Z
a]~rz
~'+ a-T dz
~ Trz
H
-..,=---O'r+ d-~---d r
r rrz a ~ - - d z
rrz+ 0 ~ - d r z
H
• r
V e r t i c a l section
O"
~ ~ o~r
2. T H E O R E T I C A L ANALYSIS
G = ~-~ (2 - v)V20 - t~Z2) (8)
The solution to be derived is for the case of a
cylindrical specimen loaded axisymmetrically. The cylin- &b
drical coordinates, r, 0, Z, and the dimensions of the a0=~ vV2(b--r~r (9)
specimen are illustrated in Fig. 1.
t~u aw
Y'==~+ 8~- (4) 2.2 Selection o f a stress function
A solution to the biharmonic equation (6) can be
~,0 = ~z0 = 0 (5) written as a polynomial even in r and odd in Z, and
also as the product of Bessel functions and terms con-
A suitable stress function ~b must satisfy the equations
sisting of sine or cosine functions; the general solution
of equilibrium as well as the equations of compatibility;
will be the sum of the two.
such a stress function can be obtained if it satisfies the
The polynomial part of the stress function used is of
biharmonic equation
the ninth order. In fact, a higher order polynomial may
be used with additional restrictions on the boundary, but
~+- + +- +
r~ \&~ r~r aZ2} the effect of higher order terms will be very small.
However, using a polynomial of less than seventh order
= V2V2~b = O. (6)
will result in a solution not satisfying the boundary
Stresses and displacements in terms of the stress func- conditions on the cylindrical surface of the specimen[5].
tion can be written as follows[lO] : To account for the degree of end-constraint, two strain
factors are introduced to force the analytical solution to
a , = ~o ( vV2~- &2] (7) fit the experimental solution. This requires a ninth order
polynomial to be used in the stress function.
Stress Distribution within Circular Cylinders 47
The following stress function will be used: tr~ = a 1 - 12C - 2(1 + v)D - (480Z 2 - 240r2)F
- [12(8 + v)Z 2 - 6(7 + v)r2]G
(480Z 4 - 1440Z2r 2 + 180r*)M
=
(1 + v) (1 - 2v)Z s + Zr 2 -
24 Z 9
+ gg 292 ZTr2 -I- ~- ZSr 4
~1 Z r3 6 ~ l r S ) Q + (~_iZ9
+ -! i iK cos (16)
24
7 ZTr2 + 9ZSr* - 5Z3r6 + ~15 Zr s \)~"_ ao = 6C + 2(1 + v)D + (240Z 2 - 60r2)F
+ [6(9 + 2v)Z 2 - 6(2 + v)r2]G
+ (240Z 4 - 360Z2r 2 + 30r*)M
+ [80(4 + 7v)Z 4 - 60(1 + 28v)ZZr 2
+BrJ,(iKR)Isin(K-~R ) (14)
-- 30(1 - 7v)r4]N
+ [~(73 + 70v)Z 6 - 4(57 + 21Or)Z% 2
where al is the average applied stress, Jo and J1 are
Bessel functions of zero and first order, respectively, and
C, D, F, G, M, N, Q, S, A, B and K, are constants to be + 9(1 + 70v)ZZr 4 + (4 - 35v)r6]Q
determined from boundary conditions, i denotes x / - 1.
The coefficients of Z and r have been chosen such that
the biharmonic equation is completely satisfied. By sub- + 48Z 6 - 180Z4r 2 + 90Z2r 4 - ~ 15
- r 6\)S
stituting equation (14) into equations (%13), we obtain:
R 1 1j1 iK
2 r-i
+ iB[(1-2V~KJ
.(r)o tK-R
r K 3.
+ 2(1 - RI
v)-~-iJ~ ( i K ~ ) ] } (~_) sin (KR_Z).
+ rljI(iKR)I}(K)ac°s(~-Z (15,
(18)
48 M. Al-Chalabi and C. L. Huang
7
~ I 0-1
E cients of terms multiplied by r are zero, and yield
--W 12C - 2(1 + 2v)DIZ
(1 + v) (1 + v)
- - (160Z 3 240ZrE)F
- -
s = 1 (67 - 7 0 v ) Q (21)
- [8(4 + v)Z 3 - 6(7 + 2v)Zr2]G
- (96Z s - 480Zar 2 + 180Zr4)M
M = g(13 - 14v)N - HEQ (22)
- [16(1 + 14v)Z s + 160(3 - 7v)Zarz
- 30(13 - 14v)Zr4]N
1 14
7HZN +
+ [ - ~ ( 1 9 + 35v)Z 7
"F- 40(7 + 2v~3
-3
H'Q(23)
(1 +
E
v~)u =
I --V0-I(1--~v) + 6C + 2D r 1 = 03 - 6C - 2(1 + v)D + ~(3 - 2v)R2G
- [420H2R 2 + 35(5 - 4vjR4]N
+ (24OZZr - 60ra)F + (54Z2r - 12ra)G
+ (240Z% - 360Z2r a + 30rS)M + [168R2H 4 + 175HER 4
+4r71Q+[48Z6r-180z4ra+90ZZr 5
154r T ] s - [ A j , ( i K R )
+ iB. (1 --2v) H Jo inII-~
..,_,,r,o(,,, r cos T)(2o)
(25)
6C + 2(1 + v ) D - t r 3 + [ 4 H 2 - ~ ( 3 - 2 v ) R 2 1 G +2,, _ H 1 . R
• /nr Z'
+ [-5-6-H4- 140(3- 2v)H2R 2 sin ~-~---) (30)
v,o + E,120,,
H 2
- 6 x 1680 - 840(3 - v)R 2 N
i--
+48x 1680(~-)]
H4Q
}=
A
.[Jo(ln17~)
" R + 6 x 6720(3 - v)
( ;1tQ
- -R 7 Jt inH
+ iB. [ ( 1 - 2 v ) ( n ~ ) J o ( i n H R )
(4) Fourth Boundary Condition.
+ - 71 inFl (28) The total axial force applied is equal to the resultant
1 of the vertical load transmitted to the specimen, i.e.
(3) Third Boundary Condition•
2II ,,,R (TZ Z = H
The shearing stress at the cylindrical surface is equal
zero. Putting r = R in equation (18), we obtain ~ = poly-
nomial in Z + series of sines of (nl-lz/H) Substitution of equations (21-24), and (29) into equation
At the periphery of the end surfaces, the shearing stress (16), and then into the equation (33) yields
is zero only if the polynomial in Z is equated to zero. 6C + (I - v)D - [280(2 - v)H 4
To satisfy this condition, we have
- 140(1 - v)HER 2 - 70vR4]N
G = ~ (210R 2 - 280HE)N - (56/-/4 - 140H2R 2
- I 1 - ~ 5 vR6 + 3 5 ( 1 - 3 v ) H 2 R 4
+ 35R4)Q (29)
- 28(13 - 9v)H4R 2 + 56(1 - 2v)H6]Q
and
[ - 3 3 6 ( 3 - v)RQ]Z 5 + {-1120(2 - v)RN
+ [1120H2R + 840(2 - v)Ra]Q}Z a
+ {-12(1 - v)RG + [l120H2R + iB,,[do(inl-I R )
+ 840(1 - v)R3]N - [448H4R + 840H2R 3
(34)
+ 210(1 - v)RS]O]Z
50 M. AI-Chalabi and C. L. Huang
IH\4 f 1 \4
1680 ( n ~ ) 5 c o s ( n H ) { I - 2 ( 2 - v ) R 2
( t]t ,40,
- 24H 2 + 192 Q
(41)
•HR
and
"( ")3
+ - Jt inH
1
,36, A, = 1680 cos(nH)(R 2 TzN + R 4 V2Q) (42)
where
3 2 6(3-v) H 2
= A~jI(inHR)-t-iB.[RJo(inHR ) q
H t )l (43)
+ 2(l - v'(H'~l[nH,i J t ( i n l l R ) 3 (37,
J
H 1
We now have the four equations (34-36), and (37) with
six unknowns: C, D, N, Q, A., and B..
After some algebraic manipulation, we can write
where
T 1 =~ (2 - v) + 8(5 - v)
(44)
iJo(inHR)l
By eliminating the constants A,, B,, C, and D, we can
t write the stress and displacement expressions in terms of
the two constants N and Q, only.
Stress Distribution within Circular Cylinders 51
q, =/1
F (Co)H,_Rl (45)
n = (1 (~,),,,R~
(E,). }
(50)
t (co). ]
where
where
(Ez)H,R = axial strain at the periphery of the end sur-
(co)n.R = tangential strain at the periphery of the end faces, and
t surfaces. (Ez)m = axial strain at r = R and Z = 0 for the case
(%)m = tangential strain at r = R and Z = 0, for the of zero friction between the surfaces of the
case of zero friction between the end surfaces cylindrical specimen and the loading machine
of the cylindrical specimen and the loading platens.
machine platens. As before, we have
Thus, we have az = el, o-, = o-o = aa, z = 0 and there-
1
fore (Cz)m = ~(O-I -- 2VO-3) (51)
1 By using equation (3), we can write equation (50) as
(%),. = ~ [(I - v)a3 - vo-,] (46)
fl = Ra~tN + R6flQ (52)
Putting r = R and Z = H into the equation (2) and where
rearranging we can write equation (45) as
= R4~N + R6rIQ (47) - (o.1 _ 2vo-3) -- 896v
where
- 560(1 + v) + 140v(1 - v)
~=
[(1 - v ) ~ 3 - w r , ] I4 (-/,
48 ~- -70(1 - v )
- _ 3360
n=l
T, Jo i n n
+ - - dl inIl (48)
, {640
fl -- (o.1 _- 2vtr3) -- ~ - v
(H)4
-- 112(1 -- 5v) ~-
1
(l- V)F320(H~6 (g) 4 +140(2+3v-v2)(H) 2
r/=[(1 -~-)~-val]L 3 ~R] -336 ~-
- 70(1-v)(H) 2
+
--~--(1105
-
]
v)I
1052 v(1 - v ) - 336 ~
n=|
_ v 3360 ~ ~-~
[(1 - v) 0"3 - w r d . = 1
V1Jo inIl ~ + ~ .
+ ~ 1680(1 + v) Jo inIl-H
n=l
(l+v) ~ (H)3(~.~)3
+ [(1 - v)o-3 vo-1].=1 1680 IF2 + 4(1- v) (RH---)(n~)]
[V'J°(inl"-1-~-l+V~ji (49, + - - J1
l
(54)
52 M. Al-Chalabi and C. L. Huang
(E=)m=
~ and (%)= = 0-'
- v ~- (58) (H)(~--H)3+~JI(inHRR)[RTx
Equations (45) and (50) can now be rewritten
= I + E(%)n,R (59)
1~0-1
7J,
= 1 E(E=)n,R (60)
0-1 r R V2
(inH-R R ) [ ( R ) V 1 - ( H ) ( 7 - ) ( ~ ) 3 ) }
Knowing the elastic constants (v and E) of the material,
the two strain factors can be determined for any end (61)
condition by making two measurements (% and E~)at the
perimeter of the end surfaces.
0-z=0-, + 0 -560(2-v) ~ +280(1 - v )
2.5 Final stress expressions
+ 140v + [1120(2- v)(H) 2 - 8 4 0 ( 1 - v ) ] .
The final stresses can be expressed as follows:
(Z)2 (H) 2 - [ 5 6 0 ( 1 - v,(H) 2
0-r =- 0"3 + 0 392 ~- -- 280 + 70(2 -- V)
+ 1680(I - v ) \ ~ j \ ~ ] '}
- 1 0 5 ( 3 - 2v)] (R) 2 - 840(3 - v,. + A J ~ - v + 70(1 - 3v) - 56(13 - %)
+ A{-7-184 ;H']6
+ \ R - J 280 (H) 2
II
I + + + + + + o + + I + I I I I I I
"< I;> ~ ~ I Ix.) I I Oo I -I-
i i
+ 4- ~ I 0 ~ ~
@0 + ~
~- I i + + ~ + :~1"~
~ ~.1 ~-,- I".J ~'~ t~
I
I I I I
, ~ + -F
~ + ~ ~ ~ +
+ ,.=., ~ -..J
I I
I , ,-t
, I ~'~'~-~,
: O"
00 r-
I I i.
0
0~
t~J
~o
r~
II
4- -I- I 4- ~ 4- I -I- I + I I I + I
- ~ ~.~ ~ -- O0 ~ ~ ~ I--1 D ~, ~, ~l ~ ~ ~ ~ .~ i
r~
r~
~ I J
(.~
0 .p.
t.,,)
54 M. AI-Chalabi and C. L. Huang
solution and that of Filon [1] for the shearing stress. The
other solutions do not satisfy the boundary condition of
no shearing stress at the cylindrical surface of the
+,,, specimen. There is a large discrepancy in the tangential
and the radial stress distribution obtained by the different
-
(
A V1 ~r Jo in H ~
l
+ -l J1
authors, but the radial stress at the edges of the specimen,
obtained by Filon, Balla and the present writers, is zero,
and thus satisfies the boundary condition of no radial
stress anywhere on the cylindrical surface of the specimen.
V1 4. EXPERIMENTAL VERIFICATION OF
THEORETICAL ANALYSIS
The validity of the present solution was checked by
The stress distribution throughout the specimen can
therefore be obtained for any type of end condition by experimental stress analysis methods [5]. A 3 by 6 in.
cylindrical specimen was fabricated from epoxy sheets
computing experimentally the two strain factors and
substituting their values into the final stress equations. which were cemented together with epoxy adhesive. A
large number of foil strain gages of lengths ranging from
In addition, strains also can be obtained from equations
(l-S).
0.03q)'125 in. were embedded at specified locations in
the interior of the cylinder during its fabrication. Addi-
tional strain gages were bonded to the cylindrical surface
3. COMPARISON OF SOLUTIONS
of the cylinder after it was completed.
A comparison of the stress distribution for the case of The cylinder was tested uniaxially in compression
perfectly confined ends, as determined by various investi- within the elastic range of the epoxy material. Testing
gators, is shown in Fig. 2. All solutions agree on the was performed for four end conditions: teflon sheet
form of the axial stress distribution, although the magni- inserts, direct contact, cardboard inserts, and perfectly
tude at the center and at the edge differ considerably. confined ends. Teflon inserts cause low friction at the
Good agreement will be noted between the present surfaces of contact; perfectly confined ends provide the
2.0
.... Filon
1,8.... Pickett
---- D'Appolonia-Newmark
I-6--- Balla
New solution //~,/'
1.4 -- . ./f.~
z~
I,z - "~
o, a .;
- - ~
_____...--
o6 - = --=- ////
0.4
0.2
0"0
0.4
~ ' -. . . ~. - - - - r -~-~. ,
__
~
R
Dist"once olong rodius - ~-
highest possible friction at the surfaces of contact, while due to the friction between the end surfaces and the
the direct contact and the cardboard inserts provide machine platens.
intermediate end conditions. (2) High friction at the specimen-platen interface
The interior and exterior strain measurements through- results in high stress concentrations in regions close to
out the specimen, for the four end conditions, were the end surfaces and also at the edges of the specimen.
compared with the strain distribution obtained from the Stress concentrations become smaller throughout the
theoretical analysis, and reasonable agreement was specimen as friction decreases.
observed for all cases [5-1. (3) For other than perfectly smooth end conditions,
The stress distribution throughout the epoxy cylinder compressive and small tensile stresses occur in the radial
was computed from the derived equations by inserting and tangential directions.
the strain factors computed from experimental measure- (4) Division of the applied load by the cross sectional
ments. Figures 3a and b show the stress distributions area of the specimen does not accurately describe the
throughout the epoxy specimen for two cases of end true state of stress, although this procedure is commonly
conditions: direct contact and teflon inserts. The stresses used in engineering practice. Uniform stress distribution
are shown in the figure in non-dimensional terms, throughout a specimen loaded in compression exists only
determined by dividing each stress by the average applied in the case when there is no friction between the end
stress al, (trl
= 200 psi). A comparison of these two cases surfaces of the specimen and the machine platens.
shows that the stresses become more uniform as friction (5) A solution is obtained in this study which takes
decreases. It is expected that, as friction is eliminated into account different degrees of friction at the specimen-
completely, the radial, tangential, and shearing stresses platen interface. The solution can be used for both axial
vanish everywhere throughout the specimen, while the and triaxial type of loading and is independent of the
axial stress approaches the average applied stress. specimen dimensions. All boundary conditions are satis-
fied everywhere on the cylindrical surface of the specimen.
5. C O N C L U D I N G REMARKS Acknowledgement--The'authors express their sincere thanks to
Professor Frank J. McCormick for his help and comments during the
(1) Specimens loaded in compression in common preparation of this paper.
laboratory tests are actually stressed triaxially. Non-
uniform stresses are developed throughout the specimen Received 14 May, 1973.
-0"5-0'7 -0.4
- o . ~
-~ 0"0 0"0
0.0 0'0
0.0 O-O
0'0 V'~°~° I ~ - 0.0
0-0070 0-0044
Radial stress distribution
0-0
0.0
i
0'0034 0002
( D i r e c t contact) (Teflon i n s e r t s )
~
-I'1 -0'85 -I.I
-i.o-o.9o \
-0"95 -1"60
/?o 71" -I.23
-0'868 -0.938
-0"915 -0-960
-0"992 -0.996
tMid-plane I J
-I.00g -1"004 -0'998 -1'004 -1'0094 -0.999
Axial stress distribution Axial stress distribution
( Direct contact) (Teflon inserts)
Fig. 3(a)
56 M. A1-Chalabi and C. k. Huang
-0.5 -0.8
-0.4
- 0.412 -0-195
-0,2
-0'001
0-0005
' 0.032
-00 0-015
0-008
0-006
0.003
/Mid-plane I I I~ Mid-plane I I I I
0.0071 0 . 0 0 6 5 0.0024 0-0034 0'0030 0"0014
Tangential stress distribution Tangential stress distribution
(Direct contact) (Teflon inserts)
-0-3
-0"1
/
0.0 0.0
-0.05
0.0 0.0
0.0
00"01 0 0-0
0.0 0.0
Mid-plane I I 1 I Mid-plane I 1 I I
0-0 0.0 0.0 0.0 0.0 0.0
Shearing stress d i s t r i b u t i o n Shearing stress d i s t r i b u t i o n
( D i r e c t contact) (Teflon inserts)
Fig. 3(b)
Fig. 3(a) and (b). Stress distributions.
REFERENCES
1. Filon N. G. On the elastic equilibrium of circular cylinders under 6. Peng S. D. Stresses within elastic circular cylinders loaded
certain practical systems of load. Phil. Trans. R. Sac. A198, 147 uniaxially and triaxially. Int. J. Rock Mech. Min. Sci., 8, 399-432
233 (1902). (1971).
2. Pickett G. Application of the Fourier method to the solution of 7. Brady B. T. An exact solution to the radially end-constrained
certain boundary problems in the theory of elasticity. Trans. Ant circular cylinder under triaxial loading. Int. J. Rock Mech. Min.
Sac. Mech. Engrs. 66, AI76-AI82 (1944). Sci., 8, 165-178 (1971).
3. D'Appolonia E. and Newmark N. M. A method for the solution of 8. A1-Chalabi M. Discussion of S. D. Peng's paper, Stresses within
the restrained cylinder under compression. Proceedings of the First elastic cylinders loaded uniaxially and triaxially. Int. J. Rock Mech.
U.S. National Congress of Applied Mechanics, 217-226 (1951 ). Min. Sci., 9, 665 (1972).
4. Balla A. A new solution of the stress conditions in triaxial com- 9. Brady B. T. Effects of inserts on the elastic behavior of cylindrical
pression, Acta tech. hung. 28, 349-387 (1960). materials loaded between rough end-plates. Int. J. Rock Mech. Min.
5. A1-Chalabi M. Stress distribution within circular cylinders in Sci., 8, 357-369 (1971).
compression, p. 199, Ph.D. Dissertation, Kansas State University 10. Timoshenko S. and Goodier J. N. Theory of elasticity, McGraw-
(1973). Hill, New York (1970).