FLOMEKO 2013

September, 24 26, 2013, Paris, France

A NEW FLOW CONDITIONER FOR 4-PATH ULTRASONIC FLOWMETERS

G J Brown1 and B W Griffith2

1
Cameron, UK, [email protected]
2
Cameron, USA, [email protected]

Abstract custody transfer applications. Most of the devices used with

ultrasonic meters today were designed for use with turbine
There are benefits, limitations and trade-offs to be taken meter or orifice plates. Common examples are tube bundles
in to account when considering using a flow conditioner with and thick perforated plates. While these conditioners are
an ultrasonic meter. These include aspects such as the generally suitable for use with ultrasonic meters, none of
pressure loss, the risk of fouling and the influence on meter them were specifically optimised for this purpose and there
repeatability, as well as the ease and reproducibility of are trade-offs to be taken into account during selection.
manufacturing.
This paper discusses different flow conditioner designs
This paper discusses different flow conditioner designs currently used with ultrasonic meters and introduces a new
and introduces a new design intended for use with 4-path design specifically intended for use with 4-path ultrasonic
ultrasonic flowmeters. Performance data and permanent flowmeters and optimised for that purpose.
pressure loss measurements are presented and compared with
a commonly used form of perforated plate flow conditioner. Flow Straigheners
It is shown that comparable flowmeter performance can be By far the most common type of flow conditioner has
achieved while reducing the pressure loss significantly. been a flow straightener of either the vane type or in the
form of a tube bundle assembly. Flow straighteners
Keywords: conditioner, straightener, ultrasonic, multipath, essentially divide the flow into a number of passages that are
flowmeter long and straight and parallel with the axis of the pipe. The
aim is that any rotational component of velocity is reduced or
Background eliminated before the flow exits the conditioner.
In the field of flow measurement it is often necessary to The tube bundle is the most commonly employed form of
condition the flow upstream of a flow metering device in flow straightener, having been standardised to some degree,
order that the flow meter will register flow with a minimal and is essentially an assembly of tubes, typically between 7
error. Bends, valves, filters and other forms of pipeline and 55 in total, arranged either in a hexagonal or circular
component distort the flow velocity profile and by changing geometry, as illustrated in Figure 1. A tube bundle using 19
the flow direction introduce non-axial velocity components tubes of equal size arranged in a circular geometry is
or swirl in the flow stream. It is well known that the included in the International Standard for differential
calibration or flow coefficient of certain types of flow meters pressure flow meters, ISO5167 [1]. Tube bundles are
are affected by distortions of the profile and/or by the typically made to be between two and three pipe diameters in
presence of swirl. length, with the result that the tubes may be 20 to 30 tube
diameters long, though studies have shown that in terms of
Flow conditioners have been employed for many years to limiting swirl, a much shorter length of bundle can still be
partially rectify distorted and swirling flows upstream of effective [2].
flow meters. The various devices deployed to date differ in
design with resulting differences in performance in terms of A recognised deficiency of the tube bundle design of
their ability to rectify flow versus the permanent pressure flow conditioner is that while it is effective at removing
loss that they impose. Most conditioners have a single swirl, the emerging axial velocity profile does not tend to be
specified geometry or a constrained set of design parameters fully developed; that is it generally tends to be flatter than the
and cannot easily be adapted to suit the requirements of a profile that would be found downstream of a long straight
particular situation. length of pipe at the Reynolds number of interest. In order to
try to overcome this limitation, Stuart [3] developed a tube
Multipath ultrasonic meters are now widely employed for bundle flow conditioner where the tube diameters used
measurement of the flow of oil and natural gas. Although within the bundle were varied in order to produce a velocity
some models of ultrasonic flow meter, such as 8-path chordal profile shape closer to the desired fully developed profile. A
meters, are able to perform accurately without the need for a disadvantage of this conditioner design in terms of
flow conditioner, meters with three, four and five paths are manufacturing, which also applies to most tube bundle
typically used with flow conditioners when deployed in
designs, is that when the pipe diameter is varied, the required Optimisation of the design of a thick-plate conditioner
tube diameters may not be readily available in standard sizes with circular passages is also complicated by particular
of tubing. The main advantage of tube bundles is that they issues associated with the chosen circular hole geometry. An
have relatively low permanent pressure loss, having a loss irregular number of holes and the circular shape of the
coefficient for fully turbulent flow in the range of 0.65 to 1.2; passages result in a complex water-shed between adjacent
see for example Baker [4]. rings of holes, and hence make the calculation of the
effective porosity in different areas difficult, as the water-
A further disadvantage of the tube bundle is its variable
shed defines the blockage area associated with each hole.
design and quality. If not constructed to the ISO standard,
Optimisation is further complicated in cases where the
the potential variations in number and size of tubes are
circular passage size is varied, as for a given thickness of
almost endless, making it difficult to predict performance or
plate as this results in variation of both the porosity and the
relate experience from one design of tube bundle to another.
ratio of the length of the passage to its hydraulic diameter.
Furthermore, variable manufacturing quality means that the
As a consequence, the steps that must be taken to optimise a
tube alignment may vary, and in some cases, for example if
conditioner with circular passages are not obvious, as when
the bundle becomes twisted during manufacture, the bundle
changes are made the shape of the water shed varies as well
can actually generate a swirling flow.
as the porosity and the length/hydraulic diameter ratio. This
likely explains some of the observations made by Karnik [7],
where in some cases small changes in hole size resulted in
significant variations in performance, but for other cases it
did not.

Fig. 1 Two typical arrangements of 19-tube tube bundle

conditioners

Flow Conditioners
The need to shape the axial velocity profile as well as
Zanker Laws (CPA/NOVA 50E)
remove swirl was first addressed properly in the design of
the Zanker flow conditioner [5]. The original Zanker
conditioner comprises a thin plate with holes designed to
produce a graded resistance to flow combined with a vane
type straightener attached to the downstream side of the
plate. In terms of the flow profile produced and level of
swirl reduction achieved by this conditioner, it is recognised
as being very effective. However, it is somewhat difficult to
manufacture and has a high pressure loss coefficient of
greater than 5.
More commonly used today are thick-plate type
conditioners. In these designs a graded resistance to flow is Spearman (NEL) Gallagher
achieved by means of making circular passages in a fairly Fig. 2 Typical thick plate conditioner geometries
thick plate. By varying the number, spacing and size of the
circular passages, the desired graded resistance is achieved.
Examples of this type of conditioner include those by Laws A particular advantage of the thick-plate conditioner is
(most common in the Nova/CPA 50E variant) [6, 7], that the manufacture and geometric scaling to different sizes
Spearman [8] and Gallagher [9], in addition to the thick plate of pipe can be achieved very easily, which overcomes the
version of the Zanker conditioner [10], where the thicker manufacturing and quality limitations associated with the
plate negates the requirement for the downstream vane-type tube bundle type of conditioner.
straightener. Common thick-plate conditioners are illustrated
As mentioned previously, the effectiveness of thick-plate
in Figure 2.
conditioners has been found to diminish when the porosity is
These thick-plate conditioners with circular passages are increased too much, with the result that most thick-plate
considered the current state-of-the-art but still have certain conditioners in use today have porosity in the region of
deficiencies. Pressure loss coefficients are typically in the 50 %. When porosity has been increased, the investigators
range of 2 to 5, greater than that available with a tube bundle. have not tended to increase plate thickness to compensate for
Attempts to produce plates of higher porosity and hence the reduction in the length/depth of the hole relative to its
lower pressure loss have generally resulted in a reduction in hydraulic diameter, which may partly explain the diminished
flow conditioning performance [11]. performance. This has led some designers to add
straightening vanes to the conditioner or to employ two A benefit of use of a segmented annular geometry over
stages of conditioning, the first being a straightening vane circular passages is easily explained by reducing one of the
and the second a graded thick-plate conditioner [9, 12]. problems of circular holes down to a simple example.
Consider a conditioner with 19 circular passages of equal
A New Conditioner Design size arranged in a hexagonal pattern. The maximum size that
these passages can be before they merge together is one fifth
Some types of flow meter are more sensitive to the
of the inner diameter of the pipe. Therefore the maximum
condition of the incoming flow velocity field than others. In
free area would be approximately 76 % of the pipe area.
the case of multi-path ultrasonic flow meters, it is often the
With segmented annular passages the free area in any portion
case that if swirl is removed effectively then the meter will
of the conditioner can be larger than this limit, whilst still
be able to perform with high accuracy in a variety of
having sufficient ligament thickness for mechanical strength
different installation conditions. Therefore it is common for
of the conditioner.
tube bundles to be used with ultrasonic meters, owing to their
lower pressure loss characteristics. However, this does not
Central channel
offset three of the disadvantages of tube bundles: first, that
Outer wall
they alter the axial velocity profile in an adverse way;
secondly the fact that they are generally manufactured to be Circular
ligaments
between 2 and 3 diameters long, and thirdly the
manufacturing issues mentioned above that can result in poor
quality conditioners. To overcome these limitations the
intention of the work described here was to produce a new
Subdividing
low pressure loss flow conditioner for use with ultrasonic ligaments
flow meters. In addition to having a low permanent pressure
loss, the requirement was set that conditioner should be easy Segmented
annular channels Plate thickness/channel length
to manufacture in a reproducible way and that it should be
possible to obtain a desirable shape of axial velocity profile.
The new flow conditioner design is based on an Fig. 3 Segmented annular conditioner geometry
arrangement of segmented annular passages, arranged
symmetrically around the centre-line of a circular conduit as
illustrated schematically in Figure 3. The choice of For practical purposes it is convenient to make the length,
segmented annular passages allows the cross-sectional area l, of each segmented annular passage is equal to the others
of the pipe to be easily divided into a predetermined number and to the thickness of the plate, though variations with
of annular rings with the width and separation of the different passage lengths can also be conceived by changing
passages to be freely varied in both radial and tangential the plate thickness associated with each ring. It is however,
directions to obtain a desired value of hydraulic diameter and important to consider the overall length of the passages and
porosity in each ring. Combined with control over the length their hydraulic characteristics. Passages that are too short in
of the passages via selecting the thickness of the conditioner, length will be ineffective at preventing the passage of swirl,
this arrangement of segmented annular passages can be whereas passages that are too long may increase the pressure
optimised to produce a conditioner that will retard swirl and loss or increase the size of the conditioner unnecessarily,
have a desired radial distribution of resistance, in with consequences for manufacturing costs. Furthermore if
combination with a specified overall pressure loss. the ratio of the length of the passage to its hydraulic diameter
Circular ligaments separate the annular areas from one differs from one set of holes to another, then the
another and subdividing ligaments partition each area into characteristics of the flow through the holes may differ
two or more segmented annular passages through which the significantly as a function of the pipe Reynolds number.
fluid can flow. The number and width of circular ligaments The importance of the ratio of hydraulic diameter of the
combined with the number and width of subdividing hole to plate depth (or passage length) is worth further
ligaments determines the overall porosity of the conditioner. consideration and is rarely mentioned in the flow conditioner
Furthermore, by varying the number of subdividing literature, which may help explain some of the observations
ligaments, or the thickness of these ligaments from one area of Karnik [7] and Langshot & Erdal [11].
relative to another, the hydraulic diameter and porosity can
be controlled in each area, and hence the flow resistance can For passages that have a very short length to hydraulic
be graded to produce a desirable flow profile. It is this diameter ratio (l/d) of less than 0.5, swirl will pass easily,
variation of the thickness of the circular and subdividing and after the flow separates at the entrance to the passage, it
ligaments to obtain a particular shape of flow profile that will not reattach inside the passage (this is termed fully
puts this new design into the flow conditioner category, separated flow). For passages of intermediate l/d in the
otherwise the device would fall into the more basic flow region of 0.5 to 1, swirl may still pass, and the flow may or
straightener category. may not reattach inside the passage depending on the flow
conditions prevailing (this is termed marginally
The outer segmented-annular ring of passages have an separated/attached flow). For passages of relatively long l/d ,
outer bounding wall included in the design of the greater than 1, swirl will be suppressed and the separated
conditioning element, but could be open at the outer flow at the entrance to the passage will re-attach (this can be
circumference such that the inside wall of the pipe forms in termed fully reattached flow). In terms of the pressure loss
the outer wall of each of those passages. for flow through a passage between two sections of pipe it
can be shown that the pressure loss is greatest for fully The new conditioner can be designed either to fit fully
separated flow, and reduces to a minimum once the flow is inside a pipe section with some means of securing it in place,
fully reattached [13]. Beyond the minimum pressure loss or it can be designed to fit between pipe flanges. For
point the pressure loss will increase again owing to increased increased mechanical strength it is preferable that the
frictional losses in the passages of the conditioner. Therefore segmented annular passages be designed with rounded
it is possible to optimise a flow conditioner in terms of the internal corners.
length to hydraulic diameter ratio.
In terms of practical preference, the conditioner would be
For the common thick-plate conditioners available today installed as a single unit. However, two or more units could
that use circular passages, the thickness of the plate is be installed in series with some separation in between in
constant and normally in the range of 0.12 to 0.15 pipe order to perform more effective flow conditioning. In terms
diameters. The hole diameters are typically in the range of of conditioning performance versus overall pressure loss, this
0.1 to 0.19 relative to the pipe diameter, with resulting l/d may be preferable to using a single unit.
values in the range of 0.63 to 1.5. As pointed out earlier, the
range of l/d corresponding to marginally separated/attached The Design Process
flow is typically between 0.5 and 1. In the likes of the Laws, The design and optimisation of a particular conditioner
Gallagher and Spearman plates, a range of values of passage geometry according to these principles begins with defining
diameter are used, with the result that under certain flow the required characteristics of the conditioner in terms of
conditions some passages may have separated flow whereas overall pressure loss and desired axial flow profile shape. At
others may have reattached flow. this stage any other constraints or requirements can be added
It is desirable to avoid the possibility of having both such as the overall thickness of the plate, the minimum
separated and reattached flow conditions occurring in length to hydraulic diameter ratio, the minimum width of
different passages of the same conditioner at the same time. ligaments, the minimum radius of the inside corners and/or a
One solution to this would be to increase plate thickness until specification that all passages have the same hydraulic
fully reattached conditions occur in the passages of largest diameter. Next the general characteristics of the conditioner
hydraulic diameter. However, when this is done the pressure are considered in terms of the approximate total number and
loss coefficient would then increase undesirably in the size of the passages. Once the number of annular rings and
passages of smaller hydraulic diameter. It is therefore the number of segments per ring has been determined, values
attractive to be able to produce a conditioner design where are chosen to produce an initial design and then the
the value of hydraulic diameter of all passages is the same. optimisation of the conditioner can begin (or indeed, it is also
conceivable that the optimisation process could include
When the hydraulic diameters of each passage are the varying the number of rings and segments per ring).
same, then with a conditioner of constant thickness the
values of l/d will also be the same, and consequently the The thickness of the circular ligaments between annular
coefficient of frictional pressure loss through the hole should rings and the thickness of the subdividing ligaments are set
also be the same. For circular passages, the hydraulic to initial values chosen from a practical and cost perspective.
diameter is simply equal to the diameter, and therefore for l/d The radial width of the segmented annular passages are set to
to be constant the passages must all have the same diameter, initial values (for example approximately equal) such that the
which imposes unwanted restrictions on the geometric total radial width of the passages plus the circular ligaments
arrangement of the holes in terms of producing the desired sums to the diameter of the conditioner. The thickness of the
graded resistance. For segmented annular passages, the plate is set to an initial value.
hydraulic diameter is equal to four times the cross-sectional The porosity and hydraulic diameter are then calculated
area divided by the perimeter. Therefore the hydraulic for each ring. This, in addition to knowledge of the thickness
diameter is a function not only of the cross-sectional area of of the conditioner, allows the pressure loss coefficient and
the passage, but also the aspect ratio of the passage. This relative velocity to be estimated for each segmented annular
provides greater flexibility in design when it is desirable to ring. In practice this can be achieved using semi-empirical
vary or optimise the conditioner design in terms of both pressure loss models that relate these terms, such as those
porosity and hydraulic diameter. described by Ward-Smith [13] or Idelchik [14].
The segmented annular conditioner can be manufactured Alternatively, the profile and pressure loss characteristics can
from a variety of materials by a method chosen to suit the be determined by means of computational fluid dynamics or
materials of construction. For conditioners that are to be by experimental testing. The geometry is then iteratively
used in small pipes it is most likely that the conditioner adjusted until the desired velocity profile and other optimal
would be manufactured from a solid part made of metal or characteristics are achieved.
plastic with the passages cut into the material using machine Some trial and error is required in terms of the starting
tools. Techniques such as water-jet cutting are appropriate conditions and constraints in order to obtain convergence and
for some materials up to a certain thickness or conventional produce a solution that has the required characteristics, and
drilling and milling techniques can be employed. In the when computational methods are used in the design, final
design process the number of rings and number of testing and some fine tuning may be required.
subdividing ligaments can be constrained in order to produce
an appropriate balance between material and manufacturing A New Flow Conditioner
cost.
This design process will now be described in detail as
applied to produce a new conditioner with the desirable
characteristics in the form of a pressure loss coefficient of passage back closer to its target value. In the fifth and final
less than 1, and each passage in the conditioner having the step the desired velocity profile and loss coefficient was
same hydraulic diameter. sought, again by means of adjusting the width of the radial
ligaments.
Common thick-plate type conditioners have between 25
and 32 circular passages, with the outer holes typically being Figures 4 and 5 illustrate the convergence of the velocity
sized at approximately 10 % of the pipe diameter. Therefore profile and the hydraulic diameters respectively towards their
for a conditioner design with a broadly similar number and target values at each of these optimisation steps.
size of passages the process began by dividing the pipe into a 0.3
central circular passage and three annular rings, the radial

Deviation from target value of normalised velocity

width of each ring being less than 14 % of the diameter. 0.2

Then using radial subdividing ligaments, and taking into

consideration a desire for symmetry, the inner annulus was 0.1

partitioned into six segments, and each of the outer annuli

0.0
into 12 segments, resulting in 31 passages in total. The 0 1 2 3 4 5 6

thickness of the circular and subdividing ligaments were set -0.1

to an arbitrary value to start, at 1 % of the pipe diameter. As Outer annulus
an initial condition a width of 0.12 D was chosen for each of -0.2
Middle annulus
the annular sections, with the result that the starting diameter Inner annulus
-0.3
of the central circular passage was 0.2 D (given the Centre hole
constraint that the outer wall and circular ligaments plus the
-0.4
width of the passages must sum to equal the pipe diameter). Iteration step

At this point the hydraulic diameter of the largest passage

Fig. 4 Graph showing convergence of the design towards the desired
is 0.2 D. Given that it is desirable to target an l/d value of velocity profile
greater than 1, a plate thickness of 0.2 D was chosen, which
ensures that the l/d requirement is met, though that could also
have been adjusted as part of the optimisation process. 0.21

Outer annulus
In the case of a single central passage its porosity is 0.20
Middle annulus
determined by its diameter and by the thickness of the 0.19
Inner annulus
Normalised hydraulic diameter

circular ligament that separates it from the first annular ring. Centre hole
0.18
Therefore the first step of the optimisation was to adjust the
other geometric parameters until the pressure loss coefficient 0.17

of this passage is close to the target value for the conditioner 0.16
as a whole. This first step was achieved by setting the
0.15
constraint that the circular ligaments should be of equal
thickness (with the exception of the outer wall, which was 0.14

fixed at 0.01 D), and then gradually increasing the thickness 0.13
of those, concurrently reducing the diameter of the central
0.12
passage, until the desired loss coefficient was achieved. 0 1 2 3 4 5 6
Iteration step
The second step was to adjust the width of the radial
ligaments in each ring until the desired velocity profile shape Fig. 5 Graph showing convergence of the design in terms of the
hydraulic diameter of the passages
was achieved, whilst also considering the target for the
overall loss coefficient. In this step the value for the width of
the radial ligaments in each annulus was adjusted iteratively The resulting conditioner design was then manufactured
until the desired profile shape and loss coefficient was complete with a flange for installation between pipe sections,
achieved. After this step the result is a design which that as shown in Figure 6. Observe the variation in ligament
should produce the desired velocity profile, and have the width from inner to outer rings.
intended overall loss coefficient, but may not yet meet some
other requirements such as every passage having the same
hydraulic diameter.
To obtain the same hydraulic diameter in each ring the
next step was adjust the radial width of each of the
segmented annular channels to satisfy the condition that all
hydraulic diameters are the same, keeping all other
dimensions constant, with the exception of the diameter of
the central passage, which was also allowed to change.
When the adjustment is made in this way, the relative
velocities diverge again from their target values, requiring
further iterations to be made.
Fig. 6 New conditioner including rounded internal corners and flange
In the next step, the widths of the circular ligaments were
adjusted again to bring the loss coefficient for the central
 0.998
Performance Tests
0.996
In order to evaluate the effectiveness of the new flow 0.994
conditioner, tests were carried out in the ISO 17025
0.992
accredited Cameron flow calibration laboratory at the Caldon
Ultrasonics Technology Centre in Pittsburgh, Pennsylvania. 0.990

Meter factor
A Caldon LEFM 280Ci ultrasonic flow meter with eight 0.988

flow velocity measuring paths was employed to determine 0.986

the effectiveness of the flow conditioner. The 8-path meter 0.984
has been described in detail elsewhere, see for example Straight pipe, no conditioner
0.982
Brown et al [15]. In its 8-path format the LEFM flow meter 10D downstream of bends, no conditioner, paths horizontal
does not normally require a flow conditioner. However in 0.980 10D downstream of bends, no conditioner, paths vertical
this case, the data from the eight measurement paths was 0.978
evaluated in the form of two 4-path combinations equivalent 0 100,000 200,000 300,000 400,000 500,000 600,000
to the Caldon LEFM 240Ci flow meter, in order to determine Reynolds number
the influence on the hydraulic correction factor of each 4-
Fig. 7a Meter factor versus Reynolds number for meter A in straight
path meter. Furthermore, all eight paths were combined to pipe and downstream of bends with no flow conditioner
give a measure of the average swirl in the form of a ratio of
the tangential velocity to the mean axial velocity.
0.998
The new conditioner was tested downstream of a long
straight pipe and then downstream of an arrangement of six 0.996

out of plane bends known to generate swirl and produce 0.994

asymmetric distortion of the axial velocity profile. The tests
0.992
were conducted using a kerosene substitute fluid with a
viscosity of approximately 3 cSt over a range of flowrates in 0.990
Meter factor

the range of 74 to 740 m3/hr in a 6-inch pipe. A Laws type 0.988

conditioner (Nova/CPA 50E variant) was also tested in the 0.986
same configuration. The arrangement of bends was kept the
0.984
same for all tests. The meter factor data was obtained by
Straight pipe, no conditioner
calibration using a unidirectional ball prover with an 0.982
10D downstream of bends, no conditioner, paths horizontal
uncertainty of +/-0.044 % as the traceable reference standard. 0.980 10D downstream of bends, no conditioner, paths vertical

Test data is presented for the following installation 0.978

combinations: 0 100,000 200,000 300,000 400,000 500,000 600,000
Reynolds number

Long straight pipe with no flow conditioning Fig. 7b Meter factor versus Reynolds number for meter B in straight
Long straight pipe with a Laws type conditioner ten pipe pipe and downstream of bends with no flow conditioner
diameters upstream of the flow meters
Long straight pipe with the new conditioner ten pipe
diameters upstream of the flow meters Figures 8a and 8b show the results for meters A and B
Flow meters at ten pipe diameters downstream of six installed ten diameters downstream of a Laws type thick-
bends with no flow conditioning, with the measurement plate conditioner. The difference between the straight pipe
paths orientated horizontally case and the case where the conditioning plate is four
Flow meters at ten pipe diameters downstream of six diameters downstream of the bends is typically on the order
bends with no flow conditioning, with the measurement of 0.1 % or less.
paths orientated vertically The data shown in Figures 8a and 8b can be summarised
Laws type conditioner four pipe diameters downstream of quantitatively in terms of a flow weighted mean error shift.
six bend, with the flow meters ten pipe diameters The results of this calculation are given in Table 1.
downstream of the conditioner
New conditioner four pipe diameters downstream of six Table 1
bend, with the flow meters ten pipe diameters
downstream of the conditioner
Flow weighted mean error shift
Laws type New
Figures 7a and 7b show the results for each of the 4-path thick-plate conditioner
meters (A and B) in the straight pipe configuration and at ten Meter A 0.08% -0.06%
diameters downstream of the bends without flow Meter B 0.09% 0.10%
conditioning, with the measurement paths in both horizontal
and vertical orientations. It can be observed that under these
conditions, with no flow conditioning, the swirl and
distortion generated by the bends results in changes in meter
factor having magnitude typically between 0.3 to 0.5 %.
 0.998 0.998

0.996 0.996

0.994 0.994

0.992 0.992

0.990 0.990
Meter factor

Meter factor
0.988 0.988

0.986 0.986

0.984 0.984
Straight pipe, Laws type conditioner
0.982 Straight pipe, New conditioner
0.982
Bends - 4D - Laws conditioner -10D - Meter Bends - 4D - New conditioner -10D - Meter
0.980
0.980

0.978
0.978
0 100,000 200,000 300,000 400,000 500,000 600,000 700,000
0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000
Reynolds number Reynolds number

Fig. 8a Meter factor versus Reynolds number for meter A in straight Fig. 9a Meter factor versus Reynolds number for meter A in straight
pipe and downstream of bends with the pipe and downstream of bends with the
4DLaws conditioner10DMeter arrangement 4DNew conditioner10DMeter arrangement

0.998 0.998

0.996 0.996

0.994 0.994

0.992 0.992

0.990 0.990
Meter factor

Meter factor

0.988 0.988

0.986 0.986

0.984 0.984

0.982 0.982
Straight pipe, Laws type conditioner Straight pipe, New conditioner
0.980 0.980
Bends - 4D - Laws conditioner -10D - Meter Bends - 4D - New conditioner -10D - Meter
0.978 0.978
0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 0 100,000 200,000 300,000 400,000 500,000 600,000 700,000 800,000
Reynolds number Reynolds number

Fig. 8b Meter factor versus Reynolds number for meter B in straight Fig. 9a Meter factor versus Reynolds number for meter B in straight
pipe and downstream of bends with the pipe and downstream of bends with the
4DLaws conditioner10DMeter arrangement 4DNew conditioner10DMeter arrangement

Figures 9a and 9b show the results obtained using the Figures 10 and 11 show the swirl quantified in terms of
new conditioner previously described. It can be readily the tangential velocity as a percentage of the mean axial
observed that the difference between the straight pipe case velocity. In Figure 10 we show the bare straight pipe case
and the case where the conditioning plate is four diameters plus the swirl generated by the bends and measured ten
downstream of the bends is similar to the case for the Laws diameters downstream with no flow conditioning. It is clear
type conditioner, typically on the order of 0.1 % or less. that the bends generate a high level of swirl. Figure 11
These results can also be summarised quantitatively in terms shows the results for the two flow conditioners tested here,
of a flow weighted mean error shift, as reported in Table 3. the Laws type conditioner and the new conditioner.
Comparing Figures 10 and 11 it is clear that both
The data recorded in Table 3 shows that in terms of the
conditioners substantially reduce swirl. At the higher
flow measurement performance of a 4-path ultrasonic meter,
Reynolds numbers it appears that the Laws type conditioner
the new conditioner matches the performance of the Laws
is slightly more effective at reducing swirl than the new
type thick plate conditioner as the flow weighted mean error
conditioner. However, when the measurement results of
shifts are of a similar magnitude, all being less than 0.1 %.
Table 3 are taken into consideration, this appears to be an
insignificant difference in terms of 4-path ultrasonic meter
performance.
 5%
to the Nova/CPA 50E variant of the Laws conditioner, which
has a loss coefficient of approximately 2 [1].
0%

Straight pipe data

DISCUSSION AND CONCLUSIONS
-5%
Downstrream of bends, no flow conditioner, paths horizontal A new design of flow conditioner using segmented
Downstream of bends, no flow conditioner, paths vertical annual passages has been described which by varying the
Swirl

-10% number and width of circular and subdividing ligaments

allows design targets to be reached via a relatively simple
-15% process of design and optimisation.
The general design is versatile and allows for variations
-20% in plate thickness, number of holes, pressure loss and
velocity profile to suit different application requirements.
-25%
0 100,000 200,000 300,000 400,000 500,000 600,000 A conditioner based on the new design with 31 passages
Reynolds number has been built and tested in an accredited calibration
laboratory. The results of the tests show that for use with a
Fig. 10 Swirl measured using the 8-path ultrasonic meter in straight
pipe and 10D downstream of bends with no flow conditioning 4-path ultrasonic meter, the measurement results with the
new conditioner are equivalent to the Laws type plate, but
are achieved with less than half the pressure loss.
5%

4%

3% REFERENCES
2% [1] ISO 5167-1 (2003) Measurement of fluid flow by means of pressure
differential devices inserted in circular cross-section conduit running
1%
full Part : General principles and requirements
Swirl

0%
[2] Laws, E M (1991) A further study of flow through tube bundles
FLUCOME 91, San Francisco, CA
-1%
New conditioner in straight pipe [3] Stuart, J W (1995) "Gradational Tube Bundle Flow Conditioner for
New conditioner downstream of bends
Providing a Natural Flow Profile to Facilitate Accurate Orifice
-2%
Metering in Fluid Filled Conduits," U. S. Patent 5.392,815
Laws type thick plate conditioner in straight pipe
-3% [4] Baker, P D (2006) Effects of flow conditioning on liquid
Laws type thick plate conditioner downstream of bends measurement, Paper 2120, the 81st International School of
-4% Hydrocarbon Measurement.
[5] Zanker, K J (1960) The development of a flow straightener for use
-5%
0 100,000 200,000 300,000 400,000 500,000 600,000 700,000
with orifice-plate flowmeters in disturbed flows, Flow Measurement
Reynolds number
in Closed Conduits, Vol 2, a Symposium held at the National
Engineering Laboratory, Scotland
Fig. 11 Swirl measured using the 8-path ultrasonic meter in straight [6] Laws, E M (1990) Flow conditioning A new development, Flow
pipe and downstream of bends with the 4D conditioner 10D Meter Measurement and Instrumentation, Vol 1, April 1990, pp. 165 - 170
arrangement, for both the Laws type conditioner and the new [7] Karnik, U (1995) A compact orifice meter/flow conditioner package
conditioner International Symposium on Fluid Flow Measurement, San Antonio,
Texas
[8] Spearman, E P, Sattary, J A, and Reader-Harris, M J (1994)
Comparison of velocity and turbulence profiles downstream of
During the tests, the pressure loss across the new perforated plate flow conditioners, Flomeko 94, Flow Measurement
conditioner was measured. The measurements of pressure in the Mid-90s, NEL, Scotland
[9] Gallagher, J E, Lanasa, P J, and Beaty, R E (1994) The Gallagher
loss can be converted into a dimensionless loss coefficient, flow conditioner North Sea Flow Measurement Workshop, Scotland,
which is a useful non-dimensional measure of the energy lost October 1994
when flowing through the conditioner. The pressure loss [10] Laws, E M and Ouazzane, A K, (1995) A further study into the
data is shown in Table 2. effect of length on the Zanker flow conditioner, Flow Measurement
and Instrumentation, Vol. 6, No. 3, pp. 217-224
[11] Langshot, M and Erdal, A (1998) Design parameters for perforated
plate flow conditioners Flomeko, Lund, Sweden, June 15-17, 1998
Table 2 Pressure loss data [12] Laws, E M and Ouazzane, A K (1994) Compact installations for
orifice pate flowmeters, Flomeko 94, Flow Measurement in the Mid-
Loss 90s, NEL, Scotland
Differential Differential
Flowrate Density Velocity coefficient, [13] Ward-Smith, A J (1971) Pressure Losses in Ducted Flows,
pressure pressure Butterworths, London
K
m3/hr kg/m3 psi m/s Pascals - [14] Idelchik, I E (2003) Handbook of Hydraulic Resistance, 3rd Edition,
Jaico Publishing House
740 800 6.9 11.3 47574 0.937 [15] Brown, G J, Augenstein, D R and Cousins, T (2006) An 8-path
607 800 4.5 9.2 31026 0.908 ultrasonic master meter for oil custody transfers, XVIII IMEKO
474 800 2.5 7.2 17237 0.827 World Congress, Rio de Janeiro, Brazil
340 800 1.5 5.2 10342 0.965
Average 0.91

For the new conditioner, the average loss coefficient is

0.91. This is less than half of the pressure loss corresponding