Lehigh University

Lehigh Preserve
Fritz Laboratory Reports Civil and Environmental Engineering

1954

Analysis and tests of a cylindrical shell roof model.,

Proc. ASCE Separate No. 434, 1954, Reprint No.
95 (54-2)
B. Thurlimann

B. G. Johnston

Follow this and additional works at: http://preserve.lehigh.edu/engr-civil-environmental-fritz-lab-

reports

Recommended Citation
Thurlimann, B. and Johnston, B. G., "Analysis and tests of a cylindrical shell roof model., Proc. ASCE Separate No. 434, 1954, Reprint
No. 95 (54-2)" (1954). Fritz Laboratory Reports. Paper 1469.
http://preserve.lehigh.edu/engr-civil-environmental-fritz-lab-reports/1469

This Technical Report is brought to you for free and open access by the Civil and Environmental Engineering at Lehigh Preserve. It has been accepted
for inclusion in Fritz Laboratory Reports by an authorized administrator of Lehigh Preserve. For more information, please contact
[email protected].
 .' . ~ .. . .
Al'TALYSIS A~'D 'IESTS OF' A
CYL!--NDRICAL SHELL ROOF MODEL

by

Bruno ThHrlima nn ~l-

and
Bruce G. Johnston"C--:l-

Fritz Engineering Laboratory Report 2l3K

Synopsis:

, Analytical expressions for the effective width of cylindrical

shells, adjacent to stiffen1n~ ribs in circumferAntial direction,

are presented. Use of these formulas is msdefor ,the analysis
of a cylindrical shell roof acted upon by horizonta 1 longi tudinal
forces •• Furthennore, the influence of foundation movements,
temperature changes, and shrinkage is studied.

The theoretical results are compared with experimental

,:/

measurements obtained on a model in the epprC'ximate scale I

to 30 of an actual cylindri cal shell roof. Good agreern en t

between analysis and test 1s established.

, Introd uct ion:
It is of ten "believed that the use of shells- as structural
elements is's modern development. But two thousand years ago
.
the Roma ns already admired th~ir Pantheon, e bui lding covered
by a sphericsl shell of about 145 ft. diameter. The cupolas of
the cathedrals of Florence and St. Peter in Rome, built during
the 15th and 16th ce~tury respectively and having a diameter of
about, 140 ft, are examples of remarkable craftmanship, based on

• experience am intuition only. 'Ihe contribution of our- age to

this type of structures consists in develop1ng a rational analysis
*Farmer Research Engineer, Fritz Engine~ring Laboratory, Lehigh
UniversiV, Bethlehem,- Fa.
**Professor- of Structural Engineering, Civil Engineering Dept.,
Universi ty of Ml chigan, Ann Arbor, M1 chigan, fo rmerly Directo r
of Fritz Engineering Laboratcry, Lehigh University.Bethlehem, PSo
 , f

2 13K -2-

pred~cting sufficiently close the state of forces in the shell.

The development of the "Shell Theory" was initiated by

Love (1)* who derived thf'J diff~rential equat:l.ons for curved
plates. Meissner (2) succeeded in integrating these equations
for sphericsl and conical shells. Bauernsfeld' s (3), Fins-
terwslder's (4), and Dischinger's (5) contributions made an
spplicsti on of the theory to the e nalysis of actua 1 structures
pos sible. A number of shells in reinforced concrete were
bUl"rt in Europe· after the fi'rst world war and parallel to this
development important theo.retical contributions were made.
It was the Roberts and Schaefer E 0' eex}ie Company,
C~cego, Il~inois who intr~ this promising type of structure

ip the Uni ted Sta tes. The Ice Arena in Hershey, ?ennsylv ani a, 1936

( 6) *..~, wa s the f:1r st c:v lindri cel she 11 roof in the Uni ted Sta tes.

In the following years and especially during and after the war,

this company was designing a number of shell roofs used as

storage houses, fectory buildings, airplane hangars, sport
arenas, armories, etc., covering a total of ten million sq. ft.
Today the largest span , of cylindrical shell roofs arrived
~

at is 340 ft. (e.g. Hangars at Rapid City, South Dakota (?),

and Limestone, Maine (B)). These large span structures

accentua ted certai n pr oblems which were of secondary importsn ce

in the smaller structures, built previoul!llyo FUrthermore, the

t,o/warq P? incree se ~.. the span, so tba t a careful·

tenden cy is
rpt~ ~ ~ ~ ~
study of ~ Eluestions becAme mandatory. iiI JSiJilSliieU!I et'

*For list of references see p. 31.

*..''Ref. (6) outlines the principles of shell structures end gives
8 number of actual eXBmpleso
 2l:3K -3- ,/

--~'--_. - .-- .-. \-.-

"The supporting ribs in cylindrical slB 11 roofs may be
designed as tee sections to include tlB addi t1 ona.l effecti va
section provided by the shell. XXl{ft~XXD« A study l6 of
tilDXJl~i;f!llIDCX th:l interaction between too ribs and shell.x
shows that an "effecti ve width" of shell can be determined
and the combined section used in a rela ti vel,.. s 1mple arch analy~is
of the ribs true tUl:' e • n

~r ell4 M pie
Lateral horizontal forces, caused) ~ )by .the wind action on X

the front and back door of large span cylindrical shell roofs,
(Fig.l) can reach a considerable magnitude. Their influence on
the structure can no longer be neglected as secondary. A difference
of tempera tUre inside and outside of the bu1lding may produce rela-

tively high thermal stresses. Very important become the effects

, of shrinkage an d pIa stic flow on the stabili ty of the structure.

These 'are just a few of the additional problems to be considered

. . ,
in the analysis of large shell roofs.

The Roberts and Schaefer Company, on the be sis of their broad

experience in designing and building shell roofs~ decided to sponsor
a two,:"yeer reses rch program on shell arch roofs at Fritz Eng:lneering
Labor'story, LehigJ1 UniverSity, Bethlehem, ·Pe., with special emphasis

.on the questtons menti0ned above. * A careful theoretical study of

the problen s was made and the results were checked experimenta lly

on 8 model of a cylindrical shell roof (Fig. 10). The work re-

sulted in a number of reports {Ref. (18)), from which in this.

Paper :3 contributions to the analysis of shell roofs are presented.

It will be shown that the model tests' are in excellent agreement
with the thee ry.

* The Project was started Jan. ~, 1949 and terminated Feb. 15, 1951.
 2l3K --4-

I. Effective width of cylindrical shells reinforced by ribs

In circumferential direction:*

It is well known that the stress distribution in the flange

of e T-Beam wi.th a streight axiS, subjected to bending in the
plane of the rib, is not constant over the width of the flenge
I

(Fig. ,2). The simple t'Beam Theory", based on the Navier-l3ernoulli

hypothesis that strain due to bending varies linearly, can be
"saved" if ~he actual width of the flange is replaced by the
effective width b. Th.v. Karman (9) was first .to derive the
correct theoretical ~xpression for thp. effective width. ($ee'
(10) for a recent resum~ of the most important articles published
about thi s subject). Shear lag is responsible for the considerable
damping of the direct strasses in the. flange in directton of the rib o
Consider now a cylindrical,shell stiffened by a rib 1n cir-
cumferential direction (Fig. 3) •. Arbitrary loeds, acting in the
plene of the rib will cause direct forces N per unit width in
1
circumferential direction as shown in this figure. The.actual
cylinder,may be replaced by 8 ring of width b for which a constant
direct force {NCf )x=o' equal to (~-.f A of the cylinder, over t,he
en tire width is assumed. By determining the width b 80 that

b<N<f lx=o - ) N1 dx
(Ring) (Cylinder)
the stresses in the ribs are obviously identical for both structures
under an equal load systen. Once the width b, called the effective
wid th of the cylinder, is known, the calculation of the rib stresses
becomes a problem of tne "Simple Beam Thea ry" (cross sections rems in
plene) 6

-
* Only a very ahort outline of this problem can be given here. For
s complete study references (15) and (16) should be consulted.
r ..'
de~lved
expressions
-~ "

fOl' the effective width of ashr:ill that 'exten.ds .

I indeflnitiGly
.

special caae (
on either
.
side of a rib..
(3f-:: 00 In Fig .. 4>
This 113 a
of the author's

I more general solution where the rib may be any

dis'tance n J" from a free boundary."
\
-------
---_._~-- '~------------~----<~---""'_"'"t
t·

...'

,,'
 2l3K 5

Analytical expressions for the effective width, based on

the g8'leral bending theory of cylindrical shells (eog. Ref. (11),
po 433), were derived in Ref. (15). Ref. (16) is a short report
on this study and compares the analytical values for b with ex-
INSc~ T
perimental results. It is m own that' essentially two effects
S€f. BAc;K
at: PAtrG.
govern tbe effective width of cylindrical shells, (1) the lag of
4.
the direct shear forces in the plane of the shell (as in the case
of 8 T-Beam with a straight axis) and (2) the radial escaping of
the shell under the circumferential direct forces N~" In
general b can be written (see Ref. (15), p. 41):
(1 )

Where: b
a
= effective' width
\

= radius of the cylinder

h
K
= thickn~s of the cy1irider
= f (~,;:It given in Fig. 4 •
. #

The par~meter~, used in Fig. 4, takes i~to account the

influence of the direct shear forces on the effective width:

"k- n~
8 ,
where n depends on the variation o~ the stresses in circumfer-
entia1 direction (n is' the number of complete waves made by e
harmonic function ,around the circumference of the cylinder, see
Fig.4). For n=O, the stress distribution has axial s~etry

(e.g. constant radial 'line load around the rib) Bnd the effective
width has its ne ximum value. The parameterA depends on the
length of the overhang ~ of the are 11 Bnd on the shell cons tan tx.
/ i: 4{3 * (3)
{Sll

As shown in Fig o 4, the shell is assumed to extend to infinity on,

. .

*See eog. Ref" (11), p. 392: ..~ 3(1

2
-'2'" ). The influence 'on b
's h
of Po'-sson's ratio "y 1,s insignificant, and hence ~ Is neglected in
Eq" (3).
 213K 6

the left side. This may be safely done, if the factor I d > 20<4,
where d is t~ distance of the end from the rib (see Ref. (15)
p. 25).

Taki~g the effective width b as flange and the rib as web of

..
a T-secti on, the Bernoulli-Navier hypothesis leads to the correct

values for ~e rib stresnes. Furthermore, calli~g the circum-

I
ferential stress along the connecting l~ne of ri b and spell () A'
the total direct force S in the shell is the integral of all direct

forces N~ =()T h over the entire length of the shell, or equal to

the product of the stressa-A' the thickness h and the effective

wi dth b:

s " fN 'f dx "O"'A hb (4)

For a given S and a g1. ven s et of boundary condit1.ons all forces

. end momenta in the shell can be calculated. Ref. (15) g1.ves a

number of tables and graphs which reduce these calculations to a

minimum.

In, summary, the knOWledge of the effective width of B cylin-

drical shell simplifies a very involved problem of elasticity to

e common problem of ~he ordinary Beam , Theoryo In the following

chapters use of this effect! ve wld th is made for the an elysis of

shell roofs.

Ito Cylindrical Shell Roof subjected to horizontal longitudinal

forces.

The action of the wind on the front end back door of cylin-
drioal shell roofs (Fig. -1.) produces considerable horizontal force.

An ana lysis fo r this type of a loading 1s presen ted and numerical

results are compared to test results obtained on a model.

1.) Problem:
213K 7

One unit* of a' cylindrical she'll roof, shown in Pig. 5 is

subjected to' horizontal forces 1n the directt on of th~ axiso

On a cross section a shear force V, a bending rnomEfl t ME and s
twisting moment MT are acttng. The problem is to find the stress
distribution due to these forces end momerite •

. Consider a part;of infinitesiJTlsI width e~d" cut out (F'ig06).

The force 8 and mom en ts acting on a infinitesimal shell elemen t
dx.e·dl':f ere defined i~ Fig. 70' Obviously the direct force Nx in
exial.direction is eq~al to the applied uniforwly distributed
)

force p at the rib x = 21 and equal to zero at the rib x : o•

.:~.~.
Assuming a li,neer varleti on of Nx along the x - axis, then
- Nx = x p
21,
and ONx - ,:: 1 p ( 6)
~ z:r-
The equilibrium of an infinitesimal shell element dx.ird<j in x -
direction g1.ves (Fig. 6)

ON x dxac'l<j '" ONtx

oX d'f
And replec:'.ng ~Nx I by its value of Eq. (6 )
~x

~~x __
-
~it

1..
(7 ).
Of.)'
By integra ting Eq. (7) end c m sidering the fact that the shear
force N is zero at the center of the spen, due to s'YJIlmetry
xC;f
conditions, N,x becomes

N yx =-h p l'
- -'
*The erect1. on of cylindrical s1::811 roof s is mede in un1 ts in order
to re-use the form work Elnd furthermore to , provide diletatioD .1oints o
,

*~fThis assumption holds certe1 nly for shell roofs whose width is'
relatively small to the span L.
 .i.
213K 8

It results that Nc:pt is ccnstant fGr a sec.tion 'J= constant and I

varies proper ti onal to the an gle ~ along the s ps n of the shell~

From theequallty of the shear 'forces '¥:<J1x =lxcr' the direct,

'",'

shear for ces N x ~ in radial direct:i on·can be taken e ~a 1 to the

, .'
dl rec t shear forces N ~x as a good approxlma ti on:'
'*
~x T'= N 'Ix ~. ~ 8 t P.<:jJ (9)
,2, '
If Imaginar:~ suppor t f0r~eS,.-'Nx'~, acting along· th~, outer ribs
x = 0 and -x::. 2"2', are assumed, the shell is in ,eqUilibrium under

the load ,P and subjected to' shear forces NX <1'snd N ~x; and di,rec·t,
forces Nx only. ActuallY ·thel!!e Imaginary fp'rce: El arenpt ~cting;. t\,
Introducing shear forQ88 -::N xr s<:tlng ~long' the, connecting I,i,ne, (.:) .'

of the ::ell .and the' ri ba th}ese ima~lnary forces N.xp are ellm~ \( ~"X'
instad. The interaction between the. rib and 'the shell
~. . is ts ken
. - - - -' - .-

, I

into account by considering 'the effect1. ve width of the she 11 sa ,8,

fl angeof the. rib. This effective section, consisting of ther,i b

I .
I

and the effective width, 1s used 'inanaly~ing the arches.

20) Analysis of the Ribs
( calculB t-l0 113 for r! b x c 21.)
llie ribs are acted upon ~y the s~ ar loads N
,xl' per unit length,
acting 81 ong ~e connectl~ng 111?e oft·he rlb and the ahe}l, posit'l y~'
.
as shown In F'ig. 80 *** The .distance of the lroads NX,Tfrom the cer:ltlrot.d .>0 '

of the effective section is called zA. Statically speaking, the

rib is ah'el1iaticallj restrained'arch, haVing two redundants only,
due to symmetry"·df the structure and. 'loed wi th respect to the center
'line. In Fig ~ 8 half of Q'l e rib is shown wi th horizontal thru'fft ,H-'c '
and the bending moment Mc 8S redundant's' at the ce'ntero The tllQ
, . *Nx.st> is not exactly equal to N~d.ue to the curvature of" tm shell. 'I
.. See. e.g. Reto (12) ,F:I/l, E1. ('
S'),.;-. .
v*Tbe procedure has some s1 milartty wi t~the '~Moment D-i,stributl on~" ,
procedure. There, the joints' ere lock.ed by inisglnary moments
wh ich afterwards are ellminat'ed l by introducing tllem in ,the OPPCi>-
site directiono '
***Positive NXdf@rce:s are ecting in tl?-;,~ opposite sense. on the r£b.
x f: 0; see ;TFig~ 6 for 'the sign convention of.the Nx~ forceso
 213K 9

geometrical c ondi ti 0 ns fur ,n1 shing the two equa t ions far the

solution of Hc and Mc ere a known horizonta I movement J K of

the foundation end a known ro'tation ¢K. l<or a full,. restrained
rib the conditions are:

J (10)

In cese of an elastically restrained arch, with known measured

foundet ion movemen ts in a ddi tion, the c ondi ti ons are:

CP - (,p.
J - J
•
K •
JK ~ (JK)~eBS
, 'K S (¢K)meas -}:MK

(J X)mea 8 and (¢K)meas ere the mea aured horizontal d1 apIa cement
and rotation of the abutment respectively. Kbeing the coefficient
of e18s~ie
restraint, the product -tM , where M is the restraining
K K
moment, represents the elastic end rotation of the rib. *

In general, the shspe of the center line of the rib ani the

variation of the moment of inertia of the effective section require

I

a numerica I prJcedure for t he e aleulation of the two redundant s He

and Me. For the special CElse of a circular center line and e con-
stant I, the calculetion of Hand M by direct integration is ahown~
c c
Extension easilyma y be made to other cases, e.g. parabolic shape
of the rib and I = Ie , or to s numeri cs I solu ti on.
cos~

For the rlb shown in Fig. 8 the normal force No end the be!lding
moment 1.1 ' due to the sl'E-a r load Nx~' may be obtained by integrati~g
0
the contribution of t~ di stri buted shear loed" Ifris the angle at
which No end Mo are determined, let Nx~oadu.>represent the sh:la~

loe d applied over an infinitesimal rib length adW a t any angle t.J,.) 'f
between 0 and r:t. 'Then: (Fig. 8b)

*Further explanation for J: (under II, 3, )p. 12.

 r
2131\ 10

NQ ., - xr • co. ('1'-"" dW

Substituting. from Eqc (9) - Nxces -!-P1' and performing the in-
J 21.
tegrat10n

No "-: -'. a~~ i (1 - co B c.f) (12)

Similarly, the contri~ution

to the moment Mo of the incre-
mental shear load 1s Nx~ a'd,w multiplied by the ~oment arm ze -

8 fJ [1- cos (1- W ) ) , wh ere a e et e· ... ze 1s the rodius of the effect1 ve

section (F'1go8b) e. The integrated total of M
o then is,
, , "'0 • itt 0 [z. - .0 (1- co (i'_ullJ] d
0 t-V
Substituting as before for -Nx~and integrating, .

. Mo ., .2;iP~-, :: i? - 1+ co. yo] (13)

\ The total normal force N and the to ta 1 bending moment M are hence
(F'ig. 8e)
. N =- No - Hc co~ '! (14)

M =Mo + Hc8e(1-c~S f) of. Me (15)

In order to. calculate H

c and Mc the horizontal disP1ecementJK .
and the roteti ctl ~K of the sbut~ent must be calculated. Choosing
8 virtual loed system as shown in 'Fig. 8c, pt s l a t the abutm(l!lt
produces the following v.1rtuel normal force Nt and virtual bending
moment M t at .. an angle, in the rib:
p' ~ 1 : N' - - coe ~ (16)

e (cos '1- coe:fk)

M' • - ( 17)
8

By means of the 'Work equation the ~isP1ecementdK of the actual

system is expressed as the work d one by ·-the virtual .loed P' : 1 * :::

(
d k
~jM'EI
tMd~S +fN EA'Nds
Where: ds = Sed t:f
The influence of the normal force N on the displacement is extremely
small and therefore, disregarded. Assuming, EI is constant, end
·S~e eny textbook about Elastic Strain E Methods
213K 11 .

replacing M' and M by their values (17) and (15) respectively,

J[-
•
becomes: f
J;.: k "e {eo0'f - eO"'/kJ &0 t He".(1-eos1 lt Me J"od r
Replacing M~ by (13) and integrating gives

The virtual. moment M'k =1 of Fig. 8c pr oduces a bending moment M'

;;; 1 for any angler:

M' k -"" 1: M' =1 (19)

lnser ting in the work equa tion

"k" fill" , do : s.d'J'

the expression (19) and (15), ¢k becomes

~k .. -iT~[Mo t Hes. {I eo" 'f i f Me Js.d y

If Mo fs substituted by its value (15) the integration can be
perfo nned

¢k ~ = Mcfk ... Hcs e (fk - si~.( k) + ;~P

(18) and (20) are two equations with the 2 unknowns He and M •
c k dl
am ¢k are given by (10) or (11) respectively. Once the r·edundante
are known the normal force N and the bendi ng moment M for any angle ~

are given by (14) and (15).' The stresses in the ribs are determined
by the usual formula for combined direct force N and the bending
moment M
"..._ N i
u - ~ rMz
 213K 12

A and I are the cross sectional area snd the moment of inertia
of the effeotive section,- cmsisting of the rib ss s web and the

effective width of the shell as e flenge. z is the distance of a

fi ber from the c"ntroidel axis of the effective sect10n.
30) Coef! ~,cl ent of efs sti c restrs int ~ :

In Eil' (11) e coefficient of elastic restraint v..... wes introduced.

In arch bridges it takes i.nte account the elast:lcity of the solI end

the abutment. In general the arch can't be considered fully re-

strained at the springing l'1ne.
In the present case another consideration led to the intro-
duction of k 0 The section A - A of Fig. 10 gives e typical con-
•
structlon detail at the springing line of'rib end shell. The latter
is supportf'ld by a reletively flexible edge-member. The rib on the

other hend terminates in a very heavy end-wall and may be regarded

as fully restrained. Consider the two extreme cases where (1) the
shell 1s held rigidly along the edge-member or (2) has a fre~ edge o

In the first case the NCf - . forces of the shell will be taken by
the support and the effective width 1s constant down to the edge-
member (Fig. 9a). If there 1s no support at all, the effective
width must reduce to zero at thesprlnging line. (f'ig.9b). The

actua 1 cond1.ti on is somewhere in between. This di s turbence of

I

the effecti ve width of the shell due to edge-member acti on berng

of very loce 1 character, ,the reduct:' on of the moment of inerti a of
the effective section can be taken into account by considering the I
as constant down to the springing line and assuming ,en elastic re-
stra int for the rib. - The higher flexibility of the effective section
in the end zone is "concentrs ted" at the springing line ..
It 1s quite obvious that the magnitude of th~ coefficient of

ele'tlc restraint depends on the effectiveness of the edge-member

 213K 13

in supporti ng the she 110 No theoretical analysis has been made

of ( in the present invest1gati on.

4.) Oa lculs ti on of the She 11 Forces:

The stress along the connecting line of rib am shell

A
is .
(j'

calcula ted by means of Eq. (21), once th e ncr rna 1 force Nand. the

bending mcment M of the effective section are known. The total

circumferentia 1 dire9t force S of the she 11 is (Eq. (4) ) ~

S &" bh(\A

The 'problem consists in finding the forces ani moments in the

shell for a given S end a given set of boundary conditions. In

the present case the boundary condi.ticns for the shell at the ribs·
x .: 0 and x = 21, are:
x =~: M
X
e 0

X 15 0 S = given
Mx being the bending moment of the shell in axial direction per
unit width, the conditionM x • 0 presupposes the neglecting of, the

torsiona 1 stiffness, of the rib. The influence of the boundaries

x =2 (middle rib) on :the stress distribution can b'e disregarded,

the forces end moments being rapidly damped out with increasing

distance from the ribs x ~ <:) and x = 22. Actual calculs·tions

may be found in reference (18), 213-0 0

5.) Deflections of ' the rib:

Knowing the aree A and the moment of inertia I of the effective

secti on, the normal force N and the bending moment M for any engle l'
along the rib, the deflect! on of the rib is readily determined 'by

the work equat ion. Putting B dummy load P t =1 at th('J point for
which the deflection is (lestred, the work of this unit load due to

the deformations caused by the actual load system, is equal to the

r
deflection under cons ideration, or
d= MtMds
EI
"'fNtNgs
EA
2l3K l<t

where Nt Bnd M' are due to P' ~ 1, end Mend N due to the actual
load Sj"stem.
6.) Numerical Example:
The foregoing theory is now applied to the analytical solution
of a model of a shell roof construction. }ilig. 10 g1 ves the dimensions
of the model approximately 1/30 of that of the actual structure
(e.g. hangar at Rapid City, S.D.). Structural steel was chosen as
materiel for the model, the dimensions (shell thickness about 1/8")
leeving practically no other choice. The following list gives the
principal dimensions and properties:
List:
Outer ribs: Height r :: 2.113"
Thickness t e 0 0505"

Angle 'fk 81
0.5866 radians
Middle rib: Height r .:' 2.113"
(Used la ter on)
'I'hickness t :: 0.625"
Shell: Radius a = 108 ft

Thickness h - 00118"
Distance 1 l~ 12"
Load: '* p :: 1~0.62 Ib/in

Measured sUPPOrnt movements:

1,,03110- 2 in
( 1) Ri b x :: 2 x.. dk=
~ =-i .07.,10- 3 radians
-1.03.10- 2 in
(2) Rib x ::; 0
dk :

~ -- 1.07•• 10- 3 radians

Material constant (E-Modu1us): E 30:10 6 1b/in 2
Coefficient of elastic restraint: ~ --
Effective Section of the Outer Ribs:
First, the effective width of the shell is.calculeted. Fig.~

gives a chert of the effective width under~eral conditions. The

*See p. J:B:- and J;4igo ~ where the actual 10ed1 ng condition 10 described ..
1'1 S"
213K 15

coeff1 c1 ent
1 0 3161 s 0 .. 3687
1/ sh'
To know the number n, the stress distribution along the ri.b
"

in circumferential direction has to be known. Anticipating the'

f1 nal. results, Fig. 12 shows the stress in the. ribs. The stress
Vt (vA has the same variation) in the lower fiber has approximately,
tb. e fo I'm. of a cosine-funct:l. on, wi th a helf-wave length of ¢: 0035 .
to 0.40 radians o !he number n becomes

n = 1f_"~_~~_/::::-'
half-wave length

And the coefficient ~ls

A,:: n~ s 0 0 26

The two outer ribs have no overhang (length of overhang 2~ 0)

and the distance to tm middle rib (l::. 12 in) is sufficiently large

to be consider ed as infini tely long

(/2= 0 .. 3687x12 :: 4.42'72 0 4; see p. 6 )

Ent~rlng the chert of Fig. 4 one may read out the coeff:i.cie- nt K:

For:/! =0
1...: 0 0 26

It may be seen, that up to )..:: 0 0 75 the coefficient K and the

effective width stay practically constant.

The effective width b becomes
b =.K {ili: 1 0 36 in
 213K 16

The area A em the momen t of inertia I of the effect! ve

section are calcu18ted next. Fig. lIe gives all pertinent 1n-

format ion:
Area A =1.227 in 2

Distances z z
u - 10186
~
in

zL = 0.926 in
ze s 0.867 in
Radius of effective section 8
e = 108.926 in.

Momen t of Inertia I = 0 .. 536 in 4

A comparison to the moment of inertia of the rib only is

1n terest ing
Ratio I (effective section) ='0.536 g 1.35
I (rib on11) 0.397

It should be noted that the ahelllncreAsea the bending stiffness·

of the rib by 35%.

The sheer l.oads N ac~ing on the r1 be are determ1ned from
xr
Eq. (9)0

All values are now available to solve the two equations (18)

and (20) for the two redundants He and Me. The solution 1s made
for the e~sumpt1on of a rigid" restraint (condi ti ons (10) ) and an
elastic ,restraint with meesured displecement and rotation of the

fourrlation (conditions (11) ). Eventually the following values

are found
a 0) Rigid restra 1 nt: H
c = -1068 1b

hI
c ::
3156 + in - 1b

b.) Elastic restraint: H

c =
-1457 1b

Mc = 4371 + in - Ib

On the basis of' these redundants the direct stresses in the

ribs were wolked out and are presented in Fig. 12.
 213K 17

The calculation of the direct forces Ntf and the cross bending
moment Mx iI( the shell fo llowed the procedure briefly outlined on
po 13. 'Ihe r,esu It s are pre sen ted in Fig. 13.

7.) Experimental investigation and comparison

An extensive experimental study on the model, shown in Fig. 10

and 15 was made., 44 rosette (typ~ AR-l), 137 cross (AX-5) and 81
single (A-5) SR-4 electrical strain gages recorded strains, about
20 Ames Dial gages measured displacements (accuracy 0.001 in), and

two level bars chflcked rotations (accuracy 0.0002 rad.). The actual
load! ng co~s iated of 10 equally spaced horizontal single loads as

shown 1n Fig. Sb. By virtue of St. Venant t s principle the. differences

,
between the theoretical. loading, assumed uniformly distributed, end
the actual loading can cause only local differences in behevior o
I
I

The stresses in the ribs ccmputed from the strain gage 'readings
I

ere shown 'in Fig: 12. Note the agreement between test and theory
if the analysis is made with arbitrary allowance for the elastic

restraint of the rib. The introduction of t is fully justified.

In Fig. 13 the experimental N~ (direct force per unit width ~f

the shell in circumferential direction) and 1J1(bending mo~ent per

x
unit'width in axial direction) of the shell are plotted.
Tl;le correspondence between the ana lyticel ara theoreticsl

results is quite close. It is interesting to note that the stress

~- .
distribution is perfect1y~symmetric with respect to the middle ~1b,

The latter 1s unstressed, therefore contributes nothing to the

strength of the structure under the present type of a loading. The

rib x : 21has exactly the opposite stresses ('If· the rib x·: 0 0
This fact is also demonstrated in Fig. 14 where the measured vertical
deflections of both outer ribs are plotted. For the center <;r= 0
the celculs ted deflections are also shown.
 213K 18

80) Conclusions:

1. The analysis of the effect of horizontal lateral wind

loads on cylindrical shell roofs, developed in this chapter, was
fully confirmed by an experimental investigation on a model.
2. For actual shell roofs of the Rapid City type (span 3~0

ft.), the wind pressure an the front door (as prescribed by

building codes) is 30 to 35 lb/ft 2 , or approximately 1000 lb/ft
of the rib. ~his results in maximum rib stresses of about ~OO

Ib/in 2 • Obviously, a stress of this magnitude cannot be dis-

regarded as a secondary one. ~y combining a number of units these
stresses can be lowered considerably.

III Found a tion Yovemen ts:

10) The Problem:

The horizontal thrusts of modern long span shell. roofs are

large enough to off~r serious frundation problems. In certain
cases, tensien ties have been found necessary to balance the hori-
zontD1 thrust acting on the abutments. (Ref. (7'), p. 35). In any
case, it usuaLly will be impossible to prevent completely any founda-
tien movements, except by very special devices, as for example arti-
ficial stressing of tension ti~s between the abutments. The importance
~.

of an analysis for the foundation movements is therefore obvious.

The general procedure for the calculation of archeS subjected
to foundati on movenents may be used if the interaction, occurring
between the ribs end the shell, is taken into eccC'unt. This will
be dene as in the previous chapter by taking as crOSB section of
the rib the effective cross section formed by the rib and the
effecti ve width. of the sh ell.
2.) Analysis of the ribs:
 213K 19

A horizontal diSPlaCementdk end symmetrical rotations ¢k

,
of the two abutments are consider~d. Unequal rotaticns ¢k l-
end ~kr of the left, and the right abutments can be solved by

superimposing sYIDJ'!letricel rotations ¢k = i- (¢k'l.'" ¢kr) and ant1-

symmetrical rotations ¢k := ±t (¢k.7. -,>¢kr). The advantage of

.
this procedure is to reduce the numbe~ of redundan ts from three to
two.
The rib, shown in F'ig. 16a, has two redundenta, the hori.zontel

thrust I1t
and the em moment M • Introducing the virtual lQsd H'k
k
:: 1 (Fig. 16b) and the virtual moment M'k 1 (Fig. 16b), the hori- =
zontal disPlacementdk end the end rotatlcn ¢k respectively can be
calculeted by means of tl:e work equation.
The actusl normal force N and bending moment M are (see Fig.
16s) :
N ~
- Hc C08 f ( 22)

M :::' . Mk - f\l!I e (C08r - co 8stk) (23)

Due to H' .- I
1
(F lg. 16b)
k -
.N' :- - cos 'f (24)

,"p
-... -Se (cos 1- COSfk) ( 25)

Hence thed becomes

k
'\

""
Sk - M'Mds+
EI
W'Nds
FA

Assuming EI am EA comtsn t, inserting (22), (23) Bnd (204),

(25) a ni per forming. the integra tion

EI S1ork} . .f. Hk 8 8 [ 9l. <i .. co.ll:/'k}- 3/4 s1011't

2e~
of. I ( t di'. .. i s 1 n 2 <;f k)] . ( 26 )
AR:!!e ,/A
 2l3{ 20

Taking Mt k l: 1 an d
Nt =0 (27 )
Mt =- 1 (28)

the work equation provides the end rotation ¢k. Taking

into account the reductin.n of the effective width in the
edge-member zone by a coeffici ent of ela stic 'restraint Ie * ,
the external momen t Mt k= 1 of the virtual load system times
, the roteti on -~Mk due to elastic restraint in the actual
load system contribute to the external work:

¢k - ~Mk = ~MtMd8 +~NtNdS"

J EI j EA
Mak~ng the necessary substitutions by (22), (23) end
(27), (28) the integration gives eventuelly~

B. ¢k
ee
~Mk (':!k + i) + Hk8e (t.fk cOB)'k, - Sb~k) (29),

Where: K = g lC
ae

For any given values of Jk and ¢k' the two redundants

Hk and Mk are determined by Eq. (26) and (29)0 The direct
stresses in the ribs are calculated by use of Eq. (21).

3.) Celculat ion of the Shell Far ces:

The, procedure is the same as described on p. 13. For

the two outer ribs end the middle rib the corresp~nd~ng

•
S-Forces ere calcula ted by "Eq. (4), once the stress (jA a long
the conre cting line of rib am shell is known (Eq. (21) ).
Knowing the boundary conditiC'ns for the shil, which are
,

Rib x - 0 l ~x =0
Rib x : 22:),

Rib x =1. ~
OX
= 0,
*See p. 12 for further explanation.
 2l3K 21

and the S-Forces, the stress distribution in the.shell can be

cal cula ted. The pre sentation of this ana lysi s, ex~eeding the
scope of the present paper, may be found in reference· (15).

40) Numericel Example:

The model (}<'1g. 10) is an alyzed for a given horizontal

dis pIa cen;tent dk a1?-d, end rotation ¢k. The effective width of
the ribs must be ca1cula ted· first. In order to know the number

n, used in t~e chart for the effective width (F~g. 4) the

. /1
stress distribution slong the rib must be known. Fig. :Mfshows
that as a first approximation the variation of the stresSOL I

in the lower fiber of the middle and outer ribs is a cosine

,--
funotion with a half-wave length ~.: 0.75. Hence

n ::.

fJrk is a first approximation for n if foundation movements

are consider ed. (Note the di fference to n in t.."'le c,ase of

horizontal lateral loads, ,po 16.) Inserting the value of ~k -

005866

ra -... 0.14
n ~i

The coeffic lentIL for the middle and outer ribs ( / being

compu ted on p. 15) are

Middle rib IL: Oo3687x12 : 4.42) 2 4 0

Outer ribs:/l= 0.3687xO ::. 0

F'rom ~'igo 4 it is obvious that the values of K for (1)

k= 0,/2.: Qoin case of ~p.e middle rib, Bnd (2»)... = 0,/2:: 0

in CBse of the outer ribs are sufficiently accurate for 'the

det~mlnetion of the effective width:

('1) Middle rib: A. = 0 1 K ~ 1052
12 = Qt?
 ·2l3K 22

b :::: K {;b ~. 5.43 in

(2) Out er
rib:jlf: 0] K--
0 .. 38
k: 0
. b :: K' feh l:J 1.,36 in

The following values for the effective section ere

eventually derived (compare Fig. lIe and lIb)
Middle rib:
Area 1,,961 in 2

Distances Z u ~ s -1.382 in
. zL e O~73l in
zA = 0.672 in
Radius of. effect:1.ve section
4
Moment of inertia I :: 0.921 in
Ratio I (effective section): 0 920 :: 1084 0

I ( r1 b only) 0.491
Outer Ribs~ (as on p. 16)
Area 1022'7 in 2
Distances ·z ~a -1 .. 186 in
t(~ \ .
zt= 00926 in
zA §
00867 in
Radius of effective section 8
e - 108 .. 926 in
MornEll t of inertie I -= 0 536 in 4
0

Note especially the considerable increase (84%) of the

bending stiffness of the middle rib by the effective width of
the shell.
On the model the follow ing horiz tonal di apIa cement cfk
and end rotaticn ¢k were induced:

dk ~ =003300 in
'" c
~k 70'753xlO-4 ra di ans
Taking the coefficient of elastic restraint
2l3K 23

Ie:: 0005 8e
EI
and inserting in Eqo (26) end (29) the proper values, the

horizontal thrust Ii< and tile em moment M are calculated.:

k
Middle Rib: .:.Hk S -1636 lb

Mk • -1'7850 in-lb
Ou tar Rib: Hk :. -n46 Ib

~k -10350 in-lb
~

Norm"l force Nand bendin g momen t M for an erbi trary

angle <r.a1cng the rib ere gtv~n by Eq. (22). and (23·)0 The
:calculated rlb stresses .(Eq. (21) ) are plotted in Fig. 17

for. the middle am

outer ri bs.
t t
i
Fig.118 presents the n~rmal force in a circumferential
i f
i
direction,Ncj end ~e bed1n~ moment in axial direction M o·f
x
I
the shell, the calculati on qr' which wes briefly outlined in 111,3.

50) Exp'erimentsl investigation ani comparison

The same model (FigolO), as used for the test of lateral

horizontal loeds, WaS subjected to.a horizontal foundation dis-

placanent (l"'-g. 19)0 '!he measured displacement and em rotations

are given on p. 24.

The stresses in the ribs and the N~-force end Mx-mcment,

computed on the besis of the SR-4 strain gege reedings are
plotted in Fig. 17 and 18 respectively. Theory and test again "
"

compare very favorably. The discrepancy of A few experimental

Mx - values 1n Fig. 19 does not influence the overall egreemento

*Note th at /( depends on the .type of loeding, for the caee of

horizontal latera 1 loads a different value for K was used (po+5).
/y'
2l3K

6 6) Con c lu s ions:

1. By taking the effective width of the shell ss flange

of a T-s6ct10n the web of which 1s the rib (effective
. ,

sectlon), the, 81a1ys18 of the ribs for foundation move-

m~nts can be done by simple arch, theory. Once the
stresses in' the ri bs are known, the stress distrlbution
in the shell can be computed also. Exper1menta1 results
on the model 'support the proposed analysis.
2. Eq. (26) and (29) show immediately that for given
founds ti on moven'en ts Jk and ¢k' the mom en t' M
k
and the

horizontal thrust Hk are proportio~al to the moment of

inertia I of the effective section of the rib. Neglecting
the influence of 'the normal force on the deforma tion s,
,
which is completely, negligible one may write:
c (c:f) I
l
] (:sO)
M = c 2 (~) I
where: cl (r), c 2 (1) are functions depending on the angle
t:J only.,
And the stress (f' in the rib becomes (Eq. (21) )

CII
-r--
+ c2 I z
I
= -c l rI -+ c
2 z (31 )

Roughly, G""is proportional to the distance z of the

fi ber from the centroids1 axis of the cross section, for
the first term of the right side of (31) is alwsys rels-
~
tive1y smello The ,importance of these QeTJisU:errs. will
~7
become apparent in a later discussion (P4 =-) •
. '

IV SOME SPECIAL PRO BLDliS:

10) Unlfonn Tempereture Change 8m Shrinkage of the 'Concrete:

Suppose B shell roof undergoes a uniform temperature c~

If' the whole strueture would be suppor ted as as imple

\
 2l3K 25

beam, this tempera ture chenge- would cause an increase /J L

of the sp en t:·
6 1 c 't.6 tal.
Where: Oc :;; coefficient of thermal expansion.
Actually this increasef)L is impossible, for the abutments are
restnlined. Therefore th! original span L must be restored
by diminishing the new spen (1·... 41.) by ~Lo The letter is

exactly the hori zontal d1 sple cern en t Jk alr-eedy tree ted under'
III. Be replacing in Eqo (26) end (29).

](32)
the horizontal thrust Hk and the end moment Mk due to s temp-
erature r1 se L1 to are thus determined.

It is common practice to consider the effect of shrinkage

of the concrete as equivalent to a fall in'temperature of a
specifiedL1 to. Hence, the stress due to shrinkage may also be
determined by the seme procedure.

20 ) Differential Temperature Change Between Ribs end Shell:

Assume a tempera ture dlfferenc e between the outside and

inside air of a shell roof structure. ~be ribs, exposed mostly

to the outside air will have a different mean - temperature than

the shell. The difference in tempe,reture between the shell and

the ri b may be ~ to. ObViously stresses in the structure will be

produced.
An analysis of this load case result eel' in thermal stresses
up to 100 Ib/ln 2 for a temperatur~ difference of 100F in 8

hangar of the Rapid City type (3~0 ft. span)o

The actual temperature distribution will be somewhat

different from the assumed one, for the temperature will vary
 ",

213K 26

continuously. Temperature measurements on an actual structure

should be made to determine the variation.

3 .. ) Stabili ty of the Structure:

The thickness h of the shell is mainly governed by
8teb~lity considerat.lons. For a certain spacing l of the ribs,
a radius "8~ of the shell, and a given distributed load, a
. minimum thickness h of the shell is requiredo * The resulting
membrane stresses reBc~ only a fraction of the allowable con-
crete stresses in compresslon~ It may be of interest to point
out that the ratio of thickness to radius of modern shell roofs
is smaller than the corresponding re tio for an egg shell. This
fact illustrates the perfection to which these structures hE;\ve
I>

been developed.
The ribs, reinforcing the shell at regular intervals l ,
increase the buckling stiffness of the shell panel in between.
Furthermore, they are l~dispensible for carrying a one-sided
live load (snow), acting on the structure. To provide B

sufficient buckling sefety for the entire structure the ribs

must have a certain minimum stiffness. The buckling loed 1s
proportional to the moment of inertia of the effective sectien
. (rib plus effective width of the shell). Previous considerations
concerning the s tresses due .to volume change led to a stiffness
of the rib aa small ss p08s1bl~. ObViously the stiffness should
be kept at the minimum required for stability end bending strength
80 that thestr~sses due to volume change do not become too exceasive<

Consequently, the question of the fector of safety against

buckling of the ribs is of prime importance. A smaller factor of
* See Ref. (11) or (12).
 213K 27

safety will require less stiffness end therefore smeller stresses

will be set up by volume changes. Plestic flow of the concrete
'affects the geometrical shape of the structure, hence the forces;
in a dangerous manner, if the factor of safety i8 decreased too
much.. In addition, the secondary moments due to deformations
of the structure can be corr.elated to the fector of safety
against buckling (Ref. 13). The scope of the present i'nvesti-
gation do es not pe rmi t specific recommend~tions for this flictOr'
.
of safety but it is a very important consideration in designo

40) Location of the shell with respect to the rib:

Recently the idea of locating the shell in the middle of

the rib wes brought forward (Ref~ 14). Considering the stresses
due. to volumechang~ VV'hitney correctly concluded that the stiffness
of the ribs should be a minimum'. By locating 'the shell in the
middle of the rib the stiffness of the effective section is in-
. r~~~.
deed diminished to a great extent (see~~, 47%). But the same
decrease can be' obtained qy .taking a smeller rib, wi th the !!hell

e t one edge of the ri b.

To illustrate, consider the .fb llowing cross sections:
Section (1) section (3)

Area Ai = 1.,9608
Momol .1 1 =0 0 9204

Dis t. : zul = 1.,382 -:: 1.056 . Zu3 :'; 1 .. i36

Z'LI - 0.731 .. i.056 zlL3 0 .. 544 =
Sections (1) er-d (2) are identical, except for the lo~etion
I
\ ,
 213K 28

of the rib. PIecing the shell in the middle of the rib

reduces ·the moment of inertia by 47%. Section (3) has e
moment of inertia practically equal to section (2) due to
a reduction in the height of the rib. The normal force
and bending moment due to volume changes are, proportional
to the mom en t of inertia of the effecti ve section (compare
Eq. (26), (29) 'and (30) ). The fiber stresses on the other
hand are add1 tlonally proportl('nal to the distance Z ot
the fiber from the neutral axis (Eq. (31) )0 The ratio of
the corresponding stresses'of the two sections 1s hence
equal to the retio of the corresronding distances z:
Uppe r fiber: ; (fu3
. ertl2
-- - 1 0 136
10056
-
Lower fiber: dL3
<1"t2

Section (3) has in the upper fiber a stress 8% higher 'and

in the lower fiber a stless 48% lower than section (2). The -
objection that in sect~on (3) important cross bending stresses
(hx are set up can be re'jected, for in the most ext:r:eme case .0-::x
is 1.73 times the lower fiber stress of the rib*:
a 1073xO.52 = 0.90
T~ cross bending st~~ess ~ is stll1 10% smaller than 01.2
and furthermore, the cross bending is concentrated on a very
short 'length.
In summary, for section (3) with a 21% lighter rib than
section (2) the bending moment and normal force due to volume
changes are identical. Neglecting the influence of the normal
.
.
force the stresses of the se'ct:lon (3) are'" 8% and - 48% of the ~
ones of se c ti on (2). Cross bending stresses ~ere~- 10% end ~
~See Ref. (15), p. 42. {fJ t'{d Q;:iV ~
 213K 29

of loc al importance (u'x has the forl'r\ of a strongly damped

oscillaticn starting from the rib). 'Iherefore, frcm a
statical point of view section (2) does not h~ve any ed-
v'antage ove'rsect,ion (3) whatsoever'. ' Contrarily, 21% of the
dead weight of the rib can be saved by using section (3).
It must be concluded that it is not only ineffective but
wasteful -to place the shell at the cen ter of the web.

SUMMARY AND CONCLUSIONS:

I
Theoretical solutions fer some problems encountered in
designirig long span cylindric~l shell roofs were presented.
Tests on a steel model in the scale 1 to 30 of an actual )

structure built in reinforced concrete (hangar Rapid City,

S.D.') confirmed the theoretical results. Fossible objection
that the test results were obtained from a perfectly elastic
structure (only elastic strains were induced. in the model)
whereas the material (reinforced concrete) in an actual
,
structure exhibits quite different properties may be answered
ae follows:
It is cornman practice to calculate the forces and moments in
statically in~etermlnate concrete structures by a$sumlng the
concrete as a perfectly elastic material. The moment of
Inerti~ for a cross sectloh is taken for the uncracked concrete
section (eog. continuous T-Beam girders, flange in tension over-
the supports etc.). On the basis of the calculated moments and
,/

norma-I forces the stresses 1n the concrete and the reinfcrcing

steel are determined under the assumption that the concrete is
ineffective in tension (~racked tension zone). This msy be
d one by using ei ther the usual n-Theory or'the so-ce lIed' "11mi t
des ign" Thee ry ~
 2l3K 30

Therefore, the similitude between the theoretical direct

forces end bending moments of the steel model end the concrete
structure c~n be made (both me teria Is are a sSU1'ned to be. per-
fectlyelastic)o 'rhese values were checked by test resultaof
the mexiel, built of B nearly perfectly elastic material. Be-
tween the stresses in the model and in the actual structure no
such direct relations can be derived. But it is a well es-

tablished fact, proved by m~ny te~ts~ that reinforced concrete

structu'res, anelyz.ed 8S elastic structures and reinforced
eceordlng to the calculated moments a ndnormal forces, behave

essentially as predicted.
/" ~0;~~~ ~~~~:;~,.

Mariy tests not reported herein were made during the course
..
-
of the investigation but, space does not permit their lnclusiono ' , .
Similar procedures of analye1sl' based on the use of ·the "effective
",

widthttl) were found adeqt";s.to to check test results to a close approximation

1n all cases 0 It Vias fC1und adequQ te to roughly estimat e the value of'
IVn", since the effective Width for the proportions of structures tested
and type of stress varis,tion ind.uced by usual loads put "nft in">~ region
where variations caused little or no change in the effective width o
Other engineering uncertainties are of much greater magnitudeo In

other applications of'this report to d:J.ffsrent problems:» it might be

1'ound necessary to carr;y' out analyses by means of a ~'ourier Ser1es
representation of the stress var;ation along the junoture of the rib
and shello In such eaSElS effective widths and stresses could be determine

1'or e.ach of. the s 19n1ticant terms or the J:i'ourier Series, and the actual
stress at any point could be determined by superpositiono

------------
',.' ..
;: .
. , .
.~: :;.'~.: -.'t"''!'t.-., .;.
"

.''1.
~ .....,.,
~.:"~ . . ""_, ;._.-. __
~i.::.
"~
,,,.,
 ACl\NOWLEDGME'N T

The report present~ a part of the theoretical studies

made during the course of a two-year research program on
shell arch roofs carried out at Fritz Engineering Laboratory,
Department of Civil Engineering and Mechanic~, lehigh Unl~

versity, Bethlehem, Pennsylvania. Professor William J. E~ey

is Head of the Department end Director of the Laboratory.

Mr. Lynn S. Beedle ,Assistant to the Director, end Mr. Kenneth R.
I

Harpel, Laboratory Foreman, gave valuable assistance through-

out the investigation.
Roberts end Schaefer Company, Chice go, I IIi no! s, wa s
sponsor of the research program. Man y thanks are expressed
to Mr. JoE. Kalinka, Executive Vice-President 0 f the Compeny,
and to Mr. Robert Zaborowski, ~anager of the New York Office,
who was representing the Company, for their continued assis-
tance. The many suggestions rece1ved from ~r. A. Tedeeko,

. .
O. Gruenwald, W. A. Renner and P. Rongved, all of Roberts end
Schaefer Company, during several meetings are sincerely 8cknow-
ledged.
 LIS'l' OF" REFERENCES

(1) Love, A. E. H. "E1a~ticlty",4th. edition

Cambridge Univ. Press, 1927
(2) Me! san er, E. "Das, E1eBtlz~tBtsprob1em far
danne Scha1en"
Phys. Zeitschrlft, vol. 14,
1913, p. 343
,. d ~~ E1astlzltBt und Festigkeit
~ d:tI"rmer Scha1en"
Vierte1jahrsschrift der Natur-
forsch. Gesellschaft, Zarich,
vol. 60, 1915, p. 23
;vrV
( 3) Beue~fe1d ; See: "Hendbuch fur Eisenbeton-Bau,
, Bend VI, 4th edition
Wilhelm Ernst und Sohn, 1928,
po 269
(~) Flnsterwe1der, U. "Die querversteiften zylindrlschen
Scha1engew8lbe mit kreissegment-
~8rrnlgem Querschnltt" .
Ingenieur Archiv, vol. 4, 1933,
p. 430

(5) Dischinger, F. "D~e strenge Theorle der

kriiszy1lndersche1e und lhre
Anwendung und dle Z.D. Schelen"
Beton & Elsen, vo1.34, 1935.
(6) Mo1ke, E.C., Ke1lnka, J.E. "Principles of Concrete Shell
Dome Des ign 1I
ACr:Proceedings, vol. 9 ·1938,
. po 649

(7) Prentiss, L.W. "Thin Concrete Arch Roof Provides

3~0 Ft. Clear Span for Bomber
Hanger"
Civil Engineering, vol. 19, no.
2, .1949, p. 34

(8) Allen, J.E. "Construction of Long-Span Con-

crete Arch Han~ar at Lirrestone
Air F.orce Base
ACI Proceedings, vol. 21, 1950,
p. 405
(9) voKarman, Th "Di·e mit tragende Brei te"
Festschr1.ft Aug. F8ppls
Springer, Berlin, 1924, p. 114

(10) Raithel, w. "The Determination of the Effect-

ive Width of Wide-Flanged Beams"
Tech. Report Nd. 61, Ordnance
Research and Development Divi-
sion, 1949
 . List of References, (Continued)

. (11) Timoshenko, S. II Thea ry of Plates and Shells tI

McGraw-Hill, New Yak, 1940
(12) Fl~gge, w. . "Statik, und Dynanlik der Scha1en n
Springer, Berlin, 1934
(13) Dischinger; F. '~nter8uchungen aber die Knick-
sicherheit, Bauing, vol. 18,
1937, p. 487
"Elastlsche und p1astische Ver-
formungen Bauing, vol. 20, 1939
p.. 53

(14) Whitney, Oh. S. -IICQst of Long-Spen Concrete Sh(l!11

Roofs"
ACI 'roceedings, vol. 21, 1950,
p. '765

(15) ThiIr1iniann, B. "The Effective Width of Circular

Cylindrical Shells Reinforced
by Rib s " '.'
PhD-Dissertation, Lehigh Unlv.,
1950
(1'6) ThiIrlimann, B. "The Effective Width of .'Cireu1~r
Bereuter, R. and b~lndrical S~el1. Adjecen t" ~o A
Johnston, B.G. C!.~cumferentia1.Reinforcing Rib",
prtsented at First National Con-
~&r~~g~~_~li~1~~t~ Meche nics, 19~
- -'-~if~ll~;~:O;;;~~~i~:~l;r- -- \:
St1f'fen~d bJ Circular Fings. n !
ProceedJ.ngs Konlnk11jke Neder... :
landsche Al(adern1e Van l\'sten- i\
i ---'-' -..--- s_c_h_e._p_p_e~e~o2~v.!I'Io ~~. ~f'
[l~)~) 'ThHrlimann, B. ~'~ ....." .)
, ,Johnston, B.G. P..rogressReP9':r'ts to Roberts &
Schaefer Company of Research
. Proj ect .2'1;3; 1949-50
2l3A-"Stress Distribution snd
Effective Width Adjacent
to Stiffeners in Cylin-
drical Shells" (tests)
..
213~"Stress Distribution end
,Effective Width Adjacent
to Stiffeners in,Cylin-
drical She11~" (analysis)
2130- "Shell Arch Roof Model
under Simulated End
Wind Load"
r
l
I
) I

t1s t of Refetences (Continued)

\ -

213D-"Shell Arch Roof Model

under Deed Load and
Half Side Live Load"
213E-"Shell Arch Roof M('del
under Si~ulated Founda-
t ion Movements"

• 213F-"Shell Arch Rocf Model

SubJected to Two Ceaes
of e Concentrated Load"
213G-"Experlmental Test Results
of a Shell Arch Roof
Model under Different
~ypes of Loading"
1 •
 _._--- -Imaginary stress distribution
I Actual stress distribution
~ b I~ r-- ~ b ---: b = Effective width
----I I~----'

.I I I
I I: !! I I ,!,
I' : I I II;

'---- Flange (flat plate)

~---'Rib

T- Beam Fig. 2

Cylinder Ring

------ ---'--

- - -I-- - - -

f----------

- X ----110 f - I- - - X

Fig. 3
 1.40
;--In finitely long cylinde'r

I. 20' I-----~

.60 f3l=O.2

.40 f3t= 0 -t----+--:..----+-----==--:--=------t

Semi-infinite
.20 I

I. Effective width b = KM I
0 .....- -........- - -........- - - - - - - -.....- - - -
----.. A
0. .50 1.00 1.50 2.00 2.50
l

--f~-}~--:--I of..

(3 =
= n~

1]1
, I--I--l S = S"cos n<P
, Fig. 4
 0) Theoretical LoadH19
. p (lbjin)
t ttttttt ~ t t t tt~ ~!~ t tt t ~ tt t tt+t tttt

b) Actual loading in test (10 51f19(e Lou(h~· p)

,
t t t t ttl t t
I t t t t -
.

I ! ~
I I
-- :.....-..J

-Fig.5-
-Fig 6 -
 • • '1--

I .

J Forces Moments Fig. 7

i
I
 section

0= radius of the shell

Q
e -== Q +"ZA = rodl us of
effective rib section

c.) ~----
I /I
I / I

II /
°e ,---
I
M'= I I
I t "....\ . ) P~1
~--f-- --+---~
r<.; J i
-4--

I 'f rotahon c?H -.on - displacement c5"~

a) .r
Rigid shell support b) Free edge of shell

Shell Shell

IEffect,ve Effective
W'dt'h Width
R,b Rib

-Fig,g
 SHELL ROOF MODEL
.• - • - <-.

FRONT VIEW I" =lO" SHOWING MEASURED

DIMENSIONS.

ALL DIMENSIONS IN
INCHES

I I
, I
59.78 59.78

10~85 II. 35 11.72

C
3.75
, I I
:
I
I
I,
I I
I I TOP VIE W I" =lO"
SECTION C-C I" =4"
II I
I I I
I I I I
I : I I
' : _I
I __ 1_
0.505 --j r-0.625
·A -- r- 0.118 2.113
I I I,
./
I I -I-
I I I
,: : I
II I I
: I I
': : I

C
B
12.0 12.0·

2.0
j~O.625 ~1t-O.505
SECTION a-a
11
1"=4
0.20
1.0 0.1\8
~--r--r--

2.5 3.5

SECTION A-A
3.5 3.5
\.0

1.5 1.5
10.0
3.0 .~-~------=~-------+-------,.------.
12.0 12.0
I.. 3.0 '
..I
FIG.IO·
a.) Outer Ri b b.) Middle Rib

0.505"

Effective section of
the ri bs
 -10000 1/ ,,_- 1--/ I '-: --v/· \
-15 0 OOt--l~'----t---+---+---+--__+--t-\----+-1- - + t - - t - : - - - (~--I---I---+---+----f---I----+--+---+---'H"

-20000H----+--+-------+-------+....,--' Itt t i itt i t ;1--- - P=1687.5.lb

~ ~'=====~===========:::::::::::t=x ===================~_- -_-_}2t -+----+----+---t---j1f------H

Edge -member:.-r-~ ,
1-+-\-+---+---+--...----,--r------ 1
----4
1
~==I===,=F===;I;:::===;to=~-~d3!!::;====1 ~-I I
5000-~
~
( 0 - (0 _
en (\J
tow -100 _ ~_ ~-
~ 00
en -
r<)
en_en __ ~ -+--+-- ~
ro to to
-- en -
to
en -
ro
en -
-
0 0 _ 00_
(J) I'-
ro- I'- -
ro
(0 -
-
~ (0
I'-~
-
(/)
o ~,.,'
~ ~X q..
'
~I
~I
~~ (\1
I'
~.
'"
='
q
,
+-.
~
~ = I'- ~ ~,
~
.
.
~
",'
~.
o'
~• x
~ LQ

. 'MIDD~E RI~ L~U=crL=~ ~

a.
--- ":4l ' 4>
CJ) -5000
CJ)
FIBER- STRESSES IN THE RIBS
w
0: 30000~
~
CJ)

FIBER-STRESS

<fsMf>
AT UPPER
AND LOWER
...
EDGES OF
..
. RIBS.

FIG. 12
 12" ------------1

MIDDLE RIB

/ / /

<t :ep=O o

o ~~~-:------,----.,......v----:::::O--X-===+===+==./=+>-=--t'J:::::;~::::;;;;;;;:~I=X=~-X-,----JO
\ 0·
~--t---t---/-+V~--+---t----t--t---t---+--4-
-400f-\-\--+-V-. 4
/' f'/'
-800 X ~ y - -8
V~, 0 ... / '
"- -r----
-1200 I-------'--.. -~ -.-._-.-._---'-_---'-_---'-_---'-_----l-_-'-_ t (62) -
o
-12

o \. Co'. I. I: ch:: 0.0580

. 1_____ ....-_ r--- a
x

.. 400
\ /V
/"-t---+---t--'---J7-!'--t---t---t---t----+---+--1 -4
~\-t-I-7"
,//
)<~ y/
_.
- 8 00 I-/---j~--+---+---:.-"./~---+----+---+----+--_+__--+--+-~
-8
/ "......... ---../
- -- . c:

"-..c
-"
c:
-1200 I------'------L- -12

-
I
c:
.-
--e-
..c

z
o
~,,~~
~
Col. 2: ep:0.1739

0 0 _--
~--t--I
0

·1)( I Ix
0

I
0

x
a ~
.c

.....
z
(/) -400 ~--~-- FORCES Nep8 MOMENT Mx in SHELL 1-----1 -4 w
w ~

'"
0::
o
lL.
- 800 I---+---+---+--_+__
THEORETICAL Nep
- - - - THEORETICAL Mx -8
0
~

x EXPERIMENTAL Nep (,!)

....o z-
w
a::
a
800 r----+-o----+---+-----._+__
~,/
400 / ~
. ~.--
--- .............

I"-
...... --
o

-.....:
EXPERIMENTAL Mxt - - - 1 8
I
I 4
a
z
w
al

x
o v/. "'------ ..... - - -
1 1--_
a
Col: 3: ep = 02898
0
~ (-6.4) 16

I 12

800 I '" "-

e
400
/ " r--... '",,-
I '- 4
/ 0 ~ "-..... : r"
l '.. . . . r--- x
a ~ -- ~
a
Col. 4: 4> =0.4057 - ~
160 a " - - r -, I-.".--r-;_- '-.---,---.I----,-----.------,-I-----,-------.-- f(-11.5 )
.......... 16

~~I
I
0;-a-

1200 0 1'-',,: I
1
12
,
1
r- ' " '
80 a / I
0 "I :
8
/
/ iI '"' " ;'"
I "-
3: 40 A ! I I " "- 4
0 Y .~.

3: ." I x
i~
0.......... " 0
0
0 ll "'-, x
"TI
",
Z
::0
0 a Co I. 5:
.J..
'fI =0.4637
- )( r- a
-t ",
Gl (J)
3:
)(
1(-10.1)
0' Z
Z -e-
.. eft l>
:I: Z
ITI 0
r
.r
 Shell-;::=========i=X=========r;.]2L
Edge-member ,
o ad¢

I Theoretical Deflection
x=o~"--
I
at center (E =29.0 x 106 )
. ./" ~"
~/ ~
x=21.~_
/d""': ,.,.,._:..---'
~
. - - __ + __---- ~
/'
h- /:--"
~.
. -".l'-""'--

EXPERIMENTAL VERTICAL DEFLECTIONS OF THE OUTER RIBS FOR

LATERAL LOADS.

FIG. 14
 Actual 0Y5tern:

. a)
d K
= horizontal . dtspl~cement
1K :; end rotation

.-
.!

~. to •
 30000 '
x
,---- ------ -------- ---- - - --- - ----- ,---- -- - - --- - - - - - - ... - - -------

CU Theoretical stress in upper fiber

------.-------------- -

-,
1 ----- 1= - --J

25000 1-1------+------ ------;-t----------e--- --- GL Theoretical stress in lower fiber ---i------------- -1-- ----------~----_H
X Experimental stress in upper fiber} - x

20000
r\.
I ''\.. :~ f-----c-----1--r 0 Experimental stress in lower f'lber. Outer, rib x = 0" and middle rib --.v-
_ . __ _
'\ I: . 6 Experimentat stress in upper fiber} Outer rib x= 21,
15000 ~ Experimental stress in lowerfiber--rr-----j----+--~/'_t_--__t__l
- 1___ ~ i---if- - 0--_
1-1----- •

I ~__ J /
I --

10000 H---+------: - - - - ----f----I----t--~=--_t__--+__-_+_---+__-_t_-=~~-+_-_t__--+__--_t_-________:v--_+--_H

5000

a
--------··1
"" 'v///'-"--/

'~//--~~TER RIBS
I
t
i ------ " '/
-',~__/'-/---+---I--------j-

-5000

-10000
(/)
0-

c: -15000
w
w
Cf)
Cf) 30000 fR
0
• w I
0::
I- 25000
Cf)

20000

15000

10000

5000

-5000

-10000

-15000

-20000 ' ,Fig,I7

 ..

OUTER RIB MIDDLE RIB-----------.-v

12" -----------~.~

1500

1000 ~ x ~
20
k* --
'\
I--
-- /
-. :><v*./
0
0 1-8-_-' - - 0
500 " 10
~
/' 0 o

~
8
'II
o ""- *
I--. ~ I ''""~
0
o
t. ep=O \,
-500 -10
\0
0
1\ -20
\
\ -30
\
\
-40
"-
~
x
V- m
~- ---1 - _ - v
-. ~
/ JO
fTI
z
~
......... ~ t--_~ 0
// ~--- ~-o-

v/ 0
i'z t'-- ~ ill'....
'", 0
Z'
G>

3:
4> = 0.1739 " 0
'", , -10 . 3:
m
z
.;
° \ -20
CJ)

\
~
THEORETICAL 3:
\ )(
- - - - - THEORETICAL \
X -30
x
0
EXPERIMENTAL Nep
EXPERIMENTAL M x
-~
•
CT
........
~
,...- 10
-S'

~ --- -----_·c-_.- --- ~

-
...............
,....
v ...
- -- ---
....... ~
.......
1"-.
.
0
........ 0
4> = 0.2898 ~,
i',
, -10
" -20
3: '"TI
0 0
m
::0
3: (") 500 I J

_ ....
10
z
-t
CJ)

~
fTI
CJ)

0 --
lL-----
ep =0.4057
+0(5.53

)(
~ .1-- .- --- I- l- .- 3- .-
-- le;
-L
~
- ,, .
o
3:
)(
» -500 -10
z0
:z
CJ) 20
I /1
fTI
r /
r / 10
ep =0.4637
0 ",
+0 (7.20~
l-- x.
_"3: ~ l-o-
v/ " o
~ ---
"-
cO
(»
-500
V
..........

./
V
X
f--.-
-- 1-0-- '>---
Ix--r---
>I. -10

-1000 -20
J