Vous êtes sur la page 1sur 15

CE 424/CE 524

Structural Steel Design

Chapter 2
Design Philosophies: LRFD/ASD

Dr. Mahmoud Reda Taha, P. Eng.


Department of Civil Engineering, University of New Mexico

© Dr. M. M. Reda Taha, 2007.

Table of contents
‰ Design philosophies

‰ Probabilistic Basis for LRFD

‰ Reliability index

‰ Determining load and resistance factors

‰ AISC load and resistance factors


„ LRFD
„ ASD

CE 424/524 - Chapter 2 Slide Number 2


Design Philosophies

‰ Allowable Stress Design (ASD)

‰ Plastic Design (PD)

‰ Load and Resistance Factor Design (LRFD)

CE 424/524 - Chapter 2 Slide Number 3

Allowable Stress Design


‰ Service loads are calculated as expected during service life.
‰ Linear elastic analysis is performed.
‰ A factor of safety (FOS) of the material strength is assumed
(usually 3-4)
Material Strength
Allowable Stress =
FOS

‰ Design is satisfactory if (maximum stress < allowable stress)


‰ Limitations
„ Case specific, no guarantee that our design covers all cases
„ Arbitrary choice of FOS?! Can this be resolved ?!

CE 424/524 - Chapter 2 Slide Number 4


New version of ASD
‰ The new AISC code introduces a new trend for ASD
Rn
Ra ≤

‰ The service loads shall be computed considering all possible load
combinations and this can result in determining the required
strength Ra

‰ A new factor of safety Ω is introduced. The new factor of safety


is > 1.0 and is derived using probabilistic methods
‰ By dividing the nominal strength Rn by the Ω factor of safety,
the concept of allowable stress is satisfied.

CE 424/524 - Chapter 2 Slide Number 5

Plastic Design
‰ Service loads are factored by a “load factor”.
‰ The structure is assumed to fail under these loads, thus,
plastic hinges will form under these loads “Plastic Analysis”.
‰ The cross section is designed to resist bending moments and
shear forces from the plastic analysis.
‰ Members are safe as they are designed to fail under these
factored loads while they will only experience service loads.
‰ Limitations
„ No FOS of the material is considered, neglecting the uncertainty in
material strength!
„ Arbitrary choice of overall FOS?!

CE 424/524 - Chapter 2 Slide Number 6


Load and Resistance Factor Design (LRFD)
‰ LRFD is similar to plastic design in that it performs design with the assumption of
failure!
‰ Service loads are multiplied by load factors (γ) and linear elastic analysis is
performed.
‰ Material strength is reduced by multiplying the nominal material strength by a
resistance factor (φ)
‰ The design rule is

This rule shall be attained


∑ γ i Q i ≤ φi R n for all limit states!!

„ Where Rn is the nominal strength and Q is the load effect for the ith limit state

‰ Advantages of LRFD
„ Non-case specific, statistical calculations guarantee population behavior.
„ Uniform factor of safety as both load and material factors are tied by reliability analysis.

CE 424/524 - Chapter 2 Slide Number 7

Probabilistic Basis for LRFD


‰ The basic statistical information we can get are the mean and the
standard deviation
„ Mean of a sample population
µ 1 n
µ= ∑ xi
n i
„ Standard deviation of a sample population
σ
1 n
σ= ∑
n i
( xi − µ )2
(n-1) for samples of less than 30 observations

‰ We can also calculate the coefficient of variation (V)

V = σµ
CE 424/524 - Chapter 2 Slide Number 8
Probabilistic Basis for LRFD

Probability density Cumulative probability


function (PDF) b

1
⎛ − ( x − µ )2

⎜ 2σ 2



P( a < x < b ) = ∫ P ( x ) dx
P( x ) = e ⎝ ⎠ a

σ 2π

a b
CE 424/524 - Chapter 2 Slide Number 9

Probabilistic Basis for LRFD

‰ In LRFD we assume that both loads and resistance are uncertain


and need to be dealt with through the theory of probability
‰ For example: Test results of a L 3 ½ x 3 ½ x 3/8 are:

Test # Load
(kips)
1 87.2
2 87.3
3 85.4
4 87
5 86.5
6 86.9
7 88
8 89
9 84.3
10 87

CE 424/524 - Chapter 2 Slide Number 10


Probabilistic Basis for LRFD
‰ The mean and the standard deviations for the test results are:
„ µ = 86.9 kips
σ = 1.29 kips Probability density function (PDF)

‰ From the previous graph


we might assume normal
distribution of the test
results
‰ Thus by using the normal
“Gaussian” distribution
we get this curve
⎛ −( x − µ )2 ⎞
⎜ ⎟
1 ⎜ 2 σ 2 ⎟⎠
P( x ) = e⎝
σ 2π

CE 424/524 - Chapter 2 Slide Number 11

Probabilistic Basis for LRFD


‰ Materials and load distributions have been shown by researchers to follow
normal or log-normal distributions
‰ If normal distribution is used, then
P( R ≤ Q ) thus if x = R − Q then POF = P( x ≤ 0 )
‰ If log-normal distribution is used, then
⎛R⎞ ⎛R⎞
P( ln⎜⎜ ⎟⎟ ≤ 0 ) thus if x = ln⎜⎜ ⎟⎟ then POF = P( x ≤ 0 )
⎝Q⎠ ⎝Q⎠
Normal or Log-normal distribution
Mean (µ)

POF = Area left to the line

x = zero
CE 424/524 - Chapter 2 Slide Number 12
Probabilistic Basis for LRFD
‰ The reliability index “β” is tied to this distribution.
‰ It simply represents how far the mean from the critical point as multiples of
the standard deviation
‰ If we consider log-normal distribution for example

⎛R⎞
ln ⎜⎜ ⎟⎟
β= T = ⎝ ⎠
x Q
σx σ ⎛R⎞
ln ⎜⎜ ⎟⎟
Mean (µ)

⎝Q⎠

xT
Each β corresponds to a
specific probability of failure
x = zero
CE 424/524 - Chapter 2 Slide Number 13

Probabilistic basis for LRFD


‰ Looking at the two distributions of resistance and load effect

Resistance “R”

R
Frequency (%)

Load effect “Q”

Q
R −Q
Q R

R-Q

CE 424/524 - Chapter 2 Slide Number 14


Probabilistic Basis for LRFD
‰ If we have the probability distribution of the load effect (Q) and the
material resistance (R) then:
„ The probability of failure can be represented by observing the probability of
the function (R-Q)
„ The probability of failure PF can be represented as the probability that Q ≥ R:

Probability
of failure

CE 424/524 - Chapter 2 Slide Number 15

Probabilistic Basis for LRFD


‰ We can view the relation between these probabilities from another
prospective

( R − Q ) Cos45 On this line Q = R,


we are interested in
Q> R

N ( R − Q ) Cos 45
β= =
R σQ σQ
The higher the β, the bigger the safety radius
o
45

Q
N

CE 424/524 - Chapter 2 Slide Number 16


Reliability Index
σR
‰ For lines inclined by an angle θ we can prove cos θ =
σ R2 + σ Q2

⎡σ ⎤
( R − Q ) ⎢ R ⎥Cosθ
⎢⎣σ Q ⎥⎦
β=
σQ
(R − Q )
β=
N
β=
σQ σ R2 + σ Q2

‰ The reliability index “β” is a function of both


„ The load effect “Q” and the resistance “R” and their probability distributions.
„ It represents how confident are we in our decision that the resistance of the
material is higher than the load effects.

CE 424/524 - Chapter 2 Slide Number 17

Reliability Index
‰ By considering the previous graph
„ The higher the parameter “β”, the lower the probability of failure “PF”
„ The parameter “β” is known as “The reliability index”
Lognormal variables Normal variables
⎡ R ⎤
⎢ ln( Q ) ⎥ Rm − Qm
β = ⎣ ⎦m
OR β=
σ⎡ R ⎤
⎢ ln( Q ) ⎥
σ R2 + σ Q2
⎣ ⎦

The formula we will use

‰ The reliability index “β” is a function of both


„ The load effect “Q” and the resistance “R” and their probability distributions.
„ By targeting a specific “Reliability index” for all the design elements, a consist
tent level of safety in design can be achieved
CE 424/524 - Chapter 2 Slide Number 18
Determining load and resistance factors

‰ As the load effect and resistance distributions can be determined by


measurements, the reliability index “β” (probability of failure) for any
combination of loads and materials can be determined.

‰ However, determining the probability of failure for a specific load and


material combination is not the design target. The target is to determine the
load and resistance factors that can achieve a specific probability of failure

‰ The following equation can be used for determining the resistance factor φ
for a specific reliability index “β”:
R m − 0 .55 β V R
φ = e
Rn
‰ VR is the coefficient of variation of the resistance

‰ The load factor “γ“ can be determined : see next example


CE 424/524 - Chapter 2 Slide Number 19

Example 1

‰ For the shown connection:


„ The 50 years wind records are used to calculate the maximum load effect
„ Experimental results of the connection resistance are recorded
Test # R (kips) Record # Pw (kips)

1 23.5 1 14.8
2 28.1 2 14.1
PW
3 24 3 18.4
4 26.5 4 16.3
Rn = 25.4 kips 5 25.3 5 16.9
6 22.2 6 19.8
7 21.2
‰ Determine 7 25.3
8 26.2 8 18.4
„ The probability of failure 9 14.8
9 24.9
„ Load and resistance factors for a 10 22.3 10 19.1
probability of failure of 0.01%

CE 424/524 - Chapter 2 Slide Number 20


Example 1

‰ To determine the probability of failure we need to determine the


probability that Qm>Rm
‰ First: Determine the statistical parameters for Q and R

Rm = 24.83 kips
σ R = 1.87 kips
VR = 0.075

Qm = 17.38 kips
σ Q = 2.36 kips
VQ = 0.136

CE 424/524 - Chapter 2 Slide Number 21

Example 1

‰ The probability of failure represents the probability that Qm>Rm

Rm − Qm
β= β = 2.47 POF = 0.67%
σ R2 + σ Q2

‰ This is relatively high POF, we need to determine the load and


resistance factors to achieve a specific POF (POF<0.01%)

POFT = 0.01 % βT ≈ 3.75 “Inverse of cumulative normal


distribution”

RT − Qm = βT σ R2 + σ Q2

RT − Qm = βT VR2 RT2 + σ Q2 Solve for RT

CE 424/524 - Chapter 2 Slide Number 22


Example 1

‰ The load factor can now be determined

RT
γ= γ ≈ 1.70
Qm

‰ We can also determine the load resistance factor φ from the equation below

Rm −0.55 β VR
φ= e φ = 0.83
Rn
‰ We can repeat the previous process for different POFs to examine the effect of
changing the POF or “β” on load and resistance factors
‰ See MATLAB code on the web
http://civilx.unm.edu/Courses/CE_424/downloads.htm
CE 424/524 - Chapter 2 Slide Number 23

AISC – Steel Design


‰ Historically,
„ LRFD concepts have been adopted in Canada since 1970s and
in Europe since early 1980s.
„ AISC adopted LRFD since 1986.
„ AISC Introduces a combined method in 2007!

‰ The new design provisions in AISC (2007) version allows the design
using one of two methods
„ Load and Resistance Factor Design (LRFD)
„ Allowable Stress Design (ASD)
‰ AISC tries to establish a uniform method by introducing a new ASD
design method, where ASD safety factors are derived probabilistically.
Therefore, both methods are similar and shall yield close results
CE 424/524 - Chapter 2 Slide Number 24
AISC – LRFD Design
‰ Design using AISC will target a specific probability of failure
“Reliability index” “β” to achieve a consistent design at the
different design load combinations and limit states

‰ AISC Reliability index (β)


Loading Conditions
D + (L or S) D+L+W D+L+E
Members 3.0 2.5 1.75
Connections 4.5 4.5 4.5

‰ Where
„ Dead loads (D)
„ Live loads (L)
„ Wind Loads (W)
„ Earthquakes (E)
CE 424/524 - Chapter 2 Slide Number 25

Load Designation
‰ Based on definitions by ASCE document on load and load combinations (2002),
AISC considers the following loads for designing of Steel structures.
‰ Dead loads (D)
‰ Live loads (LL)
„ Occupancy load (L)

„ Roof load (Lr)

„ Snow load (S)

„ Rain loads (R)

‰ Wind Loads (W)


‰ Earthquake load (E)
‰ Lateral earth pressure (H)
‰ Fluid pressures (F)
‰ Self-restraining force (T)

CE 424/524 - Chapter 2 Slide Number 26


LRFD: Load and Resistance Factors
‰ AISC considers the following load combinations in design
1− 1.4 D
∑γ i Qi ≤ φi Rn
2 − 1.2 D + 1.6 L + 0.5( Lr or S or R )

3 − 1.2 D + 1.6 ( Lr or S or R ) + 0.5L or (0.8W )


∑γ i Qi
4 − 1.2 D + 1.6 W + 0.5 L + 0.5 ( Lr or S or R )
5 − 1 .2 D ± 1 . 0 E + 0 . 5 L + 0 . 2 S

6 − 0.9 D ± (1.3 Wor 1.0 E )


‰ For garages, load factor for L in load combinations 3,4 and 5 shall be 1.0 and not 0.5 ( L = 100 psf)

φi Rn φ = 0.75 − 1.00 for yield φ = 0.9 and for bolt shear φ = 0.75

CE 424/524 - Chapter 2 Slide Number 27

ASD: Load and Resistance Factors


‰ AISC considers the following load combinations in design Rn
Ra ≤
1− D Ω
2− D+L
3− D + ( Lr or S or R )
Ra 4− D + 0.75 L + 0.75 ( Lr or S or R )
5− D ± ( W or 0.7 E )
6− D + 0.75( W or 0.7 E ) + 0.75 L + 0.75( Lr or S or R )
7− 0.6 D ± ( W or 0.7 E )

1.5
Ω Ω= for yield Ω = 1.67 and for bolt shear Ω = 2.0
φ
CE 424/524 - Chapter 2 Slide Number 28
Example 2

CE 424/524 - Chapter 2 Slide Number 29

References

‰ Segui, W. T., LRFD Steel Design, Fourth Edition, 2007, Thompson,


Brooks/Cole, USA.
‰ Manual of Steel Construction, Load and Resistance Factor Design, American
Institute of Steel Construction (AISC), 13th Edition. 2005
‰ McCormac, J. C. and Nelson, J. K., Structural Steel Design: LRFD Method,
3rd Edition, 2003, Prentice Hall, NJ, USA
‰ Kulak, G. L. and Gilmor, M. I., Limit State Design in Structural Steel, 6th
Edition, 1998, Canadian Institute of Steel Construction, Alliston, Ontario,
Canada.
‰ Loov, R. E., Structural Steel Design: Lecture Notes, 1997, Calgary, Canada.

CE 424/524 - Chapter 2 Slide Number 30