Baker R.C. Flow Measurement Handbook: Industrial Designs, Operating Principles, Performance, and Applications

Подождите немного. Документ загружается.

4.4 GAS CALIBRATION FACILITIES

Spring

Plenum

Pressure

(c)

Start of Run

Optical Switch End of Run

Optical Switch

Poppet Valve (Closed)

Actuator

Piston

Hydraulic

Solenoid

(Open)

Upstream Fluid Inlet

End of Run

Optical Switch

Fail Safe Stop

Poppet Valve (Open)

Spring

Plenum

Pressure

Hydraulic

Solenoid

(Closed)

(d)

Figure 4.11. (continued).

and so eliminate contamination. It can also be skid mounted. Wherever possible,

comparisons between ship and meter quantities should be monitored continuously

to provide warning of meter factor drift (cf. Hannisdal 1991 on alternative metering

concepts to reduce space and weight).

4.4 GAS CALIBRATION FACILITIES

4.4.1 VOLUMETRIC MEASUREMENT

The bell prover uses a standing start and stop. Figure 4.13 shows the main features.

An inverted vessel - the bell - is held by a cable and counterweight so that it dips

into a sealing fluid. This is usually water or a light oil. The flowmeter to be calibrated

is connected into a pipe that allows the gas in the gas cavity under the bell to flow

out. The bell is allowed to sink into the liquid so that the gas is expelled from

the cavity through the outlet pipe and so through the flowmeter under test. It is

therefore possible to take the height of the vessel at start and end and to deduce

the volume of gas that has passed. For high accuracy, various other factors need

to be taken into account. The volume of gas in the vessel will be affected by the

pressure and temperature within the vessel. To a lesser extent, it will be affected

CALIBRATION

1 1

1 1 1

SECOND

TIMER

ALTERNATIVE

FIRST TIMER

SECOND

(A)

TIMER

(B)

i i

i 111II

SECOND .

TIMERJ

1ST

PROVER DETECTOR

VOLUME

SWITCH

(a)

2ND

PROVER DETECTOR

VOLUME

SWITCH

FIRST

DETECTOR

SECOND

DETECTOR

-CALIBRATED VOLUME

FIRST

DETECTOR SWITCH

FINAL

DETECTOR SWITCH

MEASUREMENT

PISTON

VOLUME'D'

TIMEW/VWV

—

PROVEN VOLUME

ELAPSED TIME TO

COLLECT WHOLE

FLOWMETER PULSES

(b)

TIME

TIMEB

Figure 4.12. Interpolation of flowmeter pulses for compact prover. (a) General

concept and double chronometry method for pulse interpolation (after Reid and

Pursley 1986). (b) Double chronometry method as used by Fisher-Rosemount

(reproduced with permission).

4.4 GAS CALIBRATION FACILITIES

COUNTER

WEIGHT

GAS

MOTION

VALVE

Figure 4.13. Bell prover.

by change in the volume of the vessel. The

humidity of the gas will be important if wa-

ter is used instead of oil. The movement

of the vessel, if continuous monitoring of

the flow is taking place, may need to be al-

lowed for. This method is particularly suit-

able for small domestic gas meters and is

capable of achieving an uncertainty of as

little as ±0.2% if particular care is taken.

There are a number of elderly bell provers

still in use, but there are also some new

ones providing high accuracy in national

laboratories.

A similar approach, using a large bag

of known volume, accommodates larger volumes.

Bellinga and Delhez

(1993,

cf. Bellinga et al. 1981) of Nederlandse Gasunie,

Groningen, described the conversion of a portable piston prover for liquids to use

with gases; the prover is capable of flows from 20 to 2000 m

/h. The cylinder length

was 12 m with a measuring section of 5.22 m, diameter of 0.584 m, and volume of

1.3988

. They also described the traceability chain before the prover came into

operation. During the initial calibration of the piston prover, the volume was de-

termined by weighing the quantity of water displaced by the piston when moving

through the measuring section.

dynamic calibration was achieved by using a mas-

ter meter. As a check, the volume displaced was obtained from measurement of the

dimensions of the prover. They claimed that these different methods should agree

to within

0.03%.

The pressure difference needed to drive the piston at 150 m

/h was

20 mbar, but at lower flow rates the motion became unsteady and was corrected by

reducing the mass of the piston and adding a dynamic damping system. Bellinga

and Delhez also mentioned other small changes to valves, valve operating speed,

and electronics that include pulse interpolation. The aim was for the prover to be

accepted by the Netherlands Measurement Institute as the primary standard for high

pressure gas meter calibration. The meters calibrated from this prover should have

an uncertainty in meter factor of as good as 0.1-0.15%. Reid and Pursley (1986)

considered the use of nitrogen as a pressure-balancing gas to reduce the differential

pressure across the piston and hence to reduce the leakage and to provide stability

of motion at low flow rates.

Reid and Pursley claimed 500:1 operation in air for pressures greater than 30 bar

with agreement to 1% and repeatability of order 0.25% up to 8 bar.

Jongerius et

al.

(1993) described a new low flow test facility consisting of four mer-

cury seal piston provers at the Nederlands Meetinstituut with an operating range of

2 x 10~

-3.5 m

/h at approximately ambient conditions and operable, in principle,

on any gas. The low flow facility is traceable to the primary length standard, and

uncertainty is about 0.2% for the smallest prover and 0.15% for the others.

4.4.2 MASS MEASUREMENT

Figure 4.14 shows the scheme for a gravimetric calibration facility. The system

shown depends on the use of critical nozzles (which will be explained in Chapter 7)

FLOWMETER

UNDER

TEST

-WATER OR

LIGHT OIL

80 CALIBRATION

CRITICAL

FLOW

VENTURI

DIVERTER

UNIT

GAS SUPPLY

AT CONSTANT

PRESSURE AND

TEMPERATURE

PRESSURE

CONTROL

VALVES

WEIGHBRIIDGE

Figure 4.14. Scheme for a gravimetric gas calibration facility (Pursley 1986 reproduced with

permission of NEL).

as transfer standards. The facility is used to calibrate the sonic nozzles, then to

calibrate the flowmeter under test, and, if appropriate, to recalibrate the nozzles.

The problem with gravimetric measurement in gases is the low density of the gas

and the consequent problems of mass measurement. In addition, the flow con-

trol in the calibration facility must be such that the flow rate through the me-

ter under test is the same as that into the weighing vessel. The facility at the

National Engineering Laboratory, Scotland, is used for flows of up to 5 kg/s at

pressures of up to 50 bar to within an estimated uncertainty of ±0.3% (Brain

1978).

4.4.3 GAS/LIQUID DISPLACEMENT

The problems of calibration of gas flow has led to the use of a device that exchanges

liquid flow by volume for gas flow by volume, say, by means of a piston. The liquid

is metered to a high accuracy, and the gas at controlled temperature and pressure is

used to calibrate the flowmeter.

Lapszewicz (1991) described a flowmeter for gas mixtures in which liquid is dis-

placed, and the time and volume between two points on a tube is measured. A

valving system allows the gas to enter and be exhausted.

4.4.4 pvT METHOD

Provided that the gas under calibration obeys a well-defined relationship be-

tween the pressure p, the specific volume v, and the temperature T (e.g., for an

ideal gas),

pv = RT

where R is the gas constant, the change in quantity in a pressure vessel can be

deduced from the change in pressure and temperature, and the volume or mass

4.4 GAS CALIBRATION FACILITIES 81

Pressure

Temperature

Calibrated

Vessel

Heat

Exchanger

Pressure

Control

Valve

Regulating

Sonic

Nozzle

Heat

Exchanger

Test

Meter

Figure 4.15. Diagram of a typical facility for the pvT calibration method (Pursley 1986 repro-

duced with permission of NEL).

passing through the meter under test can be deduced. A simple facility is shown in

Figure 4.15. The flow and temperature of the gas are controlled by heat exchanger

and pressure regulation, and the flow rate may be stabilized by a critical nozzle

followed by a further heat exchanger.

4.4.5 CRITICAL NOZZLES

The design and operation of these devices will be described in Chapter 7. Here we

note only that they are rather inflexible in that the flow through them is, essen-

tially, defined by the inlet pressure and the area of the throat of the nozzle. Thus to

accommodate a range of flows, they need to be set up as a bank of flowmeters. A

typical bank of such flowmeters is shown in Figure 7.2. The approach will be to set

these in series with high quality master flowmeters. In order that the flow range is as

complete as possible, the mass flows through the bank need to be in a progression

such as 1, 2, 4, 8, 16. This will allow flow rates of 1, 2, 3, 4,

5,...,

31, by suitable

combination of the nozzles (cf. Aschenbrenner 1989 who described the calibration

of a bank of 16 nozzles). In such a system, it

will,

clearly, be essential to have a means

of checking the integrity of the valving system so that accidental leaks do not affect

the accuracy. With this range, it is then possible to calibrate the master meters over

the same range and to use these either on the facility, or as transfer standards (cf.

Bignell 1996a on comparison techniques for small sonic nozzles).

4.4.6 SOAP FILM BURETTE METHOD

Figure 4.16 shows a simple arrangement of the soap film burette calibration system

(Pursley 1986). This is suitable for very low flow rates. The gas enters the burette

after leaving the flowmeter, and a soap film, created within the gas flow, is driven up

the burette by the flow of

gas.

The movement can be observed and timed by eye, or

sensing methods, probably optical, can be used to measure its transit time. Clearly

the burette will need to be calibrated for volume. Flow range is 10~

-10~

/s at

conditions near to ambient (Brain 1978).

82 CALIBRATION

TIMING

DEVICE

Figure 4.16. Soap film burette (Pursley 1986 reproduced with permission of

NEL).

4.5 TRANSFER STANDARDS AND MASTER METERS

Calibrated master meters may also be used to measure the flow in a specially con-

structed flow facility in which the masters are in series with the meter to be calibrated.

This may be the most economical approach for many manufacturers. The cost will,

of course, include the regular recalibration of the master meters. Such a facility will

need careful design to achieve:

• steady flow,

• fully developed flow profile upstream of both the reference (master) meter and

the meter under calibration,

• adequate range for all meters to be tested,

• flexible means to accommodate flowmeters of different diameters and axial

lengths and to mount them in the line, and

• high quality sensing and recording systems to provide the comparison between

the meter under calibration and the master meter.

A good transfer standard should have the following characteristics (Harrison

1978b):

• highly repeatable,

• stable with time,

• wide flow range,

• insensitive to installation,

• simple and robust,

• compact,

4.5 TRANSFER STANDARDS AND MASTER METERS 83

• low head loss,

• suitable for a variety of fluids, and

• available in a wide range of sizes.

Meters suitable for use as masters for transfer standard work are, therefore, the

high accuracy meters, which will be described in later chapters. In particular, these

include positive displacement meters and turbine meters. For conducting liquids,

electromagnetic meters may, now, have reached a sufficient standard to be consid-

ered in some cases. Ultrasonic multibeam meters may, now, also be found to be

adequate, as may Coriolis meters. In gas flows, the ultrasonic multibeam may be

suitable, and the critical nozzle meters, as indicated earlier, are very appropriate de-

spite their inflexible flow range. Wet gas meters, although a fairly elderly design, are

still used by some standards laboratories as transfer standards.

The problem with using one meter on its own is that the stability of its signal

may not be observable, and errors may creep in without any check. For this reason,

master meters are often used in pairs, either in series, so that the consistency of their

readings is continually checked, or in parallel when one is used most of the time

and the second is kept as a particularly high accuracy meter for occasional checks.

A development to achieve very high performance and to reduce sensitivity to

installation is the calibration of the meter with the upstream pipework and a flow

straightener permanently in position, and the use of two of these packages in series

(Figure 4.17; this arrangement was suggested by Mattingly et al. 1978, with or with-

out the initial flow straightener). In this way, any deviation caused by installation

or by drift of one meter will appear as a relative shift in calibration between the two

meters. Such meters or meter pairs are referred to as transfer standards in that they

allow a calibration standard to be transferred (with only a small increase in uncer-

tainty) from a nationally certified facility to a manufacturer's facility or a research

laboratory facility.

The meters selected as transfer standards must be capable of retaining a high

performance with removal, transport, calibration, subsequent removal, transport,

reinstallation, removal, storage, etc! Several of the main flowmeter designs will meet

this requirement, and the actual selection will depend on other factors. The line

liquid should be without gas bubbles or cavitation. The meters should be selected

for optimum range; in some cases, overlapping ranges may be an advantage.

Apart from flow profile, the pipework for transfer standard installation would

have the following features:

COMPLETE

PACKAGE

FLOW FLOW

CONDITIONERS METERS

Figure 4.17. Transfer package allowing two meters, conditioners, and pipework to be calibrated

together.

CALIBRATION

PERMISSIBLE

ERROR

0.5

-0.5

•\

-2

CALIBRATION POINTS

••>

0 20 25 40 70 100

-| ). ^ |

FLQw

PATH

Figure 4.18. Maximum permissible errors: solid line, ISO

9951;

broken line, Gasunie specifica-

tion (after van der Kam and Dam 1993 with permission of Elsevier Science).

• uniquely linking meters with convenient valves to control flow path and for ease

of access;

• suitable environment, meeting safety requirements;

• freedom from vibration and cathodic protection; and

• surrounding pipework clean, in good condition, and of the same diameter as

transfer package.

Van der Kam and Dam (1993) described the replacement of orifice meters by

turbines, in export stations of Nederlandse Gasunie. Even though

ISO 9951

specifies a

Maximum Permissible Error (MPE) as

±1%

in the upper range, van der Kam and Dam

reported that, in their experience, turbines can operate within an envelope of ±0.5%

down to about 25% maximum flow and of ±1% in the lower range (Figure 4.18).

They suggested that the four-path ultrasonic meter would probably be best of all.

Pereira and Nunes (1993) described a facility with an array of ISA 1932 nozzles

as the calibration source. The sizes were such as to allow a 50:1 turndown in air, and

where nozzles were not needed, rubber balls were inserted in the inlets to seal them

off and prevent flow. Uncertainty was claimed to be in the range 0.5-0.8%.

4.6 IN SITU CALIBRATION

A new flowmeter should have some performance specification. The need to keep

documentary evidence on flowmeters should be reinforced. It may start well with

a manufacturer's minimal or full calibration. The flowmeter should, therefore, have

a known performance under reference conditions. Let us assume that this is ±1%

uncertainty with 95% confidence, a reasonable and not uncommon value. What

happens when we install this flowmeter on site?

In many cases, there may not be an adequate upstream straight length to retain

the calibration, and this may be further degraded by misalignment, etc. Let us

assume that this will cause a total uncertainty of ±2%. The meter then starts its

working life, and we cease to have any satisfactory means of predicting the likely drift

that results. Recent experience from the water industry suggests that this may range

from 5% to as much as 50%. Because of this uncertainty, it is necessary to determine

4.6 IN SITU CALIBRATION 85

a documentation, maintenance, and recalibration schedule for each meter. One way

to decide the period between recalibration is by experience. If, after the first recali-

bration, the shift is unacceptably large, then the next period to recalibration should

be half the length. If the calibration is essentially unchanged, then the period may

be increased. Descriptions such as "initial period/' "essentially unchanged/' and

"unacceptably large" are rather subjective, but they may be given a value based on

lost revenue against costs of maintenance and calibration. The outcome of all this is

the need to recalibrate, and one choice is in situ calibration with a likely uncertainty

of, say, ±3%, considerably poorer than originally obtained on a test stand.

In deciding between the merits of test stand and in situ calibration, for a specific

case,

the reader should develop a table with pros and cons in terms of ultimate

uncertainty and total costs (including removal, transport, calibration, lost revenue,

and reinstallation). Other factors such as damage to the meter should also be allowed

for in the decision.

Methods of in situ calibration include

• provers (sphere, piston, and tank);

• other meters (transfer standards);

• pipework features (inlets and bends);

• drop test;

• in situ meter measurement and inspection (dry calibration);

• clamp-on ultrasonic flowmeters (see Chapter 13); and

• probes (see Chapter 18) and tracers.

Traceability for the provers and transfer standards will be through a calibration

facility. For other methods, traceability will be more tenuous, and in situ verification

will be a more appropriate term.

Provers

Provers have already been covered but are mentioned here for completeness. The

sphere prover (Figure 4.10) is widely used in permanent systems with banks of tur-

bine meters in the oil industry. The piston prover can be trailer mounted. Claims

of order 0.1% or better are sometimes made for these devices (cf. Eide 1991 who

claimed ±0.03%). The volumetric tank prover (Figure 4.9) discussed earlier is reck-

oned capable of the same sort of accuracy level.

Other Meters

An example of

transfer standard used in a survey of domestic water consumption in

Scotland (Harrison and Williamson 1985) is shown in Figure 4.19 and was capable of

calibrating meters from 0.004 to 35.7 m

/h. This was achieved by using three meters

within the transfer standard:

• an Avery-Hardoll PD meter in series with a Kent 50-mm Helix 3000 flowmeter -

range 35.7-3.7 m

/h,

• a Kent 50-mm Master 2000 flowmeter in series with a Kent 50-mm Helix 3000

flowmeter - range 10.8-0.7 m

/h, and

• two Kent 12-mm PSM meters in series - range 2.7-0.004 m

/h.

86 CALIBRATION

Inlet

Outlet

Figure 4.19. Wide flow range reference meter transfer standard (Harrison and Williamson 1985

reproduced with permission of

NEL).

The Borda inlet chamber allowed the flow to settle prior to the Helix meter, which

is sensitive to installation.

It is obviously necessary for provers, such as this one, to be coupled into the site

pipework in series with the meter to be tested.

Pipework Features - Inlets, Bends/Elbows

The inlet from a large tank to a pipe may be used to obtain a measure of the flow

rate.

The pressure difference between the inside and the outside as flow goes round

a bend or elbow may be used to obtain flow rate. The accuracy of these methods will

depend on how well-formed the pipework features are. The equations and methods

are described in Chapter 8.

Drop Test

In the drop test method, the volumetric flow through the meter is compared with

the volume change due to a level change in a reservoir (Figure 4.20). To achieve this,

we require the following:

a. A suitable system with the pipework leading from the reservoir so as to allow

closure of all flows except that which leads from the reservoir to the meter under

test. There must be confidence that the water leaving the reservoir all passes

through the test meter. There must also be a valve suitable for controlling the

flow through the meter.

Strapping

tables,

which give the relationship between water level and the volume

of water in the reservoir.

c. A means to obtain level of water in the reservoir.