Professional Documents
Culture Documents
2.1. GENERAL ASPECTS In order to check the safety of a structure it is necessary to assess whether a dangerous situation, able to make the structure unusable, might be reached due to some extreme events. There are three types of methods to make the analysis of steel structure reliability: deterministic methods, which consider all parameters with their deterministic values; probabilistic methods, which consider all parameters and the relations among them as random variables; they are difficult to carry on and they need a very sophisticated mathematical procedure; they also need a great amount of data about loads, material properties etc.; semi-probabilistic methods, which use probabilistic models to establish the values for actions and capacities but they compare them using deterministic models; most of present day design codes for steel structures use such methods. Generally, when checking the safety of a structural element or of a whole structure, the following requirements are to be satisfied: strength requirement; stiffness requirement.
In some cases, like seismic design, ductility requirements need also to be fulfilled. 2.2. ALLOWABLE STRESS METHOD (DETERMINISTIC METHOD) In this method the strength requirement is expressed by the following relation: all In this equation (2.1) the allowable stress all is given by: all = fy c ( 2.2 ) ( 2.1 )
where c is a global safety coefficient taking into account the following possibilities:
33
2. RELIABILITY OF STEEL STRUCTURES actual nominal loads considered in calculating the effective stress in equation (2.1) could be greater than assumed; actual nominal yielding stress fy in equation (2.2) could be lower than presumed; fabrication and/or erection may produce unfavourable effects. The stiffness requirement is expressed by the following equation (same as (1.2)): a ( 2.3 )
where and a are the calculated and the allowable deformation respectively. Critical remark The method considers only a simultaneous increase of the loads that can unfavourably affect a correct analysis of the reliability, especially when permanent loads (dead loads) are significantly smaller than the imposed ones (live loads). 2.3. PROBABILISTIC ANALYSIS OF RELIABILITY 2.3.1. Probabilistic bases A more rational approach to analyse the problem of structural safety is a probabilistic one. In such a model of analysis, all the parameters whose uncertainty can influence the reliability of structures, especially those ones concerning resistance and loads, are considered as random variables. 2.3.2. Resistance randomness
The resistance is defined in EN 1990 [10] (1.5.2.15) as the capacity of a member or component, or a cross-section of a member or component of a structure, to withstand actions without mechanical failure e.g. bending resistance, buckling resistance, tension resistance. Strength is used in EN 1990 [10] (1.5.2.16) to express the mechanical property of a material indicating its ability to resist actions, usually given in units of stress.
The resistance R(s) of a structural member with respect to a certain internal force S (N, M, Q) may be expressed in a general form by: R (s ) = f (, R d ) ( 2.4 )
where is the cross-sectional characteristic corresponding to the internal force S, i.e.: =A =W for members in tension; for members in bending.
34
2. RELIABILITY OF STEEL STRUCTURES For industrially fabricated steel structural members, the cross sectional characteristic may be considered as a deterministic value. The yield stress fy must be considered as a random variable. The following steps are to be followed to define the random variable x = fy: consider the results on a sample of n = ni tensile specimen tests (i.e. n values of yield stress fy); according to the values given in table 2.1, draw the histogram in figure 2.3, noticing that the normalized area of any rectangle on the histogram represents the ratio: fi = ni n = i n ni ( 2.5 )
where ni is the number of samples satisfying the condition: fy,i < x fy,i + fy where fy = 20 N/mm2 as shown in figure 2.3. Table 2.1. Example of values of the yielding limit fy Results association Frequency of results fi xi Calculation mean value xm (N/mm2) dispersion D (N2/mm4) (xi xm)2 fi (xi xm)2 207.0461 93.42882 87.45607 6.622875 61.84293 127.0923 162.5884 D=746.0775 s = (D)0,5 = 27.31442 ( 2.6 )
Interval of association 220 240 240 260 260 280 280 300 300 320 320 340 340 360
Interval Absolute Relative ni fi central values xi 230 250 270 290 310 330 350 20 19 59 140 101 40 21 0.05 0.0475 0.1475 0.35 0.2525 0.1 0.0525
11.5 4140.923 11.875 1966.923 39.825 592.9225 101.5 18.9225 78.275 244.9225 33 1270.923 18.375 3096.923
35
0 220
Fig. 2.3. Histograms corresponding to the values in table 2.1 It is to observe that any rectangle fi represents the relative frequency of the results (simple probability) and in this case the normalized area of the whole histogram is:
=1
( 2.7 )
xm = fi xi
i =1
( 2.8 )
(for the case in table 2.1, xm = 294N/mm2) calculate the dispersion: D = s2 = fi (xi xm )
i =1 n 2
( 2.9 )
(for the case in table 2.1, D = 746N2/mm4) calculate the standard deviation:
s=
f (x x )
i =1 i i m
( 2.10 )
(for the case in table 2.1, s = 27,3N/mm2) The values xm and s define the random variable. The histogram in figure 2.3 may be represented by the normal (Gaussian) function of probability density described by (Fig. 2.4):
1 f (x ) = e 2 s 2 1 x xm s
2
( 2.11 )
The characteristic value of the yield stress fy may be defined in a probabilistic manner by the following relation:
f y,k = f y,m k s
( 2.12 )
36
Codes usually accept k = 2, which represents a probability of 2,28% (inferior fractil p) that the yield stress will not be inferior to fy,k. It means:
f y ,k = f y ,m 2s
( 2.13 )
The fractil p is defined as that value of the yield stress for which there is a probability p for the yield stress to be inferior to that value. By noting: v= s xm ( 2.14 )
f y ,k = f y , m (1 2 v )
( 2.15 )
The internal force S(Fk) in a certain cross-section of a structural member, with regard to the type of load and the structural model of calculation, may be written as: S(Fk) = (L) where: L represents the acting loads; are formulas derived from accepted principles of structural model of calculation. For a simply supported beam, the maximum bending moment is: ( 2.16 )
Example:
37
F = Mmax =
q D2 8
(x ) =
A histogram may be drawn in the same way as described for steel randomness, determining the mean value Fm and the standard deviation s for loads (Fig. 2.5). f(F) 0.02
superior fractil
ks
( 2.17 ) ( 2.18 )
The characteristic value Fk, depending on the loads, may be written as:
Codes usually accept k = 1,645, corresponding to a 5% probability for the value Fk to be exceeded (superior fractil p).
2.3.4. Safety analysis
Basically, to assess the safety of a structure in the probabilistic concept means to check that the probability p of exceeding a given limit state is not greater than an a priori chosen probability pu, depending on the consequences of reaching that limit state (Fig. 2.6). p pu ( 2.19 )
38
f(S) f(R)
f(S)
f(R)
S, R P
the semi-probabilistic method (level 1); the reliability index method (level 2); the exact probabilistic method (level 3).
2.3.6. The semi-probabilistic limit states method (level 1) 2.3.6.1. The Eurocodes Structural EUROCODES is a programme for establishing a set of
harmonised technical rules for the design of construction works in Europe. In a first stage, they were intended to be an alternative to the national design codes and in the end, they will replace the national rules. The Structural Eurocode programme comprises the following standards, each one consisting of several parts:
EN 1990 EN 1991 EN 1992 EN 1993 EN 1994 EN 1995 EN 1996 EN 1997 EN 1998 EN 1999 Eurocode 0: Basis of structural design Eurocode 1: Actions on structures Eurocode 2: Design of concrete structures Eurocode 3: Design of steel structures Eurocode 4: Design of composite steel and concrete structures Eurocode 5: Design of timber structures Eurocode 6: Design of masonry structures Eurocode 7: Geotechnical design Eurocode 8: Design of structures for earthquake resistance Eurocode 9: Design of aluminium structures
39
A limit state can be defined as the state beyond which the structure no longer fulfils the relevant design criteria (1.5.2.12 in EN1990 [10]). There are two categories of limit states: 1. ultimate limit states, which are states associated with collapse or with other similar forms
of structural failure and they generally correspond to the maximum load-carrying resistance of a structure or
structural member (1.5.2.13
a) 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; b) 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; c) GEO: Failure or excessive deformation of the ground where the strengths of soil or rock are significant in providing resistance; d) FAT: Fatigue failure of the structure or structural members.
2. serviceability limit states, which refer to the normal use of the structure and correspond to conditions beyond which specified service requirements for a structure or
structural member are no longer met (1.5.2.14 in EN1990 [10]). Serviceability limit states
NOTE 1 In the context of serviceability, the term "appearance" is concerned with such criteria as high deflection and extensive cracking, rather than aesthetics. NOTE 2 Usually the serviceability requirements are agreed for each individual project.
40
irreversible serviceability limit states (1.5.2.14.1 in EN1990 [10]) serviceability limit states where some consequences of actions exceeding the specified service requirements will remain when the actions are removed;
reversible serviceability limit states (1.5.2.14.2 in EN1990 [10]) serviceability limit states where no consequences of actions exceeding the specified service requirements will remain when the actions are removed.
The verification of serviceability limit states should be based on criteria concerning the following aspects: a) deformations that affect the comfort of people; the appearance, the comfort of users, or the functioning of the structure (including the functioning of machines or services), or that cause damage to finishes or non-structural members; b) vibrations; that cause discomfort to people, or that limit the functional effectiveness of the structure; c) damage that is likely to adversely affect
2.3.6.2. Actions
41
1.5.3.6 seismic action (AE) action that arises due to earthquake ground motions 1.5.3.8 fixed action action that has a fixed distribution and position over the structure or structural member such that the magnitude and direction of the action are determined unambiguously for the whole structure or structural member if this magnitude and direction are determined at one point on the structure or structural member 1.5.3.9 free action action that may have various spatial distributions over the structure 1.5.3.11 static action action that does not cause significant acceleration of the structure or structural members 1.5.3.12 dynamic action action that causes significant acceleration of the structure or structural members 1.5.3.13 quasi-static action dynamic action represented by an equivalent static action in a static model
42
1.5.3.20 representative value of an action (Frep) value used for the verification of a limit state. A representative value may be the characteristic value (Fk) or an accompanying value (Fk)
( 2.20 )
is the partial safety factor for the action Fi, being Fi = (Pi, Ci, Vi, Ei).
Fundamental combination
n P + n C + n n V
i i i i g i
( 2.21 )
Special combination
P + C + n
i i
d i
Vi + E1
( 2.22 )
In equations (2.21) and (2.22): ng is a factor taking into account the probability of simultaneous action of a number of variable actions (Vi) at their highest intensity:
ng = 1 ng = 0,9 ng = 0,8
for one Vi; for two or three Vi; for four or more Vi.
ni is a factor representing the long lasting part of a variable action; nid < 1. The ultimate limit states are usually examined considering the effects of the design values of actions, while for serviceability limit states the characteristic values of actions are generally used. 43
2. EN 1990 [10] uses design situations to express the requirements to be fulfilled for each limit state. Design situations (1.5.2.2 in EN1990 [10]) are sets of physical
conditions representing the real conditions occurring during a certain time interval for which the design will demonstrate that relevant limit states are not exceeded Design situations shall be classified as follows:
persistent design situations, which refer to the conditions of normal use; transient design situations, which refer to temporary conditions applicable to the structure, e.g. during execution or repair; accidental design situations, which refer to exceptional conditions applicable to the structure or to its exposure, e.g. to fire, explosion, impact or the consequences of localised failure; seismic design situations, which refer to conditions applicable to the structure when subjected to seismic events. The design working life (1.5.2.8 in EN1990 [10]) is the 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. Based on this, the design situations are defined as follows:
a persistent design situation (1.5.2.4 in EN1990 [10]) is a design situation that is relevant during a period of the same order as the design working life of the structure; a transient design situation (1.5.2.3 in EN1990 [10]) is a design situation that is relevant during a period much shorter than the design working life of the structure and which has a high probability of occurrence; an accidental design situation (1.5.2.5 in EN1990 [10]) is a design situation involving exceptional conditions of the structure or its exposure, including fire, explosion, impact or local failure; a seismic design situation (1.5.2.7 in EN1990 [10]) is a design situation involving exceptional conditions of the structure when subjected to a seismic event;
Table 2.1 - Indicative design working life
Indicative design working life (years) 10 10 to 25 15 to 30 50 100
Examples Temporary structures (1) Replaceable structural parts, e.g. gantry girders, bearings Agricultural and similar structures Regular buildings and other regular structures Monumental building structures, bridges, and other civil engineering structures
(1) Structures or parts of structures that can be dismantled with a view to being re-used should not be considered as temporary.
According to EN1990 [10], three types of combinations of actions are to be considered when designing steel members:
For persistent and transient design situations, the most unfavourable of:
j 1
G, j
( 2.23a )
44
j 1
( 2.23b )
G
j 1
k, j
( 2.24 )
G
j 1
k, j
P A Ed 2,i Qk,i
i >1
( 2.25 )
= combined with;
Gk,j = characteristic value of permanent action j; P = relevant representative value of a prestressing action; Qk,1 = characteristic value of the leading variable action 1; Qk,i = characteristic value of the accompanying variable action i; Ad = design value of an accidental action; AEd = design value of seismic action A Ed = I A Ek ; AEk = characteristic value of seismic action; I = importance factor, given in EUROCODE 8 (EN 1998) [11]; G,j = partial factor for permanent action j; P = partial factor for prestressing actions; Q,i = partial factor for the variable action i; 0 = factor for combination value of a variable action; 1 = factor for frequent value of a variable action; 2 = factor for quasi-permanent value of a variable action; = a reduction factor for unfavourable permanent actions G. The value for and factors may be set by the National annex. Some examples of recommended values of factors for buildings are given in table 2.2.
45
2. RELIABILITY OF STEEL STRUCTURES Table 2.2. Examples of recommended values of factors for buildings [10]
Table A1.1 - Values of factors for buildings
Action Imposed loads in buildings, category (see EN 1991-1-1) Category A: domestic, residential areas Category B: office areas Category C: congregation areas Category D: shopping areas Category E: storage areas Category F: traffic area vehicle weight 30kN Category G: traffic area 30kN < vehicle weight 160kN Category H: roofs Snow loads on buildings (see EN 1991-1-3)* Finland, Iceland, Norway, Sweden Remainder of CEN Member States, for sites located at altitude H > 1000 m a.s.l. Remainder of CEN Member States, for sites located at altitude H _ 1000 m a.s.l. Wind loads on buildings (see EN 1991-1-4) Temperature (non-fire) in buildings (see EN 1991-1-5:2005) NOTE The values may be set by the National annex. * For countries not mentioned below, see relevant local conditions.
0
0,7 0,7 0,7 0,7 1,0 0,7 0,7 0,71) 0,7 0,7 0,5 0,6 0,6
1
0,5 0,5 0,7 0,7 0,9 0,7 0,5 0 0,5 0,5 0,2 0,2 0,5
2
0,3 0,3 0,6 0,6 0,8 0,6 0,3 0 0,2 0,4 0 0 0
0
0,7 0,7 0,7
1
0,5 0,5 0,7
2
0,3 0,3 0,6
0,7 0,7
*
0,5 0,2
*
0,4 0
*
See SR EN 1991-1-1:2004, 3.3.2(1). * Values of factors will be available after the completion of SR EN 1991-1-5:2005 National Annex.
46
3. According to the American codes ASCE 798 [3] (the latest version is from 2010) and LRFD [4], the following combinations shall be investigated: 1,4 (D + F) 1,2 (D + F + T ) + 1,6 (L + H) + 0,5 (L r or S or R ) 1,2 D + 1,6 W + 0,5 L + 0,5 (L r or S or R ) 1,2 D + 1,0 E + 0,5 L + 0,2 S 0,9 D + 1,6 W + 1,6 H 0,9 D + 1,0 E + 1,6 H being: D F H L Lr S T E R = dead load (Pi + Ci) = load due to fluids with well-defined pressures and maximum heights = load due to lateral earth pressure, ground water pressure or pressure of bulk materials = live load (Vi imposed loads) = roof live load = snow load = self-straining force = earthquake load = rain water or ice
1,4 D 1,2 D + 1,6 L + 0,5 (L r or S or R )
( 2.26 )
Fa = flood load
W = wind load
1,2 D + 1,0 W + L + 0,5 (L r or S or R ) 1,2 D + 1,0 E + L + 0,2 S 0,9 D + 1,0 W 0,9 D + 1,0 E
being: D F = dead load (Pi + Ci)
47
Rd =
where: fk
fk M
( 2.27 )
= characteristic value of the considered material property; For the design strength R of a structural steel, equation (2.27) becomes:
R=
fk M
( 2.28 )
In the limit state method (also called the method of extreme values), the probabilistic condition in equation (2.19) p < pu is replaced by: Sd Rd the minimum probable design resistance capacity Rd. In equation (2.29): Sd = S(niFi) is the internal design effort, calculated using design values of actions and taking into account respectively the load combinations in eqs. (2.21) and (2.22) or (2.23), (2.24) and (2.25) or (2.26), depending on the code;
Rd = R(Rk/M)
( 2.29 )
which means that the maximum probable internal design effort Sd does not exceed
48
The most common serviceability limit state to be checked is the deformation check. It will be verified that:
d a
( 2.30 )
where:
d = (Fi) is the design deformation, calculated using the characteristic (nominal)
values of actions;
1. At present, the limit state method is the design method provided in most of the important codes. 2. It represents a more accurate model compared to the allowable stress method because it separates the material randomness from the load randomness and it accepts different approaches for different types of loads.
2.3.7. The reliability index method (level 2)
In a general form, equation (2.29) becomes at limit: Sd = Rd Equation (2.31) may be written:
( 2.31 )
in the subtract model Rjanitin Cornell as: E = Sd Rd = 0 in the logarithmic model Freudenthal Rosenblueth as: E = ln Sd =0 Rd ( 2.33 ) ( 2.32 )
In equations (2.32) and (2.32) E = 0 is the reliability function, expressing (Fig. 2.8):
E < 0 : safety range; E > 0 : unsafe range; E = 0 : the border between safety and unsafe range.
49
Xj fE Unsafe range E>0 Safety range E<0 mE space E EsE Xi limit hypersurface E = 0 Xn
Fig. 2.8. The reliability index method (level 2)
In the case of a simple internal effort S (= N, M or Q), the reliability index E is defined as the reverse of the coefficient of variation vE of the function E:
E = 1 m = E vE sE ( 2.34 )
( 2.35 )
In equations (2.34) and (2.35) mE and sE are the mean value and, respectively, the standard deviation of the function E. Figure 2.8 shows the physical significance of the reliability index E which represents in hyper-space E the distance calculated in standard deviations sE between the point with the abscissa mE and the point with the abscissa E = 0, located on the random hyper-surface which defines the border between safe and unsafe behaviour, corresponding to a certain probability pu = p(E). The properties of the main statistic characteristics for two variables, X1 and
X2, are given in table 2.3. Table 2.3. Main statistic characteristics Y mY DY vY
X1 C CX1
mX1 C CmX1
DX1 0 C2 DX1
vX1 0 vX1
50
X1 C
mX1 C
DX1
m X1 v X! m X1 C
2 2 2 m2 X1 v X1 + m X 2 v X 2 m X1 + m X 2 2 2 2 m2 X1 v X1 + m X 2 v X 2 m X1 m X 2
X1 + X2
mX1 + mX2
DX1 + DX2
X1 X2 X1 X2
DX1 + DX2
2 m2 X1 D X1 + m X 2 D X 2
2 v2 X1 + v X 2
X1 / X2
mX1 / mX2
1 2 m2 X 2 D X1 + m X1 D X 2 2 mX2
2 v2 X1 + v X 2
For the two models presented above, the reliability index , taking into account the relations in table 2.3, becomes: S R = mR mS DR + DS ( 2.36 )
ln
S R
mR ln m = S 2 2 vR + vS
( 2.37 )
Table 2.4 shows a correspondence between the index and the probability pu of losing the safety for SR (S and R normal distributions) and lnS/R respectively (S and R lognormal distributions). The American code provides the lnS/R index (2.37) and the following targets were selected:
=3 = 4,5 = 2,5 = 1,75 for members under dead + live and/or snow loading for connection s under dead + live and/or snow loading under dead + live + wind loading under dead + live + earthquake loading
( 2.38 )
pu
10
-1
51
1,35 10-3 2,33 10-4 3,17 10-5 3,40 10-6 2,90 10-7 1,90 10-8
Calculate the index SR and lnS/R for the beam in figure 2.9: q I24; Wy = 354cm3 L = 6m q L2 12
M=
Fig. 2.9. Example 2.4
Given: for the loading: mean value: mq = qm = 20kN/m variation factor: vq = 0,1 mean value: dispersion: mRc = Rm = 294N/mm2 DRc = 744N2/mm4
DRc = 744N2/mm4
v = DRc 744 = = 0,093 mRc 294
Calculate the index SR (2.36): S R = mR mS m m 294 169,5 = Rc = = 3,877 > 3,0 DR + DS DRc + D 744 + 287,3
ln
S R
Remarks
1. In this method, the general condition p pu (2.19) is replaced by: u ( 2.39 ) which expresses the condition E > 0 (S > R); u is a risk a priori accepted. 2. At present, this method is used especially to calibrate the partial safety factors in the limit state method and the coefficients ni in the load combinations; in the future it is to be expected that the index method will replace the limit state method. 3. In order to improve the index method two tendencies are to be observed in scientific works: a more adequate location of points on the hyper-surface E = 0; an extension of the method to various non-normal distributions.
In this method the reliability analysis is based on the general condition p pu (2.19), where p is the probability of E > 0, being:
E(S1, S2 ,K, Sn ; R1, R2 ,K, Rn ) = 0
( 2.40 )
a function of random variables Si and Ri and pu an accepted risk, depending on the consequences. At present this method is used only in scientific works.
53