TECHNICAL ISO/TS

SPECIFICATION 29041

First edition
2008-02-01

Gas mixtures — Gravimetric

preparation — Mastering correlations in
composition
Mélanges de gaz — Préparation gravimétrique — Maîtrise des
corrélations en composition

Reference number
ISO/TS 29041:2008(E)

© ISO 2008
ISO/TS 29041:2008(E)

PDF disclaimer
This PDF file may contain embedded typefaces. In accordance with Adobe's licensing policy, this file may be printed or viewed but
shall not be edited unless the typefaces which are embedded are licensed to and installed on the computer performing the editing. In
downloading this file, parties accept therein the responsibility of not infringing Adobe's licensing policy. The ISO Central Secretariat
accepts no liability in this area.
Adobe is a trademark of Adobe Systems Incorporated.
Details of the software products used to create this PDF file can be found in the General Info relative to the file; the PDF-creation
parameters were optimized for printing. Every care has been taken to ensure that the file is suitable for use by ISO member bodies. In
the unlikely event that a problem relating to it is found, please inform the Central Secretariat at the address given below.

COPYRIGHT PROTECTED DOCUMENT

© ISO 2008
All rights reserved. Unless otherwise specified, no part of this publication may be reproduced or utilized in any form or by any means,
electronic or mechanical, including photocopying and microfilm, without permission in writing from either ISO at the address below or
ISO's member body in the country of the requester.
ISO copyright office
Case postale 56 • CH-1211 Geneva 20
Tel. + 41 22 749 01 11
Fax + 41 22 749 09 47
E-mail copyright@[Link]
Web [Link]
Published in Switzerland

ii © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

Contents Page

Foreword............................................................................................................................................................ iv
Introduction ........................................................................................................................................................ v
1 Scope ..................................................................................................................................................... 1
2 Normative references ........................................................................................................................... 1
3 Symbols and abbreviated terms ......................................................................................................... 1
4 Mixture preparation by gravimetry...................................................................................................... 2
5 Unit conversion..................................................................................................................................... 5
6 Inclusion of purity data ........................................................................................................................ 5
7 Gas property calculation...................................................................................................................... 8
8 Summary and recommendations ...................................................................................................... 10
Annex A (normative) Generic approach to uncertainty calculation ........................................................... 11
Bibliography ..................................................................................................................................................... 14

© ISO 2008 – All rights reserved iii

ISO/TS 29041:2008(E)

Foreword
ISO (the International Organization for Standardization) is a worldwide federation of national standards bodies
(ISO member bodies). The work of preparing International Standards is normally carried out through ISO
technical committees. Each member body interested in a subject for which a technical committee has been
established has the right to be represented on that committee. International organizations, governmental and
non-governmental, in liaison with ISO, also take part in the work. ISO collaborates closely with the
International Electrotechnical Commission (IEC) on all matters of electrotechnical standardization.

International Standards are drafted in accordance with the rules given in the ISO/IEC Directives, Part 2.

The main task of technical committees is to prepare International Standards. Draft International Standards
adopted by the technical committees are circulated to the member bodies for voting. Publication as an
International Standard requires approval by at least 75 % of the member bodies casting a vote.

In other circumstances, particularly when there is an urgent market requirement for such documents, a
technical committee may decide to publish other types of document:

⎯ an ISO Publicly Available Specification (ISO/PAS) represents an agreement between technical experts in
an ISO working group and is accepted for publication if it is approved by more than 50 % of the members
of the parent committee casting a vote;

⎯ an ISO Technical Specification (ISO/TS) represents an agreement between the members of a technical
committee and is accepted for publication if it is approved by 2/3 of the members of the committee casting
a vote.

An ISO/PAS or ISO/TS is reviewed after three years in order to decide whether it will be confirmed for a
further three years, revised to become an International Standard, or withdrawn. If the ISO/PAS or ISO/TS is
confirmed, it is reviewed again after a further three years, at which time it must either be transformed into an
International Standard or be withdrawn.

Attention is drawn to the possibility that some of the elements of this document may be the subject of patent
rights. ISO shall not be held responsible for identifying any or all such patent rights.

ISO/TS 29041 was prepared by Technical Committee ISO/TC 158, Analysis of gases.

This document is not to be regarded as an “International Standard”. It is proposed for provisional application
so that information and experience of its use in practice may be gathered. Comments on the content of this
document should be sent to the ISO Central Secretariat.

iv © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

Introduction
ISO/TC 158 decided at its meeting in Prague (October 2002) to investigate the influence of possible
correlations (both extrinsic and intrinsic) on the uncertainty calculation(s) for mixture composition and gas
mixture property data. Methods should be developed for taking all existing correlations into account, and
exemplified appropriately.

This Technical Specification describes the tools needed for full accounting of correlations, and exemplifies
how these tools should be applied to practical examples. Some recommendations are given which are
intended to provide support to the decision on whether or not, and in which situations, the full calculatory
scheme as described herein should be applied in practice, and in which situations simplified approaches as
given in ISO 6142 are considered sufficient for the intended purpose.

© ISO 2008 – All rights reserved v

TECHNICAL SPECIFICATION ISO/TS 29041:2008(E)

Gas mixtures — Gravimetric preparation — Mastering

correlations in composition

1 Scope
In this Technical Specification, the gravimetric mixture preparation as given in ISO 6142 is investigated for
influences of a priori existing, as well as correlations introduced by data processing.

All calculations refer to an example which consists in the preparation of a synthetic natural gas of a target
composition as follows: 1,4 mol % N2, 1,8 mol % CO2, 9,4 mol % ethane, 3,4 mol % propane, 1 mol % n-
butane, and 83 mol % methane.

All considerations given for this example concerning mixture feasibility, choice of preparation procedure, and
weighing steps and sequences are the same as given in ISO 6142. This also applies to all estimates for basic
uncertainty sources and the purity tables of the gases used for preparation.

All calculations follow the principles, and use the tools and algorithms laid down in Annex A. For the sake of
simplicity, procedural steps such as matrix transformation, inversion or matrix calculus are not detailed each
time they are used in the calculations.

2 Normative references
The following referenced documents are indispensable for the application of this document. For dated
references, only the edition cited applies. For undated references, the latest edition of the referenced
document (including any amendments) applies.

ISO 6142, Gas analysis — Preparation of calibration gas mixtures — Gravimetric method

3 Symbols and abbreviated terms

CVi calorific value of gas component i, i = 1, …, n

CV calorific value of the gas mixture

G system of model equations describing the measurand

J Jacobian

m mass

mg mass of the gas component in the mixture cylinder, as determined by gravimetry

mm mass difference, as determined, between mixture and reference cylinder after the corresponding
filling step

mx mass difference corrected between mixture and reference cylinder after the corresponding filling
step

© ISO 2008 – All rights reserved 1

ISO/TS 29041:2008(E)

Mi molar mass of gas component i, i = 1, …, n

M molar mass of the gas mixture

Q transfer matrix

U(x) expanded uncertainty of a value x

u(x) standard uncertainty of a value x

u2(pi, pj) variance (for i = j) of a value pi, or covariance (for i ≠ j) of two values pi and pj

uB buoyancy correction (see Note)

uexp correction accounting for cylinder expansion

um uncertainty of balance indication (cumulated estimate)

uR correction for residual gas in the cylinder after evacuation

V variance/covariance matrix for a set of results or parameters

wg mass fraction of a component as determined by gravimetry

xi mole fraction of gas component i in the final mixture, i = 1, ..., n

x ipur
,k mole fraction of gas component k in pure gas i used for gravimetric preparation

NOTE ISO 6142 uses symbols with lower-case letter “u” (e.g. uB) for variables other than uncertainties, namely for
corrections made for influential factors in the gravimetric mixture production process, e.g. buoyancy. This symbol
assignment to variables is also retained for the purposes of this document. The reader should be careful not to confound
values and value uncertainties. The uncertainty of, say, uB is u(uB).

4 Mixture preparation by gravimetry

For the weighing steps and sequence as described in the example, the values for the mx and their
corresponding uncertainties (as given in Table 1) can be obtained from the raw data and uncertainty source
estimates.

Estimates for the corrections and their uncertainties are the same as in the CO in N2 example in ISO 6142.

2 © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

Table 1 — Component weighing data and their uncertainties for the gas mixture

Masses and Component

their
Vacuum n-Butane Propane Ethane CO2 N2 Methane
uncertainties

mm [g] 50,000 21 37,999 302 7,999 707 50,000 21 65,999 937 73,999 78 345,998 422
um [g] −0,011 0,157 −0,731 −0,299 0,052 0,065 0,13
uB [g] 0,007 45 0,005 66 0,001 19 0,007 45 0,009 84 0,001 1 0,051 58
uexp [g] 0,001 49 0,015 9
uR [g] 0,003 3
u(mm) [mg] 0,015 0,019 3 0,010 8 0,015 0,020 2 0,015 1 0,06
u(um) [mg] 2,3 2,3 2,3 2,3 2,3 2,3 2,3
u(uB) [mg] 0,019 0,014 0,003 0,019 0,025 0,028 0,13
u(uexp) [mg] 0,86 9,16
u(uR) [mg] 1,9
mx [g] 49,996 660 38,161 962 7,269 897 49,708 660 66,063 267 74,065 880 346,199 202
u(mx) [mg] 2,300 128 2,300 124 2,300 027 2,300 127 2,455 735 2,300 220 9,634 630

The following reasoning applies to possible correlations: it is assumed that um is a random variable governed
by a common distribution, and each realisation is a drawing from the distribution. Under this assumption, there
is no reason for making allowances for correlation(s) since the realisations are independent. The uncertainty
sources for the uB estimates are the same, namely pressure, temperature, and humidity. This causes
correlation, but the common sources are not quantified in ISO 6142. There is no reason for assuming
correlations for the uexp estimate, and uR occurs only once for methane. For mm, a clear correlation exists
since the same mass pieces are used, but their combinations are unknown (except for some cases). Usually,
these correlations are small and may be neglected in practice. Here, for demonstrating the principle of the
method they are included where obvious. The corresponding variance-covariance matrix for the data table is
given in Table 2.

Table 2 — Variance-covariance matrix for the weighing data in Table 1

V(mx) Vacuum n-Butane Propane Ethane CO2 N2 Methane

Vacuum 5,290 586 0 0 0,000 225 0,000 225 0,000 225 0

n-Butane 0 5,290 568 49 0 0 0 0 0

Propane 0 0 5,290 125 64 0 0 0 0

Ethane 0,000 225 0 0 5,290 586 0,000 225 0,000 225 0

CO2 0,000 225 0 0 0,000 225 6,030 633 04 0,000 225 0

N2 0,000 225 0 0 0,000 225 0,000 225 5,291 012 01 0

Methane 0 0 0 0 0 0 92,826 1

© ISO 2008 – All rights reserved 3

ISO/TS 29041:2008(E)

Values in Table 1 are exemplified for the N2 column: The uncertainty of the mx value for N2 is combined from
the contributing sources of uncertainty according to the usual uncertainty propagation rule. It holds

u 2 ( m x ) = u 2 ( m m ) + u 2 (u m ) + u 2 ( u B ) (1)

delivering a value of 2,300 3 mg. The corresponding variance is 5,291 012 01 mg2 and contained in the sixth
row of the N2 column. The mm values for the N2 and the methane filling step are quite different, it was
assumed that different mass pieces were used. Other common sources of uncertainty are the corrections um
and uB for which either an absence of correlation is assumed, or the correlations are negligible or unknown.
Thus, the covariance term is set to zero (seventh row of the N2 column). The same reasoning holds for the n-
butane/N2 and the propane/N2 pairs (second and third row of the N2 column).

The mm values for the initial weighing (vacuum) and the ethane and CO2 filling step are quite similar or at least
in the region of 50 g, so it can be assumed that the same mass pieces were used. For simplicity, the
covariance arising from this instance was estimated as the variance u2(mm) of the initial weighing of the
mixture cylinder containing only vacuum [u(mm) = 0,015 mg, u2(mm) = 0,000 225 mg2]. It is the same for all
three pairs (first, fourth and fifth row of the N2 column). Note that the values which appear in rows 1 to 7 in the
N2 column are repeated in columns 1 to 7 of the N2 row since variance-covariance matrices are symmetric.

From the set of equations:

mg(butane) − mvac + mbutane = 0

mg(propane) − mbutane + mpropane = 0

mg(ethane) − mpropane − methane = 0

mg(CO2) − mCO2 + methane = 0

mg(N2) − mN2 + mCO2 = 0

mg(methane) − mmethane + mN2 = 0

the transfer matrix Q is formed according to the recipes given in Annex A, and the gas masses and the
variance-covariance matrix of the mixture composition are calculated from the mx data of Table 1 and the
variance-covariance matrix in Table 2. This yields the values given in Table 3.

Table 3 — Gas masses in the mixture, their uncertainties and the variance-covariance matrix

n-Butane Propane Ethane CO2 N2 Methane

mg [g] 11,834 698 30,892 065 56,978 557 16,354 607 8,002 613 272,133 322
u(mg) [mg] 3,252 869 3,252 798 3,252 801 3,364 635 3,364 698 9,905 408

V n-Butane Propane Ethane CO2 N2 Methane

n-Butane 10,581 154 −5,290 568 5 0,000 225 0 0 −0,000 225

Propane −5,290 568 5 10,580 694 −5,290 125 6 0 0 0
Ethane 0,000 225 −5,290 125 6 10,580 712 −5,290 361 0 −0,000 225
CO2 0 0 −5,290 361 11,320 769 −6,030 408 0
N2 0 0 0 −6,030 408 11,321 195 −5,290 787 01
Methane −0,000 225 0 −0,000 225 0 −5,290 787 98,117 112

4 © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

5 Unit conversion
The above composition is given as gas masses in grams (uncertainties in milligrams) and should now be
converted to mol/mol. This will also enable the inclusion of purity data which are given in the tables in mol/mol.
Calculations are carried out using CONVERT [1] and yield the mole fractions and variance-covariance matrix
shown in Table 4. See also ISO 14912 [2] for more details on the mathematical and computational background.

Table 4 — Component mass fractions in the mixture, their uncertainties and

the variance-covariance matrix

n-Butane Propane Ethane CO2 N2 Methane

wg 0,009 971 64 0,034 308 29 0,092 799 1 0,018 198 9 0,0139 899 0,830 732 1
u(wg) 2,868 7 × 10−6 4,506 5 × 10−6 8,684 × 10−6 3,938 × 10−6 5,947 × 10−6 1,167 × 10−5

V n-Butane Propane Ethane CO2 N2 Methane

n-Butane 8,23 × 10−12 −4,02 × 10−12 2,21 × 10−12 5,76 × 10−13 4,42 × 10−13 −7,44 × 10−12
Propane −4,02 × 10−12 2,03 × 10−11 −1,20 × 10−12 2,13 × 10−12 1,63 × 10−12 −1,88 × 10−11
Ethane 2,21 × 10−12 −1,20 × 10−12 7,54 × 10−11 −4,38 × 10−12 4,00 × 10−12 −7,60 × 10−11
CO2 5,76 × 10−13 2,13 × 10−12 −4,38 × 10−12 1,55 × 10−11 −1,07 × 10−11 −3,08 × 10−12
N2 4,42 × 10−13 1,63 × 10−12 4,00 × 10−12 −1,07 × 10−11 3,54 × 10−11 −3,07 × 10−11
Methane −7,44 × 10−12 −1,88 × 10−11 −7,60 × 10−11 −3,08 × 10−12 −3,07 × 10−11 1,36 × 10−10

6 Inclusion of purity data

The purity data matrix for the gases used for mixture preparation and the corresponding uncertainties are
given in Tables 5 and 6 in µmol/mol units.

Table 5 — Purity table for the six gases used in mixture preparation
Values are given in µmol/mol

Content of n-Butane Propane Ethane CO2 N2 Methane

Ar 10
H2O 10 5 10 5 2 5
N2 30 20 15 999 983 10
CO 2 1
CO2 10 5 5 999 962 1 1
O2 10 10 10 5 2 5
H2 5 1 0.5 1
HC 500 0.5
n-Butane 999 470
Propane 999 700
C3H6 150
Ethane 999 745 10
C2H4 200
Methane 100 5 10 999 968

© ISO 2008 – All rights reserved 5

ISO/TS 29041:2008(E)

Table 6 — Uncertainties of purity data for the six gases used in mixture preparation
Values are given in µmol/mol

Content of n-Butane Propane Ethane CO2 N2 Methane

Ar 6 3 6 3 1 3

H2O 5 5 5 3 4

N2 1 0,5

CO 6 3 3 9 0,5 0,6

CO2 6 6 6 3 1 3

O2 3 0,5 0,3 0,6

H2 100 0,3

HC 100

n-Butane 28

Propane 25

C3H6 27 6

Ethane 25

C2H4 10 3 6 11

Methane 2

It can probably be assumed that these purity data were measured independently, i.e. say the CO2 content in
propane does not depend on (and is not correlated with) the CO2 content in nitrogen or any other of the pure
gases. Therefore, correlations between any of the purity data matrix components will not occur.

On the other hand, the governing model equations for the final mixture composition are

xk = ∑ xi ⋅ xipur
,k (2)
i

with k running over the 14 different gases contained in the 6 cylinders of “pure” gases used for mixture
preparation, and i running over these 6 “pure” gases. The xi are the “raw” fractions as determined in Clause 3,
and the x ipur
,k the purity data from the above table. According to this, one deals with 6 correlated (the xi) and
6 × 14 = 84 uncorrelated (the x ipur
,k ) input variables which are transformed by the model equation into
14 output variables xk. Thus, the initial variance-covariance matrix of input variables is a 90 × 90 matrix.
Fortunately, some of the x ipur pur
,k are equal to zero, and the rows and columns corresponding to these x i,k may
be deleted from the input variable variance-covariance matrix. After deleting zero rows and columns, there is
still a 48 × 48 matrix to deal with. The left part of the matrix containing the 6 × 6 correlated gas composition
data and 12 uncertainties of the x ipur
,k is shown in Figure 1.

Note that due to its dimensions:

a) only the upper left part of the matrix is shown; and

b) this part is reproduced in a graphics mode.

The matrix continues further to the right but is strictly diagonal in this part.

6 © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

NOTE The letter “E” is used for “times ten raised to the power”. The decimal point is shown in this screen capture
instead of the decimal comma used in the rest of the text.

Figure 1 — Variance-covariance for the purity data of the gases used for mixture composition

© ISO 2008 – All rights reserved 7

ISO/TS 29041:2008(E)

From the model equation as above, the final molar fractions of the 14 gas components and the transfer matrix
Q are determined. Using the latter, then the variance-covariance matrix for the 14 gas components is
calculated. This yields for the composition of the mixture (in µmol/mol) as in Table 7 and the variance-
covariance matrix as given in Figure 2 (matrix is displayed also in graphics mode, unit is (µmol/mol2).

Table 7 — Final molar fractions for the 14 gases in the mixture

Gas component Content u (content)

µmol/mol µmol/mol

Ar 0,140 0,028
H2O 5,472 2,557
N2 14 001,168 6,831
CO 0,050 0,019
CO2 18 199,778 3,984
O2 5,643 2,563
H2 1,320 0,571
HC 4,993 0,997
n-Butane 9 966,352 3,036
Propane 34 297,998 4,606
C3H6 5,146 0,858
Ethane 92 783,754 10,320
C2H4 18,560 2,320
Methane 830 709,593 14,826

7 Gas property calculation

From the final composition given in Table 7 above, gas properties can be calculated using the model
equations and pure gas property values given in ISO 6976 [3]. It may be preferable first to calculate the CV on
a molar basis, and the molar mass of the mixture. Both values depend on the gas composition (and are thus
correlated) and the properties of the pure gases. The latter are (still) given without uncertainties in ISO 6976,
so they can be considered as parameters. The problem has 14 input and 2 output variables, and with the
transfer matrix calculated from the model Equations (3) and (4),

CV = ∑ xi ⋅ CVi (3)
i

M = ∑ xi ⋅ M i (4)
i

the results given in Table 8 are obtained.

8 © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

NOTE The decimal point is shown in this screen capture instead of the decimal comma used in the rest of the text.

Figure 2 — Variance-covariance matrix of the final mixture composition data unit is (µmol/mol2)

© ISO 2008 – All rights reserved 9

ISO/TS 29041:2008(E)

Table 8 — Calorific value and molar mass of the gas mixture

Value [unit a] Value u(value) U(value)

correlation without correlation

CV on a molar basis [kJ/mol] 990,503 8 0,016 7 0,025 3

CV on mass basis [MJ/kg] 51,048 0 0,000 6 0,001 3
CV on volume basis [MJ/m3] 41,890 2 0,000 7 0,001 1
Molar mass [kg/mol] 19,403 4 0,000 4 0,000 6
a kJ/mol = 103 kg·m2 mol−1 s−2; MJ/kg = 106 m2 mol−1 s−2; MJ/m3 = 106 kg m−1 mol−1 s−2.

Results are given for

a) the case where correlations are fully taken into account, and

b) the case where correlations are neglected (for comparison reasons).

It is obvious that correlations have a certain and significant impact on the uncertainties of the final results.
Here, result uncertainties are considerably reduced. From the CV on a molar basis and the molar mass of the
mixture, the CV on a mass basis can be calculated and, finally, also the CV on a volumetric basis. According
to ISO 6976, the latter is a transformation of the CV on a molar basis using a transformation coefficient
calculated from state conditions. Consequently, the CV on a molar basis and its uncertainty are only scaled by
a corresponding scaling factor, no additional correlation must be taken into account.

8 Summary and recommendations

Two major conclusions can obviously be drawn from the above.

1) Correlations have a certain and significant impact on the uncertainties of the final results. Resulting
uncertainties are considerably reduced in the example discussed here. Speciality gas producers and
suppliers, claiming high-level accuracy or product applications in the field of establishment and
dissemination of metrological traceability, are recommended to fully account for correlations and
include them in uncertainty budgets. Whether a reduction by a factor of 1,5 to 2,2 of an uncertainty,
which is already in the region of 10−5 to 10−4 (relative), really matters with respect to the intended
use of the gas mixtures (“fit-for-purpose” considerations) shall only be decided for each and every
particular case, and may be challenged.

2) The calculatory efforts considerably increase when correlations are taken into account. This
additional effort should not be justified in all cases. In particular, it seems problematic to recommend
accounting for correlations in commercial environments where speciality gases are produced as
stand-alone items tailored with respect to the needs of each individual customer. On the other hand,
in the cases where pre-defined product lines are fabricated in larger batches, and (more or less)
fixed SOPs are followed, uncertainty budget calculations including covariances may be both
justifiable and feasible.

10 © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

Annex A
(normative)

Generic approach to uncertainty calculation

A.1 Introduction
In accordance with basic guidelines to the expression of measurement uncertainty, namely the GUM [4] and
the EURACHEM Guide [5], the procedure can be sub-divided into the following four basic steps:

1) definition/specification of the measurand;

2) identification of uncertainty sources;

3) quantification of uncertainty contributions;

4) calculation of the combined standard uncertainty.

This annex will not deal with step 3, the quantification of uncertainty contributions, since this is described
elsewhere in full detail, and methods for generating sensible uncertainty estimates may vary not only between
different methods and applications, but also from implementation to implementation in different laboratories.
All other steps are described below.

A.2 Definition of the measurand

Throughout this annex, definition of the measurand is meant in a more metrological sense and refers to the
establishment of operational relationships, a correct specification of the measurand (e.g. the establishment of
conditions under which a specific compound is extracted from a matrix or separated from other compounds, or
the attribution of a chromatographic peak to a specific compound) is assumed. Definition of the measurand
includes the (formal) definition of what are the output variables of an analytical procedure or a calculation, the
definition of all input variables which the output variables depend on, and the establishment of a suitable
relationship describing how the output variables depend on the input variables, i.e. the model. This model may
be deduced from physical or chemical laws or may reflect empirically established relationships, and it may
appear in the form of a single equation, a series of equations which describe, e.g. intermediate results and are
applied consecutively, a system of equations, or even a network of modules each of which consists of one or
several equations and may even include purely numerical procedures (i.e. iterative algorithms reflected in a
computer programme).

Let the vector z = (z1, ..., zk)T be the vector of the k output variables, and p = (p1, ..., pn)T the vector of the n
input variables which the output variables depend on. The model describes the relationship between output
and input variables via a system of equations G such that, in the most general case, it holds

G(z, p) = 0 (A.1)

© ISO 2008 – All rights reserved 11

ISO/TS 29041:2008(E)

A.3 Identification of uncertainty sources

Obviously, all input variables included in the model will contribute to the uncertainties of the output variables.
However, some of the input variables may depend on other variables, e.g. environmental parameters such as
pressure, temperature or air humidity. If not accounted for in the model so far, these additional parameters
should be included at this stage. Let us denote them as p′, and the extended system of equations as G′. The
final model is

G′(z, p, p′) = 0 (A.2)

and takes into account all know/identified sources of uncertainty. Since there is no need for distinguishing
between p and p′ in the sense of their mathematical treatment, and for the sake of simplicity, in all calculations
below, the model will be used in the form of Equation (A.1).

The input parameter uncertainties and possible covariances between a certain number of them are combined
in the variance-covariance matrix V(p) of the input variables:

V(p) = [u2(pi,pj)] = u2(p1,p1) u2(p1,p2)... u2(p1,pn)

u2(p2,p1) u2(p2,p2)... u2(p2,pn)
. . .
. . .
u2(pj,p1)... u2(pj,pj)... u2(pj,pn)
. . .
. . .
u2(p n,p1) u2(pn,p2)... u2(p n,pn)

A.4 Calculation of the combined standard uncertainties

A.4.1 General

The combined standard uncertainty is calculated by uncertainty propagation from the model and the matrix V
in three steps.

A.4.2 Step 1: Calculation of the Jacobians

Two Jacobians are formed: one containing the first derivatives of the model equations with respect to the
output variables, the other containing the first derivatives of the model equations with respect to the input
variables. This yields

J(z) = (∂Gi /∂zj), and (A.3)

J(p) = (∂Gi /∂pj) (A.4)

The way of generating the corresponding elements of the two Jacobians depends on the complexity of the
model. If the models consist of one or a system of several equations (what is the case throughout the example
considered in this Technical Specification), expressions for the elements can be derived in analytical form. In
the cases where the model includes numerical procedures, the only way for generating the elements of the
Jacobians is numerical differentiation. However, numerical differentiation is a general tool and may also be
helpful in the cases where analytical expressions might be obtained in principle, but seem to be too
complicated for comfortable handling.

12 © ISO 2008 – All rights reserved

 ISO/TS 29041:2008(E)

A.4.3 Step 2: Formation of the transfer matrix

The transfer matrix Q is formed from the two Jacobians according to

Q = J−1(z) ⋅ J(p) (A.5)

A.4.4 Step 3: Calculation of the ouput variance-covariance matrix

The variance-covariance matrix V(z) of the output variables is calculated from V(p) and the transfer matrix
according to

V(z) = Q ⋅ V(p) ⋅ QT (A.6)

A.5 Simplifications
If the model can be expressed in an explicit form according to

z = G(p) (A.7)

then [since J −1(z) = E with E being the unitary matrix] it holds

V(z) = J(p) ⋅ V(p) ⋅ J(p)T (A.8)

If there is only one output variable, z becomes a scalar, and the matrix J a vector, i.e. J = (∂z/∂p1, ..., ∂z/∂pn)T.
Then, the expression for the combined uncertainty of the output variable z degenerates to the well-known
form:

∂x ∂x
u 2(z) = ∑ ∂pi ⋅ ∂p j ⋅ u 2 ( pi , p j ) (A.9)
i, j

The equations as given above have accordingly been applied to all calculations throughout this document.

© ISO 2008 – All rights reserved 13

ISO/TS 29041:2008(E)

Bibliography

[1] CONVERT Software supporting ISO 14912, available from Normenasschuß Materialprüfung (NMP) im
DIN Deutsches Institut für Normung e.V., D - 10772 Berlin, Germany

[2] ISO 14912:2003, Gas analysis — Conversion of gas mixture composition data

[3] ISO 6976:1995, Natural gas — Calculation of calorific values, density, relative density and Wobbe
index from composition

[4] Guide to the expression of uncertainty in measurement (GUM). BIPM, IEC, IFCC, ISO, IUPAC, IUPAP,
OIML, 19931)

[5] QUAM:2000.P1, Quantifying Uncertainty in Analytical Measurement, EURACHEM/CITAC Guide, 2nd

edition, 2000

1) Corrected and reprinted in 1995.

14 © ISO 2008 – All rights reserved

ISO/TS 29041:2008(E)

ICS 71.040.40
Price based on 14 pages

© ISO 2008 – All rights reserved