sheikholeslami 9 stud- manifest in the 'overshoot' that they described (the
ied thin orifice plates and predicted large radial velo- maximum axial velocity occurs away from the pipe
cities and primary recirculation zones both upstream centreline). They reported the length of reattachment
and downstream, but apparently neither numerical as a function of Reynolds number, and the values
study predicted the downstream secondary recircula- appear consistent with other work on orifices and
tion. This conflicts with intuition and experimental sudden expansions. They also performed CFD simula-
evidence from studies of flow through sudden expan- tion of the flow field and compared the numerical
sions 12 1~. Mukerjee et al. ~7 evaluated the status of predictions with their experimental data and pro-
CFD modelling of orifice meters in 1984 and con- nounced excellent agreement. There were, however,
cluded that existing turbulence models were unable to some discrepancies. For instance, the computer model
provide adequate and cost-effective predictions of the predicted a longer reattachment length than was ob-
flow field. They delineated a methodology for redres- served for the experimental measurements.
sing the existing situation, much of which has been The complexity of the flow field and its depend-
incorporated into this and other more recent studies. A ence upon both upstream and downstream flow con-
major concern of theirs was the development of a ditioning was demonstrated by Shen et a l ~.~ in their
database of turbulence quantities appropriate for the study of flow near multiple orifice plates. They used a
refinement of the turbulence closure models. 2-D LDV to measure the axial and radial velocity
Very little velocity data is available for verification components, which then permitted calculation of the
of the CFD models. Dyban & Epik ~ used a single-fila- two turbulence intensities and u-v shear stress. The
ment hot-wire anemometer to measure velocity pro- upstream plate significantly affected all quantities for
files for a Reynolds number of 30000 and orifice the downstream orifice and vice versa, though to a
diameter to pipe diameter ratios (/3 ratio) of 0.25 and lesser degree.
0.67. At an axial position 0.8D downstream of the For either orifice meters or sudden expansions
orifice plate, the velocity fluctuations were 4 - 5 % in there is very little data that includes values for the
the core, increased to 27% in the shear layer, and velocity variances and covariances ( ~ , w~, w w , ~ ,
further increased to 44% in the recirculation zone. vw) 23. Studies on sudden expansions have demons-
They found that the recirculation zone was shorter for trated secondary recirculation zones 12-~5. There are
the larger /3 ratio and that the turbulence level de- similarities between the flow through an orifice and
creased with increasing Reynolds number. the flow through a sudden expansion: both demons-
For a Reynolds number of 100000, Teyssandier & trate regions of recirculation and points where the
Wilson ~9 observed a discrepancy between the location primary flow reattaches to the wall, but there are also
of the minimum cross-sectional area of flow (vena differences. Notable among the differences is the exist-
contracta) and the point of minimum wall pressure. ence of the vena contracta for flow passing through an
They measured the location of the vena contracta as a orifice, which is presumably caused by the inward
point less than one step height downstream of the flow upstream of the obstruction. A definite need
orifice, while the minimum wall pressure occurred at exists for comprehensive velocity field data for orifice
approximately three step heights downstream. They meters.
pointed out that they were measuring with only one
sensor,which made analysis difficult near the orifice
where the radial velocity is the same order of mag-
nitude as the axial. Experimental procedure
For Reynolds numbers in the range from 21 000 To ensure that the flow entering the orifice meter
to 160000 and /3 ratios of ~, ~, and ~, Teyssandier would be fully developed and devoid of swirl, it was
and Husain 2° demonstrated that the plate influenced constructed to substantially exceed the standards given
wall pressure up to 0.62D upstream of the plate. The in the American Gas Association standard (ANSI/API
pressure on the upstream surface of the plate started to 253024, 1985). The length of approach was twice that
decrease at about 50% of the step height and con- specified and the flow straighteners and orifice plate
tinued to decrease to the bore. Wall pressure measure- conformed with respect to concentricity, flatness, bev-
ments indicated that the minimum occurred 3.3 step el angle, etc. The orifice plate was 3.175 mm thick
heights downstream of the plate. and made from stainless steel. The hole was
For a Reynolds number of 22000, Durst & 25.40 mm in diameter, which yielded a/3 ratio of 0.50
Wang 21 used a 1-D LDA to measure the velocity field when used with the 50.8 mm ID cast acrylic pipe. The
within an orifice meter with /3 = 0.5. The working orifice had a 45 ° bevel, which was 1.59 mm thick.
fluid was a mixture of diesel oils, the pipe had a The pipe had a flat surface milled along the length of
50 mm ID and the meter test section was 1 m long. one side, which left an opening 7.62 mm wide where
The diesel mixture was such that the index of refrac- windows were mounted (Figure 1). The windows were
tion matched that of the transparent pipe, which was optical quality quartz treated with an anti-reflective
done to facilitate measurements with the LDA. coating to ensure minimal distortion and maximum
Flow Meas. Instrum. Vol 2 April 1991 11 7

~__Window ~ ~ Quortz -~ ~ ~ having a resolution of 2.5/~m and a repeatability of

h°l__d-ers_
- - - -- ~--f ~ 4 - - ~ - ~ ~ C)7"
6~ 10.2/~m.
2.7 Output of the signal processors was monitored in
--~q~ -Orifice--/F~l------~fl--1524
realtime to ensure proper setting of the gains and
}-q Acryl'c/ I-~-4 Acryl,c H 0 0 filters, which were adjusted to permit signals within
four standard deviations of the mean to be included.
L J pipe
127.0~ ULIflenges
127.0 - ~ U R31.8 J ~-R7.9m-4m
I~ -I I. Furthermore, if adjustment caused the mean to
Figure 1 Orifice meter modified for LDV access show- change, the band width was adjusted until no change
ing quartz windows and test section construction occurred.
Since the optical axes of the LDA system are not
all orthogonal, it was necessary to transform each
individual realization before computing the velocity
variances and covariances. This is described in detail
transmission of the projected laser beams. This modifi-
by Morrison et al. 26 Two weighting schemes were
cation created an out-of-roundness of less than 1.1%
used to calculate the means, variances and cova-
of the pipe radius. Since the measurements closest to
riances of the velocities: one used uniform weighting
the wall were 1.6 mm away, the effect of this eccentri-
and the other used the weighting proposed by
city may be considered negligible. Measurements were
McLaughlin & Tiederman 28. When the velocity profile
taken on a radial-axial plane from 1.6 mm from the
was integrated across the radius, the second scheme
window to 27 mm from the window, which placed
provided a more constant value for the volumetric
the last measurement an increment of 1.6 mm (1/32
flow rate at cross-sections along the meter length
D) past the centreline. Data was taken from 5 D
(Figure2); therefore the data reported here was re-
upstream to 10 D downstream. A valve was placed
duced using that scheme.
downstream to maintain the upstream pressure at
The flow was characterized by a pipe Reynolds
254.0 + 1.3 mm Hg (10.00 + 0.05 inchesHg). The up-
number of 54700 and a Mach number at the vena
stream pressure was monitored using a mercury mano-
contracta of 0.25, which corresponds to a maximum
meter connected to a static pressure tap mounted 5 D
velocity of 81.9 m s -~ .
upstream of the orifice plate. The flow rate was
established by maintaining a constant pressure at the
entrance of two sonic nozzles (3.23 and 4.57 mm in Results
diameter). Air was supplied by a screw-type com-
pressor capable of delivering 1 7.5 m 3 min Flow visualization
(600 scfm) at 860 kPa (125 psi) and was dehumidified Deposition of coagulated seed droplets permitted pre-
through desiccant bed driers to a dew point of - 4 0 °C liminary qualitative analysis of the flow fields. The
( - 4 0 °F). The pressure was held constant by an elec- reattachment point appeared to be about 4.8R down-
tro-pneumatically actuated control valve, which was stream of the orifice plate, which was consistent with
operated by a computer. The pressure was monitored the reattachment point indicated by velocity measure-
and held at 618.0 + 1.4 kPa (90.0 + 0.2 psi), which ments. The droplets also indicated the presence of a
was sufficient to ensure that the nozzles were choked. small secondary recirculation zone downstream and a
Velocity measurements were made by using a
3-D laser Doppler anemometer. The LDA is a six-
beam, three-colour, two-axis system configured to 2.00
receive scattered light in an off-angle backscatter
mode and described in detail elsewhere 12,23,2s. Fine
spatial resolution was obtained by using 3.75x beam 1.50
expanders producing a probe volume that was o•
m

60/~m x 60/~m x 230/~m. Extreme care was taken to ~, I,O0

ensure that the beam waists of all six beams coincided
O

precisely in space, and that the photomultiplier tubes o Statistical

were focused precisely at the measurement point. The ~0.50 A McLaughlin- Tiedermandebiasing
maximum time for validation between the three com-
ponents (coincidence window) was set at 10/~s (the 0 I I I L I I i i i I I i i ~ I I I i ll5.0
minimum obtainable at the time). All these parameters -5.0 0 5.0 I0.0
x/R
are critical to guarantee validity of the calculation of
the velocity covariances. Coincident rates between 30 Figure 2 Comparison of data reduction schemes. Data
and 60% were normal and on occasion exceeded was reduced weighting all measurements equally
even that. The alignment procedure will be described ('Statistical'), and by weighting with the inverse of the
further at a later date 27. velocity magnitude, which is the scheme proposed by
The seed was a commercially available fog fluid, McLaughlin and Tiederman to minimize the velocity
which was nebulized with an aspirating six-jet ato- bias introduced by faster moving seed particles being
mizer. The particle size distribution was measured sampled more often than slower ones. The "Statistical'
using a TSI model 3310 aerodynamic particle sizer plot uses no debiasing while the "McLaughlin-Tieder-
and was characterized by a geometric mean and man" plot uses their debiasing scheme 28, and better
standard deviation of 0.76/~m and 1.32, respectively. conserves mass. Volumetric flow rate versus axial
The LDA system was mounted on a traverse table location cross-pipe trapezoidal integration of mean
capable of motion in three orthogonal directions and axial velocities. LDV data (~ = 0.5; Re = 54 700)
118 R. E. DeOtte et al - Axisymmetric flow field near an orifice plate

small primary recirculation zone upstream of the ori- tracta. A better indication of the upstream recirculation
fice plate. All these observations are confirmed by the zone and the downstream recirculation zone can be
velocity measurements. seen in Figure 3b.
Far upstream the flow is basically fully developed
turbulent pipe flow; the velocity profile at x/R = -4.1
Vector plot
has negligible radial or azimuthal velocity components
The radial and axial velocities at each of the 656 and the centreline axial velocity is 14.5 ms -~
measurement locations were combined to produce the (47.6 fps) compared to the maximum of 81.9 ms ~. It
vector plot shown in Figure 3a. Each arrow represents is not until x / R = - 1 . 1 3 that the flow first becomes
4096 coincident velocity realizations and is non-di- affected by the presence of the orifice plate; here the
mensionalized by 8 1 . 9 m s ~ (269feets -~ (fps)), the flow begins to turn inward, resulting in a slight in-
maximum axial velocity that occurs in the vena con- crease of axial velocity near the centreline. A small

-I.0 m
-4.0 -
!
i
-3.0
i

-2.0

- I .0
i
1 l l l J l l J J ~ l * * * * o-
lji!i i
r,, 7 • ")' ~",~

m
'I~ le y,, • Iv~

'IIIVI~::::::::'II
1.0 ¸
J llJl ' . . . . . . .
Jt

2.0
l I
1 I .0~

3.0
J I I 1 1 1 [ 1 1 1 1 1 ,-i-]
J ]till"' . . . . . 0.6
r/R
1.0

4.0

I
5.0
JJJJiilll~' ..... b
ijjjjit~, .....
6.0 /JIJllll~* .....

Figure 3 Vector plot of the velocity field in an orifice

meter. The orifice ID is 25.4 mm (I.00 in) and the
pipe ID is 50.8 mm (2.00 in), which yields a ~ ratio of
0.50. The Reynolds number was 54 700. The flow is
8.0 J J l J J J ~ l ~ l ~ s , uniform upstream and reattaches at x/R -~ 5.0. (a) The
velocity field beginning from a location upstream with
fully developed, turbulent pipe flow and continuing to
9.0 ~ ~ ~ ~ ~* ~ • ~* ~* ~, •
a point downstream where the flow is again fully
developed turbulent pipe flow. (b) The vector field in
the immediate vicinity of the orifice plate. Notice the
a
. . . . 8:2 . . . . . . . . . . . ,:o upstream recirculation and the downstream secondary
r/R recirculation zones.
Flow Meas. Instrum. Vol 2 April 1991 119

primary recirculation zone occurs in the upstream 1.0

corner of the orifice plate and the pipe wall.
Immediately upstream of the orifice plate there is
a rapid acceleration of the flow through the orifice 0.8
plate with the radial velocity increasing significantly
and creating the vena contracta downstream.
Downstream of the orifice plate the inwardly 0.6
directed radial component of velocity persists. At
x/R = 1.00, the location of the vena contracta, the
direction of the flow at the centreline has become ~0.4
completely axial and reached its maximum value of
81.9 ms -1. In fact the profile is essentially flat out to
r/R = 0.25 (half of the radius of the orifice). Beyond 0.2 0-o--o.--o
the vena contracta the flow spreads radially outward
until it reattaches to the wall near x/R = 4.75. The
downstream primary recirculation zone is very obvi-
ous, but careful examination shows that there is also a
secondary recirculation zone in the corner between
the orifice plate and the pipe wall. -0.2 I I I I I
-8 -4 0 4 8 12 16
The recirculation zones are characteristic of those X/R
for fully developed flow in a small pipe that undergoes
a sudden expansion, but they are displaced down- Figure4 Axial velocities along the pipe centreline.
stream somewhat. This is due to the inwardly directed Notice the rapid rise after x/R-~ -1 , the maximum at
radial flow upstream and immediately downstream of x/R -~ 1, the rapid decrease after x/R ,~ 2.5, and finally
the orifice plate. A reattachment length of 6.5 to 7.5 the constant value after x/R ~ 9. (/3 = 0.5;
times the step height is characteristic for sudden pipe Re -- 54700; U = 81.9 ms -1)
expansions, whereas the reattachment here occurred
for 4.5-< x/R <- 5.0, which is equivalent for this /3
ratio to a step height of 9.0 to 10.0. If the vena 0.I
contracta occurs at x/R = 1.0, the number of step
heights downstream of it is 7.0 to 8.0, indicating that
once the flow has reached the vena contracta, the
downstream characteristics are very similar to those
for sudden pipe expansions.

Mean centreline axial velocities

It is instructive to examine the mean axial velocities
on the centreline of the pipe (Figure4). The flow is #-o.
essentially a fully developed turbulent flow with a "k / o -4.1
normalized axial velocity of u/U = 0.18 until approxi- \ I D -i.l
mately x/R = - 1 . 1 3 , then the axial velocity increases ,l ,, - 0 . 5 0
suddenly to u/U = 0.49 at the orifice plate inlet and \ / , -o. 25
-0.~ k / ~, 0.,25
continues to rise as it passes through the orifice. It -,~ ~' ÷ 0.25
reaches a peak at the vena contracta (x/R = 1.0). From "eK ~" , 0.50
x/R = 1.0 to 8.0 there is a fairly rapid return to the <> I.O
A 2.0
upstream velocity magnitude, u / U = 0 . 1 8 . Beyond D 4.0
that the centreline velocity appears to remain constant -0.5 I t I i I I I I i I
and it may be presumed that the flow has returned to -0. O.I 0,3 0.5 0.7 0.9
r/R
nearly fully developed. A separate set of measure-
ments 24 for the pressure field indicates that the point Figure5 Radial profiles of the radial velocity at a
of minimum wall pressure occurs at x/R = 1.5, which number of axial positions both upstream and down-
is significantly downstream from the vena contracta. stream of the orifice plate. Notice the minimum radial
For a Reynolds number greater than 100 000, Teyssan- velocity for x/R = -0.25. (Re = 54 700)
dier & Wilson ~9 observed that the minimum static wall
pressure occurred downstream of the maximum cen-
treline velocity. They referred to this as the 'paradox of maximum variation shown occurs at x / R = - 0 . 2 5 ,
the vena contracta'. where v = - 0 . 2 4 U at r/R = 0.45. It is possible that
the true minimum occurs in the range
Mean radial velocities - 0 . 2 5 -< x/R <
- 0.125, but it was not possible to per-
form LDA measurements along an entire traverse in
Figure 5 shows the radial profiles of radial velocity at that region. By x / R = 0.25 the radial velocity has
different cross sections along the pipe. At x/R = -4.1 again decreased to a minimum of - 0 . 0 5 U , but it is
and - 1 . 1 , the radial velocity does not vary with interesting to note that the profile maintains a fairly
radius. At x/R = - 0 . 5 there is a pronounced depend- pronounced deviation from uniform in the region
ence upon r with v decreasing to - 0 . 0 7 U . The extending from the centreline to the edge of the
120 R. E. DeOtte et al - Axisymmetric flow field near an orifice plate

orifice, although in the region beyond the edge, radial I oF~----¢ X/R
dependence is nearly the same as further downstream. .4.---+ 0 -4.1
In other words the flow through the orifice plate still
exhibits a radial component, which contributes to the
development of the vena contracta.
0.8
\ El
•
•
-I.I
-0.5
-0.25
'~ 0.125
+ 0.25
x O. 5 0
Mean axial velocities 0.6 0 1.0
,s 20
The radial profiles of mean axial velocity presented in t> 40
Figure6 corroborate the observations made with re-
spect to the profiles of the mean radial velocities. At X~ 0. 4
x/R = -4.1 the profile is that of a turbulent pipe flow
with u/U = ( 1 - r/R) k. As the flow approaches the
orifice plate, the centreline velocity begins to increase 0.2
and by x / R = - 0 . 2 5 the mean axial velocity has a
maximum value of 0.5U on the centreline and a value
of 0.33U at r/R= 0.45. At this location v = - 0 . 2 4 U ,
which demonstrates the significant influence the ori-
fice has on the contraction of the fluid, since u and v
are comparable in magnitude. -0.2t I I I I I I I I I
The 'cusp' at r/R-~ 0.35, x/R = 0.25, is interest- -0. I O. I 0.3 0.5 0.7 0.9
ing: it indicates a region where the velocity at the r/R
centreline is less than the velocity at the edge of the
jet issuing from the orifice. It is still apparent, but Figure6 Radial profiles of the axial velocity at a
much less pronounced, at x/R = 0.50. Recall also that number of axial positions both upstream and down-
there was an inflection in the profile of radial velocity stream of the orifice plate. The 'cusp' on the profile at
for the x/R = 0.25 cross-section, hence the 'cusp' and x/R = 0.25 indicates higher velocities away from the
the beginning of the inflection correspond. For centre/ine. (Re = 54 700)
r/R >_ 0.5 (the edge of the orifice) the axial velocities
are very small, though not necessarily precisely zero. u/U
Teyssandier & Wilson v~ had previously observed such
a 'cusp'. A final observation is that the axial velocity is
essentially flat from the centreline to r/R~ 0.25 at
x/R = 1 .O, the location of the vena contracta. o.2 I_ \ •
-4 -2 0 2 4 6 8 I0 12 14
x/R
Contour plots
Contour plots, or isograms, of the mean axial and
radial velocities, the velocity variances, the u5 covar-
lance, the ~-g correlation coefficient, and turbulence v/d
kinetic energy are presented. The azimuthal velocity,
and the ~5 and v w covariances and correlation coeffi- ~:0"75 F
cients were all measured and found to be negligible to,soF
and are not presented.
0'205F i , i i L# ~ , I I I I I
-4 -2 0 2 4 6 8 I0 12 14
x/R
Contour plots of mean velocities
The u/U contour plot in Figure 7 indicates the ac- Figure 7 Contour plots of the mean axial and radial
celeration of the flow through the orifice plate by the velocities. In both notice the acceleration as the flow
close spacing between the 0.25, 0.5 and 0.75 con- passes through the orifice. ([3 = 0.5; Re = 54 700)
tours. Closer spacing of the contours indicates more
rapid acceleration or deceleration of the flow. The
gradual increase in spacing downstream demonstrates
the v/U plot. The most significant is the separation
the relatively slower deceleration after the vena con-
from the throat of the orifice plate, which coincides
tracta (the location of maximum velocity). The very
with the u/U separation; however here the O level
close spacing of the negative upstream radial velocity
contour crosses the centreline and presumably reatt-
contours indicates the rapid radial inflow as the fluid
aches at the symmetric position on the unseen section
approaches and passes through the orifice, thereby
of the orifice plate. The positive values of v/U down-
driving the acceleration through the meter. The in-
stream indicate a spreading of the orifice jet.
tersection of O level contours with the orifice plate or
wall indicate regions of separation or reattachment. Velocity variances and covariances
The u/U plot indicates two such situations: one near
the corner where the orifice plate and pipe wall The turbulent characteristics of flow through the ori-
intersect, and the other separating from the edge of the fice meter as reflected in the velocity variances and
throat of the orifice and reattaching at x/R-~ 5.0. covariances are important for understanding the tur-
There are likewise two separation regions indicated on bulent flow field and establishing a database for
Flow Meas. Instrum. Vol 2 April 1991 121

comparison with numerical models of such flow fields. accelerates through the orifice. This suppresses the
It is this data that has been most significantly lacking. turbulence production by stretching eddies in the
In Figure 8 all three variance terms are presented direction of the acceleration, which may cause the
in separate contour plots. The contour of 1 0 0 ~ / U U , eddies to break up, but will none the less reduce the
the u/U velocity variance, shows characteristics simi- amplitude of the eddie-induced velocity fluctuations.
lar to those for most shear flows, exhibiting a max- The contours for the azimuthal velocity variance
imum in the middle of the shear layer along r/R-~ 0.4 are very similar in shape and magnitude to those just
and reaching a minimum near the wall far down- discussed for the radial velocity variance. The flow is
stream. The variance, and hence the fluctuations, of not isotropic since the variances of axial velocity in
u/U are minimal during the acceleration of the fluid the shear layer are not the same as those for either the
through the orifice; it is only after the vena contracta radial or azimuthal velocity variances, in fact the axial
at x/R = 1.0 that I O 0 ~ / U U begins to increase along velocity variance is twice the magnitude of the other
the centreline, the region where the shear layer and variances. As Tennekes & Lumley 29 point out, this is
the deceleration of the flow past the vena contracta typical of shear flows. For large Reynolds numbers the
furnish the high velocity gradients and momentum energy budget for the turbulent flow fluctuations may
exchange necessary for turbulence production. The be written in the Cartesian coordinate system as
close spacing of the ~ contours immediately down-
stream of the plate along the extension of r/R ~-0.4 0 = -ulu~ ~x2 ~x2 + - e
demonstrates the high level of axial turbulence pro-
duction at the lip of the orifice, whereas the wider Analysis of this budget indicates that only the equation
spacing of the same contour lines beyond x/R = 5.0 for the axial direction has any production, - ~ d S U / ~ r ;
indicate the slow dissipation of axial turbulence. The production for the other components being zero. As
maximum value corresponds to a turbulence intensity the flow is observed further downstream, 8 U / ~ r app-
(based upon U) of 20%, similar to the magnitude roaches zero, the production ceases and the turbul-
observed for sudden expansions. ence levels also decrease. Past x / R = 10 the turbul-
The radial velocity variance, u~, obtains a max- ence becomes isotropic.
imum dimensionless value of approximately 3.3 Neither the uw nor the uw is of particular in-
( ~ 5 = 2 1 8 m E s 2) at x / R = 1 . 5 and r / R = 0 . 4 4 and terest, because they are both essentially 0 throughout
corresponds to a turbulence intensity of u'/U = 17%. the flow field, which means that the respresentative
In the region - 0 . 8 -< x/R-< 1.0, the turbulence inten- fluctuations are uncorrelated. As shown in Figure9,
sity is less than 5%, which may be attributed to the ~5 term is nonzero, indicating a correlation be-
suppression of turbulence production by the strong tween u and u. The exchange of fluctuating momen-
acceleration through the orifice. As the flow appro- tum is between the axial and radial components. The
aches the orifice there is a redirection and accelera- maximum value occurs just downstream of the vena
tion as fluid from the wall moves radially inward and contracta in the shear layer, where the velocity va-
riances were also a maximum. Investigation of the
covariance terms indicates that the ~-5 term is 25
I 0 0 b~'/UU times the others. It is the exchange of fluctuating
~.ooI ll~r---~.~ ___.i-.z_~--Z-~..~-....<. ~q,
\ momentum that is causing the spreading of the jet and
the ultimate return to an isotropic flow past x/R = 10.
Figure 10 shows the contours of the -U-6 correlation
0 / I I I " "- I , I I I I
-4 -2 0 2 4 6 8 I0 12 14
x/R
,.oo|IOOEV/UU II
0.75p o. o ~ ~e,~

1.001
I00 ~ ' / U U
II ~ , ~ o ~ ~ .
^ __Io.2~-."~--'o II ~ ' ~ J - - - - ~ ' " - ~
.~Z a ~ _ I I - - - ~ o
o0ot
0/ ~ I I I I ~IO I ~ N " I I
-4 -2 0 2 4 6 8 I ib' i:~ I 14

"",,\,\
x/R

°"°I
o

04 -2 0 - -2 4 6 8 10
" 12
!i 14 Figure9 Contour plot of u~ covariance. ([3 --- 0.5;
x/R Re = 54 700)

correlation coefficient
,
I 00 w'-~/UU
,.oo .o
0.75p " o.,~"---"--"-"~0.2

0.25~- o
0 / i i i • ) 1~ ~ I

-4 -2 0 2 4 6 8 10 1'2 14 O-~t [ -2 ) ~0 II C OI' 4 I

" ~-r"'--"~'/"1
2 6. . . .8 I0 12 I 14.
x/R x/R

Figure 8 Contour plots of the three velocity variance Figure 10 Contour plot of the J# correlation coeffici-
terms. Notice that the o and w variances are similar in ent. Note the rapid change from - 0 . 6 to +0.4 in the
magnitude, but the u variance is a factor of two throat of the orifice. The structure follows the develop-
greater. (;6 = 0.5; Re = 54 700) ment of the shear layer. ([3 = 0.5; Re = 54 700)
 122 R. E. DeOtte et al - Axisymmetric flow field near an orifice plate

coefficient, Cu,, = ~ 5 / ( - f f ~ x S ~ ) ~/2. There is a fairly energy balance presented by Tennekes & Lumley 28,
strong negative correlation upstream as the flow accel- which indicates that the axial component in a shear
erates to pass through the orifice. The correlation flow receives all of the turbulence energy production
increases rapidly at the throat of the orifice and the is substantiated since the axial velocity variance is
highest value of slightly over 0.5 occurs on the twice the magnitude of that for the radial or azimuthal
centreline at x/R ~. 0.4. The correlation coefficient is a components.
measure of the structure of the flow, i.e. a large
magnitude of the correlation coefficient indicates that
eddies are generated in a predictable fashion and are
generally oriented in the same direction. Close to the Acknowledgements
throat of the orifice approximately 50% of the turbul-
ence is due to predictable phenomena and the rest is The authors wish gratefully to acknowledge the sup-
due to random fluctuations. The eddies break up as port of this research effort by the Gas Research
they move downstream and interact with the shear Institute. Furthermore, Dr Max Klein, the contract
layer and the flow becomes progressively less struc- monitor, has made many insightful and helpful sugges-
tured until beyond x / R = 10 the turbulence is iso- tions that are appreciated.
tropic.

Turbulence kinetic energy References

As Figure 11 shows, the dimensionless turbulence 1 Morrison, G. I.., DeOtte, R. E. Jr., Panak, D. L. and Nail,
kinetic energy, TKE = (1/2)(~u + ~ + w--~O1/2 x G. H. The flow fields inside an orifice flow meter. Chem.
I O0/UU, has a maximum value somewhat in excess Eng. Progr. (July 1990) 75-80
of 5 beginning at x/R ~-3.0 and r/R-~ 0.5. Along the 2 Morrison, G. L., DeOtte, R. E. Jr., Moen, M., Hall, K. R.
centreline the values are low prior to the vena con- and Holste, J. C. Beta ratio, swirl, and Reynold number
dependence of wall pressure in orifice flow meters. In
tracta (x/R = 1.0). Beyond that there is a relatively
2nd Int. Syrup. Fluid Flow Meas. Calgary, Alberta, Can-
rapid increase as turbulence is produced by the de- ada (6-8 June 1990)
celeration of the flow and the interaction with the 3 Morrison, G.L., DeOtte, R. E. Jr., Moen, M., Hall, K. R.
shear layers. The most rapid increase in TKE occurs in and Holste, J. C. Beta ratio, swirl, and Reynold number
the shear layer beginning from the edge of the orifice. dependence of wall pressure in orifice flow meters. Flow
TKE begins to dissipate fairly rapidly beyond the point Meas. Instrum. 1 (1990) 269
where the flow reattaches to the wall (x/R ~-4.8); this 4 Fling, W. A. The API/GPA orifice-plate data base. J.
may be attributed to the reduction of D U / ~ r and Petrol. Technol. (July 1988)
hence the turbulent production, which in turn is the 5 Husain, Z. D. and Goodson, F. D. Effects of plate
result of the momentum exchange as reflected in the thickness ~nd bevel angle on discharge coefficients of a
50 mm (2 inch) orifice meter. Trans. ASME (December
covariance term. 1986)
6 Zedan, M. F. and Teyssandier, R.G. Effect of errors in
Summary pressure tap locations on the discharge coefficient of a
flange tapped orifice plate. Trans. ASME (December
For the velocity field in a 50.8 mm (2 inch) diameter 1986)
orifice meter with /3 = 0.5 and operating at a Reynolds 7 Zedan, M. F. and Teyssandier, R. G. The effects of
number of 54 700, the quantities of interest for phys- recesses and protrusions on the discharge coefficient of a
ical understanding of the flow field as well as for flange tapped orifice plate. In ASME Winter Annual
numerical simulation, such as the mean velocity com- Meeting (1986)
ponents, the velocity variances and covariances, the 8 Mittal, R. C. and Sharma, P. K. Numerical solution of a
viscous incompressible flow problem through an orifice.
correlation coefficients, and the TKE, are reported and Appl. Sci. Res. 44 (1987) 361-375
discussed. The 'paradox of the vena contracta' re- 9 Patel, B. R. and Sheikholeslami, Z. Numerical modeling
ported earlier by Teyssandier & Wilson 19 is confirmed. of turbulent flow through orifice meters. In Int. Symp.
The primary and secondary recirculation zones are Fluid Flow Meas. Washington, D.C., USA (November
delineated. Investigation of the covariance terms indic- 1986)
ates that the -6~ term is 25 times the others. The 10 Miller, R. W. and Cullen, J. T. The Stolz and ASME-AGA
orifice equations compared to laboratory data. Con-
tributed by the Fluid Meters Division of ASME at the
IOOTKE/UU Winter Annual Meeting (1978)
I .OOj 1 1 Hall, K. R., Eubank, P. T., Holste, J. C. and Marsh, K. N.
0 7 5 I- Performance equations for compressible flow through
xCosot orifices and other AP devices: A thermodynamic ap-
0.25~- proach. AIChE JI. 32(3) (March 1986) 517-519
o! I [ I -., . I~ ~ e I } I/ II I I I~
12 Morrison, G. L., Tatterson, G. B. and Long, M. W.
-4 -2 0 2 4 6 8 I0 12 14 Three-dimensional laser velocimeter investigation of tur-
x/R
bulent, incompressible flow in an axisymmetric sudden
expansion. J. Propulsion 4(6)(1988) 533-540
Figure 11 Contour plots of turbulence kinetic energy. 13 Thompson, H. D., Stevenson, W. H. and Durrett, R. P.
The most notable features are the rapid increase Laser velocimeter measurements and analysis in turbul-
originating at the edge of the orifice and the dissipa- ent flows with combustion, Part 3. AFWAL-TR-82-2076-
tion once the flow reattaches to the wall. (/3 = 0.5; PT-3, AD-A146206/8 (July 1984)
Re = 54 700) 14 Durrett, R. P., Stevenson, W. H. and Thompson, H. D.
Flow Meas. Instrum. Vol 2 April 1991 123

Radial and axial turbulent flow measurements with an Appl. Laser Anemometry to Fluid Mech. Instituto Super-
LDV in an axisymmetric sudden expansion air flow. ior T~cnico, Av. Rovisco Pais, 1096 Lisboa Codex,
ASME J. Fluid Engng. 110 (1988) 367-372 Portugal (July 1988)
15 Kangovi, S. and Page, R. H. Subsonic turbulent flow past 23 Panak, D. L. 3-D laser Doppler velocimetry study of
a downstream facing annular step. ASMEJI. Fluid Engng. incompressible flow through an orifice plate. M.S. Thesis,
101 (June 1979) 230-236 Texas A&M University, College Station, TX 77843, USA
16 DeRossett, T. A. and Przirembel, C. E. G. Time-depend- (May 1990)
ent behaviour of separated flow regions. In Instrum. Soc. 24 Orifice Metering of Natural Gas and Other Related
Amer. Aerospace Instrum. Syrup. 74255 (1974) 273-280 Hydrocarbon Fluids, 2nd Edn, American Gas Association
17 Mukerjee, M., Singhal, A. K. and Spalding, D. B. Appli- and American Petroleum Institute, ANSI/API 2530 (1985)
cability of numerical flow models to orifice metering 25 Morrison, G. L. 3-D laser anemometer study of com-
problems. In Proc. Conf. Metering Natural Gas and pressible flow through orifice plates, GRI Rep.
Liquified Hydrocarbon Gases, Section 3, 1 & 2 February GRI-90/0036 (November 1989)
1984,London, UK, Oyez Scientific & Technical Services 26 Morrison, G. L., Johnson, M. C., Swan, D. L. and
Ltd (1984) 193-210 DeOtte, R. E. Jr. Advantages of orthogonal and non-orth-
18 Dyban, Ye. P. and Epik, E. Ya. Effect of turbulence on ogonal 3-D LDA systems. In 5th Int. Symp. Appl. Laser
calculating heat transfer downstream of an orifice in a Tech. to Fluid Mech. Instituto Superior T~cnico, Av.
tube. Heat Transfer-Sov. Res. 2(1) (January 1970) 11-16 Rovisco Pais, 1096 Lisboa Codex, Portugal (1990)
19 Teyssandier, R. G. and Wilson, M.P. Jr. The paradox of 27 DeOtte, R. E. Jr., Morrison, G. L., Wiedner, B. G.,
the vena contracta. Trans. ASME, Paper 73-WA/FM-9 Panak, D. L., Nail, G.H. and Johnson, M. C. Laser
(December 1973) 1-5 alignment and stability in 3-D laser Doppler ane-
20 Teyssandier, R. G. and Husain, Z. D. Experimental mometry. Submitted to and abstract accepted by the
investigation of an orifice meter pressure gradient. Trans. Fourth International Conference on Laser Anemomet-
ASME 109 (June 1987) 144-148 ry-Advances and Applications to be held 5-9 August
21 Durst, F. and Wang, A. B. Experimental and numerical 1991, Cleveland, Ohio, USA
investigations of the axisymmetric, turbulent pipe flow, 28 McLaughlin, D. K. and Tiederman, W. G. Biasing correc-
over a wall-mounted thin obstacle. Proc. 7th Syrup. tion for individual realization of laser anemometer
Turbulent Shear Flows, Vol 1, Stanford University (21-23 measurements in turbulent flows. The Physics of Fluids
August 1989) 10.4.1-10.4.6 16 (1973) 2022-2088
22 Shen, X., Yan, Y., Gao, J. and Ding, Z. Turbulent 29 Tennekes, H. and Lumley, J. L. 'A First Course in
velocity measurements in orifice pipe flow with an Turbulence, 12th Edn' The MIT Press, Cambridge, Mass.
improved on-axis 2-D LDV system. In 4th Int. Symp. and London, England (1972) 74-75