standardisation of pressure measurement using pressure ...metrologyindia.org/26b/sanjay...

Standardisation of Pressure Measurement Using Pressure Balance as Transfer Standard

133

Standardisation of Pressure Measurement Using PressureBalance as Transfer Standard

SANJAY YADAV*, V.K. GUPTA and A.K. BANDYOPADHYAY

National Physical Laboratory (NPLI)Council of Scientific and Industrial Research (CSIR), New Delhi - 110 012, India

*e-mail:[email protected]

[Received: 24.12.2010 ; Revised: 09.02.2011 ; Accepted: 11.02.2011]

AbstractThis paper describes the results of the interlaboratory comparison for pressure measurements of 9laboratories that are accredited by the National Accreditation Board for Testing and Calibration ofLaboratories (NABL). The artifact used for the comparison was a pressure balance covering the pressurerange (7 to 70) MPa. The primary objective of this comparison was to assess the laboratory's technicalcompetence to perform measurements and also to assess the compatibility of the results submitted by thelaboratories. The comparison began during March 2008 and ended during April 2010. For assigningthe reference values, the pilot laboratory (NPLI) carried out 3 calibrations of the transfer standard; thefirst one at the beginning, the second at the middle and the last one at the end of the programme. Thecomparison was carried out at 10 pressure points i.e. (7, 10, 15, 20, 25, 30, 40, 50, 60 and 70) MPathroughout the entire pressure range of (7 to 70) MPa. The measurements were carried out by eachlaboratory with their own resources (personnel, calibration systems, environmental conditions in theirinstallations). The deviations for each laboratory were compared against the reference values and thecompatibility of the results was calculated using the normalized error value method. Out of the total 87measurement results reported, 68 (78.2%) results are found in good agreement with the results of thereference laboratory. The normalized error (En) values of 5 laboratories out of the total 9 were foundwell within ± 1 over the entire pressure range. However, 2 other laboratories had shown good agreementwith the reference values except one pressure point each. The En values of one of the participatinglaboratory were found beyond acceptable limits at all measurements points. Another laboratory hadacceptable results only at 3 pressure points. The laboratories with unacceptable results have beenadvised to review their pressure measurement process. The deviations between laboratory values and ofthe reference values were found well within the uncertainty band of the reference values for 37%measurement results. The relative deviations for 82 measurement results were found well within 0.05%.

© Metrology Society of India, All rights reserved 2011.

1. Introduction

The quality assurance section of ISO/IEC 17025stipulates the requirement for ensuring; i) that a single

analyst within a laboratory is able to consistentlyreproduce the same result on the same sample, ii) thatthe result produced by this analyst should reflect theresult that would have come from any other analystin the laboratory and iii) that any results from the

MAPAN - Journal of Metrology Society of India, Vol. 26, No.2, 2011; pp. 133-151REPORT

Sanjay Yadav, V.K. Gupta and A.K. Bandyopadhyay

134

laboratory as a whole should reflect the results thatare agreed upon by many other laboratories. It is dueto this reason that the internal and external qualitycontrol (QC) and quality assurance (QA) clauses existwithin ISO/IEC 17025.

Although, the precise way of going about provingthe consistency and the reliability of the results is notprescribed in ISO/IEC 17025, the accreditation bodieshave built some prescriptive clauses into theirrequirements to try to facilitate meeting therequirements of ISO/IEC 17025 in an effective manner.An externally provided PT program is a useful tool inmeeting the requirements of ISO/IEC 17025. However,participating in an external PT program will notnecessarily mean that all quality assurance aspectshave been met [1-3].

Proficiency testing is used by the accreditationbody as part of the assessment processes, to evaluatethe technical competence and ability of thelaboratories to carry out tasks for which itsaccreditation has been applied for / granted. This testis a complement of the laboratory assessment carriedout by technical experts in situ. It is also mandated byaccreditation bodies that laboratories participate inthe PT exercises for all types of analyses undertakenin that laboratory, when suitable exercises exist.

To meet the requirements of MRA [1], ISO/IEC17025 [2] and APLAC MR001 [3], the NABL hasconducted several PT experiments in pressuremetrology in the pressure range (7 to 70) MPa amongstthe NABL accredited Indian pressure calibrationlaboratories in conformity with ISO/IEC Guide 43 [4]through the NMI of India i.e. NPLI, which also actedas a reference laboratory.

In a series of 7 PTs organized, the 1st PT,designated as NABL-Pressure-PT001 was organizedfor 7 laboratories, having measurement capabilitiesbetter than 0.05% of full scale pressure using deadweight tester as an artifact [5]. The (2nd and 5th) PTsi.e. NABL-Pressure-PT002 and NABL-Pressure-PT005were conducted for another 7 and 21 laboratories,having measurement capabilities coarser than 0.05 %and better than 0.25% of full scale pressure usingdigital pressure calibrator [6-7]. The (3rd and 6th )PTs i.e. NABL-Pressure-PT003 and NABL-Pressure-PT006, included (11 and 17) laboratories, respectivelyhaving measurement capabilities 0.25% or coarser

than 0.25% of full scale pressure using pressure dialgauge as an artifact [8-9]. Similarly, another PTexperiment i.e. NABL-Pressure-PT007 was carried outfor 14 laboratories having measurement capabilities0.25% or coarser than 0.25% of full scale pressureusing pressure dial gauge as an artifact in the pressurerange (6 to 60) MPa [10-11].

The final PT experiment, designated as NABL-Pressure-PT004, is recently completed during April2010. This PT programme is designed and organizedin the hydraulic pressure region covering pressurerange (7 to 70) MPa (70 to 700) bar using the pressurebalance as transfer standard. Initially, 10 NABLaccredited pressure calibration laboratoriesparticipated and finally 9 laboratories submitted theirresults.

2. Objectives

The main objectives and benefits of PT toparticipating laboratories are; i) the participatinglaboratory fulfills the requirements of ISO/IEC 17025in the area of proficiency testing, from both inter-laboratory and intra-laboratory standpoints, ii) Thelaboratory ensures the competence and capabilitiesof the staff involved in the measurements, iii) Thelaboratory collects the information that can assist infuture planning for equipment upgradation and stafftraining and iv) identification of any difficulties withmethodology, instrumentation, results and trainingneeds. It also supplements laboratory's own qualitycontrol procedures by providing additional externalaudit and also provides objective evidences that alaboratory is competent enough and can achieve thelevel of uncertainty for which accreditation is granted.This exercise gives an opportunity to accreditedcalibration laboratory to demonstrate its technicalcompetence of routine calibration services renderedto customers and to have the measurement traceabilityto the NMI.

3. Methodology

The PT programme is designed as per guidelinesstipulated in ISO/IEC 17025 [2], ISO/IEC Guide 43[4] and NABL-162 [12]. A high precision pressurebalance, Model No.- YW 1305, YANTRIKA, Sl. No.-REB 095, make Ravika Engineers, New Delhi wasused as transfer standard for this comparison. Thedetailed ‘Technical Protocol’ (TP) of the programme


135

was prepared and circulated to all the participantscontaining the information about the artifact,calibration procedure, environmental conditions tobe maintained, calculation of the results, procedurefor reporting of the results etc. The entire measurementpressure range of (7 to 70) MPa was divided into 10arbitrarily chosen measurement points of (7, 10, 15,20, 25, 30, 40, 50, 60 and 70) MPa to accommodate atleast 5 measurement points for each participant.

The programme had run smoothly and almostall the participants performed their measurementswell in time. The whole circulation programme wascompleted in two loops. After completion of the firstloop, the artifact reached NPLI, New Delhi at the endof December, 2009. During its inspection andcharacterization, it was observed that the system hadproblems of leakage and some bad handling. Theartifact was then sent for repair and overhauling toits manufacturer. This process of repair took almost10 months time to start the second loop. Since, therewas no problem with the piston and cylinderassembly of the transfer standard, the data taken

before and after the repair was well within thereported measurement uncertainty. It was recalibratedat NPLI, New Delhi and then dispatched for thecirculation of the second loop. There was no technicalproblem, fault, snag or difficulty reported by any ofthe participants. Schematic diagram of the movementof the artifact is depicted in Fig. 1.

3.1 Characterisation of the Artifact and Assigning Reference Values

The characterization of the artifact was performedthrice, first at the start of the programme duringDecember, 2007, second during (March and April)2009 and finally at the end of the programme duringMarch 2010. The characterization starts with thecalibration of individual mass values of the each deadweight and piston assembly of the artifact, collectionof pressure data, determination of effective area as afunction of applied pressure (Ap), zero pressureeffective area (A0) and distortion coefficient (λ),stability of Ap and A0 during the whole period of PTprogramme and finally assigning the reference valuewith measurement uncertainty.

Fig. 1. Circulation and movement of the artifact during comparison

March 24, 2008 NPL, New DelhiFebruary 23, 2010Dec. 29, 2009

Measure Technique (MT),Chennai

24-03-2008 to 16-04-2009

MSME Testing Centre (RTC),New Delhi

29-12-2009 to 23-02-2010

Dott. Ing. Scandura Calibration& (SCANDURA) Chennnai16-04-2008 to 25-04-2008

Electronics Regional TestLaborattory (ERTL), Delhi25-11-2008 to 10-02-2009

Nagman Instrument andElectronics (P) Ltd., (NIEPL)

25-04-2008 to 10-05-2008

Bhart Heavy Electtricals Ltd.(BHEL), NOIDA

17-10-2008 to 25-11-2008

Regiech Calibration Pvt. Ltd.(RCPL), Chennai

10-05-2008 to 04-06-2008

National Cement and BuildingMaterials (NCCBM), Ballabhgarh

12-08-2008 to 16-10-2010

Sushma Industries and CalibrationCenter (SICC), Bangalore04-06-2008 to 07-07-2008

WIKA, Pune09-07-2008 to 04-08-2008


136

Calibration of the individual mass values of theeach dead weight and piston assembly of the artifactwas performed at Mass Metrology Section of NPLI,New Delhi against the national standards of massand balances. After mass calibration, pressurecharacterization of the artifact was performed usingthe well-established and internationally acceptedmethod of 'cross-floating' of pressure balances. Theartifact was connected with national hydraulicsecondary pressure standard, designated asNPL100MPN for cross floating measurements asdiscussed elsewhere [13-16]. In a cross-floatingposition, the two gauges were considered to be inbalance when the sink rate of each of the piston wasnormal for that particular pressure. At this position,there was no pressure drop in the connecting line andconsequently no movement of the fluid. Thisprocedure was repeated for all the 10 pressure pointsi.e. (7, 10, 15, 20, 25, 30, 40, 50, 60 and 70) MPa andobservations were repeated six times, three times eachin increasing and decreasing orders of pressures, foreach pressure point. The traceability of theNPL100MPN is established by cross-floating itagainst national primary pressure standard [17-20].The NPL100MPN has also participated in several keycomparison exercises, APMP.M.P. K7 [21], CCM P.K7 [22] and APMP-SIM.M.P. K7 [23]. The values ofthe pressure generated, the effective area, repeatabilityand the expanded uncertainty were computed usingthe computer softwares developed for this purpose[24-25]. The least square curves were fitted to knowthe most probable values of the zero pressure area(A0t) and the distortion coefficients (λt) alongwith theirstandard deviations [26].

The pressure measured by laboratory standard(LS) was calculated using the following equation;

i i L air miLS 2

0 n n c p r

. (1 / )(1 )[1 ( )( )]

Σ − +=

′+ + λ + + −m g C

PA p p T T

ρ ρ γλ α α [1]

The term (l-ρair/ρmi) is the air buoyancy correctionfor weights, γC is the force exerted on the piston bysurface tension of the transmitting fluid, [1 + (αc+ αp)(T -Tr)] is the thermal expansion correction factor, theterm (1+λpn+ λ'pn

2) describes the change of the effectivearea with pressure which is the most importantcorrection term. The various terminology used in theequation are defined as follows;

mi mass of the ith weight combination (in kg) placedon the LS,

gL value of local acceleration of gravity (in m/s2) inthe measurement laboratory,

ρair density of the air (in kg/m3) at temperature,barometric pressure and humidity prevailing inthe laboratory,

ρmi density (in kg/m3) of the material of the weights,

γ surface tension (in N/m) of the pressuretransmitting fluid used,

C circumference (in m) of the piston where it emergesfrom the fluid,

A0 zero pressure effective area (in m2) of the LS,

αc &αp linear thermal expansion coefficients (in /0C)of the material of the cylinder and piston,respectively of the LS,

T measured temperature (in 0C) of the LS piston -cylinder assembly,

Tr temperature (in 0C) at which A0 (zero pressureeffective area) of LS is referred,

λ First order pressure distortion coefficient (in perPa) of the LS,

λ' Second order pressure distortion coefficient (inper Pa2) of the LS, and

Δp is the head correction (in Pa) in terms ofpressure.

The head correction term is Δp = [(ρf - ρair). gL. H],where H is the difference in height (in m) between thereference levels of the two dead weight testers and (ρf)is the density (in kg/m3) of the pressure transmittingfluid used in the measurements..

The temperature corrected forces F (in N) actingon the artifact, referred as Test herein thereafter, iscalculated using the expression;

it L air mi tTest

ct pt t rt

. (1- / )+=

[1+( + )( - )]m g C

FT T

Σ γρ ρα α [2]


137

where

mit mass (in kg) of the ith weight combinationplaced on the artifact,

Ct circumference (in m) of the piston of the Test,

αct &αpt linear thermal expansion coefficients (in 0C)of the material of the cylinder and piston,respectively of the Test,

Tt measured temperature (in 0C) of the piston-cylinder assembly of the Test,

Trt temperature at which A0t (zero pressureeffective area) of Test is to be calculated.

The effective area Ap (in m2) of the Test is thencalculated by;

Ap = FTest / PLS [3]

it L air mi tp

LS ct pt t rt

. (1 / )[1 ( )( ]

m g CA

P T TΣ − +

=+ + −

ρ ρ γα α [4]

The data thus obtained is recorded at differentpressure points and observations were repeated sixtimes on each pressure point. The pressure measuredby LS is then least square fitted against the effectivearea of Test to determine the value of A0t (zero pressureeffective area) and λt (distortion coefficient) of the Test.The measurement uncertainties in both the ranges arecomputed as per guidelines available in the literature[25-29]. The characterized values of the pressuremeasured using Eq. (1), the effective area calculatedusing Eq. (4), and the deviations from the averagevalues are plotted in Fig. 2 and data are given in Table1. The measurement uncertainties of the A0t and λt areestimated from the standard deviations of the A0t andλt obtained from the least square fitting method. Theuncertainty budget for a maximum pressure of 70 MPais shown in Table 2. The denominator in Eq. (2) isincluded as a temperature correction for effective areaof the artifact.

The detailed metrological coefficients and theassociated uncertainties of the Test for all thesuccessive calibrations performed in 2007, 2009 and2010 are shown in Table 2 for assigning the referencevalues. The average values of measured pressure 1( )p ,

effective area and relative deviation (σi) of (say for first calibration in 2007) in a particularcalibration is determined by simple statisticalcomputation as follows;

[5]

pt1

p1

nA

An

=∑

[6]

( )p1p11

p1

A A

A

∆ −σ = [7]

The reference values of the pressure measured

( )1 2 3

3

p p pp

+ += and effective area

( )p1 p2 p3

p 3

A A AA

+ += are then obtained by

arithmetic mean of the data obtained during thesethree calibrations. The relative deviation (σ) ofreference measured pressure value (p) is computed byroot sum square method as follows;

2 2 21 2 3= + +σ σ σ σ [8]

The effective area Ap of the Test is plotted as a

Fig. 2. The effective area Ap and relative deviations ofAp plotted as a function of applied pressure pfor all the three successive calibrations

pt( )A 1( )p

LS1

1

nP

pn

=∑


138

function of applied pressure (p) for all the successivecalibrations alongwith their relative deviations of theeffective area as shown in Fig. 2. The relativedeviations of the effective area, calculated as

p

p1 p2 p3p

p

, ,A

A A A AA

⎛ ⎞−= ⎜ ⎟⎜ ⎟⎝ ⎠

σ are well within ± 25 × 10-6.

The effective area Ap is also plotted as a function ofapplied pressure p. The values of the zero pressureeffective area, A0t and the distortion coefficient, λt ofthe Test, determined through least square fitting arealso shown in Table 1 for all the three calibrations.The deviation of average value of A0t are found to bewell within 37 × 10-6. Thus the deviations of Ap andA0t are found well within the standard uncertainty ofthe artifact estimated as 72 × 10-6 and shown in Table2 as detailed uncertainty budget. This suggests thatthe metrological parameters of the artifact remainedstable and well within the estimated uncertaintybudget of the artifact during the whole comparisonperiod.

Each, participating laboratory was assigned arandom code number and only these code numbersare used in this paper. As per NABL policy, the detailsof these code numbers are not divulged herein.However, these code numbers were reported to NABL,separately. However, the Code number assigned toreference laboratory, NPLI, is '1'.

3.2 Experimental Setup and Calibration Procedure

All the laboratories were advised to install theexperimental set-up as shown in Fig. 3. Usually,laboratory standard (LS) and the artifact were leveledusing the leveling screws and the sprit level. Thenecessary weights were placed on the carrier of theNPL100MPN and adjusted as per the values of thepressure generated by the artifact. This is repeatedseveral times so that the error due to the adjustment ofthe weights is minimized. Sufficient time,approximately 30 min, was provided between twosuccessive observations so that both the systems arein complete equilibrium. Participants were instructedto place the necessary weights on the carrier of the LSso that the values of the pressure generated by pistonplus carrier plus weights loaded are equal to themeasurement points equalizing the pressure withpressure generated by the artifact. This process iscalled Cross Floating of pressure balances [13-15, 25].During the cross floating, both the gauges areconnected directly. The pressure balances areconsidered to be in balance when sink rate of each ofthe piston is normal for that particular pressure. Atthis position, there was no pressure drop in theconnecting line and consequently no movement of thefluid. About 10 min time was provided between twosuccessive observations to allow the system to returnto equilibrium and 5 min time was sufficient to repeatthe observations. A waiting time of 10 min was given

Fig. 3. Experimental setup for the measurement using dead weight tester as an artifact


139

Tabl

e 1

Det

ails

of m

etro

logi

cal c

hara

cter

istic

s of t

he A

rtif

act a

nd a

ssig

nmen

t of r

efer

ence

val

ues

(All

the

valu

es re

port

ed h

ere

are

at g

NPL

= 9

.791

2393

m/s

2 and

refe

renc

e te

mpe

ratu

re o

f Tr =

23

°C)

Cha

race

risa

tions

of t

he a

rtif

act p

erfo

rmed

dur

ing

2007

2009

2010

Ref

eren

ce V

alue

s A

ssig

ned

Nom

inal

Pr

essu

reM

Pa

p 1

(MPa

)A

p1

mm

2

x 10

-6

p 2(M

pa)

Ap2

mm

2

x 10

-6

p 3(M

Pa)

Ap3

mm

2

x 10

-6M

pam

m2

x 10

-6

76.

9952

9.78

7747

7.13

6.99

5155

9.78

7876

14.0

06.

9952

349.

7878

748.

966.

9952

09.

7878

318

109.

9961

069.

7874

697.

149.

9957

669.

7878

7414

.07

9.99

6156

9.78

759

8.98

9.99

601

9.78

764

18

1514

.997

479.

7873

757.

1614

.997

19.

7876

9214

.25

14.9

975

9.78

753

9.03

14.9

9736

9.78

753

18

2019

.998

949.

7872

847.

2019

.999

029.

7873

0814

.49

19.9

9924

9.78

7301

9.10

19.9

9907

9.78

730

19

2525

.000

619.

7871

487.

2425

.000

629.

7871

9814

.79

25.0

0111

9.78

711

9.18

25.0

0078

9.78

715

19

3030

.002

489.

7869

757.

2930

.002

059.

7871

7815

.15

30.0

0307

9.78

6944

9.29

30.0

0253

9.78

703

19

4040

.006

199.

7867

787.

4240

.006

139.

7868

5616

.03

40.0

072

9.78

6682

9.55

40.0

0651

9.78

677

20

5050

.010

199.

7865

897.

5850

.010

329.

7866

3517

.10

50.0

1169

9.78

6456

9.88

50.0

1073

9.78

656

21

6060

.014

49.

7864

377.

7760

.014

649.

7864

7218

.32

60.0

1638

9.78

6275

10.2

660

.015

149.

7863

922

7070

.018

839.

7863

097.

9970

.019

439.

7863

0219

.67

70.0

2114

9.78

6138

10.7

070

.019

809.

7862

524

Para

met

er20

0720

0920

10

A0t /

mm

29.

7876

7634

9.78

8310

889.

7878

715

/ m

m2

9.78

7952

uA

A(

)/·1

00t

0t6

2.0

7.65

3.24

(-<

>)/<

>·10

AA

A0t

0t0t

628

.336

.6λ t

x(1

0 M

Pa)

6-1

.948

2-3

.152

3-2

.753

8λ t

(10

MPa

)6

-2.6

181

u() x

(10

MPa

)6

λ t0.

052

0.19

80.

084

( -<

>) x

(10

MPa

)6

λλ

tt

0.67

0.53

0.14

11

11

pp

p

AA

AΔ

−−

σ2

22

2

pp

p

AA

AΔ

−−

σ3

33

3

pp

p

AA

AΔ

−−

σ(

)+

+=

12

33

pp

pp

()

++

=1

23

3p

pp

PA

AA

Aσ=

σ+σ

+σ

22

21

23

A0t


140

Table 2Uncertainty Budget of the Artifact Used at a Maximum Pressure of 70 MPa and at Trt = 23 °C

Parameter Value Type Sensitivity Coefficient Standard Uncertainty

Relative Uncertainty (x 10 )-6

Parameter Value

Repeatability (M )

0.00142904 Type-A (1/ )p 0.014281401 5.83E-04 8.33

m ( )it kg 69.994137 Type-B (1/ ) mit 0.014286911 2.18E-04 3.11

g ( )L m/s2 9.7912393 Type-B (1/ )gL 0.102132117 1.01E-05 1.03

ρair (kg/m3) 1.15951717 Type-B (1/ )miρ 0.000125 3.52E-03 0.44

ρ ( )mi kg/m3 8000 Type-B ( / )air mi2ρ ρ 1.81175E-08 4.56E+01 0.83

ρ (N/ )m 0.0312 Type-B ( / . )C m gt it L 1.6182E-05 3.12E-03 0.05

C ( )t m 0.01109 Type-B ( / . )m git Lγ 4.55256E-05 5.00E-06 0.00

A ( )0t m2 9.7878715 Type-A (1/ )A0t 102167.2587 3.17E-05 3.24

α (/ )pt0C 0.0000045 Type-B ( )T -Tm r 0.13 4.50E-07 0.06

αct(/ C)00.0000045 Type-B ( )T -Tt rt 0.13 4.50E-07 0.06

T -T ( )t rt0C 0.13 Type-B ( + )pt ctα α 9.000000E-06 1.20E-01 1.08

λ )t(MPa-1 -2.7538E-06 Type-A ( )p 70.02114 8.39E-08 5.88

p ( )MPa 70.02114 Type-B ( )tλ -2.7538E-06 4.90E-03 -0.01

Pa

The total relative standard uncertainty evaluated through Type A method = 13.7 x 10-6

(root mean square of sum of squares all the uncertainty componentsevaluated through Type A method and stability of zero pressureeffective area A0t)

The total relative standard uncertainty evaluated through Type B method = 3.58 x 10-6

(root mean square of sum of squares all the uncertainty components evaluated through Type B method)

The relative standard uncertainty of the standard used in the measurements = 70.0 x 10-6

The combined relative standard uncertainty (at k = 1) = 72.0 x 10-6


141

after taking the reading at maximum pressure rangei.e. 70 MPa to start the observations in decreasingpressure order of pressure. This procedure wasrepeated for 10 pressure points (7, 10, 15, 20, 25, 30,40, 50, 60 and 70) MPa, selected for the presentcomparison and observations were repeated six times(3 times each in increasing and decreasing order) foreach pressure point and the values of the pressuregenerated, the repeatability and the expandeduncertainty were computed.

4. Data Compilation and Analysis

All the laboratories were advised to submit theirmeasurement results within a month after completionof the measurements. Laboratories were also asked tosubmit the copies of the calibration certificates for theLSs used in the measurements, calculation sheet fordetermining the uncertainty in measurements and thecalibration certificate issued to the customer for suchmeasurements. The values of the measured pressure,acceleration of local gravity and the referencetemperature reported by the participants are shownin Table 3. The following corrections were appliedbefore compiling and comparing the results;

4.1 Gravity Correction

The measured pressure values reported by thelaboratories are corrected for gNPL = 9.7912393 m/s2

(acceleration of gravity at NPLI, New Delhi) using thefollowing relationship;

pcorr = prep * (gNPL / gLAB) [9]

where pcorr and prep are the values of the corrected andthe reported pressure, respectively and gLAB is thevalue of the acceleration of gravity reported by thelaboratory.

4.2 Temperature Correction

The measured pressure values reported by thelaboratories were also corrected for the temperatureat 23 0C using the following formula;

p' = [pcorr / {1+(α'p+α'c)*(23 - TLAB)}] [10]

where, p' is the final corrected pressure, α'p, α'c andTLAB are the thermal expansion coefficients of piston,thermal expansion coefficients of cylinder and thereference laboratory temperature, respectively,

reported by the laboratory.

4.3 Estimation of Normalized Error (En)

The measurement performance of the laboratorywas assessed on the basis of normalized error (En)number of each measurement. The En values areestimated for each participant at each pressure usingthe equation [4-5, 30];

{ } { }R

R

n 22

'Value

( ') ( )

p pE

U p U p

−=

+ [11]

where p' is the participant's measured pressure value,pR is the reference pressure value, U(p') is theparticipant's claimed expanded uncertainty at acoverage factor k = 2 and U(pR) is the expandedmeasurement uncertainty of the reference value at acoverage factor k = 2.

5. Results

Details of the values of measured pressure (prep)and other metrological characteristics of thelaboratories standards reported by the participantsare shown in Table 3. However, the details of thecorrected pressure (p') for gravity (gNPL) and at 23 0Care shown in Table 4, the deviations from the referencevalue (pR) in (kPa) in Table 5. The measured pressurevalues, the associated uncertainties and calculatedEn values for individual pressure points are depicted inFigs. 4 (a)-(k).

In general, the performance of the laboratory isconsidered satisfactory if normalized error En is inside± 1. The plots shown in Figs. 4 (a) to (k) reveal thatthere were total 87 measurement results. Measurementresults of 5 laboratories out of total 9 laboratories werefound well within acceptable limits of the normalizederror over the entire pressure range of (7 to70) MPa.Results of the two other laboratories were also withinacceptable limit except one pressure points of 40 MPa.The En values of 68 measurement results out of thetotal 87 were found well within ± 1, which is 78.2%.These results are acceptable. The En values of thelaboratory referred as Code No. 6, were > 1 at all themeasurement points. An En value greater than unitymeans that there is a significant bias in the laboratory'sresults and that the quoted value of its associateduncertainty does not adequately accommodate that


142

Table 3Details of the reference values measured pressure (prep ) reported by the participants and other

metrological characteristics of the laboratories standards

Nominal Laboratory codepressureprep 1 2 3 4 5 6 7 8 9 10(MPa)

7 6.995196 7.007740 6.989308 6.991900 7.006520 6.997410 6.989166 6.993000 6.993290 6.99527710 9.996010 10.013790 9.988209 9.988600 10.012280 9.994660 9.986768 9.993300 9.995280 9.99561415 14.99736 15.02425 14.98599 14.98890 15.02132 14.99075 14.98347 14.99390 14.99259 15.0006220 19.99907 20.03436 19.98388 19.98750 20.03044 19.98668 19.98010 19.99360 19.99289 20.0016025 25.00078 25.04505 24.98173 24.99090 25.04017 24.98295 24.97728 24.99650 24.99253 25.0033530 30.00253 30.05539 29.98007 29.99000 30.04992 29.98000 29.97385 29.99650 29.99118 30.0024840 40.00651 40.07749 39.97616 39.99080 40.06945 39.97528 39.96857 39.99630 39.99024 40.0075550 50.01073 50.09954 49.97274 49.99040 50.08887 49.96934 49.96401 49.99395 50.0131160 60.01514 60.12077 59.97089 59.98490 60.10919 59.96754 59.95960 59.99573 60.0192670 70.01980 70.14357 69.96813 69.97160 70.12950 69.96917 69.95400 69.99858 70.02413gLAB 9.7912393 9.78269777 9.78244 9.7823794 9.80665 9.780352 9.7836664 9.7912393 9.7909591 9.7914503(m/s2)TLAB (°C) 23 23 23 23 23 23 23 23 23 23A0 mm2 9.805937 4.032 10.01621 4.0332310 9.817711 4.03181 4.03440 - 4.02896 4.02896λ x 10-6 0.822 1.4 2.06 2.0 1.02 1.2 3.0 - 3.43 8.22/ MPaαp x 10-6 4.55 5.0 4.5 5.0 4.5 5.0 11.0 - 4.9 4.5/ °Cαc x 10-6 4.55 11.5 4.5 11.5 4.5 11.5 11.0 - 11.5 4.5/ °CTrans- Sebacate VG22 Sebacate Shell Sebacate ISO ST55 SAE- Spinesstic Sebacatemitting Tellus VG22 20W/40 22Fluid Used 731Trace- NPLI-H1 DH- NIM, UKAS, NAVLAP, UKAS, DKD, NPLI, NPLI, NPLIability CCM. Buden- China U.K. USA U.K. Germany India India India

P. K7 berg,APMP. UKM.P.K7

Table 4Details of the corrected pressure (p') for gNPL and at 23 °C

Laboratory codep' (MPa) 1 2 3 4 5 6 7 8 9 10

7 6.99520 6.99673 6.99559 6.99826 6.99551 7.00520 6.99458 6.99300 6.99349 6.9951310 9.99601 9.99805 9.99719 9.99768 9.99655 10.00579 9.99450 9.99330 9.99557 9.9954015 14.99736 15.00064 14.99947 15.00252 14.99771 15.00744 14.99507 14.99390 14.99302 15.0003020 19.99907 20.00288 19.99907 20.00288 20.00185 20.00567 19.99896 20.00893 19.99557 19.9936025 25.00078 25.00569 25.00420 25.01362 25.00082 25.01076 24.99661 24.99650 24.99325 25.0028130 30.00253 30.00816 30.00704 30.01726 30.00270 30.01337 29.99705 29.99650 29.99204 30.0018340 40.00651 40.01451 40.01212 40.02715 40.00648 40.01978 39.99950 39.99630 39.99138 40.0066850 50.01073 50.0 50.01769 50.03584 50.01016 50.02496 50.00268 - 49.99538 50.0120360 60.01514 60.02629 60.02483 60.03943 60.01473 60.03429 60.00601 - 59.99745 60.0179770 70.01980 70.03334 70.03106 70.03520 70.01929 70.04706 70.00815 - 70.00058 70.02262


143

(a)

Fig. 4

(b)

( c )

(Fig. 4 continued to next page)


144

(e)

(Fig. 4 continued to next page)Fig. 4

(f)

(d)


145

(g)

(h)

(Fig. 4 continued to next page)

( i )

Fig. 4


146

(j)

Fig. 4 (a) to (k) represent the plots at (7, 10, 15, 20, 25, 30, 40, 50, 60 and 70) MPa, respectively. Black diamondsindicate the deviation of the measured pressure (p') by the laboratory from the reference value (pR). The estimatedexpanded measurement uncertainty (at k = 2) of the laboratory is depicted as error bars across the central solidline. The gap between two horizontal dotted lines shows the expanded measurement uncertainty band of thepR. Fig. 4 (k) showing the summary of the normalized error value (En) as a function of measured pressure (p') foreach laboratory. The gap between two horizontal dotted lines shows the acceptable limit of the normalizederror value

(k)


147

bias and need further investigations at the part of thelaboratory.

The larger the absolute value of the En number,the bigger the problem. The graphical representationsin Figs. 4 (a) to (k) give the agreement between theparticipating laboratories and the reference laboratory.The results lying within the uncertainty band of thereference laboratory is an indication of the satisfactoryresults without any bias in the measurement. It isclearly evident from these plots that the deviationsbetween the laboratories values and the referencevalues are well within the uncertainty bands of the

reference values for 68 measurements points.

The bias in the measurements may be due to theerrors in the measuring instrument or the estimation/ measurement of local acceleration of gravity, thevalues of thermal expansion coefficients of piston andcylinder materials reported and used by the laboratoryand under the estimation of the measurementuncertainty. The management of Laboratory withCode No.-6 is required to rectify the problems by areview of their uncertainty calculations and othersystematic affects as mentioned above.

Fig. 5 (a). Deviations (in kPa) of the measured pressure (p') by each laboratory from the reference value (pR). Thegap between two horizontal dotted lines shows the expanded uncertainty band of the reference values

Table 5Deviations (in kPa) of the measured pressure of each participating laboratory from reference values (p)

Laboratory Codep' (MPa) 1 2 3 4 5 6 7 8 9 10

7 - 1.53 0.40 3.06 0.31 10.00 -0.62 -2.20 -1.71 -0.0710 - 2.04 1.18 1.67 0.54 9.78 -1.51 -2.71 -0.44 -0.6115 - 3.28 2.11 5.17 0.36 10.08 -2.29 -3.46 -4.34 2.9420 - 3.81 2.79 6.60 -0.10 9.86 -3.50 -5.47 -5.60 2.1025 - 4.91 3.41 12.84 0.04 9.98 -4.17 -4.28 -7.54 2.0330 - 5.63 4.50 14.73 0.16 10.84 -5.48 -6.03 -10.49 -0.7040 - 8.00 5.61 20.65 -0.02 13.27 -7.00 -10.21 -15.12 0.1850 - 10.08 6.96 25.11 -0.57 14.23 -8.05 - -15.35 1.3060 - 11.15 9.69 24.29 -0.41 19.15 -9.13 - -17.69 2.8370 - 13.54 11.26 15.40 -0.51 27.26 -11.65 - -19.22 2.82


148

Fig. 6: Estimated expanded measurement uncertainties U(p') and U(pR) reported by each laboratory

Fig. 5(b). Relative deviations (%) of the measured pressure (p') by each laboratory from the reference value (pR).The gap between two horizontal dotted lines shows the expanded uncertainty band of the reference values

The deviations of the measured pressure (p') byeach laboratory from the reference value (pR) are shownin Figs. 5(a) in kPa and 5(b) in relative terms ofpercentage. It is clear from the plots that deviationsare well within 0.15% for 34 measurement points andwithin 0.05% for 82 measurement points. This suggeststhat deviations are well within 0.05% for 94%results. Since this comparison was organized basically

for those laboratories having measurementcapabilities of 0.05% or better, such results are quiteencouraging. It is also worth mentioning here that 37 %measurement results are well within the uncertaintyband of the reference values, specially for thelaboratories with Code Nos. - 3, 5 and 10. Theselaboratories deserve appreciation.


149

The expanded measurement uncertainty U(p')reported by each laboratory and the combined

expanded uncertainty { } { }R

22c ( ) ( ') ( )U p U p U p= +

estimated in this comparison are shown in Figs. 6and 7, respectively. Some of the laboratories reportedbetter measurement uncertainties then themeasurement uncertainties of the reference laboratory,U(pR)

6. Discussion and Suggested Corective Actions

Although all the participating laboratories wereasked to submit the copy of the formal calibrationcertificate issued to the customer and traceabilitycertificates of their standards, only 4 laboratories havesubmitted the copies of the formal calibrationcertificates of the dead weight tester (artifact, in thepresent case) while traceability certificates weresubmitted by only 5 laboratories. The seriousness tofollow the instructions given in TP was found lackingas only 3 laboratories with Code numbers 3, 4 and 6submitted both of these required documents.Laboratories with code numbers 2 and 10 even didnot bother to submit any of such formal certificates.The certificates thus obtained were examined andfound adequate except that there was little uniformityin the calibration certificates issued by the

participants, especially in reporting the measurementresults. Most of the laboratories submitted theirmeasurement results in time. All the laboratoriesprepared their 'Uncertainty Budget' as per instructionsgiven in the TP. Some of the laboratories reported bettermeasurement uncertainties then the measurementuncertainties of the reference laboratory.

As mentioned in Section 5, the En numbers greaterthan unity require investigations and corrective actionby the participating laboratory. The laboratory'smanagement needs to ensure that the problem isrectified and procedures are put in place to prevent arecurrence. Laboratories with Code Nos. 4 and 6 needto review their results and take appropriate correctiveactions. These laboratories are advised to improvetheir calibration facilities / modify the measurementmethod.

7. Conclusion

An interlaboratory comparison programme(proficiency testing) is carried out in the pressurerange 7 - 70 MPa using dead weight tester as anartifact. Total number of 9 NABL accreditedlaboratories participated in this programme. Thecomparison was performed at 10 pressure pointsselected arbitrarily throughout the entire pressurerange. The proficiency testing concludes that out of

Fig. 7. Combined expanded measurement uncertainty Uc(pR) estimated for each laboratory in this comparison


150

the total 87 measurement results reported here in thisreport, 68 (78.2%) are in agreement with the referencelaboratory. The results of 5 laboratories out of the total9 are acceptable and well within their reportedexpanded uncertainty at a coverage factor k = 2throughout the entire pressure range. Two otherlaboratories also have acceptable results except onepressure point. These results are quite encouragingin the country. However, after taking the appropriatecorrective actions by the participant laboratories, weexpect that the performance of the laboratories withCode Nos. - 4 and 6 would certainly improve. Theselaboratories may reassess their measurementcapabilities by participating in the future PTprogramme in this area.

Acknowledgement

We are grateful to Prof. R.C. Budhani, Director,National Physical Laboratory, New Delhi and Dr.Anil Relia, Director, National Accreditation Board forTesting & Calibration Laboratories, New Delhi fortheir support and encouragement throughout thisprogramme. The PT programme has been conductedas per NPL-NABL MoU. We are also thankful to Dr.K.P. Chaudhary, Programme Coordinator, NPL-NABLPT Programme for his constant co-operation and timeto time suggestions and discussions, which were veryhelpful during course of this comparison. Thanks arealso due to all the nine accredited laboratoriesparticipating in this interlaboratory comparisonexercise. Without their active support and co-operation this PT programme would have not beencompleted in time. We would also like to acknowledgethe help of the secretariat of NABL for theiradministrative help.

References

[1] w w w . b i p m . o r g / e n / c i p m - m r a /mra_main_text.html

[2] General Requirements for the Competence ofTesting and Calibration Laboratories", ISO /IEC 17025: (2005).

[3] Procedures for Establishing and Maintainingthe APLAC Mutual Recognition ArrangementAmongst Accreditation Bodies, APLACMRA001, Issue No. 13, (2007).

[4] Proficiency Testing by Interlaboratory

Comparison: Part - 1: Development andOperation of Proficiency Testing Schemes. Part- 2: Selection and use of Proficiency TestingSchemes by Laboratory Accreditation Bodies,ISO / IEC 43: (1997).

[5] S. Yadav and A.K. Bandyopadhyay, ProficiencyTesting Program Under NABL in the PressureRange 7-70 MPa Using a Dead Weight Tester,Med. J. Meas. Contrl., 1 (2005) 138-151.

[6] S. Yadav and A.K. Bandyopadhyay, ProficiencyTesting (PT) Program Under NABL in thePressure Range 7 - 70 MPa, Metrology andMeasurement Systems, XII (2005) 323-340.

[7] S. Yadav, V.K. Gupta and A.K. Bandyopadhyay,Standardization of Pressure Calibration (7-70MPa) using Digital Pressure Calibrator, J. Sci.and Indusl. Res., 69 (2010) 27-33.

[8] S. Yadav, V.K. Gupta, O. Prakash and A.K.Bandyopadhyay, Proficiency Testing ThroughInterlaboratory Comparison in the PressureRange 7-70 MPa Using Pressure Dial Gauge asan Artifact, J. Sci. and Indusl. Res., 64 (2005)722-740.

[9] S. Yadav, V.K. Gupta, O. Prakash and A.K.Bandyopadhyay, Evaluation of InterlaboratoryPerformance through Proficiency Testing UsingPressure Dial Gauge in the Hydraulic PressureMeasurement up to 70 MPa", Mapan-Journalof Metrology Society of India, 23 (2008) 79-99.

[10] S. Yadav and A.K. Bandyopadhyay, ExternalProficiency Testing in the Pressure Range up to60 MPa", Measure, USA, 4 (2009) 42-51.

[11] S. Yadav and A.K. Bandyopadhyay, Evaluationof Laboratory Performance throughInterlaboratory Comparison, Mapan-Journal ofMetrology Society India, 24 (2009) 125-138.

[12] Guidelines for Proficiency Testing Program forTesting and Calibration Laboratories, NABLDoc. 162, (2001).

[13] F. Pavese and G. Molinar, Modern Gas BasedTemperature and Pressure Measurements,Plenum Press, New York, (1992).

[14] S. Yadav, A.K. Bandyopadhyay, N. Dilawarand A.C. Gupta, Intercomparison of NationalHydraulic Pressure Standards up to 500 MPa,Measurement + Control, UK, 35 (2002) 47-51.


151

[15] S. Yadav, R. Agarwal, D.R. Sharma, A.K.Bandyopadhyay and A.C. Gupta, ModernInstrumentation Techniques in PressureMetrology under Static Conditions, Mapan-Journal of Metrology Society of India, 18 (2003)57-82.

[16] A.K. Bandyopadhyay, S. Yadav and N. Dilawar,Current Status of Pressure Standards at NPLIand our Experiences with the Key ComparisonData Base (KCDB), Mapan-Journal of MetrologySociety of India, 21 (2006) 127-145.

[17] S. Yadav, A.K. Bandyopadhyay and A.C. Gupta,Characterisation of National HydraulicPressure Standards in the Pressure Ranges upto 100 MPa, 200 MPa and 500 MPa, Callab: TheInternational J. Metrology, (2003).

[18] A.K. Bandyopadhyay and A.C. Gupta,Realization of a National Practical PressureScale for Pressure up to 500 MPa, Metrologia,36 (1999) 681-688.

[19] S.Yadav, O. Prakash, V.K. Gupta and A.K.Bandyopadhyay, The Effect of Pressure-Transmitting Fluids in the Characterization ofa Controlled Clearance Piston Gauge up to 1GPa, Metrologia, 44 (2007) 222-233.

[20] S. Dogra, S. Yadav and A.K. Bandyopadhyay,Computer simulation of a 1.0 GPa Piston-Cylinder Assembly Using Finite ElementAnalysis (FEA), Measurement, 43 (2010) 1345-1354.

[21] T. Kobata, A.K. Bandyopadhyay, K. Moore, A.E.Eltawil Alaaeldin, S.Y. Woo, T.K. Chan, W. Jian,J. Man, N.N. Con, C.S. Fatt, W. Permana, M.Aldammad, W. Sabuga, T. Changpan, C.C.Hung and Z. Pengcheng, Final Report on KeyComparison APMP.M.P-K7 in HydraulicGauge Pressure from 10 MPa to 100 MPa",Metrologia, 42 (2005) 07006.

[22] W. Sabuga, M. Bergoglio, T. Rabault, B. Waller,J.C. Torres, D.A. Olson, A. Agarwal, T. Kobataand A.K. Bandyopadhyay, Final Report on Key

Comparison CCM.P-K7 in the Range 10 MPa to100 MPa of Hydraulic Gauge Pressure,Metrologia, 42 (2005) 07005.

[23] R.G. Driver, D.A. Olson, S. Yadav and A.K.Bandyopadhyay, Final Report onAPMP.SIM.M.P-K7: Bilateral ComparisonBetween NIST (USA) and NPLI (India) in theHydraulic Pressure Region 40 MPa to 200MPa", Metrologia, 43 (2006) 07003.

[24] S. Yadav, D.A. Vijayakumar and A.C. Gupta,Computer Software for Calibration of Industrialand Master Simple / Reentrant Type PistonGauges", Mapan-Journal of Metrology Societyof India, 12 (1997) 101-104.

[25] S. Yadav, Characterization of Dead WeightTesters and Computation of AssociatedUncertainties: A Case Study of ContemporaryTechniques, Metrology and MeasurementSystems, XIV (2007) 453-469.

[26] S. Yadav, B.V. Kumaraswamy, V.K. Gupta andA.K. Bandyopadhyay, Least Squares Best FitLine Method for the Evaluation of MeasurementUncertainty with ElectromechanicalTransducers (EMT) with Electrical Outputs(EO)", MAPAN-Journal of Metrology Society ofIndia, 25 (2010) 97-106.

[27] S. Yadav, V.K. Gupta and A.K. Bandyopadhyay,Investigations on Measurement Uncertaintyand Stability of Pressure Dial Gauges andTransducers", Measurement Science Review, 14(2010) 130-135.

[28] Guide to the Expression of Uncertainty inMeasurement", ISO Document - ISO/TAG/WG3: (1995) (E).

[29] Guidelines for Estimation and Expression ofUncertainty in Measurement", NABL Doc. 141,(2000).

[30] H.S. Nielsen, Determining Consensus Valuesin Interlaboratory Comparisons and ProficiencyTesting, Proceedings NSCL Conference, Tampa,Florida, USA, (2003) pp. 1-16.

standardisation of pressure measurement using pressure ...metrologyindia.org/26b/sanjay...

Documents