Professional Documents
Culture Documents
An Automated Test and Measurement System For Calibration of Industrial Flowmeters
An Automated Test and Measurement System For Calibration of Industrial Flowmeters
ACKNOWLEDGMENT
In general the market shows a request that the measured I´d like to thank the contribution of my colleagues Herbert Kitzing and Marco Schäck of the HBM headquarters in
signals can be processed more accurately. In this paper it Darmstadt; Germany for their valuable contributions.
has been shown, that there is substantial progress regard-
ing compactly build precision amplifiers for strain gauge LIST OF REFERENCES
based on transducers such as force, torque and pressure
reference transducers used in metrology. [1] Hoffmann, K.; “An Introduction to Measure- [12] Schäfer, A. “Challenges in dynamic torque and
ments using Strain Gauges”, Hottinger Baldwin force measurement with special regard to indus-
In fact today also amplifiers based on resistive dividers can Messtechnik GmbH, Darmstadt, Germany, 1989, trial demands”, BIPM Workshop on Challenges
perform most tasks with sufficient accuracy. For many pp. 145 in Dynamic Measurements
decades this was not the case. Due to intensive research,
[2] Kreuzer, M. “High-precision measuring tech- [13] Schäfer, A. “Development results for static and
HBM can now offer such a dual-channel module at the
nique for strain gauge transducers” Internal pub- dynamic torque measurement“, Proceedings of
Fig. 8. Long term stability of the established calibration method same time as compact as this is necessary for a realization
lication of Hottinger Baldwin Mestechnik, torque symposium of “Zentrum für Konstruktion-
of a measuring chain consisting out of BN100 and DMP39 in the block structure the QuantumX family.
GmbH, Darmstadt, Germany, 1999 swerkstoffe“, MPA (State Material Testing Insti-
tute), Darmstadt, Germany, 2014
It is quite certain, that this a previously unknown compact- [3] Kleckers, T., Graef, M. “High capacity reference
For electrical calibration of these measuring amplifiers
ness will open up new fields of integration into applica- transducer for tensile forces”¸ IMEKO 22nd [14] Schäck, M. “Underestimated impact of measur-
bridge standards are used, which have to be traceable to
tions. QuantumX now offers the ability to process and dis- TC3, 12th TC5 and 3rd TC22 International Con- ing cables on high-precision carrier frequency
national standards.
play multiple signals of very different accuracy-bandwidth ferences 3 to 5 February, 2014, Cape Town, Re- amplifier results and compensation methods
combinations at the same time. public of South Africa therefor” XXI IMEKO World Congress, Sept.
As a matter of fact, the standards of HBM and of
2015, Prague, Czech Republic.
“Physikalisch-Technische Bundesanstalt” (PTB), the Ger- [4] Kleckers, T, “Precise radial symmetric shear
Thus the decisive progress realized with the QuantumX
man National metrology institute are not same. The reason beam force transfer transducer for compression [15] Schäfer, A.; Kitzing, H.; “DMP41- a new chapter
MX238B is, that now the cooperation with all the other
for this lies in history. HBM needed its own >>Normal<< force”; XXI IMEKO World Congress, Sept. of ultra-precision instrument for strain gauge
measurement modules of the product family QuantumX
(mV/ V) to produce first precision amplifiers such as 2015, Prague, Czech Republic. transducers”, XX IMEKO World Congress, Sep-
with PTPv2 identical timing, allows data processing abso-
DK37 and DK 38. At that time there was no existing trace- tember 2012, Busan, Republic of Korea
lutely simultaneously. Thus the isochronous bus system [5] Schäfer, A.; “Force, strain and pressure trans-
ability. Then, at the time in 1981, the PTB has set up a
(Firewire) allows synchronous processing with many other ducers based on foil type strain gauges”, Pro- [16] Schäfer, A.; “The ultra-precision instrument
normal mV/ V, these normals showed an offset.
physical quantities. ceedings of Eurosensors; XXII conference, Dres- DMP41-first experience & appropriate filter set-
As a matter of fact the standard of PTB is technically con- den, Germany, October 2008 tings”, IMEKO 22nd TC3, 12th TC5 and 3rd
The precision measurement module MX238B thus com- TC22 International Conferences, February, 2014,
structed differently than the normal in HBM; however [6] Schäfer, A. “New possibilities with a high pres-
bines the advantages of combination with all QuantumX Cape Town, Republic of South Africa
these values always can be “translated” into each other. In sure measuring chain consisting of reference
members with the possibility of taking measurements with
relation to the values acquired by the PTB the end values pressure transducer P3 TopClass BlueLine and [17] Schäck, M.; “High-precision measurement of
an accuracy class 0.0025 (25 ppm). This is achieved by the
of all HBM devices with a carrier frequency of 225 Hz, for DMP 41 high-precision instrument”, PTB Report strain gauge transducers at the physical limit
use of the carrier frequency principle and an applied carrier
reasons of continuity, are given 10 ppm lower than the TH6, Brunswick, Germany, April 2016 without any calibration interruptions” IMEKO
frequency of 225 Hz.
nominal values. There are many articles on the details [18], 22nd TC3, 12th TC5 and 3rd TC22 International
[19], [20], [21]. [7] Haller, M. Viel, W.; Schäfer, A.; “Enhancement Conferences, February, 2014, Cape Town, Re-
For the first time here HBM offers its patented background of the measurement characteristics of pressure
calibration of this precision class. This ensures that the public of South Africa
MX238B demands the connection of the transducers in a transducers up to 15000 bar through monolithic
measurement of the two integrated amplifier can be con- measuring design and foil type strain gauges”, [18] Vollmert, R. ; Ramm, G. “Realization, mainte-
six-wire-connection-circuitry (with two additional sensing
tinued without interruption during the measurement, so Proceedings of XIX IMEKO World Congress nance and dissemination of the measurand >>AC
lines), as for this class of accuracy the resistance of cable
even during the internal calibration. Lisbon, Portugal, 2009 voltage ratio in mV/V<< for strain gauge meas-
lengths in the range of only cm already counts. Subse-
quently the device offers also detailed error detection with urements“ , Proceedings of the 17th International
Furthermore, the transducer can be recognized with the so- [8] Schäfer, A.; et al.; “A new type of transducer for Conference of IMEKO TC3, Istanbul, Turkey,
wire break detection for sensor lines, feeder lines and test accurate and dynamic pressure measurement up
called Transducer Electronic Data Sheet (“TEDS”) and the 17-21, September, 2001
leads. to 15000 bar using foil type strain gauges”, XVII
individual properties of the transducer can be considered
and easily red out. IMEKO World Congress 2003, Dubrovnik, Cro- [19] Ramm, G. “ Investigation of bridge standards
Switchable shunt resistors of 100 Ohms allow an artificial atia, 2003 for use in strain-gauge measurements” Reports
bridge detuning. In a 350 Ohm Wheatstone bridge, so the in Applied Measurement RAM, Vol. 6 (1990),
The measurement results have been showing the very [9] Eichstädt. S., Esward, T. J., Schäfer, A. “On the
standard for the strain gauge pickup principle, there is an No. 1, pp. 26-30
good time as well as temperature stability of resistive necessity of dynamic calibration for improved
upset of 0.886 mV / V at 2 mV / V measuring range, i.e. a
dividers. traceability of mechanical quantities” XXI [20] Ramm, G.; Peters, M. “Investigation of the Cali-
fairly significant detuning of approximately 40%.
IMEKO World Congress, Sept. 2015, Prague, bration Signal of Force Measuring Devices”.
Nevertheless one has to say that the requirements for a Czech Republic. IMEKO Force and Mass 1986, Amsterdam,
Regarding a more reliable measurement it can be said that
top performance, so amplifiers of highest precision for Conference Digest, ISBN 90-247-3351-0, pp.
the built-in galvanic isolation is a mean to avoid ground [10] Schäfer, A.; "Dynamic measurements as an
primary calibration purposes, built to operate on the 267-273
loops and therefore the amplifier appears much more ro- emerging field in dynamic metrology"; PTB-Mit-
physical limit (such as our DMP-series), further relies on
bust. Also meeting the high EMC requirements according teilungen 125(2015) Volume 2 [21] Ramm, G. „Untersuchung des Fehlverhaltens
inductive dividers (voltage ratio derived from the number
to EN 61326 for field strengths of 10V/m are a further ad- eines Brückennormals vom Typ BN 100“
of windings of coils), as they remain still more accurate [11] IEEE 1451.4 Standard for Smart Transducers
vantage of the device. PTB Report PTB-E-30, ISBN 3-88314-655-2,
than any resistive dividers can be. Secretary, issued by IEEE 1451.4 Standard April 1987
Working Group, 1451.2WG, 2004
87 88
J.M. Dias Pereira1, Mário Alves1, Vítor Viegas1, Octavian Postolache2, Pedro Silva Girão3 IV. EXPERIMENTAL RESULTS
1
Several tests were performed in order to test the
ESTSetúbal/IPS, IT, Rua do Vale de Chaves, Estefanilha, 2910-761 Setúbal, Portugal performance of the proposed measurement and
2
ISCTE-IUL, IT, Avenida das Forças Armadas, 1649-026 Lisboa, Portugal Fig. 1. Flowmeter work bench piping diagram (H.P.- hydraulic calibration system. In the final version of the paper
3 additional results will be included for the electromagnetic
DEEC/IST/UL, IT, Avenida Rovisco Pais, 1, 1049-001 Lisboa, Portugal pump; O.F.- orifice flowmeter; C.F.- Coriolis flowmeter; E.M.-
electromagnetic flowmeter; V- ball valves). and Coriolis flowmeters.
Abstract –This paper presents a low-cost test and different type of flowmeters, from different
measurement system that can be used to test and manufacturers, is required. B. Software part A. Orifice
calibrate industrial liquid flowmeters. Different The LabVIEW graphical programming language [11] was In order to characterize the orifice flowmeter, the
flowmeter types, with different working principles, III. DESCRIPTION OF THE SYSTEM used for data processing and HMI purposes. differential pressure, between its high and low pressure
from different manufacturers, can be easily installed Measurement data from the three flowmeters is filtered taps, for different flow rates, was measured. The
and calibrated in a flexible way. The implemented The proposed flowmeter test bench includes an orifice and converted to engineering units according to the differential pressure transmitter was calibrated for an
prototype includes a software component that is based flowmeter, an electromagnetic flowmeter and a measurement range of each flowmeter. Conventional 4-20 upper range value (URV) equal to the upper range limit
tailored for laboratory test and calibration of Coriolis flowmeter. Combining electro-valves open and mA current signals or HART data streams [1] can be used (URL) and to an URV equal to URL/2, being the turn
volumetric or mass flowmeters. The final part of the close status, several fluid network configurations can be to exchange measuring and configuration data with the down factor equal to 1:15 for both measuring ranges.
paper includes several simulation and experimental implemented. In the proposed prototype, a number of 11 different flowmeters. The inverter voltage that controls Figure 3 represents the experimental results that were
results. different associations, between flowmeters, can be flow rate is normalized between 0 and 100% that obtained when the flow rate varies between 2.5 and 30
implemented. correspond to the minimum and maximum values of the l/min with increments of 0.5 l/min.
Keywords –flow measurements, calibration, virtual flow rate delivered by the hydraulic pump, respectively. A 7000
instruments, linearization A. Hardware part software implemented PID controller was used to
optimize the response of the flow rate variations
6000
Flow measurement is a topic of major importance in flowmeter [9]. The system also includes an hydraulic
every industrial processes, particularly when the pump and an inverter that is used to control the flow rate according to the dynamic characteristics of the 3000
transportation of fluids and the evaluation of energy measurement system. Graphical representations of the
intensity. In order to measure the fluid volume inside the 2000
dissipation, associated with eddies in turbulent flows, is measurement results are visible in real time and statistical
tank, a low cost gauge pressure sensor is used [10]. From
results of measurement errors, for each flowmeter, can be
1000
required. Only regular calibrations can assure that the pressure measurement it is possible to obtain the fluid
flowmeters' measurements are accurate. However, height inside the tank, whose value is given by: accessed after data processing of the measurement 0
0 5 10 15 20 25 30
calibration of flowmeters cannot avoid errors that result results. Calibration results can be obtained using a Flow (l/min)
from process conditions, namely fluid characteristics reference flowmeter or using the pressure sensor data (1). Fig. 3. Variation of the pressure loss in the flowmeter orifice as
(1)
variations, and fluid velocity profile variations inside the Figure 2 depicts the LabVIEW front panel of the a function of flow rate.
pipes. Thus, in every industrial plant, it is very important flowmeter test and calibration system.
Using the MATLAB least mean square (LMS)
to assure a regular calibration of flowmeters and to where p represents the measured pressure, ρ represents nonlinear curve-fitting function, the following orifice
develop laboratory solutions for test and calibration of the fluid's specific mass, approximately 1000 kg/m3 for parameters were obtained: CF=0.73 and A0=0.3 square
those devices. The proposed prototype fulfills this aim water, and g represents the gravity acceleration value.
and provides a flexible platform to test and calibrate inches, being the evaluated specific mass of the fluid (ρ)
Being the geometry of the tank well known, and equal to 1032 kg/m3. The maximum deviation between
different types of flowmeters that can used different performing the level measurements within its cylindrical
normalized communication signals, namely, current loop experimental and expected theoretical values of the flow
portion, it is possible to obtain the average value of the rate is lower than 2 % and the estimated orifice area is
signals, voltage signals, frequency or PWM modulated volumetric flow, during a given interval of time ( t),
signals, HART [1-2] or fieldbus digital signals [3]. almost 5 % lower than its theoretical value. The deviation
using the following relationship: between experimental results and theoretical expectations
II. RELATED RESULTS IN THE LITERATURE are acceptable considering the errors of the measurement
(2) instruments, the variations of the fluid Reynolds number
As a matter of fact, there are some commercial and also the errors caused by the variations of the
solutions to test and calibrate flowmeters [4-6] but those position of the orifice vena contracta relatively to orifice
solutions are typically very expensive and, above all, they where D represents the tank diameter, p represents the pressure taps that are always associated with flow rate
are turnkey solutions with low flexibility and without tank pressure variation that occurs during t, and the instabilities.
Fig. 2. LabVIEW front panel of the flowmeter test and
access to measurement data for data processing purposes, other variables have the same meaning that was
calibration system.
which is particularly important when the calibration of previously defined.
89 90
B. Calibration results Number of Occurrences
60
REFERENCES
The calibration procedure is based on the average value Calibration can also be performed using the mean value=25.2 l/min
>@ HART Communication Foundation, “Application Guide –
of volumetric flow measurements, during a time interval comparative method taking one of the three flowmeters, 50 standard deviation=0.15 l/min
hysteresis plus repeatability errors lower than 0.15 % linearization, a typical linearization requirement in flow Technovate Inc., U.S.A., 1976.
relatively to the span measuring range. For a series measurements is related with the square root dependence
0
applying different voltage profile variations to the input >@ ISO 5167-1:2003, "Measurement of fluid flow by means
6
linearization procedure avoids the usage of the well- of the inverter that controls the hydraulic pump. A large
4 known square-root extractor and associated signal ofpressure differential devices inserted in circular cross-
number of software routines can be developed to extend section conduits running full -- Part 1: General principles
2 conditioning circuits. the basic features of the flowmeters, without additional and requirements" (2003).
0
costs. These routines can include, for example, filtering >@ American Society of Mechanical Engineers Standard,
-2
D. Measurement uncertainty of measurement data, engineering units conversions, ASME MFC-3, “Measurement of Fluid Flows in Pipes
0 10 20 30 40 50 60
Frequency (Hz) Regarding measurement uncertainty, it is important to totalization of flow measurements, and compensation of Using Orifice, Nozzle, and Venturi” (1989).
Fig. 4. Flowmeter measurement results as a function of the underline that the main influence factors that affect its influence variables effects. >@ American Society of Mechanical Engineers Standard,
inverter frequency (square symbol- electromagnetic flowmeter; value include not only the uncertainties associated with ASME-MFC-3M-1983 (R-1988), “MeasurementUncertainty
Fluid Flow in Closed Conduits” (1988).
diamond symbol- Coriolis flowmeter; circular symbol- orifice the flowmeter, itself, but also the influence factors caused VI. CONCLUSIONS
based flowmeter). >@ International Standard, ISO/DIS 5168, “Measurement of
by fluid compression, errors associated with flow volume This paper presented a low cost automated and Fluid Flow - Evaluation of Uncertainties” (1989).
Absolute Error (l/min.)
0.3
evaluation, based on level measurements, variations of measurement system for the calibration and testing of >@ Gert Horn, Johan Huijsing,"Integrated Smart Sensors
fluid density and fluid rate stability during the calibration industrial flowmeters. Since the system includes an Design and Calibration", Kluwer Academic Publishers,
0.2 runs. In order to evaluate the uncertainty of flow rate hydraulic pump controlled by an inverter, different flow 1998.
measurements, the statistical distribution of flow was >@ P.P.L. Regtien, P.J. Trimp, "Dynamic Calibration of
0.1 profile variations can be tested in order to obtain the
obtained from a series of measurements (type A Sensors Using EEPROMs", Sensors and Actuators A, Vol.
dynamic characterization of the devices under test. The 21-23, pp. 615-618, 1990.
0 evaluation of measurement uncertainty) [19]. Using, for testing and calibration results, based on pressure
instance, the electromagnetic flowmeter, configured for >@ Joint Committee for Guides in Metrology, JCGM/WG,
-0.1 measurement or previously calibrated flowmeters, "International Vocabulary of Metrology, Basic and
the measuring range between 5 and 50 l/min, and presented low deviations between theoretical expectations General Concepts, BIPM" (2008).
-0.2 adjusting the inverter frequency to obtain a constant flow and experimental results, being these deviations
-0.3
of 25 l/min, figure 6 represents the histogram of a series acceptable considering flowmeters' measuring errors and
of 256 flow measurements. flow stability conditions.
-0.4
0 10 20 30 40 50 60
Frequency (Hz)
91 92
were applied for the RC and the L measurements. Two dependent upon the adaptor types used for the measure-
Influence of Adaptor on the Calibration of different torque drivers, N6LTDK for (5 ~ 60) cN m and
N20LTDK for (40 ~ 200) cN m were used to control the
ments. But the optimum range for the L and RC, where L
and RC values are constant within the measurement uncer-
Inductance Standards
torque between the adaptor and the inductor terminals. The tainties, is nearly the same and estimated to be 90 cN m ~
minimum incremental torque was 0.5 cN m and 1 cN m for 150 cN m.
the N6LTDK and N20LTDK, respectively. Accuracy of
Dewi Mohd Kassim1,2,3, Dan Bee Kim1, and Wan-Seop Kim1,2 the torque was ±3 % of the setting value over the full range.
All the measurements of RC and L were obtained at a
frequency of 1 kHz.
1
Korea Research Institute of Standards and Science, 267 Gajeong-ro, Yuseong-gu, Daejeon 34113,
Rep. of Korea
ws2kim@kriss.re.kr
2
University of Science and Technology, 217 Gajeong-ro, Yuseong-gu, Daejeon 34113, Rep. of
Korea
3
National Metrology Institute of Malaysia, Lot PT 4803 Bdr Baru Salak Tinggi, 43900 Sepang,
Selangor Malaysia
Abstract ě Influence of the adaptor on the calibration The standard inductors, GenRad 1482 type, are widely
of 100 H inductance standard has been studied using used in the national metrology laboratories as a low
two different homemade adaptors of BPO gold-plated Fig. 1. Homemade adaptors (a) BPO plug with gold-plated
frequency reference due to its long-term stability, low brass (BPO-Au) (b) Banana plug (1) with copper (2) adaptor
brass (BPO-Au) and banana-copper (BN-Cu) as a temperature coefficient, low internal series resistance, and
function of torque. The inductance (L) and the contact (BN-Cu).
low Q-factors. The GenRad 1482 type standard inductors
resistance (RC) showed an exponential decrease against
the increasing torque from 25 cN m to 150 cN m. The have three binding post terminals, two for the inductor Fig. 2 shows a measurement plane (shaded) of the
measured L and the calculated equivalence series leads and the third connected to the case [4]. For the adaptor for the contact resistance and inductance
resistance are dependent on the type of adaptors as well calibration of the inductance standards, a common method measurements, respectively. The potential side of the
as the RC. The analysis of RC and measurement is the substitution using a LCR meter. However, most LCR adaptor directly makes a contact with the binding post
uncertainty of L implies that the BPO-Au adaptor with Fig. 3. Measured (a) inductance L, (b) quality factor Q, and
meters have four terminals, so an adapter is needed terminal of the inductance standards.
low RC is suitable for the calibration of inductance. The (c) contact resistance RC for the two different adaptors of BPO-
between the LCR meter test leads and the inductor Au (solid symbols) and banana-copper (BN-Cu, open symbols)
measured L obtained using BPO-Au and subtracted by
the RC agreed well with the certificate value of the PTB terminals. as a function of torque . Dashed lines are fitting curves using
within the measurement uncertainty of 140 H/H. In In this study, we designed two different types of eq. (1) (s. text).
addition, the calculated inductance of the adaptors adaptors, BPO gold-plated brass (BPO-Au) and banana-
shows influence of the lumped components, stray copper (BN-Cu), for the calibration of the inductance In the case of the BPO-Au adaptor, the inductance
capacitance and series resistance, on the measurements standards and studied the influence of the adaptor on the changed by 9.3 nH from 100.1327 H to 100.1234 H
of standard inductors. calibration of the 100 H standard inductor in conjunction while the contact resistance was decreased by 0.163 m
Keywords – Adaptor, contact resistance, inductance with various torque values. The measurements results from 3.364 m to 3.301 m over the whole torque range
standard, torque, quality factor show the contact resistance and the measured inductance of (25 ~ 150) cN m. The measured RC and L can be well
are strongly dependent on the type of adaptor and torque fitted using an exponential equation below:
applied for tightening of the binding post of the inductor.
τ
I. INTRODUCTION Hence, we report the contribution of the adaptors to the y = y 0 + A ⋅ exp( − ) , (1)
contact resistance, equivalent series resistance and t
An improper contact between the test leads of a inductance measurements.
where y and y0 are L(RC) and L0(RC0) for = 0, respectively
measuring instrument and the terminals of a standard Fig. 2. Measurement plane (shaded) of the adaptor for the
contact resistance and inductance measurements, respectively. and A is a fitting parameter for the amplitude of the
inductor can considerably affect precision measurements
II. EXPERIMENTS measurements.
of inductance standards [1][2]. In particular, if the
inductance value is lower than 100 H, the effect of the III. RESULTS AND DISCUSSIONS It is worth to mention that the decrement rates of tL = 28
contact resistance can be very significant due to the fact The contact resistance (RC) was measured using HIOKI cN m for the inductance and tSC = 31 cN m for the contact
that its impedance is lower than 1 (0.63 at the BT3562 HiTester using a 4-point probe method. The Fig. 3 shows the measurement results of (a) L, (b) quality resistance are nearly the same within the fitting uncertainty
frequency of 1 kHz) and its contact resistance RC is in the inductance (L) of the 100 H standard inductor (GenRad factor Q, and (c) RC for the two adaptors of BPO-Au and of ±3 cN m. This leads to a linear dependence (dashed line)
order of m [3]. Therefore, it is clear that only a little 1482 type) was measured using a LCR meter (QuadTech BN-Cu as a function of the torque ( ) from 25 cN m to 150 between L and RC as shown in Fig. 4(a) and (c). The L
deviation in the contact resistance can lead to a 7600). Two homemade adaptors, BPO plug with gold- cN m. It is obvious that not only the inductance values linearly changes by RC with a slope of 47 nH/m .
considerable effect on the precision measurements of the plated brass adapter (BPO-Au) and banana plug with (Fig. 3(a)) but also the contact resistances (Fig. 3(b)) of the
inductance standards. adapter made from copper (BN-Cu) as shown in Fig. 1 standard inductor are significantly different and strongly
93 94