2001 Aristizabal - Semirigid Axial Loads
2001 Aristizabal - Semirigid Axial Loads
2001 Aristizabal - Semirigid Axial Loads
ABSTRACT: The stiffness and geometric stiffness matrices and load vector of a prismatic beam-column with
semirigid connections under any distributed axial load are presented. The derived matrices can be used in the
stability and second-order analyses of 2D and 3D framed structures with sidesway uninhibited, partially inhibited,
or totally inhibited in any type of construction. The effects of diagonal truss bracing and connection flexural
stiffness on the lateral stability of frames are discussed. Four examples are included that show the simplicity
and effectiveness of the proposed method in the analysis of 2D and 3D framed structures with semirigid con-
nections subjected to concentrated and distributed axial loads. It is shown that the stability and second-order
Downloaded from ascelibrary.org by Universidade Da Coruna on 02/12/14. Copyright ASCE. For personal use only; all rights reserved.
behavior of framed structures depend on (1) geometry and column layout; (2) size and type of bracing against
sidesway; (3) type and degree of restraints at the column supports; (4) type of connection (i.e., degree of restraint)
and the relative flexural stiffness (EI/L) between the beams and columns; and (5) magnitude and distribution of
the applied loads. The proposed stiffness and load matrices are presented in a practical form so that they can
be readily applied in the computer analysis of 2D and 3D framed structures.
STRUCTURAL MODEL
Assumptions
Consider the 2D beam-column element that connects points
A and B, as shown in Fig. 1(a). Element AB is made up of the
beam itself A⬘B⬘ lumped flexural connectors AA⬘ and BB⬘ at
A and B, respectively. The total number of degrees of freedom
(DOF) of the 2D beam-column element is six, as shown in
Fig. 1(b) taken at the ends A and B. It is assumed that (1)
1
125-Year Generation Prof., School of Mines, Nat. Univ., A. A. 75267,
Medellı́n, Colombia.
Note. Associate Editor: John Baugh. Discussion open until April 1,
2002. To extend the closing date one month, a written request must be
filed with the ASCE Manager of Journals. The manuscript for this paper
was submitted for review and possible publication on August 18, 2000;
revised February 27, 2001. This paper is part of the Journal of Structural
Engineering, Vol. 127, No. 11, November, 2001. 䉷ASCE, ISSN 0733- FIG. 1. 2D Beam-Column Element with Semirigid Connections: (a)
9445/01/0011-1306–1315/$8.00 ⫹ $.50 per page. Paper No. 22523. Mechanical Model; (b) DOF; (c) Definition of Fixity Factor at A a
冕
ing). The values of a and b must be determined empirically. L
Galambos and Ellingwood (1986) showed some empirical val- k ij⬘ = P(x)⌽ i⬘ ⌽ ⬘j dx (5)
ues for various steel shapes. Gerstle (1988) has indicated lower 0
and upper bounds for a and b. Xu and Grierson (1993) used
these bounds in the design of frames with semirigid connec- where P(x) = any distributed axial load. Eq. (5) indicates that
Downloaded from ascelibrary.org by Universidade Da Coruna on 02/12/14. Copyright ASCE. For personal use only; all rights reserved.
k11 symmetric
k21 k22
k11 ⫹ k21 k22 ⫹ k21 k11 ⫹ k22 ⫹ 2k21
L L L2
[K ] = (3)
k11 ⫹ k21 k22 ⫹ k21 k11 ⫹ k22 ⫹ 2k21 k11 ⫹ k22 ⫹ 2k21
⫺ ⫺ ⫺
L L L2 L2
0 0 0 0 k55
0 0 0 0 k65 k66
冕
k55 = k66 = ⫺k65 = (4) ⬘ ⫹ k ⬘21
k 11 k ⬘22 ⫹ k ⬘21
⫺ ⫺ k ⬘43 k ⬘44
L
1 L L
dx
0 EA
0 0 0 0 0
For members with flexible axial connections, the net axial 0 0 0 0 0 0
stiffness keff can be estimated as that of two springs in series;
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 2001 / 1307
clamped at both ends (i.e., with a = b = 1), which can be diagonal with the smallest AE/Ld might control buckling be-
obtained from tables available in most manuals (AISC 1989, cause the frame will sway in the direction that makes such a
1999). The axial and shear end forces can be obtained applying diagonal act in tension.
equilibrium conditions on the beam-column.
The following section shows the application of the stiffness Buckling Criterion
and geometric stiffness matrices given by (2)–(7) and (20) in
the stability and second-order analyses of 2D and 3D framed The overall stiffness matrix [K ] and the geometric stiffness
structures. matrix [K ⬘] of the frame shown in Fig. 2 were obtained from
(2)–(7). They are given by (9) and (10) as follows:
EI1 1 EIg a
12 ⫹ 12 symmetric
h1 4 ⫺ b L 4 ⫺ a c
EIg a c EI2 1 EIg c
[K ] = 6 12 ⫹ 12 (9)
L 4 ⫺ a c h2 4 ⫺ d L 4 ⫺ a c
EI1 2 ⫹ b EI2 2 ⫹ d EI1 1 ⫹ 2b EI2 1 ⫹ 2d
6 6 12 ⫹ 12 3 ⫹ S⌬
h 21 4 ⫺ b h 22 4 ⫺ d h 31 4 ⫺ b h 2 4 ⫺ d
h1 16 ⫺ 14b ⫹ 4 2b
symmetric
5 (4 ⫺ b)2
h2 16 ⫺ 14d ⫹ 4 d2
[K ⬘] = P 0 ␣ (10)
5 (4 ⫺ d)2
1 32 ⫺ 40b ⫹ 17 2b ␣ 32 ⫺ 40d ⫹ 17 2d 2 48 ⫺ 33b ⫹ 12 2b 2␣ 48 ⫺ 33d ⫹ 12 2d
⫹
10 (4 ⫺ b)2 10 (4 ⫺ d)2 5h1 (4 ⫺ b) 2
5h2 (4 ⫺ d)2
Example 1
Using the AISC LRFD approach (1999), determine the col-
umn sizes of the frame loaded as shown in Fig. 3. Assume
that (1) the columns have continuous lateral support perpen-
dicular to the plane of the structure; (2) b = 0.20 at the base
of column AB; (3) column CD is connected at the base with
a fixity factor d = 0.90; and (4) the 20-ft (6.096-m) girder AC
consists of a W16⫻31 shape connected to columns AB and
CD both with fixity factors a = c = 0.20. Use Fy = 36-ksi
(248.2-MPa) steel.
Downloaded from ascelibrary.org by Universidade Da Coruna on 02/12/14. Copyright ASCE. For personal use only; all rights reserved.
Solution:
FIG. 3. Example 1: Design of Single-Story Frame without Lateral
1. Try W14⫻120 for both columns: A = 35.3 in.2 (22,774
Bracing
mm2), Ix = 1,380 in.4 (5.744 ⫻ 108 mm4), and rx = 6.24
in. (158 mm).
2. Calculate Pu /A: (Pu /A)AB = 700/35.3 = 19.83 ksi (136.7 Pcr = 1,580.9 kip (7,032 kN) and
MPa) and (Pu /A)CD = 500/35.3 = 14.16 ksi (97.6 MPa). ␣Pcr = (5/7) ⫻ 1,580.9 kip = 1,129.2 kip (5,023 kN)
3. Determine Pcr from the characteristic equation 兩[K ] ⫺
P[K ⬘]兩 = 0 and from (9) and (10): Taking into account Therefore
冑
that ␣ = P2 /P1 = 5/7, h1 = 12 ft (3.658 m), h2 = 18 ft
(5.586 m), L = 20 ft (6.096 m), and Ig = 375 in.4 (1.561 2 ⫻ 29,000 ⫻ 1,240
KAB = = 3.29
⫻ 108 mm4), with these values a critical load of Pcr = 1,580.9 ⫻ (12 ⫻ 12)2
2,015.1 kip (8,964 kN) is obtained. Then
␣Pcr = (5/7) ⫻ 2,015.1 kip = 1,439.4 kip (6,403 kN) KCD = 冑 2 ⫻ 29,000 ⫻ 882
1,129.2 ⫻ (18 ⫻ 12)2
= 2.189
Therefore
4. Calculate c
KAB = 冑 2EI
Pcr h2
= 冑 2 ⫻ 29,000 ⫻ 1,380
2,015.1 ⫻ (12 ⫻ 12)2
= 3.075
(c)AB = 冉 冑冊
Kh
r
Fy
E AB
= 0.854 < 1.5
(c)CD = 冉 冑冊
Kh Fy
= 0.877 < 1.5
r E
4. Calculate c CD
(c)AB = 冉 冑冊
Kh
r
Fy
E AB
= 0.7942 < 1.5
therefore, use Fcr inelastic.
5. Calculate c Fcr :
2
(c Fcr)AB = 0.85 ⫻ (0.658) cFy = 0.85 ⫻ 0.759 ⫻ 36
therefore, use Fcr inelastic
冉 冑冊
= 22.56 ksi (155.5 MPa) > (Pu /A) = 21.88 ksi (150.8 MPa)
Kh Fy
(c)CD = = 0.9396 < 1.5 2
r E CD
(c Fcr)CD = 0.85 ⫻ (0.658) cFy = 0.85 ⫻ 0.725 ⫻ 36
therefore, use Fcr inelastic. = 22.18 ksi (152.9 MPa) > (Pu /A) = 20.75 ksi (143 MPa)
FIG. 4. Example 2: Parametric Study on Plane Frame: (a) Structural Model; (b) Effects of Diagonal Bracing S⌬ ; (c) Effects of Column AB Fixity
Factor at B b ; (d) Effects of Girder AC Fixity Factors a and c
6.22 in. (158 mm)] for column AB and W14⫻82 [A = 24.1 are apparent [Fig. 4(b)], particularly from S⌬ = 0 to 46
in.2 (15,548 mm2), Ix = 882 in.4 (3.6712 ⫻ 108 mm4), and rx kip/in. (80.55 kN/m). With merely 0.93 in.2 (604.5 mm2)
= 6.05 in. (154 mm)] for column CD with d = 0.90. of diagonal bracing (2.91% of the cross area of column
AB) or S⌬ = 46 kip/in. (80.55 kN/m), the K-factor is re-
Solution: Each of the three effects was studied independently duced in half in both columns. Both curves in Fig. 4(b)
as follows: are steep up to 46 kip/in. (80.55 kN/m); additional incre-
ments in S⌬ do not show significant reductions in the K-
• Effects of diagonal bracing—The lateral stiffness restrain- factors.
ing the frame against sidesway S⌬ provided by a diagonal • Effects of column AB base fixity b —The fixity of the
bracing of horizontal length L, height h, and cross area column bases also has significant benefits on the stability
Ab is given by Salmon and Johnston (1996, Chapter 14, of the frame. Fig. 4(c) indicates that, by increasing the
Eq. 14.5.16) as follows: fixity of column AB from zero (perfectly hinged) to 0.20,
the K-factor is reduced by 25% in each column, repre-
Ab EL2/h3 senting a 78% increase in the critical load of this partic-
S⌬ =
[1 ⫹ (L/h)2]3/2 ular frame. Notice that both curves in Fig. 4(c) are steep
from b = 0 up to approximately b = 0.60. This indicates
A sensitivity study was carried out on the effects of S⌬ on that any additional fixity of the base of column AB beyond
the stability of the frame of Example 1 with the fixity 0.60 shows little additional effectiveness.
factors indicated in Fig. 3. The results are shown in Fig. • Effects of girder end fixities a and c —The fixity of the
4(b). For instance, S⌬ = 100 kip/in. (175.12 kN/m) cor- girder connections has some benefits on the stability of
responds to an effective diagonal brace with cross area = the frame but they are not as significant as those at the
2.015 in.2 (1,300 mm2) that connects joints A and D. The column bases [solid curves in Fig. 4(d)]. Notice that the
benefits of diagonal bracing on the stability of this frame variations of the effective length K-factors with the di-
1310 / JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 2001
Example 3
Using (2), (3), and (20), determine the elastic buckling load
of a prismatic bar under a distributed axial load, as shown by FIG. 5. Problem 3: Simply Supported Beam-Column under Complex
Axial Load: (a) Structural Model and Loading; (b) Model with Two DOFs
Fig. 5(a). The bar is simply supported and the load distribution
(Rotation at A and Lateral Sway at B)
represents approximately the variation in the compressive
stress in the top-chord of a bridge truss, as shown by Timo-
shenko and Gere (1961, p. 107). (2), (3), and (20) any bifurcation buckling problem of beam-
Solution: In this case, the deflection curve of the buckled bar columns and framed structures subject to any concentrated or
has only one half-wave and is symmetrical with respect to the distributed axial loads applied along the members can be an-
middle (Timoshenko and Gere 1961, p. 110). Therefore, only alyzed using the matrix approach proposed herein.
half of the bar is analyzed with two DOF (the rotation at A
and the lateral sway at B), as shown by Fig. 5(b) with a = b Example 4: Stability and Second-Order Analysis of
= 1. 3D Multicolumn System
In this particular case
冋 册 冋 册
system). The frame consists of n prismatic columns, with the
2EI 2h2 ⫺3h P h2 ⫺3h centroid of column i located at (Xi , Yi) with respect to the X,Y-
[K ] = ; [K ⬘]=1 = ;
h3 ⫺3h 6 30 ⫺3h 18 plane and with individual properties including proper weight
冋 册
per unit length pi , moduli of elasticity Ei and Gi , height hi ,
P 4h2 ⫺15h cross area Ai , principal moments of inertia Ixi and Iyi , effective
[K ⬘]=2 = ⫺
420 ⫺15h 72 polar moment of inertia Jei , and two pairs of flexural restraints
with fixity factors (axi , bxi and ayi , byi) along the X- and Y-
Calling  = ph3/(840EI ), the critical load ( ph/2)cr can be axes, respectively. It is assumed that the centroidal axis of any
obtained from the characteristic equation 兩[K ] ⫺ P[K ⬘]兩 = 0, column Ai Bi is perfectly vertical and straight, with the external
as follows: axial load Pi applied along it. All n columns share the same
冏
2(1 ⫺ 50)h2
⫺3(1 ⫺ 9)h
⫺3(1 ⫺ 9)h
6(1 ⫺ 30)
=0 冏 lateral spring restraints S⌬X , S⌬Y , and SXY , and interstory side-
sways ⌬X , ⌬Y , and XY ; i.e., the top slab serves as a rigid di-
aphragm allowing only three DOF. Also all columns are as-
This determinant can be reduced to 3572 ⫺ 86 ⫹ 1 = 0 sumed doubly symmetrical with their principal cross-sectional
with the lowest root  = 0.012251. Therefore, ( ph/2)cr ⬵ axes parallel to the global X and Y axes (i.e., their shear center
20.5815EI/(2h)2, which is only 1.2% higher than the value and centroid coincide).
2.06 2EI/(2h)2 calculated by Timoshenko and Gere (1961, p. A typical column element Ai Bi in the 3D multicolumn frame
111). [Fig. 6(b)] is made up of the column itself A⬘B i ⬘i and flexural
This problem can also be solved using half the bar with connections Axi A⬘xi , Bxi B ⬘xi and Ayi A⬘yi , Byi B yi⬘ located at the top
only one DOF (the lateral sway at B) assuming a = 0 and b and bottom ends and along the X- and Y-axes, respectively.
= 1. In this particular case, [K ] = 3EI/h2, [K ⬘]=1 = 3ph/8, and The connections Axi A⬘xi and Ayi A⬘yi at the top end A have stiff-
[K ⬘]=2 = ⫺3ph/35. The critical load is obtained from the char- nesses axi and ayi along the X- and Y-axes, respectively. The
acteristic equation 3EI/h2 ⫺ 3ph/8 ⫹ 3ph/35 = 0 or ( ph/2)cr corresponding stiffness indices of column i and the top end Ai
⬵ (560/27)EI/(2h)2, which is just 2% higher than the value are Raxi = axi /(Ei Ixi /hi) and Rayi = ayi /(Ei Iyi /hi). Similarly, con-
calculated by Timoshenko and Gere (1961). nectors Bxi B xi⬘ and Byi B yi⬘ at the bottom have stiffness bxi and
By taking more column segments, and consequently more byi and stiffness indices Rbxi = bxi /(Ei Ixi /hi) and Rbyi = byi /
DOF along the member, the accuracy of the calculated buck- (Ei Iyi /hi). The fixity factors at the top and bottom ends of col-
ling load is improved even further. However, that refinement umn Ai Bi and along the X- and Y-axes are axi , bxi and ayi ,
would result in a larger determinant requiring a computer ei- byi [i.e., axi = 1/(1 ⫹ 3/Raxi), ayi = 1/(1 ⫹ 3/Rayi), bxi = 1/(1
genvalue program. It can be concluded that with the help of ⫹ 3/Rbxi), and byi = 1/(1 ⫹ 3/Rbyi)], respectively.
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 2001 / 1311
FIG. 6. Problem 4: Stability and Second-Order Analysis of 3D Multicolumn System: (a) 3D Structural Model; (b) 3D Single Beam-Column
Solution: Assuming relatively small lateral deflections and rotations, the overall static equilibrium of the top slab of the model
shown in Fig. 6(a) along the X- and Y-axes and about the Z-axis yields the following three coupled equations:
冋冉冘 冊 册 冉冘 冊
n n
冋冉冘 冊 册 冉冘 冊
n n
冉冘 冊 冉冘 冊 冉冘 冋 册 冊
n n n
Gi Jei
⫺ Yi SXi ⌬X ⫹ Xi SYi ⌬Y ⫹ X 2i SYi ⫹ Y 2i SXi ⫹ ⫹ Sxy XY = TXY (11c)
i=1 i=1 i=1 hi
where FX and FY = applied external forces along the X- and Y-axes, respectively; and TXY = applied torsional external moment
about the Z-axis
12Ei Ixi axi ⫹ bxi ⫹ axi bxi 2 Pi 40 ⫹ 8(2axi ⫹ 2bxi) ⫹ axi bxi (axi ⫹ bxi ⫹ 3axi bxi ⫺ 34)
SXi = ⫺
3
hi (4 ⫺ axi bxi) 5 hi (4 ⫺ axi bxi)2
12Ei Iyi ayi ⫹ byi ⫹ ayi byi 2 Pi 40 ⫹ 8(2ayi ⫹ 2byi) ⫹ ayi byi (ayi ⫹ byi ⫹ 3ayi byi ⫺ 34)
SYi = ⫺
3
hi (4 ⫺ ayi byi) 5 hi (4 ⫺ ayi byi)2
Notice that the terms SXi and SYi include the coefficients of the stiffness and geometric stiffness matrices corresponding to the
lateral deflection of the top end of each column (i.e., k33, [k ⬘33]=0 , and [k 33
⬘ ]=1). The last term corresponds to a uniformly distributed
axial load pi (such as the column self-weight). This term or any additional geometric stiffness term added to (12) is obtained
using (19).
1312 / JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 2001
冉冘 冊 冘
n n
SXi ⫹ S⌬ X 0 ⫺ Yi SXi
冋册冋册
i =1 i =1
冉冘 冊 冘
n n ⌬X FX
0 SYi ⫹ S⌬ Y Xi SYi ⌬Y = FY (13)
i =1 i =1
XY TXY
冘 冘 冘冋 册
n n n
Gi Jei
⫺ Yi SXi Xi SYi X 2i SYi ⫹ Y 2i SYi ⫹ ⫹ SXY
i =1 i =1 i =1 hi
calculator once the location (Xi , Yi), size (hi , Ai , Ixi , Iyi , and must be investigated first, then a comparison among the dif-
Jei), fixity factors (axi , bxi and ayi , byi), elastic moduli (Ei and ferent alternatives must be carried out before the structure is
Gi), axial loads Pi and pi on each column, applied loads (FX , finally designed and built.
FY , and TXY), and imposed restraints (S⌬ X , S⌬Y , and SXY) are The stability and second-order behavior of 3D framed struc-
given. Also, the overall bifurcation stability (i.e., buckling un- tures are very sensitive to the types symmetry of the applied
der a given load pattern) of the frame can be obtained, making loads, geometry, and column stiffnesses about the global X-
the determinant of the 3 ⫻ 3 matrix of (13) equal zero. Notice and Y-axes. Any lack of symmetry including asymmetry in the
that only three types of overall-story elastic response and axial load distribution among the columns reduces the total
buckling modes are considered by (13); pure-flexural sway, flexural critical load and introduces undesired torsion effects.
pure-torsional sway, and combined flexural-torsional sway of There is need for further research on real 3D framed structures
the entire floor. Using (13), Aristizábal-Ochoa (1997c) showed with semirigid connections. In the linear elastic range, the sta-
that (1) the stability analyses of 3D frames are related to the bility and second-order analyses of framed structures depend
types symmetry (i.e., symmetry of loading, of geometry, and on 16 parameters.
of column stiffnesses) about the X- and Y-axes; (2) the elastic
buckling loads of multistory multibay framed structures can
be determined with excellent accuracy; and (3) any lack of APPENDIX
symmetry (including any asymmetry in the loadings) intro- Consistent Stiffness Matrix Derivation for
duces torsion, causing negative effects on the stability re- Beam-Columns with Semirigid Connections
sponse of framed structures. Further details on stability anal-
ysis of framed structures with semirigid connections are given Using the consistent formulation of the FEM (Yang 1986),
by Aristizábal-Ochoa (1997a–c). the flexural coefficients of the consistent geometric stiffness
matrix k ⬘i j are given by (5). This equation indicates that the
SUMMARY AND CONCLUSIONS shape functions ⌽ and their first derivatives ⌽⬘ = (d⌽/dx)
corresponding to each of the four flexural deformations [(i.e.,
A matrix approach that determines the stability and second- DOF 1–4 in Fig. 1(b)] must be determined to evaluate each
order elastic response of beam-columns and framed structures coefficient in a consistent manner. A practical approach is pro-
with semirigid connections is presented. The stiffness matrix, posed that requires only the basic flexural stiffness coefficients
a series of consistent-geometric stiffness matrices, and the load given by (2) to determine any k ⬘i j consistently.
vector of a prismatic beam-column element are fully de-
scribed. Their use in the analysis and design of beam-columns Derivation of Deflected Shapes and First Derivatives
and frames subject to distributed axial loads along the mem-
bers or concentrated at their ends is illustrated. The proposed The conjugate beam is a fictitious beam that has the same
matrix formulation is general, versatile, and simple to imple- span length as the real beam but is supported in such a way
ment for computer usage. The effects of diagonal bracing and that, when it is loaded by the M/EI of the real beam, the shear
self-weight or any distributed axial load along the members and bending moment in the conjugate beam at any location
are also included. are equal to the slope ⌽⬘ and deflection ⌽ of the real beam at
Numerical results indicate that the stability of 2D frames the corresponding location, respectively. The conjugate-beam
increases substantially with the magnitude of the diagonal method furnishes a straightforward approach by which the
bracing and the fixity of the columns at their base supports. shape functions ⌽i and first derivatives ⌽⬘i of the real beam
The degree of fixity at the beam-column connections has less can be computed exactly, without solving complex differential
influence on the overall stability of frames if the girder’s flex- equations.
ural stiffness (EI/L) is small compared to that of the columns. To illustrate how any flexural shape function and its first
However, for frames with stiff girders (i.e., with large EI/L) derivative can be determined in closed form, consider the real
and no diagonal bracings, the flexural stiffness of the girder beam in Fig. 7(a), which is subject to a unit rotation at node
becomes more effective, particularly with fixity factors up to A. The resulting end forces and moments in Fig. 7(a) are also
0.60. The degrees of fixity at the column bases and at the shown. Also, the corresponding moment diagram and conju-
girder end connections show steep increases in the overall sta- gate beam are shown in Figs. 7(b and c) respectively. Note
bility of 2D frames when the fixity factors are increased from that the moment diagram M varies in a linear fashion, and,
zero ( = 0) to approximately = 0.60. This trend indicates because EI is constant for prismatic members, the load on the
that any additional fixity at these locations beyond 0.60 shows conjugate beam M/EI is also linear. Also, because the only
little additional effectiveness in the overall stability of a frame. deflection at the nodes of the real beam consist of a unit ro-
This is an indication that fully restrained connections (i.e., tation at A, the only reaction in the conjugate beam is a unit
‘‘totally rigid’’ connections) may not be the best choice from shear force at A, as shown in Fig. 7(c). Therefore, the shear
the point of view of economy and stability; probably semirigid force and moment in the conjugate beam can be obtained di-
connections with a fixity factor of about 0.5 might be more rectly using static equilibrium from Fig. 7(c), as follows:
JOURNAL OF STRUCTURAL ENGINEERING / NOVEMBER 2001 / 1313
Eq. (18), when placed into (5) and using simple integration,
again yields the coefficients of the consistent geometric stiff-
ness matrix. If the axial load is represented by an exponential
series P(x) = 兺 P(x/L), where is an exponent that can
assume any real value, the consistent geometric stiffness co-
efficient becomes k ij⬘ = 兺 (k ij⬘). The component corresponding
to in the series can be obtained directly by integration, re-
Downloaded from ascelibrary.org by Universidade Da Coruna on 02/12/14. Copyright ASCE. For personal use only; all rights reserved.
sulting in
(k ⬘)
ij =
P
L 冋 ai a j
⫹1
⫹
2(ai bj ⫹ aj bi)
⫹2
⫹
4bi bj ⫹ 3(ai cj ⫹ aj ci)
⫹3
册
FIG. 7. Conjugate Beam Model: (a) Unit Rotation at A and Corre-
sponding End Forces and Moments; (b) Moment Diagram; (c) Conjugate 6(bi cj ⫹ bj ci) 9ci cj
⫹ ⫹
Beam (Curvature Diagram) ⫹4 ⫹5 (19)
冕 冋 册
X For the particular case of = 0 (i.e., P = constant axial
1 z load along the member), the resulting coefficients (k ⬘)
⌽ ⬘(x)
1 =1⫹ (k11 ⫹ k21) ⫺ k11 dz (14a) ij =0 are
0 EI L given by (6) and (7). The application of (19) on the stability
冕 冋 册
X
for structural concrete and commentary.’’ ACI 318-99, ACI 318R-99,
1 z Detroit.
⌽ ⬘(x)
2 = (k22 ⫹ k21) ⫺ k22 dz (15a) American Institute of Steel Construction (AISC). (1989). Manual of steel
0 EI L construction, allowable stress design, 9th Ed., Chicago.
冕 冋 册
X construction: Nonparadoxical approach.’’ J. Struct. Engrg., ASCE,
1 z 120(4), 1272–1290.
⌽ ⬘(x)
3 = (k13 ⫹ k23) ⫺ k13 dz (16a) Aristizábal-Ochoa, J. D. (1994b). ‘‘Slenderness K-factor for leaning col-
0 EI L
冕 冋 册
umns.’’ J. Struct. Engrg., ASCE, 120(10), 2977–2991.
X
1 z Aristizábal-Ochoa, J. D. (1994c). ‘‘Stability of columns under uniform
⌽3 (x) = 1 ⫹ (k13 ⫹ k23) ⫺ k13 (x ⫺ z) dz (16b) axial load with semirigid connections.’’ J. Struct. Engrg., ASCE,
0 EI L 120(11), 3212–3222.
冕 冋 册
X Aristizábal-Ochoa, J. D. (1996). ‘‘Braced, partially braced, and unbraced
1 z columns: Complete set of classical stability equations.’’ Int. J. Struct.
⌽ ⬘(x)
4 = (k14 ⫹ k24) ⫺ k14 dz (17a)
0 EI L Engrg. and Mech., Taejon, Korea, 4(4), 365–381.
冕 冋 册
X
Aristizábal-Ochoa, J. D. (1997a). ‘‘First- and second-order stiffness ma-
1 z trices and load vector of beam-columns with semirigid connections.’’
⌽4(x) = (k14 ⫹ k24) ⫺ k14 (x ⫺ z) dz (17b) J. Struct. Engrg., ASCE, 123(5), 669–678.
0 EI L
Aristizábal-Ochoa, J. D. (1997b). ‘‘Story stability of braced, partially
Note that ⌽3(x) ⫹ ⌽4(x) = 1 and ⬘3 (x) ⫹ ⌽ ⬘4 (x) = 0. braced, and unbraced frames: Classical approach.’’ J. Struct. Engrg.,
ASCE, 123(6), 799–807.
Once the basic flexural stiffness coefficients k11, k21, and k22 Aristizábal-Ochoa, J. D. (1997c). ‘‘Amplification factor for three-dimen-
are calculated using (2), the rest of the flexural stiffness co- sional R/C framed structures: A non-paradoxical approach.’’ ACI Struct.
efficients can be determined from (3). Then using simple in- J., 94(5), 538–548.
tegration, the shape functions and the first derivatives can be Chen, W. F., and Lui, E. M. (1991). Stability design of steel frames, CRC,
determined from (14)–(17). Because the flexural shape func- Boca Raton, Fla.
tions ⌽i are cubic polynomials, they can be expressed as fol- Cunningham, R. (1990). ‘‘Some aspects of semi-rigid connections in
structural steel-work.’’ Struct. Engrg., 68(5), 85–92.
lows: Galambos, T. V., and Ellingwood, B. (1986). ‘‘Serviceability limit states:
冉冊 冉冊 冉冊
2 3 Deflection.’’ J. Struct. Engrg., ASCE, 112(1), 67–84.
x x x
⌽i (x) = ai ⫹ bi ⫹ ci ⫹ di (18a) Gerstle, K. H. (1988). ‘‘Effects of connections on frames.’’ Steel beam-
L L L to-column building connections, W. F. Chen, ed., Elsevier Science Pub-
lishers Ltd., New York, 241–267.
and the first derivatives are simply Monforton, G. R., and Wu, T. S. (1963). ‘‘Matrix analysis of semi-rigidly
冉冊 冉冊
2 connected steel frames.’’ J. Struct. Div., ASCE, 89(6), 13–42.
ai bi x ci x Salmon, C. G., and Johnson, J. E. (1996). ‘‘Frames—Braced and un-
⌽⬘(x)
i = ⫹2 ⫹3 (18b) braced.’’ Chapter 14, Steel structures: Design and behavior, 4th Ed.,
L L L L L
Harper & Collins College Publishers, New York, 917.
The coefficients ai , bi , ci , and di [with q = 3/(4 ⫺ a b)] are Timoshenko, S. P., and Gire, J. M. (1961). ‘‘Elastic Buckling of Bars and
as follows: Frames.’’ Theory of elastic stability, 2nd Ed., Engineering Societies
Monographs, McGraw-Hill, New York, 110.
Xu, L., and Grierson, D. E. (1993). ‘‘Computer-automated design of
• For ⌽1: a1 = a(4 ⫺ b)Lq/3, b1 = ⫺2a Lq, c1 = a(2 ⫹ semirigid steel frameworks.’’ J. Struct. Engrg., ASCE, 119(6), 1740–
b)Lq/3, and d1 = 0. 1760.