Beruflich Dokumente
Kultur Dokumente
ASHOK K. CHUGH*
Rock Mass Behavior Group,DenverMiningResearchCenter, US.Bureau ofMines,Denver, Colorado, U.S.A.
SUMMARY
A simple and direct procedure is presented for the formulation of an element stiffness matrix on element
co-ordinates for a beam member and a beam-column member including shear deflections. The resulting
stiffness matrices are compared with those obtained using the alternative formulation in terms of member
flexibilities:The relative effects of axial force and shear force on the stiffness coefficients are presented. The
critical buckling loads, considering the effects of shear force, are computed and compared with those
available in the literature. Only prismatic members are considered.
INTRODUCTION
In most civil engineering structures, the constituent members are either long and slender or
short and stocky. In the former type, the effect of axial force on the response of the member is
important, whereas in the latter type the effect of transverse shear force on the response of the
member is very important. Accordingly, most of the developments reported in the literature
account for the effect of one or the other, but not both, on the flexural behaviour of a member.
If a structure is composed of both slender and stocky members, then one has to appropriately
select the member load-displacement (stiffness or flexibility) characteristics. Since the slender or
stocky classification of a member is relative, it would be convenient to utilize a single
formulation of load-displacement response such that the specified dimensions and load mag-
nitudes determine the influence of the so-called secondary effects on the flexural behaviour of
the members and hence that of the structure. Furthermore, some engineering problems do not
lend themselves to the solution methods devised for typical civil engineering problems, and
hence more general and complete formulations are required. For example, a non-conventional
structural problem provided the motivation for the development presented herein, which
comprises a part of the U.S. Bureau of Mines investigation of the stability/collapse behaviour of
a naturally jointed rock mass that spans an underground opening, where the loading is
composed of both gravity and formation pressures.
In the displacement matrix method of structural analysis one constructs the element stiffness
matrix for each member in the structure and then synthesizes the element stiffness matrices to
generate the structural stiffness matrix. The structural displacements, the element displace-
ments, and hence the element forces can then be computed.
1681
1682 A. K. CHUGH
The stiffness matrix for a prismatic beam and a beam-column element can be derived in
several ways.’’2 However, the derivation of the stiffness matrix for a prismatic beam which
includes transverse shear deflection is not so straightforward. If shear deformations are not
permitted in simplebeam bending, then the requirement that plane sections remain plane can be
used and this introduces internal When this restraint is removed to allow for shear
deformation, there is an addition to the internal strain energy stored; to maintain the equality of
internal and external work, the external work must be increased by the same amount. One
possible approach to formulating the correct stiffness matrix is to form a flexibility matrix that
accounts for shear deformation and then to obtain the stiffness matrix through the flexibility-
stiffness tran~formation.~’~ In the technical literature there appears to be no formulation of a
stiffness matrix for a beam-column, prismatic or non-prismatic, with shear deflections.
The objectives of the present paper are to present
1. a simple method to construct the stiffness matrix of a beam and a beam-column element of
constant cross-section, with bending in one principal plane, including shear deflections;
2. an alternative formulation for the element stiffness matrix of a beam-column of constant
cross-section, with bending in one principal plane, including shear deflection, in terms of
member flexibilities; and
3. computation of critical buckling loads, including effects of transverse shear force, corres-
ponding to any end conditions.
Linear elasticity, small deformations, no warping, and no local cross-sectional buckling are
assumed. Figure 1showsthe general form of the member and the element co-ordinate system.
The simple method presented (objective 1.in the preceding paragraph) helps one to describe
the non-linear interaction among the axial, shear, and flexural effects intuitively.The alternative
formulation in terms of member flexibilities presents a rigorous mathematical development
which, in addition to being a contribution in itself, tends to confirm the validity of the simple
Y I v, v2
T
d
‘ P 4
Figure l(a). General layout of a beam element bending in one principal plane
method. The computation of critical buckling loads is one of the areas of application of the
element stiff neness matrix. These three items are described in the text in the order mentioned
above.
Translational stiffness
Without the shear deflection, the member forces for S translational displacement on element
co-ordinate 3 , Figure 2(a), are
1 2EIS
k33 = -k43 =-
L3
12EIS
61 =-
A,GL’
Total translational displacement on element co-ordinte 3, Figure 2(c), is
1 2EIS
A = S +S,=S+--S(I +a) (4)
A,GL~ -
where
From equations ( l ) ,(2),and (6),it follows that E 1 3 , Ez3,E33,and &43, the corresponding terms in
the element stiffness matrix [El, including shear deflections, are given by
E - E - -6EI 1
13- 2 3 - 2
L
-
l+a (7)
k-33 -
--k - 12EI 1
43- L 3 1+a
Similar expressions for member forces can be obtained for unit translational displacement on
element co-ordinate 4, as shown in Figure 2(b).
1684 A. K. CHUGH
k24
Rotational stiffness
Without the shear deflection, the member forces for 8 rotational displacement on element
co-ordinate 1 , Figure 3(a), are
The translational displacement caused by the transverse shear force k31= 6E18/L2,without
any change in the rotational displacement, if permitted, would be Figure 3(c),
.
STIFFNESS MATRIX FOR A BEAM ELEMENT 1685
The force required to cause 6' = 0, without causing any change in rotational displacement would
be, Figure 3(d),
k'31---k' - 12EI
41-3--=--
1 aL0 6EI a9
L l+a 2 L2 l+a
If, in Figure 3(e), gl1,k;', &I, and l 4 1 are the corresponding terms in the element stiffness
matrix [&I including shear deflections, then from equations (9)-(14)
-I
- -2-a- -6- 6
4+a
L L L2 L 2
2-cu -
- ( ~ -6- 6
4 +-
E' L L L2 L2
[El =-
-6 -6 12 -12
I L2 6 -
L2
- -
L3
-12 L3
l2 I
STIFFNESS MATRIX FOR A BEAM-COLUMN ELEMENT INCLUDING SHEAR
DEFLECTIONS
Translational stifness
Without shear deflection, the member forces for 8' translational displacement on element
co-ordinate 3, Figure 4(a), arelS2
-u2(1 -c)EI8'
B'3 = B23 = L2(2 - 2c -us)
u 3sEIS'
B33 =-B43 =
L3(2-2c-us)
1686 A. K. CHUGH
Y
A
Ignoring the presence of the axial force P,the transverse shear caused by B I 3and B23 is
2u2(1-c)EIs’
Bk3 = -B;3 =
L3(2-2c-us)
The additional displacement caused by the transverse shear force B53 = [2uz(1-c)EIs’]/
[ L’ ( ~- ~c- us)]is
2u2(l-C)EI8’
s; =
L2A,G(2-2c-u~)
Total translational displacement on element co-ordinate 3, Figure 4(c), is
A =s‘+s; = 6’ 1 +
[ 2u2(1-c)p
(2-2c-us) I
STIFFNESS MATRIX FOR A BEAM ELEMENT 1687
where = EI1L’A.G
.. (2- 2c - u s w
6’ =
[2- 2c - us + 2u2(1- C ) P ]
The axial force P acting through the translational displacement A’ causes the equilibratingshear
force of magnitude PA’IL, Figure 4(d). From equations (20), (22),(25) and the equilibrating
shear force with the total translational displacement A’ set equal to unity the corresponding
terms in the element stiffness matrix, [B],including shear deflections, Figure 4(e), are
- - -u’EI(l -c)
B 1 3= B23= L2[2-2c -us +2uZ(1- C ) P ]
2u2(1-c)EI P
--= u’EI[us -2u2(1 -c)P]
833=-B43= 3 (27)
L [2-2c -us +2u2(1- C ) P ] L L3[2-2c -us +2u2(1- C ) P ]
Similar expressions for member forces can be obtained for unit translational displacement on
element co-ordinate 4, as shown in Figure 4(b).
Rotational stifiess
Without the shear deflection, the member forces for 0‘ rotational displacement on element
co-ordinate 1, Figure 5(a), are
Ignoring the presence of the axial force P, the translational displacement caused by the
transverse shear force &I = [u2(l -c)E10’]/[L2(2-2c -us)], without any change in the rota-
tional displacement, if permitted, would be, Figure 5(c)
The axial force P acting through the translation1displacement 6’;causes an equilibrating shear
force of magnitude PS’;/L,Figure 5(d). The force required to restore 6’;= 0, without causing any
change in the rotational displacement, would be, Figure 5(e),
- -u2EI(1- c ) u2(i - c ) E w
B’l = B’l = B136’;=L2[2-2c -us +2u2(1-c)P] LASG(2-2c -us)
If Bll, Bzl, B31, and 841 are the corresponding terms in the element stiffness matrix, [B],
including shear deflections, then from equations (28)-(33) with 0‘set equal to unity, Figure 5(f),
one gets
-
- u(s -uc +PU2S)EI
(34)
L[2 - 2c -us +2u2(1- C ) P ]
uEI(u- S -Pu’s)
B21 -
- B21+BL=L[2-2c -us +2u2(1- c ) @ ]
(35)
STIFFNESS MATRIX FOR A BEAM ELEMENT 1689
=B31 +B;1=
-u2(1 -c)EI + [ u2EI{us-2u2P(1 - c ) } u2(1-c)EI
L3{2- 2~ -US + 2u2(1-c)P}
831 2
L (2 - 2~ -US) LASG(2-2c -US)
- -u2(1 -c)EI
-
L2[2- 2c -us + 2u2(1- c ) P ]
Similar expressions for member forces can be obtained for unit rotational displacement on
element co-ordinate 2, as shown in Figure 5(b).
1690 A. K. CHUGH
The element stiffness matrix for a beam-column element including shear deflection, for unit
displacements, can conveniently be expressed as
- uEI X
= L[2 - 2c -us +2u2(1 -C)P]
r
I (s - uc +uzs/3) (u -s - u 2sp)
-u(l-c)
L
-u(l -c)
u(1-c)
L
u(1-c)
(u - s - U Z S / 3 ) (s - uc +U 2 S P ) L L
-u(l -c) -u(l-c) u2[s-2u(1-c)p] -U2[S-2U(l-C)/3]
I L L LZ LZ
AN ALTERNATIVE FORMULATION
An alternative formulation for the stiffness matrix of a beam element considering shear
deformation is to construct a flexibility matrix that accounts for shear deformation and then
obtain the stiffness matrix. This approach is well documented''2 and is not repeated here. A
parallel approach for a beam-column element, with shear deflections, is to form the flexibility
matrix for a simply supported beam subjected to flexure plus shear, with only flexure accounting
for the axial load and to then use the flexibility-stiffness transf~rmation.~ This approach is
presented below:
The flexibility matrix, [fl,for a simply supported beam, Figure 6(a), (b), subjected to flexure
plus shear is
L(s-uc) 1 -L(U-s)
+-
(39)
-L(u-s) 1 L(s-uc) 1
+-
u'sEI LA,G u2sEI
+-LA,G
The transformation from a flexibilitymatrix to a stiffnessmatrix, without the axial force P,can be
accomplished by3
-
Pfl-??
I
L
P
T -
I
L P
Thus the transformation from a flexibility matrix to a stiffness matrix,.including an axial force P,
is accomplished by
c-usp=o (44)
Equation (2-57) in Reference 6 for the critical load of a cantilever post, including the effect of
transverse shear force, is P,, = Pe/[1 + (nP,/AG)].
General comments
Several problems were analyzed to study the relative effects of axial force and shear force on
the elements of the stiffness matrix in equation (38). Specifically, the parameter p was varied
from 0 to 0.3 in steps of 0.01,and for each value of p the parameter u was varied from 0.1 to 6-0
in steps of 0.1. Figure 8 shows the relative effect of p, and hence the shearing force, for varying
magnitudes of u, the axial load parameter, on four independent stiffness coefficients. Figure 9
represents the converse. In Figure 8, the value of u corresponding to the stiffness element Bij= 0
is the critical buckling load for the associated end conditions. Figure 8 also indicates the decay in
rotational and translational stiffnessesof a member with increasing axial and shear forces. For a
STIFFNESS MATRIX FOR A BEAM ELEMENT 1693
30 1'. aa 2'. 00
n P.
I
L
1 0
0
0 Solution of Eq. (44)to f i r s t
X
approximation
Eq.(2-57) i n R e f . 6
0.40 0.80
-
nb
(d) AG
given value of p, the carryover moments increase with increasing axial loads. However, the
carryover moments decrease for increasing value of p.
Illustration
Figure 10 shows the classical problem of a cantilever beam carrying an axial load P and a
concentrated tip load Q. If this cantilever is represented by a single element, the fixed end
condition is enforced by removal of the rows and columns from equation (38)which correspond
to the rotational and translational degrees-of-freedom at this end, i.e., by deleting row 1,column
1 and row 3, column 3 in equation (38). Inverting the resulting 2 X 2 matrix and performing the
matrix multiplication of the inverse with the load vector , the vertical deflection and
1694 A. K. CHUGH
a
CI
- ?0 . 0 0 4.00 8.00
c
'G. 0 0 4.00 8 . 00
U
( a1 (b) U
U
W
U
(C)
~ ~-C)Q( 1
8= (47)
u2(c-usp)Er
When the effect of transverse shear force is ignored, i.e., /3 = 0 in equations (46) and (47), one
obtains the usual expressionsfor deflection and rotation of the free end of a cantileverwith axial
STIFFNESS MATRIX FOR A BEAM ELEMENT 1695
8
(a)
VI
‘a. ao 0.20 a. uo
P P
(C) (d)
LEGEND
0 u.o.1
8 ---- u= 1.0
t - - - - u.2.0
x ---- ~ ~ 3 . 0
Figure 9. Sample analyses-(CalComp plots)
load P and transverse load Q applied at the tip. It is interesting to note that the effect of
transverse shear force on the response of a beam-column member is not simply additive.
CONCLUSION
An apparently simple formulation is presented for the development of the stiffness matrix of a
prismatic beam element considering shear deflections. The same ideas have been extended to
apply to a beam-column element of uniform cross-section. An alternative formulation for the
stiffnessmatrix of a beam-column, including shear effects, in terms of member flexibilitiesis also
included. Any structural synthesis scheme can be used to synthesize the element stiffness
matrices. Although the apparent advantage of the algorithm presented lies in the use of the
displacement method of matrix analysis, the information developed can also be used in the
classical methods of structural analysis. The critical buckling loads of struts with various end
conditions can be easily obtained. The results of several analyses are presented in which the
relative effects of axial force and shear force on the stiffness coefficients were investigated. The
effect of transverse shear force on the response of a beam-column is not simply additive. The
element stiffnessmatrix has been used with success in the analysis of many structural problems.
ACKNOWLEDGEMENTS
The author would like to express his sincere appreciation to Dr. Louis A. Panek for providing
the incentive for the work reported in this paper, and to Professors Sherrill Biggers, Richard
Gallagher, Hans Gesund and William Weaver, Jr., for their prompt and helpful comments on
the general subject matter of this paper in response to the author’s personal correspondence.
APPENDIX
List of symbols
A = cross-sectional area
A, = shear area = A / n
Bii = stiffness coefficient of a beam-column
Bii= i, jth beam-column stiffness including shear deflection
c =cos u
Di = displacement degree-of-freedom on element co-ordinate i
E = modulus of elasticity
[A=flexibility matrix
G = shear modulus
I = planar moment of inertia
kij= stiffness coefficient of a beam
Gi = i, jth beam stiffness including shear deflection
L = length of the member
n = numeric factor depending upon the shape of the member cross-section
P = applied axial force
P,,= buckling load for given member considering shear deflections
P, = Euler buckling load
Q = concentrated load
STIFFNESS MATRIX FOR A BEAM ELEMENT 1697
u = numeric factor = L
12EI
JG
a = numeric factor = ~
A,GL~
EI
p = numeric factor = ~
L~A,G
S = translational displacement
a1= shear deflection
0 = rotational displacement
REFERENCES
1. A. Ghali and A. M. Neville, Structural analysis, Intext Educational Publishers, 1972, pp. 364-402.
2. J. M. Gere and W. Weaver, Analysis of Framed Structures, Van Nostrand, 1965, pp. 428-430.
3. R. H. Gallagher, Finite Element Analysis, Prentice-Hall, Englewood Cliffs, N.J., 1975, pp. 37-50 and 361-367.
4. R. T. Severn, ‘Inclusion of shear deflections in the stiffness matrix for a beam element’, J. Strain Analysis, 5,239-241
(1970).
5. M. G. Salvadori and M. L. Baron, NumericalMethods in Engineering, Prentice-Hall, Englewood Cliffs, N.J., 1961,
pp. 18-20.
6. S. P. Timoshenko and J. M. Gere, Theory of Elastic Stabiliry, McGraw-Hill, 1961, pp. 132-135.