EUROMET PROJECT FINAL REPORT - EURAMET · PDF fileEUROMET PROJECT FINAL REPORT 1 ... The results of this Euromet key-comparison can now be linked to CCQM K3 ... NMi Van Swinden Laboratorium

EUROMET PROJECT

FINAL REPORT

1. Ref. No.: 485

2. Subject Field: Amount of Substance / METCHEM

3 Type of collaboration: Key-comparison 4A. Partners: (institutions) NMi, SMU, IMGC, IPQ, CSIR-NML, GUM, CMI-CHMI

4B. CEC funded? No

5. Participating countries: Netherlands, Slovakia, Italy, Portugal, South-Africa, Poland, Czech Rep. 6. Title: key comparison on automotive emission gases 7. Progress: In its second meeting the METCHEM Gas Working group approved the draft B report as presented by the coordinator. The results of this Euromet key-comparison can now be linked to CCQM K3. A summary of this key-comparison will be sent to Metrologia for publication, preferably in combination with the publication of K3. 8. Coordinator’s name: Rob Wessel Adress: NMi Van Swinden Laboratorium, P.O. Box 654, NL 2600 AR Delft Tef: tel. +31 15 2691677 E-mail: Rwessel@NMi .nl fax: +31 15 2612971 9. Completion date: 2001-03-01

10. Coordinator´s signature: 11. Date: 2001-03-06

EUROMET 485 Key Comparison “Automotives”

Draft B

Adriaan M.H. van der Veen / Rob M. Wessel

NMi Van Swinden Laboratorium B.V.

Department of Chemistry

Report number S-CH.00.06.01

Delft, 20 November 2000

2

Summary Following-up the CCQM-K3 key comparison on “automotives”, EUROMET accepted the project proposal for the organisation of a regional key comparison. The objectives of this EUROMET key comparison are essentially the same as for the CCQM-K3 comparison: to compare the measurement capabilities of national metrological institutes (NMIs) in measuring amount of substance fractions of carbon monoxide, carbon dioxide, and propane in nitrogen.

NMi Van Swinden Laboratorium B.V. operated as pilot laboratory in this comparison. The selected PSMs for this comparison were individually prepared using gravimetry and thoroughly studied for their chemical composition and stability. A long-term experience in the behaviour of these mixtures and the technical challenges in preparing batches of very similar mixtures is available at the pilot laboratory. Seven laboratories participated: CMI-CHMI (Czech Republic), GUM (Poland), IMGC (Italy), IPQ (Portugal), CSIR-NML (South Africa), SMU (Slovak Republic) and NMi (Netherlands).

Five of the seven participating laboratories have shown on all components to be able to assign values for all analysed components within ± 1% relative to the gravimetric value. Apart from propane, this is also true for IMGC. The results of CMI-CHMI show larger deviations from the gravimetric value, which are however covered by the reported uncertainty.

All participants provided satisfactory descriptions of their measurements, including the measurement technique(s), calibration method, and the evaluation method for measurement uncertainty.

III

Table of contents

1 Introduction 1

2 Participants 2

3 Design of the comparison 3

4 Measurement methods and calibration procedures 4

5 Evaluation of results 5

5.1 Uncertainty evaluation of the assigned value 5

5.2 Evaluation of results 9

5.3 Computation of degrees of equivalence and associated uncertainties 9

6 Results 11

7 Conclusions 20

IV

List of tables

Table 1: List of participants ................................................................................................................................... 2 Table 2 : Nominal amount of substance ratios ....................................................................................................... 3 Table 3 : Measurement and calibration methods ................................................................................................... 4 Table 4 : Uncertainty evaluation of wCO................................................................................................................. 6 Table 5: Uncertainty evaluation of wCO2 ................................................................................................................ 6 Table 6: Uncertainty evaluation of wC3H8 ............................................................................................................... 6 Table 7: Uncertainty evaluation of wN2 .................................................................................................................. 6 Table 8: Uncertainty components for a typical automotive mixture....................................................................... 9 Table 9: Results for CO ........................................................................................................................................ 12 Table 10: Results for CO2..................................................................................................................................... 12 Table 11: Results for propane .............................................................................................................................. 13 Table 12 : Degrees of equivalence and expanded uncertainties for CO (mmol/mol) ........................................... 17 Table 13: Degrees of equivalence and expanded uncertainties for CO2 (mmol/mol)........................................... 18 Table 14: Degrees of equivalence and expanded uncertainties for C3H8 (mmol/mol) ......................................... 19

List of figures Figure 1: Stability data of CO in nitrogen ............................................................................................................. 8 Figure 2: Results for CO ...................................................................................................................................... 14 Figure 3: Results for CO2 ..................................................................................................................................... 15 Figure 4: Results for C3H8.................................................................................................................................... 16

V

1 Introduction Following-up the CCQM-K3 key comparison on “automotives”, EUROMET accepted the project proposal for the organisation of a regional key comparison. The objectives of this EUROMET key comparison are essentially the same as for the CCQM-K3 comparison: to compare the measurement capabilities of national metrological institutes (NMIs) in measuring amount of substance fractions of carbon monoxide, carbon dioxide, and propane in nitrogen.

NMi Van Swinden Laboratorium B.V. operated as pilot laboratory both in CCQM-K3 and in this comparison. The selected PSMs for this comparison were individually prepared using gravimetry and thoroughly studied for their chemical composition and stability. A long-term experience in the behaviour of these mixtures and the technical challenges in preparing batches of very similar mixtures is available at the pilot laboratory.

A target of 1% is maintained for the required agreement between a national metrology institute and the key comparison reference value. The 1% target had a direct impact on the choice of the analytical reference methods and it was decided that the preferred analytical methods for gas analysis in general would be:

- High accuracy mass spectroscopy, - Isotope dilution mass spectroscopy, - Gas chromatography, - Chemiluminescence and Fluorescence, - Non dispersive infrared spectroscopy

Furthermore, the difference observed between the value of a national metrology institute and the key comparison reference value should not exceed the combined expanded uncertainty of this difference.

The uncertainty calculations used during this comparison are based on the experience gained in CCQM-K3. All calculations made are fully compliant to the principles of the “Guide to the expression of uncertainty in measurement” (GUM), and represent state of the art in gas analysis.

1

2 Participants The following institute participated in this key comparison (in alphabetical order).

Table 1: List of participants

Institute City Country Spirit CMI-CHMI Prague Czech Republic Best measurement capability GUM Warsaw Poland IMGC Torino Italy Best measurement capability IPQ Monte de Caparica Portugal Research measurement CSIR-NML Pretoria South Africa Best measurement capability SMU Bratislava Slovak Republic Best measurement capability The fourth column in the table indicates the “spirit”, i.e. the objective from participation. Distinction is made only between best measurement capability and research type. Best measurement capability refers to the highest level of calibration or measurement normally offered to clients. Research measurement normally implies that the laboratory tried to optimise measurement conditions to minimise the resulting measurement uncertainty. This approach usually involves considerably more work. As stated in the introduction NMi-VSL prepared the gravimetric mixtures in both CCQM-K3 and in this comparison. As in both comparisons the gravimetric value is regarded as reference value this NMi-VSL reference value safeguards a link between both comparisons. In addition SMU participated in both comparisons.

2

3 Design of the comparison The design of the comparison was adopted from CCQM-K3. The gas mixtures were prepared by means of primary methods (gravimetry) at the pilot laboratory NMi VSL and in order to do the whole comparison in a limited time frame, a batch of 6 mixtures was produced. There are small differences in the actual property values of these mixtures, which makes working with a single reference value undesirable. The differences in the compositions are of the same order of magnitude as the (expected) differences between laboratories, so that these two aspects are interfering.

Apart of the time frame in this set-up, there is another practical consideration to use a batch of very similar mixtures. From previous experience as well as from logistic problems in this key comparison, it is known that there are sometimes problems with customs or other authorities. These problems may in the worst case eventually lead to the loss of a gas mixture. Obviously, this has dramatic consequences if it is the mixture that is lost, rather than if it is ‘only’ one mixture out of the batch. For the participating national metrology institute, the latter case is still quite unsatisfactory, but in the design used, a new cylinder could be shipped without having a dramatic impact on the whole key comparison.

The nominal amount of substance ratios of CO, CO2 and C3H8 in nitrogen, as used in this key comparison, are summarised in table 2.

Table 2 : Nominal amount of substance ratios

Component x (mmol/mol) CO 32 CO2 135 C3H8 2,05 N2 Balance

The cylinders were shipped November 1999. A formal deadline for submission of results was not set. Most laboratories carried out their measurements between January and May 2000. Reports were received until September 2000.

3

4 Measurement methods and calibration procedures The following methods of measurement and calibration methods have been employed (table 3).

Table 3 : Measurement and calibration methods

Laboratory Measurement method Calibration method SMU GC; FID (C3H8), TCD (CO, CO2) Linear regression (5 points),

weighted IMGC GC; FID (C3H8), TCD (CO2) Linear regression (3 points) IPQ GC; TCD Second order polynomial (CO2);

linear regression (CO and C3H8) CSIR-NML NDIR Polynomial regression (6

points) weighted GUM GC; FID (C3H8), TCD (CO, CO2) Linear regression (3 points) CMI-CHMI No information provided

The information provided in the reports is generally of good quality. Most participants describe in considerable detail the experimental set-up and the way in which they perform their measurements. The description of the evaluation of measurement uncertainty is usually very informative. Especially IMGC and IPQ provided information to an extent, that the procedure can be reconstructed almost stepwise.

4

5 Evaluation of results

5.1 Uncertainty evaluation of the assigned value The preparation of the gas mixtures has taken place using the substitution method. The model used to describe the preparation reads as follows. The basic equation for the weighing of the reference cylinder reads as follows

S

airair

Rair

c WVWRVepCρρρρ −−+=−+ (1)

and that for the sample cylinder

S

airair

Sair

c MVMSVeqCρρρρ −−+=−+ (2)

where the symbols have the following meaning C weight contra mass (comparator) R weight reference cylinder S weight sample cylinder W total mass of added mass pieces to reference cylinder M total mass of added mass pieces to sample cylinder e calibration factor (reading → mass) p reading reference cylinder q reading sample cylinder VC volume contra mass VR volume reference cylinder VS volume sample cylinder ρair density of air ρS density of mass pieces The basic equation for the difference between reference and sample cylinder is given by

( ) ( ) ( RjSairjS

airj

jjjjjj VVMWpqew −+

−−+−= ,1 ρ

ρρ ) (3)

which follows from subtracting (2) from (1). The subscripts j denote the variables that are determined in every cycle j, where j = 0..n (n is the number of parent gases the mixture).

The expression in equation (3) forms the basis for the computation of the uncertainty in the assigned value. The assignment of the amount of substance fractions is based on the mass brought in, and the compositions of the parent gases. In the case of propane, a pre-mixture has been used, which has been prepared in a similar way as the cylinders used for the key comparison.

The complete process involves 5 weighings, a weighing for each of the four components, and for the evacuated cylinder (= w0). The results of a typical cylinder are given in tables 4-7.

5

Table 4 : Uncertainty evaluation of wCO

Parameter Estimate u sensitivity coefficient

uncertainty contribution

(mg)

Degrees of

freedom E (g) 1.000E+00 1.947E-03 8.340E-01 1.62 2∆q (g) 8.340E-01 1.477E-03 1.000E+00 1.48 3∆W (g) 7.300E+01 2.956E-05 9.999E-01 0.03 ∞ρair (g/dm3) 1.187E+00 6.706E-04 9.605E-03 0.01 ∞∆V (dm3) 4.800E-04 3.000E-03 1.187E+00 3.56 1ρS (g/dm3) 8.000E+03 2.000E-03 1.354E-06 0.00 ∞w(j) 4.18

Table 5: Uncertainty evaluation of wCO2



Degrees of


Table 6: Uncertainty evaluation of wC3H8


u(w) mg

Degrees of


Table 7: Uncertainty evaluation of wN2



Degrees of


6

From these evaluations, it has become apparent that the largest uncertainty contribution is coming from the volume expansion ∆V; it has been assumed in the calculations to be 15 ml, with an uncertainty of 3 ml. In the calculation of ∆V, it is assumed that the dependence of ∆V on the internal pressure is linear. Part of the uncertainty of 3 ml deals with this aspect; another component contained is the uncertainty in the evaluation of VS anyway.

For all cylinders of the batch, the values for the uncertainty components given in tables 4-7 are representative. The differences for the estimates (the values of the parameters) are relevant though. For the computation of the uncertainty of the components, a combined standard uncertainty of 4 mg for all components in all mixtures has been used.

In the uncertainty of the preparation, furthermore the uncertainty from purity verification/composition of the parent gases is contained. The component of uncertainty hardly plays any role at all. The data from the purity verification/composition of the parent gases has been used for computing the gravimetric composition, as an amount of substance fraction.

A second main contributor to the uncertainty of the reference value is the uncertainty from verification. The verification process is used to confirm the gravimetric value. The uncertainty estimation of the verification is complex, and in view of the discussions of the incorporation of correlations in the estimation process, not yet complete. In order to keep in the time frame of this key comparison, a value for the uncertainty from verification has been derived from stability data. In stability data on the components, in principle, the following uncertainty components are contained – stability of the reference material – uncertainty from (different suites of) PRMs/PSMs used for calibration – calibration model – position of the “unknown” on the curve – lack of fit – repeatability of measurement – number of replicates Apart from the first component, for a single verification, all other components are relevant. As it is reasonable to expect that the stability of the components of interest (CO2, CO, and C3H8) in nitrogen is not an issue, the stability data form a valuable source of information.

From the stability data, it becomes apparent that there is no drift, and it is also observed that the mean observed value is very close to the gravimetric value. Otherwise, an extra bias component in the uncertainty calculation had to be accounted for. In the stability data, ‘reproducibility’ components like calibration, operator etc. are contained, so that no “extras” are needed. A typical stability chart is shown in figure 1.

7

Stability data for CO

-0.10

-0.08

-0.06

-0.04

-0.02

0.00

0.02

0.04

0.06

0.08

0.10

0 1 2 3 4 5

Time (years)

Rel

ativ

e di

ffer

ence

(%)

6

Figure 1: Stability data of CO in nitrogen

From the theory of analysis of variance (ANOVA), the following expressions can be obtained for MSamong, the mean square “among groups”

22rAamong snsMS += (4)

and 2nsMSamong = (5)

where s2 denotes the variance of the group means, n the number of replicates in each group, sA2 the

among group variance, and sr2 the repeatability variance. Each “group” represents a data point in time.

The significance in these expressions is, that apparently the variance of the data points accounts for both the among-group effects as well as for the effects of repeatability of measurement. These among group effects include calibration, operator, suite of PSMs/PRMs etc..

The standard uncertainty in each of the data points of figure 1 can be expressed as

( yVarsnssu r

Astab ∆==+= 22

22 ) (6)

whereby ∆y denotes the relative difference between measured and gravimetric value. For the evaluation of this uncertainty component, ygrav can be taken as a constant. This approach only works if – there is no drift – there is no bias – no additional variance component due to instability – the number of calibrants in the curve and the number of replicate measurements are constant – repeatability of measurement does not change – concentration level of the stability cylinder is relevant to the mixture in question – the lack of fit, repeatability of measurement etc. of the measurement in question is in line with the

relevant parameters in the measurement of the stability cylinder These assumptions are very important. Failing to check any of them may completely invalidate the approach.

It should be noted that this approach does not provide, in itself, any positive evidence that the estimator obtained for the uncertainty from verification is valid for the present case. Therefore, it cannot substitute for the approach of evaluating the experiment. This method is –under the assumptions stated– valid as a counter check on the evaluation method of a single experiment, as there

8

is obviously a relationship between the two. Finally, the method only works if on all intermediate levels the method is performing within specifications.

For a typical mixture, the following results have been obtained, whereby for ustab the standard deviation s is used (table 8).

Table 8: Uncertainty components for a typical automotive mixture

ugrav (%,rel.) ustab (%,rel.) CO 0,020 0,030CO2 0,003 0,030C3H8 0,012 0,056

The results from table 8 have been used to compute the uncertainty in the assigned value

prepprep kuU = (7)

where

22stabgravprep uuu += (8)

and k = 2. The relative uncertainty uprep has been used to compute the combined standard uncertainty of the assigned value for all mixtures.

5.2 Evaluation of results Usually all participants perform analyses on the same artefact and the key comparison reference value is calculated from the mean of the individual results. In the current comparison on gas mixtures, measurements were performed on individually prepared gas mixtures with (slightly) different concentrations. Since the pilot laboratory prepared these mixtures using the same methods and materials, the individual gravimetric values can be adopted as reference values, despite of the small differences that exist. The problem is that these small differences are of the same order of the differences found between the national metrological institutes, and thus influencing the outcome of the key comparison if it would be operated with a single reference value.

In order to evaluate the differences between the participating national metrology institutes, the value xigrav is taken as the reference value for laboratory i.

The degree of equivalence of each laboratory with respect to the reference value is given by a pair of numbers:

Di = (xi – x igrav) (9)

and Ui , its expanded uncertainty,

Ui2 = k 2(ui

2 + uiprep2) (k=2) (10)

5.3 Computation of degrees of equivalence and associated uncertainties The model of CCQM-K1 has been adapted for this key comparison. Although the uncertainty on the assigned value comes close to the reported uncertainties from some of the participants, the model of CCQM-K1 can still be used.

9

The degree of equivalence between two laboratories is defined as

[ ] [ ]jprepjiprepijiij xxxxDDD −−−=−= (11)

The uncertainty in the degree of equivalence can be expressed as

( )( )jprepiprepjprepiprepjiij xxuuuuukU ,2222222 −+++= (12)

whereby uprep is defined as given in equation (8). looking at the results, uprep can generally not be neglected compared to ui. The degree of correlation between xiprep and xprepj has been assessed and lead to the following simplification

( )222222jgravigravjiij uuuukU +++= (13)

This simplification is the result of the establishment of ustab (see also equation (8)). From the stability data and the calibration of the equipment, it becomes clear that ustab is a combined standard uncertainty from three main contributors: uncertainty of mixtures used, position of the “unknown(s)” on the curve, including lack of fit, and the repeatability of measurement. In formula, this expression reads as

2222rlofcalibrantsstab uuuu ++= (14)

whereby ucalibrants denotes the contribution from the mixtures used for calibrating the equipment, ulof denotees the contribution due to the fitting and interpolation procedure, and ur denotes the contribution due to repeatability of measurement. In this expression, ulof is by far the dominant factor. It is also a factor, that is the same for all mixtures prepared for this comparison. As a result, the covariance between xref(A) and xref(B) can be approximated by

( ) 22, calibrantslofgravjgravi uuxxu +≈ (15)

Substitution of (8) and the approximation for the covariance in (15) leads then to (13).

10

6 Results On the following pages, all results of this key comparison are presented. The tables contain the following information

Cylinder Identification code of cylinder xi Result of measurement carried out by laboratory i ui Combined standard uncertainty of xi xigrav Gravimetric value of the amount of substance fraction in the cylinder received by lab. i uigrav Combined standard uncertainty in xigrav Di Degree of equivalence of laboratory i with respect to the reference value Ui Expanded uncertainty in Di Dij Degree of equivalence between laboratories i and j

The differences between gravimetric and reported value are given relative to the gravimetric value, allowing to check directly whether the result of the laboratory is within ± 1% relative to the gravimetric value. All laboratories reported results that fulfilled this requirement for all components.

The uncertainty bars in the graphs are given with k = 2 for all laboratories, taking into consideration both the reported uncertainty from the laboratory as well as the uncertainty from gravimetry. The combined standard uncertainty of a laboratory has been computed from Ui and ki. All laboratories used a coverage factor of 2.

The results of the laboratories of SMU, IPQ, CSIR and GUM for carbon monoxide (figure 2) are all within ± 1% relative to the gravimetric value and their uncertainties overlap with the zero-axis. CHMI deviates more, and their estimate seems to be quite conservative.

For carbon dioxide, the results of the laboratories of SMU, IMGC, IPQ, CSIR and GUM are within ± 1% relative to the gravimetric value (figure 3). The uncertainty reported by IMGC and CSIR is remarkable large in comparison to the others.

The results for propane for laboratories IMGC and CHMI differ more than 1% relative from the gravimetric value (figure 4). The uncertainties stated by these two laboratories are quite large when compared to the other laboratories (SMU, IPQ, and GUM).

11

Table 9: Results for CO

Lab Cylinder xi (mmol/mol)

ui (mmol/mol)

xgrav (mmol/mol)

uprep (mmol/mol)

SMU 8509 31.030 0.047 31.016 0.011 IMGC 0634 31.035 0.011 IPQ 7381 31.06 0.055 31.033 0.011 CSIR-NML 0601 30.78 0.23 31.040 0.011 GUM 7375 31.05 0.055 31.045 0.011 CMI-CHMI 7377 30.4 2.200 31.052 0.011

Table 10: Results for CO2


Ui (mmol/mol)

xgrav (mmol/mol)

uprep (mmol/mol)

SMU 8509 132.05 0.13 132.021 0.040 IMGC 0634 131.7 2.55 132.016 0.040 IPQ 7381 131.85 0.23 131.772 0.040 CSIR-NML 0601 132.32 2.17 131.999 0.040 GUM 7375 132.35 0.055 132.143 0.040 CMI-CHMI 7377 131.931 0.040

12

Table 11: Results for propane


UI (mmol/mol)

xgrav (mmol/mol)

uprep (mmol/mol)

SMU 8509 2.090 0.0055 2.0856 0.0012 IMGC 0634 2.013 0.036 2.0870 0.0012 IPQ 7381 2.087 0.0025 2.0868 0.0012 CSIR-NML 0601 2.0873 0.0012 GUM 7375 2.087 0.0030 2.0876 0.0012 CMI-CHMI 7377 2.21 0.075 2.0881 0.0012

13

Figure 2: Results for CO

Euromet 485: Results for Carbon Monoxide

-10%

-5%

0%

5%

10%SM

U

IMG

C

IPQ

CSI

R

GU

M

CH

MI

Laboratory

rela

tive

diff

eran

ce (%

)

14

Figure 3: Results for CO2

Euromet 485 : Results for Carbon Dioxide

-5%

-4%

-3%

-2%

-1%

0%

1%

2%

3%

4%

5%SM

U

IMG

C

IPQ

CSI

R

GU

M

CH

MI

Laboratory

Rel

ativ

e di

ffer

ence

(%)

15

Figure 4: Results for C3H8

Euromet 485 : Results for Propane

-10%

-5%

0%

5%

10%SM

U

IMG

C

IPQ

CSI

R

GU

M

CH

MI

Laboratory

Rel

ativ

e di

ffer

ence

(%)

16

Table 12 : Degrees of equivalence and expanded uncertainties for CO (mmol/mol)

Lab j ⇒

⇓Lab i SMU IMGC IPQ CSIR GUM CHMI

Di Ui mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

SMU 0.014 0.096 -0.013 0.148 0.274 0.470 0.009 0.148 0.666 4.401

IMGC

IPQ 0.027 0.112 0.013 0.148 0.287 0.474 0.022 0.114 0.679 4.401

CSIR -0.260 0.460 -0.274 0.470 0.287 0.474 -0.265 0.474 0.392 4.424

GUM 0.005 0.112 -0.009 0.148 0.022 0.114 0.265 0.474 0.657 4.401

CHMI -0.652 4.400 -0.666 4.401 0.679 4.401 -0.392 4.424 -0.657 4.401

17

Table 13: Degrees of equivalence and expanded uncertainties for CO2 (mmol/mol)

Lab j ⇒


Di Ui mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

SMU 0.029 0.272 0.345 5.108 -0.049 0.540 -0.292 4.349 -0.178 0.304

IMGC -0.316 5.101 -0.345 5.108 -0.394 5.122 -0.637 6.698 -0.523 5.102

IPQ 0.078 0.470 0.049 0.540 0.394 5.122 -0.243 4.366 -0.129 0.486

CSIR 0.321 4.341 0.292 4.349 0.637 6.698 0.243 4.366 0.114 4.434

GUM 0.207 0.136 0.178 0.304 0.523 5.102 0.129 0.486 -0.114 4.343

CHMI

18

Table 14: Degrees of equivalence and expanded uncertainties for C3H8 (mmol/mol)

Lab j ⇒


Di Ui mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

Dij Uij mmol/mol

SMU 0.0044 0.011 0.078 0.073 0.004 0.013 0.005 0.013 -0.118 0.150

IMGC -0.0740 0.072 -0.078 0.073 -0.074 0.072 -0.073 0.072 -0.196 0.166

IPQ 0.0002 0.006 -0.004 0.013 0.074 0.072 0.001 0.007 -0.122 0.150

CSIR

GUM -0.0006 0.006 -0.005 0.013 0.073 0.072 -0.001 0.007 -0.122 0.150

CHMI 0.1219 0.150 0.118 0.150 0.196 0.166 0.122 0.150 0.122 0.150

19

7 Conclusions The laboratories of SMU, IPQ, CSIR and GUM have shown on all analysed components to be able to assign values for all components within within ± 1% relative to the gravimetric value. Apart from propane, this is also true for IMGC. The results of CHMI show larger deviations from the gravimetric value, which are however covered by the reported uncertainty. IMGC, CSIR and CHMI did assign a value to only 2 of the 3 major constituents.

All participants provided satisfactory descriptions of their measurements, including the measurement technique(s), calibration method, and the evaluation method for measurement uncertainty.

20

Documents

EUROMET PROJECT FINAL REPORT - EURAMET · PDF fileEUROMET PROJECT FINAL REPORT 1 ... The results of this Euromet key-comparison can now be linked to CCQM K3 ... NMi Van Swinden Laboratorium