kelner installation effects article

8/3/2019 Kelner Installation Effects Article

1/12

Flow Meter Installation Effects Page 1 of 12

FLOW METER INSTALLATION EFFECTS

Eric Kelner, P.E.The Letton-Hall Group

Presented at the Appalachian Gas Measurement Short Course

INTRODUCTION

Meter station piping installation configuration is oneof a number of variables that may adversely affectmeter accuracy. Some piping configurations candistort the flow stream and produce flowmeasurement bias errors (i.e., offsets in the meteroutput) of up to several percent of reading. Valves,elbows, or tees placed upstream of a flow meter arejust some of the piping elements that can distort theflow stream. In this paper, installation effects arediscussed with respect to two of the four maincomponents of a flow measurement system: themeter, or primary element, and the secondary(pressure and temperature) instrumentation. The

effect of the velocity profile of the flow stream onorifice, ultrasonic, and turbine flow meters isdiscussed next. Installation conditions that mayadversely impact the accuracy of pressure andtemperature measurements are discussed afterthat. The gas chromatograph and the flowcomputer, the third and fourth components, aretreated in separate courses.

VELOCITY PROFILE EFFECTS

Typically, one or more quantities (such as thepressure differential created across an orifice, the

number of rotations produced by a turbine rotor, orthe amount of time it takes for a pulse of ultrasonicenergy to traverse the flow stream) are used, alongwith a flow equation, to determine the flow rate.These equations typically assume an ideal flowstream at the meter. Usually, the ideal flow streamis a fully-developed, symmetric, swirl-free turbulentflow profile.

This ideal flowing condition is depicted as avelocity profile in Figure 1. The velocity profiledescribes the change in velocity across the cross-section of the pipe at a given axial location. At thepipe wall, the velocity of the fluid is zero (a.k.a., the

no-slip condition). The maximum fluid velocity isat the axial centerline of the pipe. The shape of thevelocity profile very near the pipe wall is determinedby the viscosity of the fluid, the pipe wall roughness,and the Reynolds number (i.e., the ratio of inertial toviscous forces). A fully-developed turbulent velocityprofile is symmetric about the pipe centerline.

Figure 1. Fully-developed turbulent velocityprofile from the axial centerline of pipe to the

pipe wall. The profile is symmetrical about thepipe centerline

1.

A less than ideal velocity profile is one that isdistorted in some way. For example, the flow justdownstream of a 90 elbow will cause a fully-developed, symmetric, swirl-free velocity profile tochange to one that has two counter-rotating vortices(i.e., type-2 swirl) and an asymmetric or skewedvelocity profile. This is the type of flow distortionthat can introduce a bias into the flow ratemeasurement if the flow meter reads correctly whenthe flow field is ideal. The bias can be significant,on the order of several percent of reading,depending on the severity of the disturbance. Sucha condition is typically referred to as an installationeffect. In general, flow meter installation effectsoccur in three stages

2:

1. The creation of velocity profile disturbancesdue to the effects of the piping configuration

upstream of the flow meter. Commonupstream piping configurations known toproduce velocity profile distortions include asingle elbow, two elbows in an in-planeconfiguration, two elbows in an out-of-planeconfiguration, and a partially-closed valve.Disturbed velocity profiles may beasymmetric, contain swirling motions (i.e.,solid body rotation or two counter-rotatingvortices), or have a combination of the two.


2/12


2. The decay of these velocity profiledisturbances, usually as a result of turbulentdiffusion and pipe wall friction. Differenttypes of flow disturbances decay at differentrates, but all disturbances require relativelylong lengths of straight pipe to re-establisha fully-developed, symmetric, swirl-freeturbulent velocity profile. An example of the

decay rate of a disturbance associated witha single elbow is shown in Figure 2. Thevelocity profile just downstream of the elbowhas profile asymmetry plus two counter-rotating vortices. In this example, the effectpersists for approximately 59 pipediameters. Some disturbances, such as theone produced by two elbows out-of-plane(i.e., solid-body rotation or type-1 swirl), canpersist for 200 pipe diameters or more.

3. The response of a specific flow meter to thevelocity profile presented at the meter inlet.Some flow meters, such as orifice meters,

are highly sensitive to distortions in velocityprofile. Others, such as some ultrasonicmeters, may be sensitive to velocity profiledistortions, but include computationalalgorithms that may correct for someamount of flow distortion.

-1.0 -0.8 -0.6-0.4 -0.2 -0.0 0.2 0.4 0.6 0.8 1.0Normalized Horizontal Position (X)

0.90

0.95

1.00

1.05

1.10

1.15

NormalizedVe

locity

10D40D59D78D97D

n=10.13

Power Law

Figure 2. The decay of the effect of a single, 90degree long radius elbow

3.

Piping configurations showing a single elbow andtwo elbows out-of-plane are shown in Figure 3.Their associated velocity profiles at an axial distanceapproximately 10 pipe diameters downstream areshown in Figure 4. Note the combination of swirland asymmetry produced by both configurations.

These velocity profiles can cause significant flowrate measurement bias, depending on the responseof the flow meter to disturbed velocity profiles.

Flow Conditioners

A flow conditioner can be used to offset the effect offlow field disturbances caused by the upstreampiping and to reduce the amount of straight pipelength required to allow a flow distortion to fullydissipate. Flow conditioners adjust the flow field,ideally eliminating or greatly reducing the magnitudeof the flow distortion caused by the upstream pipingconfiguration. However, as will be shown later,there is no flow conditioner that can completelyisolate a flow meter from all possible flow fielddistortions. Some high-performance conditionersare effective at isolating a fairly broad range of flowdistortions propagating from upstream

4. Figure 5

shows examples of several common flowconditioners.

Figure 3. Two common upstream pipingconfigurations.

Double Elbow

Single Elbow


3/12


-0.6 -0.4 -0.2 0.2 0.4 0.6 0.8 1.0Normalized Horizontal Position (X)

-1.0

-0.4-0.2-0.00.20.40.60.81.0

NormalizedVerticalPosition(Y)

0.900 0.9280.956

0.9560.956

0.9560.983

0.983

0.9831.011

1.0391.039

-1.0 -0.8 -0.6 -0.4 -0.2 -0.0 0.2 0.4 0.6 0.8 1.0Normalized Horizontal Position (X)

-1.0

-0.8

-0.6

-0.4

-0.2

-0.0

0.20.40.60.81.0

NormalizedVertical

Position(Y)

0.900

0.928

0.956

0.956

0.983

0.983

0.983

0.983

1.011

1.039

Figure 4. Velocity profile asymmetry and swirl

10 pipe diameters downstream of (1) two elbowsout-of-plane (top) and (2) a single 90 elbow

(bottom). Note high amount of swirl representedby the velocity vectors and velocity profile

asymmetry represented by contour line shifts(and changes in color) away from the axial

centerline (i.e., center of the grid)5.

Figure 5. Example flow conditioners (from left toright: 19-tube bundle, Canadian Pipeline

Accessories CPA 50E Plate, Gallagher FlowConditioner TAS, and Vortab Flow Conditioner).

ORIFICE METER INSTALLATION EFFECTS

The orifice meter is one of the oldest and mostcommon flow meters in the natural gas industry.Tens of thousands of orifice meters are in servicethroughout the United States. Applications rangefrom low volume, remote well-head flowmeasurement to high-volume pipeline custodytransfer stations. Several types of orifice meterfittings are available. The two most common are thesenior fitting, which provides the means to changeorifice plates while the pipeline is under pressure,and the flange fitting, in which the plate ispermanently mounted between pipe flanges.

The orifice meter is categorized as a differentialpressure producer. When the flow is acceleratedthrough the orifice bore (i.e., a restriction in the flowstream), a differential pressure is produced acrossthe orifice plate that is related to the flow ratethrough the orifice. Figure 6 illustrates thedifferential pressure created across an orifice plate.

Figure 6. The differential pressure created asflow is accelerated through an orifice.

The orifice flow rate equation is as follows:

2

1 ,m d v t P q NC E Yd P = Equation 1

where:q

m is the mass flow rate, in lbm/s,N

1 is a unit conversion factor (for the units

given here, N1 = 6.30025),C

dis the orifice plate discharge coefficient

and is non-dimensional,Ev is a velocity of approach factor and is

non-dimensional,Y is an expansion factor (Y = 1.0for

incompressible fluids, but Y


4/12


d is the orifice plate bore diameter in feet,D is the meter tube diameter in feet,

is the beta ratio, d/D ( is non-dimensional),

t,P

is the density of the fluid in lbm/ft3,

P is the orifice differential pressure in

lbf/ft2.

Orifice Discharge Coefficient

The orifice equation calculates the actual flow ratewhen the assumptions associated with its derivationare true. One of the principal effects on accuracy isthe geometry of the orifice bore

6. The flow

restriction produced by the orifice creates asignificant change in flow direction commonlyreferred to as the vena contracta. In this region ofthe flow field, the streamlines of the flow contractand energy is lost to recirculation zones locatedimmediately upstream and downstream of the orificeplate. These recirculation zones are created when

the boundary layer of the flow separates from thepipe wall. An empirically-derived dischargecoefficient, Cd, accounts for this effect in the orificeflow rate equation (Equation 1).

The discharge coefficient is defined as the ratio ofthe actual flow rate to the theoretical flow rate. Fororifice meters, the discharge coefficient is typicallyaround 0.6. The U.S. gas industry uses the Reader-Harris/Gallagher (R-G) equation to calculate thedischarge coefficient. The R-G equation wasderived using a database of discharge coefficientmeasurements taken by flow laboratories around theworld. The discharge coefficients included in thedatabase were measured when the pipe flow had afully-developed, symmetric, swirl-free, turbulentvelocity profile. Figure 7 shows dischargecoefficients measured at several labs around theworld, compared with the R-G equation and its 95%confidence interval. With respect to installationeffects, the discharge coefficient calculated usingthe R-G equation will likely differ from the actualdischarge coefficient if the turbulent velocity profile isnot fully-developed, symmetric, and swirl-free.

Installation Effects Tests

Between 1987 to 2000, Southwest ResearchInstitute and several other research organizations,such as such as the NOVA Technology Center inCanada and the National Institute for StandardsTechnology (NIST), conducted a comprehensive testprogram to quantify the effect of upstream pipinginstallations on the orifice discharge coefficient.Tests were conducted under various disturbedvelocity profile conditions. The test matrix includedthe use of several beta ratios and pipe diameters,

Pipe Reynolds Number

105 106 107

OrificeC

d

0.600

0.602

0.604

0.606

0.608

0.610

0.612

0.614

0.616

0.618

0.620

= 0.57 and 0.60

250 mm Orifice Meter Tube Calibration

NIST CEAT

NELDHL Ruhr Gas

British GasGasunie

A.G.A. Report No. 3 (RG) equation for flange tapped o rifice coefficient, = 0.57

MRF - 45 D, tap #1

A.G.A. Report No. 3 (RG) equation for flange tapped orifice coefficient, = 0.60

Gas dataWater data

95% confidence interval for RG equation = 0.57

MRF - 95 D, tap #1MRF - 95 D, tap #2

Figure 7. Orifice discharge coefficient data fromseveral flow laboratories, compared with the R-G

equation and its 95% confidence interval.

with and without flow conditioners installed upstreamof the test flow meters. The purpose of this workwas to determine the appropriate length of straightpipe required upstream of the flow meter and the

appropriate placement of a flow conditioner, whenused, to produce a discharge coefficient that agreedwith the calculated discharge coefficient within the95% confidence interval of the R-G equation. Thisresearch was conducted in support of the 2000revision of AGA Report No. 3, Orifice Metering ofNatural Gas and Other Related Hydrocarbon Fluids.Figure 8 shows the results of one of the installationeffects tests conducted at SwRI.

Design Guidelines AGA Report No. 3

The gas industry standard for orifice meterinstallations, American Gas Association (AGA)

Report No. 3, Part 27

, was revised in 2000 to reflectthe results of the preceding installation effectsresearch. The revised standard included arecommendation for the minimum length of straightpipe required upstream of an orifice meter when noflow conditioner is used (up to 145 diameters of pipemay be required ahead of the meter). Figure 9shows the typical bare-tube installation configurationrecommended in the standard. The minimumrecommended length of straight pipe between theorifice meter and several common flow distortingconfigurations are also provided (see Table 1). Acatch-all category was also provided for upstream

piping configurations not specifically referenced inthe standard, such as complex headers found inmany meter stations.


5/12


Orifice ratio

0.2 0.3 0.4 0.5 0.6 0.7 0.8

C

d,

%D

eviationfrom

baselineC

dvalue

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

A = 17 D, Tap #1A = 17 D, Tap #2

MRF HPL Results for 102mm Bare Meter TubeDownstream of Two 90o Elbows Out-of-Plane

A = 45 D, Tap #1A = 45 D, Tap #2

Symbols

"High Performance" Flow Conditioner

C, Distance from exit of flow conditioner to orifice plate

0 5 10 15 20 25 30 35 40 45

C

d,

%D

eviationfro

m

BaselineC

dValue

-2.0

-1.8

-1.6

-1.4

-1.2

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

1.0

Symbols

A' = 17 D, tap #1A' = 17 D, tap #2A' = 45 D, tap #1A' = 45 D, tap #2

GRI MRF HPL sliding flow conditioner testsUpstream disturbance, two 90o elbows out-of-plane

= 0.67

Figure 8. Example of the shift in orifice Cd fortwo elbows out-of-plane. The upper figure

shows the ratio dependence of the Cd shift forbare meter tubes with 17 and 45 diameters of

straight pipe upstream of the meter. The lowerfigure shows the effect of placing a highperformance flow conditioner at various

locations between the orifice and the element(s)that produced the upstream flow disturbance.

Figure 9. Bare Orifice Meter Tube Installation.8

For installations utilizing a 19-tube bundle (Figure10), the upstream lengths of straight pipe arereduced to a maximum of 14.5 diameters for anupstream disturbance caused by two elbows out-of-plane when the upstream length is between 17 and

29 pipe diameters and the ratio is 0.67 or less(Table 2).

Distortion Source RecommendedUpstream Length (UL)for Maximum Range

(Bare Meter Tube)

Single 90 elbow 44 D ( 0.75)

Two 90 elbows out-of-plane, < 5 D apart

95 D ( 0.75)

Single 90 tee used asan elbow

44 D ( 0.75)

Gate valve at least50% open

44 D ( 0.75)

Any other configuration 145 D ( 0.75)

Table 1. Upstream Lengths Required for OrificeMeters Without a Flow Conditioner

9.

Figure 10. Orifice Meter Tube Installation with19-Tube Bundle

10.

Table 2. Upstream Lengths Required for OrificeMeters With a 19-Tube Bundle Flow

Conditioner11

.

ULTRASONIC METER INSTALLATION EFFECTS

The ultrasonic meter is now a fairly common custodytransfer meter in the gas transmission segment ofthe natural gas industry. It is gradually making itsway into applications further upstream anddownstream as well. The ultrasonic meter is arelatively new technology. It has been used fornatural gas custody transfer for only about ten years

Distortion Source RecommendedUpstream Length

(17D UL 29D) forMaximum Range(19-Tube Bundle)

Single 90 elbow 13 D ( 0.67)Two 90 elbows out-of-plane, 2 D apart

13.5 to 14.5 D ( 0.67)

Single 90 tee used as anelbow

13 D ( 0.54)

Gate valve at least 50%open

9.5 D ( 0.47)

Any other configuration 9.5 D ( 0.46)


6/12


in the United States. Initial entries into the marketcarried claims of extremely high turndown ratios (onthe order of 100:1) and insensitivity to velocity profilevariations. Research performed by SwRI and othersdetermined that these meters were capable ofaccurate measurement at turndown ratios closer to20:1 and quantified their sensitivity to velocity profiledistortions (i.e., installation effects).

An ultrasonic meter infers the flow rate by measuringthe time interval it takes for a high-frequency (i.e.,ultrasonic) pulse of energy to travel through aflowing gas stream. This interval is commonlyreferred to as the transit time. Ultrasonic pulsesare transmitted diagonally across the pipe bytransducer pairs (Figure 11). Each transduceralternates as a receiver and transmitter. When thepulse travels in the direction of the flow, the pulsevelocity equals the velocity of sound of the flowstream, plus the velocity of the flowing gas. Whenthe pulse reverses direction, the pulse velocityequals the velocity of sound minus the velocity of the

flowing gas. The transit time of the pulse traveling inthe direction of the flow is shorter than the transittime of the pulse traveling in the opposite direction.The difference in transit time is used to infer thevelocity of the fluid stream.

L

X

Profile

Direction

Flow

InletVelocity

2

Acoustic Paths

1

Exaggerated

Figure 11. Ultrasonic meter operating principle.

Ultrasonic meters used for custody transfer typicallyhave several pairs of ultrasonic transducers. Thesemulti-path meters use different path configurationsto sample a larger portion of the velocity profile thanthe simpler single-path meter discussed previously.Figure 12 shows two common ultrasonic meteracoustic path geometries: the chordal-path and the

bounce-path. The chordal-path meter essentiallyslices the velocity profile to determine the velocityalong the cross-section of the pipe. The bounce-path configuration uses reflections of the ultrasonicpulse off the pipe wall to extend the transit timeinterval. Because multi-path meters provideimproved resolution of the velocity profile,manufacturers have developed some relativelycrude methods to infer swirl and velocity profileasymmetry, thereby reducing (but not eliminating)

the sensitivity of this type of meter to changes in thevelocity profile.

Chord A

Chord B

Chord C

Chord D

Diametral Paths

Swirl paths Figure 12. Two common ultrasonic meter path

geometries. The chordal-path geometry isshown on the left. The bounce-path geometry is

shown on the right.

Since the velocity across the pipe cross-section

varies, the velocities associated with each path mustbe combined to provide an average or bulk velocitythat can be used to calculate the volumetric flowrate. To determine the average velocity, eachmanufacturer uses a proprietary method forweighting the effect of the individual paths. For thechordal-path meter, weighting factors are based ona numerical integration method that specifies thepath locations. In this case, the integration method


7/12


is independent of the velocity profile. For thebounce-path, the integration method is likelydependent on the velocity profile. Some weightingfactors may be functions of Reynolds number andothers may combine the individual pathmeasurements. The combination of the individualpath measurements may be based on recognition offlow characteristics determined by the individual

acoustic path measurements. In general, theaverage volumetric flow rate is calculated using thefollowing equation (the summation determines theaverage velocity):

22

1 1 24 2

ni i

v ii i i i

L tq D W

X t t

=

= Equation 2

where:

qv is the volumetric flow rate, in ft

3/s,

Li

is the path length of path i,

Xi is the axial length of path i,t

iis the differential transit time of path i,

ti1

is the transit time for downstream

transducer on path i,ti2 is the transit time for upstream

transducer on path i,W

i is the weighting factor on path i,

D is the meter internal bore diameter.


Research conducted by SwRI has shown that theshape of the velocity profile can introduce bias in

ultrasonic flow measurement. Flow conditionersgenerally improve meter performance but canadversely impact flow measurement accuracy aswell. Figure 13 shows the performance of a bounce-path multi-path ultrasonic meter located downstreamof two elbows out-of-plane. The graph shows thatthe bias introduced by the two elbows isapproximately 1 percent relative to the undisturbedbaseline.

Figure 14 shows relative meter performance whenflow conditioners are used. In these natural gas flowtests, the piping configuration upstream of each flowconditioner produced a fully-developed, symmetric,

swirl-free turbulent velocity profile. The test resultssuggest that each flow conditioner produces its owncharacteristic velocity profile downstream and maynot completely isolate the meter from distortions inthe upstream velocity profile. Therefore, it isgenerally recommended that ultrasonic flow metersbe flow calibrated with the meter tube and flowconditioner combined. This will ensure optimal

calibration accuracy and transferability from the flowlaboratory to the field.

0 5 10 15 20 25 30 35Velocity (ft/sec)

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

PercentError

100D Base

Meter at 20D

Figure 13. Example of the change in multi-path(bounce-path) meter performance associated

with two elbows out-of-plane with a bare meter

tube.

0 5 10 15 20 25 30 35Velocity (ft/sec)

0.0

0.2

0.4

0.6

0.8

1.0

PercentError

Bare Tube

19-tube BundleVORTAB

TMNOVA 50E

GFCTM

Meter Located Downstream of 100 D ofStraight Pipe

Figure 14. Example of the change in meterperformance associated with installing flow

conditioners. Note that each flow conditionerimpacts the meter differently.

(For best performance, the meter should becalibrated with the meter tube and the flow

conditioner.)


The gas industry standard for ultrasonic flowmeasurement is American Gas Association ReportNo. 9, Measurement of Gas by Multipath UltrasonicMeters. AGA Report No. 9 is a performance-basedrecommended practice, meaning that theperformance of the meter and meter installationmust fall within certain accuracy specifications,regardless of the meter path geometry and the


8/12


upstream piping. A performance-based standard orguideline is necessary when meters that have thesame basic operating principle are designed withdifferent physical configurations, such as thechordal-path and the bounce-path ultrasonic metergeometries. This is different from AGA Report No.3, the orifice meter standard, which specificallydefines the acceptable upstream piping

configurations.

The standard discusses protrusions into the flow,surface roughness, vibrations, thermal wells, velocityprofile considerations and the use of flowconditioners. The standard provides guidelines formeter testing under non-flowing as well as flowingconditions. The next revision of Report No. 9 mostlikely will require flow calibration for custody-transfermeters and should also provide general metercalibration requirements.

TURBINE METER INSTALLATION EFFECTS

The turbine meter is another common flow meter. Itis one of the most accurate and repeatable flowmeters available for custody transfer. Turbinemeters have fairly high flow rate capacity, good long-term and short-term repeatability, and are usuallylinear above the first 10 to 20% of the flow raterange.

Figure 15 shows the components of a typical turbineflow meter. The meter consists of a body, a rotorsupported by bearings, an inlet nose cone, and anelectronic or mechanical readout. Turbine metersmeasure flow rate by counting the number ofrevolutions of a rotor that spins as the gas flows

across its blades. The annular passage created bythe nose cone accelerates the flow, providingincreased torque to drive the rotor and improvemeter performance at lower flow rates

12. The meter

may include an electronic sensor to divide arevolution of the rotor into pulses. The readoutcounts the pulses and converts them into theindicated volume flow rate through a derived factorknown as the Reference Pulse Factor or K-factor.

Figure 15. A typical turbine meterconfiguration

13.

The reference pulse factor and indicated volumeflow rate equations for a turbine meter are:

( )

( )3N pulsesn

Reference Pulse Factor = K = =2 Q Reference Volume ft

Eq. 3

( )

( )3N Pulses

Indicated Volume Flow RateK Pulses/ft

= Eq. 4

where:

K is the meter K-factor, determined bycalibration, in pulses/ft

3,

n is the number of pulses per rotorrevolution,

is the rotational velocity of the rotor inradians/s,

Q is the reference flow measurement fromthe calibration facility, accumulatedduring the calibration, in ft

3,

N is the number of meter output pulsesaccumulated during the calibration.

The K-factor is determined by calibrating the meterunder flowing conditions. Most turbine metermanufacturers perform their own calibrations,typically with air as the test medium. Other flowlaboratories can also provide flow calibrationservices, usually with fluids such as natural gas, andat conditions more closely resembling actualoperating conditions. The flow laboratory selectedto perform the calibration should be able todemonstrate traceability of their referencemeasurements to a national standard, minimizingthe chance of introducing bias errors into the meter

K-factor through an inaccurate referencemeasurement. During a flowing calibration,reference flow rate data and accumulated meterpulses are collected and used to derive the meter K-factor in pulses per cubic foot.

The indicated volumetric flow rate is determined bydividing the total accumulated pulses by the K-factor. At a given flow rate, if the velocity profilediffers from the velocity profile that existed duringcalibration, the number of pulses accumulated whilein service may differ from the number of pulsesaccumulated during calibration, introducing a biaserror into the indicated volumetric flow rate. This

can occur if a distorted velocity profile exists at themeter inlet.


Experiments have shown that turbine meters can besensitive to velocity profile distortions, particularlyhighly swirling flows, such as those produceddownstream of two elbows oriented out-of-plane. Ingeneral, flow swirling in the direction of rotor rotation


9/12


tends to impart momentum to the rotor, causing apositive measurement bias, while flow swirling in thedirection opposing rotor rotation causes a negativebias

14,15. The effect of the upstream piping

installation on the magnitude of meter bias rangesfrom 0.35% to over 5.5% and appears to depend onmeter design (e.g., annular passage geometry, rotordesign, integral flow conditioner design, etc.).

Meters with integral flow conditioners tend to be lesssensitive to flow distortions resulting from theupstream piping configuration. In general, flowconditioners improve meter performance.

Figure 16 shows the effect of two elbows orientedout-of-plane, producing swirling flow, on a turbinemeter K-factor and the effect of a flow conditioner onthe K-factor as a function of swirl angle. In thiscase, the shift in K-factor increases as the meter islocated closer to the disturbance and as thedisturbance becomes more severe. The effect of aflow conditioner limits the K-factor shift to amaximum of approximately 1.5% of reading at a

swirl angle of 30. Without a flow conditionerinstalled upstream, the meter produces ameasurement bias error of over 5.5%.

Additional research suggests that turbine meters aresensitive to distorted velocity profiles produced byasymmetric or jetting flows. Under asymmetric or jetting flow conditions, such as is produced by apartially-closed gate valve or a pipe obstruction, thehighest velocity in the pipe cross-section tends todominate, resulting in a positive measurementbias

16.

In 2003, SwRI conducted installation effects tests on

five, commercially-available turbine meters insupport of the revision of AGA Report No. 7, GasFlow Measurement by Turbine Meters. Three pipinginstallation configurations prescribed in the standardwere tested under symmetric, swirl-free, fully-developed turbulent flow conditions as well asseverely distorted flow conditions. The pipingconfigurations included the AGA-7 recommendedmeter installation configuration, as well as theclose-coupled and short-coupled configurations.The piping upstream of the meter ranged from ahigh-performance flow conditioner followed by 30diameters of straight pipe (which provided a

symmetric, swirl-free, fully-developed turbulentvelocity profile17

) to the high-perturbationconfiguration described in ISO Standard 9951,Measurement of Gas Flow in Closed Conduits Turbine Meters (which produced an asymmetric,clockwise swirling flow leading into the clockwise-rotating turbine rotor). The research concluded thatthe total installation effect on the AGA-7 installationconfigurations, including the effect of the high-perturbation condition, was less than +/-1% of

reading. Meters with integral flow conditioning wereshown to limit the installation effect to approximately+/-0.25%

18.

Figure 16. The effect of two elbows oriented out-of-plane (top), on a turbine meter K-factor andthe effect of a flow conditioner on the K-factor

(bottom) as a function of flow swirl angle19

.


The gas industry recommended practice for turbineflow measurement is American Gas AssociationReport No. 7, Measurement of Gas by TurbineMeters. Although this report is a recommendedpractice, it essentially serves as a defacto standardfor the U.S. natural gas industry. The most recentedition was published in 1996. A revision of thereport is expected to be published sometime in2006. The document provides recommendedinstallations, such as those tested during the SwRI

research discussed earlier. It also providesguidance for installing flow conditioners, strainers,filters, and secondary instrumentation (i.e., pressureand temperature measurement devices). Inaddition, the report provides guidance on protectingturbine flow meters from being over-ranged. Itdescribes the potential adverse effect of velocityprofile distortion on meter accuracy and informationon proper meter calibration.


10/12


INSTALLATION EFFECTS ASSOCIATED WITHSECONDARY INSTRUMENTATION

Installation effects impacting flow measurementaccuracy are not limited to flow meters. Theaccuracy of secondary instrumentation, such asstatic pressure and gas temperature measurementdevices (i.e., values used in the calculation of gascompressibility or density) and differential pressuremeasurement devices (i.e., a value used in theorifice flow equation), can be adversely affected bytheir installations. These installation effects,however, are not the result of velocity profiledisturbances. In some cases, such as whenmeasuring static or differential pressure, theinstallation effect is the result of the fabricationprocess. In other cases, such as measurement ofthe gas temperature, the installation effect is theresult of the measurement method (e.g., the use ofthermal wells) and the operating conditions, (i.e., theflowing gas and ambient temperatures).

Temperature Measurement

In most gas flow applications, the temperaturemeasurement device, usually a resistancetemperature device (RTD) is inserted in a housingcalled a thermal well. This type of installation canadversely affect the accuracy of the temperaturemeasurement. Recent computational fluid dynamicsimulations

20have shown that the temperature of

the thermal well can be affected by the pipe walltemperature, when it deviates from the flowing gastemperature and when flow rates are relatively low(i.e., less than about 2 ft/s nominal or bulk gasvelocity). Since the RTD is in direct contact with thethermal well, the measured temperature can beaffected. Figure 17 shows the effect of pipe walltemperature and gas velocity on the temperature ofthe thermal well immersed in a flowing natural gasstream. Note that the pipe wall temperature and theflowing gas temperature are equal in both cases.

In addition to the effect of pipe wall temperature onthe temperature of the thermal well, there is potentialfor thermal stratification when the pipe wall orambient temperature differs from the flowing gastemperature. Figure 18 shows the temperatureprofile of a low velocity (0.5 to 1.0 ft/s) natural gas

flowing in a 12-inch diameter pipe when the ambienttemperature is 18F below the nominal flowing gastemperature. The temperature profile shows thatdifferences from the nominal flowing gastemperature ranging from 7F to less than 1F occuralong the cross-section of the pipe. Under theseconditions, a temperature error could be introducedinto the flow measurement, even if the thermal wellis unaffected by the pipe wall temperature.

Figure 17. The effect of pipe wall temperatureand velocity on thermal well temperature whenthe pipe wall temperature is below the flowinggas temperature. The tip of the thermal well isapproximately 5F (3K) cooler than the gas at

0.5 ft/s. The tip of the thermal well is about equalto the flowing gas temperature at 2.0 ft/s21

.

Temperature

50 55 60 65 70 75Elevationab

ovebottomo

fpipe(inches)

0123456789

10

11

12

Temperature Traverse

Nomin

alAmbientTemperature

NominalF

lowingGasTemperature

Test date: 12/16/2004

Temperature profile in a 12-Inch Pipe Flowing NaturalGas at Low Velocities (0.5-1.0 ft/s)

Figure 18. Temperature stratification in a 12-inchpipe flowing low velocity natural gas when theambient temperature is lower than the flowing

gas temperature.


11/12


Static and Differential Pressure Measurement

The static pressure is used in the compressibilitycalculation and to account for expansion of themeter body due to circumferential strain. Thedifferential pressure is used to measure the pressuredrop across a meter such as an orifice meter. It isusually assumed that pressure taps drilled throughholes that are perfectly normal to the pipe wall willprovide an accurate measurement of static pressureand that any deviations from this ideal can beadjusted through calibration

22. However, when a

meter calibration is not required, such as with anorifice meter, these errors become part of the overallmeter error. Figure 19 shows estimated effects ofpressure tap geometry on the static pressuremeasurement, relative to a reference condition

23. In

general, rounded pressure taps tend to cause apositive bias error, and countersunk pressure tapstend to cause a negative bias.

Figure 19. Estimated errors in measuredpressure associated with different pressure tap

geometries. The left figure is the ideal; themiddle figure shows the effect of rounded

corners, relative to the ideal. The right figure

shows the effect of countersinking, relative tothe ideal24

.

CONCLUSIONS

The configuration of the piping upstream of a flowmeter can cause distortions in the velocity profilethat may result in bias errors in the flow ratemeasurement. The velocity profile can be affectedby the upstream piping configuration through theestablishment of axial rotation or by theredistribution of momentum along the pipe cross-section. Swirling flow or velocity profile asymmetryor a combination of the two can result. Typical

piping configurations that cause these disturbancesinclude single elbows, two elbows in an in-planeconfiguration, two elbows oriented out-of-plane,partially closed valves, and other obstructions in thepipe flow. Flow distortions can persist for as muchas 200 diameters. The extended persistence ofsome types of flow disturbances suggests thatpotential sources of velocity profile distortion furtherupstream than the immediate meter run should be

investigated if piping installation effects are apotential concern.

The effect of velocity profile disturbances on flowmeasurement accuracy may be different for eachmeter type. In the case of orifice meters, the impactof velocity profile distortions is due to the fact thatthe R-G equation used to determine the dischargecoefficient value implicitly assumes that the velocityprofile is fully-developed, symmetric, swirl-free, andturbulent. In the case of ultrasonic meters, thenumber of paths and the method of integrating thevelocity profile impacts the ability of the meter toresolve and recognize distorted velocity profiles. Inthe case of turbine meters, velocity profile distortionstend to either increase or decrease rotor momentum,resulting in over- or under-measurement.Installation effects on ultrasonic and turbine meterstend to depend on the specific meter design.

Flow conditioners can be used to mitigate the effectsof upstream flow disturbances, allowing for the use

of shorter lengths of straight upstream pipe, but flowconditioners do not truly isolate the flow meter fromthe upstream flow field. While the use of a flowconditioner is recommended with all meter types, themeter station designer or operator should be awarethat some flow conditioners may freeze velocityprofile asymmetry, allowing it to continue ondownstream, while others may introduce bias errorsinto the flow measurement due to the characteristicsof the pseudo fully-developed velocity profileproduced by the flow conditioner.

While the piping installation configuration upstreamof a flow meter may adversely affect meter accuracy

through the creation of flow distortions, installationeffects may also adversely affect the accuracy ofsecondary element measurements, such as static ordifferential pressure measurement devices andflowing temperature measurement devices. Theaccuracy of these secondary measurement devicesaffects the accuracy of the total energy flow ratecalculation because of their use in the gascompressibility (i.e., gas density) calculation and theflow rate calculation.

REFERENCES

1. Roney, Jason, MAE 3130 Course Lecture 10 Spring 2003, Mechanical and AerospaceEngineering, University of Colorado at ColoradoSprings.

2. Steenbergen, W. Turbulent Pipe Flow with Swirl,Thesis, Eindhoven University of Technology.

3. Grimley, T. A., Performance Testing of 8-inchUltrasonic Flow Meters for Natural GasMeasurement, Topical Report to Gas Research


12/12


Institute, GRI Contract No. 5097-270-3937,November 2000.

4. Bowles, E. George, D. L., Effects of FlowConditioning on Gas Measurement, ISHM ClassNo. 8050.

5. Ibid.

6. Miller, R.W., Flow Measurement EngineeringHandbook, 3 ed, McGraw-Hill.

7. AGA Transmission Measurement CommitteeReport No. 3, Concentric, Square-Edge OrificeMeters, AGA, Arlington, VA, June 2000.

8. AGA Transmission Measurement CommitteeReport No. 3, Concentric, Square-Edge OrificeMeters, AGA, Arlington, VA, April 2002.


10 . AGA Transmission Measurement CommitteeReport No. 3, Concentric, Square-Edge Orifice

Meters, AGA, Arlington, VA, April 2002.


12. Miller, R. W.

13. AGA Transmission Measurement CommitteeReport No. 7, Measurement of Gas by TurbineMeters, AGA, Arlington, VA, April, 1996

14. Brennan, J.A., McFaddin, S.E., Sindt, C.F.,Kothari, K.M., The Influence of Swirling Flow onOrifice and Turbine Flowmeter Performance,Journal of Flow Measurement and

Instrumentation, Volume 1, October, 1989.15. Van Dellen, K. and Vanderkam, P.M.A., The

Effect of Double Bends Out of Plane on TurbineMeters, N.V. Nederlandse Gasunie.

16. McKee, Robert, Principal Engineer, SouthwestResearch Institute, Personal Communication.

17. George, D. L., Metering Research FacilityProgram: Turbine Meter Research in Support ofthe Planned AGA Report No. 7 Revision, TopicalReport to Gas Research Institute, GRI ContractNo. 8346, January 2003.

18. Ibid.

19. Miller, R. W.20. Morrison, Gerald R. and Brar, Pardeep, CFD

Evaluation of Pipeline Gas Stratification at LowFlow Due to Temperature Effects. Final Reportto Gas Research Institute, GRI Contract No.8733, September 2004.

21. Ibid.

22. Mattingly, G.E., Volume Flow Measurements, inFluid Mechanics Measurements, Edited byGoldstein, R.J., Hemisphere PublishingCorporation, 1983

23. Ibid.

24. Ibid.

kelner installation effects article

Documents