Eurocodes and design of structures Philosophy of partial factors calibration Philosophy of partial factors calibration Jean-Armand Calgaro Ingénieur général des ponts et chaussées e.r. Chairman of CEN/TC250 (2007-2013)
Eurocodes and design of structures
Philosophy of partial factors calibrationPhilosophy of partial factors calibration
Jean-Armand Calgaro
Ingénieur général des ponts et chaussées e.r.
Chairman of CEN/TC250 (2007-2013)
Marcus Vitruvius Pollio (born c. 80–70 BC, died after c. 15 BC), Roman
author, architect, civil engineer and military engineer during the 1st century
BC, known for his multi-volume work entitled De Architectura.
Vol. III – Section 1:
“The design of Temples depends on symmetry, the rules of which Architects
should be most careful to observe. Symmetry arises from proportion (…). should be most careful to observe. Symmetry arises from proportion (…).
Proportion is a due adjustment of the size of the different parts to each other
and to the whole; on this proper adjustment symmetry depends. Hence no
building can be said to be well designed which wants symmetry and
proportion. In truth they are as necessary to the beauty of a building as to that
of a well formed human figure”.
Fra Luca Bartolomeo de Pacioli (c. 1447–1517) was an Italian
mathematician, Franciscan friar, collaborator with Leonardo da Vinci, and
a seminal contributor to the field now known as accounting
Pacioli published several works on mathematics, and more specifically De
divina proportione (written in Milan in 1496–98, published in Venice in
1509). The subject was mathematical and artistic proportion, especially
the mathematics of the golden ratio and its application in architecture. the mathematics of the golden ratio and its application in architecture.
Leonardo da Vinci, (15 April 1452 – 2 May 1519) was an Italian
polymath. His areas of strength included painting, sculpting, architecture,
science, music, mathematics, engineering, invention, anatomy, geology,
astronomy, botany, writing, history, and cartography.
6 November 1953
CEB was founded
1970
International
Recommendations
CEB/FIP
A brief history of the reliability approach
For construction works
1906
First French
Technical « regulation »
Concrete constructions
1964
Practical
Recommendations
For constructors
1978
Model-Code 78
CEB/FIP
No specialist of geotechnical design participated to the discussions concerning
the problems of structural safety and there was no chapter in Part I of the
Model-Code 1978 devoted to concrete structures dealing with foundations. Model-Code 1978 devoted to concrete structures dealing with foundations.
This explains that the “semi-probabilistic format” has been, in reality,
“imposed” to geotechnical experts (and other experts …).
Eurocodius : Ratio firmitatisEurocodius : Ratio firmitatisEurocodius : Ratio FirmitasEurocodius : Ratio Firmitas
Latin versionLatin version
EN 1990 – EUROCODE
BASIS OF BASIS OF STRUCTURAL DESIGN
EN 1990 – 1.5 Terms and definitions (1)
1.5.2.17
reliability
ability of a structure or a structural member to fulfil the specified requirements, including
the design working life, for which it has been designed. Reliability is usually expressed in
probabilistic terms
NOTE Reliability covers safety, serviceability and durability of a structure.NOTE Reliability covers safety, serviceability and durability of a structure.
1.5.2.18
reliability differentiation
measures intended for the socio-economic optimisation of the resources to be used to
build construction works, taking into account all the expected consequences of failures
and the cost of the construction works
EN 1990 – 1.5 Terms and definitions (2)1.5.2.8
design working life
assumed period for which a structure or part of it is to be used for its intended purpose with anticipated
maintenance but without major repair being necessary
1.5.2.12
limit states
states beyond which the structure no longer fulfils the relevant design criteria
1.5.2.131.5.2.13
ultimate limit states
states associated with collapse or with other similar forms of structural failure
NOTE They generally correspond to the maximum load-carrying resistance of a structure or structural
member.
1.5.2.14
serviceability limit states
states that correspond to conditions beyond which specified service requirements for a structure or
structural member are no longer met
� Identification of undesired phenomena and situations (limit-states).
� Evaluation of the risks attached to these phenomena and situations.
The probabilistic approach of structural safety:
The heritage
� Adoption, for a construction work, of a design such that the probability of the
mentioned risks is limited to a low value in order to remain acceptable.
Two questions :
- What is a low value ?
- By whom this « low » value would be accepted ?
The semi-probabilistic verification format of construction works is based on
rules, partly empirical or historical, which introduce safety at three levels :
� Adoption of representative values of random variables (actions and
resistances) defined on clear bases;
The semi-probabilistic verification format of
construction works
resistances) defined on clear bases;
� Application of partial factors to actions and resistances;
� Introduction of safety margins more or less apparent in the various
models : models of actions, models of action effects and models of resistances.
6.4 Ultimate limit states
6.4.1 General
(1)P The following ultimate limit states shall be verified as relevant :
EQU : Loss of static equilibrium of the structure or any part of it considered as a rigid body, where :
minor variations in the value or the spatial distribution of actions from a single source are significant,
and the strengths of construction materials or ground are generally not governing ;
EN 1990 – Eurocode : Basis of Structural Design
and the strengths of construction materials or ground are generally not governing ;
STR : Internal failure or excessive deformation of the structure or structural members, including
footings, piles, basement walls, etc., where the strength of construction materials of the structure
governs ;
GEO : Failure or excessive deformation of the ground where the strengths of soil or rock are significant
in providing resistance ;
FAT : Fatigue failure of the structure or structural members.
dd RE ≤Ed is the design value of the effect of actions such a s internal force, moment or a vector representing several internal fo rces or moments ;
The semi-probabilistic format for the verification of construction works (ultimate limit states)
{ } 1≥= iaFEE direpiFd ;,,γ
Rd is the design value of the corresponding resistance.
{ }direpiFd ,,
= diM
ikid a
XRR ;
,
,
γη
6.5.1 Verifications
(1)P It shall be verified that :
The semi-probabilistic format for the verification of construction works (serviceability limit states)
dd CE ≤where :
Cd is the limiting design value of the relevant serviceability criterion.
Ed is the design value of the effects of actions specified in the serviceability
criterion, determined on the basis of the relevant combination.
∑ ∑≥ >
+++1 1
,,0,1,1,,, """"""j i
ikiiQkQPjkjG QQPG ψγγγγ
Expression (6.10)
6.4.3 - Ultimate limit-states STR/GEO - Fundamental combination for
persistent and transient design situations
EN1990 - Eurocode : Basis of structural design
≥ >1 1j i
Expressions (6.10a) and (6.10b)
+++
++
∑∑
∑∑
>≥
≥≥
1
,,0,1,1,
1
,,
1
,,0,
1
,,
""""""
""""
i
ikiiQkQP
j
jkjGj
i
ikiiQP
j
jkjG
QQPG
QPG
ψγγγγξ
ψγγγ
0,85 ≤≤≤≤ ξξξξ ≤≤≤≤ 1,00
∑>
++++1
,,01,inf,sup, 5,1""5,1""""""35,1i
ikikmkjkj QQPGG ψ
Expression (6.10)
Expressions for the fundamental combination of actions based
on recommended values of partial factors
for buildings
Expressions (6.10a) and (6.10b)
++++
+++
∑
∑
>
≥
1
,,01,inf,sup,
1
,,0inf,sup,
5,1""5,1""""""15,1
5,1""""""35,1
i
ikikmkjkj
i
ikimkjkj
QQPGG
QPGG
ψ
ψ
Expressions (6.10a) and (6.10b)
Accidental design situations : expression 6.11b
∑ ∑≥ >
++++1 1
,,21,1,21,1, "")(""""""j i
ikikdjk QQouAPG ψψψ
EN1990 - Eurocode : Basis of structural design
∑ ∑≥ ≥
+++1 1
,,2, """"""j i
ikiEdjk QAPG ψ
Seismic design situations : expression 6.12b
6.5.3 Serviceability limit states : combinations of actions
� Characteristic Combination (irreversible SLS)
∑ ∑≥ >
+++1 1
,,01,, """"""j i
ikikjk QQPG ψ
EN1990 - Eurocode : Basis of structural design
� Frequent Combination (reversible SLS)
� Quasi-permanent Combination (reversible SLS)
≥ >1 1j i
∑ ∑≥ >
+++1 1
,,21,1,1, """"""j i
ikikjk QQPG ψψ
∑ ∑≥ ≥
++1 1
,,2, """"j i
ikijk QPG ψ
Overview of reliability methods (EN 1990 – Annex C)
Deterministic
methods
Probabilistic methods
Historical methods Empirical methods
FORM (Level II)
Full probabilistic (Level III)
Calibration Calibration Calibration Calibration Calibration Calibration
Semi-probabilistic
methods (Level I)
Method c
Method a Partial factor design
Method b
The Eurocodes have been primarily based on method a. Method c or equivalent
methods have been used for further development of the Eurocodes.
NOTE An example of an equivalent method is design assisted by testing (see annex D).
In the Level II procedures, an alternative measure of reliability is conventionally defined
by the reliability index β which is related to Pf by :
β−Φ=
where Φ is the cumulative distribution function of the standardised Normal distribution.
The relationship between the target reliability index for one year, , and , the value for
the design working life equal to n years, is:
)( β−Φ=fP
1β nβ
[ ]n
n )( 1
1 ββ ΦΦ= −
Consequences
ClassDescription
Examples of buildings and
civil engineering works
Reliability
class
Recommended minimum
value for the reliability index
1 year
reference
period
50 years
reference
period
CC3
High consequence for loss of
human life, or economic, social
or environmental consequences
very great
Grandstands, public
buildings where
consequences of failure are
high (e.g. a concert hall)RC3 5,2
4,3
(= 3,8 + 0,5)
Definition of consequences classes and associated reliability classes
CC2
Medium consequence for loss
of human life, economic, social
or environmental
consequences considerable
Residential and office
buildings, public buildings
where consequences of
failure are medium (e.g. an
office building)
RC2 4,7 3,8
CC1
Low consequence for loss of
human life, and economic,
social or environmental
consequences small or
negligible
Agricultural buildings where
people do not normally enter
(e.g. storage buildings),
greenhousesRC1 4,2
3,3
(= 3,8 – 0,5)
In the « Common unified rules for different types of construction and material »
elaborated by JCSS in 1976, a target value for the probability of failure in one year,
Pf,1 = 1.10-5 was proposed for the reference case with serious economic consequences
and a medium number of persons endangered (the « acceptable » probability of
failure).
This corresponds to Pf,50 = 5.10-4 and the equivalent β value is 3.29.
Finally, the target value for β50, adopted for the reference case and the reference period R
= 50 years, is 3.8, which corresponds to Pf,50 = 7,2.10-5
For a reference period of 1 year, the corresponding β1 value is 4.7.
Model Code CEB 78 : Failure probability in 50 years (ULS)
PROBABILITY ASSOCIATED TO LIMIT STATES
Average num-ber of endan-gered people
Economic consequences
Low Medium High Low (< 0,1) 10-3 10-4 10-5
Medium 10-4 10-5 10-6 High (> 10) 10-5 10-6 10-7
Model Code CEB 78 : Failure probability in 50 years (ULS)
It should be stressed that the beta-values and the corresponding
failure probabilites are formal or notional numbers, intended
primarily as a tool for developing consistent design rules, rather thanprimarily as a tool for developing consistent design rules, rather than
giving a measure of the structural failure frequency. The CEB MC 78
should not be misinterpreted.
EN 1990 - Annex C
Basis for Partial Factor Design and Reliability Analysis
Target reliability index ββββ for Class RC2 structural members 1)
Limit state Target reliability index
1 year 50 years 1 year 50 years
Ultimate 4,7 3,8 Fatigue 1,5 to 3,8 2) Serviceability (irreversible) 2,9 1,5
1) See Annex B 2) Depends on degree of inspectability, reparability and damage tolerance.
Hypotheses :
R and E are independently Normally distributed with the usualparameters (µ R, σσσσR) and (µE, σσσσE)
⇒⇒⇒⇒ Z = R – E is also Normally distributed with the followingparameters :
Reliability - Basic approach
22)()( ERZERZ
Z
Z µµµµz
zF σσσσ
+=−=−Φ=
)(2
1
2
1)0(
_
2
0
_
2
)( 2
2
2
Z
Z
tx
Z
f
µdtedxeZPP
Z
Z
Z
Z
σππσ
σµ
σµ
−Φ===<= ∫∫
−
∞
−
∞
−−
∫∞
−=Φ
x t
dtex_
2
2
2
1)(
π
Zσ
(cumulative standardized Normal
distribution function)
ER
ey
rx
σσ==
E
EC
R
RC yx
σσµ=µ=
Geometrical interpretation (1)
Standardized coordinates
Design point
βαβα
ECP
RCP
yy
xx
−=−=
Rσα =
Likely state
C22
22
ER
EE
ER
RR
σσσα
σσσα
+−=
+=
E
E
R
R ey
rx
σσµ−=µ−=
Geometrical interpretation (2)
22
22
ER
EE
ER
RR
σσσα
σσσα
+−=
+=
P : Design point
αβ−=OP
Design values of actions and resistances
EEEdRRRd ER σαβµσαβµ 00 −=−=
0βTarget value of the reliability index :
0ββ ≥⇔≥ dd ER
)()(1
1)(Pr1)(Pr
00 βαβασ
EE
E
Eddd
µEEEobEEob
Φ=−Φ−=
−Φ−=≤−=>
)()(Pr 0βασ R
R
Rdd
µRRRob −Φ=
−Φ=≤
Design values and associated probabilities
Numerical approach
R
E
RERREER
EEERRRdd
with
t
ER
σσ
σβσσβ
σαβσαβ
= t
1 2
0
22
0
00
+−µ−µ=+−µ−µ=
+µ−−µ=−
Comparison between and T = 0,8 + 0,7t21 tT +=
0
1
2
3
4
5
6
7
T=0,8+0,7t
00 1 2 3 4 5 6 7
)7,0()8,0(
)7,08,0(
00
0
EERR
R
ERERdd ER
σβµσβµσσσβµµ
+−−=
+−−≅−
The choice of αE = -0,70 and αR = +0,80 leads to a safe-sided design point P with:
that, together with a lower limit
gives a tolerance field for practical applications, that is defined by the limits:
05,48,380,070,0' 22 =×+=β
55,380,305,4
80,3" =×=β
63,716,0 ==R
E
R
E MaxMinσσ
σσ
RR σσ
Conclusions
Dominating actions and resistances
Non-dominating actions
EEdRRd ER σβµσβµ 00 7,08,0 +=−=
)8,0()(Pr
)7,0()(Pr
0
0
ββ
−Φ=≤−Φ=>
d
d
RRob
EEob
)28,0()7,04,0()(Pr 00 ββ −Φ=−Φ=> xEEob d
Partial factors
00d
k
EEE
k
dF
EE
E σβαµγ 0−==k
FFFSd
k
dSdF
FF
F σβαµγγγ 0−==
RRR
k
d
kM
R
R
R
σβαµγ
0−==
fRf
kRd
d
kRdM
f
f
f
σβαµηγηγγ
0−==
Effects of actions and
Resistances
Individual actions and
resistances
Coefficients of variation
E
EE
R
RR VV
µµµµσσσσ
µµµµσσσσ ========
RRRk
EEEk
kµR
kµE
σσσσσσσσ
−−−−====++++====
Characteristicsafety factor
EE
RR
EEE
RRR
k
k
kVk
Vk
k
k
E
R
++++−−−−××××====
++++−−−−========
1
1γγγγσσσσµµµµσσσσµµµµγγγγ
)(1
22222ββββγγγγ
γγγγ
γγγγσσσσσσσσµµµµµµµµββββ f
VV REER
ER ====⇒⇒⇒⇒++++
−−−−====++++
−−−−====
Verification with partial factors
kMF
M
kkF
RE γγγγγγγγγγγγ
γγγγγγγγ ≤≤≤≤××××⇒⇒⇒⇒≤≤≤≤
Reliability index
Mγγγγ
kMFMFk γγγγγγγγγγγγγγγγγγγγγγγγγγγγββββ ≤≤≤≤××××→→→→→→→→→→→→ /),(
Global approach
E
E
σσσσ
CCmm
E
mE
σσσσ
P Design
point
Boundary between safety
and failure
ββββββββE
dE
σσσσ γγγγγγγγFF
γγγγγγγγ
CCkk
CCdd
kd
RR
γγγγ====
kFd EE γγγγ=====β
E
EEm
E
k kEE
σσσσσσσσ
σσσσ++++====
DDff
DDss
Reliability index
R
R
σσσσ
Eσσσσ
R
dR
σσσσ
γγγγγγγγMM
R
mR
σσσσM
dRγγγγ
====
R
RRm
R
k kRR
σσσσσσσσ
σσσσ−−−−====
M
kkF
RE
γγγγγγγγ ≤≤≤≤
Interpretation of partial factors for permanent actionsγγγγG,sup = 1,35 γγγγG,inf = 1,00
In general (simplification): Gk = Gm
With:
)7,01(sup, VGG md β+=
)7,01(sup,
sup, VG
GSd
m
d
SdG βγγγ +==
05.020.18.3 === VγβWith:
Where favourable, a permanente action may be considered as a resistance. Thus:
05.020.18.3 === VSdγβ
mmd GGG 36.1)05.08.37.01(20.1sup, =××+=
mmmd GGGG ≅=××−= 02.1)05.08.38.01(20.1inf,
Interpretation of partial factors for variable actions
(buildings)
γγγγQ = 0 or 1,50
Qk is assessed on the basis of a return period equal to 50 years (climatic actions and imposed
loads on buildings). In fact, the real return period is more 100 to 200 years if we take into account the
safety margins included in the models.
In any case, it was generally admitted that the design values of the leading actions, γSd
excluded, did correspond to a return period of about 4000 years (in fact, a return period of 1.4xk
between 1000 and 10 000 years).between 1000 and 10 000 years).
Example: X follows a Gümbel law (annual maxima). Coefficient of variation: V
Characteristic value: xk with F(xk) = 0,98 (return period T(xk) = 50 years)
Return period T(1.4xk) = 5720 years for V = 0,15
2433 years for V = 0,20
1457 years for V = 0,25
Finally:
γQ = 1,50 = γq × γSd =1,40 x1,10 = 1,54 # 1,50
Interpretation of partial factor for concrete
Traditional interpretation (CEB)
γγγγC = 1,5 = 1,10 x 1,10 x 1,24
• 1,24 : transition from the characteristic fractile 0,05 (5%) to the 0,005 (0,5%) fractile for a
coefficient of variation circa 0,15 with a Gaussian scatter and circa 0,23 with a log-normal
scatter ;
• 1,10 : average value of the conversion factor (in general between 1,0 and 1,20) from strength
of control samples to the effective strength in the structure (η factor) ;
• 1,10 : covers those uncertainties which cannot be accounted for in a purely probabilistic way
(model uncertainties in the design modelfor the resistance, inaccuracies in the execution of the
structure, more particularly deviations from the theoretical location of the reinforcement,
etc.).
Interpretation of partial factors for steel
Traditional interpretation (CEB)
Concrete reinforcement and prestressing tendonsγS = 1,15 = 1,05 x 1,05 x 1,05
· 1,05 : transition from the characteristic fractile 0,05 (5%) to the 0,005 (0,5%) fractile, for a coefficient
of variation circa 0,05 ;
· 1,05 : little loss of cross-sectional area due to long-term corrosion ;
· 1,05 : covers uncertainties on the flexural resistance of a member due to the real location of the · 1,05 : covers uncertainties on the flexural resistance of a member due to the real location of the
reinforcement.
Structural steel
γM0 = 1,0 : Cross-sections susceptible to reach plasticizing
γM1 = 1,1 : Cross-sections the resistance of which is limited by local instability.
The models of resistance of steel structures, in particular for the design of joints, are more precise than
the models of resistance of the concrete structures.
Interpretation of partial factors for material strength:
interpretation based on reliability methods
)exp(
)exp(
RR
f
d
kM
V
kV
R
R
βαµµ
γ−
−==
8,08,3645,1 === Rk αβ
222
fGmR VVVV ++=
)645,104,3exp( fR
d
kM VV
R
R −×== ηγ
Vm model uncertainty
VG geometrical uncertainty
Vf uncertainty on the property (strength)
Material Vm VG Vf VR γγγγconv γγγγM
Concrete 0,05 0,05 0,15 0,166 1,15 1,49
Partial factors for resistances
Reliability-based interpretation
Concrete 0,05 0,05 0,15 0,166 1,15 1,49
Reinforcement 0,05 0,05 0,05 0,087 1 1,20
Structural steel 0,03 0,03 0,03 0,052 1 1,115
• Partial factors are intended to ensure an appropriate reliability level to the majority of
construction works and parts of construction works, in order to limit the frequency of
accidents of structural origin.
•Partial factors (and reliability index) have not been defined by Heaven !
• Partial factors are not calibrated, and are not intended, to cover gross human errors.
• But, in fact, they cover minor human errors. Where is the limit between minor and gross
human errors (good question) ?
Conclusion: a few basic ideas
human errors (good question) ?
• And how a distinction between errors due to design, calculation (models, use of
sophisticated software) and execution may be distinguished ?