Professional Documents
Culture Documents
ABSTRACT: A survey is presented on the application of the principles of structural reliability in standardised
design of concrete structures. The ultimate objective of the survey is to consider the way in which reliability is
applied in Eurocode EN 1992 with a view of determining its suitability for adoption in South Africa. Al-
though clear guidance is given in EN 1992 on reliability management related to some aspects of quality con-
trol, little guidance is given related to various aspects of the various design situations identified in EN 1990. It
is demonstrated that EN 1992 can suitably provide for the performance requirements set by the South African
Loading Code SANS 10160:2010, which is consistent with EN 1990.
Partial Factor
1.40
used to provide for sufficient reliability across a
wide range of structural concepts (reinforced,
1.30
prestressed, lightweight, unreinforced, composite);
configurations (beams, slabs, columns, walls, foun- 1.20
dations); failure modes (flexure, axial, shear, torsion,
stability) and their combinations. Numerous assump- 1.10
tions, simplifications, empirical constants and other
RC 1 RC2 RC3
measures are imbedded in the design procedures, 1.00
without explicit treatment of reliability. Although 2 2.5 3 3.5 4 4.5
provision is made for modelling factors to provide Target Beta
for these effects, this is not applied transparently.
Figure 1. Partial factors for steel and concrete based on ECP
It is therefore somewhat surprising that the atten- model; showing three reliability classes RC1, RC2 & RC3.
tion given to the reliability assessment of structural
resistance is nowhere as extensive as it is for actions A more critical issue is the conversion of EN
on structures. This situation is however not unique to 1992-1-1 procedures for RC2 to other reliability
structural concrete, but applies also other structural classes, in particular for the important case of RC3
materials. structures. The combined effect of upwards adjust-
ment of partial factors for a higher level of reliabil-
ity, together with implementing QC measures and
4 RELIABILITY IMPLEMENTATION taking credit for such, as indicated by the bold arrow
in figure 1, indicate the effectiveness of the QC
4.1 Reliability management measures in requiring reduced partial factors {1.11;
1.43} even though the reliability class is raised.
Reliability management consists of differentiation of It should be noted that this adjustment does not
reliability classes in accordance with EN 1990 and merely represent a compensation measure to main-
adjustment of material factors in accordance with tain the reliability level, as stated in EN 1990 Annex
quality measures from EN 1992-1-1. B. If credit is taken for improved QC measures, the
By using the representative models for variability revised partial factors need to be again adjusted up-
of steel and concrete from the Eurocode 2 Commen- wards for RC3 structures, as indicated in figure 1.
tary (ECP 2008), improved QC is represented by No clear guidance is given in EN 1992-1-1 on this
changing the coefficient of variability for model un- matter, and in EN 1990 only the principles are given.
certainty, geometry and strength from {2,5%; 5%;
4%} to {1.5%; 2.5%; 4%} for steel; from {5%; 5%;
15%} to {4%; 2.5%; 10%} for concrete respectively. 4.2 Other considerations related to reliability
Figure 1 presents graphs indicating the partial fac- Although requirements for serviceability, accidental
tors for steel and concrete {γs; γc} as a function of design situations, structural integrity and robustness
the target reliability for the two cases of standard are not expressed explicitly in reliability based par-
practice and improved QC measures in accordance tial factor terms for structural behaviour/resistance,
with these adjustments. as indicated in table 1 (except that γc = 1.2 is stipu-
The partial factors are shown in figure 1 specifi- lated for accidental resistance), these limit states
cally for (i) The target levels of reliability for the form an integral part of structural performance.
three reliability classes RC1 – RC3 from EN 1990; The serviceability limit state often controls the
(ii) Adjusted for improved QC as implied by EN design, particularly in modern larger and more slen-
1992-1-1 Annex A. These two sets of partial factors der structures using higher strength materials.
represent the alternative measures available for reli- Due to large uncertainties in the serviceability
ability management. It is clear that improved QC performance of structures, reliability based optimisa-
measures are effective in managing reliability. tion can be used to derive suitable design proce-
The values for {γs; γc} of {1.09; 1.37} for im- dures. Performance levels are however mainly con-
proved QC for RC2 can be compared to values of trolled by the somewhat arbitrary serviceability
{1.05; 1.35} indicated by EN 1992-1-1 Annex A. criteria. No explicit reliability treatment of service-
This indicates that substantial credit is given for im- ability behaviour of the structure is evident in the
proved QC in the Eurocode procedures. procedures.
Specific attention is given to ductile behaviour that improved QC measures are applied to RC3 &
since it is generally accepted that a much higher RC4; for RC1 there is clearly an option of reducing
level of reliability is required for brittle failure the partial factors, particularly for concrete.
modes. The provision of sufficient ductility is one of
the critical measures to ensure sufficient robustness
of the structure. Provisions for ductility is largely 6 CONCLUSIONS
done through rules for detailing, for which there are The essential Eurocode performance requirements
a wide range of requirements, with little obvious for structural concrete have been reviewed, firstly to
linking to reliability management. establish the degree to which principles of structural
reliability are used explicitly in stipulating design
procedures; secondly to assess the way in which
5 PROVISION FOR SOUTH AFRICAN these procedures can be adopted for South African
CONDITIONS conditions.
An objective of this paper is to assess the suitability On the one hand Eurocode establishes an elabo-
of EN 1992-1-1 for adoption and to provide for the rate scheme of design situations and limit states in
basis-of-design requirements of SANS 10160:2010. the overall management of structural performance,
Since SANS 10160 is fully compatible with EN with extended treatment of actions, which applies
1990 no difficulties should be expected. The most also to resistance in general and specifically to struc-
important reliability related adaptations are to main- tural concrete.
tain the previous local reference target level of reli- On the other hand a rather simple scheme of par-
ability of βT = 3.0, which is still in accordance with tial factors is employed to cover a very broad scope
Eurocode procedures, somewhat compensated for by of design procedures for concrete resistance. A
a more elaborate set of reliability classes RC1 – scheme of model factors should serve to make the
RC4. This classification system is explained and mo- embedded effects of assumptions, expressions and
tivated by Retief & Dunaiski (2009). hosts of parameters more transparent and manage-
The required values for {γs; γc} for the four reli- able from a reliability perspective.
ability classes RC1 – RC4 is shown in Figure 2. It A reliability review of the Eurocode procedures
should be noted that although RC2 serve as refer- for concrete design in terms of South African basis-
ence, with βT = 3.0, RC2 & RC3 in fact represent a of-design requirements provide confirmation of the
division of the Eurocode RC2, generally at buildings suitability of EN 1992-1-1 for local use, and guid-
of four and five storeys. ance on how to implement such adoption.
The overall results show that partial factors of
{1.15; 1.5} would be conservative, confirming the
results of a separate analysis by Holický et al (2010). REFERENCES
This is true even for higher reliability classes, pro-
vided that improved QC measures are applied; as ECP (2008). Eurocode 2 Commentary. Downloaded 2008
should be the case for more important structures. http://www.ding.unisannio.it/ricerca/gruppi/ingciv/ceroni/co
mmentario_EC2_2004.pdf
EN 1990:2002. Basis of structural design. European Committee
for Standardization (CEN) Brussels, 2002.
1.70 Concrete EN 1992-1-1:2004. Design of concrete structures – General
Steel (QC) rules and rules for buildings. European Committee for Stan-
1.60
Concrete (QC) dardization (CEN) Brussels, 2004.
1.50 Steel FIB (2009). Draft fib Model Code 2010.
Holický, M., Retief, J.V. & Dunaiski, P.E. (2009). The reliabil-
Partial Factor
1.40
ity basis of Eurocode. In Retief, J.V. & Dunaiski, P.E. (Edi-
tors) Background to SANS 10160. SUN MeDIA, ISBN
1.30 978-1920338-10-7.
Holický, M., Retief, J.V. & Wium, J.A. (2010). Partial factors
1.20 for selected reinforced concrete members. Paper accepted
by SAICE Journal, 4 March 2010.
1.10 ISO 2394:1998. General principles on reliability for structures.
International Organisation for Standardisation.
1.00 RC1 RC2 RC3 RC4 Retief, J.V. & Dunaiski, P.E. (2009). The limit states basis of
2 2.5 3 3.5 4 4.5 structural design for SANS 10160-1. In Retief, J.V. &
Dunaiski, P.E. (Editors) Background to SANS 10160. SUN
Target Beta
MeDIA, ISBN 978-1920338-10-7.
SANS 10160:2010. Draft South African Standard: Basis of
Figure 2. Partial factors for steel and concrete for SANS 10160 structural design and actions for buildings and industrial
(RC1 – RC4) based on ECP model parameters. structures Part 1 Basis of structural design. South African
Bureau of Standards, Pretoria.
Partial factors of {1.10; 1.4} should suffice
throughout the various reliability classes, provided