Flow and Level Measurement Handbook

04 Volume 4 TRANSACTIONS

• The Flow Pioneers

• Flow Sensor Selection

• Accuracy vs. Repeatability

Figure 1-3: Faraday's Law is the Basis of the Magnetic Flowmeter

Turbulent Velocity

Flow Profile

or

E

E

D

V

Laminar Velocity

Flow Profile

Magnetic Coil

Figure 2-8: Proprietary Elements For Difficult Fluids

A) Segmental Wedge

Wedge Flow Element

DH

B) V-Cone

H L

08

TABLE OF CONTENTS

VOLUME 4—FLOW & LEVEL MEASUREMENTSection Topics Covered Page

• Primary Element Options

• Pitot Tubes

• Variable Area Flowmeters

16

• Positive Displacement Flowmeters

• Turbine Flowmeters

• Other Rotary Flowmeters

34

• Magnetic Flowmeters

• Vortex Flowmeters

• Ultrasonic Flowmeters

46

• Coriolis Mass Flowmeters

• Thermal Mass Flowmeters

• Hot-Wire Anemometers

58

Electronic Flowmeters4

Mechanical Flowmeters3

Differential Pressure Flowmeters2

A Flow Measurement Orientation1

Mass Flowmeters5

Figure 3-7:

Calibrated Volume

1st Detector 2nd DetectorFlow Tube

Flow

Disp

lace

r

Figure 4-6:

1 10 100 1,000 104 105 106 107

1.00

0.95

0.90

0.85

0.80

0.75

0.70

Re

K

K = 1 Asymptote For Flat Profile

K = 0.75 For Laminar Flow

Figure 5-5:

B)A)

C)

SupportFlanges

Mass Flowtube Enclosure

Pipe/Flowtube Junction

NOTE: Distance Between

Pipe/Flowtube Junction and

Support Must Not

Exceed 15 Inches

Flow Direction Arrow

Mass Tube Enclosure

Support (Typical)

Flow Direction

Arrow

NOTE: Distance Between Pipe/Flowtube Junction and

Support Must Not Exceed 15 Inches

'U' Rest 'V' Rest 'V' Bolt Clamp

Inverted Pipe Hanger Clamp

'V' Block Clamp (Can Be Inverted)

http://www.omega.com/literature/transactions/

WebMaster

Text Box

WebMaster

Text Box

Click here to visit the OMEGA.COM website

http://www.omega.com/

TRANSACTIONS Volume 4 05

Editorial 06

About OMEGA 07

REFERENCE SECTIONS

106 Information Resources

110 Glossary

• Level Sensor Selection

• Boiling & Cryogenic Fluids

• Sludge, Foam, & Molten Metals

Figure 6-3:

VerticalSphere

Horizontal Cylindrical

50

0 100 Volume %

100

50

Level %

Figure 7-3:

B)A)

Bimetallic Temperature Compensator

Low Pressure Side

High Pressure Side

Liquid Fill

Range Spring

Nozzle & Flapper

Feedback Bellows

Fulcrum & Seal

Force Bar

Low Pressure Side

Liquid Filled Diaphragm

Capsule

Output

High Pressure Side

Pneumatic Relay

Air Supply

72

VOLUME 4—FLOW & LEVEL MEASUREMENTSection Topics Covered Page

• Dry & Wet Leg Designs

• Bubbler Tubes

• Floats & Displacers

76

• Theory of Operation

• Probe Designs

• Installation Considerations

87

• Radar & Microwave

• Ultrasonic Level Gages

• Nuclear Level Gages

93

• Thermal Switches

• Vibrating Switches

• Optical Switches

102

Radiation-Based Level Instrumentation9

RF/Capacitance Level Instrumentation8

Pressure/Density Level Instrumentation7

A Level Measurement Orientation6

Specialty Level Switches10

Figure 8-2:

A) B)

- -- -- -- -

- -- -- -- -

- -- -

- -- -- --

+ ++ ++ ++ ++ ++ ++ +

+ ++ ++ ++ ++ ++ ++ +

- -- -- -

- -

A

A

D

Electron Flow

Ammeter

Voltmeter

#1Level

RF

#2Kv

Kl

C= KAD

C=Capacitance K=Dieletric Constant A=Area of Plates D=Dist. Between Plates

Figure 9-6:

B)A)

Reflection Microwave Detector

Microwave Window

Microwave Window

Microwave Transmitter

Transmitted Beam

Microwave Receiver

Microwave Window

Reflected Beam

Absorbed Beam

Figure 10-4:

ReceiverLED

Prism

Light from LED

Liquid Below the Sensing Prism.

Liquid Immersing the Sensing Prism.

LEDLEDReceiver

PrismLight

Lost in Liquid

Our interest in the measure-ment of air and water flowis timeless. Knowledge ofthe direction and velocity

of air flow was essential informa-tion for all ancient navigators, andthe ability to measure water flowwas necessary for the fair distribu-tion of water through the aque-ducts of such early communities as

the Sumerian cities of Ur, Kish, andMari near the Tigris and EuphratesRivers around 5,000 B.C. Even today,the distribution of water among therice patties of Bali is the sacredduty of authorities designated the“Water Priests.”

Our understanding of the behaviorof liquids and gases (including hydro-

dynamics, pneumatics, aerodynam-ics) is based on the works of theancient Greek scientists Aristotleand Archimedes. In the Aristotelianview, motion involves a medium thatrushes in behind a body to prevent avacuum. In the sixth century A.D., JohnPhiloponos suggested that a body inmotion acquired a property calledimpetus, and that the body came to

rest when its impetus died out.In 1687, the English mathematician

Sir Isaac Newton discovered the lawof universal gravitation. The opera-tion of angular momentum-typemass flowmeters is based directly onNewton’s second law of angularmotion. In 1742, the French mathe-matician Rond d’Alembert proved

that Newton’s third law of motionapplies not only to stationary bodies,but also to objects in motion.

The Flow PioneersA major milestone in the understand-ing of flow was reached in 1783 whenthe Swiss physicist Daniel Bernoullipublished his Hydrodynamica. In it, heintroduced the concept of the con-servation of energy for fluid flows.Bernoulli determined that anincrease in the velocity of a flowingfluid increases its kinetic energywhile decreasing its static energy. It isfor this reason that a flow restrictioncauses an increase in the flowingvelocity and also causes a drop in thestatic pressure of the flowing fluid.

The permanent pressure lossthrough a flowmeter is expressedeither as a percentage of the totalpressure drop or in units of velocityheads, calculated as V2/2g, where Vis the flowing velocity and g is thegravitational acceleration (32.2feet/second2 or 9.8 meters/second2

at 60° latitude). For example, if thevelocity of a flowing fluid is 10 ft/s,the velocity head is 100/64.4 = 1.55 ft.If the fluid is water, the velocity headcorresponds to 1.55 ft of water (or0.67 psi). If the fluid is air, then thevelocity head corresponds to theweight of a 1.55-ft column of air.

The permanent pressure lossthrough various flow elements canbe expressed as a percentage of thetotal pressure drop (Figure 1-1), or itcan be expressed in terms of veloc-ity heads. The permanent pressureloss through an orifice is four veloc-ity heads; through a vortex sheddingsensor, it is two; through positive


The Flow Pioneers

Flow Sensor Selection

Accuracy vs. Repeatability

FLOW & LEVEL MEASUREMENTA Flow Measurement Orientation

1

A Flow Measurement Orientation

Figure 1-1: Pressure Loss-Venturi vs. Orifice

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

90

80

70

60

50

40

30

20

10 Low Loss Venturi

Long Form Venturi

Standard Venturi

ASME Flow Nozzle

Orifice PlateRe

cove

ry—

Perc

ent

of D

iffer

enti

al

Unr

ecov

ered

Pre

ssur

e Lo

ss—

Perc

ent

of D

iffer

enti

al

Proprietary Flow Tube

Beta (Diameter) Ratio

10

20

30

40

50

60

70

80

90

O

displacement and turbine meters,about one; and, through flow venturis,less than 0.5 heads. Therefore, if an ori-fice plate (Figure 1-2) with a beta ratio

of 0.3 (diameter of the orifice to thatof the pipe) has an unrecoveredpressure loss of 100 in H2O, a venturiflow tube could reduce that pres-sure loss to about 12 in H2O for thesame measurement.

In 1831, the English scientist MichaelFaraday discovered the dynamo whenhe noted that, if a copper disk is rotat-ed between the poles of a permanentmagnet, electric current is generated.Faraday’s law of electromagneticinduction is the basis for the operationof the magnetic flowmeter. As shownin Figure 1-3, when a liquid conductormoves in a pipe having a diameter (D)and travels with an average velocity (V)through a magnetic field of B intensity,it will induce a voltage (E) according tothe relationship:

E = BVDC

where C is the constant for unitsconversion.

Over the past several years, theperformance of magnetic flowmeters

has improved significantly. Among theadvances are probe and ceramic insertdesigns and the use of pulsed mag-netic fields (Figure 1-4), but the basicoperating principle of Faraday’s law ofelectric induction has not changed.

In 1883, the British mechanical engi-neer Osborne Reynolds proposed a

single, dimensionless ratio to describethe velocity profile of flowing fluids:

Re = DVρ/µ

Where D is the pipe diameter, V isthe fluid velocity, ρ is the fluid den-sity, and µ is the fluid viscosity.

He noted that, at low Reynoldsnumbers (below 2,000) (Figure 1-5),flow is dominated by viscous forcesand the velocity profile is (elongated)parabolic. At high Reynolds numbers(above 20,000), the flow is dominatedby inertial forces, resulting in a moreuniform axial velocity across the flow-ing stream and a flat velocity profile.

Until 1970 or so, it was believedthat the transition between laminarand turbulent flows is gradual, butincreased understanding of turbu-lence through supercomputer mod-eling has shown that the onset ofturbulence is abrupt.

When flow is turbulent, the pres-sure drop through a restriction isproportional to the square of theflowrate. Therefore, flow can bemeasured by taking the square rootof a differential pressure cell output.When the flow is laminar, a linearrelationship exists between flow andpressure drop. Laminar flowmeters

1 A Flow Measurement Orientation


Figure 1-2: Conversion of Static Pressure Into Kinetic Energy

Flow

Flow

Unstable Region, No Pressure Tap Can Be Located Here

Static Pressure

(0.35-0.85)D Pressure at Vena Contracta (PVC)

Minimum Diameter

∆PCT∆PFT

∆PPT

∆PRT=∆PVC

Orifice

Flange Taps (FT), D › 2"Radius Taps (RT), D › 6"

Corner Taps (CT), D ‹ 2"D/22.5D 8D

D

D

Pipe Taps (PT)

Figure 1-3: Faraday's Law Is the Basis of the Magnetic Flowmeter

Turbulent Velocity

Flow Profile

or

E

E

D

V

Laminar Velocity

Flow Profile

Magnetic Coil

are used at very low flowrates (capil-lary flowmeters) or when the viscos-ity of the process fluid is high.

In the case of some flowmetertechnologies, more than a centuryelapsed between the discovery of a

scientific principle and its use inbuilding a flowmeter. This is the casewith both the Doppler ultrasonic andthe Coriolis meter.

In 1842, the Austrian physicistChristian Doppler discovered that, if asound source is approaching a receiver(such as a train moving toward a sta-tionary listener), the frequency of thesound will appear higher. If the sourceand the recipient are moving awayfrom each other, the pitch will drop(the wavelength of the sound willappear to decrease). Yet it was morethan a century later that the first ultra-sonic Doppler flowmeter came on themarket. It projected a 0.5-MHz beaminto a flowing stream containing reflec-tors such as bubbles or particles. Theshift in the reflected frequency was afunction of the average traveling veloc-ity of the reflectors. This speed, in turn,could be used to calculate a flowrate.

The history of the Coriolis

flowmeter is similar. The French civilengineer Gaspard Coriolis discoveredin 1843 that the wind, the ocean cur-rents, and even airborne artilleryshells will all drift sideways becauseof the earth’s rotation. In the northernhemisphere, the deflection is to theright of the motion; in the southernhemisphere, it is to the left. Similarly,a body traveling toward either polewill veer eastward, because it retainsthe greater eastward rotational speedof the lower altitudes as it passesover the more slowly rotating earthsurface near the poles. Again, it wasthe slow evolution of sensors andelectronics that delayed creation ofthe first commercial Coriolis massflowmeter until the 1970’s.

It was the Hungarian-Americanaeronautical engineer Theodorevon Karman who, as a child growingup in Transylvania (now Romania),noticed that stationary rocks causedvortices in flowing water, and thatthe distances between these travel-ing vortices are constant, no matterhow fast or slow the water runs.Later in life, he also observed that,when a flag flutters in the wind, thewavelength of the flutter is indepen-dent of wind velocity and depends

solely on the diameter of the flagpole. This is the theory behind the

vortex flowmeter, which determinesflow velocity by counting the num-ber of vortices passing a sensor. VonKarman published his findings in1954, and because by that time thesensors and electronics required tocount vortices were already in exis-tence, the first edition of theInstrument Engineers’ Handbook in1968 was able to report the availabil-ity of the first swirlmeter.

The computer has opened newfrontiers in all fields of engineering,and flow measurement is no excep-tion. It was only as long ago as 1954that another Hungarian-Americanmathematician, John Von Neumann,built Uniac—and even more recentlythat yet another Hungarian-American, Andy Grove of Intel,developed the integrated circuit. Yetthese events are already changingthe field of flowmetering. Intelligentdifferential pressure cells, for exam-ple, can automatically switch theirrange between two calibrated spans(one for 1-10%, the other for 10-100%of D/P), extending orifice accuracyto within 1% over a 10:1 flow range.Furthermore, it is possible to includein this accuracy statement not onlyhysteresis, rangeability, and linearity

effects, but also drift, temperature,humidity, vibration, over-range, and

A Flow Measurement Orientation 1


Figure 1-4: Magmeter Accuracy

Conventional Magnetic Flowmeters

Performance of Pulsed DC Magnetic Flowmeters

4.0

10 50 100

% R

ate

Acc

urac

y

% Full Scale

2.0

1.00.5

0-0.5

-2.0

-1.0

-3.0

-4.0

3.0

Flow measurement options run the gamut from simple, economical paddle wheels (shown) to

sophisticated high-accuracy devices.

power supply variation effects. With the development of super-

chips, the design of the universalflowmeter also has become feasible.It is now possible to replace dye-tagging or chemical-tracing meters(which measured flow velocity bydividing the distance between twopoints by the transit time of thetrace), with traceless cross-correla-tion flowmeters (Figure 1-6). This isan elegant flowmeter because itrequires no physical change in theprocess—not even penetration ofthe pipe. The measurement is basedon memorizing the noise pattern inany externally detectable processvariable, and, as the fluid travelsfrom point A to point B, noting itstransit time.

Flow Sensor SelectionThe purpose of this section is toprovide information to assist thereader in making an informed selec-tion of flowmeter for a particularapplication. Selection and orienta-tion tables are used to quickly focuson the most likely candidates formeasurement. Tables 1-I and 1-IIhave been prepared to make avail-able a large amount of informationfor this selection process.

At this point, one should considersuch intangible factors as familiarity ofplant personnel, their experience withcalibration and maintenance, spareparts availability, mean time betweenfailure history, etc., at the particularplant site. It is also recommended thatthe cost of the installation be comput-ed only after taking these steps. Oneof the most common flow measure-ment mistakes is the reversal of thissequence: instead of selecting a sensorwhich will perform properly, anattempt is made to justify the use of adevice because it is less expensive.

Those “inexpensive” purchases can bethe most costly installations.

The basis of good flowmeterselection is a clear understanding ofthe requirements of the particularapplication. Therefore, time shouldbe invested in fully evaluating thenature of the process fluid and of theoverall installation. The developmentof specifications that state the appli-

cation requirements should be a sys-tematic, step-by-step process.

The first step in the flow sensorselection process is to determine ifthe flowrate information should becontinuous or totalized, and whetherthis information is needed locally orremotely. If remotely, should thetransmission be analog, digital, orshared? And, if shared, what is therequired (minimum) data-update fre-quency? Once these questions areanswered, an evaluation of the prop-erties and flow characteristics of theprocess fluid, and of the piping thatwill accommodate the flowmeter,should take place (Table 1-I). In orderto approach this task in a systematicmanner, forms have been developed,requiring that the following types of

data be filled in for each application:• Fluid and flow characteristics: In

this section of the table, the nameof the fluid is given and its pressure,temperature, allowable pressuredrop, density (or specific gravity),conductivity, viscosity (Newtonianor not?) and vapor pressure atmaximum operating temperatureare listed, together with an indica-

tion of how these propertiesmight vary or interact. In addition,all safety or toxicity informationshould be provided, together withdetailed data on the fluid’s compo-sition, presence of bubbles, solids(abrasive or soft, size of particles,fibers), tendency to coat, and lighttransmission qualities (opaque,translucent or transparent?).

• Expected minimum and maximumpressure and temperature valuesshould be given in addition to thenormal operating values. Whetherflow can reverse, whether it doesnot always fill the pipe, whetherslug flow can develop (air-solids-liq-uid), whether aeration or pulsationis likely, whether sudden tempera-ture changes can occur, or whether



Figure 1-5: Effect of Reynolds Numbers on Various Flowmeters

10 102 103 104 105 106

Concentric Square-Edged

Orifice

Eccentric Orifice

Magnetic Flowmeter

Venturi TubeFlow Nozzle

Integral Orifice

Pipeline Reynolds Number

Coefficient of Discharge

Target Meter (Best Case)

Target Meter (Worst Case) Quadrant-Edged

Orifice

special precautions are needed dur-ing cleaning and maintenance, thesefacts, too, should be stated.

• Concerning the piping and the areawhere the flowmeter is to be locat-ed, the following information

should be specified: For the piping,its direction (avoid downward flowin liquid applications), size, material,schedule, flange-pressure rating,accessibility, up or downstreamturns, valves, regulators, and avail-able straight-pipe run lengths.

• In connection with the area, thespecifying engineer must know ifvibration or magnetic fields are pre-sent or possible, if electric or pneu-matic power is available, if the areais classified for explosion hazards,or if there are other specialrequirements such as compliance

with sanitary or clean-in-place(CIP) regulations.The next step is to determine the

required meter range by identifyingminimum and maximum flows (massor volumetric) that will be measured.

After that, the required flow mea-surement accuracy is determined.Typically accuracy is specified in per-centage of actual reading (AR), inpercentage of calibrated span (CS), orin percentage of full scale (FS) units.The accuracy requirements should beseparately stated at minimum, nor-mal, and maximum flowrates. Unlessyou know these requirements, yourmeter’s performance may not beacceptable over its full range.

Accuracy vs. RepeatabilityIn applications where products are

sold or purchased on the basis of ameter reading, absolute accuracy iscritical. In other applications,repeatability may be more importantthan absolute accuracy. Therefore, itis advisable to establish separatelythe accuracy and repeatabilityrequirements of each application andto state both in the specifications.

When a flowmeter’s accuracy isstated in % CS or % FS units, itsabsolute error will rise as the mea-sured flow rate drops. If meter error isstated in % AR, the error in absoluteterms stays the same at high or lowflows. Because full scale (FS) is alwaysa larger quantity than the calibratedspan (CS), a sensor with a % FS perfor-mance will always have a larger errorthan one with the same % CS specifi-cation. Therefore, in order to compareall bids fairly, it is advisable to convertall quoted error statements into thesame % AR units.

It is also recommended that theuser compare installations on thebasis of the total error of the loop. Forexample, the inaccuracy of an orificeplate is stated in % AR, while the errorof the associated d/p cell is in % CSor % FS. Similarly, the inaccuracy of aCoriolis meter is the sum of twoerrors, one given in % AR, the other asa % FS value. Total inaccuracy is calcu-lated by taking the root of the sum ofthe squares of the component inaccu-racies at the desired flow rates.

In well-prepared flowmeter specifi-cations, all accuracy statements areconverted into uniform % AR units andthese % AR requirements are specifiedseparately for minimum, normal, andmaximum flows. All flowmeter specifi-cations and bids should clearly stateboth the accuracy and the repeatabili-ty of the meter at minimum, normal,and maximum flows.

Table 1 provides data on the range



Figure 1-6: The Ultrasonic Transit-Time Flowmeter

Upstream Transducer Signal

Downstream Transducer Signal

Time. t

Time. t

Transit Time

B

Am(t)

m(t)

n(t)

n(t)

Transport Pipe

Flow

Time Delay

Position A

Position B

of Reynolds numbers (Re or RD) with-in which the various flowmeterdesigns can operate. In selecting theright flowmeter, one of the first stepsis to determine both the minimumand the maximum Reynolds numbersfor the application. Maximum RD isobtained by making the calculation

when flow and density are at theirmaximum and viscosity at its mini-mum. Conversely, the minimum RD isobtained by using minimum flow anddensity and maximum viscosity.

If acceptable metering performancecan be obtained from two differentflowmeter categories and one has

no moving parts, select the onewithout moving parts. Moving partsare a potential source of problems,not only for the obvious reasons ofwear, lubrication, and sensitivity tocoating, but also because movingparts require clearance spaces thatsometimes introduce “slippage” into



Orifice Square-Edged Honed Meter Run Integrated Segmental Wedge Eccentric Segmental V-Cone Target*** Venturi Flow Nozzle Low Loss Venturi Pitot Averaging Pitot Elbow Laminar

cP = centi Poise cS = centi Stokes SD = Some designs

? = Normally applicable (worth consideration) √ = Designed for this application (generally suitable)

URV = Upper Range Value X = Not applicable

‡ According to other sources, the minimum Reynolds number should be much higher

* Liquid must be electrically conductive ** Range 10:1 for laminar, and 15:1 for target *** Newer designs linearize the signal

Magnetic* Positive Displacement Gas Liquid Turbine Gas Liquid Ultrasonic Time of Flight Doppler Variable-Area (Rotameter) Vortex Shedding Vortex Precession (Swirl) Fluidic Oscillation (Coanda) Mass Coriolis Thermal Probe Solids Flowmeter Correlation Capacitance Ultrasonic

>1.5 (40) 0.5-1.5 (12-40) <0.5 (12) <12 (300) >2 (50) >4 (100) 0.5-72 (12-1800) <0.5(12) >2 (50) >2 (50) >3 (75) >3 (75) >1 (25) >2 (50) 0.25-16.6 (6-400) 0.1-72 (2.5-1800) <12 (300) <12 (300) 0.25-24 (6-600) 0.25-24 (6-600) >0.5 (12) >0.5 (12) ≤3 (75) 1.5-16 (40-400) <16 (400) >1.5 (40) 0.25-6 (6-150) <72 (1800) <24 (600) <8 (200) >0.5 (12)

RD > 10,000

RD > 10,000 RD > 10,000 RD > 500 RD > 10,000 RD > 10,000 RD : 8,000-5,000,000 RD > 100 RD > 75,000ŁRD > 50,000ŁRD > 12,800ŁRD > 100,000ŁRD > 40,000ŁRD > 10,000ŁRD < 500

700 (370)

150 (66)

≤600 (4,100)

≤30 (225) RD > 4,500

- No RD limit ≤ 8,000 cS - Rp > 5,000, ≤15 cS RD > 10,000 RD > 4,000 No RD limit, < 100 cS RD > 10,000, < 30 cP RD > 10,000, < 5 cP RD > 2,000, < 80 cS No RD limit No RD limit - No data available No data available

360 (180) 250 (120) 600 (315) -450-500 (268-260) -450-500 (268-260) -300-500 (-180-260) -300-500 (-180-260) 400 (200) 536 (280) 350 (175) -400-800 (-224-427) 1,500 (816) 750 (400) 300 (149) -300-250 (-180-120)

≤ 1,500 (10,800) ≤ 1,400 (10,000) ≤ 1,400 (10,000) ≤ 3,000 (21,000) ≤ 3,000 (21,000) Pipe rating Pipe rating ≤ 1,500 (10,500) Pipe rating ≤ 720 (5,000) ≤ 5,700 (39,900) Pipe rating ≤ 580 (4,000) ≤ 580 (4,000) Pipe rating

Proc

ess t

empe

ratu

re

to 10

00°F

(540

°C):

Tran

smitt

er li

mite

d to

-30-

250°

F (-3

0-12

0°C)

To 4

,000

psig

(4

1,000

kPa

)

Proc

ess t

empe

ratu

re

to 10

00°F

(540

°C):

Tran

smitt

er li

mite

d to

-30-

250°

F (-3

0-12

0°C)

To 4

,000

psig

(4

1,000

kPa

)

Glass: 400 (200) Metal: 1,000 (540)

Glass: 350 (2,400) Metal: 720 (5,000)

X

X X

SD X

X X ? √ √ X ? X X

X X

X

√ X

√ X

SD X √ √ √ X ? √ X

X X

X

X X

X X

SD X X ? ? X ? ? X

X X

X ? X

√ X

SD X X √ √ X

√ √ X

X X

X

√ X

√ X

SD X √ √ √ X

√ √ X

X X

√

X √

X √

√ X √ √ √ √

√ √ X

X X

?

X √

X X ? ? X X X X

√ ?

SD

√ ?

√

X ?

X ? ? ? ? ? ? X

√ √ X

√ √

√

X X

X X

X √ X ? X ?

√ √ ?

√ √

√

X ?

X ?

√ √ ? ? ? ? ? ? X

√ √

√

X X

X X

√ √ ? X X X ? ? X

√ √

√

X X

X X ? √ X X X X ? ?

SD

√ √

?

X X

X X

X X X X X X

X X √ ? X

?

X X

X X

X ? X X X X

√ ? X ? ?

?

X X ? ? ? X X X X X

X X X

X X

√

X X

X SD

? √ X X X X

√ ?

SD

√ √

√

X X

SD SD

√ √ X X X X ? X X

X X

√

X X

SD SD

√ √ ? X X ? ? ?

SD ? ?

?

X ? ? ?

X X ? ? ? ? ? ?

SD ? X

X

X X ? ? ? X ? ? X ? ? X X

X X

√ √ ? √ ? ? √ ? √ ? √ X √ X ?

√ √ √ √ ? ? √ √ √ √ √ √ √ √ √

X X X √ √ √ ? √ ? ? X X

SD ? X

√ √ √ √ √ √ √ √ √ √ √ √ √ √ √

√ √ √ √ √ √ √ √ √ √ √ √ √ √ √

√ √ √ √ ? ? √ √ √ √ √ √ √ √ √

X ? X ? X X ? ? ? X X X X X √

? ? ? √ ? ? √ √ √ ? ? ? ? ? ?

X X X ? ? ? ? √ ? ? X X

SD ? X

? ? ? ? ? ? ? ? ? ? √ ? ? ? ?

X X ? X X X X X X X X X X X X

X X X ? ? ? ? X √ X X X X X X

X X X ? X X ? X ? X X X X X X

SD SD SD SD SD SD X ? X X X X

SD √ X

? ? ? ? ? ? ? X ? ? ? X X X √

√ √ ? √ √ √ ? ? ? ? ? ? ? ? X

√ √ X √ √ √ ? ? ? ? ? ? ? ? X

X X X X X X X X X X X X X X X

? ? ? ? ? ? ? ? ? ? ? X X ? X

X X X X X X X X X X X X X X X

±1-4% URV ±1% URV ±2-5% URV ±0.5% URV ±2-4% URV ±2-4% URV ±0.5-1% of rate ±0.5-5% URV ±0.5-2% URV ±1-2% URV ±1.25% URV ±3-5% URV ±1-2% URV ±5-10% URV ±1% of rate

±0.5% of rate ±1% of rate ±0.5% of rate ±0.5% of rate ±0.5% of rate ±1% of rate to ±5% URV ±1% of rate to ±5% URV ±1% of rate to ±10% URV ±0.75-1.5% of rate ±0.5% of rate ±2% of rate ±0.15-10% of rate ±1-2% URV ±0.5% of rate to ±4% URV No data available ±6% of ??

FLOWMETER PIPE SIZE, in. (mm)

TYPICAL Accuracy, uncalibrated (Including transmitter)

TYPICAL Reynolds number ‡ or viscosity

TEMPERATURE °F (°C)

PRESSURE psig (kPa)

GASES (VAPORS)

LIQUIDS

PRES

S SL

URRI

ES

VISC

OUS

STEA

M

CLEA

N DI

RTY

HIGH

LO

W

CLEA

N HI

GH

LOW

DI

RTY

CORR

OSI

VE

VERY

CO

RRO

SIVE

FI

BRO

US

ABRA

SIVE

RE

VERS

E FLO

W

PULS

ATIN

G FL

OW

HIGH

TEM

PERA

TURE

CR

YOGE

NIC

SEM

I-FIL

LED

PIPE

S NO

N-NE

WTO

NIAN

S O

PEN

CHAN

NEL

Table 1: Flowmeter Evaluation Table

SQUARE ROOT SCALE: MAXIMUM SINGLE RANGE 4:1 (Typical)**

LINEAR SCALE TYPICAL RANGE 10:1 (Or better)

the flow being measured. Evenwith well maintained and calibratedmeters, this unmeasured flow varieswith changes in fluid viscosity andtemperature. Changes in temperature

also change the internal dimensions ofthe meter and require compensation.

Furthermore, if one can obtain thesame performance from both a fullflowmeter and a point sensor, it is

generally advisable to use theflowmeter. Because point sensors donot look at the full flow, they readaccurately only if they are inserted toa depth where the flow velocity is



Orifice (plate or integral cell)

Segmental Wedge

V-Cone Flowmeter

Target Meters

Venturi Tubes

Flow Nozzles

Pitot Tubes

Elbow Taps

Laminar Flowmeters

Magnetic Flowmeters

Positive Displacement Gas Meters

Positive Displacement Liquid Meters

Turbine Flowmeters

Ultrasonic Flowmeters Time of Flight Doppler

Variable Area (Rotamater)

Vortex Shedding

Fluidic Oscillation (Coanda)

Mass Flowmeters Coriolis

Mass Flowmeters Thermal Probe

Solids Flowmeters

Weirs, Flumes

0.1

1.0

10

102

103

104

Solids Flow Units

105

106

0.1

1.0

10

102

103

104kgm/hr

Sm3/hr or Am3/hr

√

√

√ √

√

√

SD

√

√

√

√ √

SD

√

√

SD

√

√

√

√

√

√

√

√

√

H

A

M

M

M

A

M

N

H

N

M

A

A

N N

M

A

H

M/H

M

-

M

20/5

20/5

2/5

20/5

15/5

20/5

30/5

25/10

15/5

5/3

N

N

15/5

20/5 20/5

N

20/5

20/5

N

20/5

5/3

4/1

3:1

3:1

3:1 to 15:1

15:1

3:1

3:1

3:1

3:1

10:1

30:1

10:1 to 200:1

10:1

10:1

20:1 10:1

10:1

10/1

12/1

20:1

20:1

5:1 to 80:1

100:1

SR

SR

SR

SR

SR

SR

SR

SR

√

√

√

√

√

√ √

√

√

√

√

√

√

SD

H

M

A

H

H

M

M

N

M

H

N N

A

A

A

N

N

M

= Non-standard Range L = Limited SD = Some Designs H = High A = Average M = Minimal N = None SR = Square Root

➀ = The data in this column is for general guidance only. ➁ = Inherent rangeability of primary device is substantially greater than shown. Value used reflects limitations of differential pressure sensing device when 1% of rate accuracy is desired. With multiple-range intelligent transmitters, rangeability can reach 10:1. ➂ = Pipe size establishes the upper limit. ➃ = Practically unlimited with probe type design.

TYPE OF DESIGN

FLOW RANGE

DIRE

CT M

ASS-

FLOW

SENS

OR

DIFF

EREN

TIAL

PRE

SSUR

E-FL

OW SE

NSO

R

VOLU

ME D

ISPL

ACEM

ENT-

FLOW

SENS

OR

VELO

CITY

-FLO

W SE

NSO

R

EXPE

CTED

ERRO

R FR

OM

VIS

COSI

TY C

HANG

E

TRAN

SMIT

TER

AVAI

LABL

E

LINE

AR O

UTPU

T

RANG

EABI

LITY

PRES

SURE

LOSS

THR

U SE

NSO

R

APPR

OX. S

TRAI

GHT

PIPE

-RUN

REQ

UIRE

MEN

T (U

PSTR

EAM

DIA

M./

DOW

NSTR

EAM

DIA

M.)

Table 2: Orientation Table For Flow Sensors

√

√

√

√

√

√

√

√

√

√

SD

SD

√

√ √

√

√

√

√

√

√

√

10-6

10-5

Gas Flow Units

10-6

10-4

10-5

10-3

10-4

10-2

10-3

0.1

10-2

1.0

0.1

10

1.0

102

10

103

102

104

103

105

104

0.05

0.3

2.8

28.3

cc/min

.004

0.04

0.4

3.8

38

379

cc/min

m3/hr

gpm—m3/hr

SCFM—Sm3/hr

10-6

Liquid Flow Units

10-6

10-5

10-5

10-4

10-4

10-3

10-3

10-2

10-2

0.1

0.1

1.0

1.0

10

10

102

102

103

103

104

104

105

106

gpm

gpm—m3/hr

gpm—m3/hr

gpm—m3/hr

gpm—m3/hr

ACFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

ACFM—Sm3/hr

gpm—m3/hr

SCFM—Sm3/hr

gpm—m3/hr

lbm—kgm/hr

SCFM—Sm3/hr

lbm—kgm/hr

SCFM—Sm3/hr

➀➄

➁

➁

➁

➁

➁

➁

➁

➁

➆

➆

➆

➇

➅

➅

➈

➂

➂

➂

➂

➂

➃

➃

➄ = Varies with upstream disturbance. ➅ = Can be more with high Reynolds number services. ➆ = Up to 100:1. ➇ = More for gas turbine meters. ➈ = Higher and lower flow ranges may be available. Check several manufacturers.

the average of the velocity profileacross the pipe. Even if this point iscarefully determined at the time ofcalibration, it is not likely to remainunaltered, since velocity profileschange with flowrate, viscosity, tem-perature, and other factors.

If all other considerations are thesame, but one design offers less pres-sure loss, it is advisable to select thatdesign. Part of the reason is that thepressure loss will have to be paid forin higher pump or compressor operat-ing costs over the life of the plant.Another reason is that a pressure dropis caused by any restriction in the flowpath, and wherever a pipe is restrictedbecomes a potential site for materialbuild-up, plugging, or cavitation.

Before specifying a flowmeter, it isalso advisable to determine whetherthe flow information will be more use-ful if presented in mass or volumetricunits. When measuring the flow ofcompressible materials, volumetricflow is not very meaningful unlessdensity (and sometimes also viscosity)is constant. When the velocity (volu-metric flow) of incompressible liquidsis measured, the presence of suspend-ed bubbles will cause error; therefore,air and gas must be removed beforethe fluid reaches the meter. In othervelocity sensors, pipe liners can causeproblems (ultrasonic), or the metermay stop functioning if the Reynoldsnumber is too low (in vortex sheddingmeters, RD > 20,000 is required).

In view of these considerations,mass flowmeters, which are insensitiveto density, pressure and viscosity vari-ations and are not affected by changesin the Reynolds number, should bekept in mind. Also underutilized in thechemical industry are the variousflumes that can measure flow in par-tially full pipes and can pass largefloating or settlable solids. T



References & Further Reading• OMEGA Complete Flow and Level Measurement Handbook and

Encyclopedia®, OMEGA Press, 1995.• OMEGA Volume 29 Handbook & Encyclopedia, Purchasing Agents

Edition, OMEGA Press, 1995.• “Advanced Process Control for Two-Phase Mixtures,” David Day,

Christopher Reiner and Michael Pepe, Measurements & Control, June, 1997.• Applied Fluid Flow Measurement, N.P. Cheremisinoff, Marcel Decker, 1979.• “Characteristics and Applications of Industrial Thermal Mass Flow

Transmitters,” Jerome L. Kurz, Proceedings 47th Annual Symposium onInstrumentation for the Process Industries, ISA, 1992.

• Developments in Thermal Flow Sensors, Jerome L. Kurz, Ph.D., KurzInstruments Inc., 1987.

• “Differential Flow Measurement of Meter-Conditioned Flow,” Stephen A.Ifft and Andrew J. Zacharias, Measurements & Control, September, 1993.

• Dry Solids Flow Update, Auburn International Inc.• Flow Measurement Engineering Handbook, R.W. Miller, McGraw-Hill, 1983.• Flow Measurement for Engineers and Scientists, N.P. Cheremisinoff,

Marcel Dekker, 1988.• Flow Measurement, Bela Liptak, CRC Press, 1993.• “Flowmeter Geometry Improves Measurement Accuracy,” Stephen A.

Ifft, Measurements & Control, October, 1995.• Flowmeters, F. Cascetta, P. Vigo, ISA, 1990.• Fluidic Flowmeter, Bulletin 1400 MX, Moore Products Co., June, 1988.• Fundamentals of Flow Metering, Technical Data Sheet 3031, Rosemount

Inc., 1982.• Guide to Variable Area Flowmeters, Application No.: T-022 Issue I,

Brooks Instrument Co., 1986.• Incompressible Flow, Donald Panton, Wiley, 1996. • Industrial Flow Measurement, D.W. Spitzer, ISA, 1984.• “Installation Effects on Venturi Tube Flowmeters”, G. Kochen, D.J.M.

Smith, and H. Umbach, Intech, October, 1989.• Instrument Engineers’ Handbook, Bela Liptak, ed., CRC Press, 1995.• “Is a Turbine Flowmeter Right for Your Application?” Michael Hammond,

Flow Control, April, 1998.• “Mass Flowmeters,” Measurements & Control, September, 1991.• Microprocessor-Based 2-Wire Swirlmeter, Bailey-Fischer & Porter Co., 1995.• “Process Gas Mass Flow Controllers: An Overview,” J. G. Olin, Solid State

Technology, April, 1988.• “Target Flowmeters,” George W. Anderson, Measurements & Control,

June, 1982.• Thermal Approach to Flow Measurement, Joseph W. Harpster and

Robert Curry, Intek, Inc. 1991.• “Ultrasonic Flowmeter Basics,” Gabor Vass, Sensors, October, 1997.• “Ultrasonic Flowmeters Pick Up Speed,” Murry Magness, Control, April, 1996.• “User Tips for Mass, Volume Flowmeters,” Donald Ginesi and Carl

Annarummo, Intech, April, 1994.

The calculation of fluid flowrate by reading the pressureloss across a pipe restriction isperhaps the most commonly

used flow measurement technique inindustrial applications (Figure 2-1). Thepressure drops generated by a widevariety of geometrical restrictionshave been well characterized over theyears, and, as compared in Table 2,these primary or “head” flow ele-ments come in a wide variety of con-figurations, each with specific applica-tion strengths and weaknesses.Variations on the theme of differen-tial pressure (d/p) flow measurement

include the use of pitot tubes andvariable-area meters (rotameters), andare discussed later in this chapter.

Primary Element OptionsIn the 18th century, Bernoulli firstestablished the relationship betweenstatic and kinetic energy in a flowingstream. As a fluid passes through arestriction, it accelerates, and theenergy for this acceleration isobtained from the fluid’s static pres-sure. Consequently, the line pressuredrops at the point of constriction(Figure 2-1). Part of the pressure dropis recovered as the flow returns to the

unrestricted pipe. The pressure differ-ential (h) developed by the flow ele-ment is measured, and the velocity (V),the volumetric flow (Q) and the massflow (W) can all be calculated usingthe following generalized formulas:

V = k (h/D)0.5

or Q =kA(h/D)0.5

or W= kA(hD)0.5

k is the discharge coefficient of theelement (which also reflects theunits of measurement), A is the cross-sectional area of the pipe’s opening,and D is the density of the flowing

fluid. The discharge coefficient k isinfluenced by the Reynolds number(see Figure 1-5) and by the “betaratio,” the ratio between the borediameter of the flow restriction andthe inside diameter of the pipe.

Additional parameters or correc-tion factors can be used in the deriva-tion of k, depending on the type offlow element used. These parameterscan be computed from equations orread from graphs and tables availablefrom the American NationalStandards Institute (ANSI), theAmerican Petroleum Institute (API),the American Society of Mechanical

Engineers (ASME), and the AmericanGas Association (AGA), and are includ-ed in many of the works listed as ref-erences at the end of this chapter.

The discharge coefficients of prima-ry elements are determined by labora-tory tests that reproduce the geome-try of the installation. Published valuesgenerally represent the average valuefor that geometry over a minimum of30 calibration runs. The uncertaintiesof these published values vary from0.5% to 3%. By using such publisheddischarge coefficients, it is possible toobtain reasonably accurate flow mea-surements without in-place calibra-tion. In-place calibration is required iftesting laboratories are not availableor if better accuracy is desired thanthat provided by the uncertainty rangenoted above. The relationshipbetween flow and pressure drop varieswith the velocity profile, which can belaminar or turbulent (Figure 2-1) as afunction of the Reynolds number (Re),which for liquid flows can be calcu-lated using the relationship:

Re = 3160(SG)(Q)/(ID)m

where ID is the inside diameter ofthe pipe in inches, Q is the volumet-ric liquid flow in gallons/minute, SGis the fluid specific gravity at 60°F,and m is the viscosity in centipoises.

At low Reynolds numbers (gener-ally under Re = 2,000), the flow islaminar and the velocity profile isparabolic. At high Reynolds num-bers (well over Re = 3,000), the flowbecomes fully turbulent, and theresulting mixing action produces auniform axial velocity across thepipe. As shown in Figure 1-5, the


Primary Element Options

Pitot Tubes

Variable Area Flowmeters

FLOW & LEVEL MEASUREMENTDifferential Pressure Flowmeters

2

TDifferential Pressure Flowmeters

Figure 2-1: Orifice Plate Pressure Drop Recovery

Vena Contracta

Line

Pr

essu

re

Flow Laminar

TurbulentFlow

transition between laminar and tur-bulent flows can cover a wide rangeof Reynolds numbers; the relation-ship with the discharge coefficient isa function of the particular primaryelement.

Today, many engineering societiesand organizations and most primary

element manufacturers offer softwarepackages for sizing d/p flow ele-ments. These programs include therequired data from graphs, charts, andtables as well as empirical equationsfor flow coefficients and correctionfactors. Some include data on thephysical properties of many commonfluids. The user can simply enter theapplication data and automatically

find the recommended size, althoughthese results should be checked forreasonableness by hand calculation.

• Accuracy & RangeabilityThe performance of a head-typeflowmeter installation is a functionof the precision of the flow element

and of the accuracy of the d/p cell.Flow element precision is typicallyreported in percentage of actualreading (AR) terms, whereas d/p cellaccuracy is a percentage of calibrat-ed span (CS). A d/p cell usually pro-vides accuracy of ±0.2% of the cali-brated span (CS). This means that, atthe low end of a 10:1 flow range (at10% flow), corresponding to a differ-

ential pressure range of 100:1, theflowmeter would have an error of±20% AR. For this reason, differentialproducing flowmeters have histori-cally been limited to use within a 3:1or 4:1 range.

Flowmeter rangeability can be fur-ther increased without adverse effect

on accuracy by operating several d/pflowmeters in parallel runs. Only asmany runs are opened at a time asare needed to keep the flow in theactive ones at around 75-90% ofrange. Another option is to stack twoor more transmitters in parallel ontothe same element, one for 1-10%,the other for 10-100% of full scale(FS) d/p produced. Both of these

2 Differential Pressure Flowmeters


Square edge concentric orifice plate

Conical/quadrant edge concentric orifice plate

Eccentric/segmental orifice plate

Integral orifice

Venturi/flowtube

Nozzle

Segmental wedge

Venturi cone

PRIMARY ELEMENT RECOMMENDED SERVICE MINIMUM SIZES ADVANTAGES LIMITATIONS RE LIMITS

≥ 2000

≥500

>10,000

>10,000

>75,000

>50,000

>500

None cited

≥ 1/2 in

1 to 6 in

4 to 14 in

1/2 to 2 in

1/2 to 72 in

>2 in

≥1/2 in

1 to 16 in

Easy to install Low cost Easy to replace



Easy to install No lead lines Low cost

Low head loss 2 to 9 times less relaxation piping than orifice Higher flow capacity than orifice for the same differential pressure Accuracy less affected by wear and installation conditions than orifice

Higher flow capacity than orifice for the same differential pressure Accuracy less affected by wear and installation conditions than orifice Good for high temperature and high velocity applications Mass transfer standard for gases

No lead lines Minimal clogging potential 40% less head loss than orifice Minimal relaxation piping

Minimal relaxation piping Low flow capability

Relaxation piping requirements High head loss Accuracy affected by installation and orifice condition

Relaxation piping requirements High head loss Accuracy affected by installation and orifice condition

Relaxation piping requirements High head loss Accuracy affected by installation and orifice condition Higher uncertainties of discharge coefficient data

Relaxation piping requirements Proprietary design requires calibration High head loss More prone to clogging than standard orifice plate

High initial cost

Harder to replace than orifice High head loss

Proprietary design needs calibration High initial cost Requires remote seal differential pressure transmitter, harder to zero

Proprietary design

Clean liquids, gases, steam

Viscous liquids

Liquids and gases containing secondary fluid phases


Clean & dirty liquids, gases, steam; slurries


Dirty liquids, gases, steam; slurries; viscous liquids

Clean & dirty liquids, gases, steam; viscous liquids

Table 3: Primary or "Head Flow" Element Comparisons

techniques are cumbersome andexpensive. Intelligent transmittersoffer a better option.

The accuracy of intelligent trans-mitters is usually stated as ±0.1% CS,which includes only errors due tohysteresis, rangeability and linearity.Potential errors due to drift, temper-ature, humidity, vibration, overrange,radio frequency interference andpower supply variation are allexcluded. If one includes them, inac-curacy is about 0.2% CS. Because

intelligent d/p transmitters can—based on their own measurements—automatically switch ranges betweentwo calibrated spans (one for 1-10%,the other for 10-100% of FS d/p), itshould be possible to obtain orificeinstallations with 1% AR inaccuracyover a 10:1 flow range.

In most flowmetering applications,density is not measured directly.Rather, it is assumed to have somenormal value. If density deviates fromthis assumed value, error results.Density error can be corrected if it ismeasured directly or indirectly bymeasuring pressure in gases or temper-ature in liquids. Flow computing pack-ages are also available that accept theinputs of the d/p transmitter and theother sensors and can simultaneouslycalculate mass and volumetric flow.

To minimize error (and the need fordensity correction) when dealing with

compressible fluids, the ratio of dif-ferential pressure (h) divided byupstream pressure (P) should notexceed 0.25 (measured in the sameengineering units).

Metering errors due to incorrectinstallation of the primary elementcan be substantial (up to 10%).Causes of such errors can be thecondition of the mating pipe sec-tions, insufficient straight pipe runs,and pressure tap and lead linedesign errors.

Under turbulent flow conditions,as much as 10% of the d/p signal canbe noise caused by disturbancesfrom valves and fittings, both up- anddownstream of the element, and bythe element itself. In the majority ofapplications, the damping providedin d/p cells is sufficient to filter outthe noise. Severe noise can bereduced by the use of two or morepressure taps connected in parallelon both sides of the d/p cell.

Pulsating flow can be caused byreciprocating pumps or compressors.This pulsation can be reduced bymoving the flowmeter away from thesource of the pulse, or downstreamof filters or other dampeningdevices. Pulsation dampening hard-ware can also be installed at thepressure taps, or dampening soft-ware can applied to the d/p cell out-put signal. One such filter is the

inverse derivative algorithm, whichblocks any rate of change occurringmore quickly than the rate at whichthe process flow can change.

• Piping, Installation, & MaintenanceInstallation guidelines are publishedby various professional organizations(ISA, ANSI, API, ASME, AGA) andby manufacturers of proprietarydesigns. These guidelines includesuch recommendations as: • When, in addition to measuring

the flow, the process temperatureor pressure is also to be measured,the pressure transmitter shouldnot be installed in the processpipe, but should be connected tothe appropriate lead line of theflow element via a tee.

• Similarly, the thermowell used fortemperature measurement shouldbe installed at least 10 diametersdownstream of the flow element, toprevent velocity profile distortions.

• Welds should be ground smoothand gaskets trimmed so that noprotrusion can be detected byphysical inspection. In order for the velocity profile to

fully develop (and the pressure dropto be predictable), straight pipe runsare required both up- and down-stream of the d/p element. Theamount of straight run requireddepends on both the beta ratio of

Differential Pressure Flowmeters 2


Figure 2-2: Flow Straighteners Installed Upstream of Primary Element

Flow

A B

7 Pipe Diameters

Profile Concentrator

Swirl Reducer

Settling Distance (4 Pipe Diameters)

the installation and on the nature ofthe upstream components in thepipeline. For example, when a single90° elbow precedes an orifice plate, thestraight-pipe requirement ranges from6 to 20 pipe diameters as the diameterratio is increased from 0.2 to 0.8.

In order to reduce the straight runrequirement, flow straighteners(Figure 2-2) such as tube bundles,perforated plates, or internal tabscan be installed upstream of the pri-mary element.

The size and orientation of thepressure taps are a function of boththe pipe size and the type of processfluid. The recommended maximumdiameter of pressure tap holesthrough the pipe or flange is G" forpipes under 2" in diameter, K" for 2"and 3" pipes, H" for 4 to 8" and I" forlarger pipes. Both taps should be ofthe same diameter, and, where thehole breaks through the inside pipesurface, it should be square with noroughness, burrs, or wire edges.Connections to pressure holesshould be made by nipples, cou-plings, or adaptors welded to theoutside surface of the pipe.

On services where the processfluid can plug the pressure taps ormight gel or freeze in the lead lines,chemical seal protectors can beused. Connection sizes are usuallylarger (seal elements can also beprovided with diaphragm exten-sions), and, because of the spacerequirement, they are usuallyinstalled at “radius tap” or “pipetap” locations, as shown in Figure 2-3. When chemical seals are used, itis important that the two connect-ing capillaries, as they are routed tothe d/p cell, experience the sametemperature and are kept shieldedfrom sunlight.

The d/p transmitter should be

located as close to the primary ele-ment as possible. Lead lines shouldbe as short as possible and of thesame diameter. In clean liquid ser-vice, the minimum diameter is G",while in condensable vapor service,the minimum diameter is 0.4". Insteam service, the horizontal leadlines should be kept as short as pos-sible and be tilted (with a minimumgradient of 1 in/ft with respect tothe piping) towards the tap, so thatcondensate can drain back into thepipe. Again, both lead lines should beexposed to the same ambient condi-tions and be shielded from sunlight.In clean liquid or gas service, the leadlines can be purged through the d/p

cell vent or drain connections, andthey should be flushed for severalminutes to remove all air from thelines. Entrapped air can offset thezero calibration.

Seal pots are on the wet leg in d/pcell installations with small ranges(under 10 in H2O) in order to mini-mize the level variation in the legs. Insteam applications, filling tees arerecommended to ensure equalheight condensate legs on both sidesof the d/p cell. If for some reasonthe two legs are not of equal height,the d/p cell can be biased to zero

out the difference, as long as thatdifference does not change.

If the process temperature exceedsthe maximum temperature limitationof the d/p cell, either chemical sealshave to be used or the lead lines needto be long enough to cool the fluid. Ifa large temperature drop is required, acoiled section of tubing (pigtail) canbe installed in the lead lines to coolthe process fluids.

The frequency of inspection orreplacement of a primary elementdepends on the erosive and corro-sive nature of the process and on theoverall accuracy required. If there isno previous experience, the orificeplate can be removed for inspection

during the first three, six, and 12months of its operation. Based onvisual inspection of the plate, a rea-sonable maintenance cycle can beextrapolated from the findings.Orifices used for material balancecalculations should be on the samemaintenance cycle.

• Sizing the Orifice PlateThe orifice plate is commonly usedin clean liquid, gas, and steam ser-vice. It is available for all pipe sizes,and if the pressure drop it requires isfree, it is very cost-effective for



Figure 2-3: Differential Pressure Tap Location Alternatives

Pipe Taps

Flange Taps

Corner Taps

8D

D D/2

1 in. 1 in.

2 1 D2

Flow

measuring flows in larger pipes (over6" diameter). The orifice plate is alsoapproved by many standards organi-zations for the custody transfer ofliquids and gases.

The orifice flow equations usedtoday still differ from one another,although the various standards orga-nizations are working to adopt a sin-gle, universally accepted orifice flowequation. Orifice sizing programsusually allow the user to select theflow equation desired from amongseveral.

The orifice plate can be made ofany material, although stainless steelis the most common. The thicknessof the plate used (J-H") is a func-tion of the line size, the process tem-perature, the pressure, and the differ-ential pressure. The traditional ori-fice is a thin circular plate (with a tabfor handling and for data), inserted

into the pipeline between the twoflanges of an orifice union. Thismethod of installation is cost-effec-tive, but it calls for a process shut-down whenever the plate is removedfor maintenance or inspection. Incontrast, an orifice fitting allows theorifice to be removed from the

process without depressurizing theline and shutting down flow. In suchfittings, the universal orifice plate, acircular plate with no tab, is used.

The concentric orifice plate(Figure 2-4A) has a sharp (square-edged) concentric bore that providesan almost pure line contact betweenthe plate and the fluid, with negligi-ble friction drag at the boundary. Thebeta (or diameter) ratios of concen-tric orifice plates range from 0.25 to0.75. The maximum velocity and min-imum static pressure occurs at some0.35 to 0.85 pipe diameters down-stream from the orifice plate. Thatpoint is called the vena contracta.Measuring the differential pressure ata location close to the orifice plateminimizes the effect of pipe rough-ness, since friction has an effect onthe fluid and the pipe wall.

Flange taps are predominantly

used in the United States and arelocated 1 inch from the orifice plate’ssurfaces (Figure 2-3). They are notrecommended for use on pipelinesunder 2 inches in diameter. Cornertaps are predominant in Europe forall sizes of pipe, and are used in theUnited States for pipes under 2 inches

(Figure 2-3). With corner taps, therelatively small clearances representa potential maintenance problem.Vena contracta taps (which areclose to the radius taps, Figure 2-4)are located one pipe diameterupstream from the plate, and down-stream at the point of vena contrac-ta. This location varies (with betaratio and Reynolds number) from0.35D to 0.8D.

The vena contracta taps providethe maximum pressure differential,but also the most noise. Additionally,if the plate is changed, it may requirea change in the tap location. Also, insmall pipes, the vena contracta mightlie under a flange. Therefore, venacontracta taps normally are usedonly in pipe sizes exceeding six inches.

Radius taps are similar to venacontracta taps, except the down-stream tap is fixed at 0.5D from the

orifice plate (Figure 2-3). Pipe taps arelocated 2.5 pipe diameters upstreamand 8 diameters downstream fromthe orifice (Figure 2-3). They detectthe smallest pressure difference and,because of the tap distance from theorifice, the effects of pipe rough-ness, dimensional inconsistencies,



Figure 2-4: Orifice Plate Openings

Vent Hole Location (Liquid Service)

Drain Hole Location (Vapor Service)

Pipe Internal

Diameter

A) Concentric B) Eccentric C) Segmental

Flow

Upstream Sharp Edge

1/8 in (3.175 mm) Maximum

1/8 in - 1/2 in (3.175-12.70 mm)

45°

Bevel Where Thickness Is Greater Than 1/8 in (3.175 mm) or the Orifice Diameter Is Less Than 1 in (25 mm) Orifice

and, therefore, measurement errorsare the greatest.

• Orifice Types & SelectionThe concentric orifice plate is rec-ommended for clean liquids, gases,and steam flows when Reynoldsnumbers range from 20,000 to 107 inpipes under six inches. Because thebasic orifice flow equations assumethat flow velocities are well belowsonic, a different theoretical andcomputational approach is requiredif sonic velocities are expected. Theminimum recommended Reynoldsnumber for flow through an orifice(Figure 1-5) varies with the beta ratioof the orifice and with the pipe size.In larger size pipes, the minimumReynolds number also rises.

Because of this minimum Reynoldsnumber consideration, square-edgedorifices are seldom used on viscousfluids. Quadrant-edged and conicalorifice plates (Figure 2-5) are recom-mended when the Reynolds numberis under 10,000. Flange taps, corner,and radius taps can all be used withquadrant-edged orifices, but onlycorner taps should be used with aconical orifice.

Concentric orifice plates can beprovided with drain holes to pre-vent buildup of entrained liquids ingas streams, or with vent holes forventing entrained gases from liquids(Figure 2-4A). The unmeasured flowpassing through the vent or drainhole is usually less than 1% of thetotal flow if the hole diameter isless than 10% of the orifice bore.The effectiveness of vent/drainholes is limited, however, becausethey often plug up.

Concentric orifice plates are notrecommended for multi-phase flu-ids in horizontal lines because thesecondary phase can build up

around the upstream edge of theplate. In extreme cases, this canclog the opening, or it can changethe flow pattern, creating measure-ment error. Eccentric and segmentalorifice plates are better suited forsuch applications. Concentric ori-fices are still preferred for multi-phase flows in vertical linesbecause accumulation of material isless likely and the sizing data forthese plates is more reliable.

The eccentric orifice (Figure 2-4B)

is similar to the concentric exceptthat the opening is offset from thepipe’s centerline. The opening of thesegmental orifice (Figure 2-4C) is asegment of a circle. If the secondaryphase is a gas, the opening of aneccentric orifice will be locatedtowards the top of the pipe. If thesecondary phase is a liquid in a gas ora slurry in a liquid stream, the openingshould be at the bottom of the pipe.The drainage area of the segmentalorifice is greater than that of theeccentric orifice, and, therefore, it ispreferred in applications with highproportions of the secondary phase.

These plates are usually used in pipesizes exceeding four inches in diame-ter, and must be carefully installed tomake sure that no portion of theflange or gasket interferes with theopening. Flange taps are used withboth types of plates, and are locatedin the quadrant opposite the openingfor the eccentric orifice, in line withthe maximum dam height for thesegmental orifice.

For the measurement of low flowrates, a d/p cell with an integral

orifice may be the best choice. In thisdesign, the total process flow passesthrough the d/p cell, eliminating theneed for lead lines. These are propri-etary devices with little publisheddata on their performance; their flowcoefficients are based on actual lab-oratory calibrations. They are recom-mended for clean, single-phase fluidsonly because even small amounts ofbuild-up will create significant mea-surement errors or will clog the unit.Restriction orifices are installed toremove excess pressure and usuallyoperate at sonic velocities with verysmall beta ratios. The pressure drop



Figure 2-5: Orifices for Viscous Flows

A) Quadrant-Edged

Flow

B) Conical

Flow

45°

across a single restriction orificeshould not exceed 500 psid becauseof plugging or galling. In multi-ele-ment restriction orifice installations,the plates are placed approximatelyone pipe diameter from one anotherin order to prevent pressure recoverybetween the plates.

• Orifice PerformanceAlthough it is a simple device, theorifice plate is, in principle, a preci-sion instrument. Under ideal condi-tions, the inaccuracy of an orificeplate can be in the range of 0.75-1.5%AR. Orifice plates are, however, quite

sensitive to a variety of error-induc-ing conditions. Precision in the borecalculations, the quality of the instal-lation, and the condition of the plateitself determine total performance.Installation factors include tap loca-tion and condition, condition of the

process pipe, adequacy of straightpipe runs, gasket interference, mis-alignment of pipe and orifice bores,and lead line design. Other adverseconditions include the dulling of thesharp edge or nicks caused by corro-sion or erosion, warpage of the platedue to waterhammer and dirt, andgrease or secondary phase depositson either orifice surface. Any of theabove conditions can change the ori-fice discharge coefficient by as muchas 10%. In combination, these prob-lems can be even more worrisomeand the net effect unpredictable.Therefore, under average operating

conditions, a typical orifice installa-tion can be expected to have anoverall inaccuracy in the range of 2 to5% AR.

The typical custody-transfer gradeorifice meter is more accurate becauseit can be calibrated in a testing

laboratory and is provided with honedpipe sections, flow straighteners,senior orifice fittings, and tempera-ture controlled enclosures.

• Venturi & FlowtubesVenturi tubes are available in sizesup to 72", and can pass 25 to 50%more flow than an orifice with thesame pressure drop. Furthermore,the total unrecovered head lossrarely exceeds 10% of measured d/p(Figure 2-6). The initial cost of ven-turi tubes is high, so they are pri-marily used on larger flows or onmore difficult or demanding flowapplications. Venturis are insensitiveto velocity profile effects andtherefore require less straight piperun than an orifice. Their contourednature, combined with the self-scouring action of the flow throughthe tube, makes the device immuneto corrosion, erosion, and internalscale build up. In spite of its high ini-tial cost, the total cost of owner-ship can still be favorable becauseof savings in installation and operat-ing and maintenance costs.

The classical Herschel venturi has avery long flow element characterizedby a tapered inlet and a diverging out-let. Inlet pressure is measured at theentrance, and static pressure in thethroat section. The pressure taps feedinto a common annular chamber, pro-viding an average pressure readingover the entire circumference of theelement. The classical venturi is limit-ed in its application to clean, non-cor-rosive liquids and gases.

In the short form venturi, theentrance angle is increased and theannular chambers are replaced bypipe taps (Figure 2-7A). The short-form venturi maintains many of theadvantages of the classical venturi,but at a reduced initial cost, shorter



Figure 2-6: Pressure Loss-Venturi vs Orifice

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

90

80

70

60

50

40

30

20

10 Low Loss Venturi

Long Form Venturi

Standard Venturi

ASME Flow Nozzle

Orifice Plate

Reco

very

—Pe

rcen

t of

Diff

eren

tial

Unr

ecov

ered

Pre

ssur

e Lo

ss—

Perc

ent

of D

iffer

enti

al

Proprietary Flow Tube

Beta (Diameter) Ratio

10

20

30

40

50

60

70

80

90

length and reduced weight. Pressuretaps are located G to H pipe diame-ter upstream of the inlet cone, and in

the middle of the throat section.Piezometer rings can be used withlarge venturi tubes to compensatefor velocity profile distortions. Inslurry service, the pipe taps can bepurged or replaced with chemicalseals, which can eliminate all dead-ended cavities.

There are several proprietary flow-tube designs which provide evenbetter pressure recovery than theclassical venturi. The best known ofthese proprietary designs is the uni-versal venturi (Figure 2-7B). The vari-ous flowtube designs vary in theircontours, tap locations, generatedd/p and in their unrecovered headloss. They all have short lay lengths,typically varying between 2 and 4pipe diameters. These proprietaryflowtubes usually cost less than theclassical and short-form venturisbecause of their short lay length.However, they may also require morestraight pipe run to condition theirflow velocity profiles.

Flowtube performance is muchaffected by calibration. The inaccuracyof the discharge coefficient in auniversal venturi, at Reynolds num-bers exceeding 75,000, is 0.5%. Theinaccuracy of a classical venturi at

Re > 200,000 is between 0.7 and 1.5%.Flowtubes are often supplied withdischarge coefficient graphs because

the discharge coefficient changes asthe Reynolds number drops. Thevariation in the discharge coefficientof a venturi caused by pipe rough-ness is less than 1% because there iscontinuous contact between thefluid and the internal pipe surface.

The high turbulence and the lack ofcavities in which material can accu-mulate make flow tubes well suitedfor slurry and sludge services.However, maintenance costs can behigh if air purging cannot preventplugging of the pressure taps and leadlines. Plunger-like devices (vent clean-ers) can be installed to periodically

remove buildup from interior open-ings, even while the meter is online.Lead lines can also be replaced withbutton-type seal elements hydrauli-

cally coupled to the d/p transmitterusing filled capillaries. Overall mea-surement accuracy can drop if the

chemical seal is small, its diaphragmis stiff, or if the capillary system isnot temperature-compensated ornot shielded from direct sunlight.

• Flow NozzlesThe flow nozzle is dimensionallymore stable than the orifice plate,particularly in high temperature andhigh velocity services. It has oftenbeen used to measure highflowrates of superheated steam.The flow nozzle, like the venturi,has a greater flow capacity than theorifice plate and requires a lowerinitial investment than a venturi

tube, but also provides less pressurerecovery (Figure 2-6). A major disad-vantage of the nozzle is that it ismore difficult to replace than the



Figure 2-7: Gradual Flow Elements

High Pressure Tap

A) Short-Form Venturi Tube B) Universal Venturi C) Flow Nozzle

Flow D

Low Pressure Tap

Inlet Inlet Cone

Throat Outlet Cone

d

D±.1D .5D±.1D

Figure 2-8: Proprietary Elements for Difficult Fluids

A) Segmental Wedge

Wedge Flow Element

DH

B) V-Cone

H L

orifice unless it can be removed aspart of a spool section.

The ASME pipe tap flow nozzle ispredominant in the United States(Figure 2-7C). The downstream endof a nozzle is a short tube having thesame diameter as the vena contrac-ta of an equivalent orifice plate. Thelow-beta designs range in diameterratios from 0.2 to 0.5, while the highbeta-ratio designs vary between0.45 and 0.8. The nozzle shouldalways be centered in the pipe, andthe downstream pressure tapshould be inside the nozzle exit. Thethroat taper should always decreasethe diameter toward the exit. Flownozzles are not recommended forslurries or dirty fluids. The mostcommon flow nozzle is the flangetype. Taps are commonly locatedone pipe diameter upstream and H

pipe diameter downstream fromthe inlet face.

Flow nozzle accuracy is typically

1% AR, with a potential for 0.25% ARif calibrated. While discharge coeffi-cient data is available for Reynoldsnumbers as low as 5,000, it is advis-able to use flow nozzles only whenthe Reynolds number exceeds 50,000.Flow nozzles maintain their accuracyfor long periods, even in difficult ser-vice. Flow nozzles can be a highly

accurate way to measure gas flows.When the gas velocity reaches thespeed of sound in the throat, thevelocity cannot increase any more(even if downstream pressure isreduced), and a choked flow condi-tion is reached. Such “critical flownozzles” are very accurate and oftenare used in flow laboratories as stan-dards for calibrating other gasflowmetering devices.

Nozzles can be installed in anyposition, although horizontal orien-tation is preferred. Vertical down-flow is preferred for wet steam,gases, or liquids containing solids.The straight pipe run requirementsare similar to those of orifice plates.

• Segmental Wedge ElementsThe segmental wedge element (Figure2-8A) is a proprietary device designedfor use in slurry, corrosive, erosive,viscous, or high-temperature applica-tions. It is relatively expensive and is

used mostly on difficult fluids, wherethe dramatic savings in maintenancecan justify the initial cost. The uniqueflow restriction is designed to last thelife of the installation without deteri-oration.

Wedge elements are used with3-in diameter chemical seals, elimi-nating both the lead lines and any

dead-ended cavities. The seals attachto the meter body immediatelyupstream and downstream of therestriction. They rarely require clean-ing, even in services like dewateredsludge, black liquor, coal slurry, flyash slurry, taconite, and crude oil.The minimum Reynolds number isonly 500, and the meter requiresonly five diameters of upstreamstraight pipe run.

The segmental wedge has aV-shaped restriction characterizedby the H/D ratio, where H is theheight of the opening below therestriction and D is the diameter. TheH/D ratio can be varied to match theflow range and to produce thedesired d/p. The oncoming flow cre-ates a sweeping action through themeter. This provides a scouring effecton both faces of the restriction,helping to keep it clean and free ofbuildup. Segmental wedges can mea-sure flow in both directions, but thed/p transmitter must be calibratedfor a split range, or the flow elementmust be provided with two sets ofconnections for two d/p transmit-ters (one for forward and one forreverse flow).

An uncalibrated wedge elementcan be expected to have a 2% to 5%AR inaccuracy over a 3:1 range. A cal-ibrated wedge element can reducethat to 0.5% AR if the fluid density isconstant. If slurry density is variableand/or unmeasured, error rises.

• Venturi-Cone ElementThe venturi-cone (V-cone) element(Figure 2-8B) is another proprietarydesign that promises consistent per-formance at low Reynolds numbersand is insensitive to velocity profiledistortion or swirl effects. Again, how-ever, it is relatively expensive. The V-cone restriction has a unique geometry



Figure 2-9: Pitot Tubes Measure Two Pressures

Static Pressure Holes Outer Pipe Only

(P)

Impact Pressure Opening (Pt)

P

Vp

Pt

Stainless Steel Tubing

Impact Pressure Connection

Tubing Adaptor

Static Pressure Connection

Vp ~ Pt - P

that minimizes accuracy degradationdue to wear, making it a good choicefor high velocity flows and ero-sive/corrosive applications.

The V-cone creates a controlledturbulence region that flattens theincoming irregular velocity profileand induces a stable differentialpressure that is sensed by a down-stream tap. The beta ratio of aV-cone is so defined that an orificeand a V-cone with equal beta ratioswill have equal opening areas.

Beta ratio = (D2 - d2).05 / D

where d is the cone diameter and Dis the inside diameter of the pipe.

With this design, the beta ratio canexceed 0.75. For example, a 3-in meterwith a beta ratio of 0.3 can have a 0 to75 gpm range. Published test results onliquid and gas flows place the systemaccuracy between 0.25 and 1.2% AR.

Pitot TubesAlthough the pitot tube is one of thesimplest flow sensors, it is used in awide range of flow measurementapplications such as air speed in rac-ing cars and Air Force fighter jets. Inindustrial applications, pitot tubesare used to measure air flow in pipes,ducts, and stacks, and liquid flow inpipes, weirs, and open channels.While accuracy and rangeability arerelatively low, pitot tubes are simple,reliable, inexpensive, and suited for avariety of environmental conditions,including extremely high tempera-tures and a wide range of pressures.

The pitot tube is an inexpensivealternative to an orifice plate.Accuracy ranges from 0.5% to 5% FS,which is comparable to that of anorifice. Its flow rangeability of 3:1(some operate at 4:1) is also similarto the capability of the orificeplate. The main difference is that,

while an orifice measures the fullflowstream, the pitot tube detectsthe flow velocity at only one point inthe flowstream. An advantage of theslender pitot tube is that it can beinserted into existing and pressurizedpipelines (called hot-tapping) with-out requiring a shutdown.

• Theory of OperationPitot tubes were invented by HenriPitot in 1732 to measure the flowingvelocity of fluids. Basically a differ-ential pressure (d/p) flowmeter, apitot tube measures two pressures:the static and the total impact pres-sure. The static pressure is the oper-ating pressure in the pipe, duct, orthe environment, upstream to thepitot tube. It is measured at rightangles to the flow direction, prefer-ably in a low turbulence location(Figure 2-9).

The total impact pressure (PT) isthe sum of the static and kineticpressures and is detected as theflowing stream impacts on the pitotopening. To measure impact pres-sure, most pitot tubes use a small,



Figure 2-10: Pipeline Installation of Pitot Tube

Impact Opening

Flow

Static Opening

Impact (High Pressure)

Connection

Static (Low Pressure)

Connection

Stuffing Box

Packing Nut

Corporation Cock

P

Pt

Figure 2-11: Traverse Point Locations

Circular Stack (10-Point Traverse)

0.916 R 0.837 R

0.707 R 0.548 R

0.316 R

Rectangular Stack (Measure at Center of at

Least 9 Equal Areas)

R

sometimes L-shaped tube, with theopening directly facing the oncom-ing flowstream. The point velocityof approach (VP) can be calculatedby taking the square root of the dif-ference between the total pressure(PT) and the static pressure (P) andmultiplying that by the C/D ratio,where C is a dimensional constantand D is density:

VP = C(PT - P)HH /D

When the flowrate is obtained bymultiplying the point velocity (VP) bythe cross-sectional area of the pipeor duct, it is critical that the velocitymeasurement be made at an inser-tion depth which corresponds to theaverage velocity. As the flow velocityrises, the velocity profile in the pipechanges from elongated (laminar) tomore flat (turbulent). This changesthe point of average velocity and

requires an adjustment of the inser-tion depth. Pitot tubes are recom-mended only for highly turbulentflows (Reynolds Numbers > 20,000)and, under these conditions, thevelocity profile tends to be flatenough so that the insertion depth isnot critical.

In 1797, G.B. Venturi developed ashort tube with a throat-like pas-sage that increases flow velocityand reduces the permanent pressuredrop. Special pitot designs are avail-able that, instead of providing justan impact hole for opening, add asingle or double venturi to theimpact opening of the pitot tube.The venturi version generates ahigher differential pressure thandoes a regular pitot tube.

• Static Pressure MeasurementIn jacketed (dual-walled) pitot-tubedesigns, the impact pressure port

faces forward into the flow, whilestatic ports do not, but are, instead,spaced around the outer tube. Bothpressure signals (PT and P) are routedby tubing to a d/p indicator ortransmitter. In industrial applica-tions, the static pressure (P) can bemeasured in three ways: 1) throughtaps in the pipe wall; 2) by staticprobes inserted in the processstream; or 3) by small openingslocated on the pitot tube itself or ona separate aerodynamic element.

Wall taps can measure static pres-sures at flow velocities up to 200ft/sec. A static probe (resembling anL-shaped pitot tube) can have fourholes of 0.04 inches in diameter,spaced 90° apart. Aerodynamic bod-ies can be cylinders or wedges, withtwo or more sensing ports.

Errors in detecting static pressurearise from fluid viscosity, velocity, andfluid compressibility. The key to accu-rate static pressure detection is tominimize the kinetic component inthe pressure measurement.



Figure 2-12: Multiple-Opening Averaging Pitot Tube

High Pressure Profile

Average High (Impact) Pressure

DP

PLPH

Average Low (Static) Pressure

Low Pressure Profile

Velocity Profile

Average Velocity

PHPt = = P

(9.5, 22, 32, or 51 mm)

PL

A = 3", 7", 1 1", or 2" 8 8 4

Pitot tube shown with associated fittings and

pressure transmitter.

• Single-Port Pitot TubesA single-port pitot tube can measurethe flow velocity at only a singlepoint in the cross-section of a flow-ing stream (Figure 2-10). The probemust be inserted to a point in theflowing stream where the flowvelocity is the average of the veloci-ties across the cross-section, and itsimpact port must face directly intothe fluid flow. The pitot tube can bemade less sensitive to flow directionif the impact port has an internalbevel of about 15°, extending about 1.5diameters into the tube.

If the pressure differential gener-ated by the venturi is too low foraccurate detection, the convention-al pitot tube can be replaced by apitot venturi or a double venturisensor. This will produce a higherpressure differential.

A calibrated, clean and properlyinserted single-port pitot tube canprovide ±1% of full scale flow accura-cy over a flow range of 3:1; and, withsome loss of accuracy, it can evenmeasure over a range of 4:1. Its advan-tages are low cost, no moving parts,simplicity, and the fact that it causesvery little pressure loss in the flowingstream. Its main limitations includethe errors resulting from velocityprofile changes or from plugging ofthe pressure ports. Pitot tubes aregenerally used for flow measure-ments of secondary importance,where cost is a major concern,and/or when the pipe or duct diam-eter is large (up to 72 inches or more).

Specially designed pitot probeshave been developed for use withpulsating flows. One design uses apitot probe filled with silicone oil totransmit the process pressures tothe d/p cell. At high frequency pul-sating applications, the oil serves asa pulsation dampening and pressure-

averaging medium.Pitot tubes also can be used in

square, rectangular or circular airducts. Typically, the pitot tube fitsthrough a 5/16-in diameter hole inthe duct. Mounting can be by aflange or gland. The tube is usuallyprovided with an external indicator,

so that its impact port can be accu-rately rotated to face directly intothe flow. In addition, the tube can bedesigned for detecting the full veloc-ity profile by making rapid and con-sistent traverses across the duct.

In some applications, such as EPA-mandated stack particulate sampling,it is necessary to traverse a pitotsampler across a stack or duct. Inthese applications, at each pointnoted in Figure 2-11, a temperatureand flow measurement is made inaddition to taking a gas sample,which data are then combined andtaken to a laboratory for analysis. Insuch applications, a single probecontains a pitot tube, a thermocou-ple, and a sampling nozzle.

A pitot tube also can be used to

measure water velocity in openchannels, at drops, chutes, or overfall crests. At the low flow velocitiestypical of laminar conditions, pitottubes are not recommendedbecause it is difficult to find theinsertion depth corresponding tothe average velocity and because

the pitot element produces such asmall pressure differential. The use ofa pitot venturi does improve on thissituation by increasing the pressuredifferential, but cannot help theproblem caused by the elongatedvelocity profile.

• Averaging Pitot TubesAveraging pitot tubes been introducedto overcome the problem of findingthe average velocity point. An averag-ing pitot tube is provided with multi-ple impact and static pressure portsand is designed to extend across theentire diameter of the pipe. The pres-sures detected by all the impact (andseparately by all the static) pressureports are combined and the squareroot of their difference is measured as



Figure 2-13: Area Averaging Pitot Station

an indication of the average flow inthe pipe (Figure 2-12). The port closerto the outlet of the combined signalhas a slightly greater influence, thanthe port that is farthest away, but, forsecondary applications where pitottubes are commonly used, this erroris acceptable.

The number of impact ports, thedistance between ports, and thediameter of the averaging pitot tubeall can be modified to match theneeds of a particular application.Sensing ports in averaging pitot tubesare often too large to allow the tubeto behave as a true averaging cham-ber. This is because the oversized

port openings are optimized not foraveraging, but to prevent plugging. Insome installations, purging with aninert gas is used to keep the portsclean, allowing the sensor to usesmaller ports.

Averaging pitot tubes offer the

same advantages and disadvantagesas do single-port tubes. They areslightly more expensive and a littlemore accurate, especially if the flowis not fully formed. Some averagingpitot sensors can be inserted throughthe same opening (or hot tap) whichaccommodates a single-port tube.

• Area AveragingArea-averaging pitot stations areused to measure the large flows oflow pressure air in boilers, dryers, orHVAC systems. These units are avail-able for the various standard sizes ofcircular or rectangular ducts (Figure2-13) and for pipes. They are so

designed that each segment of thecross-section is provided with bothan impact and a static pressure port.Each set of ports is connected to itsown manifold, which combines theaverage static and average impactpressure signals. If plugging is likely,

the manifolds can be purged to keepthe ports clean.

Because area-averaging pitot sta-tions generate very small pressure dif-ferentials, it may be necessary to uselow differential d/p cells with spansas low as 0-0.01 in water column. Toimprove accuracy, a hexagonal cell-type flow straightener and a flownozzle can be installed upstream ofthe area-averaging pitot flow sensor.The flow straightener removes localturbulence, while the nozzle ampli-fies the differential pressure pro-duced by the sensor.

• Installation Pitot tubes can be used as permanentlyinstalled flow sensors or as portablemonitoring devices providing periodicdata. Permanently installed carbonsteel or stainless steel units can oper-ate at up to 1400 PSIG pressures andare inserted into the pipe throughflanged or screw connections. Theirinstallation usually occurs prior toplant start-up, but they can be hot-tapped into an operating process.

In a hot-tap installation (Figure2-14), one first welds a fitting to thepipe. Then a drill-through valve isattached to the fitting and a hole isdrilled through the pipe. Then, afterpartially withdrawing the drill, thevalve is closed, the drill is removedand the pitot tube is inserted. Finally,the valve is opened and the pitottube is fully inserted.

The velocity profile of the flowingstream inside the pipe is affected bythe Reynolds number of the flowingfluid, pipe surface roughness, and byupstream disturbances, such asvalves, elbows, and other fittings.Pitot tubes should be used only if theminimum Reynolds number exceeds20,000 and if either a straight run ofabout 25 diameters can be provided



Figure 2-14: Hot Tap Installation of a Pitot Tube

Drill Thru Valve

Installed

Inserted

upstream to the pitot tube or ifstraightening vanes can be installed.

• Vibration DamageNatural frequency resonant vibra-tions can cause pitot tube failure.

Natural frequency vibration is causedby forces created as vortices are shedby the pitot tube. The pitot tube isexpected to experience such vibra-tion if the process fluid velocity (infeet per second) is between a lowerlimit (VL) and an upper limit (VH). Thevalues of VL and VH can be calculated(for the products of a given manufac-turer) using the equations below.

VL = 5253(M x Pr x D)/L2

VH = 7879(M x Pr x D)/L2

Where M = mounting factor (3.52 forsingle mount); Pr = probe factor (0.185for K-in diameter probes; 0.269 forH-in; 0.372 for I-in; and 0.552 for 1-in);D = probe diameter (inches); L =unsupported probe length in inches,which is calculated as the sum of thepipe I.D. plus the pipe wall thickness

plus: 1.25 in for K-in diameter probes;1.5 in for H-in; 1.56 in for I-in; and1.94 in for 1-in diameter probes.

Once the velocity limits have beencalculated, make sure that they donot fall within the range of operating

velocities. If they do, change theprobe diameter, or its mounting, ordo both, until there is no overlap.

Variable Area Flowmeters Variable area flowmeters (Figure 2-15)are simple and versatile devices thatoperate at a relatively constant pres-sure drop and measure the flow of liq-uids, gases, and steam. The position oftheir float, piston or vane is changedas the increasing flow rate opens alarger flow area to pass the flowingfluid. The position of the float, pistonor vane provides a direct visual indica-tion of flow rate. Design variationsinclude the rotameter (a float in atapered tube), orifice/rotametercombination (bypass rotameter),open-channel variable gate, taperedplug, and vane or piston designs.

Either the force of gravity or a

spring is used to return the flow ele-ment to its resting position when theflow lessens. Gravity-operated meters(rotameters) must be installed in a ver-tical position, whereas spring operatedones can be mounted in any position.

All variable area flowmeters are avail-able with local indicators. Most canalso be provided with position sensorsand transmitters (pneumatic, electronic,digital, or fiberoptic) for connecting toremote displays or controls.

• Purge-Flow RegulatorsIf a needle valve is placed at theinlet or outlet of a rotameter, and ad/p regulator controls the pressuredifference across this combination,the result is a purge-flow regulator.Such instrumentation packages areused as self-contained purgeflowmeters (Figure 2-16). These areamong the least expensive and mostwidely used flowmeters. Their mainapplication is to control small gas orliquid purge streams. They are usedto protect instruments from con-tacting hot and corrosive fluids, to



Figure 2-15: A Number of Variable Area Flowmeter Designs

Scale

Equilibrium

Float

Gravity

Flow

Tapered Tube (Rotameter)

Piston in Perforated

CylinderFlexing Vane,

Disc, or FlapperTapered Metering Tube Tapered Plug

10

20

30

40

50

60

70

80

90

100

R

protect pressure taps from plugging,to protect the cleanliness of opticaldevices, and to protect electricaldevices from igniting upon contactwith combustibles.

Purge meters are quite useful inadding nitrogen gas to the vapor

spaces of tanks and other equip-ment. Purging with nitrogen gasreduces the possibility of developinga flammable mixture because it dis-places flammable gases. The purge-flow regulator is reliable, intrinsicallysafe, and inexpensive.

As shown in Figure 2-16, purgemeters can operate in the constantflow mode, where P2 - P0 is held con-stant at about 60 to 80 in H2Odifferential. In bubbler and purgeapplications, the inlet pressure (P1) isheld constant and the outlet pres-sure (P0) is variable. Figure 2-16describes a configuration where theoutlet pressure (P0) is held constant

and the inlet pressure (P1) is variable.They can handle extremely small

flow rates from 0.01 cc/min for liq-uids and from 0.5 cc/min for gases.The most common size is a glasstube rotameter with G-in (6 mm)connections, a range of 0.05-0.5 gpm

(0.2-2.0 lpm) on water or 0.2-2.0 scfm(0.3-3.0 cmph) in air service. Typicalaccuracy is ±5% FS over a 10:1 range,and the most common pressure rat-ing is 150 psig (1 MPa).

• RotametersThe rotameter is the most widelyused variable area flowmeterbecause of its low cost, simplicity,low pressure drop, relatively widerangeability, and linear output. Itsoperation is simple: in order to passthrough the tapered tube, the fluidflow raises the float. The greater theflow, the higher the float is lifted. Inliquid service, the float rises due to a

combination of the buoyancy of theliquid and the velocity head of thefluid. With gases, buoyancy is negligi-ble, and the float responds mostly tothe velocity head.

In a rotameter (Figure 2-15), themetering tube is mounted vertically,with the small end at the bottom. Thefluid to be measured enters at thebottom of the tube, passes upwardaround the float, and exits the top.When no flow exists, the float rests atthe bottom. When fluid enters, themetering float begins to rise.

The float moves up and down inproportion to the fluid flow rate andthe annular area between the floatand the tube wall. As the float rises,the size of the annular openingincreases. As this area increases, thedifferential pressure across the floatdecreases. The float reaches a stableposition when the upward forceexerted by the flowing fluid equalsthe weight of the float. Every floatposition corresponds to a particularflowrate for a particular fluid’s densi-ty and viscosity. For this reason, it isnecessary to size the rotameter foreach application. When sized cor-rectly, the flow rate can be deter-mined by matching the float positionto a calibrated scale on the outsideof the rotameter. Many rotameterscome with a built-in valve for adjust-ing flow manually.

Several shapes of float are avail-able for various applications. Oneearly design had slots, which causedthe float to spin for stabilizing andcentering purposes. Because thisfloat rotated, the term rotameterwas coined.

Rotameters are typically providedwith calibration data and a directreading scale for air or water (orboth). To size a rotameter for otherservice, one must first convert the



Figure 2-16: Purge Flowmeter Design

Flow at P0 Outlet Pressure

Flow at P1 Inlet Pressure

Spring #1

Spring #2

Regulator Valve Flow Control

Valve (V)

Tube

Float

Diaphragm

P2

actual flow to a standard flow. For liq-uids, this standard flow is the waterequivalent in gpm; for gases, the stan-dard flow is the air flow equivalent instandard cubic feet per minute (scfm).Tables listing standard water equiva-lent gpm and/or air scfm values areprovided by rotameter manufacturers.Manufacturers also often provideslide rules, nomographs, or computersoftware for rotameter sizing.

• Design Variations A wide choice of materials is availablefor floats, packing, O-rings, and endfittings. Rotameter tubes for suchsafe applications as air or water canbe made of glass, whereas if breakagewould create an unsafe condition,they are provided with metal tubes.Glass tubes are most common, beingprecision formed of safety shielded

borosilicate glass. Floats typically aremachined from glass, plastic, metal,or stainless steel for corrosion resis-tance. Other float materials includecarboloy, sapphire, and tantalum. Endfittings are available in metal or plas-

tic. Some fluids attack the glassmetering tube, such as wet steam orhigh-pH water over 194°F (which cansoften glass); caustic soda (which dis-solves glass); and hydrofluoric acid(which etches glass).

Floats have a sharp edge at thepoint where the reading should beobserved on the tube-mountedscale. For improved reading accuracy,a glass-tube rotameter should beinstalled at eye level. The scale canbe calibrated for direct reading of airor water, or can read percentage ofrange. In general, glass tube rotame-ters can measure flows up to about60 gpm water and 200 scfh air.

A correlation rotameter has ascale from which a reading is taken(Figure 2-15). This reading is thencompared to a correlation table for agiven gas or liquid to get the actualflow in engineering units. Correlationcharts are readily available for nitro-gen, oxygen, hydrogen, helium, argon,and carbon dioxide. While not nearlyas convenient as a direct readingdevice, a correlation meter is moreaccurate. This is because a direct-reading device is accurate for onlyone specific gas or liquid at a partic-ular temperature and pressure. A cor-relation flowmeter can be used witha wide variety of fluids and gasesunder various conditions. In the sametube, different flow rates can be han-dled by using different floats.

Small glass tube rotameters are suit-able for working with pressures up to500 psig, but the maximum operatingpressure of a large (2-in diameter) tubemay be as low as 100 psig. The practi-cal temperature limit is about 400°F,but such high-temperature operationsubstantially reduces the operatingpressure of the tube. In general, thereis a linear relationship between oper-ating temperature and pressure.

Glass-tube rotameters are oftenused in applications where severalstreams of gases or liquids are beingmetered at the same time or mixed ina manifold, or where a single fluid isbeing exhausted through severalchannels (Figure 2-17). Multiple tubeflowmeters allow up to six rotametersto be mounted in the same frame.

It also is possible to operate a

rotameter in a vacuum. If therotameter has a valve, it must beplaced at the outlet at the top of themeter. For applications requiring awide measurement range, a dual-ballrotameter can be used. This instru-ment has two ball floats: a light ball(typically black) for indicating lowflows and a heavy ball (usually white)for indicating high flows. The blackball is read until it goes off scale, andthen the white ball is read. One suchinstrument has a black measuringrange from 235-2,350 ml/min and awhite to 5,000 ml/min.

For higher pressures and tempera-tures beyond the practical range ofglass, metal tube rotameters can beused. These tubes are usually madeof stainless steel, and the position ofthe float is detected by magnetic fol-lowers with readouts outside themetering tube.

Metal-tube rotameters can be



Figure 2-17: Multi-Tube Rotameter Station

Rotameters can be specified in a wide range of

sizes and materials.

used for hot and strong alkalis, fluo-rine, hydrofluoric acid, hot water,steam, slurries, sour gas, additives,and molten metals. They also can beused in applications where highoperating pressures, water hammer,or other forces could damage glasstubes. Metal-tube rotameters areavailable in diameter sizes from K into 4 in, can operate at pressures up to750 psig, temperatures to 540°C(1,000°F), and can measure flows upto 4,000 gpm of water or 1,300 scfmof air. Metal-tube rotameters arereadily available as flow transmittersfor integration with remote analog ordigital controls. Transmitters usuallydetect the float position throughmagnetic coupling and are often pro-vided with external indicationthrough a rotatable magnetic helixthat moves the pointer. The transmit-ter can be intrinsically safe, micro-processor-based, and can be provid-

ed with alarms and a pulse outputfor totalization.

Plastic-tube rotameters are rela-tively low cost rotameters that areideal for applications involving corro-sive fluids or deionized water. Thetube itself can be made from Teflon®PFA, polysulfone, or polyamide. Thewetted parts can be made from stain-less steel, PVDF, or Teflon® PFA, PTFE,PCTFE, with Viton® or Kalrez® O-rings.

• AccuracyLaboratory rotameters can be calibrat-ed to an accuracy of 0.50% AR over a4:1 range, while the inaccuracy ofindustrial rotameters is typically 1-2%FS over a 10:1 range. Purge and bypassrotameter errors are in the 5% range.

Rotameters can be used to manu-ally set flow rates by adjusting thevalve opening while observing thescale to establish the required processflow rate. If operating conditions

remain unaltered, rotameters can berepeatable to within 0.25% of theactual flow rate.

Most rotameters are relativelyinsensitive to viscosity variations.The most sensitive are very smallrotameters with ball floats, whilelarger rotameters are less sensitiveto viscosity effects. The limitationsof each design are published by themanufacturer (Figure 2-18). The floatshape does affect the viscositylimit. If the viscosity limit is exceed-ed, the indicated flow must be cor-rected for viscosity.

Because the float is sensitive tochanges in fluid density, a rotametercan be furnished with two floats (onesensitive to density, the other tovelocity) and used to approximatethe mass flow rate. The more closelythe float density matches the fluiddensity, the greater the effect of afluid density change will be on thefloat position. Mass-flow rotameterswork best with low viscosity fluidssuch as raw sugar juice, gasoline, jetfuel, and light hydrocarbons.

Rotameter accuracy is not affect-ed by the upstream piping configura-tion. The meter also can be installeddirectly after a pipe elbow withoutadverse effect on metering accuracy.Rotameters are inherently self clean-ing because, as the fluid flowsbetween the tube wall and the float,it produces a scouring action thattends to prevent the buildup of for-eign matter. Nevertheless, rotame-ters should be used only on cleanfluids which do not coat the float orthe tube. Liquids with fibrous materi-als, abrasives, and large particlesshould also be avoided.

• Other Variable-Area FlowmetersMajor disadvantages of the rotameterare its relatively high cost in larger



Figure 2-18: Rotameter Maximum Velocity

Water Equivalent Flow (GPM)

11

5

10

50

5 10 50 100

Visc

osit

y of

Met

ered

Liq

uid,

Cen

tist

okes

Xrm rc

rm = Density of metered liquid rc = Density of calibrating liquid (water)

sizes and the requirement that it beinstalled vertically (there may not beenough head room). The cost of alarge rotameter installation can bereduced by using an orifice bypass ora pitot tube in combination with asmaller rotameter. The same-sizebypass rotameter can be used tomeasure a variety of flows, with theonly difference between applicationsbeing the orifice plate and the differ-ential it produces.

Advantages of a bypass rotameterinclude low cost; its major disadvan-tage is inaccuracy and sensitivity tomaterial build-up. Bypass rotametersare often provided with isolationvalves so that they can be removedfor maintenance without shuttingdown the process line.

Tapered plug flowmeters are vari-able-area flowmeters with a station-ary core and a piston that moves asthe flow varies. In one design, thepiston movement mechanicallymoves a pointer, while in another itmagnetically moves an externalflow rate indicator. The seconddesign has a metallic meter body forapplications up to 1,000 psig.

One gate-type variable-areaflow-meter resembles a butterflyvalve. Flow through the meterforces a spring-loaded vane torotate, and a mechanical connec-tion provides local flow rate indica-

tion. The inaccuracy of such metersis 2-5% FS. The meter can be usedon oil, water and air, and is availablein sizes up to 4 inches. It also is usedas an indicating flow switch in safe-ty interlock systems. T





Edition, OMEGA Press, 1995.• “Choices Abound in Flow Measurement”, D. Ginesi, Chemical Engineering,

April 1991.• “Developments in DP Flowmeters,” Jesse Yoder, Control, April 1998.• Differential Producers - Orifice, Nozzle, Venturi, ANSI/ASME MFC,

December 1983.• Flow Measurement Engineers’ Handbook, R.W. Miller, McGraw-Hill, 1996.• Flow Measurement, D.W. Spitzer, Instrument Society of America, 1991.• Flow of Water Through Orifices, AGA/ASME, Ohio State Univ. Bulletin

89, Vol. IV, No. 3.• Fluid Meters, H.S. Bean , American Society of Mechanical Engineers, 1971.• Fundamentals of Flow Measurement, J. P. DeCarlo, Instrument Society of

America, 1984.• Instrument Engineers Handbook, 3rd edition, Bela Liptak, CRC Press, 1995.• “Orifice Metering of Natural Gas”, AGA Report 3, 1985.• “Primary Element Solves Difficult Flow Metering Problems at Water

Waste Treatment Plant,” D. Ginesi, L. Keefe, and P. Miller, Proceedings ofISA 1989, Instrument Society of America, 1989.

Discussed in this chapter arevarious types of mechanicalflowmeters that measureflow using an arrangement

of moving parts, either by passingisolated, known volumes of a fluidthrough a series of gears or chambers(positive displacement, or PD) or bymeans of a spinning turbine or rotor.

All positive displacement flowme-ters operate by isolating and count-ing known volumes of a fluid (gas orliquid) while feeding it through themeter. By counting the number ofpassed isolated volumes, a flowmeasurement is obtained. Each PDdesign uses a different means of iso-lating and counting these volumes.The frequency of the resulting pulsetrain is a measure of flow rate, whilethe total number of pulses gives thesize of the batch. While PD metersare operated by the kinetic energyof the flowing fluid, meteringpumps (described only briefly in thisarticle) determine the flow rate

while also adding kinetic energy tothe fluid.

The turbine flowmeter consists of amulti-bladed rotor mounted at right

angles to the flow, suspended in thefluid stream on a free-running bearing.The diameter of the rotor is very closeto the inside diameter of the meteringchamber, and its speed of rotation isproportional to the volumetric flowrate. Turbine rotation can be detectedby solid state devices or by mechani-cal sensors. Other types of rotary ele-ment flowmeters include the pro-peller (impeller), shunt, and paddle-wheel designs.

Positive Displacement FlowmetersPositive displacement meters providehigh accuracy (±0.1% of actual flowrate in some cases) and good repeata-bility (as high as 0.05% of reading).Accuracy is not affected by pulsatingflow unless it entrains air or gas in thefluid. PD meters do not require apower supply for their operation anddo not require straight upstream anddownstream pipe runs for their instal-lation. PD meters are available in sizesfrom G in to 12 in and can operate

with turndowns as high as 100:1,although ranges of 15:1 or lower aremuch more common. Slippagebetween the flowmeter components

is reduced and metering accuracy istherefore increased as the viscosity ofthe process fluid increases.

The process fluid must be clean.Particles greater than 100 microns insize must be removed by filtering. PDmeters operate with small clearancesbetween their precision-machinedparts; wear rapidly destroys theiraccuracy. For this reason, PD metersare generally not recommended formeasuring slurries or abrasive fluids.In clean fluid services, however, theirprecision and wide rangeability makethem ideal for custody transfer andbatch charging. They are most widelyused as household water meters.Millions of such units are producedannually at a unit cost of less than$50 U.S. In industrial and petrochem-ical applications, PD meters are com-monly used for batch charging ofboth liquids and gases.

Although slippage through the PDmeter decreases (that is, accuracyincreases) as fluid viscosity increases,

pressure drop through the meter alsorises. Consequently, the maximum(and minimum) flow capacity of theflowmeter is decreased as viscosity


Positive Displacement Flowmeters

Turbine Flowmeters

Other Rotary Flowmeters

FLOW & LEVEL MEASUREMENTMechanical Flowmeters

3

DMechanical Flowmeters

Figure 3-1: Positive Displacement Flowmeter Designs

B) Rotating ValveA) Nutating Disc

Inlet OutletInlet Outlet

Housing

Vane

Vane Slot

Rotor

Disc

Ball

increases. The higher the viscosity,the less slippage and the lower themeasurable flow rate becomes. Asviscosity decreases, the low flow

performance of the meter deterio-rates. The maximum allowable pres-sure drop across the meter con-strains the maximum operating flowin high viscosity services.

• Liquid PD MetersNutating disc meters are the mostcommon PD meters. They are used asresidential water meters around theworld. As water flows through themetering chamber, it causes a disc towobble (nutate), turning a spindle,which rotates a magnet. This magnetis coupled to a mechanical registeror a pulse transmitter. Because theflowmeter entraps a fixed quantityof fluid each time the spindle isrotated, the rate of flow is propor-tional to the rotational velocity ofthe spindle (Figure 3-1A).

Because it must be nonmagnetic,the meter housing is usually made ofbronze but can be made from plasticfor corrosion resistance or cost

savings. The wetted parts such as thedisc and spindle are usually bronze,rubber, aluminum, neoprene, Buna-N,or a fluoroelastomer such as Viton®.Nutating disc meters are designedfor water service and the materials ofwhich they are made must bechecked for compatibility with otherfluids. Meters with rubber discs givebetter accuracy than metal discs dueto the better sealing they provide.

Nutating disc meters are availablein L-in to 2-in sizes. They are suitedfor 150-psig operating pressures withoverpressure to a maximum of 300psig. Cold water service units aretemperature-limited to 120°F. Hotwater units are available up to 250°F.

These meters must meet AmericanWater Works Association (AWWA)standards for accuracy. The accuracy

of these meters is required to be±2% of actual flow rate. Higher vis-cosity can produce higher accuracy,while lower viscosity and wear over

time will reduce accuracy. The AWWArequires that residential water metersbe re-calibrated every 10 years.Because of the intermittent usepatterns of residential users, this cor-responds to recalibrating L x I inresidential water meters after theyhave metered 5 million gallons. Inindustrial applications, however, thesemeters are likely to pass this thresholdmuch sooner. The maximum continu-ous flow of a nutating disc meter isusually about 60-80% of the maxi-mum flow in intermittent service.

Rotating vane meters (Figure 3-1B)have spring-loaded vanes that entrapincrements of liquid between theeccentrically mounted rotor and thecasing. The rotation of the vanesmoves the flow increment from inletto outlet and discharge. Accuracy of

3 Mechanical Flowmeters


Figure 3-2: Piston Meter Designs

B) Single-Piston ReciprocatingA) Oscillating

Cylindrical Abutment

Measuring Chamber

Control Roller

PistonPiston Hub

Outlet PortPartition Plate

Inlet Port

Follower Magnet

Piston

Magnet Assembly

Housing

Measuring Chamber

CoverControl Roller

Cylindrical Abutment

Slide Valve

Inlet

Piston

±0.1% of actual rate (AR) is normal,and larger size meters on higher vis-cosity services can achieve accuracyto within 0.05% of rate.

Rotating vane meters are regularly

used in the petroleum industry andare capable of metering solids-ladencrude oils at flow rates as high as17,500 gpm. Pressure and temperaturelimits depend on the materials ofconstruction and can be as high as350°F and 1,000 psig. Viscosity limitsare 1 to 25,000 centipoise.

In the rotary displacement meter,a fluted central rotor operates inconstant relationship with two wiperrotors in a six-phase cycle. Its appli-cations and features are similar tothose of the rotary vane meter.

• Piston MetersOscillating piston flowmeters typical-ly are used in viscous fluid servicessuch as oil metering on engine teststands where turndown is not critical(Figure 3-2). These meters also can beused on residential water service andcan pass limited quantities of dirt,such as pipe scale and fine (viz,-200mesh or -74 micron) sand, but not

large particle size or abrasive solids. The measurement chamber is

cylindrical with a partition plate sep-arating its inlet port from its outlet.The piston is also cylindrical and is

punctured by numerous openings toallow free flow on both sides of thepiston and the post (Figure 3-2A). Thepiston is guided by a control rollerwithin the measuring chamber, andthe motion of the piston is trans-ferred to a follower magnet which isexternal to the flowstream. The fol-lower magnet can be used to driveeither a transmitter, a register, orboth. The motion of the piston isoscillatory (not rotary) since it is con-strained to move in one plane. Therate of flow is proportional to therate of oscillation of the piston.

The internals of this flowmeter canbe removed without disconnection ofthe meter from the pipeline. Becauseof the close tolerances required toseal the piston and to reduce slippage,these meters require regular mainte-nance. Oscillating piston flow metersare available in H-in to 3-in sizes, andcan generally be used between 100and 150 psig. Some industrial versions

are rated to 1,500 psig. They can meterflow rates from 1 gpm to 65 gpm incontinuous service with intermittentexcursions to 100 gpm. Meters aresized so that pressure drop is below

35 psid at maximum flow rate.Accuracy ranges from ±0.5 % AR forviscous fluids to ±2% AR for nonvis-cous applications. Upper limit onviscosity is 10,000 centipoise.

Reciprocating piston meters areprobably the oldest PD meter designs.They are available with multiple pis-tons, double-acting pistons, or rotarypistons. As in a reciprocating pistonengine, fluid is drawn into one pistonchamber as it is discharged from theopposed piston in the meter.Typically, either a crankshaft or a hor-izontal slide is used to control theopening and closing of the proper ori-fices in the meter. These meters areusually smaller (available in sizesdown to 1/10-in diameter) and areused for measuring very low flows ofviscous liquids.

• Gear & Lobe MetersThe oval gear PD meter uses twofine-toothed gears, one mounted

Mechanical Flowmeters 3


Figure 3-3: Rotating Positive Displacement Meters

A) Oval-Gear B) Rotating Lobe C) Rotating Impeller

B

B

A

A

A

horizontally, the other vertically,with gears meshing at the tip of thevertical gear and the center of thehorizontal gear (Figure 3-3A). The tworotors rotate opposite to each other,creating an entrapment in the cres-cent-shaped gap between the hous-ing and the gear. These meters can bevery accurate if slippage between thehousing and the gears is kept small. Ifthe process fluid viscosity is greaterthan 10 centipoise and the flowrate isabove 20% of rated capacity, accura-cy of 0.1% AR can be obtained. Atlower flows and at lower viscosity,slippage increases and accuracydecreases to 0.5% AR or less.

The lubricating characteristics ofthe process fluid also affect the turn-down of an oval gear meter. With liq-uids that do not lubricate well, maxi-mum rotor speed must be derated tolimit wear. Another way to limit wearis to keep the pressure drop acrossthe meter below 15 psid. Therefore,the pressure drop across the meterlimits the allowable maximum flowin high viscosity service.

Rotating lobe and impeller typePD meters are variations of the ovalgear flowmeter that do not share itsprecise gearing. In the rotating lobedesign, two impellers rotate in oppo-site directions within the ovoidhousing (Figure 3-3B). As they rotate,a fixed volume of liquid is entrappedand then transported toward theoutlet. Because the lobe gearsremain in a fixed relative position, itis only necessary to measure therotational velocity of one of them.The impeller is either geared to a reg-ister or is magnetically coupled to atransmitter. Lobe meters can be fur-nished in 2-in to 24-in line sizes. Flowcapacity is 8-10 gpm to 18,000 gpm inthe larger sizes. They provide goodrepeatability (better than 0.015% AR)

at high flows and can be used at highoperating pressures (to 1,200 psig)and temperatures (to 400°F).

The lobe gear meter is available ina wide range of materials of con-struction, from thermoplastics tohighly corrosion-resistant metals.Disadvantages of this design include aloss of accuracy at low flows. Also,the maximum flow through this meteris less than for the same size oscillato-ry piston or nutating disc meter.

In the rotating impeller meter,very coarse gears entrap the fluidand pass a fixed volume of fluidwith each rotation (Figure 3-3C).These meters are accurate to 0.5%of rate if the viscosity of theprocess fluid is both high and con-stant, or varies only within a narrowband. These meters can be madeout of a variety of metals, includingstainless steel, and corrosion-resis-tant plastics such as PVDF (Kynar).These meters are used to meterpaints and, because they are avail-able in 3A or sanitary designs, also

milk, juices, and chocolate. In these units, the passage of mag-

nets embedded in the lobes of the

rotating impellers is sensed by prox-imity switches (usually Hall-effectdetectors) mounted external to theflow chamber. The sensor transmits apulse train to a counter or flow con-troller. These meters are available in1/10-in to 6-in sizes and can handlepressures to 3,000 psig and tempera-tures to 400°F.

• Helix MetersThe helix meter is a positive dis-placement device that uses two radi-ally pitched helical gears to continu-ously entrap the process fluid as itflows. The flow forces the helicalgears to rotate in the plane of thepipeline. Optical or magnetic sensorsare used to encode a pulse train pro-portional to the rotational speed ofthe helical gears. The forces requiredto make the helices rotate are rela-tively small and therefore, in com-parison to other PD meters, thepressure drop is relatively low. Thebest attainable accuracy is about±0.2% or rate.

As shown in Figure 3-4, measure-ment error rises as either the operat-ing flowrate or the viscosity of the



Figure 3-4: Effect of Viscosity on Low-Flow Accuracy

0.1 1.0 10 100

Maximum Rated Flow, %

Erro

r

+10

+1.0

0.1

-1.0

-10

>1000cP300cP

100cP

30cP 10cP 3cP

process fluid drops. Helical gearmeters can measure the flow of highlyviscous fluids (from 3 to 300,000 cP),making them ideal for extremely

thick fluids such as glues and veryviscous polymers. Because at maxi-mum flow the pressure drop throughthe meter should not exceed 30 psid,the maximum rated flow through themeter is reduced as the fluid viscosi-ty increases. If the process fluid hasgood lubricating characteristics, themeter turndown can be as high as100:1, but lower (10:1) turndowns aremore typical.

• Metering PumpsMetering pumps are PD meters thatalso impart kinetic energy to theprocess fluid. There are three basicdesigns: peristaltic, piston, anddiaphragm.

Peristaltic pumps operate by havingfingers or a cam systematically squeezea plastic tubing against the housing,

which also serves to position the tub-ing. This type of metering pump is usedin laboratories, in a variety of medicalapplications, in the majority of envi-

ronmental sampling systems, and alsoin dispensing hypochlorite solutions.The tubing can be silicone-rubber or, ifa more corrosion-resistant material isdesired, PTFE tubing.

Piston pumps deliver a fixed vol-ume of liquid with each “out” strokeand a fixed volume enters the cham-ber on each “in” stroke (Figure 3-5A).Check valves keep the fluid flowfrom reversing. As with all positivedisplacement pumps, piston pumpsgenerate a pulsating flow. To mini-mize the pulsation, multiple pistonsor pulsation-dampening reservoirsare installed. Because of the closetolerances of the piston and cylindersleeve, a flushing mechanism must beprovided in abrasive applications.Piston pumps are sized on the basisof the displacement of the piston

and the required flow rate and dis-charge pressure. Check valves (or, oncritical applications, double checkvalves) are selected to protect

against backflow.Diaphragm pumps are the most

common industrial PD pumps (Figure3-5B). A typical configuration consistsof a single diaphragm, a chamber, andsuction and discharge check valvesto prevent backflow. The piston caneither be directly coupled to thediaphragm or can force a hydraulicoil to drive the diaphragm. Maximumoutput pressure is about 125 psig.Variations include bellows-typediaphragms, hydraulically actuateddouble diaphragms, and air-operat-ed, reciprocating double-diaphragms.

• Gas PD MetersPD gas meters operate by countingthe number of entrapped volumesof gas passed, similar to the way PDmeters operate on liquids. The



Figure 3-5: Metering Pump Designs

B) DiaphragmA) Piston

Discharge

Inlet Suction

Discharge

Suction Check Valve

Packing Gland (Stuffing Box)

Hydraulic Oil Inlet Valve

Adjustable Outlet Valve

Piston

Piston

Support Plates

Diaphragm

primary difference is that gases arecompressible.

Diaphragm gas meters most oftenare used to measure the flow of nat-ural gas, especially in metering con-sumption by households. The meteris constructed from aluminum cast-ings with cloth-backed rubberdiaphragms. The meter consists offour chambers: the two diaphragmchambers on the inlet and outletsides and the inlet and outlet cham-bers of the meter body. The passageof gas through the meter creates adifferential pressure between the twodiaphragm chambers by compressingthe one on the inlet side and expand-ing the one on the outlet side. Thisaction alternately empties and fillsthe four chambers. The slide valves atthe top of the meter alternate theroles of the chambers and synchro-nize the action of the diaphragms, as

well as operating the crank mecha-nism for the meter register.

Diaphragm meters generally are

calibrated for natural gas, which has aspecific gravity of 0.6 (relative to air).Therefore, it is necessary to re-cali-brate the flow rating of the meterwhen it is used to meter other gases.The calibration for the new flow rat-ing (QN) is obtained by multiplyingthe meter’s flow rating for natural gas(QC) by the square root of the ratioof the specific gravities of natural gas(0.6) and the new gas (SGN):

QN= QC(0.6/SGN)0.5

Diaphragm meters are usually ratedin units of cubic feet per hour andsized for a pressure drop of 0.5-2 inH2O. Accuracy is roughly ±1% of read-ing over a 200:1 range. They maintaintheir accuracy for long periods oftime, which makes them good choicesfor retail revenue metering applica-tions. Unless the gas is unusually dirty

(producer gas, or recycled methanefrom composting or digesting, forexample), the diaphragm meter will

operate with little or no maintenanceindefinitely.

Lobe gear meters (or lobedimpeller meters, as they are alsoknown), also are used for gas service.Accuracy in gas service is ±1% of rateover a 10:1 turndown, and typicalpressure drop is 0.1 psid. Because ofthe close tolerances, upstream filtra-tion is required for dirty lines.

Rotating vane meters measure theflow of gas in the same ranges as dolobe gear meters (up to 100,000ft3/hr) but can be used over a wider25:1 turndown. They also incur a lowerpressure drop of 0.05 in H2O for sim-ilar accuracy, and, because the clear-ances are somewhat more forgiving,upstream filtration is not as critical.

• High-Precision PD SystemsHigh-precision gas meters are usuallya hybrid combining a standard PD

meter and a motor drive that elimi-nates the pressure drop across themeter. Equalizing the inlet and outlet



Figure 3-6: High-Precision PD Meters Equalize Inlet and Outlet Pressures

B) Liquid ServiceA) Gas Service

Displacement Flowmeter

Displacement Flowmeter

Gas Flow

Low Sensitivity

Leaf

High Sensitivity

Leaf

Zeroing Solenoids

Displacement Transducers

PDC

M

PDC

DC Motor

Differential Pressure

Detection Piston

M

pressures eliminates slip flows, leak-age, and blow-by. In high-precisiongas flowmeter installations, high-sensitivity leaves are used to detectthe pressure differential, and dis-placement transducers are used tomeasure the deflection of the leaves(Figure 3-6A). Designed to operate at

ambient temperatures and at up to30 psig pressures, this meter isclaimed to provide accuracy to with-in 0.25% of reading over a 50:1 rangeand 0.5% over a 100:1 range. Flowcapacity ranges from 0.3-1,500 scfm.

For liquid service, a servomotor-driven oval-gear meter equalizes thepressure across the meter. Thisincreases accuracy at low flows andunder varying viscosity conditions(Figure 3-6B). This flowmeter uses avery sensitive piston to detect themeter differential and drives a vari-able speed servomotor to keep itnear zero. This design is claimed toprovide 0.25% of rate accuracy over a50:1 range at operating pressures ofup to 150 psig. High precisionflowmeters are used on engine teststands for fuel flow measurement(gasoline, diesel, alcohol, etc.). Flowranges from 0.04-40 gph are typical.

Vapor separators are usually includ-ed, to prevent vapor lock.

• Testing, Calibration & ProversAll meters with moving parts requireperiodic testing, recalibration andrepair, because wear increases theclearances. Recalibration can be

done either in a laboratory or on lineusing a prover.

Gas systems are recalibratedagainst a bell-jar prover—a calibratedcylindrical bell, liquid sealed in a tank.As the bell is lowered, it discharges aknown volume of gas through themeter being tested. The volumetricaccuracy of bell-jar provers is on theorder of 0.1% by volume, and proversare available in discharge volumes of2, 5, 10 ft3 and larger.

Liquid systems can be calibrated inthe laboratory against either a cali-brated secondary standard or a gravi-metric flow loop. This approach canprovide high accuracy (up to ±0.01%of rate) but requires removing theflowmeter from service.

In many operations, especially inthe petroleum industry, it is difficultor impossible to remove a flow-meter from service for calibration.

Therefore, field-mounted and in-lineprovers have been developed. Thistype of prover consists of a calibrat-ed chamber equipped with a barrierpiston (Figure 3-7). Two detectors aremounted a known distance (andtherefore a known volume) apart. Asthe flow passes through the cham-ber, the displacer piston is moveddownstream. Dividing the volume ofthe chamber by the time it takes forthe displacer to move from onedetector to the other gives the cali-brated flow rate. This rate is thencompared to the reading of theflowmeter under test.

Provers are repeatable on theorder of 0.02%, and can operate atup to 3,000 psig and 165°F/75°C. Theiroperating flow range is from as lowas 0.001 gpm to as high as 20,000gpm. Provers are available for bench-top use, for mounting in truck-beds,on trailers, or in-line.

• PD Meter AccessoriesPD meter accessories include strain-ers, filters, air/vapor release assem-blies, pulsation dampeners, tempera-ture compensation systems, and avariety of valves to permit dribblecut-off in batching systems.Mechanical registers can beequipped with mechanical or elec-tronic ticket-printers for inventorycontrol and point-of-use sales.Batching flow computers are readilyavailable, as are analog and intelli-gent digital transmitters. Automaticmeter reading (AMR) devices permitthe remote retrieval of readings byutility personnel.

Turbine FlowmetersInvented by Reinhard Woltman in the18th century, the turbine flowmeteris an accurate and reliable flowmeterfor both liquids and gases. It consists



Figure 3-7: Field-Mounted, In-Line Flow Prover

Calibrated Volume

1st Detector 2nd DetectorFlow Tube

Flow

Disp

lace

r

of a multi-bladed rotor mounted atright angles to the flow and suspend-ed in the fluid stream on a free-run-ning bearing. The diameter of therotor is very slightly less than theinside diameter of the meteringchamber, and its speed of rotation isproportional to the volumetric flowrate. Turbine rotation can be detect-ed by solid state devices (reluctance,inductance, capacitive and Hall-effect pick-ups) or by mechanicalsensors (gear or magnetic drives).

In the reluctance pick-up, the coilis a permanent magnet and the tur-bine blades are made of a materialattracted to magnets. As each bladepasses the coil, a voltage is generated

in the coil (Figure 3-8A). Each pulserepresents a discrete volume of liq-uid. The number of pulses per unitvolume is called the meter’s K-factor.

In the inductance pick-up, thepermanent magnet is embedded inthe rotor, or the blades of the rotor

are made of permanently magnetizedmaterial (Figure 3-8B). As each bladepasses the coil, it generates a voltagepulse. In some designs, only one bladeis magnetic and the pulse represents acomplete revolution of the rotor.

The outputs of reluctance andinductive pick-up coils are continu-ous sine waves with the pulse train’sfrequency proportional to the flowrate. At low flow, the output (theheight of the voltage pulse) may beon the order of 20 mV peak-to-peak.It is not advisable to transport such aweak signal over long distances.Therefore, the distance between thepickup and associated display elec-tronics or preamplifier must be short.

Capacitive sensors produce a sinewave by generating an RF signal thatis amplitude-modulated by themovement of the rotor blades.Instead of pick-up coils, Hall-effecttransistors also can be used. Thesetransistors change their state when

they are in the presence of a very lowstrength (on the order of 25 gauss)magnetic field.

In these turbine flowmeters, verysmall magnets are embedded in thetips of the rotor blades. Rotors are typ-ically made of a non-magnetic materi-al, like polypropylene, Ryton, or PVDF(Kynar). The signal output from a Hall-effect sensor is a square wave pulsetrain, at a frequency proportional tothe volumetric flowrate.

Because Hall-effect sensors have nomagnetic drag, they can operate atlower flow velocities (0.2 ft/sec) thanmagnetic pick-up designs (0.5-1.0ft/sec). In addition, the Hall-effect sen-sor provides a signal of high amplitude

(typically a 10.8-V square wave), per-mitting distances up to 3,000 ft.between the sensor and the electron-ics without amplification.

In the water distribution industry,mechanical-drive Woltman-type tur-bine flowmeters continue to be the



Figure 3-8: Generation of Turbine Flow Signal

B)A)

Reluctance Pickup Coil

Meter Body

Coil

Cone

Permanent Magnet

One Pulse Per Blade

One Unit Volume

Inductance Pickup Coil

Meter Body

Coil

Permanent Magnet

RotorBlade

Per Revolution

N

S

standard. These turbine meters use agear train to convert the rotation ofthe rotor into the rotation of a verti-cal shaft. The shaft passes betweenthe metering tube and the registersection through a mechanical stuff-

ing box, turning a geared mechanicalregister assembly to indicate flowrate and actuate a mechanical total-izer counter.

More recently, the water distribu-tion industry has adopted a magnet-ic drive as an improvement over highmaintenance mechanical-drive tur-bine meters. This type of meter has asealing disc between the measuringchamber and the register. On themeasuring chamber side, the verticalshaft turns a magnet instead of agear. On the register side, an oppos-ing magnet is mounted to turn thegear. This permits a completelysealed register to be used with amechanical drive mechanism.

In the United States, the AWWAsets the standards for turbineflowmeters used in water distribu-tion systems. Standard C701 pro-vides for two classes (Class I andClass II) of turbine flowmeters. Class I

turbine meters must registerbetween 98-102% of actual rate atmaximum flow when tested. Class IIturbine meters must registerbetween 98.5-101.5% of actual rate.Both Class I and Class II meters must

have mechanical registers.Solid state pickup designs are less

susceptible to mechanical wear thanAWWA Class I and Class II meters.

• Design & Construction VariationsMost industrial turbine flowmetersare manufactured from austeniticstainless steel (301, 303, 304SS),

whereas turbine meters intended formunicipal water service are bronze orcast iron. The rotor and bearingmaterials are selected to match theprocess fluid and the service. Rotorsare often made from stainless steel,and bearings of graphite, tungstencarbide, ceramics, or in special casesof synthetic ruby or sapphire com-bined with tungsten carbide. In allcases, bearings and shafts aredesigned to provide minimum fric-tion and maximum resistance towear. Some corrosion-resistantdesigns are made from plastic mate-rials such as PVC.

Small turbine meters often arecalled barstock turbines because insizes of I in to 3 in. they aremachined from stainless steel hexag-onal barstock. The turbine is sus-pended by a bearing between twohanger assemblies that also serve tocondition the flow. This design issuited for high operating pressures(up to 5,000 psig).

Similar to a pitot tube differentialpressure flowmeter, the insertion tur-bine meter is a point-velocity device.It is designed to be inserted intoeither a liquid or a gas line to a depthat which the small-diameter rotor willread the average velocity in the line.



Figure 3-9: Typical Turbine Flowmeter Calibration Curve

100

99

98

97

96

0 100 200 300 500400 600 700 800

A B

Max

imum

Lin

ear F

low

Rat

e

Met

er C

oeff

icie

nt K

- Pu

lses/

Gal.

Calibration Curve

Minimum Flow Rate for ±0.25% Linearity

Flow Rate - Gal./Min.

±0.15% Linearity Flow Rate

Nominal K Factor 98.50 +0.25%

-0.25%

This innovative turbine meter trades out a transmitted signal for local LCD indication.

Because they are very sensitive to thevelocity profile of the flowing stream,they must be profiled at severalpoints across the flow path.

Insertion turbine meters can bedesigned for gas applications (small,lightweight rotor) or for liquid (largerrotor, water-lubricated bearings).They are often used in large diame-ter pipelines where it would be cost-prohibitive to install a full size meter.They can be hot-tapped into existing

pipelines (6 in or larger) through avalving system without shuttingdown the process. Typical accuracyof an insertion turbine meter is 1% FS,and the minimum flow velocity isabout 0.2 ft/sec.

• Turbine Meter AccuracyFigure 3-9 shows a typical turbine-meter calibration curve describingthe relationship between flow andK-factor (pulses/gallon). The accu-racy of turbine meters is typicallygiven in percentage of actual rate (%AR). This particular meter has a lin-earity tolerance band of ±0.25%over a 10:1 flow range and a ±0.15%linearity in a 6:1 range. The repeata-

bility is from ±0.2% to ±0.02% overthe linear range.

Because there are minor inconsis-tencies in the manufacturingprocess, all turbine flowmeters arecalibrated prior to shipment. Theresulting K-factor in pulses per vol-ume unit will vary within the statedlinearity specification. It is possible,however, to register several K-factorsfor different portions of the flowrange and to electronically switch

from one to the other as the mea-sured flow changes. Naturally, the K-factor is applicable only to the fluidfor which the meter was calibrated.

Barstock turbine meters typicallyare linear to ±0.25% AR over a 10:1flow range. The linearity of largermeters is ±0.5% AR over a 10:1 flowrange. Turbine meters have a typicalnonlinearity (the turbine meterhump, shown in Figure 3-9) in thelower 25-30% of their range. Keepingthe minimum flow reading above thisregion will permit linearity to within0.15% on small and 0.25% on largerturbine meters. If the range of 10:1 isinsufficient, some turbine flow-meters can provide up to 100:1 turn-

downs if accuracy is de-rated to 1%of full scale (FS).

• Sizing & SelectionTurbine meters should be sized sothat the expected average flow isbetween 60% and 75% of the maxi-mum capacity of the meter. If the pipeis oversized (with flow velocity under1 ft/sec), one should select a Hall-effect pick-up and use a meter small-er than the line size. Flow velocities

under 1 ft/sec can be insufficient,while velocities in excess of 10 ft/seccan result in excessive wear. Most tur-bine meters are designed for maxi-mum velocities of 30 ft/sec.

Turbine flowmeters should besized for between 3 and 5 psid pres-sure drop at maximum flow. Becausepressure drop increases with thesquare of flow rate, reducing themeter to the next smaller size willraise the pressure drop considerably.

Viscosity affects the accuracy andlinearity of turbine meters. It is there-fore important to calibrate the meterfor the specific fluid it is intended tomeasure. Repeatability is generally notgreatly affected by changes in viscosity,



Figure 3-10: Flow Straighteners Reduce Straight Pipe Runs

Concentric Cone

Flow Straightener

Alternative Flow Straightening Vanes

D Bore Dia.

Nominal Size D Inches Coil Protection Box Concentric

Cone

Flow

Meter and Straightener Connections

5 X D2.5 D 5 X D10 X D

Bundle of Tubes Element

Radial Vane Element

and turbine meters often are used tocontrol the flow of viscous fluids.Generally, turbine meters perform wellif the Reynolds Number is greater than4,000 and less than or equal to 20,000.

Because it affects viscosity, tempera-ture variation can also adversely affectaccuracy and must be compensatedfor or controlled. The turbine meter’soperating temperature ranges from -200 to 450°C (-328 to 840°F).

Density changes do not greatlyaffect turbine meters. On low densityfluids (SG < 0.7), the minimum flowrate is increased due to the reducedtorque, but the meter’s accuracy usu-ally is not affected.

• Installation & Accessories Turbine meters are sensitive toupstream piping geometry that cancause vortices and swirling flow.Specifications call for 10-15 diametersof straight run upstream and fivediameters of straight run downstreamof the meter. However, the presenceof any of the following obstructionsupstream would necessitate thatthere be more than 15 diameters of

upstream straight-pipe runs • 20 diameters for 90° elbow, tee,

filter, strainer, or thermowell;• 25 diameters for a partially open

valve; and

• 50 or more diameters if there aretwo elbows in different planes orif the flow is spiraling orcorkscrewing. In order to reduce this straight-

run requirement, straightening vanesare installed. Tube bundles or radialvane elements are used as externalflow straighteners located at least 5diameters upstream of the meter(Figure 3-10).

Under certain conditions, the pres-sure drop across the turbine can causeflashing or cavitation. The first causesthe meter to read high, the secondresults in rotor damage. In order toprotect against this, the downstreampressure must be held at a valueequaling 1.25 times the vapor pressureplus twice the pressure drop. Smallamounts of air entrainment (100 mg/lor less) will make the meter read onlya bit high, while large quantities candestroy the rotor.

Turbine meters also can be dam-aged by solids entrained in the fluid.If the amount of suspended solidsexceeds 100 mg/l of +75 micronsize, a flushing y-strainer or a

motorized cartridge filter must beinstalled at least 20 diameters ofstraight run upstream of theflowmeter.

• New DevelopmentsDual-rotor liquid turbines increasethe operating range in small line size(under 2 in) applications. The tworotors turn in opposite directions.The front one acts as a conditioner,directing the flow to the back rotor.The rotors lock hydraulically andcontinue to turn as the flow decreaseseven to very low rates.

The linearity of a turbine meter isaffected by the velocity profile (oftendictated by the installation), viscosity,and temperature. It is now possible toinclude complex linearization func-tions in the preamplifier of a turbineflowmeter to reduce these nonlin-earities. In addition, advances infieldbus technology make it possible



Figure 3-11: Rotary Flowmeter Designs

B) PaddlewheelA) ImpellerPipe Tee

Locknut

Flow

Paddlewheel Sensor

to recalibrate turbine flowmeterscontinuously, thereby correcting forchanges in temperature and viscosity.

Flow computers are capable of lin-earization, automatic temperaturecompensation, batching, calculationof BTU content, datalogging, andstorage of multiple K-factors. Thebatching controller is set with thedesired target volume and, when itstotalizer has counted down to zero,it terminates the batch. Such pack-ages are equipped with dribble flow,pre-warn, or trickle-cut-off circuits.Whether functioning through arelay contact or a ramp function,these features serve to minimizesplashing or overfill and to accu-rately terminate the batch.

• Gas Turbine & Shunt MetersGas meters compensate for thelower driving torque produced bythe relatively low density of gases.This compensation is obtained byvery large rotor hubs, very light rotorassemblies, and larger numbers ofrotor blades. Gas turbine meters areavailable from 2" to 12" and with flowratings up to 150,000 ft3/hr. Whenoperating at elevated gas pressures(1,400 psig), a rangeability of 100:1 canbe obtained in larger size meters.Under lower pressure conditions,typical rangeability is 20:1 with ±1%linearity. The minimum upstreamstraight pipe-run requirement is 20pipe diameters.

Shunt flowmeters are used in gasand steam service. They consist ofan orifice in the main line and arotor assembly in the bypass. Thesemeters are available is sizes 2 in. andlarger and are accurate to ±2% overa range of 10:1.

Other Rotary FlowmetersOther types of rotary elementflowmeters include propeller(impeller), shunt, and paddlewheeldesigns.

Propeller meters are commonlyused in large diameter (over 4 in) irri-gation and water distribution sys-tems. Their primary trade-off is lowcost and low accuracy (Figure 3-11A).AWWA Standard C-704 sets theaccuracy criterion for propellermeters at 2% of reading. Propellermeters have a rangeability of about4:1 and exhibit very poor perfor-mance if the velocity drops below1.5 ft/sec. Most propeller meters areequipped with mechanical registers.Mechanical wear, straightening, andconditioning requirements are thesame as for turbine meters.

Paddlewheel flowmeters use arotor whose axis of rotation is par-allel to the direction of flow (Figure3-11B). Most paddlewheel metershave flat-bladed rotors and areinherently bi-directional. Severalmanufacturers, however, usecrooked rotors that only rotate in

the forward direction. For smallerpipes (H" to 3"), these meters areavailable only with a fixed insertiondepth, while for larger pipe sizes (4"to 48") adjustable insertion depthsare available. The use of capacitive-ly coupled pick-ups or Hall-effectsensors extends the range of pad-dlewheel meters into the low-flowvelocity region of 0.3 ft/sec.

Low-flow meters (usually smallerthan 1 in.) have a small jet orificethat projects the fluid onto aPelton wheel. Varying the diameterand the shape of the jet orificematches the required flow rangeand provides a flowmeter that isaccurate to 1% FS and has a range-ability of 100:1. Higher accuracycan be achieved by calibratingthe meter and by lowering itsrange. Because of the small size ofthe jet orifice, these meters canonly be used on clean fluids andthey incur a pressure drop of about20 psid. Materials of constructioninclude polypropylene, PVDF, TFEand PFA, brass, aluminum, andstainless steel. T





Edition, OMEGA Press, 1995.• Flow Measurement Engineering Handbook, Miller, McGraw-Hill, 1982.• Flow Measurement, D. W. Spitzer, ISA, 1991.• Flowmeters in Water Supply, Manual M33, AWWA, 1989.• Industrial Flow Measurement, D. W. Spitzer, ISA 1984.• Instrument Engineer’s Handbook, Bela Liptak, editor, CRC Press, 1995.• “Turbine Flowmeter Extends Flow Range”, E. Piechota, Flow Control,

February, 1997.• Water Meters—Selection, Installation, Testing and Maintenance, Manual

M6, AWWA, 1986.

While the flow measure-ment technologies dis-cussed in this chapter—magnetic, vortex, and

ultrasonic—are neither exclusivelynor exhaustively electronic in nature,they do represent a logical groupingof flow measurement technologies.All have no moving parts (well,maybe vibrating), are relatively non-intrusive, and are made possible bytoday’s sophisticated electronicstechnology.

Magnetic flowmeters, for example,are the most directly electrical innature, deriving their first principles ofoperation from Faraday’s law. Vortexmeters depend on piezoelectric sen-sors to detect vortices shed from astationary shedder bar. And today’sultrasonic flowmeters owe their suc-cessful application to sophisticateddigital signal processing.

Magnetic FlowmetersThe operation of magnetic flowme-ters is based on Faraday’s law of elec-tromagnetic induction. Magmeterscan detect the flow of conductivefluids only. Early magmeter designsrequired a minimum fluidic conduc-tivity of 1-5 microsiemens per cen-timeter for their operation. Thenewer designs have reduced thatrequirement a hundredfold tobetween 0.05 and 0.1.

The magnetic flowmeter consistsof a non-magnetic pipe lined with aninsulating material. A pair of magneticcoils is situated as shown in Figure 4-1,and a pair of electrodes penetratesthe pipe and its lining. If a conductivefluid flows through a pipe of diameter(D) through a magnetic field density

(B) generated by the coils, the amountof voltage (E) developed across theelectrodes—as predicted by Faraday’slaw—will be proportional to thevelocity (V) of the liquid. Because themagnetic field density and the pipediameter are fixed values, they can becombined into a calibration factor (K)and the equation reduces to:

E = KV

The velocity differences at differ-ent points of the flow profile arecompensated for by a signal-weigh-ing factor. Compensation is also pro-vided by shaping the magnetic coilssuch that the magnetic flux will begreatest where the signal weighingfactor is lowest, and vice versa.

Manufacturers determine eachmagmeter’s K factor by water calibra-tion of each flowtube. The K value thusobtained is valid for any other conduc-tive liquid and is linear over the entireflowmeter range. For this reason, flow-tubes are usually calibrated at only onevelocity. Magmeters can measure flowin both directions, as reversing direc-tion will change the polarity but not

the magnitude of the signal.The K value obtained by water

testing might not be valid for non-Newtonian fluids (with velocity-dependent viscosity) or magneticslurries (those containing magneticparticles). These types of fluids canaffect the density of the magneticfield in the tube. In-line calibrationand special compensating designsshould be considered for both ofthese fluids.

• Magmeter ExcitationThe voltage that develops at theelectrodes is a millivolt signal. Thissignal is typically converted into astandard current (4-20 mA) or fre-quency output (0-10,000 Hz) at ornear the flowtube. Intelligent mag-netic transmitters with digital out-puts allow direct connection to adistributed control system. Becausethe magmeter signal is a weak one,the lead wire should be shielded andtwisted if the transmitter is remote.

The magmeter’s coils can be pow-ered by either alternating or directcurrent (Figure 4-2). When ac excita-tion is used, line voltage is applied to


Magnetic Flowmeters

Vortex Flowmeters

Ultrasonic Flowmeters

FLOW & LEVEL MEASUREMENTElectronic Flowmeters

4

WElectronic Flowmeters

Figure 4-1: The Magmeter and Its Components

ES

ES

Magnetic Coil

Electrode

V

DB

the magnetic coils. As a result, theflow signal (at constant flow) willalso look like a sine wave. The ampli-tude of the wave is proportional tovelocity. In addition to the flow signal,noise voltages can be induced in theelectrode loop. Out-of-phase noise iseasily filtered, but in-phase noiserequires that the flow be stopped(with the pipe full) and the transmitteroutput set to zero. The main problemwith ac magmeter designs is thatnoise can vary with process condi-tions and frequent re-zeroing isrequired to maintain accuracy.

In dc excitation designs, a low fre-quency (7-30 Hz) dc pulse is used toexcite the magnetic coils. When thecoils are pulsed on (Figure 4-2), thetransmitter reads both the flow andnoise signals. In between pulses, thetransmitter sees only the noise signal.Therefore, the noise can be continu-ously eliminated after each cycle.

This provides a stable zero andeliminates zero drift. In addition tobeing more accurate and able tomeasure lower flows, dc meters areless bulky, easier to install, use lessenergy, and have a lower cost ofownership than ac meters. One newdc design uses significantly morepower than the earlier generationsand thereby creates a stronger flow-tube signal.

Another new design uses a uniquedual excitation scheme that pulsesthe coils at 7 Hz for zero stability andalso at 70 Hz to obtain a stronger sig-nal. Magmeter transmitters can besupplied with either ac or dc power.A two-wire, loop-powered dc mag-netic flowmeter is also available in anintrinsically safe design, but its per-formance is reduced because ofpower limitations.

Pulsed ac meters have also beenintroduced recently, eliminating the

zero stability problems of traditionalac designs. These devices contain cir-cuitry that periodically disrupts theac power, automatically zeroing outthe effects of process noise on theoutput signal.

Today, dc excitation is used inabout 85% of installations and acmagmeters claim the other 15% when

justified by the following conditions:• When air is entrained in large

quantities in the process stream; • When the process stream is a slurry

and the solid particle sizes are notuniform and/or the solid phase isnot homogeneously mixed withinthe liquid; or

• When the flow is pulsating at afrequency under 15 Hz.When any of the above three con-

ditions exist, the output of a pulseddc meter is likely to be noisy. In somecases, one can minimize the noiseproblem (hold the fluctuations with-in 1% of setpoint) by filtering anddamping the output signal. If morethan 1 to 3 seconds of damping isrequired to eliminate the noise, it isalways better to use an ac meter.

• Flowtubes, Liners, & Probes The face-to-face dimensions offlanged flowtubes (lay lengths) usual-ly meet the recommendations of theInternational Organization forStandardization (ISO). The dimensionsof short-form magmeters usually

meet these guidelines as well.Magnetic flowtubes and liners areavailable in many materials and arewidely used in all the process indus-tries, including food, pharmaceutical,mining, and metals.

Some liner materials (particularlyTeflon®) can be damaged when prybars are used while installing it orremoving it from process piping.They can also be damaged by over-torquing the flange bolts. Liner pro-tectors are available to help preventsuch damage.

Any flowtube can generally beused with any transmitter offered bythe same manufacturer. Dependingon its construction and features, thecost of a 2-in. magnetic flowmetercan range from $1,500 to $5,000. Thiscost has been coming down, but isstill higher than that of the leastexpensive flow sensors.

Magnetic flowmeters also can bepackaged as probes and inserted intoprocess pipes through taps. Theseprobes contain both the electrodesand magnetic coils. The flowingprocess fluid induces a voltage at theelectrodes, which reflects the velocityat the probe tip and not the averagefluid velocity across the pipe. Thesemagmeters are inexpensive andretractable. Therefore, the processdoes not have to be shut down toinstall or remove them. Metering accu-racy is highly dependent on the rela-tionship between the measured veloc-ity and the average velocity in the pipe.

• ElectrodesIn conventional flowtubes, the elec-trodes are in contact with the processfluid. They can be removable or per-manent if produced by a droplet ofliquid platinum as it sinters through aceramic liner and fuses with the alu-minum oxide to form a perfect seal.

4 Electronic Flowmeters


Figure 4-2: Excitation Methods

Varying Flux

ac Excitation Pulsed dc Excitation

Non-Varying Flux

This design is preferred due to itslow cost, its resistance to abrasionand wear, its insensitivity to nuclearradiation, and its suitability for sani-tary applications because there areno cavities in which bacteria cangrow. On the other hand, the ceram-ic tube cannot tolerate bending, ten-sion, or sudden cooling and cannothandle oxidizing acids or hot andconcentrated caustic.

In a more recent capacitively-coupled design, non-contactingelectrodes are used. These designsuse areas of metal sandwichedbetween layers of liner material.They are available in sizes under eightinches in diameter and with ceramicliners. Magmeters using these non-contacting electrodes can “read” flu-ids having 100 times less conductivi-ty than required to actuate conven-tional flowtubes. Because the elec-trode is behind the liner, thesedesigns are also better suited forsevere coating applications.

• Recent DevelopmentsWhen a magnetic flowmeter is pro-vided with a capacitance level sensorembedded in the liner, it can alsomeasure the flow in partially fullpipes. In this design, the magmeterelectrodes are located at the bottomof the tube (at approximately 1/10

the pipe diameter) in order to remaincovered by the fluid. Compensationis provided for wave action and cali-bration is provided for full pipe, noflow (static level), and partially filledpipe operation.

Another recent development is amagnetic flowmeter with an unlinedcarbon steel flowtube. In this design,the measuring electrodes mountexternally to the unlined flowtube

and the magnetic coils generate afield 15 times stronger than in a con-ventional tube. This magnetic fieldpenetrates deep into the processfluid (not just around the electrodeas with standard magmeter probes).The main advantage is low initial andreplacement costs, since only thesensors need be replaced.

• Selection & SizingMagnetic flowmeters can detect theflow of clean, multi-phase, dirty, cor-rosive, erosive, or viscous liquids andslurries as long as their conductivityexceeds the minimum required for theparticular design. The expected inac-curacy and rangeability of the betterdesigns are from 0.2-1% of rate, over arange of 10:1 to 30:1, if the flow veloc-ity exceeds 1 ft/sec. At slower flowvelocities (even below 0.1 ft/s), mea-surement error increases, but thereadings remain repeatable.

It is important that the conductivi-ty of the process fluid be uniform. Iftwo fluids are mixed and the conduc-tivity of one additive is significantlydifferent from that of the otherprocess fluid, it is important that theybe completely intermixed before theblend reaches the magmeter. If theblend is not uniform, the output sig-nal will be noisy. To prevent that,pockets of varying conductivity can

be eliminated by installing a staticmixer upstream of the magmeter.

Magmeter size is determined bycapacity tables or charts publishedby the manufacturer. Figure 4-3 pro-vides a flow capacity nomographfor line sizes from 0.1 in. to 96 in. Formost applications, flow velocitiesshould fall between 3 ft/sec and15 ft/sec. For corrosive fluids, thenormal velocity range should be3-6 ft/sec. If the flowtube is contin-uously operated below 3 ft/sec,metering accuracy will deteriorate,while continuous operation exceed-ing the upper limit of the normalvelocity range will shorten the lifeof the meter.

The obstructionless nature of themagmeter lowers the likelihood ofplugging and limits the unrecoveredhead loss to that of an equivalentlength of straight pipe. The lowpressure drop is desirable because it

Electronic Flowmeters 4


Figure 4-3: Capacity Nomograph for Magnetic Flowmeters

0.01 0.1 (0.023) 1 (0.23) 10 (2.3) 100 (22.7) 1,000 (227) 10,000 (2,273) 100,000

3020

10

5

2

1

Gallons per minute (m3/hr)

Fee

t p

er

seco

nd

(m

/s)

1" (2

.5 m

m)

105"

(4 m

m)

32 1" (6

.3 m

m)

41" (1

5 m

m)

21" (2

5 m

m)

2" (5

0 mm

)

3" (7

5 m

m)

4" (1

00 mm

)

6" (1

50 m

m)

8" (2

00 mm

)

24" (

600 mm

)

36" (

900 mm

)

10"

12"

14"

16"

18"

42"

54"

66"

78"

90"

48"

60"

72"

84"

95"

lowers pumping costs and aidsgravity feed systems.

• Problem ApplicationsThe magmeter cannot distinguishentrained air from the process fluid;therefore, air bubbles will cause themagmeter to read high. If thetrapped air is not homogeneouslydispersed, but takes the form of airslugs or large air bubbles (the size ofthe electrode), this will make theoutput signal noisy or even disrupt it.Therefore, in applications where airentrainment is likely, the metershould be sized so that the flowvelocity under normal flow condi-tions is 6-12 ft/sec.

Coating of the electrodes is anoth-er common magmeter problem.Material build-up on the inner sur-faces of the meter can electrically iso-late the electrodes from the processfluid. This can cause a loss of signal ora measurement error, either by chang-ing the diameter of the flowtube orby causing span and zero shifts.Naturally, the best solution is preven-tion. One preventive step is to size themeter such that, under normal flowconditions, the flowing velocity willbe relatively high: at least 6-12 ft/sec,or as high as practical considering thepossibility of erosion and corrosion.

Another method of prevention isto use electrodes that protrude intothe flow stream to take advantage ofthe turbulence and washing effect. Inmore severe service, a mechanicalcleaning system can be installed andused intermittently or continuouslyto eliminate coating and build-ups.

• InstallationThe magnetic flowmeter mustalways be full of liquid. Therefore,the preferred location for magme-ters is in vertical upward flow lines.

Installation in horizontal lines isacceptable if the pipe section is at alow point and if the electrodes arenot at the top of the pipe. This pre-vents air from coming into contactwith the electrodes. When theprocess fluid is a slurry and the mag-meter is installed at a low point, itshould be removed during long peri-ods of shutdown, so that solids willnot settle and coat the internals.

If it is essential to drain the mag-meter periodically, it should be pro-vided with an empty tube zerooption. When this option is activat-ed, the output of the transmitter willbe clamped to zero. Detection ofempty tube conditions is by circuitryconnected to extra sets of elec-trodes in the flowtube. The emptytube zero feature can also be acti-vated by an external contact, such as

a pump status contact.Magmeters require five diameters

of straight pipe upstream and twodiameters downstream in order tomaintain their accuracy and minimize

liner wear. Liner protectors are avail-able to protect the leading edge ofthe liners from the abrasive effects ofprocess fluids. If the magmeter isinstalled in a horizontal pipe exceed-ing 30 ft in length, the pipe should besupported on both sides of the meter.

The magnetic flowmeter must beelectrically grounded to the processliquid. This is because the magmeteris part of the path for any stray cur-rent traveling down the pipeline orthrough the process liquid. Bonding,by grounding the meter at both endsto the process fluid, provides a shortcircuit for stray currents, routingthem around the flowtube instead ofthrough it. If the system is not prop-erly grounded, these currents cancreate a zero shift in the magneticflowmeter output.

Electrical bonding to the process

fluid can be achieved by metalground straps. These straps connecteach end of the flowtube to theadjacent pipeline flanges, which, inturn, are in contact with the process



Figure 4-4: Vortex Meter Calculation of Flow Velocity

l

1DV

d

Still Fluid

Shear Layer

Alternative Vortices

High Velocity Fluid

Flow

liquid. Straps are used when the pip-ing is electrically conductive. Whenthe pipe is non-conductive or lined,grounding rings are used. The ground-ing ring is like an orifice plate with abore equal to the nominal size (insidediameter) of the flowtube. It isinstalled between the flanges of theflowtube and adjacent process pipingon the upstream and downstreamsides. The flowtube is bonded to theprocess fluid by being connected tothe metallic grounding rings, and isgrounded by being wired to a goodconductor, such as a cold water pipe.

In larger sizes and in exotic materi-als, grounding rings can becomeexpensive; grounding electrodes (a

third electrode placed in the flowtubefor bonding with the process fluid) canbe used instead. Another cost-savingoption is to use a plastic grounding ringwith a metal electrode insert.

Vortex FlowmetersAs a young person fishing in the moun-tain streams of the Transylvanian Alps,

Theodor von Karman discoveredthat, when a non-streamlined object(also called a bluff body) is placed inthe path of a fast-flowing stream, thefluid will alternately separate fromthe object on its two downstreamsides, and, as the boundary layerbecomes detached and curls back onitself, the fluid forms vortices (alsocalled whirlpools or eddies). He alsonoted that the distance between thevortices was constant and dependedsolely on the size of the rock thatformed it.

On the side of the bluff bodywhere the vortex is being formed,the fluid velocity is higher and thepressure is lower. As the vortex

moves downstream, it grows instrength and size, and eventuallydetaches or sheds itself. This is fol-lowed by a vortex's being formed onthe other side of the bluff body(Figure 4-4). The alternating vorticesare spaced at equal distances.

The vortex-shedding phenomenoncan be observed as wind is shed from

a flagpole (which acts as a bluffbody); this is what causes the regularrippling one sees in a flag. Vorticesare also shed from bridge piers, pil-ings, offshore drilling platform sup-ports, and tall buildings. The forcescaused by the vortex-shedding phe-nomenon must be taken intoaccount when designing these struc-tures. In a closed piping system, thevortex effect is dissipated within afew pipe diameters downstream ofthe bluff body and causes no harm.

• Vortex Meter DesignA vortex flowmeter is typicallymade of 316 stainless steel orHastelloy and includes a bluff body,a vortex sensor assembly and thetransmitter electronics, althoughthe latter can also be mountedremotely (Figure 4-5). They are typi-cally available in flange sizes fromH in. to 12 in. The installed cost ofvortex meters is competitive withthat of orifice meters in sizes undersix inches. Wafer body meters (flan-geless) have the lowest cost, whileflanged meters are preferred if theprocess fluid is hazardous or is at ahigh temperature.

Bluff body shapes (square, rectan-gular, t-shaped, trapezoidal) anddimensions have been experimentedwith to achieve the desired charac-teristics. Testing has shown that lin-earity, low Reynolds number limita-tion, and sensitivity to velocity pro-file distortion vary only slightly withbluff body shape. In size, the bluffbody must have a width that is alarge enough fraction of the pipediameter that the entire flow partic-ipates in the shedding. Second, thebluff body must have protrudingedges on the upstream face to fix thelines of flow separation, regardless ofthe flow rate. Third, the bluff body



Figure 4-5: Vortex Detecting Sensor

Sensor

Flow

Vortex Shedder Force

Force on Sensor

Pivoting Axis

Shedder Bar

length in the direction of the flowmust be a certain multiple of thebluff body width.

Today, the majority of vortexmeters use piezoelectric or capaci-tance-type sensors to detect thepressure oscillation around thebluff body. These detectorsrespond to the pressure oscillationwith a low voltage output signalwhich has the same frequency asthe oscillation. Such sensors are mod-ular, inexpensive, easily replaced, andcan operate over a wide range oftemperature ranges—from cryo-genic liquids to superheated steam.Sensors can be located inside themeter body or outside. Wetted sen-sors are stressed directly by the vor-tex pressure fluctuations and areenclosed in hardened cases to with-stand corrosion and erosion effects.

External sensors, typically piezo-electric strain gages, sense the vortexshedding indirectly through the forceexerted on the shedder bar. Externalsensors are preferred on highly ero-sive/corrosive applications to reducemaintenance costs, while internal sen-sors provide better rangeability (bet-ter low flow sensitivity). They are alsoless sensitive to pipe vibrations. Theelectronics housing usually is ratedexplosion- and weatherproof, andcontains the electronic transmittermodule, termination connections, andoptionally a flow-rate indicatorand/or totalizer.

• Sizing & RangeabilityVortex shedding frequency is directlyproportional to the velocity of thefluid in the pipe, and therefore to vol-umetric flow rate. The shedding fre-quency is independent of fluid prop-erties such as density, viscosity, con-ductivity, etc., except that the flowmust be turbulent for vortex shedding

to occur. The relationship betweenvortex frequency and fluid velocity is:

St = f(d/V)

Where St is the Strouhal number, f isthe vortex shedding frequency, d is

the width of the bluff body, and V isthe average fluid velocity. The valueof the Strouhal number is deter-mined experimentally, and is generallyfound to be constant over a widerange of Reynolds numbers. TheStrouhal number represents the ratioof the interval between vortex shed-ding (l) and bluff body width (d),which is about six (Figure 4-4). TheStrouhal number is a dimensionlesscalibration factor used to character-ize various bluff bodies. If theirStrouhal number is the same, thentwo different bluff bodies will per-form and behave similarly.

Because the volumetric flowrateQ is the product of the average fluidvelocity and of the cross-sectional

area available for flow (A):

Q = AV = (A f d B)/St

where B is the blockage factor,defined as the open area left by thebluff body divided by the full bore

area of the pipe. This equation, inturn, can be rewritten as:

Q = f K

where K is the meter coefficient,equal to the product (A f d B). As withturbine and other frequency-produc-ing flowmeters, the K factor can bedefined as pulses per unit volume(pulses per gallon, pulses per cubicfoot, etc.). Therefore, one can deter-mine flowrate by counting the pulsesper unit time. Vortex frequenciesrange from one to thousands of puls-es per second, depending upon theflow velocity, the character of theprocess fluid, and the size of themeter. In gas service, frequencies are



Figure 4-6: Installation Recommendations

Concentric Reducer

Flow

A) Length = Size of Meter

B) Upward C) Downward D) Horizontal

h

h>0h

h>0Flow

Flow

Upstream Straight Pipe Run

Vortex Meter

Downstream Straight Pipe Run

Concentric Expander

about 10 times higher than in liquidapplications.

The K factor is determined by themanufacturer, usually by water cali-bration in a flow lab. Because the Kfactor is the same for liquid, gas andvapor applications, the value deter-mined from a water calibration is valid

for any other fluid. The calibrationfactor (K) at moderate Reynolds num-bers is not sensitive to edge sharpnessor other dimensional changes thataffect square-edged orifice meters.

Although vortex meter equationsare relatively simple compared tothose for orifice plates, there aremany rules and considerations tokeep in mind. Manufacturers offerfree computer software for sizing,wherewith the user enters the fluid'sproperties (density, viscosity, anddesired flow range) and the programautomatically sizes the meter.

The force generated by the vortexpressure pulse is a function of fluiddensity multiplied by the square offluid velocity. The requirement thatthere be turbulent flow and force suf-ficient to actuate the sensor deter-mines the meter’s rangeability. Thisforce has to be high enough to be dis-tinguishable from noise. For example,a typical 2-in. vortex meter has awater flow range of 12 to 230 gpm. If

the density or viscosity of the fluiddiffers from that of water, the meterrange will change.

In order to minimize measurementnoise, it is important to select ameter that will adequately handleboth the minimum and maximumprocess flows that will be measured.

It is recommended that the minimumflow rate to be measured be at leasttwice the minimum flow ratedetectable by the meter. The maxi-mum capacity of the meter shouldbe at least five times the anticipatedmaximum flowrate.

• Accuracy & Rangeability Because the Reynolds number dropsas viscosity rises, vortex flowmeterrangeability suffers as the viscosityrises. The maximum viscosity limit,as a function of allowable accuracyand rangeability, is between 8 and30 centipoises. One can expect abetter than 20:1 rangeability for gasand steam service and over 10:1 forlow-viscosity liquid applications ifthe vortex meter has been sizedproperly for the application.

The inaccuracy of most vortexmeters is 0.5-1% of rate for Reynoldsnumbers over 30,000. As the Reynoldsnumber drops, metering errorincreases. At Reynolds numbers less

than 10,000, error can reach 10% ofactual flow.

While most flowmeters continueto give some indication at near zeroflows, the vortex meter is providedwith a cut-off point. Below this level,the meter output is automaticallyclamped at zero (4 mA for analog

transmitters). This cut-off point cor-responds to a Reynolds number at orbelow 10,000. If the minimum flowthat one needs to measure is at leasttwice the cut-off flow, this does notpose a problem. On the other hand,it can still be a drawback if lowflowrate information is desired dur-ing start-up, shutdown, or otherupset conditions.

• Recent DevelopmentsSmart vortex meters provide a digi-tal output signal containing moreinformation than just flow rate. Themicroprocessor in the flowmetercan automatically correct for insuf-ficient straight pipe conditions, fordifferences between the bore diam-eter and that of the mating pipe, forthermal expansion of the bluffbody, and for K-factor changeswhen the Reynolds number dropsbelow 10,000.

Intelligent transmitters are alsoprovided with diagnostic subroutines



Figure 4-7: Ultrasonic Flowmeter Designs

A) Doppler Shift

Transmit Transducer (Typical)

Particles in Flowstream

Receive Transducer (Typical)

Upstream Transducer (T1)

Downstream Transducer (T2)B) Transit Time

fo a

afl

a = Refraction Angle

a Flow Profile

Flow Direction

to signal component or other failures.Smart transmitters can initiate testingroutines to identify problems withboth the meter and with the applica-tion. These on-demand tests can alsoassist in ISO 9000 verification.

Some recently introduced vortexflowmeters can detect mass flow.One such design measures both thevortex frequency and the vortexpulse strength simultaneously. Fromthese readings, the density of theprocess fluid can be determined andthe mass flow calculated to within2% of span.

Another newer design is providedwith multiple sensors to detect notonly the vortex frequency, but alsothe temperature and pressure of theprocess fluid. Based on that data, itdetermines both the density and themass flow rate. This meter offers a1.25% of rate accuracy when measur-ing the mass flow of liquids and a 2%of rate accuracy for gases and steam.If knowledge of process pressure andtemperature is of value for other rea-sons, this meter provides a conve-nient, less costly alternative toinstalling separate transmitters.

• Applications & LimitationsVortex meters are not usually recom-mended for batching or other inter-mittent flow applications. This isbecause the dribble flow-rate settingof the batching station can fall belowthe meter’s minimum Reynolds num-ber limit. The smaller the total batch,the more significant the resultingerror is likely to be.

Low pressure (low density) gasesdo not produce a strong enoughpressure pulse, especially if fluidvelocities are low. Therefore, it islikely that in such services therangeability of the meter will bepoor and low flows will not be

measurable. On the other hand, ifreduced rangeability is acceptableand the meter is correctly sized fornormal flow, the vortex flowmetercan still be considered.

If the process fluid tends to coator build-up on the bluff body, as insludge and slurry service, this willeventually change the meter’s K fac-tor. Vortex-shedding flowmeters arenot recommended for such applica-tions. If, however, a dirty fluid hasonly moderate amounts of non-coating solids, the application islikely to be acceptable. This wasdemonstrated by a 2-year test on alimestone slurry. At the end of thetest, the K factor was found to havechanged only 0.3% from the originalfactory calibration, although thebluff body and flowtube were badlyscarred and pitted.

When measuring multi-phase flow(solid particles in gas or liquid; gasbubbles in liquid; liquid droplets ingas), vortex meter accuracy will drop

because of the meter’s inability todifferentiate between the phases.Wet, low-quality steam is one suchapplication: the liquid phase shouldbe homogeneously dispersed withinthe steam, and vertical flow linesshould be avoided to prevent slug-ging. When the pipe is horizontal, the

liquid phase is likely to travel on thebottom of the pipe, and thereforethe inner area of the pipe should bekept open at the bottom. This canbe achieved by installing the bluffbody horizontally. Measurementinaccuracy in such applications isabout 5% of actual flow, but withgood repeatability.

The permanent pressure lossthrough a vortex meter is about halfthat of an orifice plate, roughly twovelocity heads. (A velocity head isdefined as V2/g, where V is the flowvelocity and g is the gravitationalconstant in consistent units.) If thepipe and meter are properly sizedand of the same size, the pressuredrop is likely to be only a few psi.However, downsizing (installing asmaller-than-line-size meter) in orderto increase the Reynolds canincrease the head loss to more than10 psi. One should also make surethat the vena contracta pressuredoes not drop below the vapor pres-

sure of the process fluid, because thatwould cause cavitation. Naturally, ifthe back-pressure on the meter isbelow the vapor pressure, the processfluid will flash and the meter readingwill not be meaningful.

The main advantages of vortexmeters are their low sensitivity to



Figure 4-8: Clamp-On Ultrasonic Flowmeter

Receiving Element

Transmitting Element

Flow Direction

Reflectors

variations in process conditions andlow wear relative to orifices or tur-bine meters. Also, initial and mainte-nance costs are low. For these rea-sons, they have been gaining wideracceptance among users.

• Installation RecommendationsWhen installing a vortex flowmeter inan existing process where the flowrange is not known, it is recommended

to first make some approximatemeasurements (using portable pitotor clamp-on ultrasonic devices).Otherwise, there is no guarantee thata line-size vortex meter will work at all.

The vortex meter requires a well-developed and symmetrical flowvelocity profile, free from any distor-tions or swirls. This necessitates theuse of straight up- and downstreampiping to condition the flow. Thestraight length of pipe must be thesame size as the meter (Figure 4-6) andits length should be about the same asrequired for an orifice installationwith a beta ratio of 0.7 (see Chapter 2).Most vortex flowmeter manufactur-ers recommend a minimum of 30 pipediameters downstream of controlvalves, and 3 to 4 pipe diametersbetween the meter and downstreampressure taps. Temperature elementsshould be small and located 5 to 6diameters downstream.

About half of all vortex meterinstallations require the “neckingdown” of oversized process piping byconcentric reducers and expanders.Even if flow straighteners areinstalled, some straight (relaxation)piping will still be required.

Vortex meters can be installedvertically, horizontally, or at anyangle, as long as they are kept flood-ed. The meter can be kept flooded

by installing it in a vertical upwardflow line (Figure 4-6B). Wheninstalling the flowmeter in a down-ward (Figure 4-6C) or horizontal(Figure 4-6D) flow, the downstreampiping should be kept elevated.Check valves can be used to keep thepiping full of liquid when there is noflow. Block and bypass valves arerequired if the replacement of thesensor in the particular designrequires the stopping of the flow andthe opening up of the process.

Mating flanges (on the schedule40 or schedule 80 mating piping)must have the same diameter andsmooth bore as the flowmeter. Weldneck flanges are preferred, andreducing flanges should not be used.The inner surface of the mating pipeshould be free from mill scale, pits,holes, reaming scores and bumps fora distance of 4 diameters upstreamand 2 diameters downstream of the

meter. The bores of the meter, thegaskets and the adjacent piping mustbe carefully aligned to eliminate anyobstructions or steps.

Excessive pipe vibration can beeliminated by supporting the pipingon both sides of the meter, or byrotating the meter so that the sensoris moved out of the plane of thevibration. Process noise due to valvechattering, steam traps, or pumps can

result in high readings or non-zeroreadings under zero-flow conditions.Most meter electronics allow forincreasing the noise filter settings,but increased noise reduction usuallyalso decreases the low-flow sensitiv-ity of the meter. One option is torelocate the meter to a less noisypart of the process.

Ultrasonic FlowmetersThe speed at which sound propa-gates in a fluid is dependent on thefluid’s density. If the density is con-stant, however, one can use the timeof ultrasonic passage (or reflection)to determine the velocity of aflowing fluid.

Some manufacturers producetransducer systems that operate inthe shear-mode, sending a singlepulse and receiving a single pulse inreturn. Narrow-beam systems arecommonly subject to walk-away (the



Figure 4-9: Spool-Piece Designs for High Accuracy Ultrasonic Flowmetering

A) B)

Transducer B

Transducer A

Chordal Diametric

V θL

Flanged End

Flanged End

signal completely missing the down-stream transducer). Wide-beam sys-tems overcome beam refraction andwork better in changing liquid densi-ty and temperature. With the adventof digital signal processing, it hasbecome possible to apply digital sig-nal coding to the transmitted signal.This can eliminate many of the prob-lems associated with noise and varia-tions in liquid chemistry.

• The Doppler ShiftIn 1842, Christian Doppler discoveredthat the wavelength of sound per-ceived by a stationary observer appearsshorter when the source is approach-ing and longer when the source is mov-ing away. This shift in frequency is thebasis upon which all Doppler-shiftultrasonic flowmeters work.

Doppler flowmeter transducersoperate at 0.640 MHz (in clamp-ondesigns) and at 1.2 MHz in wettedsensor designs. The transducersends an ultrasonic pulse or beaminto the flowing stream. The soundwaves are reflected back by suchacoustical discontinuities as parti-cles, entrained gas bubbles, or evenby turbulence vortices (Figure 4-7A).For clamp-on designs, measurementinaccuracy ranges from ±1% to ±5%full scale (FS).

The meter detects the velocity ofthe discontinuities, rather than thevelocity of the fluid, in calculatingthe flow rate. The flow velocity (V)can be determined by:

V = (f0 - f1)Ct /2f0 cos(a)

Where Ct is the velocity of soundinside the transducer, f0 is the trans-mission frequency, f1 is the reflectedfrequency, and a is the angle of thetransmitter and receiver crystalswith respect to the pipe axis.

Because Ct /2f0cos(a) is a constant(K), the relationship can be simpli-fied to:

V = (f0 - f1)K

Thus, flow velocity V (ft/sec) isdirectly proportional to the changein frequency. The flow (Q in gpm) in apipe having a certain inside diameter(ID in inches) can be obtained by:

Q = 2.45V(ID)2 = 2.45[(f0 - f1)K](ID)2

The presence of acoustical dis-continuities is essential for theproper operation of the Dopplerflowmeter. The generally acceptedrule of thumb is that for proper sig-nal reflection there be a minimumof 80-100 mg/l of solids with a par-ticle size of +200 mesh (+75 micron).In the case of bubbles, 100-200mg/l with diameters between +75and +150 microns is desirable. Ifeither the size or the concentrationof the discontinuities changes, theamplitude of the reflected signalwill shift, introducing errors.

Doppler flowmeters are often usedto measure the flow of such fluids as

slurries. If the solids concentrationis too high (in excess of 45% byweight), or if too much air or gas isentrained (especially if the bubblesare very fine), these discontinuities

will attenuate the reflected Dopplersignal to the point where it cannotbe distinguished from the back-ground noise in the pipe.

The reflected Doppler signalis shifted from the transmittedfrequency by approximately 6 Hzfor every foot per second of velocity.Therefore, if the flow velocity is lessthan 1 ft/sec, ultrasonic flowmeter-ing is not practical. There seems tobe no upper limit to detectable flowvelocity, as successful installations atvelocities in the 40-50 ft/sec rangeare well documented.

• Transit Time MeasurementIn this design, the time of flight of theultrasonic signal is measured betweentwo transducers—one upstream andone downstream (Figure 4-7B). Thedifference in elapsed time going withor against the flow determines thefluid velocity.

When the flow is zero, the time forthe signal T1 to get to T2 is the same asthat required to get from T2 to T1.When there is flow, the effect is toboost the speed of the signal in thedownstream direction, while decreas-ing it in the upstream direction. The

flowing velocity (Vf) can be deter-mined by the following equation:

Vf = Kdt/TL



Figure 4-10: Axial Flowmeter

Small Pipe Configuration

Flow

where K is a calibration factor for thevolume and time units used, dt is thetime differential between upstreamand downstream transit times, and TLis the zero-flow transit time.

Theoretically, transit-time ultra-sonic meters can be very accurate

(inaccuracy of ±0.1% of reading issometimes claimed). Yet the error inthese measurements is limited byboth the ability of the signal process-ing electronics to determine the tran-sit time and by the degree to whichthe sonic velocity (C) is constant. Thespeed of sound in the fluid is a func-tion of both density and temperature.Therefore, both have to be compen-sated for. In addition, the change insonic velocity can change the refrac-tion angle (“a” in Figure 4-7B), which inturn will affect the distance the signalhas to travel. In extreme cases, the sig-nal might completely miss the down-stream receiver. Again, this type offailure is known as walk-away.

• Design VariationsClamp-on ultrasonic meters come ineither single or dual-sensor versions.

In the single-sensor version, thetransmit and receive crystals are pot-ted into the same sensor body, whichis clamped onto a single point of thepipe surface (Figure 4-8). In the dual-sensor version, the transmit crystal isin one sensor body, while the receive

crystal is in another. Clamp-on transit time meters have

been available since the early 1970s.Their aim is to rival the performanceof wetted spool-piece designs, butwithout the need to break the pipe orstop the process to install the meter.This goal has not yet been reached.

Clamp-on Doppler flowmeters aresubject to interference from the pipewall itself, as well as from any airspace between the sensor and thewall. If the pipe wall is made of stain-less steel, it might conduct the trans-mit signal far enough so that thereturning echo will be shiftedenough to interfere with the reading.There are also built-in acoustic dis-continuities in concrete-lined, plas-tic-lined, and fiberglass-reinforcedpipes. These are significant enoughto either completely scatter the

transmitted signal or attenuate thereturn signal. This dramaticallydecreases flowmeter accuracy (towithin only ±20%), and, in mostcases, clamp-on meters will not workat all if the pipe is lined.

Wetted transducer designs—bothDoppler and transit time are avail-able—overcome many of these signalattenuation limitations. The full-pipetransit-time meter originally consistedof a flanged spool section with wet-ted transducers mounted in the pipewall in transducer wells opposite toone another but at 45-degree anglesto the flow (Figure 4-9A). Transit-timeflowmeters can be either single-pathor multiple-path designs (Figure 4-9B).

Single-path flowmeters are pro-vided with a single pair of transduc-ers that make a single-line velocitymeasurement. They use a meter fac-tor that is pre-determined by calibra-tion to compensate for variations invelocity profile and for flow sectionconstruction irregularities.

In the design of multi-pathflowmeters, several sets of transduc-ers are placed in different pathsacross the flow section, therebyattempting to measure the velocityprofile across the entire cross-sec-tion of the pipe. Multi-path instru-ments are used in large-diameterconduits, such as utility stacks, and inother applications where non-uni-form flow velocity profiles exist.

Transit-time meters can also beused to measure both very hot (e.g.,liquid sulfur) and very cold (liquidnitrogen) fluids, and also to detectvery low flows. Wetted-transducerdesigns for small pipes (down to H in.)are called axial or co-axial designs(Figure 4-10). These devices permittransit-time measurement along apath length significantly greater thanthe diameter of the pipe, increasing



Figure 4-11: K-Factor Variation with Reynolds Number

1 10 100 1,000 104 10

510

610

7

1.00

0.95

0.90

0.85

0.80

0.75

0.70

Re

K

K = 1 Asymptote

For Flat Profile

K = 0.75 For Laminar Flow

low-flow sensitivity.Originally, ultrasonic flowmeters

were divided into those using theDoppler-shift principle and thoseusing the transit-time principle. Morerecently, flowmeters are capable ofmeasuring the flow of both clean flu-ids and of slurries with entrainedsolids or other acoustical discontinu-ities. Microprocessors have made itpossible to switch automaticallyfrom clean fluid mode to particulatemode based on the "correlation fac-tor". This figure of merit dramaticallyimproves the accuracy of overallperformance. In some carefully engi-neered applications, installed accu-racy to within 0.5% of reading hasbeen reported.

• Applications & PerformanceDoppler flowmeters are not recom-mended for clean fluid applications.Transit-time flowmeters, on theother hand, are often used to mea-sure the flow of crude oils and sim-ple fractions in the petroleum indus-try. They also work well with viscousliquids, provided that the Reynoldsnumber at minimum flow is eitherless than 4,000 (laminar flow) orabove 10,000 (turbulent flow).Serious non-linearities are present inthe transition region (Figure 4-11).

Transit-time flowmeters are thestandard for measuring cryogenic liq-uids down to -300°C and are alsoused in molten metal flowmetering.Measurement of liquid argon, liquidnitrogen, liquid helium and moltensulfur have often been reported.Spool-section type flowmeters aremost often used for these applica-tions, especially the axial and co-axial designs.

Raw wastewater applications usu-ally have too few acoustic disconti-nuities for Doppler flowmeters. On

the other hand, raw wastewater is notclean enough all the time for transit-time measurement. Other waste-water-related applications are equallyproblematic, as the solids concentra-tion can be too high for either transit-time or Doppler flowmeters to workproperly. In still other wastewaterapplications, the problem is that theacoustical absorbency of the mostlyorganic solids in wastewater attenu-ates the ultrasonic signals.

The use of multi-path flowmetersin raw wastewater and storm waterapplications is common, whileDoppler or cross-correlation hybriddesigns are most often used to mea-sure activated sludge and digestedsludge flows.

For mining slurries, Dopplerflowmeters typically work well.Among the few problem applications

are those in HDPE pipe, because thepipe wall flexes enough to changethe diameter of the measurementarea. This affects the accuracy of themeter. In addition, the flexure of thepipe wall can often break theacoustic coupling of the transducerto the outside of the pipe, causingfailure. Another problem area is themeasurement of slurries that areacoustically absorbent, such as limeor kaolin slurries. These applicationsfail because the highly absorbentsolids attenuate the signal belowusable strength. Lower frequency(0.45 MHz) sensors have been triedfor these applications, but successhas been limited.

Multi-path, transit-time flowme-ters also measure stack gas flows inpower-plant scrubbers, even in verylarge diameter stacks. T





Edition, OMEGA Press, 1995.• “An Intelligent Vortex Flowmeter,” T. Kamano and others, ISA/92

Proceedings, Instrument Society of America, 1992.• “Application and Installation Guidelines for Volumetric and Mass

Flowmeters,” D. Ginesi and C. Annarummo, ISA Transactions, 1994.• “Clamp-On Leak Detectors Protect Mid-Valley Line,” S. Douglas and J.

Baumoel, Pipeline & Gas Journal, April 1993.• “Committee Report: Transit Time Ultrasonic Flowmeters,” AWWA

Subcommittee on Ultrasonic Devices, AWWA Journal, July 1997.• Flow Measurement Engineering Handbook, R.W. Miller, McGraw Hill, 1996.• Flow Measurement, D.W. Spitzer, editor, Instrument Society of America, 1991.• “Flow Sensing: The Next Generation,” D. Ginesi, Control Engineering,

November 1997.• Flowmeters in Water Supply, Manual M33, AWWA, 1989.• Industrial Flow Measurement, D.W. Spitzer, ISA, 1984 • Instrument Engineers’ Handbook, Bela Liptak, editor, CRC Press, 1995.• Ultrasonic Clamp-On Flowmeters: Have They Finally Arrived?,” P. Espina,

Flow Control, January 1997.• Water Meters - Selection, Installation, Testing and Maintenance, Manual

M6, AWWA, 1986.

Mass flow measurement isthe basis of most recipeformulations, material bal-ance determinations, and

billing and custody transfer opera-tions throughout industry. With thesebeing the most critical flow measure-ments in a processing plant, the relia-bility and accuracy of mass flowdetection is very important.

In the past, mass flow was oftencalculated from the outputs of a vol-umetric flowmeter and a densitome-ter. Density was either directly mea-sured (Figure 5-1A), or was calculatedusing the outputs of process temper-ature and pressure transmitters.These measurements were not veryaccurate, because the relationshipbetween process pressure or temper-ature and density are not always pre-cisely known—each sensor adds its

own separate error to the overallmeasurement error, and the speed ofresponse of such calculations is usu-ally not sufficient to detect stepchanges in flow.

One of the early designs of self-contained mass flowmeters operated

using angular momentum (Figure 5-1B).It had a motor-driven impeller thatimparted angular momentum (rotarymotion) by accelerating the fluid to aconstant angular velocity. The higherthe density, the more angularmomentum was required to obtainthis angular velocity. Downstream ofthe driven impeller, a spring-held sta-tionary turbine was exposed to thisangular momentum. The resultingtorque (spring torsion) was an indica-tion of mass flow.

These meters all had moving partsand complex mechanical designs.First developed for the measurementof aircraft fuel, some are still in use.However, because of their complexnature and high maintenance costs,they are gradually being replaced bymore robust and less maintenance-demanding designs.

Mass flow also can be measuredby batch weighing or by combiningan accurate level sensor with a den-sitometer. Another method is tomount two d/p transmitters on thelower part of an atmospheric tank atdifferent elevations. In this case, the

output of the top d/p cell will varywith the level in the tank, while thelower one will measure the hydrosta-tic head over a fixed elevational dis-tance. This pressure differentialyields the density of the material inthe tank. Such systems have beenused to measure the total mass flowof slurries.

Coriolis Mass FlowmetersIt was G.G. Coriolis, a French engi-neer, who first noted that all bodiesmoving on the surface of the Earthtend to drift sideways because of theeastward rotation of the planet. Inthe Northern Hemisphere thedeflection is to the right of themotion; in the Southern, it is to theleft. This drift plays a principal role inboth the tidal activity of the oceansand the weather of the planet.

Because a point on the equatortraces out a larger circle per daythan a point nearer the poles, abody traveling towards either polewill bear eastward, because it retainsits higher (eastward) rotationalspeed as it passes over the more slowly


Coriolis Mass Flowmeters

Thermal Mass Flowmeters

Hot-Wire Anemometers

FLOW & LEVEL MEASUREMENTMass Flowmeters

5

MMass Flowmeters

Figure 5-1: Traditional Mass Flowmeters

Section YYMagnetic

Field

A) Magnetic Flow and Radiation Density B) Angular Momentum

Gamma Source

Detector

Field Coil Constant Speed Motor Impeller

Y

Y

Turbine

SpringAnnular SpaceField CoilProcess

FluidElectrode

rotating surface of the earth. This driftis defined as the Coriolis force.

The first industrial Coriolis patentsdate back to the 1950s, and the firstCoriolis mass flowmeters were builtin the 1970s. These flowmeters artifi-cially introduce a Coriolis accelera-tion into the flowing stream andmeasure mass flow by detecting theresulting angular momentum.

When a fluid is flowing in a pipeand it is subjected to Coriolis accel-eration through the mechanicalintroduction of apparent rotationinto the pipe, the amount ofdeflecting force generated by theCoriolis inertial effect will be afunction of the mass flow rate ofthe fluid. If a pipe is rotated arounda point while liquid is flowingthrough it (toward or away from thecenter of rotation), that fluid willgenerate an inertial force (acting onthe pipe) that will be at right anglesto the direction of the flow.

With reference to Figure 5-2, a par-ticle (dm) travels at a velocity (V)inside a tube (T). The tube is rotatingabout a fixed point (P), and the parti-cle is at a distance of one radius (R)from the fixed point. The particlemoves with angular velocity (w) undertwo components of acceleration, a

centripetal acceleration directedtoward P and a Coriolis accelerationacting at right angles to ar:

ar (centripetal) = w2rat (Coriolis) = 2wv

In order to impart the Coriolis accel-eration (at) to the fluid particle, aforce of at (dm) has to generated by

the tube. The fluid particle reacts tothis force with an equal and oppositeCoriolis force:

Fc = at(dm) = 2wv(dm)

Then, if the process fluid has density Dand is flowing at constant speedinside a rotating tube of cross-sec-tional area A, a segment of the tubeof length x will experience a Coriolisforce of magnitude:

Fc = 2wvDAx

Because the mass flowrate is dm =DvA, the Coriolis force Fc = 2w(dm)xand, finally:

Mass Flow = Fc/(2wx)

This is how measurement of theCoriolis force exerted by the flowing

fluid on the rotating tube can pro-vide an indication of mass flowrate.Naturally, rotating a tube is not prac-tical when building a commercialflowmeter, but oscillating or vibrat-ing the tube can achieve the sameeffect. Coriolis flowmeters can mea-sure flow through the tube in eitherthe forward or reverse directions.

In most designs, the tube isanchored at two points and vibratedbetween these anchors.

This configuration can be envi-sioned as vibrating a spring and massassembly. Once placed in motion, aspring and mass assembly will vibrateat its resonant frequency, which is afunction of the mass of that assem-bly. This resonant frequency is select-ed because the smallest driving forceis needed to keep the filled tube inconstant vibration.

• Tube DesignsA tube can be of a curved or straightform, and some designs can also beself-draining when mounted vertical-ly (Figure 5-3). When the design con-sists of two parallel tubes, flow isdivided into two streams by a splitternear the meter’s inlet and is recom-bined at the exit. In the single contin-uous tube design (or in two tubesjoined in series), the flow is not splitinside the meter.

In either case, drivers vibrate thetubes. These drivers consist of a coilconnected to one tube and a magnetconnected to the other. The transmit-ter applies an alternating current tothe coil, which causes the magnet tobe attracted and repelled by turns,thereby forcing the tubes towards andaway from one another. The sensorcan detect the position, velocity, oracceleration of the tubes. If electro-magnetic sensors are used, the mag-net and coil in the sensor change their

5 Mass Flowmeters


Figure 5-2: The Coriolis Principle

r

w

p dm

ar

at

x

v

T

relative positions as the tubes vibrate,causing a change in the magnetic fieldof the coil. Therefore, the sinusoidalvoltage output from the coil repre-sents the motion of the tubes.

When there is no flow in a two-tube design (Figure 5-3A), the vibra-tion caused by the coil and magnetdrive results in identical displace-ments at the two sensing points (B1and B2). When flow is present,Coriolis forces act to produce a sec-ondary twisting vibration, resulting in

a small phase difference in the rela-tive motions. This is detected at thesensing points. The deflection of thetubes caused by the Coriolis forceonly exists when both axial fluid flowand tube vibration are present.Vibration at zero flow, or flow with-out vibration, does not produce an

output from the meter.The natural resonance frequency

of the tube structure is a function ofits geometry, materials of construc-tion, and the mass of the tube assem-bly (mass of the tube plus the massof the fluid inside the tube). Themass of the tube is fixed. Since massof the fluid is its density (D) multi-plied by its volume (which is alsofixed), the frequency of vibration canbe related to the density of theprocess fluid (D). Therefore, the

density of the fluid can be determinedby measuring the resonant frequencyof oscillation of the tubes. (Note thatdensity can be measured at zeroflow, as long as the tubes are filledwith fluid and vibrating.)

Wall thickness varies considerablyfrom design to design; however, even

the sturdiest tubing will be thinnerthan the process piping. In addition,some designs use small bore tubing,which drastically increases the flow-ing velocity (from 5-10 ft/sec to morethan 25 ft/sec). Designs with thinwalls and high fluid velocities (that is,small bore tubing), may require theuse of exotic materials because oferosion concerns. One will obtain thelongest meter life by selecting thedesign with the thickest wall and theslowest flow velocity that can provide

the required accuracy and range.The Coriolis meter may need to be

made out of exotic materials becauseof corrosion considerations or to pre-vent pitting. Carbon or stainless steelcan often be used in process piping,because a small amount of pittingcan be tolerated. In case of the

Mass Flowmeters 5


Figure 5-3: Two-Tube and Straight-Tube Coriolis Meter Operation

B)A)

Measurng Tube

Flow Detector

(B2)

Flow Detector

(B1)Fi = Inertial Force Fd = Drive Force Flow Tube

Drive (A)

Fd

FiFi

FiFi

Fi

Fi

Fi

FiFd

FdFd

Fd

Fd

Fd

Fd

FlowFlow

No FlowNo Flow

Counter PendulumExciter

Electrodynamic Sensors

Nitrogen Secondary Containment

Inner Frame

vv

Fe

Fev > 0

v = 0

Coriolis meter, even a small amountof pitting cannot be toleratedbecause the walls are thin, and pittinginduces stress concentrations withinthe tube structure. Therefore, stan-dard corrosion tables (based onweight loss criteria) are not suitablefor selecting Coriolis tube materials,and the stricter guidelines of themanufacturers must be used.

• Transmitter DesignsTransmitters can operate on either acor dc power and require separatewiring for the power supply and fortheir output signals. The Coriolisflowmeter transmitter can be integrallyor remotely mounted (Figure 5-4). Thetransmitter controls the operation ofthe driver and processes and trans-mits the sensor signals. The calibra-tion factor (K) in the transmitter’smemory matches the transmitter tothe particular flow tube. This calibra-tion factor defines the constant ofproportionality between the Coriolisforce and the mass flow rate for thedynamic spring constant of the par-ticular vibrating tubes.

The transmitter does more thanconvert sensor inputs into standard-ized output signals. Most transmit-ters also offer multiple outputs,including mass flow rate, total massflow, density, and temperature.Analog and/or pulse outputs areboth available, and intelligent trans-mitters can generate digital outputsfor integration into DCS systems.

Transmitters are often providedwith a local displays and keypads toallow easy access to process data.Coriolis transmitters provide morethan just flow information and ancil-lary functions. Batch control func-tions, percent Brix or percent HFCSmonitoring, viscosity, percent solids,PID, API gravity, and specific gravity

also are available. When viscosityinformation is desired, the meter pres-sure drop needs to be measured.Other features may require informa-tion to be pre-programmed into thetransmitter memory. In addition,

transmitters have other hardware andsoftware options which allow the userto customize them to the application.

• Coriolis EvolutionThe first generation of Coriolismeters consisted of a single curvedand a thin-walled tube, in which highfluid velocities were created byreducing the tube cross-sectionalarea in relation to the process pipe.The tube distortion was measured in

reference to a fixed point or plane.The tubes were excited in such a waythat localized high amplitude bend-ing forces were created at the anchorpoints. This resulted in severe vibra-tion problems, which were alleviated

by two-tube designs (Figure 5-3A).These designs reduced external

vibration interference, decreased thepower needed to vibrate the tubes,and minimized the vibrational energyleaving the tube structure. One driverwas used to initiate tube vibration,and two sensors were used to detectthe Coriolis deflections. While thisdesign greatly improved performance,the combination of reduced bore,thin-walled tubing, and high fluid

5 Mass Flowmeters


Figure 5-4: Coriolis Transmitter with Keyboard and Display

+ 1

F1 F2 F3 F4

2 3

- 4 5 6

7 8 9

. 0

FLOW VARIABLES • Mass Flow Rate • Mass Flow Total • Volumetric Flow Rate • Volumetric Flow Total • Density • Temperature • % Solids • Dry Solids Mass Flow Rate • Dry Solids Total

CONFIGURE METER • Mass, Volume, Timebase Units • Meter Constants • Configure Outputs • Scale Outputs • Configure Inputs

DIAGNOSTICS & SIMULATION

KEYBOARD SECURITY

FUNCTION COMMANDS • Reset Total • Zero Flow Calibration • Display Mode • Clear Alarm • Stop Measurement • Start Measurement

CONFIGURABLE INPUTS/OUTPUTS

ANALOG & FREQUENCY OUTPUTS

CONTACT INPUTS

RS-485/422

DISCRETE OUTPUTS

OUTPUT ALARMS • Batch Control Capability • High/Low Limit Alarms: • Flow Rate • Flow Total • Density • Temperature • % Solids • Flow Direction • Malfunction Alarm

velocities (up to 50 ft/sec) still result-ed in premature meter failure, includ-ing potentially catastrophic spills

when the meter was used on corro-sive and erosive services. In addition,the unrecovered head losses werehigh (sometimes over 50 psid), andaccuracy was not high enough toallow users to convert batch processesinto continuous ones.

More recent design improve-ments include the introduction of avariety of new tube shapes, includ-ing ones that do not split the flow(Figure 5-3B) and the use of multipledrivers (Figure 5-5A). Thick-walledtubing (five times thicker than earlydesigns), the use of full bore diame-ters and heavy manifolds to isolatethe tube structure from stressesinduced from piping connections,and flowtube housings that doubleas secondary containment vesselshave all contributed to improvedperformance.

In some designs, torsional stressesreplaced bending, in order to preventthe concentration of stresses thatcan lead to tube cracking (Figure 5-5B). In other designs, the effects ofpipeline vibration have been mini-mized by mounting the tube struc-tures transverse to the pipeline.

These improvements increased

the number of suppliers and con-tributed to the development of anew generation of Coriolis meters

that are as reliable and rugged as tra-ditional volumetric flowmeters. Thenew designs operate at lower fluidvelocities (below 10 ft/sec) and atlower pressure drops (under 12 psid),can be installed in any orientation,and provide longer service life onslurry, viscous, corrosive, or erosiveservices. The tubes are vibrated wellbelow their endurance limits, andtypically are made of stainless steel,Hastelloy, and titanium.

• Interferences The effect of the Coriolis force onthe vibrating tube is small. Full-scaleflow might cause a deflection ofonly 0.001 inch. To obtain a flowrangeability of 100:1, sensors mustbe able to detect deflections to anaccuracy of 0.000001 inch in indus-trial environments where theprocess pressure, temperature, andfluid density are all changing, andwhere pipe vibration interferes withmeasurement.

The elasticity of metal tubeschanges with temperature; theybecome more elastic as they getwarmer. To eliminate the corre-sponding measurement error, the

tube temperature is continuouslymeasured by an RTD element and isused to continuously compensate

for variations in tube elasticity.Coriolis mass flowmeters usually

are calibrated on water, because theconstants are valid for all other liq-uids. Calibration for density is usuallydone by filling the tubes with two ormore (stagnant) calibration fluids ofknown densities.

• Accuracy & RangeabilityCoriolis meters provide 0.1-2% ofrate inaccuracy over a mass flowrange of up to 100:1. In general,curved tube designs provide widerrangeability (100:1 to 200:1), whilestraight-tube meters are limited to30:1 to 50:1 and their accuracy islower. Overall meter error is the sumof base inaccuracy and zero-shifterror, the error attributable to theirregular output signal generated atzero flow conditions. Zero-shifterror becomes the dominant portionof total error at the lower end of theflow range, where the error isbetween 1% and 2% of rate. Somemanufacturers state the overallaccuracy as a percentage of rate forthe upper portion of the flow rangeand as a percentage of span for thelower portion, while others state it

Mass Flowmeters 5


Figure 5-5: Coriolis Design Improvements

A) Torsional Bending B) Support Block and Multiple Sensors

as a percentage of rate plus a zero-shift error. There is a fair amount of“specmanship,” and one must readsales literature carefully when com-paring different devices.

When used for density measure-ment, the typical error range of aCoriolis measurement is 0.002-0.0005 g/cc.

Errors are caused by air or gaspockets in the process fluid. In thecase of homogeneously dispersedsmall bubbles, more power isrequired to vibrate the tubes, where-as, if the gas phase separates fromthe liquid, a damping effect on tubevibration (and, consequently, error)develops. Small voids also causenoise because of the sloshing of theprocess liquid in the tubes. Larger

voids will raise the energy requiredto vibrate the tubes to excessive lev-els and may cause complete failure.

Because the flowtube is subjectedto axial, bending, and torsional forcesduring meter operation, if process or

ambient temperature and pressurefluctuations alter these forces, perfor-mance may be affected and re-zeroingof the meter may be required.

Variations in the density of theprocess fluid can affect the frequencytransfer function of mechanical sys-tems, necessitating the re-zeroing ofolder designs to protect them fromdegraded performance. Because oftheir tube configurations, newerdesigns are unaffected by densitychanges over wide ranges of specificgravity variations.

• Sizing & Pressure DropBecause of the wide rangeability ofCoriolis flowmeters (30:1 to as high as200:1), the same flow can be mea-sured by two or three different sized

flow tubes. By using the smallestpossible meter, one will lower theinitial cost and reduce coating build-up, but will increase erosion/corro-sion rates and head loss, increasingpumping and operating costs.

Downsizing (using a meter that issmaller than the pipe) is acceptablewhen the pipe is oversized and theprocess fluid is clean with a low vis-cosity. On corrosive, viscous, or abra-sive slurry services, downsizing is notrecommended. A list of acceptableflow tube sizes and correspondingpressure drops, inaccuracies, and flowvelocities can be obtained from soft-ware provided by the manufacturer.

Different Coriolis meters incur dif-ferent pressure drops, but in generalthey require more than traditionalvolumetric meters, which usuallyoperate at less than 10 psid. (Theyearly electricity cost of pumping1 gpm across a differential of 10 psidis about $5 U.S.). This higher head lossis due to the reduced tubing diameter

and the circuitous path of flow.Besides pumping costs, head loss canbe of concern if the meter is installedin a low-pressure system, or if there isa potential for cavitation or flashing,or if the fluid viscosity is very high.

5 Mass Flowmeters


Figure 5-6: Installation Variations of the Coriolis Meter

B) VerticalA) Horizontal

C) Pipe Supports

SupportFlanges

Mass Flowtube Enclosure

Pipe/Flowtube Junction

NOTE: Distance Between

Pipe/Flowtube Junction and

Support Must Not

Exceed 15 Inches

Flow Direction Arrow

Mass Tube Enclosure

Support (Typical)

Flow Direction

Arrow

NOTE: Distance Between Pipe/Flowtube Junction and

Support Must Not Exceed 15 Inches

'U' Rest 'V' Rest 'V' Bolt Clamp

Inverted Pipe Hanger Clamp

'V' Block Clamp (Can Be Inverted)

The viscosity of non-Newtonianfluids is a function of their flowingvelocity. Dilettante fluids, for exam-ple, increase their apparent viscosity(resistance to flow) as their velocityis increased. This apparent viscositycan be drastically higher than their

viscosity when stagnant. In order toprovide suppliers with data on theflowing viscosity in a particular pipe,head loss per foot of pipe (used inpump sizing calculations) can beused as an approximation.

• Applications & LimitationsCoriolis mass flowmeters can detectthe flow of all liquids, includingNewtonian and non-Newtonian, aswell as that of moderately densegases. Self-draining designs are avail-able for sanitary applications thatmeet clean-in-place requirements.

Most meters are provided with

intrinsically safe circuits between theflow tube and the transmitter.Therefore, the amount of drivingpower that can be delivered to theflow tube is limited.

When fluid is unloaded from tanktrucks, drums, or railroad cars, slug

flow can occur, making the meter out-put unpredictable. If a slug-flow recov-ery feature is provided in the transmit-ter, it will stop the measurement whenslug flow is detected by the excessivedrive power drawn or by the drop inprocess density (reduction in sensoroutput amplitude).

The amount of air in the processfluid that can be tolerated by a metervaries with the viscosity of the fluid.Liquids with viscosities as high as300,000 centipoise can be meteredwith Coriolis meters. Gas content insuch highly viscous liquids can be ashigh as 20% with the small bubbles

still remaining homogeneously dis-persed. Gas content in low viscosityfluids, like milk, will separate at con-centrations as low as 1%.

The cost of an average-sized(under 2 in.) Coriolis flowmeter isbetween $4,000 and $5,000. These

mass flowmeters provide short pay-back periods on applications wheremeasurement accuracy lowers pro-duction costs (bathing, billing) orwhere multiple measurements(including density, temperature, pres-sure) are needed. On the other hand,they may not be competitive whenused in simple flow measurementapplications where volumetric sen-sors are sufficient and whererepeatability is more important thanprecision. The ability to extract dataon total mass charged, solids rate,percent solids, and viscosity from asingle instrument does lower the

Mass Flowmeters 5


Figure 5-7: Air Releases Installed Upstream From the Meter

B) Open Vent LinesA) Closed Vent Lines

C) Typical Installation

TankValve

Pump

Valve

Bidirectional Flow

Air Eliminator

Meter

Check Valve

Back Pressure Valve

Unload

Load

Check Valve

total cost of measurement, improvesprocess control, and provides redun-dancy for other instruments.

Continuous tube designs are gen-erally preferred for slurry and othermulti-phase fluid applications. Thetotal flow is divided by splitters insplit-tube designs, and the resultingtwo streams do not have to be atexactly the same mass flow rate tomaintain accuracy (they do, however,need to have the same density).Different densities in the two paral-lel tubes imbalance the system andcreate measurement errors.Therefore, if there is a secondaryphase in the stream, a simple flowsplitter may not evenly distributethe flow between the two tubes.

Continuous tube designs are alsopreferred for measuring fluids thatcan coat and/or clog the meter.

Continuous tubes, if sized to pass thelargest solid particles in the processfluid, are less likely to clog and areeasier to clean.

Straight-tube designs can be

cleaned by mechanical means, whilecurved-tube designs are usuallywashed out using cleaning solutionsat velocities in excess of 10 ft/sec.Straight-tube designs also are pre-ferred for sanitary applications dueto self-draining requirements.

Long, bent tubes twist more easilythan do short, straight tubes andtherefore will generate stronger sig-nals under the same conditions. Ingeneral, curved-tube designs providewider rangeability (100:1 to 200:1),while straight-tube meters are limitedto 30:1 to 50:1, with lower accuracy.

Straight-tube meters are moreimmune to pipeline stresses andvibration, are easy to install, requireless pressure drop, can be mechani-cally cleaned, are more compact, andrequire less room for installation.They are also preferred on services

where the process fluid can solidifyat ambient temperatures.

Not all meter housings aredesigned to withstand and containthe pressurized process fluid in case

of tube rupture, particularly if theprocess fluid is likely to vaporizeunder such conditions. If that is thecase, secondary containment hous-ings can be ordered that enclose theentire flow tube, including its hous-ing. Such secondary containmentenclosures can be provided with rup-ture disks or pressure relief valves,and with drains or vents.

• Installation RecommendationsThere are no Reynolds number limi-tations associated with Coriolismeters. They are also insensitive tovelocity profile distortion and swirl.Therefore, there is no requirementfor straight runs of relaxation pipingupstream or downstream of themeter to condition the flow.

The meter should be installed sothat it will remain full of liquid and

so air cannot get trapped inside thetubes. In sanitary installations, themeter must also drain completely. Themost desirable installation is in verticalupward flow pipes (Figure 5-6B), but

5 Mass Flowmeters


Figure 5-8: Thermal Mass Flowmeter Designs

B) Externally-Heated TubeA) Immersion Heater

T2

Constant Power

Source (q)

Resistance Thermometer

Resistance Thermometer

Resistance Heater

T1

Temperature Difference Indicator

Mass Flow Rate

Flow

Heater Supplies

q

Upstream Temperature

Sensor Measures Tf

Film

Downstream Temperature Sensor

Measures Tw

installations in horizontal lines (Figure5-6A) are also acceptable. Installationswhere the flow is downward in a verti-cal pipe are not recommended.

In newer Coriolis designs, normalpipe vibration should not affect theperformance of the Coriolis meter ifit is properly supported by theprocess piping (Figure 5-6C). No spe-cial supports or pads are needed forthe flow tube, but standard pipingsupports should be located on eitherside of the meter. If the installationinstructions require special hardwareor supports, the particular meterdesign is likely to be sensitive tovibration, and the pulsation dampen-ers, flexible connectors, and mount-ing/clamping attachments recom-mended by the manufacturer shouldbe carefully installed.

If your application requires thatyou install two Coriolis flowmetersin series or mount two Coriolis metersnear each other, the manufacturershould be consulted to preventcrosstalk between the two units.

If air bubbles are likely to be presentin the process fluid, it is recommendedto install an air release upstream of themeter. System design characteristicsthat can result in the presence of air(and which can often be eliminated atthe design stage) include:• Common piping used for pumping

into and out of storage tanks;• Allowing the formation of a vortex

in stirred vessels under low-levelconditions;

• Allowing air leakage through pack-ing glands of pumps that develophigh vacuums on the suction side(this can occur when pumpingfrom underground storage);

• Vaporization of stagnant processfluid in pipes exposed to the sun;

• High valve pressure drops causingvaporization and flashing;

• Allowing the pipe to drain for anyreason, including lack of checkvalves;

• Allowing storage tanks, trucks, orrailroad cars to drain completely;

• Using the same pipe for pumping

different materials at differenttimes; and

• Allowing foam formation by highturbulence in high velocity fluids.It is recommended to install

(upstream of the meter) strainers, fil-ters or air/vapor eliminators asrequired to remove all undesirable sec-ondary phases. Figure 5-7C illustratesan air eliminator installation. Its func-tion is to slow the velocity of the liq-uid, thereby allowing time for theentrained air to separate and beremoved by venting. The rise and fallof the liquid level in the eliminatordue to the accumulation of free aircloses and opens the vent valve anddischarges the air (Figure 5-7A&B).

Prior to zeroing the meter, all airshould be removed. This can beaccomplished by circulating the

process fluid through the meter forseveral minutes at a velocity ofapproximately 2-6 ft/sec. On batchingor other intermittent flow applica-tions, the meter should stay floodedso that it does not need to be

Mass Flowmeters 5


Figure 5-9: The Bypass Flowmeter Design

B) Temperature ProfileA) Bypass Uses Small Percent of Stream

Main Orifice

Bypass Orifice

Meter

Zero Flow

Small Flow

TC-2

L/20

Length of Tube

L/2

TC-1

Tem

pera

ture

of T

ube

repurged. All meters should be soinstalled so they can be zeroed whilefilled with liquid.

When zeroing the meter, anyassociated pumps or other equip-ment should be running so that their

noise can be zeroed out. This can beachieved in most cases by locating ashut-off value downstream of themeter and either operating thepump with its discharge blocked,which is acceptable with centrifugalpumps for a short period, or byopening the pump bypass on posi-tive displacement pumps. Valvesused in zeroing the meter shouldprovide tight shut-off; double-seat-ed valves are preferred.

Meters that are expected to becalibrated in-line must be providedwith block and bypass valves so thatthe reference standard (master)meter can be installed and discon-nected without interrupting theprocess. The requirements for in-linecalibration (for ISO 9000 verifica-tion) consist of comparing the out-put of the meter against a reference

standard of higher accuracy, such asa dead-weight calibrated weigh tank.Before Coriolis meters, the referencestandard was expected to be anorder of magnitude more accuratethan the meter being calibrated;

however, due to the high accuracy ofCoriolis meters, this is rare.

In less critical installations (whereweigh tanks are not used), volumetricprovers or master meters (typicallyanother Coriolis or a turbine metercalibrated at a flow laboratory) areused. When a volumetric reference isused in calibrating a mass flowmeter,the fluid density must be very pre-cisely determined.

Control valves should be installeddownstream of the meter to increasethe back-pressure on the meter andlower the probability of cavitation orflashing.

When the process fluid must beheld at higher temperatures, someCoriolis meters can be supplied withsteam jackets. As an alternative, elec-trical heating tape can be added tothe housing. Jackets or heating tapes

must be installed by the manufacturer.When flowmetering is not required,

the Coriolis meter can be used solelyas a densitometer. In that case, to min-imize cost, usually a small (H in.) meteris installed in a by-pass line. Such a

configuration is acceptable only inclean services that will not clog thesmall bore of the meter. In addition, arestriction must be placed in the mainpiping (between the by-pass taps) toensure a flow through the meter.

Thermal Mass Flowmeters Thermal mass flowmeters also mea-sure the mass flowrate of gases andliquids directly. Volumetric measure-ments are affected by all ambientand process conditions that influ-ence unit volume or indirectly affectpressure drop, while mass flow mea-surement is unaffected by changes inviscosity, density, temperature, orpressure.

Thermal mass flowmeters areoften used in monitoring or control-ling mass-related processes such aschemical reactions that depend on

5 Mass Flowmeters


Figure 5-10: Thermal Velocity Sensor

B) Venturi InsertionA) Probe Configuration

Temperature Sensor 0.125" Dia.

Air Velocity Sensor 0.032" Dia.

Flow

0.25" Dia.

.125".125"

.938Inlet

ScreensFlow Element

(Heated)Temperature Compensator

the relative masses of unreactedingredients. In detecting the massflow of compressible vapors andgases, the measurement is unaffected

by changes in pressure and/or tem-perature. One of the capabilities ofthermal mass flowmeters is to accu-rately measure low gas flowrates orlow gas velocities (under 25 ft. perminute)—much lower than can bedetected with any other device.

Thermal flowmeters provide highrangeability (10:1 to 100:1) if they areoperated in constant-temperature-dif-ference mode. On the other hand, ifheat input is constant, the ability todetect very small temperature differ-ences is limited and both precisionand rangeability drop off. At normalflows, measurement errors are usuallyin the 1-2% full scale range.

This meter is available in high pres-sure and high temperature designs,and in special materials includingglass, Monel, and Teflon®. Flow-through designs are used to measuresmall flows of pure substances (heatcapacity is constant if a gas is pure),while bypass and probe-type designscan detect large flows in ducts, flarestacks, and dryers.

• Theory of OperationThermal mass flowmeters are mostoften used for the regulation of low

gas flows. They operate either byintroducing a known amount of heatinto the flowing stream and measuringan associated temperature change, orby maintaining a probe at a constanttemperature and measuring the energyrequired to do so. The components ofa basic thermal mass flowmeterinclude two temperature sensors andan electric heater between them. Theheater can protrude into the fluidstream (Figure 5-8A) or can be externalto the pipe (Figure 5-8B).

In the direct-heat version, a fixedamount of heat (q) is added by anelectric heater. As the process fluidflows through the pipe, resistancetemperature detectors (RTDs) mea-sure the temperature rise, while theamount of electric heat introducedis held constant.

The mass flow (m) is calculated on

the basis of the measured tempera-ture difference (T2 - T1), the metercoefficient (K), the electric heat rate(q), and the specific heat of the fluid(Cp), as follows:

m = Kq/(Cp (T2 - T1))

• Heated-Tube DesignHeated-tube flowmeters were devel-oped to protect the heater and sen-sor elements from corrosion and anycoating effects of the process. Bymounting the sensors externally tothe piping (Figure 5-8B), the sensingelements respond more slowly andthe relationship between mass flowand temperature difference becomesnonlinear. This nonlinearity resultsfrom the fact that the heat intro-duced is distributed over some por-tion of the pipe’s surface and trans-ferred to the process fluid at differentrates along the length of the pipe.

The pipe wall temperature ishighest near the heater (detected as

Tw in Figure 5-8B), while, some dis-tance away, there is no differencebetween wall and fluid temperature.Therefore, the temperature of theunheated fluid (Tf) can be detected by

Mass Flowmeters 5


Figure 5-11: Hot Wire Anemometer

Gas Stream

Needle

Holder

Hot Wire Element

All-in-one mass flow controller provides both measurement and control of relatively low mass flow rates.

measuring the wall temperature atthis location further away from theheater. This heat transfer process isnon-linear, and the correspondingequation differs from the one aboveas follows:

m0.8 = Kq/(Cp (Tw - Tf))

This flowmeter has two operatingmodes: one measures the mass flowby keeping the electric power inputconstant and detecting the tempera-ture rise. The other mode holds thetemperature difference constant andmeasures the amount of electricity

needed to maintain it. This secondmode of operation provides for amuch higher meter rangeability.

Heated-tube designs are generallyused for the measurement of clean(e.g., bottled gases) and homoge-neous (no mixtures) flows at moder-ate temperature ranges. They are notrecommended for applicationswhere either the fluid compositionor its moisture content is variable,because the specific heat (Cp) wouldchange. They are not affected bychanges in pressure or temperature.Advantages include wide rangeability(the ability to measure very lowflows) and ease of maintenance. Thetemperature difference (or heaterpower), flowmeter geometry, thermalcapacity, specific heat, and viscosity

of the process fluid must stay con-stant when using this design.

• Bypass-Type DesignThe bypass version of the thermalmass flowmeter was developed tomeasure larger flow rates. It consistsof a thin-walled capillary tube(approximately 0.125 in diameter) andtwo externally wound self-heatingresistance temperature detectors(RTDs) that both heat the tube andmeasure the resulting temperaturerise (Figure 5-9A). The meter is placedin a bypass around a restriction in themain pipe and is sized to operate in

the laminar flow region over its fulloperating range.

When there is no flow, the heatersraise the bypass-tube temperature toapproximately 160°F above ambienttemperature. Under this condition, asymmetrical temperature distribu-tion exists along the length of thetube (Figure 5-9B). When flow is tak-ing place, the gas molecules carry theheat downstream and the tempera-ture profile is shifted in the directionof the flow. A Wheatstone bridgeconnected to the sensor terminalsconverts the electrical signal into amass flow rate proportional to thechange in temperature.

The small size of the bypass tubemakes it possible to minimize electricpower consumption and to increase

the speed of response of the mea-surement. On the other hand,because of the small size, filters arenecessary to prevent plugging. Oneserious limitation is the high pressuredrop (up to 45 psi) needed to developlaminar flow. This is typically accept-able only for high pressure gas appli-cations where the pressure needs tobe reduced in any case.

This is a low accuracy (2% fullscale), low maintenance, and lowcost flowmeter. Electronic packageswithin the units allow for data acqui-sition, chart recording, and computerinterfacing. These devices are popular

in the semiconductor processingindustry. Modern day units are alsoavailable as complete control loops,including a controller and automaticcontrol valve.

• Air Velocity ProbesProbe-style mass flowmeters areused to measure air flows and areinsensitive to the presence of mod-erate amounts of dust. They maintaina temperature differential betweentwo RTDs mounted on the sensortube. The upper sensor measures theambient temperature of the gas(Figure 5-10A) and continuouslymaintains the second RTD (near thetip of the probe) at 60°F above ambi-ent. The higher the gas velocity, themore current is required to maintain

5 Mass Flowmeters


Figure 5-12: Circuling and Rectangular Measuring Stations

A) B)

1

1

1

2

22

33

3 4

44

5

55

6

66

60° 60°

the temperature differential.Another version of the velocity

probe is the venturi-type thermalmass flowmeter, which places aheated mass flow sensor at the mini-mum diameter of a venturi flow ele-ment and a temperature compensa-tion probe downstream (Figure 5-10B). An inlet screen mixes the flowto make the temperature uniform.This design is used for both gas andliquid measurement (including slur-ries), with flow range a function ofthe size of the venturi. Pressure dropis relatively low and precision isdependent upon finding the properprobe insertion depth.

A flow switch version is also avail-able that contains two temperaturesensors in the tip. One of the sensorsis heated and the temperature differ-ence is a measure of velocity. Theswitch can be used to detect high orlow flow within 5%.

• Uses & LimitationsThermal mass flowmeters can havevery high rangeability and reasonableaccuracy, but they also have seriouslimitations. Potential problems includethe condensation of moisture (in satu-rated gases) on the temperaturedetector. Such condensation willcause the thermometer to read lowand can lead to corrosion. Coating ormaterial build-up on the sensor alsowill inhibit heat transfer and causethe meter to read low. Additionalpotential sources of error includevariations in the specific heat causedby changes in the gas’s composition.

Some common gas-flow applica-tions for thermal mass flowmetersinclude combustion air measurementin large boilers, semiconductorprocess gas measurement, air sam-pling in nuclear power plants,process gas measurements in the

chemical and petrochemical indus-tries, research and developmentapplications, gas chromatography,and filter and leak testing. While hot-wire anemometers are best suited forclean gases at low velocities, venturimeters can also be considered forsome liquid (including slurry) flow

applications. Thermal mass flowme-ters are well suited for high range-ability measurements of very lowflows, but also can be used in mea-suring large flows such as combus-tion air, natural gas, or the distribu-tion of compressed air.

Hot-Wire AnemometersThe term anemometer was derivedfrom the Greek words anemos,“wind,” and metron, “measure.”Mechanical anemometers were firstdeveloped back in the 15th centuryto measure wind speed.

A hot-wire anemometer consistsof an electrically heated, fine-wireelement (0.00016 inch in diameterand 0.05 inch long) supported byneedles at its ends (Figure 5-11).Tungsten is used as the wire materi-al because of its strength and hightemperature coefficient of resis-tance. When placed in a movingstream of gas, the wire cools; the

rate of cooling corresponds to themass flowrate.

The circuitry of the heated sensingelement is controlled by one of twotypes of solid-state electronic circuits:constant-temperature or constant-power. The constant-temperature sen-sor maintains a constant temperature

differential between a heated sensorand a reference sensor; the amount ofpower required to maintain the differ-ential is measured as an indication ofthe mass flow rate.

Constant-temperature anemome-ters are popular because of theirhigh-frequency response, low elec-tronic noise level, immunity fromsensor burnout when airflow sud-denly drops, compatibility with hot-film sensors, and their applicabilityto liquid or gas flows.

Constant-power anemometers donot have a feedback system.Temperature is simply proportionalto flowrate. They are less popularbecause their zero-flow reading isnot stable, temperature and velocityresponse is slow, and temperaturecompensation is limited.

• Air Duct TraversingAnemometers are widely used for airduct balancing. This is accomplished

Mass Flowmeters 5


Air velocity probe provides 1.5% accuracy for local flow rate measurement.

by placing multiple anemometers in across-section of the duct or gas pipeand manually recording the velocityreadings at numerous points. The massflow rate is obtained by calculating themean velocity and multiplying this bythe density and by the cross-sectionalarea measurement of the duct.

For cylindrical ducts, the log-linearmethod of traversing provides thehighest accuracy because it takesinto account the effects of frictionalong the walls of the duct. Becauseof the number of measurements(Figure 5-12), air duct traversing is atime-consuming task. Microprocessor-based anemometers are available toautomate this procedure.

Because of the small size andfragility of the wire, hot-wireanemometers are susceptible to dirtbuild-up and breakage. A positive con-sequence of their small mass is fastspeed of response. They are widelyused in HVAC and ventilation applica-tions. Larger and more ruggedanemometers are also available formore demanding industrial applica-tions. To ensure the proper formationof the velocity profile, a straight ductsection is usually provided upstreamof the anemometer station (usually 10diameters long). A conditioning nozzleis used to eliminate boundary layereffects. If there is no room for thestraight pipe section, a honeycombflow straightener can be incorporatedinto the sensor assembly. T

5 Mass Flowmeters




Edition, OMEGA Press, 1995.• “Air Elimination Techniques for Accurate Liquid Measurement,” J. R.

Chester, Mechanical Engineering, February 1983.• “Application and Installation Guidelines for Volumetric and Mass

Flowmeters,” D. Ginesi and C. Annarummo, ISA Transactions, InstrumentSociety of America, 1994.

• Automated Process Control Electronics, John Harrington, DelmarPublishing Inc., 1989.

• “Coriolis for the Masses,” G. J. Blickley, Control Engineering, June 1995.• “Coriolis Mass Flowmeter is Ready for the Tough Jobs,” W. Chin, I&CS,

February 1992.• “Field Proving Coriolis Mass Flowmeter,” R. Harold and C. Strawn, ISA/91

Proceedings, Instrument Society of America, 1991.• Flow Measurement, D.W. Spitzer (editor), Instrument Society of America,

1991.• “Flow Sensing: The Next Generation,” D. Ginesi, Control Engineering,

November 1997.• Instrument Engineers’ Handbook, Bela Liptak, CRC Press, 1995.• Instrumentation for Process Measurement and Control, 3rd edition,

Norman A. Anderson, Chilton Co., 1980.• Instruments of Science, Robert Bud and Deborah Jean Warner, Garland

Publishing Inc., 1998.• “Metering Mass Flow,” H. van der Bent, Process Engineering, May 1993.• “On-line Viscosity Measurement with Coriolis Mass Flowmeters,” P.

Kalotry and D. Schaffer, ISA/91 Proceedings, Instrument Society ofAmerica, 1991.

• Process/Industrial Instruments and Controls Handbook, 4th edition,Douglas M. Considine, McGraw-Hill, 1993.

• “Technical Application Guide to Mass Flow Measurement,” WayneShannon, Magnetrol International, 1998.

• The McGraw-Hill Encyclopedia of Science and Technology, 8th edition,John H. Zifcak, McGraw-Hill, 1997.

On the 28th of March, 1979,thousands of people fledfrom Three Mile Island (nearHarrisburg, PA) when the

cooling system of a nuclear reactorfailed. This dangerous situation

developed because the level con-trols turned off the coolant flow tothe reactor when they detected thepresence of cooling water near thetop of the tank. Unfortunately, thewater reached the top of the reac-tor vessel not because there wastoo much water in the tank, butbecause there was so little that itboiled and swelled to the top. Fromthis example, we can see that levelmeasurement is more complex thansimply the determination of thepresence or absence of a fluid at aparticular elevation.

Level Sensor SelectionWhen determining what type oflevel sensor should be used for agiven application, there are a seriesof questions that must be answered:• Can the level sensor be inserted

into the tank or should it be com-pletely external?

• Should the sensor detect the levelcontinuously or will a point sensorbe adequate?

• Can the sensor come in contact

with the process fluid or must itbe located in the vapor space?

• Is direct measurement of the levelneeded or is indirect detection ofhydrostatic head (which respondsto changes in both level and den-sity) acceptable?

• Is tank depressurization or processshut-down acceptable when sen-sor removal or maintenance isrequired?By evaluating the above choices,

one will substantially shorten the listof sensors to consider. The selectionis further narrowed by consideringonly those designs that can be pro-vided in the required materials ofconstruction and can function at therequired accuracy, operating temper-ature, etc. (Table 4). When the levelto be measured is a solid, slurry,foam, or the interface between two

liquid layers, it is advisable to consultnot only Table 4, but other recom-mendations, such as Table 5.

If it is found that a number of leveldetector designs can satisfy therequirements of the application, oneshould also consider the traditionsor preferences of the particular plantor the particular process industry,because of user familiarity and theavailability of spare parts. For exam-ple, the oil industry generally prefersdisplacement-type level sensors,while the chemical industry favorsdifferential pressure (d/p) cells. (Thepetroleum industry will use d/p cellswhen the span exceeds 60-80 in.)

If the tank is agitated, there isoften no space in which to insertprobe-type sensors. Plus, becausethe liquid surface is not flat, sonic,ultrasonic, or radar devices typicallycannot be used, either. Even with dis-placer or d/p sensors, agitation cancause the level signal to cycle. Thesepulses can be filtered out by firstdetermining the maximum rate at

which the level can change (due tofilling or discharging) and disregard-ing any change that occurs faster


Level Sensor Selection

Boiling & Cryogenic Fluids

Sludge, Foam, & Molten Metals

FLOW & LEVEL MEASUREMENTA Level Measurement Orientation

6

OA Level Measurement Orientation

Figure 6-1: Relationship Between Level and Volume for Different Tanks

VerticalSphere

Horizontal Cylindrical

50

0 100 Volume %

100

50

Level %

Figure 6-2: Intelligent Multi-Transmitter Package

"Top" Pressure

"Middle" Pressure

Mass ± 0.2% Density ± 0.3% Volume ± 0.28%

"Bottom" Pressure

Temperature

PT

PT RTD

PT

than that.The relationship between level

and tank volume is a function of thecross-sectional shape of the tank.With vertical tanks, this relationshipis linear, while with horizontal orspherical vessels, it is a non-linearrelationship (Figure 6-1).

If the level in a tank is to beinferred using hydrostatic pressuremeasurement, it is necessary to use

multi-transmitter systems when it isdesirable to: • Detect the true level, while either

the process temperature or densi-ty varies;

• Measure both level and density;and

• Measure the volume and the mass(weight) in the tank.By measuring one temperature and

three pressures, the system shown in

Figure 6-2 is capable of simultane-ously measuring volume (level), mass(weight), and density, all with anaccuracy of 0.3% of full span.

Boiling & Cryogenic FluidsWhen a d/p cell is used to measurethe level in a steam drum, a reverse-acting transmitter is usually installed(Figure 6-3). An uninsulated condens-ing chamber is used to connect the

6 A Level Measurement Orientation


TYPE

LIMITATIONS

APPLICATIONS

LIQUIDS

SOLIDS

MAX

. TEM

P. (°F

)

AVAI

LABL

E AS

NONC

ONT

ACT

INAC

CURA

CY

(1 in

. = 25

.4 m

m)

CLEA

N

VISC

OUS

SLUR

RY/S

LUDG

E

INTE

RFAC

E

FOAM

POW

DER

CHUN

KY

STIC

KY

Table 4: Orientation Table for Selecting Level Sensors

Air Bubblers Capacitance Conductivity Switch Diaphragm Differential Pressure Displacer Float Laser Level Gages Microwave Switches Optical Switches Radar Radiation Resistance Tape Rotating Paddle Switch Slip Tubes Tape-Type Level Sensors Thermal TDR/PDS Ultrasonic Vibrating Switches

Introduces foreign substance into process; high maintenance

Interface between conductive layers and detection of foam is a problem

Can detect interface only between conductive and nonconductive liquids. Field effect design for solids

Switches only for solid service

Only extended diaphragm seals or repeaters can eliminate plugging. Purging and sealing legs are also used

Not recommended for sludge or slurry service

Moving parts limit most designs to clean service. Only preset density floats can follow interfaces

Limited to cloudy liquids or bright solids in tanks with transparent vapor spaces

Glass is not allowed in some processes

Thick coating is a limitation

Refraction-type for clean liquids only; reflection-type requires clean vapor space

Interference from coating, agitator blades, spray, or excessive turbulence

Requires NRC license

Limited to liquids under near-atmospheric pressure and temperature conditions

Limited to detection of dry, non-corrosive, low-pressure solids

An unsafe manual device

Only the inductively coupled float is suited for interface measurement. Float hangup is a potential problem with most designs

Foam and interface detectiom is limited by the thermal conductives involved

Limited performance on sticky process materials

Presence of dust, foam, dew in vapor space; sloping or fluffy process material interferes with performance

Excessive material buildup can prevent operation

UL

2,000

1,800

350 1,200

850

500

UL

700 400 260

450

UL 225

500

200 300

850

221

300

300

X

X X

X

X

X

1-2% FS

1-2% FS

1/8 in

0.5% FS 0.1% AS

0.5% FS

1% FS

0.5 in

0.25 in 0.5 in

0.25 in

0.12 in

0.25 in 0.5 in

1 in

0.5 in 0.1 in

0.5 in

3 in

1% FS

0.2 in

G

G F

G E

E

G

L

G G G

G

G G F E

G F

F-G F

F

F-G

P F

G-E

P

P

G F G F

G

E G

P F F F

G

G

P F F F G

P

P

G

P F E F

E G

P P F F

G

G

F

G-L

L

P

F-G F F G

F-G

P

G

G

P

F-G F

P

L F F F F F

P

L

P

F F F

P

E

P

F

F

G

G

F

L F

F

G P F

E F

F

G F

G

F

L F

F

G F

P

G

G

G

G F F

TDR = Time Domain Reflectometry PDS = Phase Difference Sensors AS = in % of actual span E = Excellent FS = in % of full scale

F = Fair G = Good L = Limited P = Poor UL = Unlimited

high pressure (HP) side of the d/pcell to the vapor space on the top ofthe drum. The steam condenses inthis chamber and fills the wet legwith ambient temperature water,while the low pressure (LP) side ofthe d/p cell detects the hydrostatichead of the boiling water inside thedrum. The output of the d/p cell

reflects the amount of water in thedrum. Output rises as the mass ofwater in the drum drops (because thesteaming rate and the associatedswelling increase). It is for this reasonthat a reverse acting d/p cell is rec-ommended for this application.

When the process fluid is liquidnitrogen (or some other cryogenicmaterial), the tank is usually sur-rounded by a thermally insulatedand evacuated cold box. Here, thelow pressure (LP) side of a directacting d/p cell is connected to thevapor space above the cryogenicliquid (Figure 6-4). As the liquidnitrogen approaches the HP side ofthe d/p cell (which is at ambienttemperature outside the cold box),its temperature rises. When thetemperature reaches the boilingpoint of nitrogen, it will boil and,

from that point on, the connectingline will be filled with nitrogenvapor. This can cause noise in thelevel measurement. To protectagainst this, the liquid filled portionof the connecting line should besloped back towards the tank. Thecross-section of the line should belarge (about 1 inch in diameter) to

A Level Measurement Orientation 6


Figure 6-3: Wet Leg on a Steam Drum

Condensing Chamber

HPLT

LP(Reverse)

Slope

Steam Drum

P

C

LIQUID/ LIQUID

INTERFACE

FOAM

SLURRY

SUSPENDED SOLIDS

POWDERY SOLIDS

GRANULAR SOLIDS

CHUNKY SOLIDS

STICKY MOIST SOLIDS

LIQUIDS

Table 5: The Applicability of Level Sensors for Different Services

Beam Breaker

Bubbler

Capacitance

Conductive

Differential Pressure

Diaphragm

Displacer

Float

Float/Tape

Paddlewheel

Weight/Cable

Glass

Magnetic

Inductive

Microwave

Radiation

Sonar

Sonic

Ultrasonic

Thermal

Vibration

-

1

1

1

2

1

1

1

3

-

3

1

1

-

1

1

-

1

1

1

2

-

-

2

-

-

3

-

-

-

-

1

-

-

3

1

1

-

1

2

-

-

1

-

1

1

-

2

-

-

-

2

-

-

-

3

1

1

-

2

2

-

1

-

-

2

-

-

-

-

-

-

-

1

-

-

2

1

1

-

1

2

-

-

3

-

2

3

-

3

-

-

-

3

-

-

-

2

1

1

-

1

1

-

2

-

-

2

-

-

-

-

-

-

-

1

-

-

2

1

1

-

1

2

-

-

1

-

1

3

-

1

-

-

-

1

-

-

-

2

1

1

-

1

2

-

1

-

-

2

-

3

3

-

-

-

-

1

-

-

2

2

1

-

3

3

-

-

1

-

2

3

3

1

-

-

-

2

-

-

-

2

1

1

-

-

-

-

1

-

-

-

-

-

-

-

-

-

-

1

-

-

-

-

-

1

2

1

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

-

1

2

1

-

1

-

2

2

-

2

2

2

-

3

-

1

3

3

-

1

1

3

1

2

-

-

-

3

1

1

2

2

3

3

-

3

-

3

3

2

1

1

-

1

1

2

2

-

-

2

-

-

-

-

-

-

-

-

3

3

-

-

-

-

-

-

-

-

2

-

1

1

-

-

-

-

-

-

-

3

3

-

-

-

-

-

-

2

-

-

-

1

-

2

-

2

-

-

-

-

2

-

-

-

-

2

3

2

-

-

-

-

1

2

2

2

2

2

-

-

-

2

-

-

-

-

2

3

2

1

3

-

1

1

-

1

1

2

-

1

-

1

1

1

-

1

1

-

1

2

-

-

P

C

P

C

P

C

P

C

P

C

P

C

P

C

P

C

Elec

trom

echa

nica

l Ga

ges

Soni

c Ech

o

Figure 6-4: Thermally Insulated Cold Box

Superheated N2 Vapor

Boiling Liquid

HPLP

N2 Liquid

Thermal Insulation in the "Cold Box"

LT

minimize the turbulence caused bythe simultaneous boiling and re-condensing occurring at the liquid-vapor interface.

Sludge, Foam, & Molten MetalsMany process fluids are aggressive ordifficult to handle and it’s best to

avoid physical contact with them.This can be accomplished by placingthe level sensor outside the tank(weighing, radiation) or locating thesensor in the vapor space (ultrasonic,radar, microwave) above the processfluid. When these options are notavailable or acceptable, one must aimto minimize maintenance and physi-cal contact with the process fluid.

When the process fluid is a sludge,slurry, or a highly viscous polymer,and the goal is to detect the level atone point, the design shown in Figure6-5A is commonly considered. Theultrasonic or optical signal sourceand receiver typically are separatedby more than six inches so that theprocess fluid drains freely from theintervening space. After a high-level

episode, an automatic washing sprayis activated.

When the sludge or slurry level isdetected continuously, one of thegoals is to eliminate dead-ended cavi-ties where the sludge might settle. Inaddition, all surfaces which areexposed to the process fluid should

be covered with Teflon®. Figure 6-5Bshows such an installation, employingTeflon®-coated extended diaphragmsto minimize material buildup.

In strippers, where the goal is todrive off the solvent in the shortest

period of time, one aims to keep thefoam level below a maximum. Inother processes, it is desirable to sep-arately control both the liquid levelbeneath the foam and the thicknessof the foam. In the paper industry,beta radiation detectors are used forsuch applications (Kraft processing),while other industries detect thedegree of foaming indirectly (by mea-suring related variables, such as heatinput or vapor flow), or they usecapacitance, conductivity, tuningfork, optical, or thermal switches, allprovided with automatic washers.

Measuring the level of moltenglass or metals is another specialapplication. The most expensive(but also most accurate) techniqueavailable is proximity capacitance-based level measurement, whichcan provide a resolution of 0.1 mmover a range of 6 in. Laser-basedsystems can provide even betterresolution from distances up to 2 ft.If such high resolution is notrequired and cost is a concern, onecan make a float out of refractorymaterial and attach a linear variabledifferential transformer (LVDT), ormake a bubbler tube out of refrac-tory material and bubble argon ornitrogen through it. T

6 A Level Measurement Orientation


Figure 6-5: Level-Instrumentation Design For Sludges, Slurries, & High Viscosity

B) Continuous DetectionA) Level Switch

1:1 Repeater

Differential Pressure Transmitter

To Controller

Pv



Edition, OMEGA Press, 1995.• Instrument Engineer’s Handbook, Bela G. Liptak, editor, CRC Press, 1995.• Instrumentation for Process Measurement and Control, Third Edition, N.

A. Anderson, Chilton, 1980.• Measurement and Control of Liquid Level, C. H. Cho, Instrument Society

of America, 1982.• Principles of Industrial Measurement for Control Applications, E. Smith,

Instrument Society of America, 1984.

One of the primary principlesunderlying industrial levelmeasurement is that differ-ent materials and different

phases of the same material have dif-ferent densities. This basic law ofnature can be utilized to measurelevel via differential pressure (that atthe bottom of the tank relative tothat in the vapor space or to atmos-pheric pressure) or via a float or dis-placer that depends on the densitydifferences between phases.

Level measurement based on pres-sure measurement is also referred toas hydrostatic tank gaging (HTG). Itworks on the principle that the differ-ence between the two pressures (d/p)

is equal to the height of the liquid (h,in inches) multiplied by the specificgravity (SG) of the fluid (see Figure 7-1):

d/p = h (SG)

By definition, specific gravity is theliquid’s density divided by the densi-ty of pure water at 68° F at atmos-pheric pressure. A pressure gage ord/p cell can provide an indication of

level (accurate to better than 1%)over wide ranges, as long as the den-sity of the liquid is constant. When ad/p cell is used, it will cancel out theeffects of barometric pressure varia-tions because both the liquid in thetank and the low pressure side of thed/p cell are exposed to the pressureof the atmosphere (Figure 7-1B).Therefore, the d/p cell reading willrepresent the tank level.

Dry & Wet Leg DesignsWhen measuring the level in pressur-ized tanks, the same d/p cell designs(motion balance, force balance, orelectronic) are used as on open tanks.It is assumed that the weight of the

vapor column above the liquid is neg-ligible. On the other hand, the pres-sure in the vapor space cannot beneglected, but must be relayed to thelow pressure side of the d/p cell.Such a connection to the vapor spaceis called a dry leg, used when processvapors are non-corrosive, non-plug-ging, and when their condensationrates, at normal operating tempera-tures, are very low (Figure 7-1C). A dry

leg enables the d/p cell to compen-sate for the pressure pushing down onthe liquid’s surface, in the same way asthe effect of barometric pressure iscanceled out in open tanks.

It is important to keep this refer-ence leg dry because accumulation ofcondensate or other liquids wouldcause error in the level measurement.When the process vapors condense atnormal ambient temperatures or arecorrosive, this reference leg can befilled to form a wet leg. If the processcondensate is corrosive, unstable, orundesirable to use to fill the wet leg,this reference leg can be filled with aninert liquid.

In this case, two factors must be

considered. First, the specific gravityof the inert fluid (SGwl) and theheight (hwl) of the reference columnmust be accurately determined, andthe d/p cell must be depressed bythe equivalent of the hydrostatichead of that column [(SGwl)(hwl)].Second, it is desirable to provide asight flow indicator at the top of thewet leg so that the height of that ref-erence leg can be visually checked.


Dry & Wet Leg Designs

Bubbler Tubes

Floats & Displacers

FLOW & LEVEL MEASUREMENTPressure/Density Level Instrumentation

7

OPressure/Density Level Instrumentation

Figure 7-1: Hydrostatic Tank Gage

A) Local B) Remote C) Compensation

dp = h (SG)

Pneumatic Supply

dp = h (SG)

LI

h h

LT

LI

Pneumatic Supply

Dry Leg

h

LT

LI

Any changes in leg fill level (due toleakage or vaporization) introduceerror into the level measurement. Ifthe specific gravity of the filling fluidfor the wet leg is greater than that ofthe process fluid, the high pressureside should be connected to the ref-erence leg and the low to the tank.

If the condensate can be used tofill the reference leg, a condensatepot can be mounted and piped bothto the high level connection of thetank and to the top of the vaporspace. The condensate pot must bemounted slightly higher than thehigh level connection (tap) so that itwill maintain a constant condensatelevel. Excess liquid will drain backinto the tank. It is also desirableeither to install a level gage on thecondensate pot or to use a sight flowindicator in place of the pot, so thatthe level in the pot can convenientlybe inspected.

Either method (wet or dry)assures a constant reference leg forthe d/p cell, guaranteeing that theonly variable will be the level in the

tank. The required piping and valv-ing must always be provided onboth the tank and the reference legside of the d/p cell, so that drainingand flushing operations can easilybe performed. When a wet refer-ence leg is used, a low thermalexpansion filling fluid should beselected. Otherwise, the designermust correct for the density varia-tions in the reference leg caused byambient temperature variations.

If smart transmitters are used andif the filling fluid data is known, wet-leg temperature compensation canbe provided locally. Alternatively,the host or supervisory control sys-tem can perform the compensationcalculations.

If it is desired to keep the processvapors in the tank, a pressure repeatercan be used. These devices repeat thevapor pressure (or vacuum) and sendout an air signal identical to that ofthe vapor space. The measurementside of the repeater is connected tothe vapor space and its output signalto the low pressure side of the d/p

cell. If the tank connection is subjectto material build-up or plugging,extended diaphragm Type 1:1 repeaterscan be considered for the service(Figure 7-2).

While repeaters eliminate theerrors caused by wet legs, they dointroduce their own errors as a func-tion of the pressure repeated. Forexample, at 40 psig, repeater error isabout 2 in. At 400 psig, it is 20 in. Inmany applications, the former isacceptable but the latter is not.

• d/p Cells Because the designs of the variousd/p cells are discussed in detail inanother issue of Transactions, only abrief overview is provided here.

The motion balance cell is wellsuited for remote locations whereinstrument air or electric power arenot available. If a bellows is used asthe sensing element in a motion bal-ance d/p cell, an increase in the pres-sure on either side causes the corre-sponding bellows to contract (Figure7-3A). The bellows is connected to a

7 Pressure/Density Level Instrumentation


Figure 7-2: Extended Diaphragm Pressure Repeaters

B) VacuumA) Positive Pressure

Process Tank

To Vacuum Source

Atmospheric Air

20# Air Supply

Zero

Repeated Vacuum Pressure

Air Supply at 5 psi More Than Pressure

To Be Repeated

Relay

Vent

Low Pressure

Side Zero

High Pressure

Side

Repeated Positive Pressure

linkage assembly that converts the lin-ear motion of the bellows into arotary indicator motion, which can be

calibrated to indicate the tank level.In a force-balance type of d/p

cell, the sensing element (often adiaphragm) does not move. A forcebar is provided to maintain theforces acting on the diaphragm inequilibrium (Figure 7-3B). In pneumaticd/p cells, this is often achieved bythe use of a nozzle and flapperarrangement that guarantees that thepneumatic output signal will alwaysbe proportional to the differentialpressure across the cell. The output ofpneumatic d/p cells is linear and isusually ranged from 3 to 15 psig. Thelevels represented by such transmit-ted signals (pneumatic, electronic,fiberoptic or digital) can be displayedon local indicators or remote instru-ments. Pneumatic transmitters requirea compressed air (or nitrogen) supply.

Electronic d/p cells provide±0.5% of span or better precisiontypically conveyed via a 4-20 mAsignal. The range of these simpleand robust cells can be as narrow as

a draft range of 0-H inH2O or aswide as 0-1,000 psid. Some elec-tronic d/p cells can operate at line

pressures up to 4,500 psig at 250°F.The drift and inaccuracy of some ofthese units have been tested forperiods of up to 30 months, and theerrors did not exceed the ±0.5% ofspan limit.

• Difficult Process FluidsWhen the process fluid is a sludge, aviscous polymer or is otherwise hardto handle, the goal is to isolate thedirty process from the d/p cell. Aflat diaphragm can be bolted to ablock valve on the tank nozzle sothat the d/p cell can be removed forcleaning or replacement without tak-ing the tank out of service. If it isacceptable to take the tank out ofservice when d/p cell removal isneeded, an extended diaphragmdesign can be considered. In thiscase, the diaphragm extension fillsthe tank nozzle so that thediaphragm is flush with the insidesurface of the tank. This eliminatesdead ends or pockets where solids

can accumulate and affect theperformance of the cell. Flat andextended diaphragm-type d/p cells,

pressure repeaters, and chemicalseals are available to protect d/pcells under these conditions.

Chemical seals, or diaphragmpressure seals, are available with fillliquids such as water, glycol, alco-hol, and various oils. These seals areused when plugging or corrosioncan occur on both sides of the cell.A broad range of corrosion-resistantdiaphragm and lining materials isavailable. Teflon® lining is often usedto minimize material build-up andcoating. Level measurement accuracydoes suffer when these seals areused. Capillary tube lengths shouldbe as short as possible and thetubes should be shielded from thesun. In addition, either low thermalexpansion filling fluids should beused or ambient temperature com-pensation should be provided, asdiscussed in connection with wetlegs. If the seals leak, maintenanceof these systems is usually doneat the supplier’s factory due to

Pressure/Density Level Instrumentation 7


Figure 7-3: Differential Pressure Cell Designs

B) Force BalanceA) Motion Balance

Bimetallic Temperature Compensator

Low Pressure Side

High Pressure Side

Liquid Fill

Range Spring

Nozzle & Flapper

Feedback Bellows

Fulcrum & Seal

Force Bar

Low Pressure Side

Liquid-Filled Diaphragm

Capsule

Output

High Pressure Side

Pneumatic Relay

Air Supply

the complex evacuation and back-filling procedures involved.

Bubbler TubesBubbler tubes provide a simple andinexpensive but less accurate (±1-2%)level measurement system for corro-sive or slurry-type applications.Bubblers use compressed air or aninert gas (usually nitrogen) intro-duced through a dip pipe (Figure 7-4A).Gas flow is regulated at a constantrate (usually at about 500 cc/min). Adifferential pressure regulator acrossa rotameter maintains constant flow,while the tank level determines theback-pressure. As the level drops, the

back-pressure is proportionallyreduced and is read on a pressuregage calibrated in percent level or ona manometer or transmitter. The dippipe should have a relatively largediameter (about 2 in.) so that the pres-sure drop is negligible. The bottom

end of the dip pipe should be locatedfar enough above the tank bottom sothat sediment or sludge will not plugit. Also, its tip should be notchedwith a slot or “V” to ensure the for-mation of a uniform and continuousflow of small bubbles. An alternativeto locating the dip pipe in the tank isto place it in an external chamberconnected to the tank.

In pressurized tanks, two sets ofdip pipes are needed to measurethe level (Figure 7-4B). The twoback-pressures on the two dippipes can be connected to the twosides of a u-tube manometer, a dif-ferential pressure gage or a d/p

cell/transmitter. The pneumatic pip-ing or tubing in a bubbler systemshould be sloped toward the tank sothat condensed process vapors willdrain back into the tank if purge pres-sure is lost. The purge gas supplyshould be clean, dry, and available at

a pressure at least 10 psi greaterthan the expected maximum totalpressure required (when the tank isfull and the vapor pressure is at itsmaximum). An alternative to a con-tinuous bubbler is to use a handpump (similar to a bicycle tirepump) providing purge air onlywhen the level is being read.

Bubblers do consume inert gases,which can later accumulate andblanket processing equipment. Theyalso require maintenance to ensurethat the purge supply is always avail-able and that the system is properlyadjusted and calibrated. When allfactors are considered, d/p cells

typically are preferred to bubblers inthe majority of applications.

• Elevation & SuppressionIf the d/p cell is not located at an ele-vation that corresponds to 0% level inthe tank, it must be calibrated to



Figure 7-4: Bubbler Tube Measurement System

B) Closed TankA) Open Tank

Remotely Located

Components

SS

PCVP1

F1

SS F1dPCV

dPCV

L1

Equalizing LineTransmission Line

Dip Tube

Manometer

N2

P2

PNEUMATIC SUPPLY

(N2)

Flow @ P1 Inlet Pressure

Spring #2

Regulator Valve

Diaphragm

Flow Control Valve (V)

FloatTubeSpring

#1

L1

account for the difference in eleva-tion. This calibration adjustment iscalled zero elevation when the cell islocated above the lower tap, and iscalled zero suppression or zero

depression when the cell is locatedbelow the lower tap. Most d/p cellsare available with elevation and sup-pression ranges of 600% and 500% ofcalibrated span, respectively, as longas the calibrated span does notexceed 100% of the upper range limitof the cell.

For example, assume that an elec-tronic d/p cell can be calibrated forspans between 0-10 psid (which is itslower range limit, LRL) and 0-100 psid(which is its upper range limit, URL).The cell is to be used on a 45-ft tallclosed water tank, which requires ahydrostatic range of 0-20 psid. Thecell is located about 11 feet (5 psid)above the lower tap of the tank;therefore, a zero elevation of 5 psid isneeded. The d/p cell can handle thisapplication, because the calibratedspan is 20% of the URL and the eleva-tion is 25% of the calibrated span.

On interface level measurementapplications with a wet leg reference,

the high pressure side of the d/p cellshould be connected to the tank ifthe specific gravity of the wet leg fill-ing fluid is close to that of the lightlayer. It should be connected to the

reference leg if the wet-leg fluid’s SGis closer to that of the heavy layer.

• Special ApplicationsWhen the process fluid is boiling,such as in a steam drum, a wet refer-ence leg is maintained by a conden-sate pot, which drains back into thesteam drum so that the level of thewet leg is kept constant. Changes inambient temperature (or sun expo-sure) will change the water density inthe reference leg and, therefore,temperature compensation (manualor automatic) is needed.

Figure 7-5 describes a typicalpower plant steam drum level appli-cation. The differential pressuredetected by the level d/p cell is:

d/p = h1SG1 + h2SG2 - h3SG3d/p = 0.03h1 + 0.76h2 - 0.99h3

Note that the SG of the saturatedsteam layer (0.03) and that of the

saturated liquid layer (0.76) vary notonly with drum pressure but alsowith steaming rate. This causes theswelling of bubbles when thesteaming rate rises (and SG2 drops),as well as their collapse when thesteaming rate drops (and SG2 rises).Therefore, to make an accuratedetermination of both the level andthe mass of the water in the steamdrum, the calculation must considernot only the d/p cell output, butalso the drum pressure and the pre-vailing steaming rate.

• Tank FarmsComputerized tank farm systems usu-ally accept level signals from severaltanks through field networks. Thesesystems perform the level monitoringtasks using a variety of compensationand conversion algorithms. The algo-rithms provide density corrections,volumetric or mass conversions, andcorrections to consider the shapes ofhorizontal, vertical or spherical tanks.These systems can perform safetyfunctions, such as shutting off feedpumps to prevent overfilling.

Floats & DisplacersIt was more than 2,200 years ago thatArchimedes first discovered that theapparent weight of a floating object isreduced by the weight of the liquiddisplaced. Some 2,000 years later, inthe late 1700s, the first industrial appli-cation of the level float appeared,when James Brindley and SuttonThomas Wood in England and I. I.Polzunov in Russia introduced the firstfloat-type level regulators in boilers.

Floats are motion balance devicesthat move up and down with liquidlevel. Displacers are force balancedevices (restrained floats), whoseapparent weight varies in accor-dance with Archimedes’ principle:



Figure 7-5: Steam Drum Level Measurement

LT

Steaming Rate

Feed Water

Steam Drum

525°F, 850 psigh3

h1h2

120°F

the buoyant force acting on an objectequals the weight of the fluid dis-placed. As the level changes around

the stationary (and constant diameter)displacer float, the buoyant forcevaries in proportion and can be detect-ed as an indication of level. Regularand displacer floats are available asboth continuous level transmitters andpoint-sensing level switches.

In industrial applications, displacerfloats are often favored because theydo not require motion. Furthermore,force can often be detected moreaccurately than position. However,regular floats are also used, mostlyfor utilities and in other secondaryapplications.

• Float Level SwitchesThe buoyant force available to operatea float level switch (that is, its netbuoyancy) is the difference betweenthe weight of the displaced fluid (grossbuoyancy) and the weight of the float.Floats are available in spherical (Figure7-6A), cylindrical (Figure 7-6B), and avariety of other shapes (Figure 7-6C).They can be made out of stainlesssteel, Teflon®, Hastelloy, Monel, andvarious plastic materials. Typical tem-perature and pressure ratings are -40

to 80°C (-40 to 180° F) and up to 150psig for rubber or plastic floats, and -40 to 260°C (-40 to 500°F) and up to

750 psig for stainless steel floats.Standard float sizes are available from1 to 5 inches in diameter. Custom floatsizes, shapes, and materials can beordered from most manufacturers. Thefloat of a side-mounted switch is hor-izontal; a permanent magnet actuatesthe reed switch in it (Figure 7-6B).

Floats should always be lighter thanthe minimum expected specific gravity

(SG) of the process fluid. For clean liq-uids a 0.1 SG difference might suffice,while for viscous or dirty applications,

a difference of at least 0.3 SG is rec-ommended. This provides additionalforce to overcome the resistance dueto friction and material build-up. Indirty applications, floats should alsobe accessible for cleaning.

Floats can be attached to mechan-ical arms or levers and can actuateelectrical, pneumatic, or mechanicalmechanisms. The switch itself can be



Figure 7-6: Float-Based Level Switches

B)

A) Magnetic Piston B) Reed Switch C) Mecury Switch

Float Reed Switch

1/2 NPT (13 mm)

Permanent Magnet

Switching Element

Bias Spring

Magnet

Float

Rising LevelFalling Level

Level Differential

Pivot

Figure 7-7: Tilt Float Control of a Single Pump

On

85°

Off

x

y

mercury (Figures 7-6A and 7-6C), drycontact (snap-action or reed type,shown in Figure 7-6B), hermeticallysealed, or pneumatic. The switch can

be used to actuate a visual display,annunciator, pump, or valve. The elec-tric contacts can be rated light-duty (10-100 volt amps, VA) or heavy-duty (up to15 A @ 120 Vac). If the switch is to oper-ate a circuit with a greater load than therating of the switch contacts, an inter-posing relay needs to be inserted. If theswitch is to be inserted in a 4-20 mA dccircuit, gold-plated dry contacts shouldbe specified to ensure the required verylow contact resistance.

• Applications & InstallationsIn the tilt switch (Figure 7-6C), a mer-cury element or relay is mountedinside a plastic float; the float’s elec-trical cable is strapped to a pipeinside the tank or sump. As the levelrises and falls, the float tilts up anddown, thus opening and closing itselectric contact. The free length ofthe cable determines the actuationlevel. One, two, or three switchescan be used to operate simplex and

duplex sump-pump stations. A sim-plex (one pump) system will use asingle switch wired in series with themotor leads so that the switch

directly starts and stops the pumpmotor (Figure 7-7).

A duplex (two pump) applicationmight use three switches: one at thetank bottom (LO) to stop bothpumps, another in the middle (HI) tostart one pump, and the last at thetop (HI-HI) to actuate the secondpump, as well as perhaps an audible

and/or visual alarm.Figure 7-8A illustrates how a side-

mounted float switch might actuatean adjacent, sealed reed switch. Themain advantage of this design is thatthe lever extension tends to amplifythe buoyant force generated by thefloat. Therefore the float itself canbe rather small. The main disadvan-tage is that the tank must be openedin order to perform maintenance onthe switch. If the buoyant force ofthe float is used mechanically toactuate a snap-action switch, a forceof only one ounce is needed.

In top (or bottom) mounted mag-netic float switches (Figure 7-8B), themagnet is in the cylindrical float thattravels up or down on a short verticalguide tube containing a reed switch.The float’s motion is restrained byclips and can be only H in or less.These float and guide tubes are avail-able with multiple floats that candetect several levels. The switchassembly itself can be either inserteddirectly into the tank or side-mount-ed in a separate chamber.

A magnetic piston operatedswitch also can be mounted in anexternal chamber (Figure 7-8C). Asthe magnet slides up and down



Figure 7-8: Typical Level Switch Applications

Top Mounted

Side Mounted Float

Cage

A)

B)C)

Figure 7-9: Float Level Switch Configurations

Guard Cage

inside a non-magnetic tube, it oper-ates the mercury switch outside thetube. These switches are complete-ly sealed and well suited for heavyduty industrial applications up to900 psig and 400°C (750°F), meetingASME code requirements. These

switches can be side, top, or cagemounted (Figure 7-9) and can serveboth alarm and control functions onsteam drums, feedwater heaters,condensate pots, gas/oil separators,receivers, and accumulators. Light-duty caged float switches are alsoavailable for service ratings up to250 psig at 200°C (400°F) and 400psig at 40°C (100°F)—suitable formany boilers, condensate receivers,flash tanks, day tanks, holding tanks,and dump valve controls. The cagescan be provided with level gages.Multiple switches are available formultiple-switching applications suchas boiler level alarms and controls.

• Displacer SwitchesWhereas a float usually follows theliquid level, a displacer remains par-tially or completely submerged. Asshown in Figure 7-10A, the apparentweight of the displacer is reduced asit becomes covered by more liquid.

When the weight drops below thespring tension, the switch is actuat-ed. Displacer switches are more reli-able than regular floats on turbu-lent, surging, frothy, or foamy appli-cations. Changing their settings iseasy because displacers can bemoved anywhere along the suspen-sion cable (up to 50 ft). These switch-es are interchangeable betweentanks because differences in processdensity can be accommodatedby changing the tension of the sup-port spring.

Testing the proper functioning of aregular float switch may require fill-ing the tank to the actuation level,

while a displacer switch can be testedsimply by lifting a suspension (Figure7-10A). Displacer switches are avail-able with heavy-duty cages andflanges for applications up to 5000psig at 150°C (300°F), suitable for useon hydraulic accumulators, natural

gas receivers, high pressure scrub-bers, and hydrocarbon flash tanks.

• Continuous Level DisplacersDisplacers are popular as level trans-mitters and as local level controllers,particularly in the oil and petro-chemical industries. However, theyare not suited for slurry or sludgeservice because coating of the dis-placer changes its volume and there-fore its buoyant force. They are mostaccurate and reliable for servicesinvolving clean liquids of constantdensity. They should be temperature-compensated, particularly if varia-tions in process temperature cause



Figure 7-10: Displacement Level Detection

B) Continuous TransmitterA) Switch

Torque Arm

Limit Stop

Knife Edge

Displacer

Torque Arm

BlockTorque Tube

Torque Tube Flange Torque

Rod

NozzleFlapper

Testing Cable

DisplacersFlexible Cable

Low Level

High Level

significant changes in the density ofthe process fluid.

When used as a level transmitter,the displacer, which is always heavierthan the process fluid, is suspendedfrom the torque arm. Its apparent

weight causes an angular displace-ment of the torque tube (a torsionspring, a frictionless pressure seal).This angular displacement is linearlyproportional to the displacer’sweight (Figure 7-10B).

Standard displacer volume is 100cubic inches and the most commonlyused lengths are 14, 32, 48, and 60 in.(Lengths up to 60 ft are available inspecial designs.) In addition totorque tubes, the buoyant force canalso be detected by other force sen-sors, including springs and force-bal-ance instruments. When the buoyant

force is balanced by a spring, there issome movement, while with a force-balance detector, the displacer staysin one position and only the levelover the displacer varies.

Displacer units are available with

both pneumatic and electronic out-puts and can also be configured aslocal, self-contained controllers.When used in water service, a 100cubic inch displacer will generate abuoyant force of 3.6 pounds.Therefore, standard torque tubes arecalibrated for a force range of 0-3.6lbf and thin-walled torque tubes for a0-1.8 lbf range.

For oil refineries and otherprocesses that are operated continu-ously, the American PetroleumInstitute recommends (in API RP 550)that displacers be installed in external

standpipes with level gages and iso-lating valves (Figure 7-11). This way itis possible to recalibrate or maintainthe displacer without interruptingthe process.

• Interface ApplicationsWhen measuring the interfacebetween a heavy liquid and a lightliquid (such as oil on water), the topconnection of the displacer isplaced into the light and the bot-tom connection into the heavy liq-uid layer. If the output of such atransmitter is set to zero when thechamber is full of the light liquid,and to 100% when it is full with theheavy phase, the output will corre-spond to the interface level.Naturally, when interface is beingmeasured, it is essential that thetwo connections of the displacerchamber be located in the two differ-ent liquid layers and that the chamberalways be flooded. Displacer diame-ter can be changed to match the dif-ference in liquid densities, and dis-placer length can be set to match thevertical range of the level interfacevariation.

Regular floats can also be usedfor interface detection if the differ-ence in SG between the two processliquids is more than 0.05. In suchapplications, a float density is need-ed that is greater than the lighterliquid and less than the heavier liq-uid. When so selected, the float willfollow the interface level and, inclean services, provide acceptableperformance.

• Continuous Level FloatsOf the various float sensor designsused for continuous level measure-ment, the oldest and arguably mostaccurate is the tape level gage (Figure7-12A). In this design, a tape or cable



Figure 7-11: Installation to API Standards

2" or Larger Nozzle

3/4" Gate Valve (Vent) or Plug

2" or 3" Standpipe

3/4" Gate Valve (Vent) or Plug

1 1/2" or Larger Gate Valve

1 1/2" or Larger Gate Valve

These Assemblies May Be Elbows

3/4" Gate Valve (Drain)

Reducer to 3/4" Pipe

Overlapping Gage Glasses

3/4" Coupling-6000 LB Tapped One End Only

1 1/2" or Larger Screwed or Flange Connections

Automatic Gage Cocks or Tees

Gate Valve2 " Gate Valve

Pressure Vessel

connects the float inside the tank toa gage board or an indicating take-upreel mounted on the outside of thetank. The float is guided up and downthe tank by guide wires or travelsinside a stilling well. These level indi-cators are used in remote, unattend-ed, stand-alone applications, or theycan be provided with data transmis-sion electronics for integration intoplant-wide control systems.

To install the tape gage, an open-ing is needed at the top of the tankand an anchor is required at its bot-tom. When properly maintained,tape gages are accurate to ±G in. It isimportant to maintain the guidewires under tension, clean and freeof corrosion, and to make sure thatthe tape never touches the protec-tive piping in which it travels. If thisis not done, the float can get stuckon the guide wires or the tape canget stuck to the pipe. (This can hap-pen if the level does not change forlong periods or if the tank farm is

located in a humid region.)Another continuous level indicator

is the magnetic level gage, consist-ing of a magnetic float that travelsup and down on the inside of a long,non-magnetic (usually stainless steel)pipe. The pipe is connected toflanged nozzles on the side of thetank. The pipe column is provided

with a visual indicator, consisting ofG-in triangular wafer elements.These elements flip over (fromgreen to red, or any other color)when the magnet in the float reachestheir level (Figure 7-12B). Alarmswitches and transmitter optionsare available with similar magneticcoupling schemes (Figure 7-12C). In asimilar design, a series of reedswitches is located inside a stand-pipe. The change in output voltageas the individual reed switches areclosed by the rising magnet is mea-sured, giving an indication of level.

The operation of magnetostrictivesensors is based on the Villari effect.In the magnetic waveguide-type con-tinuous level detector, the float (orfloats, when detecting interface) trav-els concentrically up and down out-side a vertical pipe. Inside the pipe isa concentric waveguide made of amagnetostrictive material. A low cur-rent interrogation pulse is sent downthe waveguide, creating an electro-

magnetic field along the length of thewaveguide. When this field interactswith the permanent magnet inside thefloat, a torsional strain pulse (or wave-guide twist) is created and detected asa return pulse. The difference in theinterrogation time and the returnpulse time is proportional to theliquid level in the tank.

This tank level sensing method ishighly accurate, to ±0.02 in, and there-fore is ideal for precision inventorymanagement operations. The sensor isavailable in lengths of 2-25 ft and canbe inserted into the tank from the topof the vessel through flanged,screwed, or welded connections. Forthe simultaneous measurement ofboth interface and total level, a two-float system is available (Figure 7-12D).A resistance temperature detector(RTD) is also available for temperaturecompensation. Like all other floatlevel instruments, this design too isfor clean liquids. Rating is up to 150°C(300° F) and 300 psig. The transmitteroutput can be 4-20 mA dc analog orfieldbus-compatible digital.

• Float Control ValvesFloat-operated control valves com-bine level measurement and levelcontrol functions into a single levelregulator. While simple and inexpen-sive, they are limited to applications

involving small flows and small pres-sure drops across the valve. This isbecause the force available to throt-tle the valve is limited to that pro-vided by the buoyant force acting onthe float, multiplied by the leveraction of the float arm. This does notsuffice to close large valves againsthigh pressure differentials.



Figure 7-12: Continuous Float Level Detectors

A) C)B) D)

Top Guide Wire

Anchor

Oil Seal Assembly

Head

FloatCrank

Assembly

Float

Indicator WafersMagnet

Cross Section

Front View

Follower Magnet and Gaging Rod

Float and Magnet

Guide Tube

Yet, for simple and unattendedapplications (like controlling themake-up water supply into a coolingtower basin or draining condensatefrom a trap), they are acceptable. It isimportant to understand that floatregulators are simple proportionalcontrollers: they are incapable ofholding level at a single setpoint.What they can do is open or close avalve as the float travels through itscontrol range. Therefore, instead of asetpoint, regulators have a throttlingrange. If the range is narrow (floatsusually fully stroke their valve over afew inches of float travel), it gives theimpression of a constant level.

In fact, level will vary over thethrottling range because the onlyway for the regulator to increase thefeed flow (say into a cooling tower

basin) is to first let the level drop sothat the sinking of the float will fur-ther open the valve. The relationshipbetween the maximum flow througha linear valve (Qmax) and the range inliquid level (h) is called the propor-tional sensitivity of the regulator

(Kc = Qmax/h), expressed in units ofGPM/inch. The offset of a float regu-lator is the distance (in inches)between the center of the float rangeand the amount of elevation of thefloat required to deliver the flowratedemanded by the process. T





Edition, OMEGA Press, 1995.• Instrument Engineer’s Handbook, Bela G. Liptak, editor, CRC Press, 1995.• Instrumentation for Process Measurement and Control, Third Edition, N.




Capacitance level detectors arealso referred to as radio fre-quency (RF) or admittancelevel sensors. They operate in

the low MHz radio frequency range,measuring admittance of an alter-nating current (ac) circuit that varieswith level. Admittance is a measure

of the conductivity in an ac circuit,and is the reciprocal of impedance.Admittance and impedance in an accircuit are similar to conductanceand resistance in a direct current (dc)circuit. In this chapter, the termcapacitance level sensor will be usedinstead of RF or admittance.

Table 6 lists some of the industriesand applications where capacitance-type level sensors are used.

Theory of OperationA capacitor consists of two conduc-tors (plates) that are electrically iso-lated from one another by a non-conductor (dielectric). When thetwo conductors are at differentpotentials (voltages), the system iscapable of storing an electric

charge. The storage capability of acapacitor is measured in farads. Asshown in Figure 8-1, the capacitorplates have an area (A) and are sepa-rated by a gap (D) filled with a non-conducting material (dielectric) ofdielectric constant (K). The dielectricconstant of a vacuum is 1.0; the

dielectric constants of a variety ofmaterials are listed in Table 7.

The dielectric constant of a sub-stance is proportional to its admit-tance. The lower the dielectric con-stant, the lower the admittance of the

material (that is, the less conductive itis). Capacitance (C) is calculated as:

C = KA/D

If the area (A) of and the distance (D)between the plates of a capacitorremain constant, capacitance willvary only as a function of the dielec-tric constant of the substance fillingthe gap between the plates. If achange in level causes a change in thetotal dielectric of the capacitancesystem, because (as illustrated inFigure 8-1B) the lower part of area (A)is exposed to a liquid (dielectric Kl)while the upper part is in contact witha vapor (dielectric Kv, which is close to1.0), the capacitance measurementwill be proportional to level.

In the case of a horizontallymounted level switch (Figure 8-2), aconductive probe forms one of theplates of the capacitor (A1), and thevessel wall (assuming it is made froma conductive material) forms theother (A2). An insulator with a lowdielectric constant is used to isolatethe conductive probe from the hous-ing, which is connected to the vessel


RF/Capacitance Level InstrumentationTheory of Operation

Probe Designs

Installation Considerations

FLOW & LEVEL MEASUREMENTRF/Capacitance Level Instrumentation

8

C

Chemical/Petrochemical

Feed & Grain

Food

Pet Food

Plastics/Rubber

Foundries

Beer/Breweries

Pharmaceuticals

Power/Utilities

Water/Waste Treatment

Charcoal

Saw Mills/Woodworking

Mining & Miscellaneous

Soda Ash, Fuel, Oil , Clay, Liquids & Powders

Pellets, Granules, Flakes, Fats, Molasses, Calcium Dust

Sugar, Salt, Flour, Powdered Milk, Various Liquids

Pellets, Rawhides, Grains

Plastic Pellets, Resin, Regrind, Powders, Rubber

Silica Sand, Foundry Sand

Malt, Barley Liquids

Various Powders & Liquids

Coal, Wood, Sawdust, Petro-Coke

Limestone, Hydrated Lime, Water

Charred Sawdust, Wood

Wood Shavings, Sawdust

Various Minerals, Clay, Metals, Stone, Glass, Bentonite

INDUSTRY MATERIALS SENSED

Table 6: Applications of Capicitance Level Sensors

Figure 8-1: Capacitance Theory of Operation

A) Capacitor B) Capacitance Circuit

- -- -- -- -

- -- -- -- -

- -- -

- -- -- --

+ ++ ++ ++ ++ ++ ++ +

+ ++ ++ ++ ++ ++ ++ +

- -- -- -

- -

A

A

D

Electron Flow

Ammeter

Voltmeter

#1Level

RF

#2Kv

Kl

C= KAD

C=Capacitance K=Dieletric Constant A=Area of Plates D=Dist. Between Plates

wall. The probe is connected to thelevel sensor via the conductive threadsof the housing. Measurement is made

by applying an RF signal between theconductive probe and the vessel wall.

The RF signal results in a minutecurrent flow through the dielectricprocess material in the tank from theprobe to the vessel wall. When thelevel in the tank drops and the probeis exposed to the even less conduc-tive vapors, the dielectric constantdrops. This causes a drop in thecapacitance reading and a minutedrop in current flow. This change isdetected by the level switch’s inter-nal circuitry and translated into achange in the relay state of the levelswitch. In the case of continuouslevel detectors (vertical probes), theoutput is not a relay state, but ascaled analog signal.

The total area is the combinedarea of the level sensor probe andthe area of the conductive vesselwall (A = A1 + A2), and the distance (D)is the shortest distance between thesensor probe and the vessel wall.Both of these values are fixed.Therefore, when the probe is no

longer surrounded by vapors (K1), butby the process material (K2), the result-ing capacitance change is directly

related to the difference in dielectricconstant between the two media:

Change in C = (K2 - K1)(A/D)

The sensitivity of a capacitancesensor is expressed in pico-farads (pF).The capacitance unit is the farad,defined as the potential created whena one-volt battery connected to acapacitor causes the storage of one

coulomb of electric energy. A pico-farad is one trillionth of that, and thesensitivity of an accurate capacitancedetector is 0.5 pF. This is the minimum

detectable change in capacitanceresulting from a change in dielectricconstant (K2 -K1).

In most level-sensing applications,the reference material is air (K1 = 1.0).Table 7 gives the K2 values of a varietyof process materials. As the dielectricconstant of the process material getsclose to that of air (K2 for plastic pel-lets, for example, is 1.1), the measure-ment becomes more difficult.

Probe DesignsThe most common probe design is astainless steel rod of G in. or H in.diameter, suitable for most non-con-ductive and non-corrosive materials.The probe is insulated from thehousing and bin wall by an low-dielectric insulator, such as Nylon orRyton. These polymers have maxi-mum operating temperatures of 175-230°C (350-450°F). Ceramics can beused for higher temperature applica-tions or if abrasion resistance isrequired. For applications where theprocess material is conductive andcorrosive, the probe must be coatedwith Teflon® or Kynar.

Some point level sensors are avail-able with build-up immunity, or coat-ing rejection functionality. This is

required when the process material iswet or sticky and likely to causepermanent coating. Build-up immu-nity is provided by the addition of a

RF/Capacitance Level Instrumentation 8


Figure 8-2: Horizontally Mounted Capacitance Switch

K1

A2

K2

A1

- -

- -

+ +

+ +

- -- -

- -- -

+ ++ +

+ ++ +

D

Insulation

A = A1 + A2 ∆C = (K2 + K1) A/D

Figure 8-3: Coating Rejection Design

Coating

Driven Shield (+5 V)

Measuring Section (+5 V)

Vessel Wall

Insulation

No Current Flow

Current Path Through Level

second active section of probe and asecond insulator (Figure 8-3). This sec-ond active section (the driven shield)is driven at the same potential andfrequency as the measuring probe.Because current cannot flow betweenequal potentials, the measuring probe

does not sense material build-upbetween the probe and vessel wall.

Typical insertion lengths of stan-dard capacitance probes range from7 to 16 in. These probes typically areside-mounted (Figure 8-4A). Verticalprobes can be extended by solid rodsup to a length of 1.2 to 1.5 m (4 to 5 ft),or a steel cable with a weight can be

used to suspend the probe up to 15 m(50 ft) (Figure 8-4B). Most capacitancelevel sensors are provided with I to1-H in NPT mounting connectors. Thematching female coupling is usuallywelded to the vessel wall and thecapacitance probe is screwed into the

mating connector. Low profile capac-itance sensors also are available(Figure 8-4C) and are flange-mounted.

In applications where the vessel isnon-conductive and unable to formthe return path for the RF signal, asecond probe placed parallel to theactive one or a conductive strip canbe installed.

• Electronics & HousingsThe electronic circuitry of the probeperforms the functions of: 1) rectifyingand filtering the incoming power, 2)generating the radio frequency signal,3) measuring the changes in currentflow, and 4) driving and controllinginterface devices such as relays,analog signal generators and displaymeters. The circuitry is usually of solidstate design and provided with poten-tiometer adjustments for setting sen-sitivity and time delays.

Because the level sensor will ulti-mately drive an external device, it isadvisable to evaluate for systemcompatibility the number of relaysrequired, their capacities, or theanalog signals required, time delays,and power supply requirements.More advanced microprocessor-based units are self-calibrating; sen-sitivity and time delay adjustmentsare under pushbutton control.These units are often supplied withself-test capability and built-in tem-perature compensation.

8 RF/Capacitance Level Instrumentation


Figure 8-4: Design and Installation of Capacitance Level Sensors

A) Horizontal B) Vertical C) Low Profile

Pipe Section Steel Plates

175 mm (7")

95 mm (3.75")

D

1-14" NPT

1-14" NPT

Flexible Cable

Cable Weight

D

Vessel Wall

200 mm (7.8")

115 mm (4.5")

15 m (50') Max. Customer Specified

Length

25 mm (1")

50 mm (2") Flange

Suggested Baffle Types

34" NPT

Capacitance probes typically are coated with Teflon® (shown), Kynar, or polyethylene

The more advanced designs arealso two-wire, intrinsically safe, andsupply your choice of standard 4-20mA or digitally enhanced outputusing the HART (Highway

Addressable Remote Transducer)protocol. Accuracy (including linearity,hysteresis, and repeatability, butexcluding temperature and supplyvoltage effects) is typically 0.25% ofrange. Minimum span is 4 pF, and theupper range limit (URL) is 2,500 pF.

Level switches are usually provided

with time delays for filtering outfalse readings caused by materialshifts or splashing liquids. In addition,the feature of failsafe selectabilityprovides a predetermined state for

the relay output in the event of apower failure or malfunction.

Sensor housings are typicallymade from cast aluminum, steel, orsynthetic materials such as glass-reinforced nylon. Most housings aresuitable for outdoor installations industy or wet environments.

• The Dielectric ConstantThe dielectric constant of the processmaterial is the most important aspectof the process data. The higher thedifference between the dielectricconstants (of the process material andthe vapor space or between the twolayers in the case of an interface mea-surement), the easier the measure-ment. If the difference is low (K2-K1 <1.0 in Figure 8-2), a high sensitivitydesign (0.5 pF) must be used.

Each sensor has a capacitance threshold, defined as the amount ofcapacitance change required tocause a change in the sensor output.The dielectric constant of a materialcan change due to variations in tem-perature, moisture, humidity, materi-al bulk density, and particle size. Ifthe change in dielectric constantresults in a greater capacitancechange than the calibrated capaci-tance threshold of the sensor, a falsereading will result. This condition canusually be corrected by reducing thesensitivity (increasing the capaci-tance threshold) of the sensor.

As shown in connection withFigure 8-3, sensitivity can beincreased by increasing the probelength (A) or by decreasing the sizeof the gap (D). Either or both changeswill minimize the effect of dielectricconstant fluctuations or increasesensitivity to low dielectrics. It isusually more practical to specify alonger probe than to decrease thedistance (D) from the vessel wall.When the probe is installed from theside (Figure 8-4A), D is fixed, whereas ifthe probe is inserted from the top ofthe tank, D can be changed (if otherconsiderations permit) by moving theprobe closer to the wall of the vessel.

If the same vessel will hold differ-ent materials at different times, thecapacitance sensor must be



Acetic Acid Asbestos Asphalt Bakelite Calcium Carbonate Cellulose Ferrous Oxide Glass Lead Oxide Magnesium Oxide Naphthalene Nylon Paper

4.1 4.8 2.7 5.0 9.1 3.9

14.2 3.7

25.9 9.7 2.5

45.0 2.0

SOLIDS

DIELECTRIC CONSTANT

Phenol Polyethylene Polypropylene Porcelain Quartz Rubber (Hard) Sand Sulphur Sugar Urea Zinc Sulfide Teflon®

4.3 4.5 1.5 5.7 4.3 3.0 3.5 3.4 3.0 3.5 8.2 2.0

DIELECTRIC CONSTANT

Table 7: Dielectric Constants

Acetone Ammonia Aniline Benzene Benzil Bromine Butane Carbon Tetrachloride Castor Oil Chlorine Chloroform Cumene Cyclohexane Dimethylheptane Dimethylpentane Dowtherm Ethanol Ethyl Acetate Ethyl Benzene Ethyl Benzene Ethyl Ether Ethylene Chloride Formic Acid Freon 12 Glycol

21.4 22.4

7.8 2.3

13.0 3.1 1.4 2.2 4.7 2.0 5.5 2.4 2.0 1.9 1.9 3.3

24.3 6.4 2.5 3.0 4.3

10.5 58.5 2.4

41.2

71/22 -27/-33

32/0 68/20

202/94 68/20

30/-1 68/20 60/16

32/0 32/0

68/20 68/20 68/20 68/20 70/21 77/25

68/20 68/20 76/24 68/20 68/20 60/16 70/21

68/20

LIQUIDS

DIELECTRIC CONSTANT

TEMP °F/°C

DIELECTRIC CONSTANT

TEMP °F/°C

Heptane Hexane Hydrogen Chloride Iodine Kerosene Methanol Methyl Alcohol Methyl Ether Mineral Oil Naphthalene Octane Pentane Phenol Phosgene Propane Pyridine Styrene Sulphur Toluene Urethane Vinyl Ether Water Water Water Xylene

68/20 68/20 87/28

224/107 70/21 77/25

68/20 78/26 80/27 68/20 68/20 68/20 118/47

32/0 32/0

68/20 77/25

752/400 68/20 74/23

68/20 32/0

68/20 212/100

68/20

1.9 1.9 4.6

118.0 1.8

33.6 33.1 5.0 2.1 2.5 2.0 1.8 9.9 4.7 1.6

12.5 2.4 3.4 2.4 3.2 3.9

88.0 80.0 48.0

2.4

equipped with local or remote recal-ibration capability.

Light density materials under 20lb/ft3 and materials with particlesizes exceeding H in. in diameter canbe a problem due to their very lowdielectric constants (caused by thelarge amount of air space betweenparticles). These applications mightnot be suited for capacitance-typelevel measurement.

• Application ConsiderationsMaterials that are conductive (water-based liquids with a conductivity of100 micromhos/cm or more) cancause a short circuit between a barestainless steel probe and the vesselwall. As the liquid level drops, theprobe remains wetted, providing aconductive path between the probeand the vessel wall. The faster the levelchanges, the more likely this false indi-cation is to occur. It is advisable to useTeflon® or Kynar insulator coating onthe conductive probe surface whenthe process fluid is conductive.

Temperature affects both the sen-sor components inside the vessel(active probes and insulators) and theelectronic components and housingoutside. An active probe is typicallymade from stainless steel and, assuch (unless it is coated), it is suitablefor most applications. Probe insula-tors can be Teflon®, Kynar, or ceramic,and should be selected for the oper-ating temperature of the application.The housing and the electronics areaffected by both the internal andexternal vessel temperatures.

Ambient temperature limits usuallyare specified by the manufacturer,but heat conduction from a high-temperature process is more difficultto evaluate. Heat conduction can bereduced by using an extendedmounting coupling or one made of a

8 RF/Capacitance Level Instrumentation


Figure 8-5: Capacitance Probe Installation Recommendations

Recommended

Good Probe Coverage Poor Probe Coverage

In Main FlowOut of Main Flow

Probes Do Not Touch the Vessel Structure

Complete Probe Protruding into Material

Portion of Probe Within Mounting Neck

Material Will Fall Away

Material Can Build Up, Creating A False Signal

Probes Touch the Vessel Structure

Not Recommended

Sensors Acceptable (> 18") Sensors Too Close (< 18")

A

C

D

E

B

low thermal conductivity material. Ifsuch methods are insufficient, theelectronics may be mounted up to20 ft away and connected via coaxialcable. The cable’s inherent capaci-tance, however, reduces the overallsensitivity of the system.

Housings must also be compati-ble with the requirements for haz-ardous, wash-down, wet, and/ordusty environments. Explosion-proof environments may require thehousing to be certified. In addition,the active probe might need to beintrinsically safe.

If the process material is corrosiveto stainless steel, the probe shouldbe coated with Kynar or Teflon® forprotection. Ryton is a good choicefor abrasive materials, and, for foodgrade or sanitary applications, stain-less steel and Teflon® are a goodprobe-insulator combination.

Installation ConsiderationsThe capacitance probe should bemounted in such a way that itsoperation is unaffected by incom-ing or outgoing material flow(Figure 8-5A). Material impacts cancause false readings or damage to

the probe and insulator. When mea-suring low-dielectric materials, it'simportant that the entire probe becovered, not just the tip (Figure 8-5C). When rod or cable extensionsare used, allow for 8-12 in. of activeprobe coverage.

Install the probe so that it doesnot contact the vessel wall (Figure 8-5B) or any structural elements of thevessel. If a cable extension is used,allow for swinging of the cable as thematerial level in the vessel rises, sothat the plumb bob on the end ofthe cable does not touch the vesselwall. The probe should not bemounted where material can form a

bridge between the active probe andthe vessel wall. In addition, the probeshould not be mounted at an upwardangle (Figure 8-5D), to avoid materialbuild-up.

If more than one capacitance levelsensor is mounted in the vessel, a min-imum distance of 18 in. should be pro-vided between the probes (Figure 8-5E).Closer than that and their electro-magnetic fields might interfere. If acapacitance probe is installed throughthe side wall of a vessel and theweight of the process material actingon the probe is sometimes excessive,a protective baffle should be installedabove the sensor (Figure 8-4A). T





Edition, OMEGA Press, 1995. • Instrument Engineer’s Handbook, Bela Liptak, Third Edition, CRC Press, 1995.• Instrumentation for Process Measurement and Control, Third Edition, N.




An entire class of level instru-mentation devices is basedon a material’s tendency toreflect or absorb radiation.

For continuous level gages, the mostcommon types of radiation used areradar/microwave, ultrasonic, andnuclear. Optical electromagneticradiation also can be used, but thishas found its way primarily into thepoint-switch applications discussedin the next chapter.

The main advantage of a radia-tion-based level gage is the absenceof moving parts and the ability todetect level without making physi-cal contact with the process fluid.Because they can in effect “see”through solid tank walls, nuclearradiation gages are perhaps the ulti-mate in non-contact sensing.Because they require a gamma radi-ation source and are relativelyexpensive, however, nuclear gagesare often considered the level gageof last resort.

Radar & MicrowaveIn 1925, A. Hoyt Taylor and Leo Youngof the U.S. Navy used radar (RAdioDetection And Ranging) to measurethe height of the earth’s ionosphere.By 1934, they were developing radarfor Navy ships. In 1935, RobertWatson-Watt of England used radarto detect aircraft. The first radar levelsensors were introduced in 1976, butthey did not become economicallycompetitive until a decade later.

Both radar signals and microwavestravel at the speed of light, but aredistinguished by their frequencies(FM radio broadcast frequencyis from 88 to 108 MHz, while

microwaves range from 1-300 GHz) andby their power levels (radar is around0.01 mW/cm2, while microwaves rangefrom 0.1-5 mW/cm2). Becausemicrowaves operate at a higher energylevel, they can withstand more coatingthan can radar-type sensors.

Radar sensors consist of a trans-mitter, an antenna, a receiver with

signal processor, and an operatorinterface. The transmitter is mountedon top of the vessel. Its solid-stateoscillator sends out an electromag-netic wave (using a selected carrierfrequency and waveform) aimeddownward at the surface of theprocess fluid in the tank. The fre-quency used is typically 10 GHz.

The signal is radiated by a parabolicdish or horn-type antenna (Figure 9-1A) toward the surface of the processliquid (Figure 1B). A portion is reflect-ed back to the antenna, where it iscollected and routed to the receiver.Here, a microprocessor calculatesthe time of flight and calculates the

level. Time of flight is the periodbetween the transmission of theradar pulse and the reception of thereturn echo. It is determined by theradar detector, which is simultane-ously exposed to both the sent andthe reflected signal. The detectoroutput is based on the difference. Thefrequency-modulated (FM) signal

varies from 0 to 200 Hz as the dis-tance to the process fluid surfacevaries between 0 and 200 ft. Becausethis measurement takes place in thefrequency domain, it is reasonablyfree of noise interference.

The depth of the vapor space (thedistance between the datum pointand the level in the tank, identified as“d” in Figure 9-1B) is calculated fromthe time of flight (t) and the speed oflight (c = 186,000 miles/sec):

d = t/2c

The level (L in Figure 9-1B) is calculatedby figuring the difference between


Radiation-Based Level GagesRadar & Microwave

Ultrasonic Level Gages

Nuclear Level Gages

FLOW & LEVEL MEASUREMENTRadiation-Based Level Instrumentation

9

A

Figure 9-1: Electronics Calculate Time of Flight

B)A)

Large Size Parabolic Dish

Antenna

Small Size Horn

Antenna

Datum Point

E d

L

the total tank height (E) and thevapor space depth (d):

L = E-d

Knowing the signal velocity (c) and thedielectric constant (dc) of the vapor(that is, the relative ability of the vaporto oppose and reflect electromagneticwaves), the velocity of the radar wavetransmission (V) can be calculated:

V = c/(dc)0.5

• Antenna Designs and MountingThe two commonly used antennas arethe horn and the parabolic dish anten-na. When the radar level gage sendsout its signal, the microwaves spreadout. The larger the antenna diameter,the smaller the divergence angle andthe greater the signal strength (Figure9-1A). The disadvantages of smallerantennas include higher beam spread-ing and the correspondingly increasedpossibility of reflection from obsta-cles within the tank. On the positiveside, there is a greater chance that theemitted beam will be reflected backto the detector. Therefore, alignmentof the sensor is not as critical.

Large antennas generate a morefocused signal, helping to eliminatenoise interference from flat and hor-izontal metal surfaces. On the otherhand, they are more prone to errorscaused by unwanted reflections fromturbulent or sloping surfaces. A fullyisolated antenna mounted outsidethe tank (Figures 9-2 and 9-4) pro-vides both sealing and thermal isola-tion. If the antenna is positionedbelow the process seal, it is exposedto the process vapors, but gains theadvantages of stronger signal ampli-tudes and suitability for higher oper-ating pressures.

• Contact & Non-Contact RadarNon-contact radar gages either usepulsed radar waves or frequency-modulated continuous waves(FMCW). In the first, short-durationradar pulses are transmitted and thetarget distance is calculated usingthe transit time. The FMCW sensorsends out continuous frequency-modulated signals, usually in succes-sive (linear) ramps. The frequency

difference caused by the time delaybetween transmittal and receptionindicates the distance.

Radar beams can penetrate plas-tic and fiberglass; therefore, non-contact radar gages can be isolatedfrom the process vapors by a seal.The seal can be above the parabolicdisc (Figure 9-1A) or can totally iso-late the sensor (Figure 9-2A). Thebeam’s low power allows for safeinstallation in both metallic andnon-metallic vessels. Radar sensorscan be used when the process mate-rials are flammable or dirty andwhen the composition or tempera-ture of the vapor space varies.

Contact radar gages send a pulsedown a wire to the vapor-liquid inter-face. There, a sudden change in thedielectric constant causes the signal

to be partially reflected. The time-of-flight is then measured (Figure 9-2B).The unreflected portion travels on tothe end of the probe and provides azero-level reference signal. Contactradar technology can be used on liq-uids and on small-grained bulk solidswith up to 20-mm grain size.

Reflection-type microwave switch-es measure the change in amplitudeof a reflected signal (Figure 9-3A). Air

and vapors return a small percentageof the signal because of their lowdielectric constants, while highdielectric materials such as waterreturn almost all the signal. More sen-sitive switches can distinguish liquid-liquid or liquid-solid interfaces havingas little as 0.1 difference in dielectricconstant. Low dielectric materials likeplastic pellets (dielectric 1.1) can bemeasured if the particle diameter isless than 0.1 in (larger than that, exces-sive beam scattering occurs).

The beam-breaker switch sends amicrowave beam from a transmitterto a receiver located on the oppositeside of the tank. When the beam isblocked, the signal is weakened(Figure 9-3B). Beam-breaker align-ment is not critical, and separationdistance can be up to 100 ft.

Radiation-Based Level Instrumentation 9


Figure 9-2: Radar-Based Antenna Installations

B) ContactA) Non-Contact

Cutaway View of Horn Antenna

Cutaway View of Parabolic Dish Antenna

Process Seal

Process Seal

Impu

lse

Tank Nozzle

Tank Ceiling

Both reflection and beam-breakermicrowave switches are typically usedin applications where it is desirablenot to penetrate the tank. These non-intrusive sensors send electromagnet-ic radio waves through plastic, ceramicor glass windows, or through fiber-glass or plastic tank walls.

• Advantages & LimitationsThe reflective properties of theprocess material affect the returnedradar signal strength. Whereas liquidshave good reflectivity characteristics,solids do not. Radar can detect theliquid level under a layer of light dustor airy foam, but if the dust particlesize increases, or if the foam or dustgets thick, it will no longer detect theliquid level. Instead, the level of thefoam or dust will be measured.

Internal piping, deposits on theantenna, multiple reflections, orreflections from the wall can all inter-fere with the proper operation of aradar sensor. Other sources of inter-ference are rat-holing and bridging ofsolids, as well as angled process

material surfaces that can reflect theradar beam away from the receiver.

In comparison to other radiationreflection sensors, radar has some

advantages. For example, ultrasonicsensors are affected by the composi-tion of the vapor space. On the otherhand, ultrasonic sensors performbetter in dirty applications, or withsolids when the grain size is largerthan 20 mm.

Ultrasonic Level GagesThe origin of ultrasonic level instru-mentation goes back to the echome-ters used in measuring the depth ofwells by firing a blank shell and tim-ing the return of the echo. SONARdetectors used in naval navigationalso predate industrial applicationsof this principle.

The frequency range of audiblesound is 9-10 kHz, slightly below the20-45 kHz range used by industriallevel gages. The velocity of an ultra-sonic pulse varies with both the sub-stance through which it travels andwith the temperature of that sub-stance. This means that if the speedof sound is to be used in measuring alevel (distance or position), the sub-stance through which it travels must

be well known and its temperaturevariations must be measured andcompensated for.

At room temperature, the speed of

sound in atmospheric air is 340 m/s or762 mph. At that same temperature, anultrasonic pulse travels through waterat 1,496 m/s or 3,353 mph. If the air isheated to 100°C, the speed of soundrises to 386 m/s. Indeed, the speed ofsound is proportional to the squareroot of temperature. At near ambienttemperatures, the speed rises by 0.6m/s per each 1°C increase, correspond-ing to an increase of 0.18%/°C.

Ultrasonic level switches (pointsensors) operate by detecting eitherdampening of ultrasonic oscillation orby sensing the absorption or transmis-sion of an ultrasonic pulse. Ultrasoniclevel transmitters measure actual dis-tance by issuing an ultrasonic pulseand measuring the time required forthe reflected echo to be received.

• Ultrasonic TransducersThe transducer that generates theultrasonic pulse is usually piezoelec-tric, although in the past electrostaticunits also were used. An electrostatictransducer is constructed of a thin,flexible gold-plated plastic foil,

stretched over an aluminum back-plate and held in place by a leafspring. This design was used in earlyPolaroid auto-focus cameras and is

9 Radiation-Based Level Instrumentation


Figure 9-3: Microwave Switches Send Pulses Through a Window or Tank Wall

B) Beam-BreakerA) Reflection

Reflection Microwave Detector

Microwave Window

Microwave Window

Microwave Transmitter

Transmitted Beam

Microwave Receiver

Microwave Window

Reflected Beam

Absorbed Beam

still utilized in clean environments.Piezoelectric transducers utilize

ceramic or polymer crystals vibratedat their natural frequency. Theseunits are much more rugged, canwithstand wash-down pressures of1,200 psig and can conform toNEMA-6P (IEC IP67) standards.

Generally, the larger the diameterof the transducer, the longer therange and the lower the frequency.This is because, after releasing anultrasonic pulse, the transducer needstime for the vibration to settle. Theoscillation frequency is inversely pro-portional to the element’s diameter,so smaller diameter transducer ele-ments generate higher frequencies.Standard transducers have a beamangle of about 8°, require a connec-tion size between G in and 2.5 in NPT,and are suited for operating tempera-tures in the range of -20 to 60°C (-30to 140°F). Accuracy is typically within0.25-0.5% of full range, up to about 30ft. Output typically is 4-20 mA with a12-amp relay output.

• Level Transmitter Configurations The ultrasonic level sensor assemblycan consist of separate transmitter and

receiver elements (Figure 9-4A). Mostoften, however, a single transducer is

cycled on and off at regular intervalsto listen for the reflected echo (Figure9-4A). When mounted on the top ofthe tank, the sensor detects the depthof the vapor space. Accurate knowl-

edge of the shape of the tank’s cross-section is required in order to

determine the volume of liquid. If it is desired to measure the

height of the liquid column directly,the transducer can be mounted inthe bottom of the tank (Figure 9-4A).However, this configuration exposesthe transducer to the process fluidand limits accessibility for mainte-nance. Alternately, the transducercan be mounted on the outside ofthe wall of the vessel bottom, butthe ultrasonic pulse is likely to besubstantially weakened by theabsorbing and dispersing effects ofthe tank wall (Figure 9-4A).

Stagnant, unagitated liquids andsolids consisting of large and hardparticles are good reflectors, andtherefore good candidates for ultra-sonic level measurement. Fluff,foam, and loose dirt are poor reflec-tors, and dust, mist, or humidity inthe vapor space tend to absorb theultrasonic pulse. The ultrasonic sig-nal also is attenuated by distance. Ifa 44-kHz sound wave is traveling indry, clean ambient air, its soundpower drops by 1-3 decibels (dB) foreach meter of distance traveled.Therefore it is important, particularlywhen measuring greater depths,



Figure 9-4: Ultrasonic Level Installation Configurations

A) Returned Echo Timing B) Signal Absorption C) Contact

Ground Level

Casing

Brine Pipe

Hydrocarbon

Interface

Brine

Transducer

Cavity

Figure 9-5: Calibration Target

that the transducers generate astrong and well-focused ultrasonicpulse (Figure 9-4B).

It is also desirable that the surfacebe both flat and perpendicular to thesound wave. In liquid-level applica-tions, the aiming angle must be with-in 2 degrees of the vertical. If the sur-face is agitated or sloping (as in thecase of solids), the echo is likely tobe dispersed. Therefore, the key to

successful ultrasonic level sensorinstallations is the careful analysis ofthe reflection, propagation, andabsorption characteristics of thetank’s contents.

When detecting the interfacebetween two liquids, such as thehydrocarbon/brine interface in asalt dome storage well, the trans-ducer is lowered down to the bot-tom of the well. The ultrasonicpulse is sent up through the heavybrine layer to the interface. The

time it takes for the echo to returnis an indication of the location ofthe interface (Figure 9-4C).

• Special FeaturesMost modern ultrasonic instrumentsinclude temperature compensation,filters for data processing andresponse time, and some even pro-vide self-calibration. Figure 9-5 illus-trates a fixed target assembly that

provides a point reference to auto-matically recalibrate the level sensor.Multiple calibration targets can beprovided by calibration ridges insounding pipes. This can guaranteemeasurement accuracy of within5 mm over a distance of 30 meters.

Intelligent units can performautomatic self-calibration or convertthe level in spherical, irregular, or hor-izontal cylindrical tanks into actualvolume. They can also be used inmulti-tank or multi-silo installations,

which, through multiplexing, canreduce the unit costs of obtaininglevel measurements.

• Level Switches When it is sufficient to detect the pres-ence or absence of level at a particularelevation, dampened or absorption-type level switches can be considered.In the dampened design, a piezoelec-tric crystal vibrates the sensor face at

its resonant frequency. The vibration isdampened when the probe face is sub-merged in process fluid. As shown inFigure 9-3A, these switches can bemounted outside or inside the tank,above or below the liquid level. Theprobe can be horizontal or vertical.These switches are limited to clean liq-uid installations because coating candampen the vibration. Solids may notprovide sufficient dampening effects toactuate the switch.

In the absorption-type level switch,



Figure 9-6: Transmission of Gamma Rays Generated by Cesium 137

% T

rans

miss

ion

Water SG=1.0

Thickness - Inches

10080

605040

30

20

108

654

3

2

11 2 43 5 6 8 107 9 12 14 16 18 20 22 24 26 28 30

SG=0.5

SG=1.5

SG=2.0

Steel SG=7.6

Aluminum SG=2.8

Lead SG=11.3

one piezoelectric crystal serves as atransmitter and another as the receiver.When the gap between them is filledwith liquid, the sonic wave passesfrom one crystal to the other. Whenvapors fill the gap, however, the ultra-sonic pulse does not reach the

receiver. The crystals can be mountedon opposite sides of the tank, con-tained in the fingers of a fork-shapedsensor, or located on the two sides ofone or more 0.5-in gaps in a horizon-tal or vertical probe. When theprocess fluid is a sludge or slurry, it isdesirable to provide a large gapbetween the transmitter and receiverin order to make sure that sticky orcoating fluids will drain completelyfrom the gap when the level drops.

Typical accuracy of these switchesis H-in or better. Connection size isI-in NPT. Operating temperaturerange is 40-90°C (100 to 195°F) (withspecial units capable of readings upto 400°C/750°F) and operating pres-sure to 1000 psig. Standard output isa 5 or 10 amp double-pole/double-throw (DPDT) relay, but voltage and

current outputs are also used. The presence or absence of an

interface between clean liquids canbe measured by inserting an absorp-tion (gap) probe at a 10° angle belowthe horizontal. In this configuration, aslong as the probe is immersed in the

heavy or light liquid, the ultrasonicpulse will reach the receiver. Whenthe interface moves into the gap,however, it is reflected away and doesnot reach the receiver.

When a sludge or slurry interface isto be detected or when the thicknessof the light layer is of interest, anultrasonic gap sensor can be attachedto a float. As long as the absorptioncharacteristics of the two layers differ,the sensor will signal if the layer isthicker or thinner than desired.

Nuclear Level SensorsIn 1898 Marie Curie discovered radi-um by observing that certain ele-ments naturally emit energy. Shenamed these emissions gamma rays.Gamma rays exhibited mysteriousproperties—they could pass through

a seemingly solid, impenetrable massof matter. In the passage, however,the gamma rays lost some of theirintensity. The rays were predictablyaffected by the specific gravity andtotal thickness of the object, and bythe distance between the gamma raysource and the detector.

For example, Figure 9-6 showsthat, if radiation from Cesium 137 ispassing through an 3-in thick steelobject, 92% of the radiation energywill be absorbed and only 8% will betransmitted. Therefore, if the observ-er can hold all variables except thick-ness constant, the amount of gammatransmission can be used to measurethe thickness of the object. Assumingthat the distance between the sourceand detector does not change, onecan make accurate measurements ofeither thickness (level), or, if thick-ness is fixed, then of the density of aprocess material.

• Radiation SourcesThe development of nuclear levelsensors began when this technologymoved from the lab to the industrialenvironment. This necessitated thedesign and manufacture of suitabledetectors and the mass productionof radioisotopes. Both occurred inthe 1950s.

The penetrating power of nuclearradiation is identified by its photonenergy, expressed in electron volts (eV)and related to wavelength (Figure 9-7).The most common isotope used forlevel measurement is Cesium 137,which has a photon energy level of0.56 MeV. Another isotope that isoccasionally used is Cobalt 60, whichhas an energy level of 1.33 MeV. Whilethe greater penetrating power of thishigher energy radiation appears attrac-tive at first, the penalty is that it alsohas a shorter half-life. As any isotope



Figure 9-7: Wavelengths and Photon Energies

Nonionizing Ionizing

Photon Energy eV

Electric Waves

Radio Waves Infrared

Visible

Gamma Rays

X-Rays

Cosmic Rays

Wavelength cm

105 100

10-5 10-10 10-15

10-10 10-5 100 105 1010

Ultraviolet

decays, it loses strength—the time ittakes to lose half of its strength iscalled its half-life.

The half-life of Cobalt 60 is 5.3years. This means that, in 5.3 years, theactivity of a 100 millicurie (mCi)Cobalt 60 source will be reduced to50 mCi. (One mCi is defined as therate of activity of one milligram ofRadium 226.) When used for levelmeasurement, the continuous loss ofsource strength requires not only con-tinuous compensation, but, eventually(in the case of Cobalt 60, in about 5years), the source must be replaced.This means not only the expense ofpurchasing a new source, but also thecost of disposing of the old one.

In contrast, the 33-year half-life ofCesium 137 is long enough that thesource may well outlive the process.Another likelihood is that technologi-cal advances will increase the sensitiv-ity of the detector faster than the rateat which the source is decaying. Thisprovides users the option of replacingor upgrading the detector while keep-ing the source in place for the future.

• Radiation SafetyThe Nuclear Regulatory Commission(NRC) limits radiation intensity to amaximum of 5 milliroentgens per hour(mr/hr) at a distance of 12 in from thenuclear gage. If it is more, the arearequires Radiation Area posting. Thedistance of 12 in is critical, because radi-ation intensity decreases by the inversesquare of distance. Nuclear level gagesare sized to provide radiation intensityat the detector that exceeds the mini-mum required, but is under the 5 mr/hrmaximum. For ion chamber detectors,the minimum is 1 mr/hr. For Geiger-Mueller switches, it is 0.5 mr/hr. Andfor scintillation detectors, it is 0.1-0.2mr/hr. Because the nuclear gage isbasically measuring the vapor space

above the liquid, as the level rises in thetank, the intensity at the detectordrops. When the tank is full, radiationintensity is practically zero.

When used as a tank level sensor,radiation must pass through several lay-ers of material before reaching thedetector. At the detector, the maximumradiation must be less than some safetylimit (such as 5 mr/hr) to avoid the needfor “posting.” Other criteria can be used,such as keeping a yearly dosage under 5rems (roentgen + equivalent + man). Ifsomebody is exposed to radiationthroughout the year, such a dosage willresult from exposure to radiation at anintensity of 0.57 mr/hr, while if an oper-ator is exposed for only 40 hrs/wk, 5rem/yr will correspond to what that

person would receive if exposed to 2.4mr/hr in the work area. As it is the totallifetime dosage of radiation exposurethat really matters (maximum of 250rems), the acceptability of the 5 rem/yr,or any other limit, is also a function ofage (Figure 9-8). On the other hand, theradiation at the detector must still besufficient to produce a usable change indetector output when the level changes.

This can be illustrated by an example:

• Source SizingA point source of 10 mCi Cesium 137(source constant for Cesium 137 isK=0.6) is installed on a high-pressurewater tank having H-in steel walls(Figure 9-9). Usually, two criteria needto be satisfied: First, the radiationintensity at the detector must drop byat least 50% as the level rises from 0-100%. The second and more importantcriterion is that the maximum radiationdose at the detector (when the tank isempty) must not exceed the safetylimit (say, 2.4 mr/hr). It must exceed 1.0mr/hr, however, in order to actuate theintended ion chamber detector.

First the in air intensity (Da in

mr/hr) is calculated at the detector,for the condition when there is notank between the source and receiver.Assume distance (d) is 48 in:

Da = 1000 K(mCi)/d2 = 1000(0.6)(10)/482 = 2.6 mr/hr

Because the source is shielded in alldirections except towards the tank,



Figure 9-8: Accumulation of Radiation Exposure

Operator's Age

Unsafe

SafeAccu

mul

ated

Rem

s

250

200

150

100

50

010 20 30 40 50 60 70

Rate: 5

Rems/

Yr

the operator who is working near thedetector will receive the maximumdosage when the tank is empty. Thetwo H-in steel walls will reduce Da(% transmission of 1-in steel in Figure1 is 49%) to 0.49 x 2.6 = 1.27 mr/hr.This is below the allowable maxi-mum but above the minimum need-ed by the detector.

When the tank is full, the pres-ence of 30 in of water in the radia-tion path will reduce this maximumintensity to 0.045 mr/hr (0.035 x 1.9= 0.045). This reduction in intensitywell exceeds the required 50% dropneeded for sensitive measurement.Note that the source size couldhave been cut in half if a Geiger-Mueller detector were used. A scin-tillation detector would reducesource size 5- to 10-fold.

The source size can also bereduced by locating the source in thetip of a probe inside the tank andmoving it relatively close to the wall.When large level ranges are to bemeasured, a strip source can be usedinstead of a point source. The accura-cy of most nuclear level gages isaround 1% of range. If accountingaccuracy is desired, the source andthe detector can both be attached tomotor driven tapes and positioned atthe level (or at the interface level, ifthe tank contains two liquids).

Fortunately, today’s computers caneasily crunch the numbers and formu-las of any combination of geometryand design criteria. The biggest chal-lenge is not the calculation, but theobtaining of accurate inputs for thecalculations. Therefore, it is veryimportant that your vessel’s wall mate-rials, thicknesses, other tank compo-nents such as baffles, agitator bladesor jackets, and all distances be accu-rately determined. In short, the perfor-mance of a nuclear gage installation is

very much a function of the accurateknowledge of the installation details.

• Detector OptionsThe simplest and oldest type of radia-tion detector is the Geiger-Mullertube. This instrument is most oftenidentified with the Geiger countersthat make a loud and dramatic clickingsound when exposed to radiation. Theworking component of this detector isa metal cylinder that acts as one of theelectrodes and is filled with an inertgas. A thin wire down the center actsas the other electrode. Glass caps are

used as insulators, and a high voltage(700-1000 vdc) nearly sufficient tocause current flow between the elec-

trodes is applied. When the tube isexposed to gamma radiation, the gasionizes and the ionized particles carrythe current from one electrode to theother. The more gamma radiationreaches the gas in the tube, the morepulses are generated. The resultingpulse rate is counted by the associatedelectronic circuitry, which makes mea-surements in pulses per second.

This detector can be used as alevel switch if it is calibrated toengage or disengage a relay whenradiation intensity indicates a high orlow level condition. The G-M tube

detector can only be used as a singlepoint detection device. Its advan-tages include its relatively low cost,



Figure 9-9: Radiation Source Sizing

48"

30"

1/2"

Min. Liquid Level

Radiation Path

45 Max

Max. Liquid Level

Plan View

Platform

Elevation

Source and

Holder

"U" Bolt

Detector

Vessel Clip

Radiation Path

Vessel Clips and Support

Plate

small size, and high reliability. The ion chamber detector is a con-

tinuous level device. It is a 4 to 6-indiameter tube up to twenty feet longfilled with inert gas pressurized toseveral atmospheres. A small biasvoltage is applied to a large elec-trode inserted down the center ofthe ion chamber. As gamma energystrikes the chamber, a very small sig-nal (measured in picoamperes) isdetected as the inert gas is ionized.This current, which is proportional tothe amount of gamma radiationreceived by the detector, is amplifiedand transmitted as the level mea-surement signal.

In level measurement applications,the ion chamber will receive the mostradiation and, therefore, its output willbe highest when the level is lowest. Asthe level rises and the greater quantityof measurand absorbs more gammaradiation, the output current of thedetector decreases proportionally.The system is calibrated to read 0%level when the detector current out-put is its highest. 100% level is set tomatch the lowest value of the outputcurrent. Non-linearities in betweencan usually be corrected with the useof linearizing software. This softwarecan correct for the effects of steamcoils, agitator blades, baffles, stiffeningrings, jackets and other componentsinside or outside the tank.

Scintillation counter detectors arefive to ten times more sensitive thanion chambers. They also cost more,yet many users are willing to acceptthe added expense because it allowsthem either to use a smaller sourcesize or to obtain a more sensitivegage. When gamma energy hits a scin-tillator material (a phosphor), it is con-verted into visible flashes comprisedof light photons (particles of light).

These photons increase in number

as the intensity of gamma radiationincreases. The photons travel throughthe clear plastic scintillator mediumto a photo multiplier tube, whichconverts the light photons into elec-trons. The output is directly propor-tional to the gamma energy that isstriking the scintillator.

Scintillators are available in a mul-titude of shapes, sizes, and lengths.One of the latest is a fiber opticcable that allows one to increasedetector sensitivity by installingmore filaments in the bundle.Another advantage of the fiber opticcable is that it is manufactured inlong lengths flexible enough toform-fit to the geometry of the ves-sel. This simplifies the measurementof levels in spherical, conical, orother oddly shaped vessels.

• Nuclear ApplicationsRadiation gages typically are consid-ered when nothing else will work, orwhen process penetrations requiredby a traditional level sensor present arisk to human life, to the environ-ment, or could do major damage toproperty. The liquids and bulk solids

measured by nuclear gages are amongthe most dangerous, highly pressur-ized, toxic, corrosive, explosive, andcarcinogenic materials around.Because the nuclear gage “sees”through tank walls, it can be installedand modified while the process is run-ning—without expensive down timeor chance accidental release.

Because the installation of nuclearsensors requires a NuclearRegulatory Commission (NRC)license, associated procedures aredesigned to guarantee that theinstallation will be safe. The best wayto look at the safety aspects ofradioactive gaging is to compare thewell defined and understood riskrepresented by exposing the opera-tors to radiation against the possiblylarger risk of having an unreliable orinaccurate level reading on a danger-ous process.

As detectors become more sensi-tive and are aided by computers,radiation source sizes and the result-ing radiation levels continue to drop.Therefore, the safety of these instru-ments is likely to continue toimprove with time. T





Edition, OMEGA Press, 1995. • Automated Process Control Electronics, John Harrington, Delmar

Publishing Inc., 1989.• Fundamentals of Radar Techniques for Level Gauging, Detlef Brumbi,

Krohne Metechnik GmbH & Co. KG, 1995.• Industrial Applications of Radar Technology for Continuous Level

Measurement, W. L. Hendrick, Instrument Society of America, 1992.• Instrument Engineer’s Handbook, Bela Liptak, Third Edition, CRC Press, 1995.• Process/Industrial Instruments and Controls Handbook, 4th Edition,

Douglas M. Considine, McGraw-Hill, 1993.• Theoretical Nuclear Physics Volume I: Nuclear Structure, New York, A.

deShalit, H. Feshback,: John Wiley & Sons, 1974.

Thermal, vibrating, and opticallevel switches are specialtydevices developed to solvespecific level detection prob-

lems. Typically, they are used in appli-cations that either cannot be handledby the more common float and probe-type devices, or when ultrasonic,nuclear, radar or microwave designswould be too sophisticated, expensive,or otherwise unsuited for the task.

All three types can be used todetect liquid levels or interfacesbetween liquids. The optical levelswitch is also suited for detectinghigh foam levels, if it is spray washedafter each event. In some specializedapplications, all three of theseswitches have been tuned to identifyspecific materials or to determinewhen a material reaches a particularviscosity, density, opacity, or thermalconductivity condition.

All three level switch designs aresimple, straightforward, and reliable.Although some can detect otherprocess properties besides level,their main purpose is to measure thepresence or absence of material at aparticular level in a tank.

These switches are good candi-dates for use in multiple purposeprocessing equipment where theymust be compatible with a variety ofprocess materials and process condi-tions. They do not require recalibra-tion between batches and can becleaned in place.

Vibrating probe-type sensors areoften used to detect solid materialssuch as powders, bulk solids, grain,flour, plastic granules, cement, and flyash. They provide excellent perfor-mance as high or low level switches

and can be mounted from the tops orsides of tanks. The low thermal con-ductivity of solids and the dustyatmospheres that are likely to exist inthe vapor space of solids bins tend toexclude the use of optical and ther-mal switches from most solids levelmeasurement applications.

When solid materials rat-hole orbridge, few level sensors (except load

cells or radiation devices) work well.The performance of vibrating probeand tuning-fork sensors is also ques-tionable in such services, but theirvibrating nature can help to collapsethe bridges or to break up the rat-holes.

Vibrating and tuning fork probescan tolerate a fair amount of materi-al build-up, or, if coated with Teflon®,can be self-cleaning in some less dif-ficult services. Optical level switchesare available with automatic washersto remove the build-up of coating

after each high level episode.Thermal switches can continue towork when lightly coated, but build-up does usually add a thermally insu-lating layer, ultimately slowingresponse time.

Of the three level-switch designsdiscussed in this chapter, only thelaser-based optical level switch isappropriate for use in molten metal

level detection. Of the other levelsensor technologies, refractory floats,refractory bubbler tubes, and proxim-ity-type capacitance detectors alsoare used in molten metal service.

Thermal SwitchesThermal level switches sense either thedifference between the temperaturesof the vapor space and the liquid or,more commonly, the increase in ther-mal conductivity as a probe becomessubmerged in the process liquid.


Thermal Switches

Vibrating Switches

Optical Switches

FLOW & LEVEL MEASUREMENTSpecialty Level Switches

10

TSpecialty Level Switches

Figure 10-1: Thermal Conductivity Level Switch

Unheated Probe

Heated Probe

O + DC Voltage

To Switch Heated Probe

Unheated Probe

Heater

Heater

One of the simplest thermal levelswitch designs consists of a temper-ature sensor heated with a constantamount of heat input. As long as theprobe is in the vapor space, theprobe remains at a high temperature,because low-conductivity vapors donot carry much heat away from theprobe. When the probe is sub-merged, the liquid absorbs more heatand the probe temperature drops.The switch is actuated when thischange in temperature occurs.

Another type of thermal sensoruses two resistance temperaturedetectors (RTDs), both mounted atthe same elevation. One probe isheated and the other provides anunheated reference. The outputs ofboth sensors are fed into aWheatstone bridge (Figure 10-1).While the sensor is in the vaporphase, the heated probe will bewarmer than the reference probe,and the bridge circuit will be unbal-anced. When both probes are sub-merged in the process liquid, theirtemperatures will approach that of theliquid. Their outputs will be nearlyequal and the bridge will be in balance.This level switch is actuated when achange in bridge balance occurs.

Since all process materials have acharacteristic heat transfer coeffi-cient, thermal level switches can becalibrated to detect the presence orabsence of any fluid. Therefore, theseswitches can be used in difficult ser-vices, such as interfaces, slurry, andsludge applications. They can alsodetect thermally conductive foams ifspray-cleaned after each operation.

Thermal level and interface switcheshave no mechanical moving parts andare rated for pressures up to 3,000psig and process temperatures from-75 to 175°C (-100 to 350°F). Whendetecting water level, response time is

typically 0.5 second and accuracy iswithin 2 mm. In general, thermal levelswitches work best with non-coatingliquids and with slurries having 0.4-1.2specific gravity and 1-300 cP viscosity.

A third type of thermal switchalso uses two sensors inside thesame vertical probe. One is mount-ed above the other and both areconnected to a voltage source.When both are in the vapor or bothin the liquid phase, the current flow

through the two sensors is the same.If, on the other hand, the lower oneis in liquid and the upper in vapor,more current will flow through thelower sensor. A current comparatorcan detect this difference and signalthat the sensor has reached thevapor/liquid interface.

One interesting feature of thisdesign is that the sensor capsule canbe suspended by a cable into a tank orwell, and the sensor output can beused to drive the cable take-up motor.In this fashion, the level switch can beused as a continuous detector of thelocation of the vapor/liquid interface.

Thermometers also can be used todetect level in higher temperatureprocesses, such as measuring thelevel of molten steel in casting molds.The thermometers do not actuallytouch the molten metal; instead, theyidentify the place where the temper-ature on the outside of the mold

suddenly increases. This is the levelinside the mold. Using multiple sen-sors spaced vertically, the systemcan determine the level of moltenmetal in the mold to within a frac-tion of an inch.

Vibrating SwitchesVibrating level switches detect thedampening that occurs when avibrating probe is submerged in aprocess medium. The three types ofvibrating sensors—reed, probe, andtuning fork—are distinguished bytheir configurations and operatingfrequencies (120, 200-400, and 85 Hz,respectively). Their methods of oper-ation and applications are similar.The reed switch consists of a paddle,a driver and a pickup (Figure 10-2).The driver coil induces a 120-Hzvibration in the paddle that isdamped out when the paddle getscovered by a process material. Theswitch can detect both rising andfalling levels, and only its actuationdepth (the material depth over thepaddle) increases as the density ofthe process fluid drops. The variationin actuation depth is usually less thanan inch. A reed switch can detect liq-uid/liquid, liquid/vapor, andsolid/vapor interfaces, and can alsosignal density or viscosity variations.

When used on wet powders, thevibrating paddle has a tendency tocreate a cavity in the granular solids.If this occurs, false readings willresult, because the sensor will con-fuse the cavity with vapor space.

It is best to use a reed switch onnon-coating applications or to pro-vide automatic spray washing aftereach immersion in a sludge or slurry.Probe-type vibrating sensors are lesssensitive to material build-up or coat-ing. The vibrating probe is a roundstainless steel element (resembling a

10 Specialty Level Switches


Figure 10-2: Vibrating Reed Switch

Pickup End Driver End

NodePaddle

thermowell) that extends into thematerial. If Teflon®coated and insert-ed at an angle, these devices tend tobe self-cleaning. Both the drive andthe sensor are piezoelectric ele-ments: one causes the vibration andthe other measures it. When theprobe is buried under the process

material, its vibration is dampenedand this decrease triggers the switch.

Vibrating probe sensors can beused to monitor powders, bulksolids, and granular materials such asgrain, flour, plastic pellets, cement,and fly ash. Their vibrating naturetends to minimize the bridging thatoccurs in solid materials. Tuning forksensors are vibrated at about 85 Hzby one piezoelectric crystal, whileanother piezoelectric crystal detectsthe vibration. As the process fluidrises to cover the tuning forks, thevibration frequency changes.

Like vibrating probes, tuning-forkdesigns can be self-cleaning ifTeflon® coated and installed at anangle. They can also be calibrated todetect a wide range of materials,including lubricating oils, hydraulicfluids, water, corrosive materials,sand, thick and turbulent fluids, pow-ders, light granules, and pastes.

Tuning-fork sensors can be con-structed with components made of

PVDF, polypropylene, stainless steel,carbon steel, and aluminum. Theyare available with Teflon®coatings orin hygienic versions for sanitaryapplications.

Vibrating sensors can be used toascertain liquid, solid, and slurry lev-els. Reed switches can operate at

pressures up to 3,000 psig, while tun-ing forks and vibrating probes arelimited to 150 psig. Operating tem-peratures range from -100 to 150°C(-150 to 300°F) and response time isabout 1 second.

Optical SwitchesUsing visible, infrared, or laser light,optical sensors rely upon the lighttransmitting, reflecting, or refractingproperties of the process materialwhen measuring its level. The opticallevel switch can be of a contacting or

non-contacting design.In a non-contacting, reflecting

optical sensor, a beam of light isaimed down at the surface of theprocess material. When the level ofthis surface rises to the setpoint ofthe switch, the reflected light beam isdetected by a photocell. Both theLED light source and photodetectorare housed behind the same lens.

By adjusting the photocell or thedetection electronics, the sensor canbe calibrated to detect levels at dis-tances 0.25 to 12 in below the sensor.These reflective switches can measurethe levels of clear as well as translu-cent, reflective, and opaque liquids.Some solids also can be detected. Byusing multiple photocells, a sensorcan detect several levels.

Laser light also can be used whenmaking difficult level measurements,such as of molten metals, moltenglass, glass plate, or any other kind ofsolid or liquid material that has areflecting surface. If the receivermodule is motor driven, it can trackthe reflected laser beam as the levelrises and falls, thereby acting as acontinuous level transmitter.

A refracting sensor relies on theprinciple that infrared or visiblelight changes direction (refracts)when it passes through the interfacebetween two media. When the sen-sor is in the vapor phase, most of the

Specialty Level Switches 10


Figure 10-3: Optical Level Switch

ReceiverLED

Prism

Light from LED

Liquid Below the Sensing Prism

Liquid Immersing the Sensing Prism

LEDLEDReceiver

PrismLight

Lost in Liquid

Ultrasonic liquid level switches provide a 300:1 signal ratio from dry to wetted state.

light from the LED is reflected backwithin a prism (Figure 10-3). When theprism is submerged, most of the lightrefracts into the liquid, and theamount of reflected light that reachesthe receiver drops substantially.Therefore, a drop in the reflectedlight signal indicates contact withthe process liquid.

A refracting sensor cannot be usedwith slurries or coating liquids, unlessit is spray-washed after each submer-sion. Even a few drops of liquid on theprism will refract light and cause erro-neous readings. Refracting sensors aredesigned to be submerged in liquids;therefore, any number of them can beinstalled on a vertical pipe to detect anumber of level points.

Transmission optical sensors send abeam of light across the tank. Asludge level sensor of this design usesan LED and a photocell at the end ofa probe, located at the same eleva-tion and separated by a few inches. Tofind the sludge level, a mechanism (oran operator, manually) lowers theprobe into the tank until the sensors

encounter the sludge layer.Other transmission sensors rely on

the refraction principle utilizing anunclad, U-shaped fiber optic cable. Alight source transmits a pulsed lightbeam through the fiber cable, andthe sensor measures the amount oflight that returns. If liquid covers thecable, it will cause light to refractaway from the cable. The use of fiber-optics makes the system imperviousto electrical interference, and somedesigns are also intrinsically safe.

Optical sensors can operate atpressures up to 500 psig and tem-

peratures up to 125°C (260°F).Response time is virtually immedi-ate, and detection accuracy of mostdesigns is within 1 mm. Optical levelswitches are also designed for spe-cific or unique applications. Forexample, Teflon® optical levelswitches are available for sensingthe level of ultra-pure fluids. Otherunique designs include a levelswitch that combines an opticalwith a conductivity-type level sen-sor to detect the presence of bothwater (conductive) and hydrocar-bons (nonconductive). T

10 Specialty Level Switches




Edition, OMEGA Press, 1995.• Industrial Control Handbook, E.A. Parr, Butterworth-Heinemann Ltd., 1995.• Instrument Engineer’s Handbook, Bela Liptak, Third Edition, CRC Press, 1995.• Process/Industrial Instruments and Controls Handbook, 4th Edition,

Douglas M. Considine, McGraw-Hill, 1993.• The McGraw-Hill Encyclopedia of Science and Technology, 8th Edition,

John H. Zifcak, McGraw-Hill, 1997.

Flow and Level Measurement Handbook

Documents

Transcript of Flow and Level Measurement Handbook