Received: 30 May 2017 Revised: 11 December 2017 Accepted: 16 January 2018

DOI: 10.1002/suco.201700108

TECHNICAL PAPER

Experimental and numerical assessment of deflections in circular

reinforced concrete beams
Indu Geevar1 | Bijily Balakrishnan1 | Fayaz Habeeb2 | Devdas Menon1

1
Department of Civil Engineering, Indian Institute
of Technology Madras, Chennai, India Methods for calculating deflections (at service loads) prescribed in codes such as
2
WS Atkins, Bangalore, India EC2 and ACI 318 are largely based on semi-empirical methods validated for
Correspondence beams with rectangular cross sections. The applicability of these methods to
Devdas Menon, Department of Civil Engineering, beams with circular cross section is validated in this paper using experimental
Indian Institute of Technology Madras, Chennai
studies on eight circular beams with different grades of concrete and different per-
600036, India.
Email: [email protected] centages of reinforcing steel. Similar beams with square cross section have also
been studied for comparison and it is found that the code estimates have more-or-
less the same degree of conservatism for square and circular beams. The complete
load-deflection behavior generated by the test specimens could also be simulated
numerically using a layered approach, considering material nonlinearity in con-
crete and steel rebar. Using the proposed procedure, the influences of percentage
of reinforcing steel and axial compression on the load-deflection behavior have
also been studied. This study shows that the postcracking stiffness increases and
ductility decreases with increase percentage of reinforcing steel. The circular
beams are found to have less strength and ductility when compared to square
beams (with the same area of cross section) owing to differences in distribution of
steel rebar.

KEYWORDS

axial compression, circular, deflection, load-deflection, moment-curvature,

nonlinear, square

1 | INTRODUCTION been developed to predict the complete load-deflection

behavior of RC beams.1–3
Reinforced concrete (RC) members must not only have ade- Design code recommendations for the estimation of
quate margin of safety against collapse but should also short-term deflection at service loads are based on studies
exhibit satisfactory performance at service, in terms of carried out on RC rectangular members. The present study
acceptably low deflections and crack widths. In addition to explores the suitability of such simplified methods (based
short-term deflections (caused by service loads), long term on linear elastic analysis including the effect of cracking
deflection due to creep and shrinkage effects also need to be and tension stiffening)4,5 to RC circular members, which are
considered. These are difficult to predict accurately on commonly encountered in piles, bridge piers and columns
account of various uncertainties related to stiffness degrada- in buildings. It is necessary to estimate the lateral drift in
tion (due to cracking), material properties, environmental such elements in order to achieve deflection control and
effects, etc. Nevertheless, several numerical methods have estimate secondary moments due to P-Δ effects induced by
the axial compressive force.6 However, before exploring the
Discussion on this paper must be submitted within two months of the print
publication. The discussion will then be published in print, along with the effects of axial load on the lateral stiffness of circular beam-
authors' closure, if any, approximately nine months after the print publication. columns, it is necessary to first establish methods to assess

Structural Concrete. 2018;19:1633–1648. wileyonlinelibrary.com/journal/suco © 2018 fib. International Federation for Structural Concrete 1633
1634 GEEVAR ET AL.

lateral deflections in beams without axial load. This is 318 and EC2 overestimate the deflection prediction for per-
attempted in the present study, which explores the validity centages of steel higher than 0.8%, and showed variability
of using existing code-based procedures, originally formu- in deflection prediction for lower percentages of
lated for rectangular RC members, to circular RC members. steel (<0.8%).
This paper presents the results of experimental studies
on eight circular beams subjected to constant moment and 2.1 | ACI 318-14
loaded till failure. For comparison of existing formulations
of deflection calculation, six square beams were also tested. ACI 318 adopts the expression proposed by Branson14,15 to
The test specimens had different percentages of reinforcing compute the effective moment of inertia, Ie, for cracked RC
steel and grades of concrete. It is shown that short-term beams and one-way slabs, subject to a maximum moment,
beam deflections can be estimated at service load using ACI M (greater than the cracking moment, Mcr) and is given as
318-147 and EC28 formulations with reasonable conserva- below:
tism. A method to generate the complete load-deflection
n 
n 
Mcr Mcr
behavior based on layered approach, considering material Ie = Ig + 1 − Icr ð1Þ
M M
nonlinearity in concrete and steel rebar is developed and
validated with test results. The proposed procedure was where the exponent n was assigned a value of 3 by Branson
used to study the influences of percentage of reinforcing based on studies on simply supported beams subjected to
steel and axial compression on the load-deflection behavior. uniformly distributed load. Experimental studies have
shown considerable scatter in the value of the parameter n,
ranging from around 1 to 4.12,16,17 The actual value of n is
2 | CO M P U T A T I O N O F S HO R T - T E R M found to depend on the tensile reinforcement ratio, pt and
D E F L E CT IO N S AT S E R VI C E L OA DS the degree of flexural cracking induced along the length of
the beam by the applied loading. With an increase in pt,
Conventionally, deflection control in RC flexural members there will be an increase in Ie. Al Shaik and Al Zaid16 have
is ensured by limiting the span-to-depth ratio. This is proposed n = 3–0.8pt. Ashour12 observed an increase of Ie
approximate and is not strictly applicable for large spans with increase in grade of concrete and proposed a modifica-
and high intensity of loads.9 Therefore, design codes recom- tion of n for fc > 33 MPa to consider this. Al Zaid et al17
mend improved methods to explicitly compute deflections observed a 20% enhancement in Ie for beams subjected to
at service loads based on the linear elastic analysis. These central point loads as compared to beams subjected to uni-
methods are typically based on the “effective moment of formly distributed load at the same level of maximum
inertia” method (adopted in ACI 318-14) or by “effective moment. The value of n was suggested as 2.8 and 1.8 for
curvature” formulation (adopted in EC2). In “effective beams with uniformly distributed load and concentrated
moment of inertia” method, degradation of stiffness due to load at mid-span, respectively.17 However, for simplicity
cracking is considered by adopting an effective moment of ACI 318 recommends n = 3 for all cases. When M < Mcr,
inertia (Ie), which is a weighted average of the cracked Ie is taken as Ig.
moment of inertia (Icr) and the gross moment of inertia (Ig). The cracking moment, Mcr is given by
In the “effective curvature” method10,11, the deflections are
Ig
computed from an effective curvature (φe) which takes into Mcr = fr ð2Þ
y
account the extent of cracking in the member. The deflec-
tion (Δ) at service load for a beam subjected to constant where Ig is the moment of inertia of the gross section, y is
moment M can then be calculated as cML2/(EIe) or cφeL2, the distance from the centroid to the extreme tension fiber
where c is a constant which depends on the loading and of cross section and fr is the modulus of rupture (in MPa)
boundary conditions. which for normal strength concrete is given by
Several studies on short-term deflection have been car- pffiffiffiffi
fr = 0:62 fc ð3Þ
ried out on beams with rectangular cross section to validate
the applicability of these methods.12,13 Ashour12 conducted where fc is the specified cylinder strength of concrete
nine experiments on RC rectangular beam subject to two- in MPa.
point loading for different grades of concrete and percent- Branson’s equation14,15 is generally accepted to be valid
ages of steel. He proposed a correction factor for high- for predicting deflections in RC beams up to service loads
strength concrete in the expression for effective moment of and is not intended for load levels much above service load
inertia, Ie. Gribniak et al13 compared service load- where large strains in concrete and plastic deformation in
deflections of 80 rectangular beams subject to two-point steel rebar occur. To compute the short-term deflection,
loading using ACI 318 and EC2 and found that accuracy of effective moment of inertia, Ie and short-term modulus of
deflection prediction varies significantly with load levels elasticity of concrete, Ec are used in the expression for elas-
and percentage of reinforcement. It was observed that ACI tic deflection, where Ec (in MPa) is given by
GEEVAR ET AL. 1635

pffiffiffiffi
Ec = 4700 fc ð4Þ
where fc is the specified cylinder strength in MPa.

2.2 | Eurocode 2
In eurocode 2 (EC2), the deflection calculation is based on
FIGURE 1 Simply supported beam subject to constant moment
an effective deformation parameter, α, which can be chosen
as strain, curvature, rotation or deflection; it is given by also suggests a “simplified method” as an alternative, where
α = ð1 −ξÞαI + ξαII ð5Þ the deflection is chosen as the deformation parameter. Here,
the service load deflection (Δ) for short term loading is cal-
where αI is the deformation parameter in uncracked condi-
culated based on the equation below:
tion, αII is the deformation parameter in fully cracked condi-

2 "
#
tion and ξ = 1 − β(Mcr/M)2 for flexural members, which Mcr Mcr 2
indicates the level of cracking in the member. Here, β is Δ= Δg + 1− Δcr ð9Þ
M M
taken as 1.0 for short-term loading and 0.5 for sustained
loads or many cycles of repeated loading, M is the bending where Δg is the elastic deflection considering the whole
moment at service loads at the section considered (unlike in member to be uncracked and Δcr is the elastic deflection
Equation (1), where it is the maximum moment in the mem- assuming the whole member to be fully cracked. For exam-
ber) and Mcr is the cracking moment which may be calcu- ple, in a simply supported beam subject to central point load
lated using Equation (3) with modulus of rupture, fr which results in a maximum bending moment, M, the Δg
(in MPa) which is given by and Δcr at mid-span are given by ML2/(12EcIg) and ML2/
( 2=3
) (12EcIcr), respectively. It is important to note that the rigor-
0:30 ð f Þ
fr = max 
c ous and simplified methods give the same results when the
 ð6Þ
1:6− h 1, 000 × 0:30ðfc Þ2=3 beam is subject to constant moment.

where h is the total member depth in millimeters, fc is the

specified cylinder strength of concrete in MPa. The above 2.3 | Comparison between ACI 318-14 and EC2
expression indicates an enhancement in fr due to “size
effect” for beam depths less than 600 mm. To evaluate the differences in the approaches given by ACI
Even though the most suitable deformation parameter 318 and EC2, maximum deflection, Δ is calculated for two
(α) is curvature for flexural members, the code also permits simply supported beams of span, L, with square and circular
the use of rotation or deflection as the deformation parame- cross sections subjected to a constant moment, M (Figure 1).
ter. When curvature is chosen as the deformation parameter, The diameter, D, of the circular cross section is chosen as
the Equation (5) (for M > Mcr) for short term loading 1.13 times the size, a, of the square cross section, to ensure
becomes: that the cross-sectional area (a2) is identical for the two
"
 # beams. The effective cover to the outermost bar is taken as

2
Mcr Mcr 2 0.1a and 0.1D for square and circular sections, respectively.
φe = φg + 1 − φcr ð7Þ The span of the beams is chosen as 12a. Three different per-
M M
centages of total reinforcing steel (p = 100As/a2 = .75, 1.5,
where φe is the effective curvature, φg is the curvature in 2.25) are chosen, where As is the total area of reinforcing
the uncracked state, which is given by M/EcIg and φcr is the steel. For the square beam, the areas of tension steel, Ast,
curvature in the fully cracked state which is given by M/ and compression steel, Asc, are taken as 2As/3 and As/3,
EcIcr. Here, Ec is the short-term modulus of concrete respectively (Figure 2a). For the circular beam, the steel
(in GPa), which for normal strength concrete is rebar is typically uniformly distributed around the circum-

0:3 ference, as shown in Figure 2c. The cylindrical compressive
fc
Ec = 22 ð8Þ strength of concrete is taken as 30 MPa for both the beams.
10
The maximum deflection (Δ) for the simply supported
where fc is the specified cylinder strength in MPa. When beam, subject to constant moment, is given by φe L2/8. As
M < Mcr, φe is taken as φg. per ACI 318, φe is given as M/(EcIe) where Ie is evaluated
For the most accurate estimate of deflection, the “rigor- using Equation (1). In the case of EC2, φe is computed
ous method” is to be used, where the effective curvatures directly using Equation (7). It is important to note that when
are computed at frequent sections along the member and M < Mcr, φe = φg, and Ie = Ig. However, the estimate of
then deflections are calculated by numerical integration. the cracking moment is different according to the two
This method considers the variation of flexural rigidity codes, owing to differences in the expressions for fr
along the length of the member due to cracking. The code Equations (3) and (6). Figure 3a shows the variation of
1636 GEEVAR ET AL.

FIGURE 2 Transformed cross sections for square and circular cracked sections (to find Icr)

Mcr,ACI/Mcr,EC2 (equal to fr,ACI/fr,EC2) with fc for the circular fc = 20 MPa, but reduces with increase in fc (to 3.4% for
beams (Figure 2c). It may be noted that the square beam fc = 60 MPa).
(Figure 2a) gives similar results with slight differences After the onset of cracking (M ≥ Mcr), the flexural stiff-
owing to the difference in the depth of cross sections ness degrades with an increase in the moment. This stiffness
(Figure 2). It is seen that for moderate depths degradation is considered in EC2 in terms of the weighted
(a = 450 mm) the difference in prediction by the two codes average of the curvatures in the uncracked and fully cracked
is within 10%. However, for larger depths, ACI is found conditions (Equation (7)), while in the case of ACI 318, the
to predict higher values of Mcr, for lower grades of concrete weighted average of the moment of inertia of the gross and
and for lower depths, EC2 is found to give higher estimates cracked sections is considered (Equation (1)). The expres-
of Mcr especially for higher grades of concrete. sions for computing the cracked moment of inertia, Icr for
Owing to the difference in expressions for Ec as pre- this purpose is shown in Figure 2 using the concept of trans-
scribed in EC2 and ACI 318, the deflection estimates by the formed section.
two codes in the uncracked condition (M < Mcr) are likely The degradation of stiffness can be seen in the normal-
to be different. This is depicted in Figure 3b in terms of var- ized load-deflection plot (Figure 4) for circular and square
iations in ΔACI/ΔEC2 (equal to Ec,EC2/Ec,ACI) for various cross sections having the same total percentage of reinfor-
grades of concrete. It is seen that ACI 318 invariably pre- cing steel (p = 1.5%) with a = 450 mm and fc = 30 MPa.
dicts lower values of Ec and hence higher values of Δ com- The applied moment, M is normalized here with respect to
pared to EC2. The difference is as high as 28.8% for Mcr,EC2. It is seen that EC2 predicts marginally higher

FIGURE 3 Comparison of estimates of Mcr and Δ by EC2 and ACI 318

GEEVAR ET AL. 1637

A comparison of the normalized moment-deflection plots

using the rigorous and simplified methods of EC2 and the
ACI simplified method is depicted in Figure 7a for a typical
case of a circular beam subject to point load at center (with
p = 1.5%, a = 450 mm, and fc = 30 MPa). The deflection
predicted using EC2 rigorous method is expected to be the
most accurate, compared to which the EC2 method is found
to predict higher deflections by 8% and ACI 318 (which is
most conservative) by 13% (at M/Mcr,EC2 = 4). A similar plot
is presented for the same beam subject to uniformly distrib-
uted load in Figure 7b. Here, the maximum deflection (at
mid-span) is given by ML2/48EIe. For this case, EC2 method
is found to predict higher deflections by 3% and ACI 318 by
8% (at M/Mcr,EC2 = 4), which is slightly less, compared to
FIGURE 4 Deflection estimate for circular and square cross section by the beam with point load.
EC2 and ACI 318 subject to constant moment

stiffness at all stages of loading compared to ACI 318. The 3 | E XP ER IME NT AL IN VE STIGATION
deflection predicted by ACI 318 method is found to be
higher by about 6 and 5% (at M/Mcr,EC2 = 4) for square and Experiments on 14 specimens were conducted at the Struc-
circular beams, respectively, compared to EC2. The square tural Engineering Laboratory, Indian Institute of Technol-
beam is found to have a higher flexural stiffness than the ogy Madras. The parameters considered for the study are
circular beam (with the same cross-sectional area), primarily grade of concrete (mean cylinder strength, fcm) and percent-
because the tension reinforcement is concentrated close age of longitudinal reinforcement (p). The experiments were
to the extreme tension fiber, unlike the circular carried out on six square beams and eight circular beams
section (Figure 2). This difference becomes more pro- subject to a region of zero shear and constant bending
nounced at higher percentages of reinforcing steel. With ref- moment. All specimens were laterally loaded up to failure
erence to the ACI model, it is interesting to also explore the using a displacement controlled actuator. The test specimens
influence of the parameter n (Equation (1)), whose default were provided with sufficient shear reinforcement to ensure
value was taken as 3. It is seen that the postcracking stiff- that the shear mode of failure would not precede flexural
ness reduces marginally when the value is increased to failure. Strains in longitudinal reinforcement at mid-span
4, but increases significantly when the value of n takes on were measured using electrical strain gauges.
lesser values, particularly at n = 1, as shown in Figure 5. Although various alternative (statically determinate) test
The shift in the load-deflection plots is especially more con- setups are possible to achieve a constant bending moment
spicuous in the case of circular beams, compared to square zone, in the present study, it was found convenient to adopt
beams, and this may be attributed to the fact that the ratio a symmetrical arrangement comprising a simply supported
Ig/Icr is invariably much higher for circular beams, com- segment (of span 900 mm) with overhangs on both sides, as
pared to equivalent square beams (refer Equation (1)). shown in Figure 8, with concentrated loads applied at the
When the bending moment diagram varies along the tips of the overhangs. The six square beams were of the
length of the beam, EC2 offers a “rigorous method” (con-
sidering effective curvatures at small intervals along the
length of the beam), as an alternative to the “simplified
method.” Considering the same simply supported beam
shown in Figure 1 subject to concentrated load at mid-span
(Figure 6a), the maximum deflection is given by ML2/12EIe
as per ACI code, where Ie is calculated using Equation (1).
For EC2 rigorous method, the beam is dividing into
k elements, and the effective curvatures at the intersections
of elements are determined using Equation (7) (Figure 6c)
from the bending moment diagram (Figure 6b). The maxi-
mum deflection, Δ can now be easily determined from the
effective curvature diagram using any of the conventional
methods such as the conjugate beam method as shown in
Figure 6d. The maximum deflection can also be calculated FIGURE 5 Influence of parameter n (in ACI-318 formulation) for circular
by the simplified method in EC2 using Equation (9). and square beams subject to constant moment
1638 GEEVAR ET AL.

The deflection at the mid-span (Δ) was recorded using lin-

ear variable differential transducers at every load increment,
which was applied using displacement controlled actuator.
As depicted in Table 1, two different grades of concrete
(fcm = 19.0 and 34.6 MPa) with different percentages of
longitudinal reinforcement (with arrangement as shown in
Figure 9) were used for both square and circular beams.
The beams with a square cross section are designated as
SBi and beams with a circular cross section are designated
as CBj, where “S” represent “square cross section,” “C”
represent “circular cross section,” “B” represent “beam” and
i and j represent the sequence number. The six square
beams were provided with equal longitudinal steel rebar at
top and bottom (with clear cover of 30 mm) and two-legged
stirrups of 8 mm diameter (with 270 mm spacing in the
main span and 150 mm in the overhangs). The eight circu-
lar beams were provided with longitudinal reinforcement
uniformly distributed around the circumference (with a clear
cover of 20 mm) and circular ties of 8 mm diameter (with
250 mm spacing in the main span and 120 mm in the over-
hangs). The measured proof stress values of the reinforcing
steel are also summarized in Table 1.
FIGURE 6 Steps in “rigorous method” of analysis as per EC2

cross-sectional size of 250 × 250 mm and the eight circular 4 | E XP ER IME NT AL RES ULT S A ND
beams were with a diameter of 230 mm with details as DISCUSSION
shown in Table 1. The beams with circular cross
section were cast using a formwork of two half-cut 6 mm All experimental specimens exhibited flexural failure initi-
thick PVC pipes with an internal diameter of 230 mm ated by yielding of tension steel (as observed from strain
joined with clamps. To achieve the required grades of con- gauges at mid-span) followed by crushing of concrete in
crete, a mix proportion of 1:2.30:3.45 with water cement compression. At load levels beyond cracking load, the
ratio of 0.53, and 1:2.09:3.12 with water cement ratio of experimental specimens exhibited a series of uniformly
0.43, were used for fcm = 19.0 and 34.6 MPa, respectively. spaced cracks in the main span, as this region was subjected
The curing was done using wet gunny bags for 14 days. At to constant moment. Wide cracks and large deflections were
the supports and load locations, circular saddles (150 mm) observed at failure. Figure 10 shows typical failure patterns
wide were provided to facilitate load transfer (Figure 8). observed in the test specimens.

FIGURE 7 Deflection estimate for circular beams by ACI 318 and EC2 (simplified and rigorous)
GEEVAR ET AL. 1639

FIGURE 8 Experimental setup

4.1 | Comparison of deflections at service loads moment at collapse. Corresponding to the various values of
A comparison of deflections at service loads, calculated by M, the deflection estimates ΔACI, and ΔEC2 are compared
with experimental deflection, Δexp. From the results of
the ACI and EC2 methods with the experimental results is
summarized in Tables 2 and 3 for square and circular ΔACI/ Δexp and ΔEC2/Δexp listed in Tables 2 and 3, it can be
beams, respectively. Here, the simplified and rigorous observed that the theoretical estimates by both ACI and
EC2 are reasonably conservative, and the degree of conser-
methods of EC2 give the same deflection estimate, as beam
is subject to constant moment. The deflections are calcu- vatism is more-or-less the same for both square and circular
lated at three levels of service load moment, M taken as beams. In square beam specimens SP4 and SP5, the experi-
0.3Mur,exp, 0.4Mur,exp, and 0.5Mur,exp, where Mur,exp is the mental cracking moment was less than that predicted by

TABLE 1 Experimental details

Mean cylindrical Longitudinal Percentage of longitudinal Measured proof stress of

Sl. no. Specimens strength (fcm) MPa reinforcement reinforcement (p)* reinforcement (fy) MPa
Beams with square cross section
1 SB1 19.0 4-12Y 0.72 550
2 SB2 19.0 6-12Y 1.09 550
3 SB3 19.0 4-16Y 1.29 530
4 SB4 34.6 6-12Y 1.09 550
5 SB5 34.6 4-16Y 1.29 510
6 SB6 34.6 6-16Y 1.93 510
Beams with circular cross section
7 CB1 19.0 6-12Y 1.63 502
8 CB2 19.0 8-12Y 2.18 502
9 CB3 19.0 6-16Y 2.90 504
10 CB4 19.0 8-16Y 3.87 504
11 CB5 34.6 6-12Y 1.63 502
12 CB6 34.6 8-12Y 2.18 502
13 CB7 34.6 6-16Y 2.90 504
14 CB8 34.6 8-16Y 3.87 504

*Percentage of longitudinal reinforcement is calculated based on total cross-sectional area.

1640 GEEVAR ET AL.

FIGURE 9 Typical cross section of the test specimens

ACI and EC2, due to which the deflection prediction was axis), is invoked. The M-φ relationship is then utilized to
found to be unconservative in these two cases. generate the effective curvature φe at any stage of loading,
The prediction based on ACI 318 depends on the for which the bending moment M is known. Then, the cor-
parameter n whose value is adopted as 3 for the deflection responding value of maximum deflection at mid-span in the
calculations, for rectangular sections. However, as observed simply supported can be easily generated as Δ = φeL2/8.
earlier (Figure 5), the load-deflection behavior is more sen- The proposed method, which accounts for material stiff-
sitive to the value of n in circular sections. An analysis of ness degradation due to progressive cracking and nonlinear-
the results of the present experimental study to fit the appro- ities, is expected to be far more accurate than the simplified
priate value of n at service loads, showed a range of values code-based methods to predict deflections at service loads.
between 1.0 and 1.5 for circular beams, and between 1.5 Moreover, it is capable of generating the complete load-
and 4 for square beams. deflection behavior up to failure, which is not possible in
the code-based methods. An alternative method is one using
nonlinear finite element analysis; however, this is computa-
tionally rigorous and has issues related to solution conver-
5 | P R E D I C T I O N O F L OA D - D E F L EC T I O N gence, especially while using different commercially
BEHAVIOR available software packages. The proposed method is simple
and elegant in comparison.
The simplified code procedures used for predicting deflec-
tions is suitable only for load levels below service loads. At
higher loads, significant nonlinearity is expected to occur
5.1 | Material models
and a nonlinear analysis is required for an accurate estimate
of deflection. In this section, a procedure for the generation The bilinear stress–strain curve for steel rebar (in tension and
of the complete load-deflection behavior using nonlinear compression), including strain-hardening, adopted in this
material models is outlined and is compared with the experi- study is shown in Figure 11a and can be expressed as follows:
mental results.
σ s = Es εs : εs ≤ εy ð10Þ
Initially, a moment-curvature (M-φ) relationship is gen-


erated for the given cross section, based on the material σ s = fy + Eh εs −εy : εy < εs < εu ð11Þ
models for concrete and steel, using equilibrium and com-
patibility equations. 18 In this exercise, for the purpose of where εs is the strain in steel rebar, εy is the yield strain, εu
generating the resultant force in compression in circular sec- is ultimate strain; σ s is the stress in steel rebar, fy is the yield
tions, a numerical procedure, summing up the contributions stress; Es is the elastic modulus and Eh is the modulus at
of a large number of very thin strips (above the neutral strain-hardening of steel rebar.

FIGURE 10 Failure patterns observed in experiments

GEEVAR ET AL. 1641

TABLE 2 Comparison of deflection at various service load stages (M = 0.3Mur,exp, 0.4Mur,exp, 0.5Mur,exp) for square beams

Specimens Mur,exp (kNm) M (kNm) ΔACI (mm) ΔEC2 (mm) Δexp (mm) ΔACI/Δexp ΔEC2/Δexp
SB1 27.3 8.2 0.18 0.18 0.19 0.95 0.95
10.9 0.40 0.48 0.39 1.03 1.23
13.7 0.68 0.75 0.51 1.33 1.47
SB2 40.3 12.1 0.44 0.45 0.25 1.76 1.80
16.1 0.73 0.70 0.37 1.97 1.89
20.2 1.01 0.95 0.52 1.94 1.83
SB3 43.2 12.9 0.47 0.45 0.35 1.34 1.29
17.3 0.74 0.70 0.54 1.37 1.30
21.6 0.99 0.92 0.72 1.38 1.28
SB4 40.8 12.3 0.24 0.21 0.43 0.56 0.49
16.3 0.50 0.51 0.59 0.85 0.86
20.4 0.78 0.79 0.79 0.99 1.00
SB5 43.2 12.9 0.27 0.24 0.31 0.87 0.77
17.3 0.53 0.53 0.51 1.04 1.04
21.6 0.79 0.77 0.70 1.13 1.10
SB6 56.4 16.9 0.42 0.38 0.20 2.10 1.90
22.6 0.66 0.61 0.36 1.83 1.69
28.2 0.88 0.82 0.52 1.69 1.58
Average 1.34 1.30
C.O.V. 0.33 0.32

TABLE 3 Comparison of deflection at various service load stages (M = 0.3Mur,exp, 0.4Mur,exp, 0.5Mur,exp) for circular beams

Specimens Mur,exp (kNm) M (kNm) ΔACI (mm) ΔEC2 (mm) Δexp (mm) ΔACI/Δexp ΔEC2/Δexp
CB1 24.7 7.4 0.66 0.6 0.41 1.61 1.46
9.9 0.97 0.88 0.54 1.80 1.63
12.4 1.25 1.14 0.74 1.69 1.54
CB2 31.7 9.5 0.77 0.69 0.54 1.43 1.28
12.7 1.08 0.96 0.69 1.57 1.39
15.9 1.34 1.22 1.01 1.33 1.21
CB3 37.6 11.3 0.77 0.7 0.67 1.15 1.04
15.0 1.04 0.95 0.87 1.20 1.09
18.8 1.31 1.2 1.05 1.25 1.14
CB4 48.4 14.5 0.82 0.75 0.72 1.14 1.04
19.4 1.10 1.01 0.94 1.17 1.07
24.2 1.37 1.27 1.18 1.16 1.08
CB5 26.6 7.8 0.54 0.51 0.35 1.54 1.46
10.6 0.87 0.81 0.55 1.58 1.47
13.3 1.17 1.1 0.76 1.54 1.45
CB6 35.9 10.8 0.75 0.68 0.49 1.53 1.39
14.4 1.07 0.99 0.75 1.43 1.32
18.0 1.38 1.28 1.01 1.37 1.27
CB7 42.7 12.8 0.78 0.71 0.56 1.39 1.27
17.1 1.08 1.00 0.74 1.46 1.35
21.4 1.36 1.28 0.96 1.42 1.33
CB8 52.0 15.6 0.8 0.74 0.48 1.67 1.54
20.8 1.08 1.01 0.72 1.50 1.40
26.0 1.35 1.30 0.97 1.39 1.34
Average 1.43 1.32
C. O. V. 0.13 0.13
1642 GEEVAR ET AL.

FIGURE 11 Assumed uniaxial stress–strain curves for (a) reinforcing steel; (b) concrete in compression; (c) concrete in tension

For concrete in uniaxial compression and tension, the approach (numerical integration).21–24 As shown in Figure 12,
expressions proposed by Bažant and Oh19 and Carrera and the forces in each strip of concrete (Cj, in the jth strip) and steel
Chu20 are used in the study, respectively, including the effects rebar layer (Si, in the ith layer) can be calculated from the cor-
of strain softening (in compression) and tension stiffening responding strains (obtained from linear strain profile) using
(in tension) (Figure 11b,c). They can be expressed as follows: nonlinear material models discussed above.
For concrete in compression, Using equilibrium equations (Equations (15) and (16),
the neutral axis depth, x and the moment, M can be calcu-
Ec εc
σc =    2 ð12Þ lated for each increment of strain at the extreme compres-
1 + Ecfεc c0 −2 εεc0c + εεc0c sion fiber, εc.
X X
where Ec is the elastic modulus of concrete, fc is the peak P= Cj + Si ð15Þ
stress in compression (cylindrical strength) and εco is the j i

strain at peak compressive stress, taken as 2fc/Ec 20 whose X X

M= Cj yj + Si ei ð16Þ
value is close to 0.002. j i
For concrete in tension,
 Si = Asi ðfsi −fci Þ ð17Þ
ft η εεtpt
σt =  η : 0 < εt ≤ εtf ð13Þ Cj = fcj bj Δx ð18Þ
η −1 + εεtpt where P is the axial force in the member (is zero for
beams); yj and ei are the perpendicular distances of the jth
σ t = 0 : εt > εtf ð14Þ
concrete strip and ith steel rebar layer, respectively, from
where ft is the direct tensile strength of concrete, εtp is the the centroidal axis; Asi is the area of steel rebar in the ith
strain at peak tensile stress, which is taken as 0.00018; εtf is layer; fsi is the stress in steel rebar, fci is the stress in con-
the final strain at which tensile stress drops to zero, which crete at the ith layer of steel rebar, fcj is the stress in the jth
is taken as 0.002; η is a parameter that depends on the shape concrete strip; bj is the length of the jth strip which can be
of stress–strain curve whose value is found to vary from calculated from geometry in terms of diameter, D and yj.
1.45 to 2.26 depending on various parameters like percent- The curvatures can be obtained from x using the follow-
age of reinforcing steel, spacing, clear cover, etc. For the ing expression:
εc
present study, a value of 1.8 is chosen for η.20 φ= ð19Þ
x
For a rectangular cross section, the same procedure
5.2 | Generation of M-φ curve described for circular cross section can be used, with bj = b,
Strain compatibility method is popularly used for the gener- breadth of the section.
ation of moment vs. curvature diagram (M-φ curve), assum- The M-φ curves are generated using the described pro-
ing a linear strain distribution within the cross section.18 cedure for all the experimental specimens. The modulus of
Calculation of the resultant compressive force in the con- elasticity, Ec is taken from EC2 recommendations
crete above the neutral axis (Figure 12) is rendered difficult (Equation (8) and ft is taken as 0.34(fc)0.5.25 Es and Eh
in the case of circular cross section as the compressive stress (Figure 11a) for reinforcing steel are chosen as
block has to be integrated over a varying breadth of the sec- 2 × 105 MPa and 7,000 MPa, respectively, based on ten-
tion. To overcome this difficulty the section is divided into sion tests on the reinforcing bars used. The strip thickness,
rectangular strips with small thickness, Δx using the layered Δx is choosen as 0.1 mm.
GEEVAR ET AL. 1643

FIGURE 12 Calculation of forces in concrete and steel layers in circular cross section

5.3 | Validation of load-deflection behavior provided a reasonably low thickness for each layer is cho-
The load-deflection behavior generated using the described sen, for both square and circular beams.
procedure from the M-φ curves for all test specimens and A noticeable difference between the responses of square
are validated with experimental results (Figures 13 and 14). and circular beams is with regard to the point of yielding,
The predicted load-deflection behaviors for the tested square which is relatively sharp and well defined in the case of the
square beam but gradual and smooth in the case of the cir-
and circular beams are found to match with the experimen-
tal results reasonably well. The adopted layered approach is cular beam. This may be attributed to the gradually varying
able to capture the overall trend of the member response, effective depth in the case of tension bars in the circular
beam, compared to the uniform effective depth in case of

FIGURE 13 Validation of load-deflection curve for square beams

1644 GEEVAR ET AL.

FIGURE 14 Validation of load-deflection curve for circular beams

square beams. In general, it can be seen from the plots A comparison of the normalized moment-deflection
(Figures 13 and 14) that the proposed method tends to plots is depicted in Figure 15a for beams with three differ-
slightly underestimate the stiffness of the beam in the post- ent total percentages of reinforcing steel (p = .75, 1.5 and
cracking phase up to yielding, for both square and circular 2.25%) and in Figure 15b for beam-columns (with
beams, indicating that the models used are not able to capture p = .75%) subject to different levels of nominal axial com-
accurately the “tension stiffening effect.” In the postyielding pressive stress (σ = 0, 0.04fc and 0.08fc). These axial load
phase up to collapse, the trend is reasonably well captured. levels were chosen such that the member undergoes a duc-
tile failure initiated by yielding of steel rebar. The following
inferences may be drawn from Figure 15:
5.4 | Influence of percentage of reinforcing steel and
axial load • The load-deflection behavior is seen to be somewhat tri-
The proposed procedure can now be used to study the influ- linear for the square beams, with the yield points well
ence of percentage of reinforcing steel and axial load level defined. In the circular beams, the postyielding phase is
on the load-deflection behavior. Beams subject to pure multilinear owing to the nature of distribution of tension
bending with square and circular cross sections (Figure 2a, steel around the circumference.
c), with dimensions shown in Figure 1 are considered here • In the absence of axial load, all the beam specimens,
for the parametric study. having the same cross-sectional area are found to have
GEEVAR ET AL. 1645

FIGURE 15 Influence of percentage of steel (p) and axial compressive stress (σ) on load-deflection behavior of square and circular beams

more-or-less the same cracking load, which appears to respectively, for an axial compressive stress, σ = 0.04fc
be around 1.8 times the value estimated by EC2. The with respect to zero axial compressive stress).
presence of axial compression enhances the crack-
ing load.
• All the beams exhibit more-or-less same initial stiffness
up to first crack, but the degradation in stiffness in the 6 | C O NC L U S IO N S
postcracking phase is dependent on the percentage of
reinforcing steel and its distribution in the cross section. Experimental studies are reported in this paper on beams
The stiffness increases significantly with increase in per- with circular and square sections with the objective of veri-
centage of reinforcing steel (due to improved tension fying the applicability of simplified code-based methods
stiffening effects) and is found to be higher in the case (ACI 318 and EC2) to compute service load deflections.
of square beam compared to circular beams due to These procedures, originally developed for rectangular sec-
higher effective depth. tions, are validated in this paper for circular beams with dif-
• The strength of square beams is found to be more than ferent grades of concrete and different percentages of
circular beams and the difference increases with increase reinforcing steel. Furthermore, a numerical procedure based
in percentage of reinforcing steel. (about 9, 14, and 17% on layered approach, considering material nonlinearity in
more for p = .75, 1.5, and 2.25%, respectively). This is concrete and steel rebar, has been successfully developed to
due to higher moment of inertia and effective depth for generate the complete load-deflection behavior of the 14 test
square beams. With the increase in axial compression, specimens. Using this numerical procedure, a parametric
the ultimate strength increases both for circular and study was carried out to assess the influences of percentage
square beams (about 23 and 27% increase for square reinforcing steel and axial compression on the load-
and circular beams, respectively, for an axial compres- deflection behavior. The following salient conclusions may
sive stress, σ = 0.04fc with respect to zero axial com- be drawn from the study:
pressive stress).
• Considerable ductility is found to occur in the beams at • The tests on circular and square beams subjected to con-
lower percentage of reinforcing steel as the sections become stant moment have shown that the simplified procedures
more under-reinforced. Circular beams are found to have for finding service load deflections in EC2 and ACI
significantly less ductility compared to square beams as 318 (derived for rectangular beams) are equally applica-
they have less effective depth for the same overall percent- ble for circular beams, having more-or-less the same
age of reinforcing steel (about 86, 62, and 51% less for degree of conservatism (ranging from 29 to 39% conser-
p = .75, 1.5, and 2.25%). The presence of axial compres- vatism). The conservatism observed was marginally
sion reduces the ductility at failure as the neutral axis shifts higher for ACI 318, compared to EC2.
to increase the cross-sectional area subject to compression • The proposed numerical method, based on layered
(about 25 and 18% decrease for square and circular beams, approach, considering material nonlinearity for concrete
1646 GEEVAR ET AL.

and steel rebar, is found to predict the full load-

fr modulus of rupture
deflection behavior for the square and circular beams
fsi stress in ith layer of steel
reasonable well. However, a slight under-estimation of
ft direct tensile strength of concrete
postcracking stiffness was observed. This numerical pro-
fy yield stress of steel
cedure can be used to investigate the behavior of RC
h total member depth in millimeters
beam-columns of any symmetric cross section, although
Icr cracked moment of inertia
further experiments are needed to validate the influence
Ie effective moment of inertia
of axial loading.
Ig gross moment of inertia
• The parametric study on circular and square beams
IT transformed moment of inertia
shows that the postcracking stiffness increases and duc-
I0 moment of inertia of the circular cross
tility decreases with increase percentage of reinforcing
section (without steel) about its centroidal axis
steel. Presence of axial compression reduces the ductil-
k number of elements along the length of the beam
ity at failure. The circular beams are found to have less
kd neutral axis depth calculated based on linear elastic
strength and ductility, when compared to square beams
analysis
(with the same area of cross section), due to the more
L length of the beam
effective distribution of steel rebar in the square
M bending moment at service load
beams.
Mcr cracking moment
Mcr,ACI cracking moment calculated based on ACI 318
7 | NOTATIONS Mcr,EC2 cracking moment calculated based on EC2
Mur,exp bending moment at collapse for the experimental
AT transformed area of cross section specimens
As total area of longitudinal steel m modular ratio
Asc area of compression steel n coefficient used in the Branson’s equation
Ascj area of compression steel in the jth steel layer P axial compressive load
Asi area of steel in the ith steel layer p total percentage of longitudinal reinforcement (cal-
Ast area of tension steel culated based on total cross-sectional area)
Asti area of tension steel in the ith steel layer pt tensile steel reinforcement ratio
a size of the square cross section Si force in the ith steel layer
b breadth of the cross section y distance from the centroidal axis to the extreme
bj length of the jth strip tension fiber of cross section
c a constant depending on loading and boundary yj distance of the jth concrete strip from the centroidal
conditions axis
Cj compressive force in the jth strip y1, y2 distances used in the expression of cracked moment
D diameter of the circular cross section of inertia for circular section
di distance from the extreme compressive fiber to the x neutral axis depth
ith tension steel layer α effective deformation parameter
dj distance from the extreme compressive fiber to the αI deformation parameter in uncracked condition
jth compression steel layer αII deformation parameter in fully cracked condition
Ec modulus of elasticity of concrete β coefficient taking account of the duration of
Ec,ACI modulus of elasticity of concrete calculated based loading
on ACI 318 Δ maximum deflection in the member
Ec,EC2 modulus of elasticity of concrete calculated based ΔACI maximum deflection calculated based on ACI-318
on EC2 Δcr elastic deflection assuming the whole member to
Eh modulus at strain-hardening of steel be fully cracked
Es modulus of elasticity of steel ΔEC2 maximum deflection calculated based on EC2
ei distance from the centroidal axis to the ith steel Δexp maximum deflection measured for the experimental
layer specimens
fc cylinder compressive strength of concrete Δg elastic deflection assuming the whole member to
fci stress in concrete at the ith layer of steel be uncracked
fcj stress in concrete in the jth strip Δx strip thickness
fcm mean cylinder compressive strength in concrete εc compressive strain in concrete
fr modulus of rupture εcj strain in the jth concrete strip
fr,ACI modulus of rupture calculated based on ACI-318 εco strain at peak compressive stress of concrete
fr,EC2 modulus of rupture calculated based on EC2 εs strain in steel
GEEVAR ET AL. 1647

15. Branson DE. Deformation of Concrete Structures. New York: McGaw-Hill;

εsi strain in ith steel layer 1977.
εt tensile strain in concrete 16. Al-Shaikh AH, Al-Zaid RZ. Effect of reinforcement ratio on the effective
εtp strain at peak tensile stress in concrete moment of inertia of reinforced concrete beams. ACI Struct J. 1993;90(2):
144-149.
εu ultimate strain in steel 17. Al-Zaid RZ, Al-Shaikh AH, Abu-Hussein MM. Effect of loading type on
εy yield strain in steel the effective moment of inertia of reinforced concrete beams. ACI Struct J.
η coefficient that depends on the shape of compres- 1991;88(2):184-190.
18. Park P, Paulay T. Reinforced Concrete Structures. John Wiley & Sons,
sive stress–strain curve of concrete New York, USA; 1975.
θ angle used to calculate cracked moment of inertia 19. Bazant ZP, Oh BH. Deformation of progressively cracking reinforced con-
for circular section crete beams. ACI Struct J. 1984;81(3):268-278.
20. Carreira DJ, Chu KH. Stress–strain relationship for reinforced concrete in
ξ coefficient which indicates the level of cracking in
tension. ACI J Proc. 1986;83(1):21-28.
the member 21. Hand FR, Pecknold DA, Schnobrich WC. A Layered Finite Element Non-
σ axial compressive stress linear Analysis of Reinforced Concrete Plates and Shells. University of Illi-
σc compressive stress in concrete nois Engineering Experiment Station, College of Engineering, University of
Illinois at Urbana-Champaign, Illinois, USA; 1972.
σs stress in steel 22. Bashur FK, Darwin D. Nonlinear model for reinforced concrete slabs. J
σt tensile stress in concrete Struct Div. 1978;104(1):157-170.
φ curvature 23. Kwak HG, Kim SP. Nonlinear analysis of RC beams based on
moment-curvature relation. Comput Struct. 2002;80:615-628.
φg curvature in the uncracked condition 24. Srinivasan S, Menon D. RC rectangular column sections under biaxial
φcr curvature in the cracked condition eccentric compression- an improved design recommendation. J Struct Eng.
φe effective curvature 2003;29(4):205-211.
25. American Concrete Institute. Prediction of Creep, Shrinkage, and Tempera-
ture Effects in Concrete Structures. ACI 209R-92, Farmington Hills,
ORCID Michigan, USA, 1997.

Devdas Menon https://orcid.org/0000-0003-3476-0553

AUTHOR'S BIOGRAPHIES

REFERENC ES Indu Geevar,

1. Gaston, J. R., Siess, C. P. and Newmark, N. M.. An investigation of the MTech, PhD Research Scholar,
load-deformation characteristics of reinforced concrete beams up to the
Department of Civil Engineering,
point of failure. Illinois University at Urbana Engineering Experiment Sta-
tion. 1952. Indian Institute of Technology
2. Batoz JL, Dhatt G. Incremental displacement algorithms for nonlinear prob- Madras,
lems. Int J Numer Methods Eng. 1979;14(8):1262-1267. Chennai 600036,
3. Shariff MN, Menon D. Displacement-controlled nonlinear analysis of RC
frames and grids. J Struct Eng. 2015;42(5):393-404.
India
4. Broms BB. Stress distribution, crack patterns, and failure mecha-
nisms of reinforced concrete members. ACI J Proc. 1964;61(12):
1535-1558.
5. Broms BB. Crack width and crack spacing in reinforced concrete members.
ACI J Proc. 1965;62(10):1237-1256.
Bijily Balakrishnan,
6. Menon D. Advanced Structural Analysis. New Delhi, India: Narosa Publish-
ing House; 2009. MTech, PhD Research Scholar,
7. American Concrete Institute. Building Code Requirements for Structural Department of Civil Engineering,
Concrete and Commentary. Farmington Hills, MI: ACI Committee Indian Institute of Technology
318, American Concrete Institute; 2014 ACI 31814.
8. British Standards Institution. Design of Concrete Structures: Part 11: Gen-
Madras,
eral Rules and Rules for Buildings, 2004, Eurocode 2, Brussels, Belgium: Chennai 600036,
part 1. India
9. Lee YH, Scanlon A. Comparison of one-and two-way slab minimum thick-
ness provisions in building codes and standards. ACI Struct J. 2010;107(2):
157-163.
10. Pillai U, Menon D. Reinforced Concrete Design. 3rd ed. New Delhi, India:
Tata McGraw Hill Publication Co. Ltd.; 2009.
11. Balázs GL, Bisch P, Borosnyoi A, et al. Design for SLS according to fib Fayaz Habeeb,
model code 2010. Struct Concr. 2013;14(2):99-123. M. Tech., Structural Engineer,
12. Ashour SA. Effect of compressive strength and tensile reinforcement ratio
on flexural behavior of high-strength concrete beams. Eng Struct. 2000;
WS Atkins,
22(5):413-423. Bangalore,
13. Gribniak V, Cervenka V, Kaklauskas G. Deflection prediction of reinforced India
concrete beams by design codes and computer simulation. Eng Struct.
2013;56:2175-2186.
14. Branson, DE: Instantaneous and time-dependent deflections of simple and
continuous reinforced concrete beams. Alabama Highway Department,
HPR Report, 1963, Vol. 7, pp. 1–78.
1648 GEEVAR ET AL.

Devdas Menon,
PhD, Professor, Dept of Civil
Engineering, How to cite this article: Geevar I, Balakrishnan B,
Indian Institute of Technology Habeeb F, Menon D. Experimental and numerical
Madras, assessment of deflections in circular reinforced con-
Chennai 600036, crete beams. Structural Concrete. 2018;19:
India 1633–1648. https://doi.org/10.1002/suco.201700108
Copyright of Structural Concrete is the property of John Wiley & Sons, Inc. and its content
may not be copied or emailed to multiple sites or posted to a listserv without the copyright
holder's express written permission. However, users may print, download, or email articles for
individual use.