1998 Investigationo F Stressesa T The Fixed End of Deep Cantilever Beam
1998 Investigationo F Stressesa T The Fixed End of Deep Cantilever Beam
1998 Investigationo F Stressesa T The Fixed End of Deep Cantilever Beam
& Structures
PERGAMON
Abstract
A numerical investigation for the stresses and displacements of a two-dimensional elastic problem with
mixed boundary conditions is reported in this paper. Specifically, it is on the analysis of stresses at the fixed
end of deep cantilever beams, subjected to uniformly distributed shear at the free end. An ideal rnathematical
model, based on a displacement potential function, has been used to formulate the problem. The solutions are
presented in the form of graphs. Results are compared with the elementary solutions and the discrepancy
appears to be quite noticeable, specifically at the fixed end. The present solution shows that the fixed end of
a short cantilever beam is an extremely critical zone and the elementary theory of beams completely fails to
predict stressesin this zore. e 1998 Published by Elsevier Science Ltd. All rights reserved.
1. Notation
x,y
E
1)
o
,t
l,
ox
oy
o,..,
a
b
h
k
R
m
n
0
{t
rectangular coordinates
elastic modulus of the rnaterial
Poisson'sratio
stress
displacement component in the x-direction
displacement component in the y-direction
normal stress component in the x-direction
bending stress
shearing stress
beam length
beam depth
mesh length in the x-direction
mesh length in the y.direction
ratio of the mesh lengths klh
number of mesh points in x-direction
number of mesh points in y-direction
Airy's stress function
displacement potential function.
* Corresponding author.
2. Introduction
The elementary theories of strength df materials are
unable to predict the stressesin the critical zones of engineering structures. They are very inadequate to give
information regarding local stressesnear the loads and
near the supports of the beam. They are only approximately correct in some casesbut most of the time, violate conditions which are brought to light by the more
refined investigation of the theory of elasticity.
Among the existing mathematical models for two
dimensional boundary-value stress problems, the two
displacement function approach [1] and the stress function approach [9] are noticeable. The solution of practical problems started mainly after the introduction of
Airy's stress function [9]. But the difficulties involved
in trying to solve practical problems using the stress
function are pointed out by Uddin [1] and also by
Durelli [2]. The shortcoming of @-formulation [9] is
that it accepts boundary conditions in terms of loading
oniy. Boundary restraints specified in terms of u and v
can not be satisfactorily imposed on the stress function
d. As most of the problems of elasticity are of mixed
boundary conditions, this approach fails to provide
any explicit understanding of the stress distribution in
330
the region of restrained boundaries which are, in general, the most critical zones in terms of stress. Again,
the two displacement function approach that rs the u,
v-formulation involves flnding two functions simultaneously from two second order elliptic partial differential equations [1]. But the simultaneousevaluation of
two functions, satisfying two simultaneous differential
and this problem
of managing the
boundary conditions imposed on them. The age-old SVenant's principle is still applied and its merit is evaluated in solving problems of solid mechanics [3,4] in
which full boundary effects could not be taken into
account satisfactorily. Actually, management of boundary conditions and boundary shapes are the main obstacles to the solution of practical problems. The
reason for the birth and dominance of the finite element method is merely its superiority in managing the
boundary conditions. In circumventing this problem,
0o",,
0o
Eo,.u
(1)
;ox
- + - ; -dy
:0,
;0y- "+ - ; =
:U,
dx
(2)
o,):
o
G+.#)r.r
(3)
If we replace the stress functions in Eqs. (1)-(3) by displacement functions u (x, y) and v (r, y), which are related to stress functions through the expressions
o,:
fau
Evl
(4)
r_ v2L**'d,
E
0r1
6 t : 1 - r , z lfSu
r r *,' * ) ,
ot!
(5)
E f\u Evl
4+v)LO"*l'
'
then Eq. (3) is redundant and Eqs. (1) and (2) transform to
02u, (l - v\02u , /l + v\
02v
/l + v\ E2u
t_-0
0 y 2 '\ 2 ) a x z ' y 2 ) s x s y
(t)
(8)
suitcon-
brium Eqs. (7) and (8) have to be solved now for the
case of a two-dimensional problem when the body
used
has
u(x, y) - 0,
n ) t
d-vl
u :;--;,
oxoy
and
1 l A',]t,
l(1-u)
r-l-vL
a y ' - '"A',1t1
af l'
v:-
JJI
(9)
yfa=0,
for
In order to solve the problem using Eq. (9), the boundary conditions are also needed to be expressedin terms
of {/ and thus the corresponding relations between
known functions on the boundary and the function ry'
are,
a2lr
u(x,y): -,
0xov
(10)
|
v,( x ,\ y ) : - ; :
f ,,
,A2rl,, ^A2rlr1
'dY'
l * u f 'l t t - v ) ; + + 2 ;d+x1' l,
o , ( x r''
,Y)::l
"r\""'
(l
4. Boundary conditions
The practical problems in elasticity are normally of
the boundary-value type where the conditions that are
imposed on the boundary of the elastic body are visualized either in terms of edge-flxity or edge-loading,
that is, known values of displacements and stressesat
the boundary. Referring to Fig. 1, which illustrates the
present problem of cantilever beam, both the top and
bottom edges are free from loading, the left lateral
edge is fixed and the right lateral edge is subjected to
uniformly distributed shear.
For both the top and bottom edges, AB and CD,
the normal and tangential stress components, stated
mathematically, are given by
o,(x, y) :0,
6 * y ( x ,! ) : 0 ,
f A3rl/
'l
' - v' A3{/1
v)2
+
lax2ay oy3l'
E
la3l./ .^
, a3l/l
o/, 'I x'J. 'v \ - ^l^ -+Q*v\=-^:-|.
dxzdY)
(l + v)' YaY'
" u ( x ,'/y' ) :
"6 ^Y\""
a3{r
.Iv-
(11)
(12)
( .^.
13)
a3{/1
ax3l
'
As far as numerical computation is conce.rned,it is evident from the expressions of above conditions that all
the boundary conditions of interest can easily be discretized in terms of the displacement function r/ by the
method of finite-difference.
5. Solution procedure
and
for
l.
and
1.
For the left lateral edge, AC, the normal and tangential displacement components are, respectively,
B
T:,
rE
TH
t ,li
rB
x
Fig. l. Deep cantilever beam subjected to uniformly distributed shear at the free end.
The limitation and complexity associated with analytical solutions [7] leads to the conclusion that a numerical modeling for this class of problem is the only
t a a
)JZ
klh.
Considering an interior mesh point O (i, j), it is seen
that the algebraic Eq. (15) contains the discretized
variable of the 13 neighboring mesh points, and when
O becomes an immediate neighbor of the physical
boundary mesh points, this equation will contain mesh
points exterior to the boundary as well as on the
boundary itself (Fig. 2). Thus, the application of the
central difference expression of the bi-harmonic
equation to the points in the immediate neighborhood
of the physical boundary will cause no difficulties,
provided an imaginary false boundary exterior to the
physical boundary is introduced.
PhYsical
Boundary
Boundary
Top, AB
Bottom, CD
Left, AC
Right, BD
6x,
oxy
6x,
oxy
Utl
o,,- 6-,,
Correspondence between
mesh-points and given boundary conditions
Condition/mesh-point
Condition/mesh-point
6-lQi)
o,l(m- lj)
ul(i,2)
o,rl(i,n - I)
- s)r,rz,i>
o*(Z,i):
sfio,,6t(+
1)+ (, - +) trQ,i
+ t)
+ t.sv(z,i. (+ - z)'t'tz'i
+ 2)+ o's,'l(2'i
+ 3)
-t+t{,(r,i)+ v(3,i)l
+ r)}
.+{{/(r,i+l)+ {/(3,i
fitt,o,i+z)+l,e,i*,)]],
6'ylQi)
6*yl(mi)
vl(i,l)
orl(i,n)
Q,.il
lmaginary Boundary
+i
(16)
i=2
i=3
a - A
t - a
6,v(2,
i) :
i-2 i-t i
j+3 j+4
i+l
i
(a)
(17)
The discretization scheme using the neighboring gridpoints as required for expressing the above conditions
on the top boundary of the beam, AB, is illustrated in
Fig. 3. Special treatments are also adopted for the cor'tranner mesh points which are generally points of
sition' in the boundary conditions. Referring to Fig. 4,
assuming ,B as the corner mesh point, it is seen that B
is a common point of both the edges AB and BD and
thus it has four boundary conditions-two from each
edge. In solving the present beam problem, three conditions out of the four are used, the remaining one is
treated as redundant. The three conditions mentioned
above are organized in such a way that the values of ry'
at three points, namely, I, B, and 2 are evaluated from
these equations-points 1 and B from the boundary
conditions coming from edge AB and point 2 from the
+i
i:3
i=4
i-5
i=6
i-z i-t
i+l
(b)
Fig. 3. Grid-points for expressing the boundary conditions on
the top edge at points closer to A, (a) for normal stress component, o, (b) for tangential stress component, oxl.
Physical Boundary
+i
i:l
i:1
t=5
i:4
(2,n)
V
i
n-5
l=n
o r ( 2n, - l )' : . .
2 y)}
, ! : ; . ; l t . 5 V ( 2 , n )- { 5 + 3 R 2 ( +
( l * v ) ' Rt ht
x V(2,n - I)+ {6 + 4R2Q* v)}rl(2,n - 2)
-{3+ R2(2+v)}l/Q,n-3)
n - 4) + 1 .5 R 2 (2 +v X/(l , n - l )
+ 0. 5{ / ( 2,
+ l/ ( 3,n - l) j - 2 R 2 Q + v ){ (/1 ,n - 2 )
+ l/ ( 3,n - 2) l + 0 .5 R 2 (2
+ v X /(l , n - 3 )
+{/(3,n-3)ll.
(18)
Bottom right, D
Possible
boundary
conditions
lor, o*y) on AB
[u, v] on AC
lo* o,yf on AB
loy, 6,y] on BD
lo", o*yl on CD
lu, vf on AC
for,orr) on CD
lo, oryT on BD
Conditions used
Corresponding
mesh-points for
evaluation of ry'
lo,o*r,v]
( 2 , 2 ) , ( 1 , 2 ) ,( 2 , 1 )
fo"ro*rroy)
( 2 , n- l ) , ( l , n - l ) , ( 2 , n )
fo *,o*r,v)
( m - 1 , 2 ) ,( m , 2 ) , ( m - l , l )
fo
",o
ryo r)
335
+4lb=1.0
"#
2.0
+
3.0
---3.5
0.05
5
0.04
tS
s o.o3
o.o2
0.01
0
0.5
t l a
't
.
t l a
0.0015
+xlb=0.00
--tr0.10
r'--.F
0.25
-.+
0.50
+y/a=0.00
"-o0.05
----r
0.10
+
0.50
0.90
+
1.00
0.001
5x10-a
tq
0
-5x10-a
-0.001
5
y l a
Fig. 6. Distribution of the displacement component y at various longitudinal sections of the cantilever beams (alb : 2).
-0.0015
0.5
x l b
336
+a/b=1.0
+
2.0
#
3.0
-#
3.5
lrl
\ "
+yla=0.00
.---.o0.05
+
0.25
+
0.50
0.7s
0.90
*
1.00
l.$
-0.001
-0.002
5
xl b
0.5
x l b
Fig. I l. Distribution of bending stress o' over various transverse sections of a deep cantilever beam (alb : 2.s).
0.012
0.006
0.004
Fr
tq o.oo3
61 0.002
0.001
-0.006
0
-0.012
0.5
x l b
Fig. 10. Distribution of bending stress on at the flxed end over
the depth of deep cantilever beams.
-0.001
0
0.5
x l b
0.003
+yla=0.00
*
0.05
.-.---Cr0.10
#
0.50
0.90
0.002
t
1.oo
s 0.001
h
5
x l b
J J I
Fig. 13. Distribution of shearing stress o,, over various transverse sections of a deep cantilever beam (alb : 2.5).
7. Conclusions
References
[] Uddin MW. Finite differencesolution of two-dimensional elasticproblemswith mixed boundary conditions.
M.Sc. thesis.CarletonUniversity,Canada,1966.
B. Parametricsolution of
[2] Durelli AJ, Ranganayakamma
stresses
in beams.J EngngMech l9B9;115(2):401.
[3] Horgan CO, Knowels JK. Recentdevelopmentconcerning S-Venant'sprinciple. Adv Appl Mech 1983;23:179269.
[4] Parker DF. The role of S-Venant'ssolutionsin rod and
beamtheories.J Appl Mech 1979;46:861-6.
[5] Dow JO, JonesMS, Harwood SA. A new approachto
boundary modeling for finite differenceapplicationsin
solid mechanics.Int J Numer Meth Engng 1990;30:9911 3 .
'
338
l6l
Durelli AJ, Ranganayakamma B. On the use of photoglasticity and some numerical methods. Photomecfr Speck
Metrol, SPIE 1987;!14:1:-8.
171Idris
,ABM.
boundary-value
,
new approagh to solution
elastic
Bangladesh,University
problems.
M.Sc.
..
of mixed
thesis.
[8] Ahmed SR. Numerical solutions of mixed boundaryvalue elastic problems. M.Sc. thesis. Bangladesh
University of Engineering and Technology, Dhaka,
Bangladesh,1993.
[9] TimoshenkoSP, Goodier JN. Theory of elasticity,3rd
ed.,New York: McGraw-Hlll,1979.
[10] Leipholz H. Theory of elasticity. Gronigen: Noordhoff,
1974:219-221.
.r1
*t
.;