Professional Documents
Culture Documents
2015
Handbook
NON-MANDATORY DOCUMENT
STRUCTURAL RELIABILITY
HANDBOOK
2015
Handbook: Structural Reliability
However, neither the ABCB, the participating Governments, nor the groups which have
endorsed or been involved in the development of the Handbook, accept any
responsibility for the use of the information contained in the Handbook and make no
guarantee or representation whatsoever that the information is an exhaustive treatment
of the subject matters contained therein or is complete, accurate, up-to-date or reliable
for any particular purpose.
The ABCB, the participating Governments and groups which have endorsed or been
involved in the development of the Handbook expressly disclaim all liability for any loss,
damage, injury or other consequence, howsoever caused (including without limitation by
way of negligence) which may arise directly or indirectly from use of, or reliance on, this
Handbook.
Users should exercise their own skill and care with respect to their use of this Handbook
and should obtain appropriate independent professional advice on any specific issues
concerning them.
In particular, and to avoid doubt, the use of this Handbook does not–
This work is the copyright of the Australian Government and States and Territories of
Australia and, apart from any use as permitted under the Copyright Act 1968, no part
may be reproduced by any process without prior written permission. Requests and
enquiries concerning reproduction and rights should be directed in the first instance to:
Preface
The Inter-Government Agreement (IGA) that governs the ABCB places a strong
emphasis on reducing reliance on regulation, including consideration of non-regulatory
alternatives such as non-mandatory information handbooks.
This Handbook is one of a series produced by the ABCB. The series of Handbooks is
being developed in response to comments and concerns expressed by government,
industry and the community that relate to the built environment. The topics of
Handbooks expand on areas of existing regulation or relate to topics which have, for a
variety of reasons, been deemed inappropriate for regulation. The aim of the
Handbooks is to provide construction industry participants, non-mandatory advice and
guidance on specific topics.
Structural reliability has been identified as an issue that requires consistent uniform
guidance. The Structural Reliability Handbook has been developed to foster a greater
understanding of Verification Methods BV1 and V2.1.1 contained within the National
Construction Code (NCC) Volumes One and Two respectively. This Information
Handbook addresses the methodology in developing the Verification Methods in generic
terms, and is not a document that sets out a specific process of using the Verification
Methods or an alternative structural reliability process. It is expected that this Handbook
will be used to develop solutions relevant to specific situations in accordance with the
generic principles and criteria contained herein.
Table of Contents
Important Notice and Disclaimer .................................................................................. ii
Preface........................................................................................................................... iv
1 Introduction ........................................................................................................... 1
1.1 Structural Reliability .................................................................................... 1
1.2 Limitations .................................................................................................. 2
2 Background ........................................................................................................... 3
2.1 Notation ...................................................................................................... 3
2.2 Importance Levels ...................................................................................... 4
2.3 Hierarchy of Performance Specification ..................................................... 5
2.4 Structural Performance Requirements ....................................................... 6
5 Model of Resistance............................................................................................ 16
5.1 Developing resistance models .................................................................. 16
References ................................................................................................................... 39
1 Introduction
Reminder:
This Handbook is not mandatory or regulatory in nature and compliance with it will not
necessarily discharge a user's legal obligations. The Handbook should only be read and
used subject to, and in conjunction with, the general disclaimer at page ii.
The Handbook also needs to be read in conjunction with the building legislation of the
relevant State or Territory. It is written in generic terms and it is not intended that the
content of the Handbook counteract or conflict with the legislative requirements, any
references in legal documents, any handbooks issued by the Administration or any
directives by the Building Control Authority.
BP1.1 and BP1.2 have a comprehensive list of documents to support the Deemed-to-
Satisfy (DtS) Provisions, while P2.1.1 (a), (b) and (c) have supporting DtS Provisions
through Acceptable Construction Manuals, and Acceptable Construction Practices.
These manuals and documents cover most aspects of the Limit State Design Method
for most construction materials, however if designers wish to or have to operate outside
DtS they must develop an Alternative Solution (Performance Solution). BV1 and V2.1.1
are designed to support those who wish to follow this Alternative Solution (Performance
Solution) path.
1.2 Limitations
This Handbook is not intended to:
This Handbook is intended to make users aware of provisions that may affect them, not
exactly what is required by those provisions. If users determine that a provision may
apply to them, the NCC should be read to determine the specifics of the provision.
2 Background
2.1 Notation
The units and notation used in this Handbook are designed specifically for use with the
structural reliability Verification Methods. The symbols used are outlined in Table 2.1.1.
Table 2.1.1 Symbols
Symbol Meaning
C Aerodynamic shape factor to convert wind speed to wind pressure
CD Factor to cover the effects of the ground condition
CE Factor to cover the effects of exposure for snow action
CF Factor to cover the geometrical effects such as the roof shape for snow
action
COV Co-efficient of variation
CQ Correction factor for action ( = 1+VQ2)
CR Correction factor for resistance ( = 1+VR2)
CW Factor to cover the effects of mass distribution of a building for an
earthquake action
E Earthquake action effect
g Permanent action
G Permanent action effect
Hx Factor to convert action to action effect for ‘x’ action
Ki Uncertainty factors for resistance
M Factor to cover all multipliers for wind speed: direction, exposure,
shielding and topographic
Q Imposed action effect
q Imposed action
Qm Mean action
Qn Nominal design value of the action
Rm Mean resistance
Rn Nominal design value of the resistance
sG Ground snow load
V Basic wind speed
VQ Coefficient of variation with respect to action
Symbol Meaning
VR Coefficient of variation with respect to resistance
W Wind action effect
β Beta, with respect to structural reliability, structural reliability index
There are various levels of performance specifications, from prescriptive, which involve
detailed descriptions of how the process should be completed, to pure performance
which allows a greater degree of freedom in achieving the same requirements or
objectives. describes the relationship of prescriptive and performance based
specifications and where a Verification Method sits within this relationship. The
structural reliability Verification Methods are predominately performance based
solutions, but they are prescriptive in the determination of actions in order to provide
comparable indices and ensure a level of safety in line with the current NCC
requirements.
Figure 2.3.1 Level of Performance Specification
(b) a listing of the factors to be considered, namely the actions to which a building
‘may reasonably be subjected’.
The Verification Methods are specifically arranged to cover strength performance that is
the relationship between the actions (described in BP1.1 and P2.1.1(a) and (b)) and the
resistance (described in BP1.2 and P2.1.1(c)).
The Verification Method is one way, but not the only way, to demonstrate compliance
with the NCC Performance Requirements. There are other Structural Performance
Requirements in the NCC that are not covered by these Verification Methods.
3 Structural Reliability
The NCC target structural reliability indices are set as the averages of the reliability
indices found in current design practice for steel, concrete and timber. The reference
period for this target is one year.
The distance between these two curves is the performance of the component under
question. The Verification Methods provide the action models in line with the
appropriate parts of the Joint Australian Standard and New Zealand Standard (AS/NZS)
1170 as described in Section 4 Models of Actions, of this Handbook.
The basic information required for the Action and Resistance Models are:
The performance of the system can be thought of as the distance between the two
curves in . One method of quantifying this and taking into account all the above
variables is to define a reliability index, β. The NCC defines reliability index in BV1 and
V2.1.1 as:
𝑅𝑚 𝐶𝑄
𝛽 = ln[( ) √ ] /√ln(𝐶𝑅 . 𝐶𝑄 )
𝑄𝑚 𝐶𝑅
with
𝑅𝑚 𝛾 𝑅𝑚 𝑄𝑚
( ) = ( ) ( )/( )
𝑄𝑚 𝜙 𝑅𝑛 𝑄𝑛
and
𝐶𝑅 = 1 + 𝑉𝑅2
𝐶𝑄 = 1 + 𝑉𝑄2
Then
𝑋̅ = 𝑋1
̅̅̅̅ . 𝑋2
̅̅̅̅ . ̅̅̅̅̅
𝑋3𝑛
2
𝑉𝑋2 = (1 + 𝑉𝑋1
2 )( 2 )
1 + 𝑉𝑋2 2 𝑛
(1 + 𝑉𝑋3 ) −1
where;
̅̅̅̅̅̅
(𝑋𝑖) = 𝑚𝑒𝑎𝑛 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 (𝑋𝑖)
and
and
𝑉𝑥 = 𝜎𝑥 /𝑥̅
Reference to other distributions can be found in other technical papers and are not
discussed in this Handbook.
4 Models of Actions
A set of probabilistic Action Models used in the Verification Methods are given below for
the purpose of computing the reliability indices. The models represent typical
characteristics of the actions as related to Australian conditions, but not specific to a
particular location or a type of occupancy.
Numerical values for the models of actions to be used in the computation of the
reliability indices are given below. Appendix B provides further details on how these
figures are derived.
G = HG . g
where:
g = permanent action
The Mean and COV values for the parameters have been assessed as follows:
Therefore:
This is the model for permanent action to be used in the calculation of the reliability
indices.
Q = HQ . q
where:
q = imposed action
The Mean and COV values for the parameters have been assessed as follows:
W = HW . C . (M . V)2
where:
V = the basic wind speed whose statistics are available and given in
AS/NZS 1170.2 in terms of annual probability of exceedance
M = factor to cover all multipliers for the wind speed: direction, exposure,
shielding and topographic
The corresponding nominal design wind action effect is: W n = HWn . Cn . (Mn . Vn)2
The Mean and COV values for the parameters have been assessed as follows:
The values for (V/Vn) are given in Appendix B4 for all non-cyclonic and cyclonic regions
of Australia. For the purpose of calculation of the reliability indices, the following figures
in Table 4.3.1 are used.
Table 4.3.1 Mean and Coefficient of Variation values for peak annual wind actions for non-cyclonic
and cyclonic regions of Australia
S = HS . CE . CF . sG
where:
sG = the ground snow load whose statistics are available and given in
AS/NZS 1170.3 in terms of annual probability of exceedance
The corresponding nominal design snow action effect is: Sn = HSn . CEn . CFn . sGn
The MEAN and COV values for the parameters have been assessed as follows:
The values for (sG/sGn) are given in Appendix B5. The Mean and COV values for peak
annual snow action are given in Table 4.4.1.
Table 4.4.1 Mean and Coefficient of Variation values for peak annual actions for snow
E = HE . CR .CS . CW . a
where:
The MEAN and COV values for the parameters have been assessed as follows:
The values for (a/an) are given in Appendix B6. The Mean and COV values for peak
annual earthquake action are given in Table 4.5.1.
Table 4.5.1 Mean and Coefficient of Variation values for peak annual earthquake action
These are the values to be used in the calculation of reliability indices for earthquake
action.
5 Model of Resistance
R = Km. Kf. Ks … Rn
The sources of uncertainties must include, but are not limited to the following;
If all the above variables are assumed to be of lognormal distribution and statistically
independent, then the mean value and the coefficient of variation of R can be
established as follows:
The value of RN is usually established by identifying the major parameters that affect the
behaviour of the component and constructing appropriate structural models to account
for their effects. RN must be formulated using five percentile characteristic material
properties in accordance with BP1.2 and P2.1.1(c) of the NCC. Examples can be found
in structural design standards for steel, concrete and timber.
The value of Km to account for variability of the relevant mechanical properties is usually
obtained from test data used for quality control of the material manufacturing process.
The value of Ks to account for variability in structural modelling is obtained from the test
research data used in the construction of the structural model.
A flow chart for the process of establishing the Resistance Model is outlined in Figure
5.1.1.
By using lognormal distribution for both the actions and resistances, it is possible to
have a close-form expression for the reliability indices. Other more sophisticated
methods for computing the reliability indices can be used but it is up to the proposer to
justify their appropriateness.
The method is expected to be used in situations where there are no appropriate DtS
references. The targets are set at the average values of those found in current practice
using steel, concrete or timber.
A flowchart for the method of BV1 and V2.1.1 is shown in Figure 6.1.1.
Product Data:
1155 full size samples, measured and tested to a 5 minute duration test.
Mean bending stress, fu = 37.7 Mega Pascals (MPa)
Coefficient of variation, Vfu = 0.40
Dimensional variations:
Width b (Mean/Nominal) = 1.0, Vb = 0.02
Depth d (Mean/Nominal) =1.0, Vd = 0.02
Solution:
Establish the Resistance Model: The moment capacity of the section is the key
parameter in assessing bending resistance following the model used in AS 1720.
R = kt . Z . fu
where:
Z = section modulus
fu = failure stress
RN = ktn . Zn . fun
Therefore;
𝑉𝑅2 = 𝑉𝑘𝑡
2
+ 𝑉𝑍2 + 𝑉𝑓𝑢
2
Where LDF = load duration factor as given in AS 1720 as shown in the table below;
VZ = 0.04
Step 3 Evaluation of (fu/fun): With the number of tests over 1000, there is no need to
have a sampling factor. The five percentile value is used as the design stress f un as
required under BP1.2 and P2.1.1(c).
Thus the 5-percentile ranges from 16.6 to 18.5 MPa. Selecting a central value of fun =
17 MPa. It does not matter what the selected value is as higher fun will require a lower φ
and vice versa.
2
𝑉𝑅 = √(𝑉𝑘𝑡 + 𝑉𝑍2 + 𝑉𝑓𝑢
2
) = 0.42
Table 6.3.2 and Table 6.3.3 present the reliability indices for both primary and other
structural components.
Table 6.3.2 Reliability index calculation outcomes for primary structural components
Target β
Importance Calculated β Load Duration
Action type (primary
Level for φ = 0.9 Factor Used
structural)
Permanent 1, 2, 3, and 4 3.8 4.0 0.57
Imposed 1, 2, 3, and 4 3.8 3.8 0.8-0.94
Non-cyclonic wind 1 3.2 2.9 1.0
Non-cyclonic wind 2 3.4 3.2 1.0
Non-cyclonic wind 3 3.6 3.3 1.0
Non-cyclonic wind 4 3.8 3.4 1.0
Cyclonic wind 1 3.2 3.2 1.0
Cyclonic wind 2 3.4 3.5 1.0
Cyclonic wind 3 3.6 3.6 1.0
Cyclonic wind 4 3.8 3.8 1.0
Target β
Importance Calculated β Load Duration
Action type (non-primary
Level for φ = 0.9 Factor Used
structural)
Permanent 1, 2, 3, and 4 3.5 4.0 0.57
Imposed 1, 2, 3, and 4 3.5 3.8 0.8-0.94
Non-cyclonic wind 1 2.9 2.9 1.0
Non-cyclonic wind 2 3.1 3.2 1.0
Non-cyclonic wind 3 3.3 3.3 1.0
Non-cyclonic wind 4 3.5 3.4 1.0
Cyclonic wind 1 2.9 3.2 1.0
Cyclonic wind 2 3.1 3.5 1.0
Cyclonic wind 3 3.3 3.6 1.0
Cyclonic wind 4 3.5 3.8 1.0
Discussion
It is not important for the verifier to know the derivation of the proposed design stress
and φ, capacity factor. Looking at the data, the proposed design stress is approximately
the five percentile value which is a requirement under BP2 and P2.1.1(c). In this
example, we did not question the validity of M = Z . Fu. ignoring the structural modelling
factor in this case. For more complex situations, uncertainty in structural analysis could
be a significant factor. Uncertainty in structural modelling is high when the structural
action is not well understood and empirical factors are introduced to reconcile the
structural model with experimental data such as shear strength, or anchors. In this
example, the variability in the material strength is the controlling factor, which is normal
for materials like timber.
From Table 6.4.2 and Table 6.4.3, it is seen that the adoption of a basic bending design
stress of 17 MPA and a Φ =0.9 will achieve the target reliability for non-primary
structural members for permanent, imposed and wind actions, that is it will be
satisfactory for housing and other similar applications. However, for a primary structural
member, it will be prudent to reduce the Φ factor to 0.7-0.8 to meet the reliability target.
These findings are in line with AS 1720 recommendations.
Two series of tests were conducted with products from two different factories, factory A
and factory B with the same product specification. The test bending capacities in
Kilonewton-metre/metre (kNm/m) are given in Table 6.4.1. The data statistics are given
in Table 6.4.2.
In this example, we will establish a number of solutions (DtS and Alternative Solutions)
and use the verification method to assess their appropriateness.
Sample 1 2 3 4 5 6 7 8 9 10
A 2.556 1.955 1.901 1.814 1.553 1.504 1.205 0.995 1.656 1.894
B 1.109 1.333 1.453 1.409 1.573 0.889 1.273 1.188 1.459 1.282
Table 6.4.2 Data statistics
Solution:
Assuming the data is a fair representation of the properties of the products from
factories A and B, the COV of the samples are taken as COV of the products.
A DtS solution
Alternative Solutions
Option 1: Instead of using the minimum sampling factors, we use the average sampling
factors, but maintain the same criteria as AS/NZS 1170 of a five-percentile value with
99% confidence and a Weibull distribution. The outcomes are in Table 6.4.4.
Option 2 Instead of using minimum or average factors we can nominate a design value
and φ factor. In this case we have selected the five-percentile value with a lognormal
distribution, 99% confidence and φ = 1. The outcomes are in Table 6.4.5.
Note: For the derivation of kt in the above refer to Wang & Pham ‘Sampling factor for
prototype testing of structures’ Australian Journal of Structural Engineering Vol.12, No2,
2012.
Discussion
Table 6.4.6 outlines all three solutions in comparison to target values within the
Verification Methods BV1 and V2.1.1. It is noted that the DtS solution is conservative,
while the alternate solutions are not conservative, but they can be used with an adjusted
Φ factor (< 1.0) calibrated to enable the target indices to be reached. This example is
used to demonstrate the Verification Method can be used to assess the appropriateness
of various alternate solutions.
Table 6.4.6 Verification of Target Reliability
To achieve the target reliability, the Φ value for Option 1 is 0.9 and for Option 2 is 0.8.
Design properties of FRP square hollow section 100 x 100 x 5.2 SHS is to be
established with three series of tests for bending capacity, longitudinal tension and
modulus of elasticity. From the test data, establish the design properties and
appropriate capacity factors, φ, for the tube using BV1.
One series of bending tests were conducted for one size of tube. Failure was initiated
with the local buckling of the compression flange. The test local buckling capacities M b
and ultimate bending capacities Mu (in kN-m) are given in Table 6.5.1.
Moment 1 2 3 4 5
Mb 14.80 14.44 16.38 15.07 15.28
Mu 16.14 15.10 16.38 16.20 15.49
Solution:
The COV of the samples is very small, probably because all samples were taken from
the same batch. Samples from different production times may well exhibit much larger
variation. We shall explore the possibilities that the COV of the product could be 10% to
20%.
The design values for Mb should be the five-percentile values in accordance with BP1.2.
These values could be established at varying levels of confidence and assumed
distributions as given in Table 6.5.5.
Weibull
Lognormal
Population Sample Degree of Weibull 5- Lognormal
Moment 5-percentile
COV Average Confidence kt percentile kt
value
value
10% 15.2 Mb 75 1.26 12.1 1.22 12.5
(kN-m)
Note: For the derivation of kt in the above refer to Wang & Pham ‘Sampling factor for
prototype testing of structures’ Australian Journal of Structural Engineering Vol.12, No2,
2012.
Table 6.5.4 Reliability index calculation for product COV = 10% and design value 11.6 kN-m with
varying capacity factors
Importance Target
Action type
Level
Qm/Qn VQ ϒ Rm/Rn VR φ β
β
Table 6.5.5 Reliability index calculation for product COV=20% and design value 9.0 kN-m with
varying capacity factors
Importance Target
Action type Qm/Qn VQ ϒ Rm/Rn VR φ β
β
Level
Discussion
For production variability of 10%, we select Mb = 11.6 kN-m and adjust φ to achieve the
target reliability. The outcomes are shown in Table 6.5.4 where it is seen that φ = 0.7 is
necessary for primary structural components and φ = 0.85 is necessary for other
structural components. For production variability of 20%, we select Mb = 9.0 kN-m and
adjust φ to achieve the target reliability.
The outcomes are shown in Table 6.5.5 where it is seen that φ = 0.8 is necessary for
primary structural components and φ = 0.95 is necessary for other structural
components. This example illustrates the change in Φ values required to cope with the
change in production variability to maintain an appropriate level of reliability.
Appendices
The following is an extract of the structural Verification Method in NCC Volume One.
The Verification Method for NCC Volume Two is identical except that it is limited to the
Importance Levels covered by Volume Two, Importance Levels 1 and 2 only.
Importance Level (see Permanent and Imposed Wind, Snow and Earthquake
Table B1.2a) Actions Actions
1 3.2
2 3.4
3.8
3 3.6
4 3.8
Note:
The structural reliability indices shown in this table are for primary structural
components and connections whose failure could result in collapse of the building,
structure or other property. For other structural components, the target structural
reliability indices can be reduced by 0.3.
(b) the structural reliability index (β) is calculated in accordance with the following
formula:
𝑅𝑚 𝐶𝑄
𝛽 = ln[( ) √ ] /√ln(𝐶𝑅 . 𝐶𝑄 )
𝑄𝑚 𝐶𝑅
where—
𝛾 𝑅
𝑅𝑚 (𝜙) ( 𝑅𝑚 )
𝑛
( )=
𝑄𝑚 𝑄𝑚
𝑄𝑛
𝐶𝑅 = 1 + 𝑉𝑅2
𝐶𝑄 = 1 + 𝑉𝑄2
where—
CQ = correction factor for action; and
CR = correction factor for resistance; and
Qm = mean action; and
Qn = nominal design action; and
Rm = mean resistance; and
Rn = nominal design resistance; and
VQ = coefficient of variation with respect to action; and
VR = coefficient of variation with respect to resistance; and
Φ = capacity factor; and
γ = load factor; and
c) the action models for calculation of the structural reliability index are
determined in accordance with Table BV1.2; and
d) the resistance model for the structural component is established after taking
into account variability due to material properties, fabrication and construction
processes, and structural modelling.
Table BV1.2 – Action Models
B.1 Introduction
This Appendix describes the derivation of the action models presented in Section 4 of
this Handbook. The models, in general, include two components:
(a) a factor to convert action into action effect; and
(b) the intensity of the action.
The former is largely based on judgement and the latter on statistical data. Lognormal
distribution has been assumed for all parameters for ease of combination. It is important
to note the models (except for Permanent Action) were established based on a
reference time period of one year. Thus the resulting reliability indices are also for a one
year time reference.
Different types of building occupancy result in different imposed action models. For the
purpose of reliability assessment, it is desirable to have only one representative model
for imposed action. The statistical characteristics of office, residential and school
buildings are sufficiently close to have a common model as the average of the three
types of building occupancy. Factor H for permanent and imposed load is kept to the
same value since the same structural modelling is used for both gravity types of load.
Joint Committee on Structural Safety (JCSS) reported a number of statistical data sets
for different types of building occupancies (e.g. office, residential, schools, shops etc.).
A number of models were established for different types of occupancies using JCSS
parameters. A hybrid model was then established from the models for office, residential
and school buildings to represent a typical Imposed Action.
Separate probabilistic wind speed models for region A, B, C and D were established
using the 14 quartile values specified in AS/NZS 1170.2. Lognormal distribution was
used for ease of combination with other parameters. Values for other factors are based
on judgement or as given by JCSS.
The Mean and COV values for peak annual wind speed have been assessed from wind
statistical records in accordance with the Importance Level requirements of the NCC
and tabulated below:
Table B4.1 Mean and Coefficient of Variation values for peak annual wind speed for regions A and
B
The Mean and COV values for peak annual wind actions are described in the tables
below.
Table B4.3 Mean and Coefficient of Variation values for peak annual wind actions for regions A
and B
Table B4.4 Mean and Coefficient of Variation values for peak annual wind actions for regions C
and D
The factor H for wind action is reduced to 0.8 because the wind action is essentially a
dynamic action which has been conservatively transformed into an equivalent static
action. The Coefficient of Variation for other factors is based on the JCSS assessment.
The MEAN and COV values for peak annual ground snow have been assessed from
snow statistical records in accordance with the Importance Level requirements of the
NCC and tabulated below:
Table B5.1 Mean and Coefficient of Variation values for peak annual ground snow
Probability factors for ground snow action were fitted to a lognormal distribution for the
construction of the snow action model. The factor H was kept the same as for other
gravity loads and the factors CE and CF were as given in JCSS.
Probability factors for peak ground acceleration were fitted to a lognormal distribution for
the construction of the earthquake action model. All other factors are based on
judgement.
The MEAN and COV values for peak annual acceleration coefficient have been
assessed from earthquake statistical records in accordance with the Importance Level
requirements of the NCC and tabulated below:
Table B6.1 Mean and Coefficient of Variation values for peak annual acceleration of earthquakes
References
Wang & Pham ‘Sampling factors for prototype testing of structures’ Australian Journal of
Structural Engineering Vol.12, No2, 2012
Wang C. ‘An investigation on structural reliability Verification Method Final Report’ Feb
2014
Joint Committee on Structural Safety (JCSS) - 2000 ‘Probabilistic Model Code Part 1 –
Basis of Design’