Académique Documents
Professionnel Documents
Culture Documents
5/6/2003
Paper 1
Laterally supported aluminum flexural members with
symmetric cross sections
Yongwook Kim and Teoman Pekz
Abstract
The Specification for Aluminum Structures published by the Aluminum
Association does not fully account for the plastic-ultimate bending capacity: For
compressive component elements, the allowable stress is based on the yield stress.
In this study, possible modifications to the current specification are suggested so
that rather compact extruded flexural members can be evaluated more precisely. As a first
step, limit state stress equations are modified so that the plastic-ultimate capacity is fully
incorporated. Closed form solution of the ultimate shape factor is provided to set a new
cut-off corresponding to the ultimate stress. The ultimate shape factor is simplified for
practical design purposes. A parametric study of component elements using the finite
element analyses suggests a complete form of the modified limit state stress equations.
Using the modified limit state stress equations, more accurate approaches are proposed to
compute the plastic-ultimate moment capacity.
A parametric study of doubly symmetric I-shaped sections using the finite
element method shows that the approaches developed in this study are significantly better
than the current specification provisions. Flexural tests conducted in this research also
support the approaches.
To maintain a certain factor of safety against yielding in the allowable stress
design equations, an approach to use only a potion of the ultimate-plastic capacity is
suggested. Using this approach, a larger inelastic reserve capacity is recognized for the
material with a larger margin between the ultimate and yield stresses.
Paper 1 - 1
5/6/2003
Tension flange of
structural shapes
Fty/ny
Ftu/nu
Tension web of
1.42Ftu/nu
1.30Fty/ny
structural shapes
a
Except for some of 2014-T6, 6066-T6 and 6070-T6 family materials, for which nu
is replaced with ktnu due to notch sensitivity. See AA (2000a) for details.
Fty = tensile yield stress
Ftu = tensile ultimate stress
ny = factor of safety on yield stress
nu = factor of safety on ultimate stress
However, the check against the limit state of the ultimate stress is not available when a
limit state stress is computed for a compressive component element. Compressive limit
state stress equations for a component element refer to yielding, inelastic buckling, and
post buckling ranges as shown in Figure 1.
For this reason, the ultimate-plastic capacity of aluminum members has not been
fully incorporated in the specifications, although aluminum alloys are hardening and
ductile materials.
The limit state stress equations in this figure are expressed in terms of the equivalent
slenderness ratio (p) as defined in Equation (1), originating from Equation (2).
Paper 1 - 2
p =
5/6/2003
3.266 b
b
= for = 1/3
t
kp t
(1)
2E
2E
2
p2
12(1 2 ) ( b / t )
(2)
Fcr = k p
For a flange or web with junctions between component elements idealized as simplysupported boundary conditions, equivalent slenderness ratios depending on plate buckling
coefficient kp are listed in the AA Specification as exampled in Table 2.
Table 2. Equivalent slenderness ratios in the AA Specification
(S.S denotes simply-supported boundary condition)
(a) AA 3.4.16
p = 1.6
t
(b) AA 3.4.15
p = 5.1
t
(c) AA 3.4.18
p = 0.67
t
The post-buckling equation seen in Figure 1 was proposed by Jombock and Clark
(1968) to take account of the non-linear stress distribution of a buckled plate. B and D
factors were determined by Clark and Rolf (1966) simplifying the inelastic plate buckling
equation proposed by Stowell (1948) based on experimental studies.
The cut-off in the yielding range was set against the yield stress without a
scientific basis. It can be assumed that the rather arbitrarily chosen cut-off for column
members to prevent a catastrophic failure after yielding was also employed for the
component elements according to Alcoa (1955). Due to the inelastic reserve capacity,
however, component elements of a flexural member are not as critical as the ones of a
column after yielding. For this reason, the cut-off equations for compressive component
elements are to be investigated.
The ultimate cut-off and shape factor for the ultimate and plastic
flexural capacity
Rigorous analytic ultimate shape factor for rectangular web elements
The ultimate moment capacity of a member can be obtained if ultimate capacities
of component elements are known. For example, in the case of a doubly symmetric Ishaped section, the ultimate capacity of the flange element is the ultimate stress of the
used material, obviously. On the other hand, the ultimate capacity of the web is the
ultimate bending capacity. Since the stress distribution on the web is non-linear, the
moment capacity can be obtained through integration of the stress-distribution.
For the integration, the stress-strain relation is curve-fitted by the modified
Ramberg-Osgood equation:
f
f
= + 0.002
Fy
E
Paper 1 - 3
(3)
5/6/2003
(4)
n=
F
u
log F
However, the statistical data of the ultimate strain is not available. Instead, the minimum
percent elongation is only available in the Aluminum Standards and Data (2000c),
abbreviated to ASD. Since the percent elongation is somewhat larger than the ultimate
strain, in general, the ultimate strain may be assumed from the percent elongation.
As Eberwien et al. (2001) showed, a moment capacity for a rectangular block
with non-linear stress distribution expressed by the modified Ramberg-Osgood equation
can be obtained using the Bernoulli-hypothesis. Based on the idea, the ultimate moment
capacity is computed and normalized by the yield moment capacity, resulting in a closed
form solution of the ultimate shape factor for a web element of a doubly symmetric
section:
2n
n
2
bh 2 Fu 1 Fu
1 n Fu
1 n + 1 Fu Fu
(5)
Mu =
+
2 u2 3 E 5002 2n + 1 Fy
500 n + 2 Fy E
2
bh Fy
My =
(6)
6
2n
n
2
Mu
3 Fu 1 Fu
1 n Fu
1 n + 1 Fu Fu
(7)
w =
= +
500 n + 2 Fy E
M y u2 Fy 3 E 5002 2n + 1 Fy
As seen in DOD (1994) and tensile coupon tests in this study, the typical value of
the ultimate strain (u) is approximately 6 to 8% for 6061-T6. For the same material, the
typical value of percent elongation is 12% as tabulated in the Aluminum Design Manual,
ADM in short, (2000b). Similarly, since the minimum value of the percent elongation is
8% for 6061-T6 as shown in ASD (2000c), the minimum value of the ultimate strain may
be approximately 4 to 6%. To understand the effect of the ultimate strain variation, shape
factors are computed for various ultimate strain values using Equation (7) in Figure 2.
From the figure, it is found that the shape factor variation is less than 1% if the ultimate
strain is set to either 50% larger or smaller than the minimum percent elongation.
Fig. 2. Variation of the shape factor for a rectangular web element with respect to the
ultimate strain (6061-T6)
Paper 1 - 4
5/6/2003
Thus, the conjecture that the ultimate strain may be assumed from the percent
elongation is reasonable. In this study, the recommended value of the ultimate strain is
between the half and the same amount of the minimum percent elongation. The ultimate
shape factors are computed using Equation (7) for some 6000 series alloys in Table 3,
which are more frequently used for extrusions.
The obtained shape factors in Table 3 can be directly used as the proposed cut-off
for the limit state of the ultimate stress when those are multiplied by the yield stress.
Since the compressive stress distribution is approximately the same as the tension one for
doubly symmetric compact sections, the shape factors (w) in this study are also
compared to those for the tension web available in AA: Equation (8). It can be concluded
that the shape factor in AA (2000a) is quite close to the rigorous one proposed in this
study for 6000 series alloys.
Table 3. Material properties and ultimate shape factors for rectangular sections (webs) for
some 6000 series alloys (FIX DATA WITH 10000KSI E)
Alloy-temper
Fy a
(MPa)
Ea
(MPa)
Fu a
(MPa)
u b
nc
w d
6005-T5
241.15
69589
261.82
0.04
35.23
1.5976
1.0363
1.42 Ftu Fy
241.15
69589
261.82
0.04
35.23
1.5976
1.0363
6063-T5
110.24
69589
151.58
0.04
9.23
1.9468
0.9971
6063-T5
103.35
69589
144.69
0.04
8.74
1.9768
0.9944
6063-T6, T62
172.25
69589
206.70
0.04
16.01
1.7362
1.0189
310.05
69589
344.50
0.04
27.18
1.6227
1.0284
6070-T6, T62
310.05
69589
330.72
0.03
39.29
1.562
1.0313
6105-T5
241.15
69589
261.82
0.04
35.23
1.5976
1.0363
6351-T5
241.15
69589
261.82
0.04
35.23
1.5976
1.0363
6463-T6
172.25
69589
206.70
0.04
16.01
1.7362
1.0189
a.
b.
c.
d.
e.
f.
g.
Paper 1 - 5
5/6/2003
from specimens and product types of specimens (plates, pipes or extruded shapes).
Among these data, 986 alloy-temper combinations are chosen for the investigation. The
bases to choose the data are as follows: First, minimum material properties required for
Equation (7) should be available. Second, the ultimate strain should exceed 1.5%, when it
is assumed to be the half of the minimum percent elongation. Third, the orientation of the
test coupon should be longitudinal or not specified.
5
Equation (7)
4.5
Equation (8)
w = 1.42
4
3.5
Ftu
Fty
(current AA)
Equation (9)
2.5
w = 1.25
Ftu
+ 0.2
Fty
(proposed)
1.5
1
0.5
1.5
2.5
u > 1.5%
3.5
Ftu Fty
Fig. 3. Comparison of the ultimate shape factor approximations using Equations (8) and (9)
for a plate under bending
For the selected alloy-temper combinations, the shape factors are computed using
Equation (7) and plotted with respect to the Ftu/Fty ratio in Figure 3. From the figure, it is
clear that Equation (8) becomes unconservative as the Ftu/Fty ratio increases. Thus, a
more precise curve-fit is proposed for the ultimate shape factor:
F
u = w = 1.25 tu + 0.2
(9)
Fty
For practical design purposes, Equation (9) could be used instead of Equation (7)
for the ultimate shape factor of a rectangular web element under bending.
5/6/2003
models. The alloy and temper is assumed to be extruded 6061-T6 with minimum material
properties in Table 3 except for the ultimate strain. To observe the variation of the
analysis results, two ultimate strain values are used; 4% and 8%. According to the
variation of the ultimate strain, the Ramberg-Osgood exponent (n) and the shape factor
for the web (w) vary.
For material model, isotropic hardening is used, since this study deals with
monotonic loading. In this case, once a stress reaches the ultimate stress, the stress
remains constant as the plastic strain exceeds the ultimate strain. Due to uncertainty after
this stage, it is assumed that the whole member reaches the failure when the von-Mises
stress at a single point of a member reaches the ultimate stress. This occurs when the
member is too compact to buckle. On the other hand, the failure of the member can also
be initiated by buckling when the member is less compact. In this case, the peak load of
the member is obtained before any point of the member reaches the ultimate stress. In this
study, these two possibilities of failure are considered simultaneously to find an ultimate
load factor.
Fig. 5. Parametric study results for component elements (a) web (b) flange
Paper 1 - 7
5/6/2003
D
B Fu
Flange
Ff =Fu
t
D
ultimate stress
(proposed AA)
B w Fy
Web
Fw = w Fy
D
Note. 1. F = Ff or Fw
2. B,D, k1 and k2 for a flange differ from those for a web. See AA (2000a) for details.
3. Equation (1) for (Table 2)
4. Equation (7) for w
Paper 1 - 8
5/6/2003
As an alternative in AA, limit state stresses obtained from all component elements
can be averaged according to contributory area. The averaged stress is multiplied by the
section modulus to compute the moment capacity. This is called weighted average stress
approach (denoted WASA).
The weighted average stress equation was first introduced by Jombock and Clark
(1968) to compute the crippling strength of aluminum trapezoidal formed sheet members:
Equation (10),
F A +F 1 A
Fwt = f f 1 w 6 w
(10)
Af + 6 Aw
where Ff = limit state stress for the flange, Fw = limit state stress for the web, Af = entire
compression side flange area, and Aw = entire web area.
Although the WASA was verified through experiments by the researchers, a
theoretical basis of the weighted average method has never been investigated. Thus, the
accuracy of the method is questionable for various geometric shapes other than the
formed sheet members.
The theoretical basis of the WASA is investigated for a doubly symmetric Isection shown in Figure 6. Multiplication of both the denominator and numerator of
Equation (10) by (hc/2)2 and simplification assuming that the flange thickness is
relatively small result in Equation (11):
M + M w hc M u hc
Fwt = f
(11)
=
I f + Iw 2
I 2
where Mf = moment capacity of the flanges, Mw = moment capacity of the web, If =
moment of inertia of the flange elements, Iw = moment of inertia of the web element, Mu
= total moment capacity and I = total moment of inertia.
Paper 1 - 9
5/6/2003
elements, falling into the yielding or, at least, inelastic buckling ranges. Thus, the
modifications shown in Table 5 as well as Table 4 are significant for extruded sections.
Table 5. Correction in the current WASA
current WASA
proposed WASA
(WASA)
(WASA2)
h
Mu = (Fwt)(S)
hc
Note. S = section modulus = I /(h/2)
Mu = (Fwt)(S)
5/6/2003
Ultimate stress, while MAA-Y-WASA is the moment capacity considering the limit state of
the Yield stress. The horizontal axis is the slenderness factor. The vertical axis is the
moment capacity obtained from the finite element analysis normalized by the moment
capacities using the approaches specified in the graph.
As seen in this figure, an approximately 7% improvement in the member capacity
is observed by the consideration of the ultimate-plastic capacity. In addition, the variation
of the data significantly decreases when the ultimate-plastic capacity is considered.
/M
approaches
1.3
1.2
1.1
FEM
M FEM / M approaches
M FEM / M A A -Y -WA SA
mean
c.o.v
0.9
1.209
0.056
0.8
M FEM / M A A -U-WA SA
1.141
0.029
0.3
0.4
0.5
0.6
0.7
0.8
= Fy Fcr
Fig. 8. Influence of the ultimate-plastic capacity consideration in AA (u = 4%)
FEM
/M
approaches
1.3
1.2
1.1
1
mean
c.o.v
0.9
M FEM / M approaches
M FEM / M A A -U-WA SA
1.141
0.029
0.8
M FEM / M A A -U-WA SA 2
1.04
0.024
0.3
0.4
0.5
0.6
0.7
0.8
= Fy Fcr
Fig. 9. Influence of the modification in WASA (u = 4%)
5/6/2003
employed. The moment capacities obtained from all four approaches as well as the finite
element analyses are listed in Table 6.
Table 6. Parametric study results
Fy
MAA-Y-MMCA MAA-Y-WASA MAA-U-WASA2 MAA-U-TMCA MFEM-4% MFEM-8%
=
My
My
My
My
My
My
Fcr
tw (mm)
tf (mm)
25.400
42.342
1.000
1.023
1.309
1.305
1.375
1.378
0.233
25.400
31.750
1.000
1.031
1.277
1.279
1.347
1.340
0.289
25.400
21.158
1.000
1.047
1.223
1.229
1.290
1.280
0.384
25.400
15.875
0.971
1.037
1.157
1.163
1.240
1.230
0.473
25.400
12.700
0.892
0.990
1.105
1.111
1.200
1.190
0.566
12.700
42.342
1.000
1.012
1.289
1.281
1.314
1.305
0.259
12.700
31.750
1.000
1.017
1.251
1.249
1.280
1.270
0.335
12.700
21.158
1.000
1.025
1.171
1.172
1.210
1.200
0.469
12.700
15.875
0.954
0.992
1.085
1.087
1.150
1.140
0.581
12.700
12.700
0.871
0.929
1.013
1.014
1.090
1.090
0.679
8.458
42.342
1.000
1.008
1.282
1.272
1.290
1.275
0.268
8.458
31.750
1.000
1.011
1.241
1.238
1.256
1.240
0.351
8.458
21.158
1.000
1.017
1.151
1.151
1.180
1.170
0.506
8.458
15.875
0.948
0.975
1.056
1.057
1.100
1.100
0.644
8.458
12.700
0.863
0.905
0.975
0.975
1.020
1.030
0.767
6.350
42.342
1.000
1.006
1.276
1.267
1.276
1.260
0.349
6.350
31.750
1.000
1.009
1.233
1.230
1.240
1.220
0.358
6.350
21.158
1.000
1.013
1.136
1.136
1.170
1.150
0.524
6.350
15.875
0.945
0.966
1.034
1.035
1.080
1.080
0.677
6.350
12.700
0.860
0.892
0.946
0.947
0.998
1.000
0.817
1.5
M TEST / M approaches
1.4
FEM
/M
approaches
1.3
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.3
M FEM / M approaches
mean
c.o.v
M FEM / M A A -Y -MMCA
M FEM / M A A -Y -WA SA
1.248
0.055
1.209
0.056
M FEM / M A A -U-WA SA 2
M FEM / M A A -U-TMCA
1.04
0.024
1.04
0.021
0.4
0.5
0.6
0.7
0.8
= Fy Fcr
Fig. 10. Comparison between the current and proposed approaches (u = 4%)
All the finite element computations in Figures 8 to 10 are made when the ultimate
strain is 4%. The percent symbol in the subscript of FEM analyses in Table 6, 4% or 8%,
Paper 1 - 12
5/6/2003
represents the ultimate strain value used. When the ultimate strain is 8%, the results do
not change significantly as compared in Table 6.
Fig. 11. (a) Dimensions of section -3x1.64 (b) schematic test setup side-view
All dimensions are in mm and not to scale
Paper 1 - 13
5/6/2003
Fig. 13. Deformed shape near failure using finite element method simulation (SOLID)
1.4
1.2
LF = Mu /My
0.8
0.6
0.4
0.2
50
100
SCVD (Span Center Vertical Displacement, mm)
150
A deformed shape near the failure from one of the finite element simulations
exampled in Figure 13 is similar to the ones from the physical tests shown in Figure 12.
The load factor-displacement curves obtained from both physical tests and the finite
element simulations show close agreements each other as plotted in Figure 14. The
variation in the test results would be mainly due to the variations of the material
properties as well as initial geometric imperfections.
Paper 1 - 14
5/6/2003
The average of the maximum load factors obtained from the physical tests is
compared to the current AA approaches (MMCA and WASA) and those developed in
this study (WASA2 and TMCA) in a dashed oval of Figure 10. The test results follow the
trend of the parametric study, which supports the validation of the proposed approaches
in this study.
approaches
current AA
Specification
Fa = Fay
Procedure I
proposed
approaches
(14)
Procedure II
Paper 1 - 15
(15)
5/6/2003
Table 8. Details of allowable stress equations for (a) tension element (b) compression
element and (c) shape factors
(a)
AA Section
allowable stresses
Fay = y Fty n y
3.4.2
3.4.4
allowable stresses
AA
b t S1
Section
Fay =
3.4.15
3.4.16
3.4.18
Fau =
Fau = u Fty
(b)
allowable stresses
S1 b t S2
limit
S1
y Fcy
B y Fcy
ny
D
n
B n u Fcy
u Fcy
nu
( kt nu )
Fay =
1
ny
B D
Fau =
1
ny
B D
t
k1 B
k BE
Fay = Fau = 2
b
D
n y
t
(c)
AA Section
3.4.2
y = 1.0
u = Ftu Fty
3.4.4
y = 1.3
3.4.15, 3.4.16
y = 1.0
u = Ftu Fcy
3.4.18
y = 1.3
b
b
2.2
k t = 1.00
k t = 1.10
k t = 1.25
2
1.8
2
1.8
~
ny = 1.65
1.6
1.6
n y
~
ny = 1.32
1.4
1.4
1.2
1.2
~
ny = 1.00
1
0.8
0.6
~
n
y
1
1.5
mean stdev
min
max
1.54
0.129
1.32
2.01
2.5
Ftu Fty
3.5
(a)
1
0.8
0.6
200
400
numbers of alloy-temper
(b)
Fig. 15. Safety factor of the tensile allowable stress (AA 3.4.4) for a plate under bending
Paper 1 - 16
5/6/2003
In Figure 15a, the safety factors against yielding of a tension side rectangular web
element under bending are computed and plotted using solid circles for the previously
selected 986 alloy-temper combinations when Procedure II is used (AA Section 3.4.4). It
is found that the average safety factor (1.54) is only 6.7% smaller than the current fixed
safety factor (1.65). Solid and dashed curves represent the analytic expressions of the
safety factors for the materials with different notch sensitivities (kt). It is also noted that
the majority of the alloy-temper combinations are concentrated near the average as shown
in Figure 15b. In addition, the minimum safety factor is set to 1.32.
Conclusions
The cut-off based on the yield stress in the AA Specification is replaced with the
ultimate shape factor developed in this study, which is eventually simplified for practical
design purposes. With the shape factor, the compressive limit state stress of a component
element can be evaluated more precisely. In addition, the modification is more consistent
with the specification approach for tension elements.
Secondly, the empirically developed weighted average stress approach (WASA)
is investigated. From the study, it is found that the weighted average stress equation is an
approximation of a linearized ultimate bending stress at the centroid of the flange, while
the current specification equation is considered to be the bending stress at the extreme
fiber; a simple modification is proposed to the WASA.
As a result of the study, moment capacities of laterally supported flexural
members are evaluated more accurately, when a series of parametric study is conducted
for a wide range of slenderness using the finite element method. The study is also further
validated by the flexural tests of standard sections.
In addition, two procedures are suggested for the final form of the future
specification. In these approaches, a certain factor of safety against yielding is
maintained. Procedure II is more reasonable, since the amount of additional inelastic
reserve capacity used in the approach depends on the margin between the ultimate and
yield stresses. In the proposed approaches, the current frame of the specification is
maintained.
Acknowledgement
This study is sponsored by the U.S. Department of Energy and the Aluminum
Association. Their sponsorship is gratefully acknowledged.
References
The Aluminum Association (2000a). The Specification for Aluminum Structures.
The Aluminum Association (2000b). The Aluminum Design Manual.
The Aluminum Association (2000c). Aluminum Standards and Data.
Aluminum Company of America (Alcoa) (1955). Alcoa Structural Handbook, A Design
Manual for Aluminum.
Clark, J.W., Rolf, R.L. (1966). Buckling of Aluminum Columns, Plates, and Beams.
Journal of the Structural Division, ASCE, Vol. 92, Proc. Paper 4838.
Eberwien, U., Valtinat, G. (2001). The fullness method: A direct procedure for
calculation of the bending moment of a symmetrical aluminum cross section. The
Paper 1 - 17
5/6/2003
Nomenclature
f = the ultimate shape factor for flange
u = the ultimate shape factor
w = the ultimate shape factor for web
y = the yield shape factor
w8% = the ultimate shape factor for web when the ultimate strain is 8%
= variable strain
u = strain at the ultimate stress = the ultimate strain
= constant for equivalent slenderness ratio depending on the plate buckling coefficient
= slenderness factor = (Fy/Fcr)0.5
p = equivalent slenderness ratio
= Poissons ratio
Af = entire compression side flange area
Aw = entire web area
B = buckling formula constant intersecting vertical axis for zero width-to-thickness ratio
b = the width of a plate element or a flange element
D = buckling formula constant for slope of the inelastic buckling range
E = Youngs modulus
f = variable stress
F = limit state stress for the flange or web
Fa = the (final) allowable stress
Fau = the allowable stress based on the ultimate limit state
Fay = the allowable stress based on the yield limit state
Fcr = minimum buckling stress
Ff = limit state stress for the flange
Fp = limit state stress of a component plate element
Ftu = tensile ultimate stress
Fty = tensile yield stress
Fu = the ultimate stress
Fw = limit state stress for the web
Fy = the yield stress
h = depth of the web element or the entire depth of an I-shaped section
hc = depth of an I-shaped section between centerlines of flanges
ho = depth of an I-shaped section between inner surfaces of flanges
I = total moment of inertia
Paper 1 - 18
5/6/2003
Paper 1 - 19