You are on page 1of 8

Journal of Advanced Concrete Technology Vol. 2, No.

2, 249-256, June 2004 / Copyright 2004 Japan Concrete Institute 249



Design of Non-Prismatic RC Beams Using Strut-and-Tie Models
Kiang Hwee Tan
1

Received 20 October 2003, accepted March 2004
Abstract
This study deals with the application of the strut-and-tie models in the analysis and design of non-prismatic reinforced
concrete beams. Seven beams were designed, fabricated and tested to failure. Test results showed that the ultimate
loads exceeded the design loads for all beams. Non-prismatic beams with a recess through the web performed satisfac-
torily, compared to beams with equivalent transverse rectangular openings. For non-prismatic beams with a recess at
the bottom, an increase in the recess width resulted in a decrease in the stiffness and an increase in the beam deflection.
Non-prismatic beams with a recess in the compression zone performed better with regards to cracking but not deflection,
compared to beams with a recess in the tensile zone. Also, beams strengthened with carbon fibrereinforced polymer
(FRP) plates performed satisfactory with regard to strength; however, the deflection and crack widths increased rapidly
thereafter, leading to a sudden and non-ductile failure of the beam.


1. Introduction
There are many instances where beams can be made
non-prismatic in cross-section along its length. For
example, in modern buildings where utility ducts and
pipes are being accommodated below the floor beams in
the space above the false ceiling, the use of a
non-prismatic beam with a recess would allow these
ducts to pass through the beam, eliminating a significant
amount of dead space [Fig. 1(a)]. This would reduce the
height of each storey, leading to substantial savings in
the material and construction costs.
Similarly, non-prismatic beams could be appropri-
ately used as ground beams [Fig. 1(b)] in residential
up-grading projects, where existing utility pipes often
obstruct the construction of tie beams that connect the
newly constructed columns to existing ones. The use of
non-prismatic tie beams allows the construction to pro-
ceed without the need of relocating these pipes.
A non-prismatic beam in the form of a stepped beam
can also be applied to support a split-level floor [Fig.
1(c)]. This application is commonly found in theatres
and in private housing for aesthetic reasons.
Finally, in buildings that are being retrofitted, there
might be a need to add new service ducts, and this is
often a problem due to limited headroom. In this case,
a recess can be made in the existing beam to accommo-
date the new service ducts. These non-prismatic beams
could then be strengthened by externally bonded rein-
forcement such as fibre-reinforced polymer (FRP) rein-
forcement or steel plates [Fig. 1(d)].
The behaviour of non-prismatic beams is very differ-
ent from ordinary prismatic beams, and current codes do
not give specific provisions for the design of such
beams. Frequently, actual designs are based on rule of
thumb or are empirical in nature and are not adequately
backed by research findings. As a result, the design may
be too conservative in certain cases while in others,
critical issues may be overlooked.
In view of the above, the present study is carried out
primarily to evaluate the strut-and-tie method as a de-
sign tool for non-prismatic beams. The development
and application of strut-and-tie models for design and
detailing of structural concrete have been described in
detail by Schlaich et al. (1987), Schlaich and Schafer
(1991), Marti (1991), and more recently by ACI
Sub-Committee 445 (Reineck 2002). For the purpose
of the present study, a test programme was carried out
on seven non-prismatic beams. As a secondary objec-
tive, the behaviour of beams with a recess was com-
pared with those with a transverse rectangular opening
in the web, and the effect of recess size and location and

1
Associate Professor, Department of Civil Engineering,
National University of Singapore, Singapore
E-mail: cvetankh@nus.edu.sg
Existing
structure
Column
G.L.
(c)
(a)
(b)
(d)
Extension
Column
Fig. 1 Examples of Recess Beams: (a) Floor Beams; (b)
Ground Beam; (c) Stepped Beam; (d) Retrofitted Beam.
250 K. H. Tan / Journal of Advanced Concrete Technology Vol. 2, No. 2, 249-256, 2004
strengthening scheme, on the overall behaviour and
strength of non-prismatic beams was investigated.

2. Analytical considerations
A non-prismatic beam may be divided into regions
known as the beam regions or the B-regions, in which
beam theory, including linear strain distribution across
sections, applies, and other regions termed the disconti-
nuity regions or D-regions. Based on St. Venants prin-
ciple, the D-region extends from the location of discon-
tinuity for a distance equal to the member depth.
In general, the design of a non-prismatic beam begins
with the isolation of the critical D-regions, followed by
the construction of a strut-and-tie model that would
transfer the boundary forces through these D-regions.
The rest of the beam, that is the B-regions, can be de-
signed using the usual procedure for prismatic beams.
On the other hand, the beam could also be designed
using a strut-and-tie model for the entire beam, as
adopted in this study.

2.1 Formulation of strut-and-tie model
The formulation of the strut-and-tie models is based on
the following assumptions:
(1) Strut members are formed by concrete while tie
members are formed by steel or FRP reinforce-
ment.
(2) The strut members have a prismatic width. The
capacity of the strut members is equal to the effec-
tive compressive strength of concrete multiplied by
the available concrete area.
(3) The capacity of the tie members is equal to the
cross sectional area multiplied by the yield or effec-
tive strength of the reinforcement. All tie members
are provided with adequate anchorage and there is
no bond slip of reinforcement.
(4) The joints (nodal zones), where the strut and tie
members meet, do not fail at ultimate.

2.1.1 Design of strut members
The effective compressive strength of the concrete struts
is taken as:
f
ce
=
1

2
f
c
' (1)
where the product v
1
v
2
is an efficiency factor between 0
and 1. The factor v
1
depends on the cracked condition
of the strut member, and for the case where the strut is
cracked longitudinally and is not confined by transverse
reinforcement, v
1
is equal to 0.65 (MacGregor 1997).
The factor v
2
accounts for the increased brittleness of
concrete as the strength increases, and is given by

2
= 0.55 + 1.25/( f
c
')
1/2
(2)
where f
c
' is the cylinder compressive strength in MPa.
For the case of f
c
'= 30 MPa, v
2
is equal to 0.778; hence,
the value of f
ce
is about 0.5 f
c
', and was used in the de-
sign of all the strut members for the test beams in this
study.
The required width of the strut is
a

= F
s
/tf
ce
(3)
where F
s
is the force acting on the strut, and t is the
thickness of the member. All struts must fit within the
beam, and must also not overlap. Otherwise, additional
steel reinforcement would be needed to reinforce the
strut so that the required width could be reduced. Such
compression reinforcement must be laterally restrained
to prevent buckling before failure of the beam occurs.

2.1.2 Design of tie members
The required reinforcement for each tie member is cal-
culated from:
A
s
f
y
T
n
(4)
where T
n
is the calculated force for the tie member, A
s
is
the required area of reinforcement and f
y
is the yield
strength of the steel reinforcement or effective strength
of the FRP reinforcement. For carbon FRP plates, the
plate would debond before the full potential of its ten-
sile strength is reached. Based on a previous study
(Tan 2001), an effective strength of 0.45 times the rup-
ture strength may be used.

2.2 Deflection of beam under service load
The deflection under service load, assumed as the ulti-
mate load divided by a factor of 1.7, is calculated using
the conjugate beam method. To account for cracking, an
effective moment of inertia, I
e
, is

used, where
I
e
= I
cr
+(I
g
- I
cr
)(M
cr
/M
a
)
3
(5)
in which M
cr
and M
a
are the cracking moment and
maximum applied moment, respectively, I
g
is the gross
moment of inertia, and I
cr
is the cracked moment of in-
ertia of the section. The cracking moment for a section
is given by
M
cr
= (f
r
I
g
)/y (6)
where y is the distance from the neutral axis to the ex-
treme tensile fibre and f
r
is the modulus of rupture taken
as equal to 0.12(f
cu
)
0.7
(BS 8110, 1997), where f
cu
is the
cube compressive strength of concrete. In general, f
cu

may be taken as f
c

divided by a factor of 0.8.


The beam is divided into segments of constant
cross-sections. For each prismatic segment, the maxi-
mum moment is expressed in terms of applied load, P.
This is then substituted for M
cr
in Eq. (6) with the re-
spective I
g
and y values. The lowest value of P thus de-
rived is the cracking load for the beam.
The conjugate beam is loaded by distributed loads
given by M/EI
c
where I
c
corresponds to the respective
prismatic segment. The bending moment evaluated
from force equilibrium at any section of the conjugate
beam gives the deflection at that section of the actual
beam. The bending moments at critical sections, typi-
K. H. Tan / Journal of Advanced Concrete Technology Vol. 2, No. 2, 249-256, 2004 251
cally, at the high-moment end of recess and under the
load were thus calculated for the conjugate beam and
the higher value gives the maximum deflection.

3. Test programme
Seven beams with dimensions shown in Fig. 2 were
designed, fabricated and tested to failure. Beams ST-1,
ST-2 and ST-3 each had a recess in the tensile zone (that
is, at the bottom) of the beam with widths of 400 mm,
800 mm and 1200 mm respectively, with the center of
recess at 1,000 mm from one support. They were de-
signed to take a load of 204 kN, 132 kN and 90 kN,
respectively, applied at one-third span length from the
other support.
Beam ST-2T was designed with a recess in the com-
pression zone (that is, at the top) at also 1,000 mm from
one support to take a load of 132 kN, applied at
one-third span length from the other support. Beam
ST-4 had two recesses, one at the top and the other at
the bottom and ST-5 was designed as a stepped beam.
Both beams were designed for a total load of 150 kN, as
a point load at the mid-span for Beam ST-4, and as two
point loads, one each at the mid-length of the upper and
lower stepped regions, for Beam ST-5.
Beam ST-2R was meant to simulate a beam that was
strengthened with externally bonded carbon FRP plates
after a recess has been introduced. The geometry of
ST-2R was the same as that of beam ST-2. The internal
steel reinforcement of ST-2R was first designed assum-
ing a solid beam carrying a load of 89 kN. This rein-
forcement was curtailed at the faces of introduced recess
and was welded to steel plates that lined the recess.
The strengthened beam was designed with external car-
bon FRP plates to transfer the forces over the recess so
that the beam would carry the original design load of 89
kN.
The strut-and-tie-models for the test beams are pre-
sented in Figs. 3(a) to (d), with the solid lines indicating
the tie members and the dotted lines representing strut
members. The reinforcement was designed accord-
ingly to resist the forces in the tie members, derived
from force equilibrium at the nodes. Typical rein-
forcement layout is shown in Fig. 4. Nominal links
were provided in accordance with code requirements.

3.1 Materials
The concrete mix was designed for a 28-day cylinder
compressive strength of 30 MPa. Ordinary Portland
cement, natural sand and crushed granite of 10 mm
maximum size were mixed in the ratio of 1 : 1.32 : 1.98
by weight. The watercement ratio was 0.45 and the
cement content was 495 kg/m
3
. To increase workability,
a superplasticizer was added at a dosage of 0.3 kg per
100 kg of cement to give a slump of about 150 mm.
Mild steel bars designated R6 and R8, and high yield
deformed bars designated T10, T13, T16, T20 and T25,
were used as internal reinforcement. Tensile tests were
carried out on three specimens of each bar size and the
results are presented in Table 1.
The carbon FRP plates had a thickness of 1.2 mm and
width of 100 mm. They were stiff in the longitudinal
direction, having a high tensile modulus of 150 GPa, but
weak in the transverse direction. The plate has a low
density of 1.6 g/cm
3
. Properties of the carbon FRP
plates and the epoxy mortar are shown in Table 2.
The plates were cut to the required length and width
using an ordinary cutting blade. The concrete surface
was ground and cleaned of dust and loose particles be-
fore the plates were bonded to the beam using the epoxy
mortar.

3.2 Preparation of test beams
A wooden prismatic formwork was used. The recess
was formed by attaching rectangular boxes of the re-
quired dimensions into the formwork. The inner surface
of the formwork was oiled to facilitate demoulding.
Six 100 mm cubes were cast for each beam. The beam
C-C
Sections
A-A
200 mm
B-B
200 mm
200 mm
220mm
(140mm
for ST-5)
400mm
(500mm
for ST-5)
Fig. 2 Dimensions of Test Beams (all dimensions in mm).
220mm
(140mm
for ST-5)
400
150
ST-2T
P
1000 600 600 800
B
B
C
C
800 800 1000
150 3000
ST-1
ST-2, 2R
ST-3
A
A
B
B
B
B
ST-4
P
400 600 500 500 600 400
B
B
C
C
A
A
650
P/2
650 650 650 400
A B
ST-5
P/2
A
C
C B
P
252 K. H. Tan / Journal of Advanced Concrete Technology Vol. 2, No. 2, 249-256, 2004
and the cubes were stripped off the moulds after one day.
Then they were placed under damp hessian for about a
week before being left in the laboratory under ambient
conditions. The beams were also whitewashed for easy
identification of cracks.
For beam ST-2R, holes, 12 mm in diameter and 50
mm deep, were drilled through the beam and the carbon
FRP plates after bonding the latter. These holes were
vacuum cleaned to remove the dust and metal bolts
were inserted and subsequently tightened with nuts, to
fasten the carbon FRP plates. Washers were used to
cushion the applied pressure on the plates.

3.3 Test instrumentation and procedure
All test beams were simply supported over a span of 3.0
m and tested under a one-point load except Beam ST-5,
which was tested under two point loads. The strains in
the internal steel bars, external carbon FRP plates and
concrete surface were monitored using electrical resis-
(b) ST-5
2 T16
2 T16
2 T20
R8 links at
50 mm
R8 links at
200 mm
R8 links at
50 mm
2 T20 + 1T13
2 T20 + 1 T13
2 T16
2 T16 + 1 T10
2 T13
3
4
2
1
1,3
2,4
2 T20
2 T20
100
mm
70 mm
50
mm
R8 links at 70 mm
50 mm 2 T16 50 mm width
Steel
Plate
(c) ST-2R
Fig. 4 Typical Reinforcement Details of Test Beams.
2 T10
2 T16
2 T10
2 T25
+1 T10
2 T10
(a) ST-4
R8 links at 75 mm
2 T20
2 T20
2 T16
2 T16
2 T25
2 T20
2 T25
160
180
160
180
160
180
140
(a) ST-1
[similar for ST-2, 3, 2R (after introduction of recess)]
P/3 2P/3
P
400 1200
400
1000
(b) ST-2T
P/3
2P/3
P
1000 800 600 600
(c) ST-4
400 600
400
600 500 500
P/2 P/2
P
300
140
340
(d) ST-2R (Before introduction of recess)
P/2
P/2
40 650 650 650 650

P/2
300
P/3 2P/3
P
1000 2000

(e) ST-5
Fig. 3 Strut-and-Tie Models for Test Beams.
Table 1 Steel Properties.
Type
Mild Steel
(Round)
High Tensile Steel
(Deformed)
Designa-
tion
R6 R8 T10 T13 T16 T20 T25
Actual
Diameter
(mm)
6.0 8.0 9.7 12.8 15.3 19.6 24.1
Crosssec-
tion Area
(mm
2
)
28.3 50.3 78.5 129 183 302 454
Yield
strength
f
y
(MPa)
296 346 478 489 532 552 458
Ultimate
Strength
(MPa)
393 433 571 584 617 684 670
Youngs
Modulus
E (GPa)
188 160 159 187 165 184 186
Strain at
Yield,
y

( x 10
-3
mm/mm)
1.58 2.16 3.01 2.62 3.22 3.71 2.49

P/2
K. H. Tan / Journal of Advanced Concrete Technology Vol. 2, No. 2, 249-256, 2004 253
tance strain gauges. The vertical deflections of the test
beams were measured using linear variable displace-
ment transducers. Crack widths were measured using a
hand-held microscope having a accuracy of 0.02 mm.
The loads on the beam were applied in increments. At
every load increment of 1 kN or a deflection increment
of 1 mm, strain gauge readings and transducer dis-
placements were recorded. After the first crack has
occured, the development of cracks was marked and the
maximum crack widths were measured at every 20 kN
load increment. Finally, the ultimate load and the mode
of failure were noted.

4. Test results and discussion
The crack patterns shown in Fig. 5 for six of the beams,
corresponded well with the orientation of the tie mem-
bers in the strut-and-tie model, shown in Fig. 3. It was
also observed that the strains in the longitudinal steel
bars were very close to the predicted values, and the
measured strains and hence stresses followed closely the
general trend predicted by the proposed strut-and-tie
model.
A unique crack pattern was observed for beam ST-1
above the recess, indicating the presence of arch action
over the recess [Fig. 5(a)]. The first crack appeared as
a diagonal crack at the high moment end of the recess
for beams ST-3 [Fig. 5(b)], ST-4 [Fig. 5(d)] and ST-2R
[Fig. 5 (f)], but as a flexural crack at the recess for the
beams ST-2 and ST-2T [Fig. 5(c)]. As for the beam
ST5 [Fig. 5(e)], the first crack appeared as a diagonal
crack at the corner below the upper stepped portion of
the beam.
The cracking loads are indicated in Table 3. As the
beam surface of beam ST-2R was covered by carbon
FRP plates, the observation of cracks was only possible
at the later stages of loading when the cracks had
propagated above the plates. Therefore, the recorded
cracking load of beam ST-2R was much higher than the
other beams. According to BS 8110 (1997), the maxi-
mum crack width under service loads should not exceed
0.3 mm. It is clear from Table 3 that this is satisfied in
all beams except for beam ST-2, and in particular beam
ST-5.
The maximum deflection occurred at the middle of
the recess for beams ST-1, ST-2 and ST-3, at the high
moment end of the recess for beam ST-2R and ST-4 and
Table 2 Properties of CFRP System.
Carbon FRP Plates Epoxy mortar
Colour Black Colour Comp. A: White
Comp. B: Black
Comp. A + B:
Light grey
Tensile
strength
> 2400 MPa Mix ratio Comp. A : Comp.
B = 3 : 1
(by wt. or vol.)
Modulus of
elasticity
> 150000 MPa Static
E-modulus
12800 MPa
Breaking
strain
1.4 % Open time 30 minutes (at
35 C )
Apparent
density
1.6 g/cm
3
Density 1.77 kg/lit. (A +
B)
Temperature
resistance
> 500 C Coef. of
expansion
9 x 10
-5
per C
(-10 C to
+40 C)
Shelf life Unlimited (no
exposure to
direct sunlight)
Shelf life 1 year in original
packaging at
+5 C to 25 C



(a)
(c)
ST-2T
ST-1
(d)
ST-4
(f)
ST-2R
(e)
ST-5
(b)
ST-3
Fig. 5 Appearance of Test Beams After Failure.
(a) (b) (c)
(d) (e) (f)
254 K. H. Tan / Journal of Advanced Concrete Technology Vol. 2, No. 2, 249-256, 2004
at the mid-span for beam ST-5. Table 3 shows that the
maximum service load deflections for all the test beams
were less than span/350 or 20 mm (BS 8110 1997) ex-
cept for beam ST-2R.
Beam ST-2 failed in flexure at the section directly
under the applied load. Beams ST-2T, ST-3 and ST-4
failed in flexure with the crushing of concrete at the
high moment end of the recess whereas ST-1 failed by
concrete crushing at the top of the arch that developed
above the recess. ST-2R failed by debonding of the
carbon FRP plates from the concrete surface and ST-5
failed by steel yielding at the connection between the
upper and the lower stepped regions.

4.1 Comparison of test results with design val-
ues
The test results are compared with the design or pre-
dicted values in Table 3. The predicted cracking loads
are in general less than the test values, except for beam
ST-1. This is because the first cracks were diagonal
cracks originating from the corners of recess whereas
the prediction assumed a flexural crack. The predicted
service load deflections are also less than the observed
values, indicating that further refinement of the calcula-
tion method is necessary.
Excluding ST-5, the ratio of the observed to design
ultimate loads varies from 1.07 to 1.69, with an average
of 1.29 and standard deviation of 0.21. The main reason
for the higher observed ultimate strength is that the
strut-and-tie method gives a lower bound solution, al-
though the higher concrete strength compared to the
design strength of 30 MPa might have some effect.
Beam ST-5 did not satisfy both the ultimate strength
and serviceability requirements, due to premature fail-
ure. This was the result of a detailing error for the di-
agonal bars, which had not been sufficiently anchored.
Table 3 Test Results.
f
c


(MPa)
P
cr, test
(kN)
P
cr, pred
(kN)

s,test

(mm)

s,pred

(mm)

max,s

(mm)
P
u,test

(kN)
P
u,design
(kN)
P
u,test
/
P
u,design

ST-1 51.9 7.5 12.7 8.2 4.6 0.15 259 204 1.27
ST-2 42.5 28.0 10.2 5.6 5.0 0.35 220 132 1.67
ST-2T 36.5 15.0 7.8 7.5 5.4 0.14 156 132 1.18
ST-2R 39.2 50.0 6.6 7.0 8.4 0.09 95 89 1.07
ST-3 48.0 9.7 8.0 5.7 5.0 0.14 152 90 1.69
ST-4 42.3 17.3 8.5 8.3 3.8 0.14 188 150 1.25
ST-5 37.0 12.6 9.6 17.0 9.1 0.76 133 150 0.89
Note: f
c
= concrete cylinder compressive strength;
P
cr
= cracking load; P
u
= ultimate load;

s
= maximum deflection under service load;

max,s
= maximum crack width under service load.
300
50
100
0
200
10 20 30 40 50
Service Load
R1
ST-1
250
150
Deflection (mm)
L
o
a
d

(
k
N
)

Maximum Crack Width
0
50
100
150
200
250
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Service
Load
ST-1
R1
L
o
a
d

(
k
N
)

0
20
40
60
80
100
120
140
0.2 0.4 0.6 0.8 1.0 1.2 1.4
R5
ST-3
Service Load
(a) Deflection Characteristics



0
50
100
150
200
10 20 30 40 50 60 70 80
Service Load
ST-2
R3
250
Deflection (mm)
0
10 20 30 40 50 60
Deflection (mm)
40
60
80
120
140
100
160
20
Service Load
ST-3
R5
ST-2
0
50
100
200
250
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Service Load
R3 150
Maximum Crack Width
(b) Cracking Characteristics
Maximum Crack Width
Fig. 6 Beams with Recess vs. Beams with Opening.
K. H. Tan / Journal of Advanced Concrete Technology Vol. 2, No. 2, 249-256, 2004 255
That is, the diagonal bars indicated by 2-2 and 3-3 in
Fig. 4(b) should have been extended to the bottom and
top edges of the beam, respectively, so as to effectively
control cracking at the re-entrant corners of the beam.

4.2 Comparison of beam performance
The beams are first compared with beams with an
opening in place of the recess. Next, the effect of re-
cess width is investigated using the results of ST1, ST-2
and ST-3. The effect of recess location across the beam
depth is examined using the results of ST-2 and ST-2T.
Finally, beam ST-2R is compared with beam ST-2 to
investigate the effect of strengthening.

4.2.1 Recess vesus web opening
Beams ST-1, ST-2 and ST-3, are compared to Beams R1,
R3 and R5, respectively, which were tested by Mansur
et al. (1985). The latter beams had the same overall
cross-section dimensions, beam span and were designed
to carry the same ultimate load under the same test
set-up as ST-1, ST-2 and ST-3. The only difference
between the two groups of beams is that instead of a
recess, R1, R3 and R5 each had an opening through the
web at mid-depth, having the same dimensions and lo-
cation along the beam as the recess in ST-1, 2, and ST-3
respectively.
As shown in Figs. 6(a), the load-deflection character-
istics of beams ST-1, ST-2 and ST-3 are similar to those
of beam R1, R3 and R5 respectively. The maximum
service load deflections are similar for each pair of
beams. Figs. 6(b) show the load versus maximum
crack width relations for the beams. Beams with a re-
cess had smaller crack widths at service load and hence,
more desired cracking characteristics than beams with
an opening.
All beams with a recess exhibited a ductile failure.
Beams R3 and R5 both failed with the crushing of con-
crete on the top and bottom faces of the chord members
at the high and low moment ends of the opening respec-
tively, while Beam R1 failed at the solid section under
the applied load. As beams R1, R3 and R5 were tested
using a load-control actuator, the post-peak behaviour
could not be obtained.
It is concluded that the provision of recesses offers an
alternative solution to openings, and such beams per-
form satisfactorily with respect to deflection, cracking
and ultimate load behaviour.

4.2.2 Effect of recess width
The load-deflection characteristics of Beams ST-1, ST-2
and ST-3, with recess widths of 400 mm, 800 mm and
1200 mm, respectively, are compared in Fig. 7(a). All
three beams have recesses at the same location at the
bottom of the beam. The maximum deflection oc-
curred at the middle of the recess for ST1 and ST-3, and
under the load for beam ST-2. The service load deflec-
tion decreases with an increase in the recess width.
The load versus maximum crack width curves are
compared in Fig 7(b). The maximum crack widths of
Beams ST-1, ST-2 and ST-3 under the service load are
0.14 mm, 0.35 mm and 0.15 mm respectively. The
larger value for Beam ST-2 is probably due to the failure
occurring under the applied load whereas it occurred at
the high moment end of the recess in ST-1 and ST-3.
50
100
150
200
250
0 20 40 60 80
Deflection (mm)
ST-2R
ST-2
(e)
0
50
100
150
200
250
300
10 20 30 40 50 60 70 80
ST-3
ST-1
ST-2
(a)
Deflection (mm)
L
o
a
d

(
k
N
)

L
o
a
d

(
k
N
)

0
50
100
150
200
250
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
ST-2
ST-1
ST-3
(b)
Maximum crack width
0
50
100
150
200
250
0.2 0.4 0.6
0.8 1.0 1.2 1.4
ST-2R
ST-2
Maximum crack width
(f)
0
50
100
150
200
10 20 30 40 50 60 70 80
ST-2
ST-2T
250
(c)
Deflection (mm)
Service Load
0
50
100
150
200
250
0.2 0.4 0.6 0.8 1.0 1.2 1.4
Service Load
ST-2T
ST-2
(d)
Maximum crack width
Fig. 7 Effect of Recess Width and Location, and Beam Strengthening.
256 K. H. Tan / Journal of Advanced Concrete Technology Vol. 2, No. 2, 249-256, 2004

4.2.3 Effect of recess location
Beams ST-2 and ST-2T had the same recess width of
800 mm. The recess is located in the tensile zone for
Beam ST-2 but in the compression zone for Beam ST-2T.
The deflection characteristics of Beams ST-2 and ST-2T
are compared in Fig. 7(c). The deflection at the service
load for Beam ST-2T was larger than for ST-2, with
values of 7.50 mm and 5.61 mm respectively.
The load versus maximum crack width curves are
compared in Fig. 7(d). The maximum crack widths
under the service load are 0.14 mm and 0.35 mm for
Beams ST-2T and ST-2 respectively. Thus, the provi-
sion of the recess in the compression zone offers better
cracking characteristics but less desired deflection
characteristics, compared to recess in the tensile zone.

4.2.4 Effect of strengthening
Beams ST-2 and ST-2R had exactly the same dimen-
sions. Beam ST-2R was designed as a truncated beam
that was subsequently strengthened with externally
bonded carbon FRP plates. Beam ST-2, on the other
hand, had been designed from the beginning to accom-
modate the recess and was reinforced by internal steel
reinforcement bars only. However, the design loads
were different, being 132 kN and 89 kN respectively for
beams ST-2 and ST-2R. The deflection characteristics
are compared in Fig. 7(e), which gives the service load
deflections as 7.0 mm and 5.6 mm for ST-2 and ST-2R
respectively. The load versus maximum crack width
relations shown in Fig. 7(f), give the crack widths at the
service loads as 0.35 mm and 0.09 mm, respectively, for
beams ST-2 and ST-2R. Beam ST-2R failed in a sudden
manner, losing its load carrying capacity once the
debonding of carbon FRP plates occurred. On the other
hand, beam ST-2 failed in a ductile manner.

5. Conclusion
Strut-and-tie models were presented to design the rein-
forcement for non-prismatic reinforced concrete beams.
Seven beams designed using these models were fabri-
cated and tested. Recess width and location, and
strengthening scheme, were considered. The test re-
sults were compared with design values and the effects
of these parameters on the strength and behaviour of
beams were discussed in detail. Also, the performance
of the beams was compared to similar beams with web
openings.
The following conclusions may be drawn from the
investigations carried out:
1. The strut-and-tie method of design was shown to be
suitable for application in non- prismatic beams as (a)
the crack pattern and measured strains in the rein-
forcement agreed with the strut-and-tie model; (b) the
strut-and-tie model gives lower bound values for the
ultimate load; and (c) the method offers a simple and
straightforward solution that is based on established
principles to an otherwise complicated problem.
2. Non-prismatic beams with a recess exhibit compara-
ble performance to beams with a transverse rectangu-
lar opening with respect to deflection and cracking
characteristics, and ultimate load behaviour.
3. For non-prismatic beams with a recess in the tensile
zone, an increase in the recess width results in smaller
ultimate load, higher cracking load and smaller ser-
vice load deflection.
4. Beams with a recess introduced and subsequently
strengthened with carbon FRP plates performed sat-
isfactory with regard to strength, deflection and crack
width. However, the failure tends to be non-ductile
and sudden.

References
British Standards Institution (1997). Structural Use of
Concrete. BS 8110, London.
MacGregor, J. G. (1997). Reinforced Concrete
Mechanics and Design. 3
rd
Ed., Upper Saddle River,
N.J.: Prentice Hall.
Mansur, M. A., Tan, K. H. and Lee, S. L. (1985).
Design method for reinforced concrete beams with
large openings. ACI Journal, USA, 82 (4),
517-524.
Marti, P. (1991). Dimensioning and Detailing.
IABSE Colloquium on Structural Concrete, Stuttgart,
411-443.
Reineck, K.-H. (Ed.). (2002). Examples for the
design of structural concrete with strut-and-tie
models. ACI Special Publication SP-208,
American Concrete Institute, 244 pp.
Schlaich, J., Schafer, K. and Jennewein, M. (1987).
Toward a consistent design of structural concrete.
PCI Journal, 32 (3), 74-150.
Schlaich, J. and Schafer, K. (1991). Design and
detailing of structural concrete using strut-and-tie
model. The Structural Engineer, UK, 69 (6), 13 pp.
Tan, K. H. (2001). Shear strengthening of dapped-end
beams using FRP systems. Fifth International
Symposium on FRP for Reinforced Concrete
Structures (FRPRCS-5), Cambridge, UK, July 16-18,
Vol. 1, 249-258.

You might also like