Professional Documents
Culture Documents
Iso 11929 4 2022
Iso 11929 4 2022
ISO 11929‑4:2022
National foreword
This British Standard is the UK implementation of ISO 11929‑4:2022. It
supersedes BS ISO 11929‑4:2020, which is withdrawn.
The UK participation in its preparation was entrusted to Technical
Committee NCE/2, Radiation protection and measurement.
A list of organizations represented on this committee can be obtained on
request to its committee manager.
Contractual and legal considerations
This publication has been prepared in good faith, however no
representation, warranty, assurance or undertaking (express or
implied) is or will be made, and no responsibility or liability is or will be
accepted by BSI in relation to the adequacy, accuracy, completeness or
reasonableness of this publication. All and any such responsibility and
liability is expressly disclaimed to the full extent permitted by the law.
This publication is provided as is, and is to be used at the
recipient’s own risk.
The recipient is advised to consider seeking professional guidance with
respect to its use of this publication.
This publication is not intended to constitute a contract. Users are
responsible for its correct application.
© The British Standards Institution 2022
Published by BSI Standards Limited 2022
ISBN 978 0 539 20954 9
ICS 17.240
Compliance with a British Standard cannot confer immunity from
legal obligations.
This British Standard was published under the authority of the
Standards Policy and Strategy Committee on 31 August 2022.
Third edition
2022-07
Reference number
ISO 11929-4:2022(E)
© ISO 2022
BS ISO 11929‑4:2022
ISO 11929-4:2022(E)
Contents Page
Foreword.......................................................................................................................................................................................................................................vii
Introduction............................................................................................................................................................................................................................viii
1 Scope.................................................................................................................................................................................................................................. 1
2 Normative references...................................................................................................................................................................................... 2
3 Terms and definitions..................................................................................................................................................................................... 2
4 Quantities and symbols................................................................................................................................................................................. 3
5 Summary of this document........................................................................................................................................................................ 5
5.1 Procedures according to ISO 11929 (all parts)........................................................................................................... 5
5.2 Survey on the examples................................................................................................................................................................... 5
5.3 General stipulations............................................................................................................................................................................ 8
6 Counting measurements with small or moderate uncertainties...................................................................... 9
6.1 Definition of the task and general aspects..................................................................................................................... 9
6.2 Model of evaluation and standard uncertainty.......................................................................................................... 9
6.3 Available information, input data, and specifications.......................................................................................... 9
6.4 Evaluation of the measurement and characteristic limits according to ISO 11929-1........... 10
6.4.1 Background effect........................................................................................................................................................... 10
6.4.2 Primary result and its associated standard uncertainty............................................................. 10
6.4.3 Standard uncertainty as a function of an assumed true value............................................... 10
6.4.4 Decision threshold.......................................................................................................................................................... 10
6.4.5 Detection limit.................................................................................................................................................................... 11
6.4.6 Limits of coverage intervals................................................................................................................................... 11
6.4.7 The best estimate and its associated standard uncertainty..................................................... 11
6.5 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2...................................... 11
6.6 Assessment and explanations................................................................................................................................................. 13
7 Counting measurement with small count numbers..................................................................................................... 14
7.1 Definition of the task and general aspects.................................................................................................................. 14
7.2 Model of evaluation and standard uncertainty....................................................................................................... 14
7.3 Available information, input data, and specifications....................................................................................... 14
7.4 Evaluation of the measurement and characteristic limits according to ISO 11929-1........... 15
7.4.1 Background effect........................................................................................................................................................... 15
7.4.2 Primary result and its associated standard uncertainty............................................................. 15
7.4.3 Standard uncertainty as a function of an assumed true value............................................... 16
7.4.4 Decision threshold.......................................................................................................................................................... 16
7.4.5 Detection limit.................................................................................................................................................................... 16
7.4.6 Limits of coverage intervals................................................................................................................................... 17
7.4.7 The best estimate and its associated standard uncertainty..................................................... 17
7.5 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2...................................... 17
7.6 Assessment and explanations................................................................................................................................................. 18
7.7 An alternative example of a measurement with small count numbers............................................. 19
7.7.1 General...................................................................................................................................................................................... 19
7.7.2 Background effect........................................................................................................................................................... 20
7.7.3 Primary result and its associated standard uncertainty............................................................. 20
7.7.4 Standard uncertainty as a function of an assumed true value............................................... 20
7.7.5 Decision threshold.......................................................................................................................................................... 20
7.7.6 Detection limit.................................................................................................................................................................... 21
7.7.7 Limits of coverage intervals................................................................................................................................... 21
7.7.8 The best estimate and its associated standard uncertainty..................................................... 21
7.8 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2...................................... 22
7.9 Assessment of the alternative example and explanations............................................................................ 23
8 Counting measurements with large uncertainties in the numerator of the
calibration factor............................................................................................................................................................................................... 23
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.
The procedures used to develop this document and those intended for its further maintenance are
described in the ISO/IEC Directives, Part 1. In particular the different approval criteria needed for the
different types of ISO documents should be noted. This document was drafted in accordance with the
editorial rules of the ISO/IEC Directives, Part 2 (see www.iso.org/directives).
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. Details of
any patent rights identified during the development of the document will be in the Introduction and/or
on the ISO list of patent declarations received (see www.iso.org/patents).
Any trade name used in this document is information given for the convenience of users and does not
constitute an endorsement.
For an explanation on the voluntary nature of standards, the meaning of ISO specific terms and
expressions related to conformity assessment, as well as information about ISO's adherence to the
World Trade Organization (WTO) principles in the Technical Barriers to Trade (TBT) see the following
URL: www.iso.org/iso/foreword.html.
This document was prepared by Technical Committee ISO/TC 85, Nuclear energy, nuclear technologies,
and radiological protection, Subcommittee SC 2, Radiological protection.
This third edition of ISO 11929-4 cancels and replaces the second edition (ISO 11929-4:2020), of which
it constitutes a minor revision.
The main changes are as follows:
— Editorial changes were done in text and formulae
A list of all parts of ISO 11929 can be found on the ISO website.
Any feedback or questions on this document should be directed to the user’s national standards body. A
complete listing of these bodies can be found at www.iso.org/members.html.
Introduction
Measurement uncertainties and characteristic values, i.e. characteristic limits such as the decision
threshold, the detection limit and limits of the coverage interval for measurements as well as the best
estimate and its associated standard measurement uncertainty, are of importance in metrology, in
general, and for radiological protection, in particular. The quantification of the uncertainty associated
with a measurement result provides a basis for the trust an individual can have in a measurement
result.
NOTE 1 Conformity with regulatory limits, constraints or reference values can only be demonstrated taking
into account and quantifying all sources of uncertainty. Characteristic limits provide – in the end – the basis for
deciding accepting results under uncertainty.
ISO 11929 (all parts) provides characteristic values of a non-negative measurand of ionizing radiation.
It is applicable for a wide range of measuring methods extending beyond measurements of ionizing
radiation.
The limits to be provided according to ISO 11929 (all parts) for specified probabilities of wrong decisions
allow detection possibilities to be assessed for a measurand and for the physical effect quantified by
this measurand as follows:
— the “decision threshold” allows a decision to be made on whether or not the physical effect quantified
by the measurand is present;
— the “detection limit” indicates the smallest true quantity value of the measurand that can still be
detected with the applied measurement procedure; this gives and allows for a decision on whether
or not the measurement procedure satisfies the requirements and is therefore suitable for the
intended measurement purpose;
— the “limits of the coverage interval” enclose, in the case of the physical effect recognized as present,
a coverage interval containing the true quantity value of the measurand with a specified probability.
Hereinafter, the limits mentioned are jointly called “characteristic limits”.
NOTE 2 According to ISO/IEC Guide 99 updated by JCGM 200:2012, the term “coverage interval” is used
here instead of “confidence interval” in order to distinguish the wording of Bayesian terminology from that of
conventional statistics.
All the characteristic values are based on Bayesian statistics and on the ISO/IEC Guide 98-3 as well as
on the ISO/IEC Guide 98-3:2008/Suppl.1 and ISO/IEC Guide 98-3:2008/Suppl.2. As explained in detail in
ISO 11929-2, the characteristic values are mathematically defined by means of moments and quantiles
of probability distributions of the possible measurand values.
Since measurement uncertainty plays an important role in all parts of ISO 11929, the evaluation of
measurements and the treatment of measurement uncertainties are carried out by means of the general
procedures according to the ISO/IEC Guide 98-3 and to the ISO/IEC Guide 98-3:2008/Suppl.1; see also
References [21] to [25]. This enables the strict separation of the evaluation of the measurements,
on the one hand, and the provision and calculation of the characteristic values, on the other hand.
ISO 11929 (all parts) makes use of a theory of uncertainty in measurement[26] to [28] based on Bayesian
statistics (e.g. References [29] to [36]) in order to allow taking into account also those uncertainties
that cannot be derived from repeated or counting measurements. The latter uncertainties cannot be
handled by frequentist statistics.
Because of developments in metrology concerning measurement uncertainty, laid down in the
ISO/IEC Guide 98-3, ISO 11929:2010 was drawn up on the basis of ISO/IEC Guide 98-3, but using
Bayesian statistics and the Bayesian theory of measurement uncertainty. This theory provides a
Bayesian foundation for the ISO/IEC Guide 98-3. Moreover, ISO 11929:2010 was based on the definitions
of the characteristic values[21], the standard proposal[22], and the introducing article[23]. It unified
and replaced all earlier parts of ISO 11929 and was applicable not only to a large variety of particular
measurements of ionizing radiation but also, in analogy, to other measurement procedures. Some
explanatory material about the basics of ISO 11929 (all parts), in general, and its application in has been
published elsewhere[42][43].
Since the ISO/IEC Guide 98-3:2008/Suppl.1 has been published, the Monte Carlo method has been
used to deal comprehensively with a more general treatment of measurement uncertainty in complex
measurement evaluations. This development provided an incentive for writing a corresponding
Monte Carlo supplement[24] to ISO 11929:2010. The revised ISO 11929 (all parts) is also essentially
founded on Bayesian statistics and can serve as a bridge between documents ISO 11929:2010 and
the ISO/IEC Guide 98-3:2008/Suppl.1. Moreover, more general definitions of the characteristic values
(ISO 11929-2) and the Monte Carlo computation of the characteristic values make it possible to go a step
beyond the present state of standardization laid down in ISO 11929:2010 since probability distributions
rather than uncertainties can be propagated. It is thus more comprehensive and extending the range of
applications.
The revised ISO 11929 (all parts), moreover, is more explicit on the calculation of the characteristic
values. Reference [25] gives a survey on the basis of the revision. Further, in ISO 11929-3, it gives
detailed advice how to calculate characteristic values in the case of multivariate measurements using
unfolding methods. For such measurements, the ISO/IEC Guide 98-3:2008/Suppl.2 provides the basis of
the uncertainty evaluation.
Formulae are provided for the calculation of the characteristic values of an ionizing radiation measurand
via the “standard measurement uncertainty” of the measurand (hereinafter “standard uncertainty”)
derived according to the ISO/IEC Guide 98-3 as well as via probability density functions (PDFs) of the
measurand derived on the basis of the ISO/IEC Guide 98-3:2008/Suppl.1. The standard uncertainties
or probability density functions take into account the uncertainties of the actual measurement as well
as those of sample treatment, calibration of the measuring system and other influences. The latter
uncertainties are assumed to be known from previous investigations.
1 Scope
This document specifies a procedure, in the field of ionizing radiation metrology, for the calculation of
the “decision threshold”, the “detection limit” and the “limits of the coverage interval” for a non‑negative
ionizing radiation measurand when counting measurements with preselection of time or counts are
carried out. The measurand results from a gross count rate and a background count rate as well as from
further quantities on the basis of a model of the evaluation. In particular, the measurand can be the net
count rate as the difference of the gross count rate and the background count rate, or the net activity of
a sample. It can also be influenced by calibration of the measuring system, by sample treatment and by
other factors.
ISO 11929 has been divided into four parts covering elementary applications in ISO 11929-1, advanced
applications on the basis of the ISO/IEC Guide 98-3:2008/Suppl.1 in ISO 11929-2, applications to
unfolding methods in ISO 11929-3, and guidance to the application in ISO 11929-4.
ISO 11929-1 covers basic applications of counting measurements frequently used in the field of ionizing
radiation metrology. It is restricted to applications for which the uncertainties can be evaluated on the
basis of the ISO/IEC Guide 98-3 (JCGM 2008). In ISO 11929-1:2019, Annex A the special case of repeated
counting measurements with random influences and in ISO 11929-1:2019, Annex B, measurements
with linear analogous ratemeters are covered.
ISO 11929-2 extends ISO 11929-1 to the evaluation of measurement uncertainties according to the
ISO/IEC Guide 98-3:2008/Suppl.1. ISO 11929-2 also presents some explanatory notes regarding general
aspects of counting measurements and Bayesian statistics in measurements.
ISO 11929-3 deals with the evaluation of measurements using unfolding methods and counting
spectrometric multi-channel measurements if evaluated by unfolding methods, in particular, alpha- and
gamma-spectrometric measurements. Further, it provides some advice how to deal with correlations
and covariances.
ISO 11929-4 gives guidance to the application of ISO 11929 (all parts), summarizing shortly the general
procedure and then presenting a wide range of numerical examples. The examples cover elementary
applications according to ISO 11929-1 and ISO 11929-2.
The ISO 11929 (all parts) also applies analogously to other measurements of any kind if a similar model
of the evaluation is involved. Further practical examples can be found in other International Standards,
for example, see References [1 to 20].
NOTE A code system, named UncertRadio, is available allowing for calculations according to ISO 11929-1
to ISO 11929-3. UncertRadio[40][41] can be downloaded for free from https://w ww.thuenen.de/en/f i/f ields-of
-activity/marine-environment/coordination-centre-of-radioactivity/uncertradio/. The download contains a
setup installation file that copies all files and folders into a folder specified by the user. After installation one
has to add information to the PATH of Windows as indicated by a pop‑up window during installation. English
language can be chosen and extensive “help” information is available. Another tool is the package ‘metRology’[44]
which is available for programming in R. It contains the two R functions ‘uncert’ and ‘uncertMC’ which perform
the GUM-conform uncertainty propagation, either analytically or by the Monte Carlo method, respectively.
Covariances/correlations of input quantities are included. Applying these two functions within iterations for
decision threshold and the detection limit calculations simplifies the programming effort significantly. It is also
possible to implement this document in a spreadsheet containing a Monte Carlo add-in or into other commercial
mathematics software.
2 Normative references
The following documents are referred to in the text in such a way that some or all of their content
constitutes requirements 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 3534-1, Statistics — Vocabulary and symbols — Part 1: General statistical terms and terms used in
probability
ISO 11929-1, Determination of the characteristic limits (decision threshold, detection limit and limits of
the coverage interval) for measurements of ionizing radiation — Fundamentals and application — Part 1:
Elementary applications
ISO 11929-2, Determination of the characteristic limits (decision threshold, detection limit and limits of
the coverage interval) for measurements of ionizing radiation — Fundamentals and application — Part 2:
Advanced applications
ISO 80000-1, Quantities and units — Part 1: General
ISO 80000-10, Quantities and units — Part 10: Atomic and nuclear physics
ISO/IEC Guide 98-3, Uncertainty of measurement — Part 3: Guide to the expression of uncertainty in
measurement (GUM:1995)
ISO/IEC Guide 98-3:2008/Suppl.1, Evaluation of measurement data — Supplement 1 to the “Guide to the
expression of uncertainty in measurement” — a Propagation of distributions using a Monte Carlo method,
JCGM 101:2008
ISO/IEC Guide 98-3:2008/Suppl.2, Evaluation of measurement data — Supplement 2 to the “Guide to the
expression of uncertainty in measurement” — Extension to any number of output quantities, JCGM 102:2011
ISO/IEC Guide 99, International vocabulary of metrology — Basic and general concepts and associated
terms (VIM)
IEC/TR 62461, Radiation protection instrumentation — Determination of uncertainty in measurement,
Ed. 2.0, IEC 23.1.2015
ISO and IEC maintain terminological databases for use in standardization at the following addresses:
— ISO Online browsing platform available at https://w ww.iso.org/obp
— IEC Electropedia available at https://w ww.electropedia.org/
m number of input quantities, also used for the mass of the test sample
G model of evaluation connecting the input quantities with the measurand: Y = G( X1 ,..., Xm )
xi measured value of the input quantity Xi , estimate of the true value xi of Xi
xi possible or assumed true values of the input quantity Xi
u( xi ) standard uncertainty of the input quantity Xi associated with the estimate Xi
∆xi width of the region of the possible values of the input quantity Xi
urel ( w ) relative standard uncertainty of a quantity W associated with the estimate w
y possible or assumed true values of the measurand; if the physical effect of interest is
not present, then y = 0 otherwise, y > 0
y , y lower and upper limit of the symmetric coverage interval, respectively, of the measurand
y< , y > lower and upper limit of the shortest coverage interval, respectively, of the measurand
f ( y y ) posterior probability density function (PDF) for a true value y given the estimate y
taking NOT into account the condition that the measurand Y is non-negative
f ( y y ,Y ≥ 0 ) posterior probability density function (PDF) for a true value y given the estimate y
taking into account the condition that the measurand Y is non-negative
f ( y y = 0 ) predictive probability density function (PDF) to obtain a measured value y if a true
value y = 0 of the measurand Y is assumed
R( x ; xL , xU ) Rectangular distribution as the probability density function (PDF) of the true value
x of a quantity X given the lower and upper limits xL and xU
T( x ;a , b ) Triangular distribution as the probability density function (PDF) of the true value x
of a quantity X being the sum of two quantities, A and B being assigned rectangular
probability distributions with the lower and upper limits aL and bL respectively aU
and bL and with a = aL + bL and b = aU + bU .
H( x ) 0 x < 0
Heaviside step function: H ( x ) = for
1 x ≥ 0
rg , r0 estimate of the gross count rate and of the background count rate, respectively
rn estimate of the net count rate
ni number of counted pulses obtained from the measurement of the count rate Ri
ng , n0 number of counted pulses of the gross measurement and of the background meas-
urement, respectively
ti measurement duration of the measurement of the count rate Ri
t g , t0 duration of the gross and the background measurement, respectively
ri estimate of the count rate Ri
A activity
a estimate of the activity A
w calibration factor
am activity per unit mass
as activity per unit surface
A criterion whether ISO 11929-1 or ISO 11929-2 is to be preferred in practice is given in IEC/TR 62461.
It recommends “the results of both methods should be given in order to display their difference. When
the 95 % coverage intervals of the Monte Carlo method and of the analytical method do not deviate by
more than 10 %, then the analytical one may be used for the uncertainty determination in similar cases,
i.e. a similar model function and similar or smaller values of the uncertainty of the input quantities”. See
also ISO 11929-1:2019, 5.3.
The examples, the models, and the data are exemplary and not normative.
In ISO 11929-1, a quite general model is specified as follows:
x6 ⋅ x 8 ⋅⋅⋅
y = ( x1 − x 2 ⋅ x3 − x 4 ) ⋅ = ( x1 − x2 x3 − x 4 ) ⋅ w (1)
x5 ⋅ x7 ⋅⋅⋅
with
x6 ⋅ x 8 ⋅⋅⋅
w= (2)
x5 ⋅ x7 ⋅⋅⋅
where x1 = rg is the gross count rate and x2 = r0 is the background count rate. The other input
quantities, x i , are calibration, correction or influence quantities, or conversion factors, for instance the
emission or response probability. In particular, x3 is a shielding factor and x 4 an additional background
correction quantity. If some of the input quantities are not involved, x i = 1 ( i = 3; i > 4) , x 4 = 0 and
u( x i ) = 0 shall be set for them.
NOTE 2 Generally, physical quantities are denoted by uppercase letters and have to be carefully distinguished
from their values, denoted by the corresponding lowercase letters. However, for sake of simplicity no distinction
is made between uppercase letters for quantities and lowercase letters for values throughout this document.
Only lowercase letters are used because the examples deal with values rather than with quantities.
In this document, the simplest form of Formula (1) used for most examples for ease of understanding is:
y = ( x1 − x2 ) ⋅ w = ( rg − r0 ) ⋅ w (3)
This simple model according to Formula (3) of a counting measurement with small or moderate
relative uncertainties below 25 % such as in the majority of laboratory experiments is dealt with in
Clause 6.
Small count numbers provide a particular problem in counting measurements since the information
about the associated count rates is scarce. An example describing the measurement of alpha-particles
with low count numbers according to ISO 11929-2:2019, A.1 is given in Clause 7.
Clause 8 gives an example of a counting measurement with uncertain counting geometry where the
uncertainties of the calibration factor are generally large and where, in consequence, a dominating
uncertainty exists for a quantity in the numerator of the calibration factor of Formula (2).
Clause 9 gives an example of a counting measurement in form of a wipe test where the uncertainties
are generally large and where, in particular, a dominating uncertainty exists for a quantity in the
denominator of the calibration factor of Formula (2).
Clause 10 is an example where large uncertainties occur due to shielding of the background as in
the case of measurements with a portal monitor.
Clause 11 exemplifies also the general model according to Formula (1) by dealing with the clearance
measurement of bulk samples of concrete with a special counting device and with consideration of
natural radioactivity in the background.
Clause 12 exemplifies the general model according to Formula (1) by dealing with the γ-spectrometry
of 235U with an interference from 226Ra.
Clause 13 deals with the case of measurements of ambient dose rates with a measuring device of
unknown functionality and algorithm, i.e. a so-called black box measurement according to A.4
of ISO 11929-1 where no technical details about the measuring procedure are known and only readings
of a device are available.
The particular cases of counting measurements with unknown random influences from the sample
treatment according to ISO 11929-1:2019, A.2 are dealt with in Clause 14.
The case of counting measurements with known random influences from the sample treatment
according to of ISO 11929-1:2019, A.3 is dealt with in Clause 15.
The examples in Clause 16 and Clause 17 demonstrate the applicability of ISO 11929-1 and ISO 11929-2
to non-counting measurements. The case of an active personal dosemeter is treated according
ISO 11929-1 and ISO 11929-2 in Clause 16. The particlular feature of this example is that the standard
uncertainty associated with the gross quantity is constant, thereby allowing to calculate the standard
uncertainty as a function of an assumed true value of the measurand.
The case of a dose rate measurement using a neutron area monitor treated according to ISO 11929-1
and ISO 11929-2 is dealt with in Clause 17. In this example it is known that the measurement is a
counting measurement. However, the actual gross and background count numbers are unknown. In
spite of that, the standard uncertainty as a function of an assumed true value of the measurand can be
calculated using a general feature of the relative uncertainty of the gross indication.
This case is characterized by y ∗ < y and y # < yr . A primary result above the decision threshold was
obtained. One decides to conclude that an effect of the sample was recognized. The detection limit is
lower than the guideline value. One concludes that the measurement method is suitable for the
measurement purpose.
See Clauses 6, 7, 8, 9, 10, 11, 12, 13, 15, 16, and 17 for examples.
A frequent misunderstanding results from comparison of the primary measurement result with the
detection limit. Such a comparison is not according to the stipulations of ISO 11929 (all parts). The only
characteristic limit to be compared with the primary measurement result is the decision threshold, not
the detection limit. If the primary measurement result exceeds the decision threshold one decides to
conclude that an effect of the sample was recognized.
This decision to be made does not depend on whether the primary measurement result is below or
above the detection limit. The only purpose of the detection limit is to characterize the quality of the
measurement procedure by comparison with a guideline value. If the detection limit is lower than the
guideline value, one concludes that the measurement method is suitable for the measurement purpose.
This case is characterized by y < y ∗ and y # < yr . A result below the decision threshold was obtained.
One decides to conclude that no effect of the sample was recognized, though it might be present. The
detection limit is lower than the guideline value. One concludes that the measurement method is
suitable for the measurement purpose.
This case is characterized by y ∗ < y and yr < y # . A result above the decision threshold was obtained.
One decides to conclude that an effect of the sample was recognized. However, the detection limit is
higher than the guideline value. Consequently, the measurement method is not suitable for the
measurement purpose in principle.
Nevertheless, y > y ∗ is a primary measurement result with its associated standard uncertainty, which
needs no justification regarding to the suitability of the procedure.
See Clause 14 for an example.
This case is characterized by y < y ∗ and yr < y # . A primary result below the decision threshold was
obtained. One decides to conclude that no effect of the sample was recognized. The detection limit is
higher than the guideline value. One concludes that the measurement method is not suitable for the
measurement purpose and does not satisfy the measurement purpose. In total, the procedure is not
suitable and the measurement was not successful.
There is no example of such a pathological case in this document.
NOTE 4 In this document, the guideline value, yr , in comparison with the detection limit, y# , simply serves
to demonstrate the decision whether or not the measuring procedure is suitable for the measurement purpose.
This document does not provide stipulations whether or not a measurement result, y , conforms to requirement
from legal or other sources. Such assessments are beyond the scope of this document. An application of
ISO 11929-1 to such conformity assessments can be found elsewhere[45].
ng n0
a= − · w = ( rg − r0 ) · w = rn · w (4)
t
g t0
The standard uncertainty u ( a ) associated with the primary result a is calculated by:
ng n
u2 ( a ) = rn2 ⋅ u2 ( w ) + w 2 ⋅ u2 ( rn ) = w 2 ⋅ + 0 + a2 ⋅ urel
2
( w ) (5)
t2 t2
g 0
The PDFs of the calibration factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a detailed description.
The background effect of the sample has been measured. The number of the counted pulses is
n0 = 73 150 . This gives a count rate of
n0
r0 = = 6 , 096 s −1 (6)
t0
The standard uncertainty of the count rate is given by:
n0
u ( r0 ) = = 0 , 022 5 s −1 (7)
t 02
The activity of the sample is calculated by using the Formula (4). This relation depends on the calibration
factor w, which needs to be evaluated first, as well as its standard uncertainty u ( w ) .
w = 4 , 1 Bq ⋅ s , u ( w ) = 0 , 6 Bq ⋅ s , urel ( w ) = 0 , 146 (8)
ng
a = rn ⋅ w = − r0 ⋅ w = 49 , 05 Bq (9)
t
g
with the standard uncertainty
ng n
u ( a ) = rn2 ⋅ u2 ( w ) + w 2 ⋅ + 0 = 7 , 20 Bq (10)
t2 t2
g 0
The standard uncertainty u ( a ) as a function of the true value a is needed to calculate the decision
n
threshold and the detection limit. For a true value a = rn w = rg − 0 w of the measurand, one expects
t0
a n
a gross count rate rg = + 0 and with Formula (5) one obtains
w t0
a n n
u 2 ( a ) = rn2 · u2 ( w ) + w 2 · u2 ( rn ) = w 2 ⋅ + 0 + 0 + a 2 · urel
2
( w ) (11)
t g w t g t0 t 2
0
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
n n
a∗ = k1−α ⋅ u ( 0) = k1−α · w 2 ⋅ 0 + 0 = 0 , 504 Bq (12)
t g t0 t 2
0
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit is calculated by iteration, which leads to
( )
a# = a∗ + k1− β ⋅u a#
a# (13)
∗ #2 n0 n0
= a + k1− β ⋅ a ⋅ urel
2
( w ) + w 2 ⋅
+ + = 1 , 08 Bq
2
t g w t g t0 t0
Since α = β , k1−α = k1− β = k . The detection limit can also be calculated by the explicit Formula (14):
2 ⋅ a * + ( k 2 ⋅ w )/ t g
# ∗
( )
a = a + k1− β ⋅u a# =
1 − k 2 ⋅ urel
2
(w )
= 1 , 08 Bq (14)
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result a exceeds the decision threshold a∗ . With ω = Φ [ a / u( a )] = 1 , 00 and γ = 0 , 05 one
obtains the probabilities by p = ω ⋅ (1 − γ / 2) = 0 , 975 and q = 1 − ω ⋅ γ / 2 = 0 , 975 , and hence the quantiles
k p and kq are equal to 1,960. Then the lower and upper limits of the probabilistically symmetric
coverage interval are given by:
a = a − k p ⋅ u ( a ) = 34 , 94 Bq and a = a + kq ⋅u ( a ) = 63 , 15 Bq (15)
Alternatively, the lower and upper limits of the shortest coverage interval are calculated according to
Formula (16). With ω = Φ [ a / u( a )] = 1 , 00 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 975 , and hence the
quantile k p is equal to 1,960. With this, the limits of the shortest coverage interval are
The best estimate â of the activity of the sample and its associated standard uncertainty u ( â ) are
given by:
aˆ = a +
{
u ( a ) · exp − a2 2u2 (a)
}
= 49, 05 Bq (17)
ω 2π
and its associated standard uncertainty
u ( aˆ) = u2 ( a ) − ( aˆ − a ) ⋅ aˆ = 7, 20 Bq (18)
The available information is a = {ng , t g , n0 , t 0 , w , u( w )} ; see 6.3 and Table 1. The results and
characteristic limits are summarized in Table 2.
Applying ISO 11929-2 the marginal posterior for the true value of the measurand taking into account
that the measurand is non-negative is given by:
f A ( a a ) =
+∞
(19)
C ⋅ H( a ) ⋅ ∫ Ga(rg ;ng ,1 / t g )⋅ Ga(r0 ;n0 ,1 / t0 )⋅ N[w ;w , u(w )]⋅δ a − (rg − r0 )· w drg dr0 dw
−∞
Analogous formulae are valid for the marginal posteriors of the other examples in this document if
the PDFs under the integrals are replaced by those indicated in the tables in the clauses “Available
information, input data, and specifications.” Therefore, the respective posteriors are not repeated for
each example.
The integral is solved using Monte Carlo methods using the code UncertRadio[40]. Annex B gives an
exemplary description of the procedure following ISO 11929-2. The best estimate and its associated
standard uncertainty are calculated as the mean and the square root of the variance of the PDF according
to Formula (19); the limits of the coverage intervals are calculated as suitable quantiles of this PDF.
The PDFs required for the calculation of the primary measurement result and its associated standard
uncertainty as well as for the decision threshold and the detection limit are also given in Annex B.
The PDFs obtained by the Monte Carlo simulations are presented in Figure 1.
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 1 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The lower and upper limits of the probabilistically symmetric coverage interval are a = 34 , 94 Bq
and a = 63 , 15 Bq .
— The lower and upper limits of the shortest coverage interval are calculated as a < = 34 , 94 Bq and
a > = 63 , 15 Bq .
— The best estimate is aˆ = 49, 05 Bq with an associated standard uncertainty u( aˆ) = 7, 20 Bq .
Because the model of evaluation is linear and the relative uncertainty of the primary measurement
result is small, the results obtained by application of ISO 11929-2 are practically identical with those
obtained by ISO 11929-1:2019, Table 2. The ISO/IEC Guide 98-3, approximations assuming normal PDFs
and using a Taylor expansion truncated after the linear term, hold.
The probabilistically symmetric coverage interval is equal to the shortest coverage interval. The best
estimate and its associated standard uncertainty equal the primary measurement result and its
associated standard uncertainty. According to Formula (B.3) in ISO 11929-2:2019, Annex B, this holds
of for all results of measurements with y / u( y ) ≥ 4 , i.e. for measurement procedures with relative
uncertainties urel ( y ) ≤ 0 , 25 .
For larger relative uncertainties of the primary measurement result deviations between the limits of
the two coverage intervals and between the primary measurement result and the best estimate and
their respective standard uncertainties occur. This can be due to small count numbers (see Clause 7) or
due to large uncertainties of the calibration factor. With respect to uncertainties of the calibration factor
according to Formula (2), one has to distinguish whether they result from quantities in the numerator
or the denominator of Formula (2). These cases are dealt with in Clause 9 and Clause 10, respectively.
A further origin of large uncertainties can be due to those of the shielding factor; see Clause 11 for an
example.
ng n0
a= − · w = ( rg − r0 ) · w = rn · w (20)
t t 0
g
ng n
u2 ( a ) = w 2 ⋅ u2 ( rn ) + rn2 ⋅ u2 ( w ) = w 2 ⋅ + 0 + a2 ⋅ urel
2
( w ) (21)
t
g t0
2 2
The PDFs of the calibration factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also of ISO 11929-2:2019, 6.3 for a detailed description.
The background effect of the sample has been measured. The number of the counted pulses is n0 = 3 .
This gives a count rate of
n0
r0 = = 0 , 00250 s −1 (22)
t0
The standard uncertainty of the count rate is given by:
n0
u ( r0 ) = = 0 , 001 44 s −1 (23)
t 02
The activity of the sample is calculated by using the Formula (20). This relation depends on the
calibration factor w, which needs to be evaluated first, as well as its standard uncertainty u(w).
ng
a = rn ⋅ w = − r0 ⋅ w = 0 , 0103 Bq (25)
t
g
with the standard uncertainty
ng n
u ( a ) = rn2 ⋅ u2 ( w ) + w 2 ⋅ + 0 = 0 , 010 4 Bq (26)
t2 t2
g 0
and
urel ( a ) = u( a )/ a = 1 , 01 (27)
The standard uncertainty u ( a ) as a function of the true value a is needed to calculate the decision
threshold and the detection limit. For a true value a = rn ⋅ w of the measurand, one expects a gross count
a n
rate rg = + 0 and with Formula (21) one obtains
w t0
a n n
u 2 ( a ) = rn2 · u2 ( w ) + w 2 · u2 ( rn ) = w 2 ⋅ + 0 + 0 + a 2 · urel
2
( w ) (28)
t g w t g t0 t 2
0
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
n n
a∗ = k1−α u ( 0) = k1−α · w 2 ⋅ 0 + 0 = 0 , 013 8 Bq (29)
t g t0 t 2
0
The measurement result a is below the decision threshold a∗ .
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit is calculated by iteration, which leads to
2 a# n n
a# = a∗ + k1− β ⋅ a# ⋅ urel
2
( w ) + w 2 ⋅ + 0 + 0 = 0 , 039 Bq (30)
t g w t g t0 t0
2
Since α = β , k1−α = k1− β = k . The detection limit can also be calculated by the explicit Formula (31):
2 ⋅ a * + ( k 2 ⋅ w )/ t g
( )
a# = a∗ + k1− β ⋅u a# =
1 − k 2 ⋅ urel
2
(w )
= 0 , 039 Bq (31)
The lower and upper limits of the probabilistically symmetric coverage interval are calculated
exemplarily though the measurement result a is below the decision threshold a∗ . With
ω =Φ ( a / u( a )) = 0 , 837 and γ =0 , 05 one obtains the probabilities p = ω ⋅ (1 − γ / 2) = 0 , 816 and
q = 1 − ω ⋅ γ / 2 = 0 , 979 , and hence the quantiles k p and kq are equal to 0,908 and 2,034, respectively.
Then the lower and upper limits of the probabilistically symmetric coverage interval are given by:
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω =Φ ( a / u( a )) = 0 , 837 , one obtains p = (1 + ω ⋅(1 − γ ))/ 2 = 0 , 897 and hence the quantile k p is equal
to 1,27. With this, the limits of the shortest coverage interval according to Formula (33) are
Since the lower limit of the shortest coverage interval according to Formula (33) is negative, one has to
proceed according to Formula (34) and obtains with q = 1 − ω ⋅ γ = 0 , 958 the quantile kq = 1 , 728
aˆ = a +
( (
u ( a ) · exp − a2 2u2 (a) )) = 0, 013 3 Bq (35)
ω 2π
with its associated standard uncertainty u ( â )
Table 4 (continued)
Results Symbol Unit ISO 11929-1 ISO 11929-2
Relative uncertainty associated with urel ( aˆ) 1 0,62 0,68
the best estimate
Lower limit of the probabilistically a Bq 8,54E-4 8,24E-4
symmetric coverage interval
Upper limit of the probabilistically a Bq 3,13E-2 3,38E-2
symmetric coverage interval
Lower limit of the shortest coverage a< Bq 0 0
interval
Upper limit of the shortest coverage a> Bq 2,82E-2 2,93E-2
interval
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 2 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The primary measurement result a = 0 , 0103 Bq is below the decision threshold a * = 0 , 013 8 Bq . It
is decided to conclude that no effect of the sample was recognized. In spite of that all characteristic
values were calculated and documented in this example.
— The detection limit a# = 0 , 039 Bq is below the guideline value ar = 0 , 1 Bq . It is decided to conclude
that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are a = 0 , 000 854 Bq
and a = 0 , 0313 Bq .
— The lower and upper limits of the shortest coverage interval are calculated as a < = 0 Bq and
a > = 0 , 028 2 Bq .
— The best estimate is aˆ = 0, 013 3 Bq with an associated standard uncertainty u( aˆ) = 0, 008 2 Bq .
There are just minor differences between the results obtained by applying ISO 11929-1 and ISO 11929-2,
which are due to the deviation between the Gaussian distributions and the actual posterior distributions
calculated according to ISO 11929-2; see Figure 2.
If large relative uncertainties of the primary result are due to the uncertainties of the count rates, this
does not affect the existence of the detection limit calculated according to ISO 11929-1. Also in this
extreme case of low count numbers, the application of ISO 11929-1 is justified.
7.7.1 General
The data of an alternative example are given in Table 5. The input data differ from the example of Table 3
just with respect of the count numbers, which are now n0 = 1 and ng = 4 .
NOTE This case with just one background count is regarded as a limiting case. In any reasonable single-
channel counting measurement it is expected that at least one count is obtained in both, the gross and the
background measurement. The case of a single-channel counting measurement with zero background counts is
considered as a pathological case where the measurement or even the measurement method is insufficient.
The PDFs of the calibration factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also of ISO 11929-2:2019, 6.3 for a detailed description.
The background effect of the sample has been measured. The number of the counted pulses is n0 = 1 .
This gives a count rate of
n0
r0 = = 0 , 000 833 s −1 (37)
t0
The standard uncertainty of the count rate is given by:
n0
u ( r0 ) = = 0 , 000 833 s −1 (38)
t 02
The activity of the sample is calculated by using the Formula (20). This relation depends on the
calibration factor w, which needs to be evaluated first, as well as its standard uncertainty u(w).
The calibration factor is given by:
w = 4 , 1 Bq ⋅ s with u ( w ) = 0 , 6 Bq ⋅ s and urel ( w ) = 0 , 146 (39)
ng
a = rn ⋅ w = − r0 ⋅ w = 0 , 0103 Bq (40)
t
g
with the standard uncertainty
ng n
u ( a ) = rn2 ⋅ u2 ( w ) + w 2 ⋅ + 0 = 0 , 007 79 Bq (41)
t2 t2
g 0
and
urel ( a ) = u( a )/ a = 0 , 756 (42)
The standard uncertainty u ( a ) as a function of the true value a is needed to calculate the decision
threshold and the detection limit. For a true value a = rn ⋅ w of the measurand one expects a gross count
a n
rate rg = + 0 and with Formula (21) one obtains
w t0
a n n
u 2 ( a ) = rn2 · u2 ( w ) + w 2 · u2 ( rn ) = a 2 · urel
2
( w ) + w 2 ⋅ + 0 + 0 (43)
2
t g w t g t0 t0
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
n n
a∗ = k1−α u ( 0) = k1−α · w 2 ⋅ 0 + 0 = 0 , 007 95 Bq (44)
t g t0 t 2
0
For β = 5 %, the value of the corresponding quantile k1− β of the standardised normal distribution is
1,645. The detection limit is calculated by iteration, which leads to
2 a# n n
a# = a∗ + k1− β ⋅ a# ⋅ urel
2
( w ) + w 2 ⋅ + 0 + 0 = 0 , 026 7 Bq (45)
t g w t g t0 t0
2
Since α = β , k1−α = k1− β = k . The detection limit can also be calculated by the explicit Formula (46):
2 ⋅ a * + ( k 2 ⋅ w )/ t g
( )
a# = a∗ + k1− β ⋅u a# =
1 − k 2 ⋅ urel
2
(w )
= 0 , 026 7 Bq (46)
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result a exceeds the decision threshold a∗ . With ω = Φ [ a / u( a )] = 0 , 907 and γ = 0 , 05 one
obtains the probabilities by p = ω ⋅ (1 − γ / 2) = 0 , 884 and q = 1 − ω ⋅ γ / 2 = 0 , 977 , and hence the quantiles
k p and kq are equal to 1,20 and 2,00, respectively. Then the lower and upper limits of the probabilistically
symmetric coverage interval are given by:
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ a / u( a )] = 0 , 907 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 931 , and hence the quantile k p is equal
to 1,48. With this the limits of the shortest coverage interval according to Formula (48) were
Since the lower limit of the shortest coverage interval according to Formula (48) is negative, one has to
proceed according to Formula (49) and obtains with q = 1 − ω ⋅ γ = 0 , 955 the quantile kq = 1 , 692
aˆ = a +
{
u ( a ) · exp − a2 2u2 (a)
}
= 0, 011 7 Bq (50)
ω 2π
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 3 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The primary measurement result a = 0 , 0103 Bq exceeds the decision threshold a * = 0 , 007 95 Bq .
It is decided to conclude that an effect of the sample was recognized.
— The lower and upper limits of the probabilistically symmetric coverage interval are a =0 , 000 97 Bq
and a = 0 , 026 Bq .
— The lower and upper limits of the shortest coverage interval are calculated as a< =0 Bq and
a> =0 , 023 Bq .
— The best estimate is aˆ = 0, 011 7 Bq with an associated standard uncertainty u( â ) = 0, 006 6 Bq .
There are just minor differences between the results obtained by applying ISO 11929-1 and ISO 11929-2,
which are due to the deviation between the Gaussian distributions and the actual posterior distributions
calculated according to ISO 11929-2; see Figure 3.
If large relative uncertainties of the primary result are due to the uncertainties of the count rates, this
does not affect the existence of the detection limit calculated according to ISO 11929-1. Also in this
extreme case of low count numbers the application of ISO 11929-1 is justified.
The comparison between the primary measurement result and the decision threshold provides
information on whether or not an activity of the sample is observed. If the primary measurement result
is larger than the decision threshold, it is concluded that an activity has been observed.
A guideline value ar = 10 Bq is to be compared with the detection limit. If the detection limit is smaller
than the guide value, it is concluded that the measuring method is suitable for the measurement
purpose.
ng n0
a = ( rg − r0 ) · w = rn · w = − · w (52)
t t 0
g
ng n
u2 ( a ) = w 2 ⋅ u2 ( rn ) + rn2 ⋅ u2 ( w ) = w 2 ⋅ + 0 + a2 ⋅ urel
2
( w ) (53)
t
g t0
2 2
The PDFs of the calibration factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a detailed description. It is assumed
in this example that the calibration is badly defined and that only the limits of a range of calibration
factors (3 s·Bq to 20 s·Bq) are known. This leads to assuming a rectangular PDF for the calibration
factor and a large relative standard uncertainty of the calibration factor of urel ( w ) = 0 , 43 .
The count rate of the background effect of the sample measurement has been measured. The number of
counted pulses is ng = 73150 . This gives a background count rate of
n0
r0 = = 6 , 096 s −1 (54)
t0
Its standard uncertainty associated with the background count rate is given by:
n0
u ( r0 ) = = 0 , 0225 s −1 (55)
t 02
The activity is calculated from Formula (52). This formula depends on the calibration factor, the input
quantities of which and their associated standard uncertainties were determined independently. This
yielded a calibration factor of
w =11,5Bq ⋅ s (56)
2
ng n ng n
u( a ) = w 2 ⋅ + 0 + − 0 ⋅ u2 ( w ) = 58 , 7 Bq
q (59)
t2 t2 t t
g 0 g 0
8.4.3 Standard uncertainty as a function of an assumed true value
The standard uncertainty u( a ) as a function of the true value a of the measurand is needed to calculate
the decision threshold and the detection limit. For a true value a = ( rg − r0 ) ⋅ w one expects
n g = t g ⋅ ( a w + r0 ) . This yields with Formula (53)
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit is calculated by iteration, which leads to
a# = a * + k1− β ⋅ w 2 ⋅ (( a# w + r0 ) t g + r0 t 0 ) + a#2urel
2
( w ) = 5 , 63 Bq (62)
Since α = β , k1−α = k1− β = k , the detection limit can also be calculated by the explicit Formula (63):
2 ⋅ a * + ( k 2 ⋅ w )/ t g
( )
a# = a∗ + k1− β ⋅u a# =
1 − k 2 ⋅ urel
2
(w )
= 5 , 63 Bq (63)
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result a exceeds the decision threshold a* . With ω =Φ [ a / u( a )] = 0 , 990 and γ =0 , 05 one
obtains the probabilities by p = ω ⋅(1 − γ / 2) = 0 , 966 and q = 1 − ω ⋅ γ / 2 = 0 , 975 , and hence the quantiles
k p and kq are equal to 1,821 and 1,964, respectively. Then the lower and upper limits of the
probabilistically symmetric coverage interval are given by:
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ a / u( a )] = 0 , 990 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 970 , and hence the quantile kp is equal
to 1,887. With this, the limits of the shortest coverage interval according to Formula (65) were:
aˆ = a +
{
u ( a ) ⋅ exp − a2 2u2 ( a )
}
= 139, 09 Bq (66)
ω 2π
with its associated standard uncertainty u( aˆ)
8.5 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2
The results and characteristic limits are summarized in Table 8 and the PDFs obtained by the Monte
Carlo simulations are presented in Figure 4.
Table 8 (continued)
Results Symbol Unit ISO 11929-1 ISO 11929-2
Standard uncertainty associated with the prima-
u( a ) Bq 59 59
ry result
Relative standard uncertainty associated with
urel ( a ) 1 0,428 0,431
the primary result
Decision threshold a* Bq 1,41 1,56
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 4 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The detection limit a# = 5 , 63 Bq is below the guideline values ar =10 Bq . It is decided to conclude
that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are a = 30 , 7 Bq
and a = 253 Bq .
— The lower and upper limits of the shortest coverage interval are calculated as a < = 26 , 7 Bq and
a > = 248 Bq .
— The best estimate is â = 139 Bq with an associated standard uncertainty u( aˆ) = 57 Bq .
The results obtained by applying ISO 11929-1 and ISO 11929-2 differ with regard to the detection limits
and the limits of the coverage intervals. This is due to the deviation between the Gaussian distributions
and the actual posterior distributions calculated according to ISO 11929-2 (see Figure 4). The
rectangular PDF of the calibration factor is responsible for the differences. The differences between
the best estimates and their associated standard uncertainties are small because the central estimates
of the rectangular distribution of the calibration factor is identical to that of the normal distribution
applied in ISO 11929-1.
Considering the differences in the limits of the coverage intervals, the application of ISO 11929-2 is to
be preferred if there are large and dominating uncertainties in the numerator of the calibration factor.
It is assumed that an area s of a surface has been wiped with a wiping efficiency ew . The wipe, i.e. the
sample, is measured by an unspecified detector with the counting efficiency eD . For simplicity it is
assumed that the surface emission rate measured is equal to the count rate caused by the surface
related activity as . This measurement is an example for a case where the uncertainty of one of the
input quantities is dominating.
The comparison between the primary measurement result and the decision threshold provides
information on whether or not an activity of the sample is observed. If the primary measurement result
is larger than the decision threshold, it is concluded that an activity of the sample has been observed.
A guideline value as,r = 0 , 5 Bq ⋅ cm −2 is to be compared with the detection limit. If the detection limit is
smaller than the guide value, it is concluded that the measuring method is suitable for the measurement
purpose.
The area wiped was chosen as s = 100 cm2 and its associated relative standard uncertainty was set to
urel ( s ) = 10 % by expert guess.
For the wiping efficiency ew past experience was that it randomly varied in the interval ew ∈[0 , 06 ; 0 , 62]
with an average ew = 0 , 34 and an associated standard uncertainty u( ew ) = 0 , 162 derived from a
rectangular PDF.
The input data and their associated uncertainties are given in Table 9. The PDFs assigned to the input
data and used in the application of ISO 11929-2 are also given in Table 9.
Table 9 (continued)
Quantity Symbol xi u ( xi ) PDF Unit
Count time of the background effect t0 7 200 — — s
Efficiency of the detector eD 0,30 0,015 5 N( eD ;eD , u( eD )) Bq-1 s-1
Wiping efficiency eW 0,34 0,162 R( ew ;0 , 06 , 0 , 62) 1
Wiped area s 100 10 N( s ; s , u( s )) cm2
Intermediate values
Gross count rate rg 7,20 0,141 Ga( rg ;ng ,1 / tg ) s-1
Count rate of the background effect r0 5,80 0,028 4 Ga( r0 ;n0 ,1 / t0 ) s-1
The PDFs of the efficiency of the detector, the wiping efficiency, the wiped area, and of the count rates
were chosen according to the ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a
detailed description. It is assumed in this example that for the wiping efficiency only the limits of a
range (0,06 to 0,62) are known from experience. This leads to assuming a rectangular PDF for the
wiping efficiency and a large relative standard uncertainty of the wiping efficiency of urel ( ew ) = 0 , 48 .
The particular point of this example compared to that in Clause 8 is that now the quantity with a large
uncertainty is in the denominator.
r0
u ( r0 ) = = 0 , 028 4 s −1 (72)
t0
u( w ) = w ⋅urel ( w ) = w ⋅ urel
2 ( e )+ u2 ( s )+ u2 ( e ) = 0 , 048 s Bq ⋅cm-2 (74)
D rel rel w
as = ( ng / t g − n0 / t 0 ) ⋅ w = 0 , 137 Bq ⋅ cm −2 (75)
The standard uncertainty u( as ) as a function of the true value aS of the measurand is needed to
calculate the decision threshold and the detection limit. For a true value as = ( rg − r0 ) ⋅ w one expects
n g = t g ⋅ ( aS w + r0 ) . This yields with Formula (76)
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
The probability β is set to 5 %. Then, the value of the corresponding quantile k1−β of the standardised
normal distribution is 1,645. The detection limit is calculated by iteration, which leads to
Since α = β , k1−α = k1− β = k , the detection limit can also be calculated by the explicit Formula (80):
2 ⋅ aS* + ( k 2 ⋅ w )/ t g
( )
aS# = aS* + k1− β ⋅u aS# =
1−k 2
⋅ urel
2
(w )
= 0 , 121 Bq ⋅ cm-2 (80)
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result as exceeds the decision threshold as* . With ω =Φ [ as / u( as )] = 0 , 977 and γ =0 , 05
one obtains the probabilities by p = ω ⋅(1 − γ / 2) = 0 , 953 and q = 1 −ω ⋅γ / 2 = 0 , 976 , and hence the quantiles
kp and kq are equal to 1,674 and 1,970, respectively. Then the lower and upper limits of the
probabilistically symmetric coverage interval are given by:
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ as / u( as )] = 0 , 977 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 964 , and hence the quantile k p is equal
to 1,802. With this, the limits of the shortest coverage interval according to Formula (82) are
The best estimate âs of the activity of the sample is given by:
aˆs = as +
{ }
u ( as ) ⋅ exp − as2 2u2 ( as )
= 0, 140 Bq ⋅ cm−2 (83)
ω 2π
with its associated standard uncertainty u( aˆs )
9.5 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2
The results and characteristic limits are summarized in Table 10 and the PDFs obtained by the
Monte Carlo simulations are presented in Figure 5.
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 5 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The detection limit as# = 0 , 121 Bq ⋅ cm −2 is below the guideline values as,r = 0 , 5 Bq ⋅ cm −2 . It is decided
to conclude that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are as = 0 , 022 Bq ⋅ cm −2
and as = 0 , 271 Bq ⋅ cm −2 .
— The lower and upper limits of the shortest coverage interval are calculated as as< = 0 , 014 Bq ⋅ cm −2
and as> = 0 , 260 Bq ⋅ cm −2 .
— The best estimate is aˆs = 0, 140 Bq ⋅ cm−2 with an associated standard uncertainty
u( aˆs ) = 0, 064 Bq ⋅ cm−2 .
The results obtained by applying ISO 11929-1 and ISO 11929-2 differ strongly with regard numerical
values (Table 10) and the PDFs (Figure 5). This effect is mainly due to the rectangular PDF in the
denominator. Considering these differences, the application of ISO 11929-2 is to be preferred if there
are large and dominating uncertainties in the denominator of the calibration factor.
The standard uncertainty u( rn ) associated with the primary measurement result rn is calculated by:
The PDFs of the shielding factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a detailed description. It is assumed
in this example that for the shielding factor only the limits of a range of shielding factors (0,1 to 0,7) are
known. This leads to assuming a rectangular PDF for the calibration factor and a large relative standard
uncertainty of the calibration factor of urel ( w ) = 0 , 43 .
r0
u ( r0 ) = = 0 , 087 8 s −1 (88)
t0
Since there is no calibration of the detector with respect to an activity, the measurand is the net count
rate. The primary result of the measurement is given by:
rn = ng / t g − s ⋅ n0 / t 0 = 10 , 56 s −1 (89)
The standard uncertainty u( rn ) as a function of the true value rn of the measurand is needed to
calculate the decision threshold and the detection limit. For a true value rn = rg − s ⋅ r0 one expects
n g = t g ⋅ ( rn + s ⋅ r0 ) . This yields with Formula (86)
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the net count rate is calculated by:
1 s
rn* = k1−α ⋅ u( 0 ) = k1−α ⋅ s ⋅ r0 ⋅ + + r02 ⋅ u2 ( s ) = 3 , 97 s-1 (92)
t
g t0
The measured primary result rn exceeds the decision threshold rn* .
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit is calculated by iteration, which leads to
rn# + s ⋅r0 r0
rn# = rn* + k1−β ⋅ + s2 ⋅ + r02 ⋅u2 ( s ) = 7 , 95 s-1 (93)
tg t0
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result rn exceeds the decision threshold rn* . With ω = Φ [ rn / u( rn )] = 1 , 000 and γ = 0 , 05
one obtains the probabilities by p = ω ⋅ (1 − γ / 2) = 0 , 975 and q = 1 − ω ⋅ γ / 2 = 0 , 975 , and hence the
quantiles k p and kq are both equal to 1,96. Then the lower and upper limits of the probabilistically
symmetric coverage interval are given by:
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ rn / u( rn )] = 1 , 00 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 975 , and hence the quantile k p is equal to
1,96. With this the limits of the shortest coverage interval according to Formula (95) are
The best estimate r̂n of the net count rate of the sample is given by:
rˆn = rn +
{
u ( rn ) ⋅ exp − rn2 2u2 ( rn )
}
= 10, 56 s−1 (96)
ω 2π
with its associated standard uncertainty u(ˆ
rn )
10.5 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2
The results and characteristic limits are summarized in Table 12 and the PDFs obtained by the
Monte Carlo simulations are presented in Figure 6.
Table 12 (continued)
Results Symbol Unit ISO 11929-1 ISO 11929-2
Detection limit rn# s-1 7,95 7,51
Lower limit of the shortest coverage interval rn< s-1 5,80 6,54
Upper limit of the shortest coverage interval rn> s-1 15,3 14,58
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 6 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The detection limit rn# = 7 , 95 s −1 is below the guideline values rn,r = 10 s −1 . It is decided to conclude
that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are rn = 5 , 8 s-1 and
rn = 15 , 3 s-1 .
— The lower and upper limits of the shortest coverage interval are calculated as rn< = 5 , 8 s-1 and
rn> = 15 , 3 s-1 .
— The best estimate is rˆn = 10, 6 s-1 with an associated standard uncertainty u(ˆ
rn ) = 2, 4 s-1 .
The results obtained by applying ISO 11929-1 and ISO 11929-2 exhibit differences up to 10 %. This is due
to the deviation between the Gaussian distributions and the actual posterior distributions calculated
according to ISO 11929-2; see Figure 6. The differences are mainly due to the lack of knowledge about
the shielding factor and its associated rectangular PDF.
The coverage intervals calculated with ISO 11929-2 are smaller than those based on ISO 11929-1. The
differences between the best estimates and their associated standard uncertainties are small because
the central estimates of the rectangular distribution of the shielding factor is identical to that of the
normal distribution applied in ISO 11929-1.
Considering the differences in the limits of the coverage intervals, the application of ISO 11929-2 is to
be preferred if there are large and dominating uncertainties in the numerator of the shielding factor.
NOTE 2 These monitors consist of a lead shielding of at least 5 cm thickness surrounding the waste as a cuboid
and several plastic scintillation detectors (typically 6 to 24 detectors) in the inner volume in front of the shielding.
The detectors and the shielding are built up in a way to cover the material in a so-called 4π-geometry. According
to the large volumes of the detectors, uncertainties due to counting statistics can practically be neglected. Due
to the gross measurement of gamma radiation no nuclide specific information can be obtained; it is a counting
measurement of nuclide equivalents.
NOTE 3 To ensure metrological traceability this leads to the fact, that a lot of efficiency calibrations are
performed (typically 10 to 100) for different combinations of the parameters mentioned above. For practical
reasons a somewhat conservative parameter is chosen to calculate the results (e. g. if there is a calibration for
250 kg and 300 kg and the net mass is 280 kg, the calibration with 300 kg is used).
NOTE 4 In comparison to spectrometric measurements (e. g. with drum scanners) using HPGe-detectors
the disadvantages of non-nuclide-specific measurements stand in contrast to advantages, in particular, of
significantly shorter measurement duration times and lower measurement uncertainties due to inhomogeneity
of the activity distribution.
A guideline value ar = 30 000 Bq according to the clearance level for unconditional clearance
of 0,1 Bq/g = 30 000 Bq/300 kg is assumed.
ng n0
a= − ⋅ x3 − x 41 + x 42 ⋅ w = ( rg − r0 ⋅ x3 − x 41 + x 42 ) ⋅ w (98)
t
g t0
with
x3 correction of the background count rate for its variability due to work activities near the
device for clearance measurements,
x41 correction for natural radionuclides, i.e. 40K and the 226Ra and 232Th decay series, in the
material to be measured. The concrete was characterized in independent measurements for
its natural radioactivity and the resulting contribution x4 to the gross count rate and its
associated standard certainty u( x4 ) were determined.
( )
u2 ( a ) = w 2 ⋅ rg t g + x32 ⋅ r0 t 0 + r02 ⋅ x32 ⋅ u2 ( x3 ) + u2 ( x 41 ) + u2 ( x 42 ) + a2 ⋅ urel
2
( w ) (99)
The PDFs of the calibration factor, of the correction factors, and of the count rates were chosen according
to the ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a detailed description.
r0
u ( r0 ) = = 3 , 07 s −1 (101)
t0
ng n0
a= − ⋅ x3 − x 41 + x 42 ⋅ w = 12 477 Bq (102)
t
g t0
and the associated standard uncertainty by
The standard uncertainty u( a ) as a function of the true value a of the measurand is needed to calculate
the decision threshold and the detection limit.
a n
For a true value a one expects n g = + 0 ⋅ x3 + x 41 − x 42 ⋅ t g . This yields with Formula (99)
w t0
a n0
+ ⋅ x3 + x 41 − x 42 / t g + x32 ⋅ r0 t 0
2 w t
u (a) = w ⋅
2
0 + a ⋅ urel ( w ) (104)
2 2
2 2
+ r0 ⋅ u ( x3 ) + u ( x 41 ) + u ( x 42 )
2 2
11.4.4 Decision threshold
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
a * = k1−α ⋅ u( 0 )
n
= k1−α ⋅ w ⋅ 0 ⋅ x3 + x 41 − x 42 / t g + x32 ⋅ r0 t 0 + r02 ⋅ u2 ( x3 ) + u2 ( x 41 ) + u2 ( x 42 ) (105)
t0
= 1747 Bq
The measured primary result exceeds the decision threshold a * .
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit is calculated by iteration, which leads to
a # n0
w 2 ⋅ + ⋅ x3 + x 41 − x 42 / t g + x32 ⋅ r0 t 0 + r02 ⋅ u2 ( x3 ) + u2 ( x 41 ) + u2 ( x 42 )
a = a + k1− β ⋅
# * w t 0
+ a #2 ⋅ urel
2
(w )
= 4 618 Bq
(106)
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result a exceeds the decision threshold a* . With ω =Φ [ a / u( a )] = 0 , 999 and γ =0 , 05 one
obtains the probabilities by p = ω ⋅(1 − γ / 2) = 0 , 974 and q = 1 −ω ⋅γ / 2 = 0 , 975 , and hence the quantiles kp
and kq are equal to 1,95 and 1,96, respectively. Then the lower and upper limits of the probabilistically
symmetric coverage interval are given by:
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ a / u( a )] = 0 , 999 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 975 , and hence the quantile k p is equal to
1,954. With this the limits of the shortest coverage interval according to Formula (108) were
aˆ = a +
{ }
u ( a ) ⋅ exp − a2 2u2 ( a )
= 12 486 Bq (109)
ω 2π
with its associated standard uncertainty u( aˆ)
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 7 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The primary measurement result a = 12 477 Bq exceeds the decision threshold a * = 1747 Bq . It is
decided to conclude that an effect from the sample was recognized.
— The detection limit a# = 4 618 Bq is below the guideline values ar =30 000 Bq . It is decided to
conclude that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are a = 4 894 Bq
and a =20104 Bq .
— The lower and upper limits of the shortest coverage interval are calculated as a < = 4 872 Bq and
a > = 20 081 Bq .
— The best estimate is â = 12 486 Bq with an associated standard uncertainty u( aˆ) = 3 876 Bq .
The results obtained by applying ISO 11929-1 and ISO 11929-2 exhibit differences of less than 10 %.
The differences are due to the deviation between the Gaussian distributions and the actual posterior
distributions calculated according to ISO 11929-2; see Figure 7.
The coverage intervals calculated with ISO 11929-2 differ only slightly from those based on ISO 11929-1.
The differences between the best estimates and their associated standard uncertainties are small
because the central estimates of the rectangular distribution of the calibration factor is identical to that
of the normal distribution applied in ISO 11929-1.
Considering the differences between the results obtained by ISO 11929-1 and ISO 11929-2 the
ISO/IEC Guide 98-3 approximation and consequently ISO 11929-1 are sufficient.
NOTE 1 Since this document does not deal with the problem of conformity with requirements, as e.g. with
the clearance level for unconditional clearance, the question whether or not the material measured can be
unconditionally cleared is not answered. For an answer to this question see Reference [45].
NOTE 2 It is necessary to take into account that in the present case favourable boundary conditions with
regard to the mass of the material to be measured and the photon energies (60Co) been selected. By increasing
the mass to several 100 kg or smaller photon energies (e. g. 137Cs) it can be necessary to use an approach as given
in A.3.
n 2
∂a
2
∑
u ( am ) = m ⋅ u2 ( x i ) (113)
∂x i
i =1
which yields
1
The relative standard uncertainty of the calibration factor w = is given by:
eU ,186 ⋅ m
urel
2
( w ) = urel
2
( eU ,186 ) + urel
2
( m ) (116)
t count time
ng,186 number of gross counts in the peak at 186 keV
nn,609 Number of counts in the net peak of 214Bi at 609 keV
nU,186 number of counts in the background under the peak at 186 keV
nn ,186 ,0 number of net counts in the peak at 186 keV due to activity of the impurity in the detector
rg,186 gross count rate in the 186 keV peak
Table 15 (continued)
Symbol
Quantity xi u ( xi ) Unit PDF
for xi
Background count rate of rU,186 0,412 0,005 24 s-1 Ga( r ;n ,1 / t )
the 186 keV peak
Count rate in the net peak
at 186 keV due to the activity of rn ,186 ,0 0,013 8 0,000 959 s-1 Ga( r ;n ,1 / t )
the impurity in the detector
The PDFs of the count rates and of the input data, except the given actual numbers of counts, were
chosen according to the ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a detailed
description.
The background under the peak at 186 keV is nU,186 = 618 1 which yields a count rate of
nU,186
rU,186 = = 0 , 412 s-1 (117)
t
The standard uncertainty of background under the peak at 186 keV is
nU,186
u ( rU,186 ) = = 0 , 00524 s-1 (118)
t 2
For the calculation of the primary measurement result am, the activity per unit mass of 235U, first the
correction factor k and its associated standard uncertainty u(k) are determined.
urel ( w ) = urel
2
( eU ,186 ) + urel
2
( m ) = 0 , 051 4 (120)
The correction factor k is calculated with the emission probabilities and the detector efficiencies
eRa ,186
k= = 0 , 143 (122)
eRa ,609 · ε 609
and its relative standard uncertainty is
urel ( k ) = u2rel ( eRa ,186 ) + u2rel ( eRa ,609 ) + u2rel ( ε 609 ) = 0 , 0633 (123)
With the correction factor k and the calibration factor w the primary measurement result am is
ε 186
2
= 0 , 030 2 Bq ⋅ g -1
12.4.3 Standard uncertainty as a function of an assumed true value
The standard uncertainty u( am ) as a function of the true value am of the measurand is needed to
calculate the decision threshold and the detection limit. For a true value am one expects
a
rg ,186 = ε186 ⋅ m + ε186 ⋅rn,609 · k + rU,186 + rn ,186 ,0 .
w
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
am
*
= k1−α ⋅ u( 0 )
1 rn,609 rU,186 1
( rn,609 ⋅ ε 186 · k + rU,186 + rn ,186 ,0 ) ⋅ +
t
⋅ k2 +
t
⋅
ε 186 ⋅t ε 186
2 2
= k1−α ⋅ w ⋅ (128)
rn,186,0 1
+ ⋅ + urel
2
( ε 186 ) ⋅ r 2 · k 2 + u2 ( k ) ⋅ rn,2 609
t ε 186
2 n,609
= 0 , 0455 Bq ⋅ g -1
The primary measurement result am exceeds the decision threshold am
*
.
For β = 0,05 is k1− β = 1 , 65 and by solving the implicit Formula one obtains the detection limit
am
#
= am
*
+ k1− β ·u( am
#
)
ε 186 ⋅ a# 1
( m
+ rn,609 ⋅ ε 186 · k + rU,186 + rn ,186 ,0 ) ⋅
w ε 186
2
⋅t
rn,609 rU,186 1 rn,186,0 1
= am
*
+ k1− β ⋅ w ⋅ + ⋅ k2 + ⋅ + ⋅ (129)
t t ε 186
2 t ε 186
2
2
a#
a# 2 2
+ u rel ( ε 186 ) ⋅
2 m
+ rn,609 · k + u2 ( k ) ⋅ rn,2 609 + m ⋅ urel (w )
w w2
= 0 , 099 2 Bq ⋅ g -1
12.4.6 Limits of coverage intervals
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result a exceeds the decision threshold a*. With ω = Φ [ am / u( am )] = 0 , 966 7 and γ = 0 , 05
one obtains the probabilities by p = ω ⋅ (1 − γ / 2) = 0 , 9425 and q = 1 − ω ⋅ γ / 2 = 0 , 975 8 , and hence the
quantiles k p and kq are equal to 1,576 and 1,974, respectively. Then the lower and upper limits of the
probabilistically symmetric coverage interval are given by:
am
= am − k p ⋅ u ( am ) = 0 , 007 79 Bq ⋅ g -1 and am ( )
= am + kq ⋅u am = 0 , 115 Bq ⋅ g -1 (130)
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ am / u( am )] = 0 , 966 7 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 959 2 , and hence the quantile k p is
equal to 1,741. With this, the limits of the shortest coverage interval according to Formula (131) are
< >
am = am − k p ⋅ u ( am ) = 0 , 002 81 Bq ⋅ g -1 and am = am + k p ⋅u ( am ) = 0 , 108 Bq ⋅ g -1 (131)
The best estimate âm of the activity of 235U per unit mass of the sample is given by:
aˆm = am +
{ }
u ( am )·exp − am2 / 2u2 ( am )
= 0, 057 7 Bq ⋅ g-1 (132)
ω 2π
with its associated standard uncertainty
12.5 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2
The results and characteristic limits are summarized in Table 16 and the PDFs obtained by the
Monte Carlo simulations are presented in Figure 8.
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 8 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
in the sample.
— Since no guideline value is given the assessment of the measurement procedure is omitted.
— The lower and upper limits of the probabilistically symmetric coverage interval are calculated as
am = 0 , 007 8 Bq ⋅ g -1 and am = 0 , 115 Bq ⋅ g -1 .
— The lower and upper limits of the shortest coverage interval are calculated as am < = 0 , 002 8 Bq ⋅ g -1
and am > = 0 , 108 Bq ⋅ g -1 .
— The best estimate is calculated as aˆm = 0, 057 7 Bq ⋅ g-1 with an associated standard uncertainty
u ( aˆm ) = 0, 027 9 Bq ⋅ g-1 .
The model of evaluation is linear, and there are no dominating large uncertainties so that the
application of ISO 11929-1 is justified. In spite of the fact that the relative uncertainty of the primary
estimate exceeds 50 %, many results obtained by application of ISO 11929-2 are practically identical
with those obtained by ISO 11929-1:2019, Table 16. Large differences are seen between the limits of the
probabilistically symmetric and the shortest coverage intervals, the upper limits being more affected
than the lower limits. The ISO/IEC Guide 98-3 approximations, assuming normal PDFs and using a
Taylor expansion truncated after the linear term, hold.
ng nb
∑ ∑ x b ,i , and
1 1
with the means of the gross and background measurements, x g = x g ,i and x b =
ng i =1
nb i =1
ng 1/2
1
with the standard deviations of the gross and background measurements, sg =
ng − 1 i =1 ∑
( x g ,i −x g )2
1/2
1 nb
and sb = ∑
nb − 1 i =1
( x b ,i −x b )2 , respectively.
According to the ISO/IEC Guide 98-3:2008/Suppl.1:2008, 6.4.9 the repeated measurements are assumed
to be drawn from a Gaussian distribution with unknown mean and variance. Consequently, the PDF is
the t-distribution t n−1 ( x , s 2 / n ) . The variance of this distribution then yields the standard uncertainties
associated with the means of the gross and background measurements
1/2
ng − 1 sg n −1
1/2
sb
u( x g ) = ⋅ and u( x b ) = b ⋅ (135)
n − 3 ng
nb − 3 nb
g
NOTE Obviously, ng , nb > 3 is required.
Then the standard uncertainty associated with the primary measurement result y is calculated by:
u2 ( y ) = w 2 ⋅[u2 ( x g ) + u2 ( x b )] + y 2 ⋅ urel
2
( w ) (136)
The PDFs of the calibration factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a detailed description and
ISO 11929-1:2019, A.4 for the general methodology.
The background effect was obtained by repeated nb readings. The values of the individual readings x b,i
are given in Table 17. One calculates the mean background measurement by
nb
∑ x b ,i = 75,70 nSv ⋅ h−1 (137)
1
xb =
nb i =1
with a standard deviation
1/2
1 nb
sb =
nb − 1 i =1
∑
( x b ,i −x b )2
= 5 , 90 nSv ⋅ h −1 (138)
The gross measurement was obtained by repeated ng readings. The values of the individual readings
x g,i are given in Table 17. One calculates the mean gross effect by
ng
The standard uncertainty associated with the primary measurement result is given by:
u( y ) = w 2 ⋅[u2 ( x g ) + u2 ( x b )] + y 2 ⋅ urel
2
( w ) = 37 , 71 nSv ⋅ h −1 (144)
The standard uncertainty as a function of an assumed true value is not explicitly available. Therefore,
the following procedure is applied.
For a true value of the measurand y =0 one expects with ISO 11929-1:2019, Formula (A.24)
ng −1 nb −1 2
u2 ( y = 0 ) = w2 ⋅ + ⋅ sb (145)
ng ⋅( ng −3) nb ⋅( nb −3)
and the decision threshold
ng −1 nb −1
y* = k1−α ⋅w ⋅ sb ⋅ + (146)
ng ⋅( ng −3) nb ⋅( nb −3)
According to ISO 11929-1:2019, 5.5, the following linear interpolation often suffices, if only u(
0 ) , one
result, y > 0 , and its associated standard uncertainty, u( y ) , are known
u 2 ( y ) = u 2 ( 0 ) ⋅ (1 − y y ) + u2 ( y ) ⋅ y y (147)
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the dose rate is calculated by:
ng − 1 nb − 1
y * = k1−α ⋅ w ⋅ sb ⋅ + = 2 , 72 nSv ⋅ h −1 (148)
ng ⋅ ( ng − 3) nb ⋅ ( nb − 3)
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The linear interpolation according to Formula (147) leads with the
auxiliary quantity a according to Formula (149)
a = k1−α ⋅ u( 0 ) +
1
2
{( k12−β y ) (u2 ( y ) − u2 (0)} (149)
to the detection limit
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result y exceeds the decision threshold y * . With ω = Φ [ y / u( y )] = 0 , 999 and γ = 0 , 05
one obtains the probabilities by p = ω ⋅ (1 − γ / 2) = 0 , 974 and q = 1 − ω ⋅ γ / 2 = 0 , 975 , and hence the
quantiles are k p = 1 , 944 and kq = 1 , 960 . Then the lower and upper limits of the probabilistically
symmetric coverage interval are given by:
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ y / u( y )] = 0 , 999 0 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 974 5 , and hence the quantile kp is
equal to 1,96. With this, the limits of the shortest coverage interval according to Formula (152) are
ˆy = y +
{
u ( y ) ⋅ exp − y2 2u2 ( y )
}
= 116, 67 nSv ⋅ h−1 (153)
ω 2π
with its associated standard uncertainty u( ˆ)
y
Lower limit of the shortest coverage interval y< nSv h-1 42,93 43,78
Upper limit of the shortest coverage interval y> nSv h-1 190,16 191,80
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 9 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The primary measurement result y = 116 , 55 nSv ⋅ h-1 exceeds the decision threshold y * = 2 , 72 nSv ⋅ h-1
. It is decided to conclude that a difference between the two investigated areas was recognized.
— The detection limit y # = 38 , 39 nSv ⋅ h-1 is below the guideline values y r = 50 nSv ⋅ h -1 . It is decided to
conclude that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are y = 43 , 25 nSv ⋅ h-1
and y = 190 , 48 nSv ⋅ h-1 . They are equal to those of the shortest coverage interval.
— The best estimate is ˆy = 116, 67 nSv ⋅ h-1 with an associated standard uncertainty u( ˆ)
y = 37, 52 nSv ⋅ h-1
.
The model of evaluation is linear, and there are no dominating large uncertainties so that the application
of ISO 11929-1 is justified. This is a case with moderate relative uncertainties. The results obtained by
application of ISO 11929-2 are practically identical with those obtained by ISO 11929-1:2019, Table 18.
The ISO/IEC Guide 98-3 approximations, assuming normal PDFs and using a Taylor expansion truncated
after the linear term, hold.
ng n0 rg − r0
am = − ⋅w = = ( rg − r0 ) ⋅ w (155)
t
g t0 m ⋅ κ ⋅ε
where
1
w= (156)
m ⋅ κ ⋅ε
The mean value of the background counts n0 and the square of its standard uncertainty u2 ( n0 ) of the
values n0,i are given by:
m0 m0
1 m0 − 1
∑ n0 ,i ∑
1 1
n0 = ; u ( n0 ) =
2
n0 + n0 + ( n0 ,i − n0 )2 (157)
m0 m0 m0 − 3 m0 − 3 i =1
i =1
The mean value of the gross counts ng and its uncertainty u2 ( ng ) of the values ng ,i are given by:
mg mg
1 mg − 1
∑ ∑
1 1
ng = ng ,i ; u ( ng ) =
2
ng + ng + ( ng ,i − ng )2 (158)
mg i =1
mg mg − 3 mg − 3 i =1
NOTE Obviously, m0 , mg > 3 is required.
Table 19 (continued)
Standard
Quantity Symbol Value PDF
uncertainty
Number of recorded events for the 966; 676; 911;
blanks (background effect)
n0,i
856; 676
N[ n0 ;n0 , u( n0 )]
Measurement durations tg , t 0 30 000 s -
Sample mass (general) m with u(m) 0,100 kg 0,001 kg N[ m
;m , u( m )]
Detection efficiency κ with u(κ ) 0,510 0,020 s-1 Bq N[κ ;κ , u(κ )]
Chemical yield ε with u( ε ) 0,570 0,040 N[ε ;ε , u( ε )]
Intermediate values
Quantity Symbol Value
Mean values n0 , ng 817,00; 2 039,6
Empirical standard deviations s 0 , sg 288,14; 134,46
Standard uncertainty u( n0 ), u( ng ) 87,88; 185,57
The PDFs of the calibration factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3 for a detailed description and
ISO 11929-1:2019, A.2 for the general methodology.
The count rate of the background effect has been measured by repeated mb analyses of blank samples.
The numbers of counted pulses n0,i are given in Table 19. One calculates the mean background count
rate by
n0
r0 = = 0 , 068 0 s −1 (159)
t0
The standard uncertainty associated with the background count rate is given by:
u2 ( n0 )
u ( r0 ) = = 0 , 002 93 s −1 (160)
t 02
The mass–related activity is calculated from Formula (155). This Formula depends on the calibration
factor, the input quantities of which and their associated standard uncertainties were determined
independently.
urel ( w ) = urel
2
( m ) + urel
2
( ε ) + urel
2
(κ ) = 0 , 081 0 (162)
which yields
u( w ) = 2 , 79 s Bq ⋅ kg −1 (163)
The standard uncertainty u( am ) associated with the primary measurement result am is calculated
by:
u( am ) = w 2 ⋅ u2 ( ng ) t g2 + u2 ( n0 ) t 02 ) + am
2
⋅ urel
2
( w )= 0 , 261 Bq ⋅ kg −1 . (165)
The standard uncertainty as a function of an assumed true value is not explicitly available. Therefore,
the following procedure is applied
For a true value of the measurand y = 0 one expects x g = x b and hence one receives
u2 ( 0 ) = 2 ⋅ w 2 ⋅ u2 ( r0 ) (166)
The value of the quantile k1−α of the standardised normal distribution Φ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is then calculated by:
am
*
= k1−α ⋅ u( 0 ) = k1−α ⋅ w ⋅ 2 ⋅ u2 ( n0 ) t 02 = 0 , 234 Bq ⋅ kg −1 (168)
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The linear interpolation according to Formula (167) and the auxiliary
quantity a according to Formula (169)
a = k1−α ⋅ u( 0 ) +
1
2
{( k12−β ) }
am ,1 ( u2 ( y1 ) − u 2 ( 0 ) (169)
am
#
= a + a2 + ( k12− β − k12−α ) u 2 ( 0 ) = 0 , 546 Bq ⋅ kg −1 (170)
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result am exceeds the decision threshold am *
. With ω = Φ [ am / u( am )] = 0 , 999 and
γ = 0 , 05 one obtains the probabilities p = ω ⋅ (1 − γ / 2) = 0 , 975 and q = 1 − ω ⋅ γ / 2 = 0 , 975 , and hence the
quantiles k p and kq are equal to 1,96 and 1,96, respectively. Then, the lower and upper limits of the
probabilistically symmetric coverage interval are given by:
am = am − k p ⋅ u ( am ) = 0 , 890 Bq ⋅ kg −1 and am
= am + kq ⋅u ( am ) = 1 , 91 Bq ⋅ kg −1 (171)
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ am / u( am )] = 0 , 999 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 975 , and hence the quantile k p is equal
to 1,96. With this, the limits of the shortest coverage interval according to Formula (172) are
<
am = am − k p ⋅ u ( am ) = 0 , 890 Bq ⋅ kg −1 and am
>
= am + k p ⋅u ( am ) = 1 , 91 Bq ⋅ kg −1 (172)
The best estimate âm of the activity of the sample is given by:
aˆm = am +
u ( am ) ⋅ exp − am {2u ( am )
2 2
}
= 1, 40 Bq ⋅ kg −1 (173)
ω 2π
with its associated standard uncertainty u( aˆm )
Table 20 (continued)
Results Symbol Unit ISO 11929-1 ISO 11929-2
Standard uncertainty associated with the primary
u( am ) Bq·kg−1 0,261 0,265
result
Relative standard uncertainty associated with the
urel ( am ) 1 0,186 0,188
primary result
Decision threshold am
* Bq·kg−1 0,234 0,237
Detection limit am
# Bq·kg−1 0,546 0,545
Best estimate âm Bq·kg−1 1,40 1,411
Standard uncertainty associated with the best esti-
u( aˆm ) Bq·kg−1 0,261 0,264
mate
Relative uncertainty associated with the best esti-
urel ( aˆm ) 1 0,186 0,187
mate
Lower limit of the probabilistically symmetric cover-
am Bq·kg−1 0,890 0,918
age interval
Upper limit of the probabilistically symmetric cover-
am Bq·kg−1 1,91 1,96
age interval
Lower limit of the shortest coverage interval < Bq·kg−1 0,890 0,902
am
Upper limit of the shortest coverage interval > Bq·kg−1 1,91 1,94
am
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 10 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
— The best estimate is aˆm = 1, 40 Bq ⋅ kg-1 with an associated standard uncertainty u( âm ) = 0, 26 Bq ⋅ kg-1
.
The model of evaluation is linear, and there are no dominating large uncertainties so that the application
of ISO 11929-1 is justified. This is a case with small or minor relative uncertainties. The results obtained
by application of ISO 11929-2 are practically identical with those obtained by ISO 11929-1:2019,
Table 20. The ISO/IEC Guide 98-3 approximations, which assume normal PDFs and make use of a Taylor
expansion truncated after the linear term, hold.
Note that in spite of the fact that according to the comparison of the detection limit and the guideline
value the measurement procedure is not suited for the measurement purpose, an effect of the sample
has been recognized. The primary measurement result and its associated standard uncertainty can
well be used to calculate limits of coverage intervals and the best estimate and its associated standard
uncertainty.
the sample treatment from measurements on reference samples or on other samples. The latter samples
should be similar to the current samples and be measured under similar conditions, in order that
they can be taken as reference samples although they need not be examined specifically for reference
purposes. Since the influence of the sample treatment is now known one measurement, each of a sample
and a blank is sufficient in the actual measurement. This is an example according to the general model
of ISO 11929-1:2019, A.3.
This procedure, appropriate when small random influences are present, is based on the approach
n + u2 n + ϑ 2 ⋅ n 2
u2 ( n ) = = with ϑ = u / n (175)
m m
m
m −1
∑
1
and with the auxiliary quantity u = 2
n+ ( ni − n )2 for any counts of the measurement
m −3 m − 3 i =1
procedure.
The influence parameter ϑ can be calculated from the data of counting measurements of mr reference
samples by
ϑ 2 = ( mr ⋅ u2 ( nr ) − nr ) nr2 or ϑ = ur / nr (176)
If ϑ 2 < 0 results, the approach and the data are not compatible. The number mr of the reference samples
should then be enlarged or ϑ = 0 shall be set. Moreover, ϑ < 0 , 2 should be obtained. Otherwise, one can
proceed according to the example in Clause 14.
ng n0 rg − r0
am = − ⋅w = = ( rg − r0 ) ⋅ w (177)
t
g t0 m ⋅ κ ⋅ε
where
1
w= (178)
m ⋅ κ ⋅ε
Then the standard uncertainties of the background and the gross count rate are given by:
The relative standard uncertainty of the calibration factor urel ( w ) = u( w )/ w is calculated by:
urel
2
( w ) = urel
2
( m ) + urel
2
( ε ) + urel
2
(κ ) (180)
u( am ) = w 2 ⋅ u2 ( rg ) + u2 ( r0 ) + am
2
⋅ urel
2
( w ) (181)
Table 21 (continued)
Quantity Symbol Data
Detection efficiency κ with u(κ ) 0,51 s-1 Bq-1 0,02 s-1 Bq-1 N[κ ;κ , u(κ )]
Chemical yield ε with u( ε ) 0,57 0,04 N[ε ;ε , u( ε )]
Intermediate values
Quantity Symbol Value
Mean values nr 73 946,5
Auxiliary quantity ur 10 771,3
Influence parameter ϑ = ur / nr 0,146
The PDFs of the calibration factor and of the count rates were chosen according to the
ISO/IEC Guide 98-3:2008/Suppl.1; see also ISO 11929-2:2019, 6.3, for a detailed description and
ISO 11929-1:2019, A.3 for the general methodology.
By analysing mr reference samples, the relative uncertainty of the sample treatment was determined
by
ϑ = ur / nr (182)
m
mr − 1 r
∑
1 1
ϑ= nr + ( nr ,i − nr )2 = 0 , 146 (184)
nr mr − 3 mr − 3 i =1
The count rate of the background effect has been measured by an analysis of a blank sample. The
number of counted pulses n0 is given in Table 21. One calculates the background count rate by
n0
r0 = = 0 , 027 2 s −1 (185)
t0
Its standard uncertainty associated with the background count rate is given by:
n0 + ϑ 2 ⋅ n02
u ( r0 ) = = 0 , 004 08 s −1 (186)
t 02
The mass–related activity is calculated from Formula (177). This formula depends on the calibration
factor the input quantities of which and their associated standard uncertainties were determined
independently.
urel ( w ) = urel
2
( m ) + urel
2
( ε ) + urel
2
(κ ) = 0 , 081 0 (188)
which yields
u( w ) = 2 , 79 s Bq ⋅ kg −1 (189)
ng n0
am = − ⋅ w = 1 , 40 Bq ⋅ kg −1 (190)
t
g t0
The standard uncertainty u( am ) associated with the primary measurement result am is calculated as
The standard uncertainty u( a m ) as a function of the true value a m of the measurand is needed to
calculate the decision threshold and the detection limit.
For an assumed true value of the measurand a m one expects n g = a m w + n0 t 0 and one obtains with
Formula (181) and u2 ( n0 ) and u2 ( ng ) from Formula (179)
2 2
2n0 ⋅ϑ 2 1 n0 n02 ⋅ϑ 2 n0 + ϑ 2 ⋅ n02
u( a m ) = a m ϑ + u2
( w ) + a
⋅ w + + w 2
+ + (192)
rel m t t g t0 ⋅ t g t 02 t 02
0
The value of the quantile k1−α of the standardised normal distribution Φ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold of the activity is calculated by:
n n2 ⋅ϑ 2 n0 + ϑ 2 ⋅ n02
am
*
= k1−α w 2 0 + 0 + = 0 , 327 Bq ⋅ kg −1 (193)
t0 ⋅ t g t 02 t 02
The measured primary result am exceeds the decision threshold am
*
.
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit is calculated by:
am
#
= am
*
+
#2 2
2n0 ⋅ϑ 2 1 n0 n02 ⋅ϑ 2 n0 + ϑ 2 ⋅ n02
k1− β am ϑ + u2
( w ) + a #
⋅ w + + w 2
+ + (194)
rel m t t g t0 ⋅ t g t 02 t 02
0
= 0 , 826 Bq ⋅ kg −1
The guideline value am,r exceeds the detection limit am
#
.
The lower and upper limits of the probabilistically symmetric coverage interval are calculated since the
measurement result am exceeds the decision threshold am * . With ω = Φ [ a / u( a )] = 1 , 00 and
m m
γ = 0 , 05 one obtains the probabilities by p = ω ⋅ (1 − γ / 2) = 0 , 975 and q = 1 − ω ⋅ γ / 2 = 0 , 975 , and hence
the quantiles k p and kq are both equal to 1,96. Then, the lower and upper limits of the probabilistically
symmetric coverage interval are given by:
am = am − k p ⋅ u ( am ) = 0 , 641 Bq ⋅ kg −1 and am
= am + kq ⋅u ( am ) = 2 , 165 Bq ⋅ kg −1 (195)
Alternatively, the lower and upper limits of the shortest coverage interval are calculated. With
ω = Φ [ am / u( am )] = 1 , 00 , one obtains p = [1 + ω ⋅ (1 − γ )]/ 2 = 0 , 975 , and hence the quantile k p is equal
to 1,96. With this, the limits of the shortest coverage interval according to Formula (196) are
<
am = am − k p ⋅ u ( am ) = 0 , 640 Bq ⋅ kg −1 and am
>
= am + k p ⋅u ( am ) = 2 , 16 Bq ⋅ kg −1 (196)
The best estimate âm of the activity of the sample is given by:
aˆm = am +
{
u ( am ) ⋅ exp − am 2u ( am )
2 2
}
= 1, 40 Bq ⋅ kg −1 (197)
ω 2π
with its associated standard uncertainty u( aˆs )
15.5 Documentation of the results obtained by ISO 11929-1 and ISO 11929-2
The results and characteristic limits are summarized in Table 22 and the PDFs obtained by the
Monte Carlo simulations are presented in Figure 11.
Table 22 (continued)
Results Symbol Unit ISO 11929-1 ISO 11929-2
Relative standard uncertainty associated with
urel ( am ) 1 0,277 0,278
the primary result
Decision threshold am
* Bq·kg−1 0,327 0,330
Detection limit am
# Bq·kg−1 0,826 0,823
Best estimate âm Bq·kg−1 1,40 1,41
Standard uncertainty associated with the best
u( aˆm ) Bq·kg−1 0,389 0,393
estimate
Relative uncertainty associated with the best
urel ( aˆm ) 1 0,277 0,278
estimate
Lower limit of the probabilistically symmetric
am Bq·kg−1 0,641 0,665
coverage interval
Upper limit of the probabilistically symmetric
am Bq·kg−1 2,17 2,210
coverage interval
Lower limit of the shortest coverage interval < Bq·kg−1 0,640 0,647
am
Upper limit of the shortest coverage interval > Bq·kg−1 2,16 2,190
am
NOTE The blue PDFs are the respective Gaussian approximations. The blue vertical lines indicate the best
estimate and the limits of the coverage interval (upper panel), the decision threshold (middle panel) and the
decision threshold and the detection limit (lower panel).
Figure 11 — PDFs calculated by Monte Carlo methods according to ISO 11929-2 in red colour
.
The model of evaluation is linear, and there are no dominating large uncertainties so that the application
of ISO 11929-1 is justified. This is a case with small or minor relative uncertainties. The results obtained
by application of ISO 11929-2 are practically identical with those obtained by ISO 11929-1:2019,
Table 22. The ISO/IEC Guide 98-3 approximations, assuming normal PDFs and using a Taylor expansion
truncated after the linear term, hold.
M b background dose rate during the time of exposure in µSv per hour;
te time of exposure in h;
N0
w= composite calibration factor;
rrel
u constant standard uncertainty of Mg .
The standard uncertainty u[ H p (10 )] associated with the personal dose equivalent H p (10 ) is given by:
u2 ( H p (10 )) = u2 ( M g − M b ⋅ t e ) ⋅ w 2 + ( M g − M b ⋅ t e ) 2⋅ u2 ( w )
(200)
= u ( M g ) ⋅ w + u ( M b ) ⋅ t e2 ⋅ w 2 + H p (10 ) 2⋅ urel
2 2 2 2
(w )
with
urel
2
( w ) = urel
2
( N0 ) + urel
2
( rrel ) (201)
With regard to the calculation of the decision threshold and the detection limit, the gross indication
M g is the analogous quantity to the gross count rate x1 in the general model of evaluation given in
ISO 11929-1. However, the actual example is not a counting measurement and demonstrates that the
possible application of ISO 11929 (all parts) extends far beyond counting measurements. The particular
feature of this example is that the gross indication M g has a constant uncertainty u , which allows
calculating the standard uncertainty of the measurand as a function of an assumed true value in 16.4.3.
The background dose rate M b is set by assuming an average ambient dose rate of 0,1 µSv per hour. The
standard uncertainty was obtained from a Gaussian distribution, since the value was determined by
several repeated measurements resulting in u( M b ) = 0 , 016 0 µ Sv ⋅ h −1 .
With the data of Table 23 the primary result of the personal dose equivalent is calculated to be
N
H p (10 ) = ( M g − M b ⋅ t e ) ⋅ 0 = ( M g − M b ⋅ t e ) ⋅ w = 19 , 20 µSv (202)
rrel
and its associated standard uncertainty
The standard uncertainty u[ H p (10 )] as a function of the true value H p (10 ) of the measurand is needed
to calculate the decision threshold and the detection limit.
This function is available in this example. Since u( M g ) = u = const. , u[ H p (10 )] is given by:
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold is calculated from Formula (204). With the data
from Table 23 the decision threshold is
The measurement result H p (10 ) exceeds the decision threshold H p (10 )* and hence a personal dose
equivalent greater than zero has been detected.
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit follows from Formula (204) with the decision threshold
as given in Formula (205).
The implicit Formula (206) is solved by iteration. As the initial value for the iteration is selected
2 ⋅ H p* (10 ) ≈ 1 µ Sv . This iteration is almost always convergent; otherwise the measurement procedure
is completely unsuitable. Using the data from Table 23 the detection limit is
Since α = β , k1−α = k1− β = k . The detection limit can also be calculated by the explicit Formula (208):
2 ⋅ H p (10 )*
H p (10 )# = = 1,27 µSv (208)
1 − k 2 ⋅ urel
2
(w )
The detection limit is compared with the limit for legal or other requirements. If the detection limit is
smaller than the limit, the method for determining the photon dose is useful in the personal dosimetry.
The limits of the probabilistically symmetric coverage interval, H p (10 ) and H p (10 ) and γ = 0 , 05 of
the personal dose equivalent is calculated for the probability 1 - γ = 0,95 for the obtained value of the
personal dose equivalent H p (10 ) = 19 , 20 µ Sv with the associated standard uncertainty
u[ H p (10 )] = 4 , 97µ Sv . From the standard normal distribution results the parameter
and with
p = ω ⋅ (1 − γ / 2) = 0 , 975 and q = 1 − ω ⋅ γ / 2 = 0 , 975 (210)
and the quantiles of the standardised normal distribution k p = 1,96 and kq = 1,96 , respectively, one
obtains the limits of the probabilistically symmetric coverage interval
and
The limits of the shortest coverage interval, H p< (10 ) and H p> (10 ) , of the personal dose equivalent is
calculated for the probability 1 - γ = 0,95 for the obtained value of the personal dose equivalent
H p (10 ) = 19 , 2 µ Sv with the associated standard uncertainty u[ H p (10 )] = 5 , 0 µ Sv .
With ω = Φ (( H p (10 )/ u( H p (10 ))) = Φ (3 , 85) = 1 , 00 , the limits of the shortest coverage interval are
identical to those of the probabilistically symmetric coverage interval.
The calculation of the best estimate and its associated standard uncertainty can be omitted since
{ }
ω = Φ H p (10 )/ u[ H p (10 )] = Φ (3 , 85) = 1 , 000 . Consequently, the best estimate and its associated
standard uncertainty are equal to the primary measurement result and its associated standard
uncertainty, respectively. The results are given in Table 24 anyway.
NOTE The blue PDFs are the respective Gaussian approximations according to ISO 11929-1. The blue vertical
lines indicate the best estimate and the limits of the probabilistically symmetric coverage interval (upper panel),
the decision threshold (middle panel) and the decision threshold and the detection limit (lower panel).
— The detection limit H p# (10 ) = 1 , 3 µ Sv is below the guideline values H p,r (10 ) = 10 µ Sv . It is concluded
that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are H p (10 ) = 9 , 5 µ Sv
and H p (10 ) = 28 , 9 µ Sv .
— The lower and upper limits of the shortest coverage interval are calculated as H p< (10 ) = 9 , 5 µ Sv and
H p> (10 ) = 28 , 9 µ Sv .
— The best estimate is Hˆp (10) = 19, 2 µ Sv with an associated standard uncertainty
u[ Hˆp (10)] = 4, 9 µ Sv .
Because the model of evaluation is not linear, the results obtained by application of ISO 11929-2 differ
from those obtained by ISO 11929-1:2019, Table 23. The GUM approximations, assuming normal PDFs
and using a Taylor expansion truncated after the linear term, is not sufficient (see Figure 12) and the
application of ISO 11929-2 is to be preferred.
There are also differences between the limits of the probabilistically symmetric coverage interval and
those of the shortest coverage interval. The decision, which of the coverage intervals should be used for
further evaluations, depends on the regulations and is not a matter of this document. The best estimate
and its associated standard uncertainty are equal to the primary measurement result and its associated
standard uncertainty.
The model function for the gross indication is M g = M net + M CR + M ib with the indication due to the net
H * (10 )CR
dose rate, M net , the indication due to cosmic radiation, M CR = , and the indication due to
NCf ⋅ K E ,CR
the intrinsic background of the monitor M ib .
The dose rate due to cosmic neutron radiation at the place and time of the measurement varies with the
actual air pressure p and with the solar activity (sa). These two effects are considered with correction
factors, K p for the influence caused by the actual air pressure, and K sa for the influence due to the
actual time of the measurement with a certain solar activity (sa) H * (10 ) = H * (10 )
CR ⋅K ⋅K ) .
0,CR p sa
The dose rate due to cosmic neutron radiation at reference conditions of air pressure (p0) and solar
activity (sa) at the position of the measurement in the workplace is H * (10 )0,CR . In this example, a value
of H * (10 )
0,CR = 11 , 2 nSv ⋅ h-1 ± 1 , 1 nSv ⋅ h-1 is used, which is based on a measurement of the cosmic
neutron component with a Bonner sphere spectrometer for a reference pressure of
p0 = (1 004 , 0 ± 1 , 0 ) hPa at the reference date 2005-08-09 at a position that can be seen as being
representative for the measurement in the workplace. The correction factor due to air pressure is
K p = 1 − c b ⋅ ( p − p0 ) .
The actual measurement in the workplace was performed 2018-11-15. For this date, the cosmic neutron
radiation was higher compared to 2005 and a correction factor K sa has to be used to correct for this
effect: K sa = count rate 2018-11-15/count rate 2005-08-09 .
This correction factor can be determined from the datasets of cosmic neutron monitors, which are
distributed worldwide (http://w ww01.nmdb.eu/station/). A station similar in geomagnetic cut off
rigidity to the position of the workplace should be chosen, if possible.
The barometric coefficient c b can be extracted from literature data or is given by the operators of
specific cosmic neutron monitors, respectively.
To calculate the dose rate in the workplace field from the background corrected reading of the monitor
M net , the instrument specific calibration factor NCf in a reference field, e.g. 252Cf and a correction
factor for the instruments response in the actual workplace field K E ,net have to be considered. From
this, the model equation for the net dose rate follows with the correction factor for the actual neutron
spectrum K E ,net :
with
H * (10 )CR
M CR = (214)
NCf ⋅ K E ,CR
K p = 1 − c b ⋅ ( p − p0 ) (216)
H * (10 )net
M net = (217)
NCf ⋅ K E ,net
H * (10 ) 0,CR ⋅ K p ⋅ K sa
H * (10 ) net = M g − − M ib ⋅ NCf ⋅ K E ,net
NCf ⋅ K E ,CR (218)
= ( M − M ) ⋅ w
g b
with
H * (10 )0,CR ⋅ K p ⋅ K sa
M b = + M ib (219)
NCf ⋅ K E ,CR
and
w = NCf ⋅ K E ,net (220)
M CR indication due to cosmic radiation at the time and position of the measurement
H * (10 )CR dose rate due to cosmic radiation at the time and position of the measurement
H * (10 )0,CR dose rate due to cosmic radiation at reference conditions of air pressure ( p0 ) and solar
activity (sa) at the position of the measurement
To facilitate the calculations, first M CR and u( M CR ) are calculated, then w and u( w ), as well as M b
and u( M b ) .
H * (10 )0,CR ⋅ K p ⋅ K sa
M CR = (221)
NCf ⋅ K E ,CR
urel
2
( M CR ) = urel
2
[ H * (10 )0,CR ] + urel
2
( K p ) + urel
2
( K sa ) + urel
2
( NCf ) + urel
2
( K E ,CR ) (222)
urel
2
( w ) = urel
2
( NCf ) + urel
2
( K E ,net ) (224)
H * (10 )0,CR ⋅ K p ⋅ K sa
M b = + M ib = M CR + M ib (225)
NCf ⋅ K E ,CR
u2 ( M b ) = u2 ( M CR ) + u2 ( M ib ) (226)
After this, the calculations are performed for the model of evaluation in the simple form
With regard to the calculation of the decision threshold and the detection limit, the gross indication
M g is the analogous quantity to the gross count rate x1 in the general model of evaluation given in
ISO 11929-1. However, the actual example is not a counting measurement with given gross and
background count numbers. It demonstrates that the possible application of ISO 11929 (all parts)
extends far beyond mere counting measurements. The particular feature of this example is that the
gross indication M g for a constant measurement time is proportional to the unknown number of counts
ng .
1 1
Consequently, for the relative uncertainty, urel ( M g ) , follows urel ( M g ) ∝ ∝ .
ng M g
This allows calculating the relative uncertainty urel ( M g ) for any reading and thereby calculating the
standard uncertainty of the measurand as a function of an assumed true value in 17.4.3.
The indication due to the intrinsic background of the monitor was determined independently with an
average of M ib = 2 , 5 nSv ⋅ h −1 and a standard uncertainty of u( M ib ) = 0 , 5 nSv ⋅ h −1 . A Gaussian PDF with a
coefficient of variation of 20 % is assumed.
The dose rate due to cosmic neutron radiation at the time of the measurement with an actual air
pressure p is H * (10 )CR = H * (10 )0,CR ⋅ K p ⋅ K sa . For H * (10 )0,CR a value of 11 , 2 nSv ⋅ h −1 with an
associated standard uncertainty of 1 , 1 nSv ⋅ h −1 was obtained from a previous measurement with a
Bonner sphere spectrometer. The correction factor due to air pressure is calculated with the help of the
barometric coefficient c b from the difference of the actual air pressure p and the reference air pressure
p0 : K p = 1 − c b ⋅ ( p − p0 ) .
This results in a reading of the instrument due to cosmic neutron radiation
H * (10 )0,CR ⋅ K p ⋅ K sa
MCR = = 6 , 04 nSv ⋅ h −1 with NCf being the calibration factor of the instrument in
NCf ⋅ K E ,CR
the reference field of a reference neutron source, e.g. 252Cf, and K E ,CR being the field correction factor
of the instrument in the field of cosmic neutrons. As neutron area dosemeters are usually designed for
the neutron energy range up to 20 MeV, these devices show typically a large underestimation of the
dose rate caused by the spectrum of cosmic neutrons. The factor of K E ,CR = 1 , 81 was actually determined
for the measurement device without lead shield. A value of K E ,CR in the order of 2 is typical for all
moderator based devices, which do not contain additional heavy metal in or nearby the moderator to
enlarge the response of the instrument at neutron energies higher than 20 MeV. The standard
uncertainty associated with M CR is calculated as u( M CR ) = 0 , 690 nSv ⋅ h −1 .
NOTE The results in this example depend on the calibration source used. They are different if an 241Am/Be
source is used for calibration instead of a 252Cf source. A different calibration source also results in a different
value of K E ,cr .
With the data of Table 25 the primary result of the ambient dose equivalent is calculated to be
H * ( 10 ) 0,CR ⋅ K p ⋅ K sa
H * ( 10 ) net = ( M g − − M ib ) ⋅ NCf ⋅ K E ,net
NCf ⋅ K E ,CR (229)
= ( M g − M b ) ⋅ w = 30 , 20 nSv ⋅ h -1
and its associated standard uncertainty
( )
The standard uncertainty u H * (10 )net as a function of the true value H * (10 )net of the measurand is
needed to calculate the decision threshold and the detection limit.
For H * (10 )net one expects M g = H * (10 )net / w + M b . Given a constant measurement time, for any
2
M g
H (10 )net the corresponding indication M g has the standard uncertainty u ( M g ) = u ( M g ) ⋅
* 2
.
M g
Then the function u H * (10 )net is given by:
M g
u H (10 ) net = u ( M g ) ⋅
2 * 2
+ u2 ( M b ) ⋅ w 2 + H * (10 ) net2⋅ urel
2
(w ) =
Mg
H * (10 ) net (231)
+ MCR + M ib
= u2 ( M g ) ⋅ w + u2 ( M b ) ⋅ w 2 + H * (10 ) net2⋅ urel
2
(w )
M g
17.4.4 Decision threshold
The value of the quantile k1−α of the standardised normal distribution Φ ( k1−α ) = 1 − α is 1,645
assuming a probability α of 5 %. The decision threshold is calculated from Formula (231) with use of
Formula (232). With the data from Table 25 the decision threshold is
M + M ib
H * (10 )net
*
= k1−α ⋅ u ( 0) = k1−α ⋅ w ⋅ u2 ( M g ) ⋅ CR + u2 ( M b ) = 1 , 93 nSv ⋅ h-1 (232)
Mg
The measurement result H * (10 )net exceeds the decision threshold H * (10 )*net and hence an ambient
dose equivalent greater than zero has been detected.
The probability β is set to 5 %. Then, the value of the corresponding quantile k1− β of the standardised
normal distribution is 1,645. The detection limit follows from Formula (231) with the decision threshold
as given in Formula (232)
H * (10 ) #net
+ MCR + M ib
+ k1− β ⋅ u2 ( M g ) ⋅ w + u2 ( M b ) ⋅ w 2 + H * (10 ) #net 2⋅ urel
2
(w )
M g
The implicit Formula (233) is solved by iteration. As the initial value for the iteration is selected
2 ⋅ H * ( 10 )*net ≈ 3 , 8 nSv ⋅ h -1 . This iteration is almost always convergent; otherwise the measurement
procedure is completely unsuitable. Using the data from Table 25 the detection limit is
Since α = β , k1−α = k1− β = k . The detection limit can also be calculated by the explicit Formula (235):
2 ⋅ H * (10 )*net
H * (10 )#net = = 4,0 nSv ⋅ h-1 (235)
1 − k ⋅ urel ( w )
2 2
The detection limit is compared with the limit for legal or other requirements. If the detection limit is
smaller than the limit, the method is suitable for the measurement purpose.
The limits of the probabilistically symmetric coverage interval, H * (10 )net and H * (10 )net , of the
ambient dose equivalent rate is calculated for the probability 1 - γ = 0,95. From the standard normal
distribution result the parameters
and with
p = ω ⋅ (1 − γ / 2) = 0 , 975 and q = 1 − ω ⋅ γ / 2 = 0 , 975 (237)
and the quantiles of the standardised normal distribution kq = 1,96 respectively k p = 1,96 one obtains
the limits of the probabilistically symmetric coverage interval
and
H * (10) * * −1
net = H (10)net + k p ⋅ u( H (10)net ) = 34 , 81 nSv ⋅ h (239)
The limits of the shortest coverage interval, H * (10 )<net and H * (10 )net
>
are calculated for the probability
1 – γ = 0,95.
With ω = Φ H * (10 )net / u( H * (10 )net = Φ (13 , 1 ) = 1 , 00 , the limits of the shortest coverage interval are
identical to those of the probabilistically symmetric coverage interval.
The calculation of the best estimate and its associated standard uncertainty can be omitted since
ω = Φ H * (10 )net / u( H * (10 )net = Φ (13 , 1 ) = 1 , 00 . Consequently, the best estimate and its associated
standard uncertainty are equal to the primary measurement result and its associated standard
uncertainty, respectively. The results are given in Table 26 anyway.
NOTE The blue PDFs are the respective Gaussian approximations according to ISO 11929-1. The blue vertical
lines indicate the best estimate and the limits of the probabilistically symmetric coverage interval (upper panel),
the decision threshold (middle panel) and the decision threshold and the detection limit (lower panel).
— The primary measurement result H * (10 )net = 30 , 2 nSv ⋅ h -1 exceeds the decision threshold
H * (10 )*net = 2 , 0 nSv ⋅ h -1 . It is decided to conclude that an ambient dose equivalent greater than zero
was recognized.
— The detection limit H * (10 )#net = 4 , 2 nSv ⋅ h -1 is below the guideline values H r* (10 )net = 5 nSv ⋅ h -1 . It is
concluded that the measurement procedure is suited for the measurement purpose.
— The lower and upper limits of the probabilistically symmetric coverage interval are
H * (10 )net = 25 , 6 nSv ⋅ h -1 and H * (10 )net = 34 , 8 nSv ⋅ h -1 .
— The lower and upper limits of the shortest coverage interval are identical to those of the
probabilistically symmetric coverage interval.
— The best estimate and its associated standard uncertainty are identical to the primary result and its
associated standard uncertainty.
Because the relative uncertainty of the primary estimate as well as of the input quantities are small,
the results obtained by application of ISO 11929-2 are practically identical with those obtained
by ISO 11929-1:2019, Table 26. The GUM approximations, assuming normal PDFs and using a Taylor
expansion truncated after the linear term, hold.
The probabilistically symmetric coverage interval is equal to the shortest coverage interval. The best
estimate and its associated standard uncertainty are equal to the primary measurement result and its
associated standard uncertainty. According to of ISO 11929-2:2019, Figure B.3, this holds of for all
results with H * (10 )net / u[ H * (10 )net ] > 5 .
Annex A
(informative)
urel ( w ) = urel
2
( aK ) + urel
2
( rK,n ) (A.2)
with
rK,0 count rate of the background effect of the calibration source, in s–1
The activity of the calibration source is given by ak = 25 , 035 kBq with the standard uncertainty
u( ak ) = 0 , 015 kBq at the date of the measurement. Additional data are shown in Table A.1.
The background effect of the calibration source has been measured. The number of the counted pulses
is nk ,0 = 1 381 . This gives a count rate of
nK,0
rK,0 = = 2 , 302 s-1 (A.3)
t K,0
The standard uncertainty of the count rate is given by
nK,0
u ( rK,0 ) = = 0 , 061 9 s-1 (A.4)
t K,
2
0
The gross effect of the calibration source has been measured. The number of counted impulses is
nk ,g nk ,0
nK,g = 4 932 . This leads to a net count rate of rK,n = − = 5 , 918 s −1 . Then, the calibration factor is
t k ,g t k ,0
given by
aK
w= = 4 , 230 k ⋅ Bq ⋅ s (A.5)
rK,n
The relative standard uncertainty urel ( w ) = u( w )/ w is calculated by
u2 ( aK )
⋅ u2 ( rK,g ) + u2 ( rK,0 ) (A.6)
1
urel
2
( w ) = urel
2
( aK ) + urel
2
( rK,n ) = +
aK2 rK,n
2
From this the relative standard uncertainty urel ( w ) and the standard uncertainty u ( w ) are calculated:
urel ( w ) = urel
2
( w ) = 0 , 022 4 (A.7)
and
u ( w ) = w ⋅ urel ( w ) = 0 , 094 7 k ⋅ Bq ⋅ s (A.8)
1 /( ε U − ε L ) for ε L ≤ ε ≤ ε U
fε ( ε ℑ) = (A.10)
0 otherwise
1
with the expectation E( fε ( ε ℑ)) = ( ε U + ε L ) (A.11)
2
1
and the variance Var( fε ( ε ℑ)) = ( ε U − ε L ) (A.12)
12
transforms to the distribution
f w ( w ℑ) =
{
1 / w 2 ⋅ ( ε U − ε L ) } for ε U−1 ≤ w ≤ ε L−1
(A.13)
0 otherwise
with the expectation
ln(1 / ε L ) − ln(1 / ε U )
E( f w ( w ℑ)) = (A.14)
εU − εL
and the variance
2
1 ln(1 / ε L ) − ln(1 / ε U )
Var( f w ( w ℑ)) = − (A.15)
εL ⋅ε U εU − εL
Since the difference between both distributions is apparently significant and the choice between a
numerator or denominator is arbitrary or determined by convention, the rectangular distribution is to
be used with care in practice, despite its repeated use in the previous examples for the sake of
convenience. Nevertheless, the benefit of the rectangular distribution either for the calibration factor
or the efficiency is the warranted existence of the detection limit y # due to the limited range of this
distribution.
α
with the expectation E[ fε ( ε ε i , i = 1 ,..., n )] = (A.17)
α +β
α ⋅β
and the variance Var( fε [ε ε i , i = 1 ,..., n )] = (A.18)
(α + β )2 (α + β + 1 )
From this, the parameters of the Beta distribution are calculated as
1 − E[ fε ( ε ε i , i = 1 ,..., n )] 1
α = {E[ fε ( ε ε i , i = 1 ,..., n )]}2 ⋅ − (A.19)
Var[ fε ( ε ε i , i = 1 ,..., n )] E[ fε ( ε ε i , i = 1 ,..., n )]
and
1
β = α ⋅ − 1 (A.20)
E[ fε ( ε ε i , i = 1 ,..., n )]
NOTE In case of a single counting calibration measurement α corresponds approximately to the number of
detected particles and β corresponds approximately to the number of missed particles (e. g. derived from a
calibration standard or a second detector with known efficiency). So, in Formula (A.18) for a reasonable number
n of events the detection efficiency is calculated and in Formula (A.19) the corresponding statistical variance. For
low event numbers see the corrections derived in Reference [46].
n n
1
∑ ∑
1
With the mean ε = ε i and the squared standard deviation sε2 = ( ε i −ε )2 of the measured
n − 1 i =1 n − 1 i =1
efficiency data the parameters of the Beta distribution are calculated by
1−ε 1 1
α =ε 2 − and β = α ⋅ − 1 (A.21)
s2
ε ε
ε
If the efficiency is not limited to the interval [0 , 1] but to [ε L , ε U ] the Beta distribution can be expanded
to the rescaled four parameter Beta distribution
1 ε − ε L
f ( ε ( ε i , i = 1 ,..., n ), ε L ≤ ε ≤ ε U ) = B( ε ,α , β , ε L , ε U ) = ⋅ B ,α , β (A.22)
εU − εL εU − εL
Annex B
(informative)
This annex exemplarily describes the procedure according to ISO 11929-2. For this purpose, a simple
model of evaluation according to Formula (B.1) is assumed.
ng n0
a = ( x1 − x2 ) · w = ( rg − r0 ) · w = rn · w = − · w (B.1)
t
g t0
Formula (B.1) is just a simple special case of the more general model in Formula (1)
x ⋅ x ⋅⋅⋅ x ⋅ x ⋅⋅⋅
y = ( x1 − x2 ⋅ x3 − x 4 ) ⋅ 6 8 = ( x1 − x2 x3 − x 4 ) ⋅ w with w = 6 8
x5 ⋅ x7 ⋅⋅⋅ x5 ⋅ x7 ⋅⋅⋅
or, more generally, for m input quantities,
y = G( x1 ,..., x m ) (B.2)
The uncertainty associated with y, u( y ) , is determined from the uncertainties u( x j ) by the method of
uncertainty propagation according to the ISO/IEC Guide 98-3.
Applying instead the Monte Carlo simulation method according to the ISO/IEC Guide 98-3:2008/Suppl.1
means to simulate a large number n M of measurements yi ( i = 1,..., nM ). This is achieved by generating
n M random values for each of the input quantities. The respective PDFs for the different input quantities
are derived according to the ISO/IEC Guide 98-3:2008/Suppl.1. They are documented for the examples
in the tables with the available information. The necessary information is summarized for the
application of ISO 11929-1 by a = {x1 ,..., x m ; u ( x1 ) ,..., u ( x m )} . For the application of ISO 11929-2, PDFs
have to be assigned to each of the input quantities X 1 ,..., X m according to the ISO/IEC Guide 98-3:2008/
Suppl.1 depending on the available information, which is simply summarized by the symbol a .
ISO 11929-2:2019, 6.3, Annex A and Annex B give a detailed description of the methodology.
Assuming here for simplicity and as an example that the x j follow normal distributions, a random
value of x j is generated by
x ji = x j + u( x j )z ji (step A) (B.3)
where z ji is a N( 0 , 1 ) deviate, i.e., a random number following a standard normal distribution. The
simulated values
yi = G( x1i ,..., x mi ) ( i = 1,..., nM ) (step B) (B.4)
which after sorting the array ascendingly represents a discrete distribution of y from which the
characteristic values are derived as follows. For the following always sorted arrays are assumed.
The probability pk associated with a value yk is given by pk = k / nM . The pk -values represent the
cumulative probability density, i.e. a function increasing from zero to unity if plotted against the yk
nM M n
∑ ∑
1 1
y= yi and u( y ) = ( yi − y )2 (B.9)
nM i =1
nM − 1 i =1
A quantile qp of the array { y1 ,..., ynM } , associated with a given probability p, is the value yk with
k = p nM .
The integral form analogues of the two expressions in Formula (B.8) are
∞ ∞
y = E (Y ) = ∫ ∫
y f Y ( y | x , u( x )) d y ; u( y ) = E (Y 2 ) − y 2 = y 2 f Y ( y | x , u( x )) d y − y 2 (B.10)
−∞ −∞
Comparing Formula (B.8) and Formulae (B.9, B.7) demonstrates how the Monte Carlo method simplifies
the calculations.
The best estimate, refers to values yi ≥ 0 ( nP number of positive yi , 0 ≤ nP ≤ nM ). It is calculated
together with its associated standard uncertainty as
nM nM
∑ ∑
1 1
ŷ = yi and u( ˆ)
y = ( yi − ˆ)
y 2 (B.11)
nM − nn i = nn + 1
nM − nn − 1 i = n +1
n
A quantile qp of the sub-array { ynn +1 ,..., ynM } , associated with a given probability p, is the value yk
with k = p ⋅( nM − nn ) .
The integral form analogues of the two expressions in Formula (B.10) are also represented by
Formula (B.9), but with the lower integral limit raised to zero.
The limits of the coverage intervals are obtained as suitable quantiles of the sub-array { ynn +1 ,..., ynM }
.
For the calculation of the decision threshold and the detection limit, the MC simulations refer to possible
true values a of the measurand, A, which differ from y. In this case, the input quantity value rg differs
from the measured one or, in other words, the information a = {x1 ,..., x m ; u ( x1 ) ,..., u ( x m )} is changed to
a´ = {x ´1 ,..., x m ; u ( x ´1 ) ,..., u ( x m )} . It is also assumed, that ng represents the first of the input quantity
values.
The integral form of the associated probability distribution is given by the so-called prospective
posterior
∞
m
f A ( a | a ´ ) = C ∫ ∏
fX1 ( x ´1 | x ´1 , u( x ´1 ))
i =1
fXi ( xi | xi , u( xi )) ⋅ δ ( a − G( x ´1 ,..., xm )) d x1 d x2 d xm (B.12)
−∞
The calculation of the decision threshold is performed by an iterative procedure searching for a set of
information a´ = {x ´1 ,..., x m ; u ( x ´1 ) ,..., u ( x m )} for which a =0 results. If this set is found, the (1−α )
-quantile of the prospective posterior according to Formula (B.11) is the decision threshold y* . The
decision threshold value is obtained as its upper quantile: y * = yk , with k = p nM at p = 1 −α .
The distribution associated with the decision threshold may become asymmetric (e.g., if a rectangular
distributed input quantity is involved) and the mean value then may deviate from zero. In that case it is
recommended to implement an iteration method also for the decision threshold, which, by applying
small deviations of x1 from zero, tries to diminish the distribution mean deviation from zero.
The calculation of the detection limit is performed by an iterative procedure through which the center
of the associated distribution is stepwise adjusted such that its lower quantile ( p = β ) finally converges
to the value y * . Therefore, the MC simulation needs to be repeated several times. In each iteration
step, the iteration procedure generates a specific value a of the measurand. From the latter, a new
value
∞
P ( y > y * y = 0 ) = ∫ fY ( y ′ a ', y = 0) dy ′ = α (B.13)
y*
is calculated. At this stage within a single iteration step, the simulation procedure (steps A-C) is
repeated. After conversion, the detection limit value is given as the mean value of the array
y = { y 1 ,..., ynM } , being the last one within the iteration.
The iteration requires using a numerical root finding method. Due to Monte Carlo uncertainties, the
functions supplied to such a root finding method are “noisy” (fluctuating). Therefore, such routines
should be more robust with respect to these fluctuations than for instance, the simpler bisection
method.
Bibliography
[1] ISO 7503 (all parts), Measurement of radioactivity — Measurement and evaluation of surface
contamination
[2] ISO 9696, Water quality — Gross alpha activity — Test method using thick source
[3] ISO 9697, Water quality — Gross beta activity — Test method using thick source
[4] ISO 9698, Water quality — Tritium — Test method using liquid scintillation counting
[5] ISO 10703, Water quality — Gamma-ray emitting radionuclides — Test method using high
resolution gamma-ray spectrometry
[6] ISO 10704, Water quality — Gross alpha and gross beta activity — Test method using thin source
deposit
[7] ISO 11483, Nuclear fuel technology — Preparation of plutonium sources and determination of
238Pu/239Pu isotope ratio by alpha spectrometry
[8] ISO 11665 (all parts), Measurement of radioactivity in the environment — Air: radon-222
[9] ISO 13160, Water quality — Strontium 90 and strontium 89 — Test methods using liquid scintillation
counting or proportional counting
[10] ISO 13161, Water quality — Polonium 210 — Test method using alpha spectrometry
[11] ISO 13162, Water quality — Carbon 14 — Test method using liquid scintillation counting
[12] ISO 13163, Water quality — Lead-210 — Test method using liquid scintillation counting
[13] ISO 13164-1, Water quality — Radon-222 — Part 1: General principles
[14] ISO 13165-1, Water quality — Radium-226 — Part 1: Test method using liquid scintillation counting
[15] ISO 13166, Water quality — Uranium isotopes — Test method using alpha-spectrometry
[16] ISO 13167, Water quality — Plutonium, americium, curium and neptunium — Test method using
alpha spectrometry
[17] ISO 13168, Water quality — Simultaneous determination of tritium and carbon 14 activities — Test
method using liquid scintillation counting
[18] ISO 18589 (all parts), Measurement of radioactivity in the environment — Soil
[19] ISO 21847, Nuclear fuel technology — Alpha spectrometry
[20] ISO 28218, Radiation protection — Performance criteria for radiobioassay
[21] Weise K.1998, , Bayesian-statistical decision threshold, detection limit and confidence interval
in nuclear radiation measurement. Kerntechnik 63; 214-224
[22] Weise K., Hübel K., Michel R., Rose E., Schläger M., Schrammel D. et al. 2005, Determination
of the detection limit and decision threshold for ionizing-radiation measurements: Fundamentals
and particular applications; Proposal for a standard. Report FS-05-129-AKSIGMA, Fachverband
für Strahlenschutz (TÜV-Verlag, Cologne)
[23] Weise K., Hübel K., Rose E., Schläger M., Schrammel D., Täschner M., Michel R.2006,
, Bayesian decision threshold, detection limit and confidence limits in ionizing-radiation
measurement, Radiat. Prot. Dosimetry 121(1), 52 – 63
[24] Weise K., Kanisch G., Michel R., Schläger M., Schrammel D., Täschner M., Monte Carlo
determination of the characteristic values in measurement of ionising radiation: Fundamentals
and numeric, Radiation Protection Dosimetry 135 No. 3 (2009) 169 – 196; doi:10.1093/rpd/
ncp105
[25] Weise K., Kanisch G., Michel R., Schläger M., Schrammel D., Täschner M., Characteristic
values in measurement of ionizing radiation – material for a critical discussion on fundamentals
and alternatives, Fachverband für Strahlenschutz e.V., FS-2013-167-AKSIGMA, TUEV Media
GmbH, Köln, 2013, ISBN 978-3-8249-1709-9
[26] Weise K., Wöger W. 1992, Eine Bayessche Theorie der Messunsicherheit. PTB-Bericht N-11.
(Physikalisch-Technische Bundesanstalt, Braunschweig)
[27] Weise K., Wöger W. 1993, A Bayesian theory of measurement uncertainty. Meas. Sci.
Technol. 4; 1-11
[28] Weise K., Wöger W., 1999, Messunsicherheit und Messdatenauswertung. ( Wiley-VCH,
Weinheim)
[29] Bernardo J.M., Smith A.F.M., 1994, Bayesian Theory. ( Wiley, Chichester)
[30] Bernardo J.M. 2003, Bayesian Statistics, in: Probability and Statistics (R. Viertl, ed.) of the
Encyclopedia of Life Support Systems (EOLSS). (UNESCO, Oxford)
[31] Gelman A., Carlin J.B., Stern H.S., Rubin D.B., 2003, Bayesian Data Analysis ( 2nd ed.). (
Chapman, London)
[32] Gregory P.C., 2005, Bayesian logical data analysis for the physical sciences. ( Cambridge
University Press, Cambridge)
[33] Lee P.M., 1989, Bayesian Statistics: An Introduction. ( Oxford University Press, New York)
[34] Robert C.P., 2001, The Bayesian Choice ( 2nd ed.). ( Springer, New York)
[35] Jaynes E.T., 2003, Probability Theory: The Logic of Science, Bretthorst G.L., ed. ( Cambridge
University Press, Cambridge)
[36] Berger J.O., 1985, Statistical Decision Theory and Bayesian Analysis. ( Springer, New York)
[37] Abramowitz M., Stegun I., 1968, Handbook of Mathematical Functions. 5th ed., Chap. 26. (
Dover Publications, New York)
[38] Bé M.M. et al., 2011, Table of Radionuclides, Monography BIPM-5, Vol. 6 (2011), ISBN 978‑92-
822-2242-3 (Vol. 6) https://w ww.bipm.org/utils/common/pdf/monographieRI/Monographie
_BIPM-5_Tables_Vol8.pdf (downloaded Dec. 13, 2017) and http://w ww.radiochemistry.org/
periodictable/gamma_ spectra/
[39] Mameghani R, 2010, Konsequenzen der ISO 11929 für die Begutachtung der Strahlungs-
und Aktivitätsüberwachung in kerntechnischen Anlagen, Bericht ETS1 - 28/2010, TÜV SÜD,
München
[40] Kanisch G., 2015, Generalized evaluation of environmental radioactivity measurements with
UncertRadio Part I: Methods without linear unfolding https://d x.doi.org/10.1016/j.apradiso
.2015.12.003
[41] Kanisch G., 2015, Generalized evaluation of environmental radioactivity measurements with
UncertRadio Part II: Methods with linear unfolding, https://d x.doi.org/10.1016/j.apradiso.2015
.12.046
[42] Michel R., 2016, Measuring, Estimating, and Deciding under Uncertainty. Applied Radiation
and Isotopes (2016), pp. 6-11, DOI information: 10.1016/j.apradiso.2015.12.013
Buying standards PLUS is an updating service exclusive to BSI Subscribing Members. You will
automatically receive the latest hard copy of your standards when they’re
You can buy and download PDF versions of BSI publications, including British and revised or replaced.
adopted European and international standards, through our website at bsigroup.
com/shop, where hard copies can also be purchased. To find out more about becoming a BSI Subscribing Member and the benefits
of membership, please visit bsigroup.com/shop.
If you need international and foreign standards from other Standards Development
Organizations, hard copies can be ordered from our Customer Services team. With a Multi-User Network Licence (MUNL) you are able to host standards
publications on your intranet. Licences can cover as few or as many users as you
Copyright in BSI publications wish. With updates supplied as soon as they’re available, you can be sure your
documentation is current. For further information, email [email protected].
All the content in BSI publications, including British Standards, is the property
of and copyrighted by BSI or some person or entity that owns copyright in the Revisions
information used (such as the international standardization bodies) and has
formally licensed such information to BSI for commercial publication and use. Our British Standards and other publications are updated by amendment or revision.
Save for the provisions below, you may not transfer, share or disseminate any We continually improve the quality of our products and services to benefit your
portion of the standard to any other person. You may not adapt, distribute, business. If you find an inaccuracy or ambiguity within a British Standard or other
commercially exploit or publicly display the standard or any portion thereof in any BSI publication please inform the Knowledge Centre.
manner whatsoever without BSI’s prior written consent.
Useful Contacts
Storing and using standards Customer Services
Standards purchased in soft copy format: Tel: +44 345 086 9001
• A British Standard purchased in soft copy format is licensed to a sole named Email: [email protected]
user for personal or internal company use only.
Subscriptions
• The standard may be stored on more than one device provided that it is Tel: +44 345 086 9001
accessible by the sole named user only and that only one copy is accessed at Email: [email protected]
any one time.
• A single paper copy may be printed for personal or internal company use only. Knowledge Centre
Tel: +44 20 8996 7004
Standards purchased in hard copy format: Email: [email protected]
• A British Standard purchased in hard copy format is for personal or internal
company use only. Copyright & Licensing
Tel: +44 20 8996 7070
• It may not be further reproduced – in any format – to create an additional copy.
This includes scanning of the document. Email: [email protected]
If you need more than one copy of the document, or if you wish to share the BSI Group Headquarters
document on an internal network, you can save money by choosing a subscription
product (see ‘Subscriptions’). 389 Chiswick High Road London W4 4AL UK