SP 596
By Bruno ThUrlimann
93
                   ~
1. Introduction
Some 70 years ago W. Ritter and E. Morsch developed design formulas for the
shear reinforcement of concrete beams assuming a truss model with 4S 0 compression
diagonals. About 30 years later E. Rausch extended this approach to beams sub
jected to torsion. During the past 25 years extensive experimental research has
been conducted in many countries. Today we are faced with results of tests
exceeding very likely a few thousand. Nevertheless, we have still not developed
a generally accepted approach to the design of reinforced and prestressed concrete
beams subjected to shear (e.g. (1) ). Dne reason is that an interpretation of all
the empirical results offers almost insurmountable difficulties. Some difficulties
are due to the complexity of the problem. Many, however, are caused by insuffi
ciencies in the planning of the test programs (no systematic separation of the
variables), in the design of the test specimens (improper details, unrealistic
proportions between shear and longitudinal reinforcement, etc. l or in the incom
plete reporting of the results. It is hence not surprising that many different
approaches have been developed leading to considerable differences in the design.
They can be classified into three categories:
1. Empirical methods
2, Semiempirical methods
3. Methods based on physical models.
'
The rules of the CEBRecommendations, 1970, and also the "Standard Method" of
the new Model Code, 1978 [2] fall under 1. or 2. depending on more or lessar
bitrary physical interpretation of the different terms in the design formulas
(e.g. shear resistance of the compression zone, inclined compression chords,
dowel effect, etc. [3], [5]).
94
ed critical tests and if necessary improved. The "Refined Procedure" of the
1978 CEBModel Code [2) (also [4) l is based on such an approach.
In the following a generalized space truss model will be described which allows
. ) . d . t . [ 9]
such a unified treatment for shear, c1rculatory tors1on an warp1ng ors1on ,
[10), [17). Fig. 1 shows a box girder subjected to'a normal farce N, bending
moments M and M , shear forces V and V and a torsional moment T. At ultimate
y z y z
load a system of inclined cracks develops in the walls. The static system of a
side wall (Fig. 2) consists of two stringers acting as compression or tension
chords, stirrups as posts and a continuous compression field made up of the
concrete compression diagonals at an angle a. In the following such an element
is called a shear wall. It can be a side wall of a box girder, the web of an
T
~
' '
' ........
Shear Wall I
I
fz
95
&
"
Sheor
Woll
Stringers  Chords
4. At ultimate load, i.e. after initial elastic and inelastic displacements and
the redistribution of the internal forces have taken place, the following
flow rules for reinforced concrete hold: Uniaxial yielding of the reinforcing
bars; opening of the final cracks perpendicular to the crack direction.
5. The detailing of the reinforcement is such that no local failures are possible
[i.e. the compression diagonals are properly anchored"by the stringers and
stirrups, etc.).
6. Against a premature failure of the concrete an upper limit for the concrete
stresses is introduced (e.g. max. compressive stress, max. nominal shear
stress). In addition, limits on the inclination a of the concrete compression
field and, hence, on the redistribution of the internal forces are given.
96
3. Strength of a Shear Wall
The strength of a shear wall will be determined using the lower bound (static)
theorem. On an element (Fig. 3) the forces M, N and V are acting. The truss
forces are the upper and lower stringer forces Fu and F , the stirrup forces S
1
and the resultant diagonal farce O of the concrete compression field, a , under
e
the variable inclination a. Equilibrium between the actions and truss forces
furnishes the following relations:
Diagonal farce O:
V
o =
sina (1 )
. F
upper s t r1nger: u = 2N  hM + V
2cota (3)
.
1ower s t r1nger: F = N MV+ cota
+  (4)
1 2 h 2
Stirrup farce S:
0
t
s = vtana
h
( 5)
Truss Forces
Actions
Fu  
f ~:;
ovvo j).
V
Fl Gc, '''' I I
~ hco~~ ~
97
;;}i.
I
' f
For an underreinforced section the following plasticity conditions hold:
I I
upper stringer: F < F
lower stringer: F
u  yu
= A o
u yu
< F = A o ( 6)
I I
f
1  yl 1 yl I
I
stirrup : s < sy = A o
S yS
Fl F
yl (7)
s sy iW'
Introduced into Eqs. (4) and (5) M and V will reach the ultimate values M
p
and V :
p
M V
P + _E. cota F (8)
h 2 yl
t
V tana = S (9)
p h y
s .h i
_Y_ J
':Jith: tana ( 1 o) ~
V t
p l
~
tq. la l gives i.
~
3
~
M v2 t
1 p
_E_ +   F
h 2 s .h
y
yl i
Far convenience the equations will be nondimensionalized using the following
reference values.
M F h ( 11 )
pO yl
98
The f inal interaction equation takes the form:
M V
_P_ + ( p
M
pO
vl
pO
2
( 13)
0.82 3
00..n~1<=
V , /,
:::.near
0.6 1<=112 }
' Eq.(21)
I( =3
I
\
0.4 \
\
r K > 3
0.2
Mp
Mp0
The inclination a, given in Eq. (10), has to be limited as the following kine
matic considerations will show. Fig. 5 illustrates the plastic strain increments
of a shear wall element at the onset of the plastic flow. The perpendicular
crac~ openings described by the crack parameter ER as defined in the figure
/
/
/
/
[F~, l ,j . ,.,,,.. (,,
  ;. 
/
,:t
/
/
/
coso. t,. ,;.. I ,
V .,,. ,, : 'C fJ..,(.
/
/
/
\;'
coso.
(, .
~
\
/
/
L\ l : 'I:.
!
/
  l . ,,.,..S~ 1 " M e __ , 1.
(5
,;.,.,.:;..
'" i' 
I
t:L cotono. _____ 1 ,, .) .
r coton o. /,
6 ,/,
('& 99
L
te
e:s e: cot 2 a I
L
( 14)
e:R = e:S + e:L
The total strains at ultimate load depend on the entire loading process covering
the uncracked and cracked state including the redistribution of forces and cracks.
The above relations can nevertheless be used to get an estimate of the strains.
To reach the yield strain e: of the reinforcement the following values of the
y
crack parameter e:R are required:
The relationship can be discussed with the aid of Fig. 6. Far an inclination of
the compression diagonals a = 45 the crack parameter e:R and, hence, the crack
(R
Ey
I on,.. ot Yilding I
10 in Stringer in Stirrup
6
5
4
0 I V;" + ,. 11 o.
r:J' 15 0 45 w 75 90"
  ~
05 5 tono. 5 2.0
~
Fig.6: Crack Width ( (R) vs. Y1eld Strain 1n Re1nforcement '!'
~
~
100 f
l
I
width become a minimum for yielding of both the longitudinal and stirrup rein
forcement. A smaller angle a requires asymptotically increasing crack openings
and stirrup strains to obtain yielding of the longitudinal reinforcement. Far
o
angles a > 45 yielding of the stirrups demands larger and larger strains in
the other two elements. Hence, it becomes obvious that a can only vary between
certain limits. From test results the limits
0.5 < tana < 2 (17)
have been deduced (10), [17). The following considerations may lead to a physical
understanding of the problem. For small crack openings the aggregate interlock
remains intact allowing on the one side only a perpendicular crack opening. On
the other side a shear transfer in the originally formed cracks above the crack
ing load remains possible. Such a transfer is necessary to obtain a redistribu
tion of the forces up to the formation of the failure mechanism. If, however,
large cracks develop the aggregate interlock disintegrates. The conditions are
illustrateci in Fig. 7. At the lower limit tana = 1/2 the crack strain and the
stirrup strain are equal to 5 and 4 times the longitudinal strain, respectively.
Hence, it can be expected that yielding of the stirrups alone with a crack strain
approaching ES will lead to a shear failure without yielding of the longitudinal
steel. The apposite holds for tana = 2. Here the stirrups will no longer yield
and a bending failure will result.
~ f7
Mf.s ~"ER
()
VEL
ton cx 112 2
Lonq. Stroin EL I Ey 4
101
.3 Interaction Diagram
The Interaction Diagram of Fig. 4 can now be discussed. Eq. (13) is limited
by the condition (17)
An upper limit for the shear farce V follows from Eq. (10) with tana = 1/2
p
V = 2. s .~ ( 18)
p y t
or in nondimensional form
h
V 2s 
_e__
V
pO
=
I
y t
h'
 2 F S 
={< ( 19)
yl y t
M 1 V
p
+  p
4F=1
MpO yl
~ ~
1
with V = F
pO yl ~K
M V
__e_ + f~~ (I= 1 ( 21 )
11
pO pO ~K
In Fig. 4 the limits for different values of the ratio K are shown. For K > 2
which holds far most practical cases the limit from Eq. (19) and, hence,
tana = 1/2
will always lead to a maximum value V < V whereas the limit from
p pO
Eq. (21) has no noticeable influence on the maximum moment M. The three types
p
of failure mechanisms, shear, combined bending  shear. and bending, are indicateci
in the figure far the case K = 3.
It should be noted that the values tana = 1/2 and tana = 2 are not exact bounda
ries but should be regarded as regions where a rapid transition from one type
of mechanism to another takes place.
102
'f
t,~
i
3.4 Introduction of concentrated Laads
I I I fI f 1
:y/7
I
: "
I I
(~  . \ihi
/:~111
2~  1.ll"
1 tR=4Sy JIlL ____ JI
L _____ Rt
i
!~Syh
St1rrups: Y1eld1ng S =Sy
Spac1ng I = h/2
Crocks : toncx = 1/2
I
V= 4 Sy J
F1g 8 Concentroted Loads Compress1on Fans
// ,,. I  l > M 1
I / /
i:a= ton 1 (1!2) ~ ' Ft =vcorano.~2
j; = 4 Sy
2h
x
i
'.... ...
4 ::::::1 ?sy ... . ... ./  hM .. 21 V colon o.
",._ M
81 '<: h
12Syi
i
1 .... JI
'. ,,;r4S,
Fl f Lower Stringer
103
I)
+J: h/2.1
_,' /t/';7'I
,,. ,..
.,,,,."' / /
/I
I

Fu =
M
h = 20 S)
_,.,,. ,,.,,. ,,, ,. I I
/~'!t~,,&l f:i,
h
"' ,,,. ,,. ,,,. // I
y _,. ,,, S' .

M
I F1 =h =20 Sy
4 3 2 16 Fu
~Zhi
x'
12
~ 16
M
h 20Sy
M 1 h
h+'2" Vcotonc:x
Lower Stringer f F
1
It may appear that a design based on a truss model with a variable inclination a.
of the compression field could become tao complicated. However, the apposite is
the case. In design the bending reinforcement A (Ml is determined first
1
M
A (M) = hc1 l (22)
1 y
with h equal to the distance between the tension and compressian resultants or
approximately the distance between the longitudinal reinforcements enclosed by
the stirrups.
Then the shear reinforcement A (Vl is calculated. In many cases the limiting
5
value tana = 1/2 will be chosen as it will lead ta the least total cast far bath
the longitudinal and stirrup reinforcement (g). From Eq. (10):
Vt
AS(V) tana. (23)
ho
yS
I
strength. Crack control under working load may require under extreme conditions
some additional shear reinforcement.
So far only complete truss action has been considered requiring diagonal crack
ing of the web. If, however, the nominal shear stresses
V
T (25)
bh
in the web are small at ultimate load, no or very limited diagonal cracking will
take place. Hence, the web provides an additional continuously diminishing shear
resistance in the transition range between the uncracked state and the fully de
veloped truss action. Theoretical studies far a qualitative explanation of this
behavior have been made (e.g. reference [14), 3.3 and Fig. 12).
i.e. the difference between the yield stresses of the prestressing and the ordi
nary steel. The two oEdiagrams have been superimposed to reflect this differ
ence. After decompression of the concrete the strains of both reinforcements
s ~6'
N/mm2 Nlmm 2 ~
1500
'
I
I
I .e u.?'Y
600
'G'y
I
I .
1 Ord1nory Reinforcemenr
  I
10
[
0
/o
l
1000 EY .5 1.'5
Oecomp<m,oo of Concrete
105
I
~
1
increase simultaneously and reach yielding at the same time. T~e strains and
cracks in the concrete will be the sane as far an ordinarily reinforced beam
with a yield farce equal to the yield farce of both reinforcements (e.g. [16) ).
Therefore, all previously derived equations hold if the yield farce of the j
combined reinforcement is introduced. i
Far small nominal shear stresses the transition range between the uncracked
state and full truss action will, however, be affected by prestressing. Accord
ingly an increased additional resistance of the web concrete over an extended
transition range will result (e.g. [4] ). l
4. CEBSpecifications far Shear
f
4.2 Members with Web Reinforcement i
r
I
Two procedures are distinguished. The "Standard Method" is a semiempirical f
method based essentially on test results. The "Refined Method" is theoretically
founded on the plastic analysis of the truss model described in this paper. It
has been checked and calibrateci with test results. It is applied to more
I
~
special cases like main girders in buildings and bridges and in particular to
cases involving combined bending, torsion and shear.
For the sake of clarity only the essentials are given in the following leaving
out all additional considerations such as influence of normal forces, variable
depth of section, prestressing, inclined cables, etc. {
In both methods the design shear farce V must be equal to or less than the
(
Sd f
resistance VRd made up of the resistance Vwd carried by truss action (inclined f
~
concrete struts and shear reinforcementl and the resistance V attributed to
cd
the shear resistance of the concrete compression zone and secondary effects
V V (26)
VSd ~ VRd3 wd
+
cd
1. Standard Method:
A
V = ~0.9df (1+ cotalsina ( 27l
wd1 s ywd
i
V = 2 . 5 T Rd b wd ' TRd  C\
..J.
75f
~ ctd (28) ~
cd1 i
The symbols are defined in Fig. 12. Attention should be given to the fact that
106 l
'
... s
I
\/ Sd
j
_l __
,
_iJ I
A sw
A l
~~
lncl1no11on c;f St1rrups (). (In Poper 90 l
lncl1no11on of Compression F1eld 8 (In Poper O<
they differ from the symbols used in the paper. The design stress f for the
ywd
web reinforcement is the characteristic value of the yield stress divided by the
appropriate resistance safety factor. The design concrete tensile stress f
ctd
is equal to the characteristic value of the tensile strength divided by the
appropriate resistance safety factor.
On the other hand the shear resistance reaches an upper limit controlled by
crushing of the concrete compression diagonals
V = 0.30f b d [29)
Rd2 cd w
with f the design concrete stress, i.e. the characteristic value of the con
cd
crete strength divided by the appropriate resistance safety factor.
107
O 9 fyw<l
Asw O 30 fcd . b.,. s
100
Refe rene es :
~~
~("
o [ 7] [ 8 /.;:..'
./,<o ~
V [11] , [12) / ,, ~
/, .;;.I~
o [13) .,''o'
. ,,
'i/
o [15) / V
/ Dg~
/ ~~OPJ V
/ ~'JoJo v "7
'J% Vsd 0.30 fcd. b.,. d
100
2. Refined Method:
A
V
wd2
= ~0.9df
s ywd
(cot8 + cotalsina (30)
o
The equation follows directly from Eq. [9) if the stirrups are vertical, a = 90 ,
The value of the inclinations of the concrete compression diagonals can be varied
between
3/5 < cot8 < 5/3 ( 31)
Compared to Eq. (17) the indicated values far cot8 have been chosen cautiously.
They assure a reasonable control of cracking under working load conditions.
The value of cot8 has a direct influence on the design of the longitudinal rein
forcement. According to Eq. (8) the bending reinforcement A CM l has to be
s 1 Sd
increased by an additional longitudinal reinforcement 6A rv dl due to the design
51 5
shear farce, i.e. the working shear farce multiplied by the appropriate load
safety f actor
V 2 s
Sd
Msl (VSdl (32)
2A f d f
sw ywd yld
108
i.
wit~, f the design stress of the longitudinal reinforcement, i.e. the charac
y 1d
teristic value of the yield stress divided by the appropriate resistance safety
factor. Eq. (32) follows from Eq. (24) and Eq. (10) using the appropriate CEB
notation. In the standard procedure the increase of the longitudinal reinforce
ment is indirectly made by the socalled "shiftrule".
The contribution Vcd varies linearly with the intensity of the nominal shear
2
stress in the transition range between no diagonal cracks and fully developed
truss action as discussed under 3.5.
Range V
cd2
where: T = 0.25f
Rd ctd
109
,
~
~
0.'Jf~
As .. 0.30\d. b., . s
1.00 4~
Un I Truss
crocked Tronsition
/
0.50
/ V'
V
~
'V
1
VSd 0.30 fcd'bw d
088 100
. \~'''
~o~~
Q20 \c.\
1
Vcd2 0.30fcd bwd
0.88 1.00
........... ~~~coro,..
.... ~.~
~ ~., n
CO ~
OA5
~
~C'OIJ S
.J References
''\ o [ 7] [ 8]
'Q (11). (12]
0.75i fyld o [13]
Asl 0.30fcd b,,d
0.90t~
o [15]
110
4.3 Comments on the Design Methods
Thc uscfulncss of a design method depends on its accuracy judged from comparison
with test results, its applicability to different problems, its ease of han
~ling dn~ last but not least on its scientific foundation.
The two CEBProcedures give very similar results for low shear stresses. At high
values the refined procedure can lead to a considerable reduction in the shear
reinforcement if a low value of tan8 is chosen. In the practical use there
should be no difference in ease of handling. However, the second approach has
the decided advantage of being compatible with the torsion design as both are
derived from the same truss model.
Finally, it is felt that the refined method has a sound theoretical foundation
being based on a clear truss model and the theory of plasticity. Despite of the
fact that simplifying assumptions are made reasonable correspondence with test
results is obtained. It shows that the theory of plasticity can be sucessfully
applied to determine the resistance of reinforced concrete sections to the com
oined action of shear and bending.
Sumc,dry
The "Refin~d Mctncj" of tne CEBModel Code, 1975, for the design of beams sub
jected to shear and bencin~ is bassd on such a model.
111
Notation
1. Used in Paper:
Fp prestressing farce
F farce in the upper stringer
u
F yield farce of the lower stringer
yl
F yield farce af the upper stringer
yu
h depth of shear wall, i.e. distance between the langitudinal reinforce
ments enclased by the stirrups; distance between the tension and com
pressian resultant causedby M
'
M bending moment
M plastic bending mament
p
M
pO
plastic bending mar.ent 1~ shear farce V o
N normal farce
P concentrateci laad
R reactian
S stirrup farce
S yield farce af a stirrup
y
T torsianal moment
t spacing af stirrups
V shGar farce
V plastic shear farce
p
V plastic shear farce if bending mament M = O
pO
a angle af inclinatian af the concrete compressian field
y shear strain
L langitudinal strain
R crack parameter ("mean crack strain"l
s stirrup strain
5
yield strain of reinfarcement
y
K ration between the yield forces af the lawer stringer and the stirrup,
Eq. (20)
112
o normal stress
o stress in prestressing steel
oc concrete stress of the compression field
a yield stress of the ordinary reinforcement
y
a yield stress of prestressing steel
y
a yield stress of the lower stringer or the yield stress of the
yl
longitudinal reinforcement
a yield stress of the stirrup reinforcement
yS
a yield stress of the upper stringer
yu
T nominal shear stress
2. CEBNotation
113
List cf ~ef Erences
[1] ASCEACI Tas~ Comn~ttee ~2E, "The S~ear Str8ng~~ of ~einforced Con
crete "'~err,te~s"; Jourrs~ =f the ::J:ructur31 [ ivi5icn,
1
A5C[, \'cl. 99,
No. ST6, June 1973 and .cl. 100, ~ic. STB, f;ugust 1974.
[2] Comit EuroInternatioral du Bton, "Systrne international de rgle
mentation technique unifie de structures"; Information Bulletin
~Jo. 124/125, Paris, Ai:;ril 1975.
115
,..
SP 597
By Bruno Thurlimann
117
ACI fellow Bruno Th~rlimann is professor of structural engineering at the
Swiss Federal Institute of Technology Zurich, Swit~erland. He is the author
of many technical papers especially on the application of plasticity to
steel and concrete structures. He is President of the International Associa
tion for Bridge and Structural Engineeering (IABSEJ. He serves as Chairman
of Commission V, 'Shear and Torsion', Comit EuroInternational du Bton
(CEB). He is a member of ACIASCE Committee 445 'Shear and Torsion'.
1. Introduction
The intensive research over the past 15 years into the torsional strength of
reinforced concrete memhers has lead to design procedures based on two ap
proaches, the "SkewBending Theory" [e.g. [1), (5), (10), [24)), and the
"Space Truss Model" (e.g. (3), [7], (13), (14], (24] J. In many cases they
lead to the same results.
However, it is felt that with the truss model certain scientific shortcomings
of the Skew Bending approach are avoided (19]. Furthermore, it is applicable
to any type of cross section. Finally, it allows a uniform treatment of other
loading cases such as shear and combined bending, shear and torsion [1B], [22].
Recently, it has been used to determine the warping resistance of thinwalled
reinforced concrete beams with open cross sections (9). It has also served to
determine the torsional stiffness of members after cracking (11], (23].
For these reasons the "Space Truss Model" has been adopted by CEB in arder to
derive appropriate design procedures [2].
2. Torsion e
The shear flow is a function of the torsional moment and the enclossd area
A , i.e. area enclosed by the perimeter connecting the longitudinal stringers
0
118
Stringer A1 , G'v 1 , F1 , Fv 1
(8) 7
:r Cr
1T
bz
Sz (2)
I: lbl
Diagonal farce O:
O= (Tb)s (2)
sina
o
o 
(Pb)
sina:osa
 T
( 3)
e bscosa b sinacosa
F3
t/o Se',,,,,,
(2) : s'
_l
'
3
...,_~ 1="~1
I I I ";
r1
s cotona b
2F = (T"b)scota ( 5)
Stirrup farce S:
Sscota (pb)s
t
s (Tb)ttana (6)
Stirrup forces S:
Tt ,
s = 2At. .. J. ( 8)
o
Concrete compresson stresses cr :
e
T
cr ~~~~~~~
2 A b sinacosa
( 9)
e
o
The state of stress descrbed by Eqs. (7), (8) and (9) s statcally admssible,
i.e. it fulfills all equilibrium equations. Assuming an underrenforced cross
section, i.e. yelding of the steel wll take place prior to crushing of the
concrete, R and S are limited by the yeld forces R of all strngers and S
y y
of the stirrups, respectively.
120
Resultant of stringers: R < R
 y
( 1 o)
Stirrup s <

sy Asoys J
The ultimate torsional resistance T is reached if both the stringers and
p
stirrups yield. Eqs. (7) and (8) give:
2A R
T = O Ytana
p u
( 11 )
2A S
T Q__zcota
p t
tana
T
p
=
*f?i
2A
o
y
t
( 12)
( 13)
In [20] limitations on the value of tana have been introduced and discussed
(further references [14], [22] J. Kinematic considerations concerning the rela
tionships between crack width and the strains in the stringers and stirrups
and also the rapid deterioration of the aggregate interlock in the concrete
with increasing crack width require such limits. From tests the following values
have been deduced:
1
 < tana < 2 (14)
2  
Case 1:
Fully syrrmetric section, i.e. all stringers have the same cross sectional
area AL and the same yield farce FyL' The stirrup reinforcernent A with yield
5
stress oyS is constant around the cross section, the corresponding yield farce
is S
y
121
Stringers: Areas A1 A AS A AL
3 7
Yield forces: Fy F F F F
1 y3 yS y7 yl
From Eqs. (12) and (13) with R 4F the following expressions are derived:
y yl
4F S
T 2A ~ ( 1 5)
p o ut
tana ( 16)
Case 2:
The section has a "bendingtype" reinforcement, i.e. the yield forces of the
upper stringers F are smaller than F of the lower stringers. Hence, the
yu y1
upper stringers will yield and the section will rotate about an axis 35 through
~he lower strins~r3.
F 4F ( 17l
y yu
The result can be generalized. If the stringers are irregularly placed around
the periMeter of a section [Fig. 3) the value of the resultant R ~inimum
y
governs the resi5tance. Taking an arbitrary axis BB through a side the resultant
R in the centroij
y
eu of the perimeter can be determined as follows:
l::F 7
i
R =~ ( 18)
y z
'.J
The axis 6B has to be varied to find the minimum of Ry. An axis 6B inter
secting the section as shown in Fig. 3 is not admissible because it would put
the stringer i and hence the concrete of that region into compression.
122
/B
 Fy,
of Perimeter
Perimeter u
s/
Fig. 3: Generai Cross Sect1on
Case 3:
With the shear flow (Tbl constant around the section the inclinations a and
2
a will differ in arder to meet this condition. Using relations (4) and [6)
4
together with the yield conditions the following expressions result:
123
Side 2:
1
nF nF  (T b) s cata
3 yl 2 2 2
s2
 = s (Tb)tana
t2 y 2
Side 4:
1
(1nJF (1nJFyl  (T"b) s cata
3 2 4 4
s4
S (T"b) tana
t4 y 4
n F
1
=  (T b) s (Tb)

yl 2 2 s
y
1 (Tb)
(1n)F = (T"b)s  
yl 2 4 s
y
The sum af these twa expressians equals the yield farce of the stringer 3:
2
1
1 (T"b) (s + s
F
yl 
2  2 4) ~=
sy ~.
The three treated cases can be extended to more general cross sectians such
that any canfiguration can be analyzed [22). The limits far tana given by Eq.(14)
hald far any one side. If they are reached no further redistributian of the in
ternal forces is possible and the circulatory torsional strength provided by a
constant shear flow (Tb) is reached. *)
124
However, an increase in warping resistance may ther. still be possible (9). In
many practical cases this additional resistance can be neglected.
s
\
z

T
d tJ
 s. CiX

i)
~ s di.!
2
Y
2"'cix
Kinematic relations furnish the following expression far the bending strain
on the surface of a wall with thickness b:
b dlJ
Eb ( lJ) = + (20)
 
2 dx

where is the angle of twist. The uniform compression a (T) of the concrete
1J
e
diagonals (Eq. (9)) is hence increased by bending stresses ob(lJ)
a = oc(Tl ob(1JJ ( 21 )
In tests maximum values of ob e1ual to two to three times the value of oc have
been observed [22]. Far convenience in design the influence of the Ob(lJ)
stresses can be indirectly considered by selecting an appropriate value for the
effective wall thickness b discussed in the following paragraph, i.e. a some
e
what smaller value such that the o stresses will become larger and include
e
approximately the wall bending contribution. In such a way it becomes possible
to introduce the same upper limits for the a stresses in tofsion as well as in
e
shear. Under combined loading the a stresses due to torsion and shear can then
e
be directly superimposed.
125
2.4 Solid Cross Sections
~
Tests of beams with solid cross sections (e.g. (15)) have shown that their
torsional strength is made up of the closed reinforcing cage, consisting of
the longitudinal stringers and the transverse stirrups, dnd an effective con
crete shell forming the necessary compression field under an inclination a.
On the basis of test results CEB [2) and others (4) take as effective wall
thickness b of the resisting concrete shell
e
do
b (22)
e 6
where d is the diameter of the circle inscribed into the perimeter connecting
0
the corner stringers of the section (fig. 5). For box sections the minimum \_
8
do
6 I I
/ ._~, . I
'~.....
bmin
._
'!

I
I !
~j
d0 ~
do b _ /bm1n
Sol1d Section be Box
6 e  "" do
6
In most applications torsion will be combined with bending and/or shear. First,
the simpler case torsion  bending will be treated.
3.1 Torsion  Bending
F l < Fy = A o
1 yl
s < s y = AS o yS
126
r
Fy
a  ~
I
" (81 Fyu
rr, ~i . t 7 
aJ,fJi: \2)
I ,,,
'' : (6) h
Far pure bending the plastic moment MpO will be reached if yielding of the
lower stringers starts. Hence,
M 2F h (23)
pO yl
The height h is the distance between the tension and compression resultants.
Far underreinforced sections h can be taken as the distance between the upper
and lower stringers enclosed by the stirrups.
In pure torsion the weaker upper stringers will yield first and the section
will then rotate about the lower axis 3  5. The resultant axial farce of all
stringers is
R = 4F yu
The case torsion  bending is considered next. Fig. 7 shows visually the super
position of the stringer forces F(Tl and F[MJ due to torsion and bending. It
is first assumed that the combination is such that yielding of the lower
tringers and the stirrups will take place at failure. The stirrups contr1bute
Tors1on Bending Tors1on . Bend1ng
, '
F(T) F(T1F\Mi
FTF(M)
127
only to the torsional resistance. One part of the lower stringers, 2nFyl'
resist bending, the other part, 2[1nlFyl' torsion. Hence,
M 2nF h
p r.:::.
y1
4[1nlF S
T 2A '\/ yl y
p o V ut
'
The term 4F of Eq. [24) is replaced by 4(1nlF as far each stringer only
yu yl
the part (1n!F is resisting torsion. Failure takes place by rotation a.Jout
yl "
the upper axis 1  7. If n is eliminateci and MpO and TpO from Eqs. (23) and [24)
are used as reference values the following interaction equation can be derived:
F T 2 M
Fl F ~( ___E_ ) + _P_ (25)
yl Fyl T M
pO pO
I
\'
Next the case of yielding of the upper stringers is considered. The tensile fo~ce
M
+;__p_lS
'
yu , h y
T
p
= 2A
O
l 
u t
T F M
p ~._P_ (26)
F F T
u yu F M
pO yu pO
Fli
2_. [ F  F l
2 yl yu
~
2.[F TF ) S
yl yu y (27)
Max.: T 2.A T + 1)
p max 0 u t pO
Far: M = [F  F )h [28)
p yl yu
128
To Ro!10 Fyl I Fyu 0 3
Tpo
Str 1nger Y1eld1ng
13 /
',,_ // D
o Stringer R191d
/
f2' I
o
~L+
1
3 3
1 Mp
Mp0
The section of Fig. 6 with the indicateci values for the yield forces of the
stringers and stirrups is now investigated far the general combination T, M, V.
~he shear flows due to torsion T and shear V are given in Fig. 9. By super
position the flows for the different sides are:
Side 2 (Tb) =  T + 
V
2 2A 2h
o
T
Sides 4 and 8 [Tb) = = [T"b) (29)
4 2A 8
0
T V
Side 6 [Tb) =   
6 2A 2h
o
129
f
I
I
\
Torsion Sheor
I'
lllllllilltllllllllll 1 1 I
t1 (8)71 f
o
1
~
7'
I} I
(8l
1111111111:ti111111111
T V
(Tb): 2AO (Tb)=2h
The corresponding static system is shown in Fig. 10. All sides will have differ
ent inclinations far the compression field such that yielding of two stringers
and the stirrups will occur. As a first failure mechanism it will be assumed
that the two upper stringers will not yield, hence, the section will rotate
about the axis 1  7. Taking moments about this axis furnishes
h h
M = 2F h  (Tbl hcota   (Tbl acota h  (pb) hcota  (30)
p yl 2 2 2 4 4 6 6 2
F3
Eq. (6) establishes the relation between the stirrup forces and inclination of
the compression field, a. If yielding occurs
S = lTblttana = S
y
130
(T"b) t
2 t T V
cota = = ( + )
2 sy S
y
2A
o 2h
(Tb) t
4 t T
cota = = 
4 sy S
y
2A
o
(T"b) t
6 t T V
cota = = (  )
6 sy S
y
2A
o 2h
Replacing the cota in Eq. (30) by the above expressions and making use of the
reference values ""
M 2F h (Plastic Moment)
pO yl
T
pO
2A
o ~ h
y (Plastic Torsional Momentl ( 31)
F3 FS F
yl
F
~. ~( T
_P_
2
+ ( __E )
V 2 J +
M
p
MpO (32)
Fyl TpO VpO
F =F = Fyu
1 u
and
F =F =F
3 1 yl
should be investigateci. In this case rotation will take place about a vertical
axis 5  7 through the right side, the moment which respect to this axis being
zero. In such a way the third interaction equation is found
F1 F
yu V F
p p ~
T 2
2TV 2
_P_ + 2h +

p 1 yl
) = ( + 1) ( 34)
T T V u V O 2 F
F3 F pO pO pO p yu
yl
131
The three interaction equations are graphically presented in Fig. 11 far the
o /. " . . . . . . X o
Tp0 11
o Stringer Rig1d
A
. ........... \
/ ......... . \
_,v,;__. .... \ \ Eq (32)
>\ \
\ \
\ \o
.
,.\'~
)
~
.
 ), I Y4J j
~
o DEQ(34)
VpQ
The interaction equations are usful far an understanding of the rather complex
interrelations between torsion  bending  shear and the different failure
mechanisms. However, for practical design a direct approach is much simpler.
The bending reinforcement A (Ml and A (M) is determined first (Fig. 6):
3 5
M
A (M) = A (M) = 2ho (35)
3 5 yl
Then the shear flows (Tbl due to torsion and shear are determined. The inclina
tion a of the compression field can be assumed within the limits given by Eq.(14).
1
Economical considerations are generally leading to the lower limit tana = 2
The additional longitudinal reinforcement due to torsion and shear follows from
Eq. (4) far the different sides:
132
1 T h V
Side 2: A CT,V) = A3 CT,Vl = ~( .A + lcota
1
yl
2
o 2
1 Ta
Side 4: A [T,V)
3
= A5 (T_,VJ =  .2Acota
2 0 yL o
(36)
1 T h V
Side 6: A [T,Vl = A (T,Vl = z:o( ~  )cota
5 7
yl o 2
1 Ta
Side 8: A (t,Vl =A [T,V) = cota
7 1 2o 2A
yl o
The reinforcements on the lowCTr side,i.e. tension side far bending, must al
ways be added. On the upper side, i.e. compression side far bending, the ten
sion farce due to shear and torsion is counteracted by the compression farce
due to bendi11g. Hence, the tensile reinforcement can be proportionately re
duced
M
A = LA [T,Vl  2ho
1 1 yl
(37)
M
A = ZA (T,Vl 
7 7 2 .ho yl
Introducing into Eq. (38) the shear flows of the different sides, Eq. (29),
the required stirrup reinforcement far each side is obtained.
(T' b)
o   [39j
e b sinacosa
The minimum value is obtained far a= 45. For the two limits, Eq. [14),
tana = 1/2 and tana = 2, the stresses increase by 25%. In a box section the
o stresses due to shear and torsion can be directly superimposec for each
e
side. TaKing the case of a solid section, however, far shear b is the width
of the web, whereas far torsion the effective wall thicKness b must be used.
' e
Oespite of the fact that the two states can not be directly superimposed it
is felt that a safe value far o (V, Tl is obtained if o (V) and o (T) are cal
e e e
culated separately, using the corresponding width b and b , and then added.
e
This value should not exceed a specified limit in arder to avoid a concrete
failure.
133
So far only complete truss action at high nominal shear stresses
1 T V
1: = ( :z;;;: + 2h ) (40)
b o
leading to cracking of the web and a redistribution of the internal forces has
been considered. However, in the transition range between the uncracked state
and full truss action the concrete of the web provides an additional continuous
ly dominishing rcsistance. :ue to the aggregate interlock a shear transfer in
the first formed cracks remains possible if the nominal shear stresses are rela
tively small.
The inf luence of prestressing on the shear resistance of a beam under combined
bending and shear has been discussed in [20] (and also [21) ). The same observa
tions hold if in addition torsion is acting. For high nominal shear stresses at
ultimate load a prestressed beam is acting like a reinforced beam as the strain
and crack pattern of the two beams will be similar. Hence, all previously derived
equations are applicable if the yield forces of the ordinary reinforcements are
replaced by the yield forces of the cornbined reinforcements.
The CEBApproach is based on the plastic analysis of the space truss model de
scribed in this paper. lo simplify the discussion only the essentials are pre
sented leaving out all acd~tional considerations such as prestressing, normal
forces, etc.
4 .1 Defin tions
The following definitions are introduced:
 Compatibility Torsion
 Equilibrium Torsion
 Circulatory Torsion
 Warping Torsion
134
Compatibility torsion is caused by deformations of adjacent members in statical
ly indeterminate structures. It produces secondary effects which should be con
sidered in evaluating the serviceability. It is however neglected in designs
based on ultimate strength.
The design torsional moment TSd must be equal to or less than the resistance
value TRd
TSd :_ TRd
An upper limit of the resistance is given by the strength of the concrete com
pression field:
Eq. (41) is based on Eqs. (1) and (3). The symbols are defined in Fig. 12.
Attention is given to the fact that they differ from the notation used previous
ly in the paper. The design concrete stress fcd is equal to the characteristic
value of the strength divided by the appropriate resistance safety factor. The
inclination G of the compression field can be chosen between
3/5 < cot8 < 5/3 (42)
Compared to Eq. (14) the indicateci values far cotO have been chosen cautiously.
They assure a reasonable control of cracking under working load conditions.
The other limit of the resistance is provided by the space truss action of the
reinforcing cage consisting of longitudinal stringers, closed stirrups and con
crete compression diagonals, i P

..eI J I,
U
a r1 ,..J




I I!~
 adui tiune:JJ r~~ '_J i :..:i t ::in ce of thL~ c1mcn::te,
T , in the transition range between the uncracked state and full truss action:
cd
T + T (43)
TRd2 ef,d cJ
135
het
Uet ( Perimeter)
Aef( Enclosed Areo)
As
with f d the design stress of the stirrup reinforcement, i.e. the characteristic
yw
value of the yield stress divided by the appropriate resistance safety factor.
li
Range Tcd t
l1
Uncracked : T < 5T A h 501 0 0
Sd  Rd ef E:f Rc Aef hef
1 ( 1 5 T A h  T )

Transition : S1 ~ h < T < 151 A h [ l
Rd ef ef Sd cd ef ef 2 Rd ef ef Sd
Truss .~ction: 151 A h < T o
Rd ef ef  Sd
where: 1 Rd '.J.25fctd
136
The design tensile concrete stress, f , is equal to the characteristic value
ctd
of the tensile strength divided by the appropriate resistance safety factor.
The results of the CEBdesign procedure are summarized in Fig. 13. In dimension
less form the stirrup reinforcement A , the total longitudinal 'reinforcement A
s s1
and the additional resistance T of the concrete in the transition range are
cd
shown as functions of the design torsional moment TSd and the inclination of
the concrete compression field, i.e. cot8. The influence of the design con
crete tensile stress f on T d is also shown. Comparison with Fig. 14 of
ctd c
reference [20) shows that the procedure is completely identical with the CEB
Refined Method for the shear desigr.
1.0()1~__,
4fywd
As f . h t' s
cd ,e
Uncrock.ed Transition
I
I
0.50
~~!
Tsd
0.50 fcd Aef hef
0.88 1.00
0.88 1.00
Tsd
\~. 0.50 fcd'Ae1hef
\ \
""co,,~&_, .j
\ CoQ.., 6l . ~
\('I
\o... "~
~6) ~
0.50 \''
\ \
\
\
A~
sl f .
\
1.00 cd he(Uef
\
Fig.13: CEBMethod:
2
St1rrup and Long Reinforcements vs Tors1onal MJment [1N/rnm =145 psi)
137
......
References:
[6] 1 [1ZJ, [15], [16], [17)
2.00 //0.7
A._ /
//
/V/
1.0
1.00
_._.1.5
_..,.....o
0.50
Tsd
0.5 focsAerhet
0.88 1.00
TSd
0.5fcdA.fhet
0.50
1.00
1.0
. ."""o
Selecled Tesi Poinls: ""
4fytd
"'l 0.71 < colon 8 < 0.82
Asi. fcd. hef. Uef
2.00 o 0.95 c:coton 8 < 0.99
o colon 8 = 1.5 " " 1.5
138
4.3 Combined Torsion  Shear
As discussed under 3.3 the reinforcements are determined separately for bend
ing, torsion and shear. In the tension zone caused by bending all longitudinal
reinforcements must be added. In the compression zone, however, the reinforce
ments can be reduced proportionally to the compression resultant of the bend
ing moment.
The stirrup reinforcements for shear and torsion are added. It should be noted
that the transition range is governed by the intensity of the combined shear
flow. In arder to simplify the design procedure the additional resistances of
the concrete, Vcd for shear and Tcd for torsion, are neglected.
Finally, the acting design torque and shear farce, TSd and VSd' must fulfil
the condition:
V = 0.30f b dsin20
Rd2 cd w
Only the case of beams with Ushaped cross sections consisting of three walls
is treated. The shear components in each wall due to a shear farce and/or a
torsional moment can be calculated from purely static considerations. The
stirrup reinforcement in each wall and the longitudinal reinforcement due to
bending, shear and torsion can then be designed. Far the strength analysis of
beams with arbitrary thinwalled open cross sections references [9] and [22]
may be consulted.
5. Summary
The same model is used in the CEB Shear Design. Hence, the general loading case
of combined bending  shear  torsion can be treated in a uniform and consist
ent manner.
139
Notati on
1) Used in Paper
A area enclosed by the perimeter u connecting the corner stringers
o of a cross section
2 l CEBNotation
VSd design shear farce, i.e. shear farce multiplied by load safety factor
142
(11] Hsu, T., "PostCracking Torsional Rigidity of Rcinforced Concrete
Sections"; Journal of the American Concrete Institute, Val. 70,
No. 5, May 1973.
(12] Hsu, T., "Torsion of Structural ConcreteBehavior of Reinforced
Concrete Rectangular Members"; Publication SP18, ACI, 1968.
[13] Lampert, P .. Collins, M.P., "Torsion, Bending and Confusion 
An Attempt to Establish the Facts"; Journal of the American Concrete
Institute, Val. 69, No. 8, August 197=.
143