U.S. patent application number 11/542710 was filed with the patent office on 2008-04-10 for method, apparatus and computer medium for correcting transient flow errors in flowmeter proving data.
Invention is credited to Donald R. Augenstein, Herbert Estrada, Matthew Mihalcin.
Application Number | 20080083262 11/542710 |
Document ID | / |
Family ID | 39271209 |
Filed Date | 2008-04-10 |
United States Patent
Application |
20080083262 |
Kind Code |
A1 |
Augenstein; Donald R. ; et
al. |
April 10, 2008 |
METHOD, APPARATUS AND COMPUTER MEDIUM FOR CORRECTING TRANSIENT FLOW
ERRORS IN FLOWMETER PROVING DATA
Abstract
An apparatus for increasing the accuracy of meter factors of a
flow instrument in conjunction with a prover having upstream and
downstream prover detection switches and a flow computer including
a computer program for correcting errors in a meter factor measured
in a proving run and/or correcting effects of flow rate changes
during proving. A computer readable medium whose contents causes a
processor to increase the accuracy of meter factors of a flow
instrument in conjunction with a prover having upstream and
downstream prover detection switches and a flow computer, by
performing the steps of receiving signals from the upstream and
downstream prover detection switches and the flow instrument. There
is the step of correcting errors in a meter factor measured in a
proving run. A method for increasing the accuracy of the meter
factors of a flow instrument in conjunction with a prover having
upstream and downstream prover detection switches and a flow
computer. The method includes the steps activating a processor.
There is the step of correcting errors in the meter factor measured
in a proving run with a computer program in the processor.
Inventors: |
Augenstein; Donald R.;
(Pittsburgh, PA) ; Estrada; Herbert; (Annapolis,
MD) ; Mihalcin; Matthew; (Pittsburgh, PA) |
Correspondence
Address: |
Ansel M. Schwartz;Attorney at Law
Suite 304, 201 N. Craig Street
Pittsburgh
PA
15213
US
|
Family ID: |
39271209 |
Appl. No.: |
11/542710 |
Filed: |
October 4, 2006 |
Current U.S.
Class: |
73/1.16 |
Current CPC
Class: |
G01F 25/0007
20130101 |
Class at
Publication: |
73/1.16 |
International
Class: |
G01F 25/00 20060101
G01F025/00 |
Claims
1. An apparatus for increasing the accuracy of meter factors of a
flow instrument in conjunction with a prover having upstream and
downstream prover detection switches and a flow computer, the
apparatus comprising a flow instrument having a time constant, and
a computer program embodied on a computer readable medium for
correcting errors in a meter factor of the flow instrument measured
in a proving run produced by flow changes in a proving run.
2. An apparatus as described in claim 24 including a container
having a processor and the program.
3. An apparatus as described in claim 24 wherein the flow
instrument includes the program.
4. An apparatus as described in claim 24 comprising a flow computer
wherein the flow computer includes the program.
5. An apparatus as described in claim 24 wherein the program
includes a filter time constant which reduces the number of proving
runs required to establish a meter factor within a specified
accuracy.
6. An apparatus as described in claim 24 wherein the program is
configured to detect flow rate changes prior to or during a proving
run, which flow rate change would result in an unacceptable error
in the meter factor determined by the proving run, whether or not
the effect of the flow rate change is accounted in the meter factor
determination.
7. An apparatus as described in claim 24 wherein the program
calculates the error produced by a flow rate change, taking into
account signal processing, including data filtering, employed by
the flow instrument that measures the flow rate.
8. An apparatus as described in claim 24 wherein the program
corrects the meter factor obtained by a proving run during which
flow rate changes by an amount that, without correction, would lead
to an unacceptable error in the meter factor, such that the
corrected meter factor has essentially no error.
9. An apparatus as described in claim 24 wherein the program
corrects error from a signal indicating a position of the upstream
prover detection switch, a signal indicating the position of the
downstream prover detection switch, a signal equal to flow rate as
measured and filtered by the flow instrument to be proved,
synchronized with the actuation of the upstream prover detection
switch, a signal equal to flow rate as measured and filtered by the
flow instrument to be proved, synchronized with the actuation of
the downstream prover detection switch, an input equal to a value
of a pulse train frequency/measured flow constant of the flow
instrument to be calibrated, and an input equal to the value of a
time and a constant of a data filter employed by the flow
instrument being proved.
10. An apparatus as described in claim 24 wherein the program
corrects error from a signal indicating the position of the
upstream prover detection switch, a signal indicating a position of
the downstream prover detection switch, an input equal to
instantaneous pulse train frequencies synchronized with actuations
of the upstream and downstream prover detection switches, an input
equal to the value of a pulse train frequency/measured flow
constant of the flow instrument to be calibrated, and an input
equal to the value of a time and a constant of a data filter
employed by the flow instrument being proved.
11. An apparatus as described in claim 24 wherein the program
includes an algorithm as follows:
N.sub.mc=N.sub.m+[.differential.f/.differential.Q.times.(Q.sub.b-Q.sub.a)-
.times..tau.]; Here N.sub.mc is the number of meter pulses produced
during a proving run corrected for the filtering used by an
ultrasonic or other meter; N.sub.m is the integer number of pulses
measured during the proving run, as previously defined;
.differential.f/.differential.Q is a pulse frequency/flow rate
constant employed by a meter; Q.sub.b is flow rate measured by a
meter at the instant the downstream prover detection switch 18 is
actuated; Q.sub.a is flow rate measured by a meter at the instant
the upstream prover detection switch 16 is actuated; .tau. is a
time constant of a single pole low pass filter used to smooth an
instantaneous flowmeter output.
12. An apparatus as described in claim 24 including a memory
containing the program.
13. A computer program embodied on a computer readable medium whose
contents causes a processor to increase accuracy of meter factors
of a flow instrument in conjunction with a prover having upstream
and downstream prover detection switches and a flow computer
comprising the computer implemented steps of: receiving signals
from the upstream and downstream prover detection switches and the
flow instrument having a time constant; and correcting errors in a
meter factor of the flow instrument measured in a proving run
produced by flow changes in a proving run.
14. A method for increasing accuracy of meter factors of a flow
instrument in conjunction with a prover having upstream and
downstream prover detection switches and a flow computer comprising
the steps of: activating a processor; and correcting errors in a
meter factor of a flow instrument having a time constant measured
in a proving run produced by flow changes in a proving run by
implementing a computer program in a processor.
15. A method as described in claim 25 wherein the activating step
includes the step of activating the processor in a container.
16. A method as described in claim 25 wherein the activating step
includes the step of activating the processor in the flow
instrument.
17. A method as described in claim 25 wherein the activating step
includes the step of activating the processor in a flow
computer.
18. A method as described in claim 25 wherein the correcting step
includes the step of correcting errors with the computer program
using a filter time constant which reduces the number of proving
runs required to establish a meter factor within a specified
accuracy.
19. A method as described in claim 25 wherein the correcting step
includes the step of correcting errors with the computer program
that detects flow rate changes prior to or during a proving run,
which flow rate change would result in an unacceptable error in a
meter factor determined by a proving run, whether or not the effect
of a flow rate change is accounted in a meter factor
determination.
20. A method as described in claim 25 wherein the correcting step
includes the step of correcting errors with the computer program
that calculates the error produced by a flow rate change, taking
into account signal processing, including data filtering, employed
by the flow instrument that measures flow rate.
21. A method as described in claim 25 wherein the correcting step
includes the step of correcting errors with the computer program
that corrects a meter factor obtained by a proving run during which
the flow rate changes by an amount that, without correction, would
lead to an unacceptable error in a meter factor, such that a
corrected meter factor has essentially no error.
22. A method as described in claim 25 including the steps of
receiving a signal indicating a position of the upstream prover
detection switch, receiving by the processor a signal indicating
the position of the downstream prover detection switch, receiving
by the processor a signal equal to flow rate as measured and
filtered by a meter to be proved, synchronized with an actuation of
the upstream prover detection switch, receiving by the processor a
signal equal to the flow rate as measured and filtered by a meter
to be proved, synchronized with the actuation of the downstream
prover detection switch, receiving by the processor an input equal
to the value of a pulse train frequency/measured flow constant of a
meter to be calibrated, and receiving by the processor an input
equal to the value of a time and a constant of a data filter
employed by a meter being proved.
23. A method as described in claim 25 wherein the program employs
an algorithm as follows:
H.sub.mc=N.sub.m+[.differential.f/.differential.Q.times.(Q.sub.b-Q.sub.a)-
.times..tau.]. Here N.sub.mc is the number of meter pulses produced
during a proving run corrected for the filtering used by an
ultrasonic or other meter; N.sub.m is the integer number of pulses
measured during the proving run, as previously defined;
.differential.f/.differential.Q is a pulse frequency/flow rate
constant employed by a meter; Q.sub.b is flow rate measured by a
meter at the instant the downstream prover detection switch 18 is
actuated; Q.sub.a is flow rate measured by a meter at the instant
the upstream prover detection switch 16 is actuated; .tau. is a
time constant of a single pole low pass filter used to smooth an
instantaneous flowmeter output.
24. An apparatus as described in claim 1 wherein the flow
instrument is either an ultrasonic flow meter or a coriolis or
vortex shredding meter.
25. A method as described in claim 14 wherein the correcting errors
step includes the step of correcting errors in the meter factor of
either an ultrasonic flow meter or a coriolis or vortex shredding
meter.
Description
FIELD OF THE INVENTION
[0001] In accordance with certain embodiments, the present
invention is related to the ability to "prove" transit time
ultrasonic flowmeters as well as meters based on other
technologies, using any standard but particularly using small
volume provers.
BACKGROUND OF THE INVENTION
[0002] Small volume provers are used extensively in the petroleum
industry as a means to prove custody transfer meters--meters that
account the volume of product delivered from one party to another.
(See the American Petroleum Institute (API) Manual of Petroleum
Measurement Standards, Chapter 4, Section 3 for a description of
the design of small volume provers.) The proving process confirms
or modifies the meter calibration through the application of a
meter factor. It is carried out as follows: [0003] Flow through the
meter to be calibrated is also directed through the prover. The
flow causes a piston or similar device within the prover to pass
between two position detector switches. The product volume measured
by the meter to be calibrated during the period between detector
switch actuations is compared against the volume of the prover
between the detector switches. (Prior to using the prover, the
volume between detectors is measured using a standard volume
container traceable in the US to NIST. In other countries the
standard volume is traceable to other national standard
laboratories.) The ratio of the prover volume to the uncorrected
volume measured by the meter is the meter factor. Each such meter
factor determination is referred to as a "proving run". Typically
multiple proving runs are performed, and a meter factor for future
operations is determined from the average of the individual prove
measurements. The accuracy of the meter factor determination is
enhanced by the use of multiple prove runs.
[0004] Proving transit time ultrasonic meters with small volume
provers presents a problem that is inherent in the application of
the technology to the measurement of fluid flow over a period of
short duration. In order to understand this problem, it is useful
first to describe, as an example, the operation of ultrasonic
meters.
[0005] Transit time ultrasonic meters determine fluid velocity by
measuring the transit times of pulses of ultrasonic energy
traveling along clearly defined paths, with and against the
direction of the flow. Ultrasonic meters used for custody transfer
typically employ multiple paths, sometimes in a chordal
arrangement, so that the multiple velocity measurements can be
combined according to an appropriate algorithm to determine
volumetric flow rate. A flow rate measurement is thus performed
using a set of transit time measurements. In the majority of
petroleum product flow measurements, the flow is turbulent, which
means that local fluid velocities vary spatially and temporally
about some average. As a result a single flow rate measurement,
comprised of a set of chordal measurements, will not, in general,
represent the true average flow rate present at that time, but may
vary over a range of about .+-.2%. If many flow rate measurements
are made, it will be found that they form a normal distribution,
centered on the average flow rate and having a standard deviation
of about 1%. The uncertainty of the mean of a normally distributed
population of measurements is equal to the standard deviation of
the distribution divided by the square root of the number of sample
measurements taken from the population. The average of multiple
measurements can therefore yield an accurate measure of the average
flow rate prevailing during the time over which the measurements
are made.
[0006] The contained volume of a small volume prover is, as the
name implies, small; consequently the duration of a prove--the
elapsed time between actuations of the first and second detector
switches--is short, typically between 1/2 and 1 second. The rate at
which transit time ultrasonic flowmeters sample the flow rate
varies among manufacturers, but is usually between 5 and 100 Hz.
Using a sample rate of 25 Hz as an example, the number of flow rate
samples obtained during a 1 second prove will be 25. If, as
described in the preceding paragraph, the individual sample
measurements obtained during the prove vary randomly about the
mean, in a normal distribution with a standard deviation of 1%, the
uncertainty in the mean meter factor obtained during the prove is
1%/(25).sup.1/2=.+-.0.2% (one standard deviation). Put another way,
the meter factors determined from repeated proves will be randomly
distributed about the true mean in a distribution having a standard
deviation of 0.2%. API standards for custody transfer require that
the meter factor be determined with an accuracy of .+-.0.027% at a
95% confidence level, which is equivalent to two standard
deviations. Application of statistical analyses shows that over 200
prover runs would be required to determine a meter factor with the
requisite accuracy.
[0007] Coriolis and vortex shedding meters, as well as other meters
whose instantaneous outputs fluctuate about the true flow are
subject to similar constraints. Like ultrasonic meters, these
meters create a pulse train representative of volume artificially,
via a variable frequency oscillator whose frequency is set by the
measured flow rate.
[0008] These discussions illustrate an important difference between
the meters listed in the paragraph above and flow measurement
devices traditionally used in the petroleum industry. As described
above, ultrasonic flowmeters compute volumetric flow rate from a
finite sample of velocities measured along acoustic paths whereas
traditionally used instruments, specifically turbine meters and
positive displacement meters, respond continuously to the flow
field as a whole--the determination of volumetric flow is inherent
in their principles of operation. Turbine meters and positive
displacement meters are therefore less sensitive to turbulent
variations and can usually be proved in a relatively small number
of prover runs, even with a small volume prover.
[0009] Despite this advantage, turbine meters and positive
displacement meters are gradually being replaced by ultrasonic
meters because the maintenance costs of the latter meters, which
have no moving parts, are far lower. Extensive testing by the API
and others, using large provers and master meters, has demonstrated
that transit time ultrasonic meters are capable of delivering an
accuracy of .+-.0.027% in petroleum applications. There is
therefore a significant incentive to find a way whereby they can be
calibrated effectively with small volume provers.
[0010] To address the proving shortcomings of ultrasonic meters,
designers have turned to filtering of the raw flow rate samples,
processing multiple samples to form a "smoothed" flow rate
measurement. Similar measures have been taken by designers of
coriolis meters. In most instances, the signal processing amounts
to a single time constant low pass filter. With ultrasonic meters
time constants as short as 0.1 seconds or as long as 10 seconds may
be employed. This practice has the effect of extending the proving
period (because more prove samples are incorporated in the
determination of flow during the prove). It has however a
significant weakness: If the actual flow rate changes just before
or during a proving run, the meter factor determined from the
proving data will be biased by an amount dependent on the sign and
magnitude of the flow change. (The subject of errors due to sample
delays as well as smoothing time constants has been explored by an
API Task Group. See "Proving Liquid Meters with Microprocessor
Based Pulse Outputs", K. D. Elliot, North Sea Workshop, October,
2005, incorporated by reference herein.) The meter factor bias
(which must be viewed as an error since it will not be detected) is
shown as a function of the filter time constant in FIG. 1 for both
step and ramp changes during a 1 second prove.
[0011] Some discussion of FIG. 1 is warranted. First, it should be
noted that flow rate can and usually does change during a prove.
The change is sometimes caused by the hydraulic resistance inserted
by the prover itself in the flow circuit--in this case the bias
produced by the flow change will be systematic and will not be
removed by repeated proves. The API recognizes that flow changes
can be a problem; changes can introduce errors in turbine meters as
well as ultrasonic meters. Consequently, a 5% flow change during a
prove is a generally accepted limit. Interpreting FIG. 1 on this
basis: If a step change of 5% took place at the beginning of the
prove of an ultrasonic meter having a 5 second time constant, the
meter factor determined from the data of that prove would be in
error by 0.9.times.5%=4.5%. This error is obviously far in excess
of the .+-.0.027% allowable uncertainty in the meter factor (and
the allowable uncertainty must also accommodate the prover
uncertainty and the statistical variations in measured meter factor
due to turbulence).
[0012] Even with shorter meter time constants the bias is
significant. Suppose for example a meter employs a smoothing time
constant of 0.1 seconds. FIG. 1 indicates that a step change of 5%
during a 1 second prove of this meter would introduce an error of
0.1.times.5%=0.5%--still far outside the desired accuracy
bound.
Existing Corrections to the Small Volume Proving Process
[0013] Because rotational speed of turbine and positive
displacement meters is proportional to flow rate, the number of
rotations produced by these meters during a proving run is a direct
indication of the volume of fluid that has passed through them
during the run. The rotors of these meters are equipped with a
proximity detecting arrangement which produces a pulse train that
allows the turns to be counted. In turbine meters the proximity
devices are often affixed to each blade, so that fractions of a
turn may be counted; similar measures may be taken with positive
displacement meters. Despite these measures, the number of pulses
produced by turbine meters and positive displacement meters during
a small volume proving run may not be consistent with the precision
requirement for the calibration: .+-.0.01%. Consequently, it is
industry practice with small volume provers to employ double
chronometry: a process which enhances the number of pulses measured
by a fraction to achieve the requisite precision. (The principles
of double chronometry are described in Chapter 5, section 6 of the
API Manual of Petroleum Measurement Standards, previously
cited.)
[0014] Specifically, the number of pulses measured is increased as
follows:
N.sub.1=N.sub.m(T.sub.2/T.sub.1) 1)
Here
[0015] N.sub.1 is the corrected number of pulses and is typically
not an integer [0016] N.sub.m is the integer number of pulses
measured during the prove by the meter to be calibrated [0017]
T.sub.2 is the time measured between the actuations of the upstream
and downstream prover detection switches of the prover [0018]
T.sub.1 is the time measured from the leading edge of the first
pulse produced following the actuation of the upstream detector
switch of the prover to the leading edge of the first pulse
following the actuation of the downstream detector switch of the
prover.
[0019] The meter factor, in volume per pulse, is then computed as
the volume of the prover divided by N.sub.1.
[0020] Because their operating principles lead to a flow rate
measurement (as opposed to a volume measurement), ultrasonic
flowmeters, as well as other meters that may benefit from the means
disclosed herein, produce a pulse train representative of the
volume of product passing through them by means of a controlled
oscillator whose frequency is made proportional to the measured
flow rate. Although the meter designer has some control over the
frequency of the pulse train, it is rarely high enough to achieve
the requisite precision during a prove of 1/2 to 1 second duration,
so that double chronometry, as described above is applied to these
meters also. (It should also be noted that the use of a frequency
high enough to provide the requisite resolution may not be
consistent with the capability of the flow computer which receives
the pulses.)
[0021] The meter factor determination and the double chronometry
correction functions typically are not performed in the meter being
calibrated but are carried out in a flow computer which also
controls the proving process and, during normal operation, corrects
meter volumetric output to standard temperature and pressure
conditions.
BRIEF SUMMARY OF THE INVENTION
[0022] In accordance with certain embodiments, the present
invention pertains to an apparatus for increasing the accuracy of
meter factors of a flow instrument in conjunction with a prover
having upstream and downstream prover detection switches and a flow
computer. The apparatus comprises a computer program for correcting
errors in a meter factor measured in a proving run.
[0023] In accordance with further embodiments, the present
invention pertains to computer readable medium whose contents
causes a processor to increase the accuracy of meter factors of a
flow instrument in conjunction with a prover having upstream and
downstream prover detection switches and a flow computer, by
performing the steps of receiving signals from the upstream and
downstream prover detection switches and the flow instrument. There
is the step of correcting errors in a meter factor measured in a
proving run.
[0024] In accordance with yet further exemplary embodiments, the
present invention pertains to a method for increasing the accuracy
of the meter factors of a flow instrument in conjunction with a
prover having upstream and downstream prover detection switches and
a flow computer. The method comprises the steps activating a
processor. There is the step of correcting errors in the meter
factor measured in a proving run with a computer program in the
processor.
BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWING
[0025] In the accompanying drawings, exemplary embodiment of the
invention and exemplary methods of practicing the invention are
illustrated in which:
[0026] FIG. 1 is a draft of calibration error versus filter time
constant.
[0027] FIG. 2 is a graph of change in actual and measured flow
rates versus time.
[0028] FIG. 3 is a graph of actual measured flow rates versus
time.
[0029] FIG. 4 is a block diagram of a first embodiment of the
present invention.
[0030] FIG. 5 is a block diagram of a second embodiment of the
present invention.
[0031] FIG. 6 is a block diagram of a third embodiment of the
present invention.
[0032] FIG. 7 is a block diagram of a fourth embodiment of the
present invention.
[0033] FIG. 8 is a block diagram of the present invention.
DETAILED DESCRIPTION OF THE INVENTION
[0034] Referring now to the drawings wherein like reference
numerals refer to similar or identical parts throughout the several
views, and more specifically to FIGS. 4-8 thereof, there is shown
an apparatus 10 for increasing the accuracy of meter factors of a
flow instrument 12 in conjunction with a prover 14 having upstream
and downstream prover detection switches 16, 18 and a flow computer
20. The apparatus 10 comprises a computer program 22 for correcting
errors in a meter factor measured in a proving run. The computer
program 22 may be disposed in one or more tangible media, such as
non-volatile or volatile memory, for example. Moreover, the
computer program 22 may be effectuated through hardware, software,
or any combination thereof.
[0035] The apparatus 10 can include a container 26 (see FIGS. 6 and
7) having a processor 24 and the program 22. Alternatively, the
flow measuring instrument can include the program 22.
Alternatively, the flow computer 20 includes the program 22. There
is a memory 28 containing the program 22 in communication with the
processor 24.
[0036] The program 22 can use a filter time constant which reduces
the number of proving runs required to establish a meter factor
within a specified accuracy. The program 22 can be used to detect
flow rate changes prior to or during a proving run, which flow rate
change would result in an unacceptable error in the meter factor
determined by the proving run, whether or not the effect of the
flow rate change is accounted in the meter factor determination.
The program 22 can calculate the error produced by a flow rate
change, taking into account signal processing, including data
filtering, employed by the meter that measures the flow rate. The
program 22 can correct the meter factor obtained by a proving run
during which the flow rate changes by an amount that, without
correction, would lead to an unacceptable error in the meter
factor, such that the corrected meter factor has essentially no
error.
[0037] The program 22 can correct error from a signal indicating
the position of the upstream prover detection switch 16, a signal
indicating the position of the downstream prover detection switch
18, a signal equal to the flow rate as measured and filtered by the
meter to be proved, synchronized with the actuation of the upstream
prover detection switch 16, a signal equal to the flow rate as
measured and filtered by the meter to be proved, synchronized with
the actuation of the downstream prover detection switch 18, an
input equal to the value of a pulse train frequency/measured flow
constant of the meter to be calibrated, and an input equal to the
value of a time and a constant of the data filter employed by the
meter being proved. The program 22 can correct error from a signal
indicating the position of the upstream prover detection switch 16,
a signal indicating the position of the downstream prover detection
switch 18, an input equal to instantaneous pulse train frequencies
synchronized with the actuations of the upstream and downstream
prover detection switches 16, 18 an input equal to the value of a
pulse train frequency/measured flow constant of the meter to be
calibrated, and an input equal to the value of a time and a
constant of the data filter employed by the meter being proved.
[0038] The program 22, in certain embodiments, employs an algorithm
as follows:
N.sub.mc=N.sub.m+[.differential.f/.differential.Q.times.(Q.sub.b-Q.sub.a-
).times..tau.].
Here
[0039] N.sub.mc is the number of meter pulses produced during the
proving run corrected for the filtering used by the ultrasonic (or
other) meter. [0040] N.sub.m is the integer number of pulses
measured during the proving run, as previously defined. [0041]
.differential.f/.differential.Q is the pulse frequency/flow rate
constant employed by the meter. [0042] Q.sub.b is the flow rate
measured by the meter at the instant the downstream prover
detection switch 18 is actuated. [0043] Q.sub.a is the flow rate
measured by the meter at the instant the upstream prover detection
switch 16 is actuated. [0044] .tau. is the time constant of the
single pole low pass filter used to smooth the instantaneous
flowmeter output.
[0045] In certain embodiments, the present invention pertains to
computer readable medium whose contents causes a processor 24 to
increase the accuracy of meter factors of a flow instrument 12 in
conjunction with a prover 14 having upstream and downstream prover
detection switches 16, 18 and a flow computer 20, by performing the
steps of receiving signals from the upstream and downstream prover
detection switches 16, 18 and the flow instrument 12. There is the
step of correcting errors in a meter factor measured in a proving
run.
[0046] In certain embodiments, the present invention pertains to a
method for increasing the accuracy of the meter factors of a flow
instrument 12 in conjunction with a prover 14 having upstream and
downstream prover detection switches 16, 18 and a flow computer 20.
The method comprises the steps activating a processor 24. There is
the step of correcting errors in the meter factor measured in a
proving run with a computer program 22 in the processor 24.
[0047] The activating step can include the step of activating the
processor 24 in a container 26. Alternatively, the activating step
includes the step of activating the processor 24 in the flow
instrument 12. Alternatively, the activating step includes the step
of activating the processor 24 in the flow computer 20.
[0048] The correcting step can include the step of correcting
errors with the computer program 22 using a filter time constant
which reduces the number of proving runs required to establish a
meter factor within a specified accuracy. The correcting step can
include the step of correcting errors with the computer program 22
that detects flow rate changes prior to or during a proving run,
which flow rate change would result in an unacceptable error in the
meter factor determined by the proving run, whether or not the
effect of the flow rate change is accounted in the meter factor
determination. The correcting step can include the step of
correcting errors with the computer program 22 that calculates the
error produced by a flow rate change, taking into account signal
processing, including data filtering, employed by the meter that
measures the flow rate. The correcting step can include the step of
correcting errors with the computer program 22 that corrects the
meter factor obtained by a proving run during which the flow rate
changes by an amount that, without correction, would lead to an
unacceptable error in the meter factor, such that the corrected
meter factor has essentially no error.
[0049] The correcting step can include the steps of receiving a
signal indicating the position of the upstream prover detection
switch 16. There is the step of receiving by the processor 24 a
signal indicating the position of the downstream prover detection
switch 18. There is the step of receiving by the processor 24 a
signal equal to the flow rate as measured and filtered by the meter
to be proved, synchronized with the actuation of the upstream
prover detection switch 16. There is the step of receiving by the
processor 24 a signal equal to the flow rate as measured and
filtered by the meter to be proved, synchronized with the actuation
of the downstream prover detection switch 18. There is the step of
receiving by the processor 24 an input equal to the value of a
pulse train frequency/measured flow constant of the meter to be
calibrated. There is the step of receiving by the processor 24 an
input equal to the value of a time and a constant of the data
filter employed by the meter being proved.
Theory for the Correction of UFM Calibration Data
[0050] As was noted above, ultrasonic meter designers usually
process raw flow rate data through a smoothing filter, thereby
reducing the effects of turbulence on their measurement. As was
also noted, this process introduces errors in the meter factor
measured in a proving run, if the flow changes during that run.
This invention provides a technique effectively to correct these
errors. The invention allows raw data to be filtered using time
constants that significantly extend the effective duration of a
proving run. It also eases substantially the requirement on the
hydraulic circuit that flow rate be maintained constant during a
proving run.
[0051] Specifically, certain embodiments of the invention correct
proving data according to the following algorithm:
N.sub.mc=N.sub.m+[.differential.f/.differential.Q.times.(Q.sub.b-Q.sub.a-
).times..tau.] 2)
Here
[0052] N.sub.mc is the number of meter pulses produced during the
proving run corrected for the filtering used by the ultrasonic (or
other) meter. [0053] N.sub.m is the integer number of pulses
measured during the proving run, as previously defined. [0054]
.differential.f/.differential.Q is the pulse frequency/flow rate
constant employed by the meter. [0055] Q.sub.b is the flow rate
measured by the meter at the instant the downstream prover
detection switch 18 is actuated. [0056] Q.sub.a is the flow rate
measured by the meter at the instant the upstream prover detection
switch 16 is actuated. [0057] .tau. is the time constant of the
single pole low pass filter used to smooth the instantaneous
flowmeter output.
[0058] The term in brackets [ ] in equation (2) is the correction
count and is the preferred embodiment of this disclosure. The
correction pulse count and/or count fraction is added or subtracted
(depending on the sign of the Q.sub.b-Q.sub.a term) to the pulses
measured during the proving run, before applying the double
chronometric correction of equation (3):
N.sub.1=N.sub.mc(T.sub.2/T.sub.1) 3)
Here N.sub.1, T.sub.2 and T.sub.1 have been defined following
equation (1).
[0059] The theoretical basis underlying the invention can be
understood by examining the response of an idealized measurement
system whose output passes through a single pole low pass filter.
FIG. 2 shows the response of a meter having a 0.2 second filtering
time constant to an incremental step .DELTA. in the actual flow,
occurring just after the initiation of the prove. This transient,
if negative, is roughly representative of the transient introduced
by a prover 14 having more hydraulic resistance than the normal
through-flow path.
[0060] The meter response to the step, .DELTA..sub.m is described
by an exponential function, specifically:
.DELTA..sub.m=.DELTA.(1-e.sup.-t/.tau.) 4)
[0061] The proving error is the integral of the difference between
the step and the exponential--the area A, bounded by the step and
the rising output function. It may be shown that A is given by:
A=-.DELTA..tau.e.sup.-t/.tau. evaluated, for the 1 second prove, at
t=1 and at t=0 5)
Substituting
[0062]
A=-.DELTA..tau.(e.sup.1/.tau.-1)=.DELTA..tau.(1-e.sup.1/.tau.)
6)
[0063] But what is measured is not .DELTA. but .DELTA..sub.m. At
any time t, .DELTA. can be expressed in terms of .DELTA..sub.m:
.DELTA.=.DELTA..sub.m/(1-e.sup.t/.tau.) 7)
Thus A at t=1 is given by
[0064] A=.DELTA..sub.m.tau. 8)
[0065] This is exactly the correction term proposed herein. Note
that on a theoretical basis, the correction works for a step
disturbance regardless of the filter time constant relative to the
prove time.
[0066] It should be noted that, because the ultrasonic flowmeter is
a sample data system typically producing a digital output, the
smoothing filter is likely to be digital. As a consequence, there
may, in practice, be a small residual error in the correction whose
magnitude is dependent on the flowmeter sample rate as against the
filter time constant--the faster the sample rate relative to the
time constant, the smaller the error. The small residual error
shown for the corrected response in FIG. 2 is due to the sample
rate used in the calculation; higher sample rates or longer time
constants will lead to smaller residual errors. For example, the
proposed correction for the same step, sampled at the same rate,
will, with a 5 second filter, produce an error of only 0.02% of the
volume change. Computer simulations demonstrate that the correction
works regardless of the timing of the step relative to the
initiation of the prove.
[0067] The correction also works for ramp changes. FIG. 3 depicts
the idealized response of a meter with a 1 second time constant to
a 1 second prove to a ramp change of 1 per unit volume per second.
For this example the ramp has been initiated 7 seconds before the
prove begins. As theory predicts for times long after the
initiation of the transient, the measured response is essentially
parallel to the actual response, lagging behind it by a time given
by the time constant .tau., which in the case of FIG. 3 is 1
second.
[0068] Again the theoretical basis for the correction may readily
be demonstrated. The discrepancy in the uncorrected output of the
meter with a 1 second time constant is the area of the
parallelogram bounded by the input flow, the measured flow, and the
prove start and finish times (0 and 1 seconds in FIG. 3). This area
represents the difference in the actual and measured volumes of the
prove, V.sub.act and V.sub.meas. Expressing these volumes
algebraically:
V.sub.act=[(Q.sub.a+Q.sub.b)/2](t.sub.b-t.sub.a) 9)
V.sub.meas=[(Q.sub.am+Q.sub.bm)/2](t.sub.b-t.sub.a) 10)
[0069] Here Q.sub.a and Q.sub.b are the actual flow rates at times
t.sub.a, the beginning of the prove, and t.sub.b, the end of the
prove. Q.sub.am and Q.sub.bm are the measured flow rates at these
times.
[0070] The actual flow rate, Q, can be expressed algebraically as
follows:
Q=Q.sub.0+(dQ/dt)t 11)
[0071] Here Q.sub.0 is the value of the flow rate at the initiation
of the transient, which occurs at t=0. The term (dQ/dt) is of
course the slope of the ramp.
[0072] At the time of the prove, the measured flow, Q.sub.meas, can
be expressed as follows:
Q.sub.meas=Q.sub.0+(dQ/dt)(t-.tau.) 12)
[0073] From FIG. 3, it will be noted that the slopes of the ramps
in actual and measured flow rates can be expressed as:
dQ/dt=(Q.sub.b-Q.sub.a)/(t.sub.b-t.sub.a)=(Q.sub.bm-Q.sub.am)/(t.sub.b-t-
.sub.a) 13)
[0074] The error in the volumetric measurement can be determined
from the difference in actual and measured volumes (equations 9 and
10), using equations 11, 12, and 13.
V act - V meas = 1 2 [ ( Q a - Q am ) + ( ( Q b - Q bm ) ] ( t b -
t a ) = 1 2 [ 2 .tau. ( Q t ) ] ( t b - t a ) = .tau. ( Q bm - Q am
) ##EQU00001##
[0075] Thus, for the idealized ramp of FIG. 3, the error in the
prove is exactly corrected by the proposed algorithm.
[0076] Simulations demonstrate that the correction works regardless
of when the ramp is initiated relative to the initiation of the
proving run. As with the step response, there may be a small
residual error depending on the filter time constant and the sample
rate. Again, if the sample rate is high relative to the proving
time, the error is small and will diminish as the time constant
increases. For example, computer simulations show that if the ramp
is initiated simultaneously with the start of the prove, a filter
time constant in the 0.1 to 0.2 second range, with a meter having a
moderately fast sample rate, may produce an error, after
correction, of 0.18% of the net volume change. If, for the same
sample rate, the filter time constant is increased to 5 seconds,
the net error after correction is 0.02% of the volume change.
[0077] Simulations of other disturbances, such as a pulse in flow
rate during a proving run, show that the proposed algorithm
effectively eliminates proving biases from these disturbances as
well.
[0078] The analyses and descriptions of the present invention have
used, as an example, a meter having a first order linear filter to
remove unwanted fluctuations--noise--from the raw flow rate data.
However, the same approach can be applied to correct the proving
data for other filtering systems such as a sliding average filter.
For alternative data filters the value of .tau. would be selected
as appropriate to the specific filtering methodology used.
Implementation
[0079] The software and hardware to carry out the meter factor
correction described herein may conveniently be located in the flow
computer 20, which, as noted above, controls the proving process,
and receives the prover 14 detector switch actuation signals and
the volumetric pulse train from the meter to be calibrated.
Uncorrected measured flow rates are also typically supplied to the
flow computer 20, so that, if the meter manufacturer supplies the
meter's frequency/flow rate constant
(.differential.f/.differential.Q) and filter time constant, .tau.,
to the flow computer 20 manufacturer, the correction can readily be
carried out in the flow computer 20. Alternatively, the flow
computer 20 can calculate the correction from the instantaneous
pulse train frequency, continuously supplied to the flow computer
20 from the meter to be calibrated, along with .tau.. A diagram of
this embodiment is shown in FIG. 4.
[0080] As a second arrangement, if the prover 14 switch actuation
signals and prover 14 volume are provided to the meter
manufacturer, the meter factor computation, including the
correction, can be carried out in the meter itself. A diagram of
this embodiment is shown in FIG. 5. For this arrangement,
correction of meter factor data to standard temperature and
pressure conditions (S.T.P.) remains in the flow computer 20 as
does all custody transfer functions.
[0081] A third arrangement employs a separate "black box", which is
supplied with the pulse train from the meter to be calibrated along
with the instantaneous pulse train frequency, .tau., and the prover
14 switch actuation signals. The black box performs the meter
factor calculation which is fed to the flow computer 20. A diagram
of this embodiment is shown in FIG. 6. Again correction to S.T.P.
and custody transfer functions would remain with the flow computer
20.
[0082] A fourth arrangement consists of an ultrasonic or other
flowmeter configured to output each sample flow rate it measures,
unfiltered, in a continuous data stream to a "black box". The black
box: [0083] (a) Filters the flow rate data using a single pole low
pass filter or similar, the time constant of said filter being user
selectable to enhance the ability to prove the meter in a limited
number of proving runs, [0084] (b) Converts the filtered flow rate
data to a pulse train whose instantaneous frequency is directly
proportional to the filtered flow rate, [0085] (c) Accept input
signals from the actuations of the upstream and downstream prover
detection switches 16, 18, [0086] (d) Counts the number of pulses
between the actuation of the upstream and downstream prover
detection switches 16, 18, [0087] (e) Records the filtered flow
rates at the times of actuation of the upstream and downstream
prover detection switches 16, 18, [0088] (f) Computes the
correction to the pulse count from the [difference in pulse
frequencies at the times when upstream and downstream prover
detection switches 16, 18 actuate] times [the time constant of the
instantaneous flow sample filter], [0089] (g) Computes a corrected
pulse count as the algebraic sum of the number of pulses counted in
(d) and the correction calculated in (f), [0090] (h) Computes a
meter factor from the quotient of the (input) prover 14 volume and
the corrected pulse count as calculated in (g), [0091] (i)
Transmits same meter factor to a flow computer 20 for use in
custody transfer operations.
[0092] Although the invention has been described in detail in the
foregoing embodiments for the purpose of illustration, it is to be
understood that such detail is solely for that purpose and that
variations can be made therein by those skilled in the art without
departing from the spirit and scope of the invention except as it
may be described by the following claims.
* * * * *