Вы находитесь на странице: 1из 8

International Journal of Steel Structures

September 2013, Vol 13, No 3, 487-494


DOI 10.1007/s13296-013-3008-z

www.springer.com/journal/13296

Finite Element Investigation on Deflection of Cellular Beams


with Various Configurations
Pattamad Panedpojaman1,* and Thaksin Thepchatri2
1

Lecturer, Department of Civil Engineering, Faculty of Engineering, Prince of Songkla University, Songkhla, 90110, Thailand
Professor, Department of Civil Engineering, Chulalongkorn University, Phayathai Road, Pathumwan, Bangkok, 10330, Thailand.

Abstract
The effect of the cellular beam configuration on the deflection has been investigated by using the three-dimensional finite
element analysis of 408 cellular beams. Cross-section dimension, beam slenderness, opening size and opening spacing are
found to affect the stiffness of the analytical load-deflection curve. The parametric study of the FE stiffness and normalized
stiffness is conducted. The stress distribution of the finite element (FE) model reveals the strut stress in the web-post contributes
to the increasing deflection in addition to the regular bending deflection. The deformation of the web-post and the relative
movement between the upper and lower tee-sections due to strut stress is the main reason of the additional deflection in the
cellular beams. The effect of the strut stress is found to be significant for the deflection of the short-span beams but less for
the long-span beams. To convert the theoretical bending deflection to be the overall deflection, the calibrating coefficient
function is established by using the empirical study. The function is formulated in terms of the slenderness, spacing and cross
section ratio. The deflection prediction is useful to check serviceability condition for the design purpose.
Keywords: cellular beam, deflection, finite element analysis, strut stress, web-post

1. Introduction
Cellular beams are made by cutting a beam web of a
steel rolled beam in a half circular pattern along its
centerline and re-welding the two halves. The regular
circular openings along its span are made; as a result, the
rolled beam section is modified and improved its
structural capacity by increasing the depth through the
openings. Cellular beams are approximately 40-60%
deeper and 40-60% stronger than its original section
(Erdal et al., 2011). The use of cellular beams provides
lightened and longer-span structures. The openings are
useful for allowing technical installations (pipes and
ducts) to pass through the openings. The basic geometry
and notations used for cellular beams are described in
Fig. 1. The beam is characterized by its rolled beam depth
d, cellular beam depth H, span L, flange width b, flange
thickness tf , web thickness tw , circular opening diameter
do, and uniformly spacing between the adjacent opening
Note.-Discussion open until February 1, 2014. This manuscript for
this paper was submitted for review and possible publication on January 3, 2012; approved on August 14, 2013.
KSSC and Springer 2013
*Corresponding author
Tel: +6674-28-7140; Fax: +6674-45-9396
E-mail: ppattamad@eng.psu.ac.th

at distance s along the span of the beam.


The various designs of cellular beams have been proposed
(Ward, 1990; Dionisio et al., 2004; Chung et al., 2003;
Lawson et al., 2006). The ultimate state design of cellular
beams requires checking their strength and serviceability.
The strength design of cellular beams considers overall
beam flexural capacity, beam shear capacity, overall
beam buckling strength, web-post flexure and buckling,
Vierendeel bending of upper and lower tees and local
buckling of compression flange. Moreover, deflection is
also considered as the serviceability requirements.
According to SCI (Ward, 1990; Knowles, 1985), the
deflection of cellular beams computed by using the
virtual work method was proposed. Benitez et al. (1998)
proposed the matrix stiffness method to predict the
deflection. Deflection calculations were also presented in
a number of researches and design recommendations
(Lawson, 1987; Lawson et al., 1997; Darwin, 1990). In
summary, each web opening contributes to the increased
deflection caused by the variations in bending and shear
stiffness along the span of the beam. Generally, the
influence of the variation of bending stiffness dominates
the deflection (Lawson et al., 2006). However, the
detailed calculations of deflection, such as the virtual
work method (Ward 1990, Knowles 1985) or the matrix
stiffness method (Benitez et al., 1998), are not usually
proper as they are both complex and potentially inaccurate

488

Pattamad Panedpojaman and Thaksin Thepchatri / International Journal of Steel Structures, 13(3), 487-494, 2013

Figure 1. Geometrical parameters of cellular beam.

Figure 2. Determination of the coefficient K1 and K2.


Table 1. Additional deflection due to individual openings
Case
Circular openings
Unstiffened openings
Stiffened opening

Opening depth/Beam depth


0.5

0.6

0.7

1%
2%
1.5%

1.5%
3%
2.5%

2%
3.5%

(Chung and Lawson, 2001). Therefore, simple methods to


predict the deflection are developed such as the typical
design table method (Fabsec, 2006) and the simple
coefficient method (ArcellarMittal, 2008). The predicted
deflection depends on geometrical variables of cellular
beams. The methods are generally derived based on the
empirical study of Finite Element Analysis (FEA).

1.1. Typical design table method (Fabsec, 2006)


The additional deflection due to individual openings is
expressed as percentage of the maximum deflection of an
imperforated beam as shown in Table 1. For beams with
regular circular openings, the method may be conservative.
And the additional deflection due to the regular openings
should not exceed 25%. For cellular beams having
opening depth smaller than 40% of the beam depth, the
effect of the openings on deflection can be neglected.
1.2. Simple coefficient method (ArcellarMittal, 2008)
The mid-span deflection of a simply-supported cellular
beam under uniform load is computed based on a

theoretical bending deflection in elastic stage.


4

5wL
TB = -------------------384EInet

(1)

where
3

Inet = I twd0 12

(2)

in which I is moment of inertia of the imperforated


section; Inet is moment of inertia of the upper and lower
tee section; w is uniform load; and E is Youngs modulus
of steel. Note that Inet is used for a conservative
consideration of the moment of inertia for the perforated
section. To compute the deflection of cellular beams CB,
the formulation is multiplied by a coefficient which
represents the opening effect.
4

5wL CB = K1K2------------------384EInet

(3)

in which K1 and K2 are a coefficient, enabling the


slenderness (L/H) and the sensitivity of the number of
openings (L/s) of cellular beams, respectively, as shown
in Fig. 2.
However, the previous study does not clearly explain
the effect of the openings on the deflection. The simple
methods are still potentially inaccurate for some geometry
of cellular beams, especially for cellular beams with the
short spacing that is the narrow web-post. Therefore, the

Finite Element Investigation on Deflection of Cellular Beams with Various Configurations

489

Figure 3. Cellular-beam configuration.

purpose of the study is to investigate effect of the


openings on the deflection and to develop a new simple
coefficient function for predicting the deflection of cellular
beams covering various geometries through the FEA.

2. Cellular Beam Model and Investigated


Parameters
To provide the insight into the structural behavior of
cellular beams, the study is conducted numerically by
using a three-dimensional finite element model to simulate
behavior of cellular beams. The scope of the analysis is a
simply supported beam under uniform load. Poissons ratio
of 0.3 and bi-linear material with Youngs modulus of
2105 MPa and yield stress of 245 MPa without hardening
behavior are assumed in the analysis. The beam is cabbed
at both ends as normally used with plate having thickness
as same as the flange thickness of the beam.
A total of 408 non-linear finite element models of
cellular beams is used in the study that covers various
opening configurations and beam section sizes as follows:
- slenderness ratio L/H in range of 5 to 40;
- opening depth ratio do /d in range of 0.8 to 1.2;
- spacing ration s/do in range of 1.1 to 1.8; and
- beam section sizes of H400300107, H500200
89.6, H700300185 and H800300210.
The spacing between the adjacent openings is related to
the web opening diameter based on the half circular
pattern for the cutting. The beam parameters of L/H, do /d,
s /do and Af /Aw are used to investigate their effect on the
deflection. Note that Af /Aw in Eq. (4) is the parameter of
the cross section.

2btf
Af Aw = ---------------------tw(H 2tf)

(4)

in which Af is area of the flange of cellular beams which


can be considered as a representative of the ability to
resist the moment; and Aw is area of the web of cellular
beams which can be considered as a representative of the
ability to resist the shear force.
The commercial nonlinear finite element analysis
program, ANSYS, is used to investigate behavior of the
cellular steel beams. The FE model is simulated using the
eight-node solid element (Solid45). The element consists
of three degrees of freedom (DOF) at each node-nodal
translations. The finite element discretization yields a set

Figure 4. Comparison of the load-deflection curve of the


rolled beams obtained from the finite element model and
the theoretical formulation.

of simultaneous equations. Since the coefficient matrix is


a function of the unknown DOF values (or their derivatives),
the analysis is nonlinear. The Newton-Raphson iterative
method is used to solve the nonlinear equations (ANSYS
2009). The structural analysis is controlled by time step
analysis. The load varies within each time step but the
self load is constant. Element size of the model is 25, 30,
35, and 40 mm for the rolled beam of H400300107,
H50020089.6, H700300185 and H800300210,
respectively. Before the provided element size is selected,
several mesh configurations are attempted to set until
providing convergence of the load-deflection curve within
reasonable execution time.
The beam model as presented in Fig. 3 is defined by
using a global Cartesian coordinate system. The origin is
located at the mid-height of the cross-section at the left
end of the beams. The beams with a simply supported
condition are imposed. Vertical displacement and out-ofplane displacement are restrained at the central node of
the cross-section at both ends, whereas longitudinal
displacement is only restrained at the left end. The
uniform load is applied on the top surface of the beam. In
order to verify the ability of the model to simulate a
simply supported condition, the analyzed load-deflection
curve of the rolled beams with L/H of 12 is compared
with the theoretical bending deflection (Eq. (1)) as shown
in Fig. 4. The good agreement of the curve below
proportional limit between the model and the formulation
is found in the comparison.

490

Pattamad Panedpojaman and Thaksin Thepchatri / International Journal of Steel Structures, 13(3), 487-494, 2013

Figure 5. Variation of kFE and kFE /kTB with the parameters of do /d, L/H, s /do and Af /Aw: (a) variation with do /d for the
cellular beams of H800300210 and L/H=18; (b) variation with L/H for the cellular beams of H800300210 and do /d
=0.9; (c) variation with s/do for the cellular beams of H800300210 and do /d=0.9; and (d) variation with Af /Aw for the
cellular beams of s/do=1.5, do /d=0.9.

3. Finite Element Analysis of Cellular Beams


Since the prediction of the deflection for serviceability
requirements in structural design is considered in the
elastic stage, the discussion of the deflection is focused in
the elastic stage. The influence of do /d, L/H, s /do and
Af /Aw on the deflection is investigated in terms of the
stiffness of the load-deflection obtained from the FE
model, kFE, in Eq. (5) and the normalized kFE with the
theoretical stiffness kTB, kFE /kTB. Note that kFE /kTB is also
used to investigate the effect of the parameters on the

ability of the bending formulation to predict the deflection


of cellular beams. Due to analyzing a number of models,
the analytical influence of the parameters is described
based on given representative groups of the cellular
beams as shown in Fig. 5.

w
kFE = ---------max

(5)

in which max is the maximum deflection of the beam


which is at the mid-span obtained from the FE model.

Finite Element Investigation on Deflection of Cellular Beams with Various Configurations

491

Table 2. Type of the cellular beam


Type

Definition

Details

Representative beam

L/H

s /do

Long-span beam

>12

1.1 to 1.8

H800300210 with L/H=18,


s /do= 1.8 and do /d =0.8

Short-span beam with wide web-post

5 to 12

>1.2 to 1.8

H800300210 with L/H=7,


s /do= 1.3 and do /d =1.1

Short-span beam with narrow web-post

5 to 12

1.1 to 1.2

H800300210 with L/H=7,


s /do= 1.1 and do /d =1.2

Figure 6. Stress distribution of the long-span beam for the half-span length.

The stiffness value is increased with the smaller value


of do /d and L/H but with the larger value of s /do as
shown in Fig. 5(a) to 5(c). The variation of do /d slightly
affects kFE /kTB as shown in Fig. 5(a). However, do /d
implicitly affects kTB through the equation of Inet. The
value of kFE likely converts to the value of kTB, which is
kFE /kTB1, for the large value of L/H (higher than 25) as
shown in Fig. 5(b). With the smaller value of s /do, the
value of kFE /kTB is lower however kFE can still be
converted to kTB for the large value of L/H as shown in
Fig. 5(c). In other words, the stiffness, computed depending
on the pure bending effect-that is kTB, does not dominate
the structural behavior for the small value of s /do and L/H.
The shear force has more effect on either the stiffness or
the deflection for such beams.
As shown in Fig. 5(d), Af /Aw cannot directly be used to
observe its influence on the stiffness due to the moment
of inertia of each section highly influences the stiffness
value. The normalized stiffness describes that value of
kFE /kTB is smaller with larger value of Af /Aw. For the large
value of Af /Aw that is Af > Aw, the ability to resist the shear
force is low comparing with the ability to resist the
moment. Due to the weakness of the shear ability, the
shear force again takes more effect on deflection.
The bending and shear effect on the cellular beams are also
investigated through the stress distribution. Corresponding
to the parametric study, the beams can be identified into
three types as shown in Table 2. The stress distribution of
the cellular beams at the proportional limit is described
through the representative beam of each type.

For the long-span beam, the stress distribution of the


representative beam is shown in Fig. 6. As a result of the
bending moment effect, the longitudinal stress is generated.
Compressive stress and tensile stress are developed in the
upper tee-section and the lower tee-section, respectively.
The longitudinal stress at the upper and lower tee-sections
is very high comparing with the shear and vertical (Y
direction in Fig. 3) stresses around the openings. The high
bending stress dominates the beam behavior. Therefore,
the analytical deflection or stiffness of the beam can
approximately be predicted by using the bending
formulation.
The stress distribution of the short-span beam with
wide and narrow web-post is shown in Fig. 7. The less
longitudinal stress is observed in these beams, especially
for the beam with narrow web-post. The high shear and
vertical stresses are observed in the figure. Note that the
vertical stress is significantly higher than the shear stress.
The shear force can be considered to be more effect on
the short-span beam. Comparing with the beam with
narrow web-post, the shear and vertical stresses of the
beam with wide web-post are better distributed through
the wider web-post while the bending stress is developed.
Due to the bending stress does not dominate the beam
behavior, the analytical deflection or stiffness of the beam
differs from the bending formulation.
The effect of shear force, V, on the deflection can be
considered into the direct and indirect effects. The shear
stress is used to describe the direct effect on the deflection.
As the indirect effect of shear force, compressive and

492

Pattamad Panedpojaman and Thaksin Thepchatri / International Journal of Steel Structures, 13(3), 487-494, 2013

Figure 7. Stress distribution of the short-span beam for the half-span length.

Figure 8. Forces and deformation in the web-post based on a strut analogy.

tensile forces act across the web-post on opposite diagonals


based on a strut analogy (Lawson et al., 2006) as illustrated
in Fig. 8(a). The strut stress can approximately be observed
through the vertical stress. Due to the stress acting on the
strut, the deformation of the web-post and the relative
movement between the upper and lower tee-sections are
occurred as shown in Fig. 8(b). The behavior accounts for
the additional deflection. Because of the significantly
higher vertical stress comparing with the shear stress, the
strut stress is considered as the major effect of the shear
force on the deflection.
In summary, once the span is not long enough, the
bending moment does not dominate the deflection
behavior. The shear force takes more effect on such
beams, especially for the short-span beam with narrow
web-post which the bending stress is not well developed.
As the major effect of the shear force, the strut stress
contributes to increase the deflection or soften the
stiffness in addition to the bending formulation.

4. Coefficient for Predicting the Deflection


As described in the previous section, the bending
formulation Eq. (1) cannot directly be used to predict
either the deflection or the stiffness of cellular beams.
The difference between kFE and kTB of the 408 finite
element models is illustrated in Fig. 9. To predict the
stiffness of cellular beams, kTB must be calibrated by a
coefficient, c, as

Figure 9. Comparison of kFE with kTB.

kFE=ckTB

(6)

To identify the calibrating coefficient, the correlation


between kFE and kTB is investigated through a regression
analysis. The stiffness data of the finite element models
with various parameters of L/H, do /d, s /do and Af /Aw is
separated into the three types before the empirical
analysis is conducted.
The analysis clearly assures that do /d is less effective
on the coefficient function. The coefficient is found to be
a function of L/H, s /do and Af /Aw in terms of a
multilinear relationship. According to the multilinear

Finite Element Investigation on Deflection of Cellular Beams with Various Configurations

493

Figure 10. Comparison of kFE with the calibrating kTB.

Figure 11. Comparison of the FE stiffness and the


predicted stiffness.

regression analysis, the coefficient function can be


formulated as follows:
c=0.101+0.486s /do+0.014L/H0.090Af /Aw 1
for L/H >12

(7)

c=0.034+0.404s /do+0.039L/H0.139Af /Aw


for s /do >1.2 and 7 L/H 12

(8)

c=1.583+1.692s /do+0.040L/H0.163Af /Aw


for 1.1s /do 1.2 and 7 L/H 12

(9)

The comparison of kFE with the calibrating kTB is shown


in Fig. 10(a). Note that the large error found in the
comparison is only for the results of the beams with L/H<7
which is the very short beams. The comparison of kFE
with the calibrating kTB for the beam with L/H7 is
shown in Fig. 10(b). The coefficient function can be used
to predict the deflection of cellular beams as
4

5wL
max = 1--- -------------------c 384EInet

(10)

where Inet is describe in Eq. (2). Note that Inet can be

simplified by using Inet =0.93I based on the average Inet /


I of the investigated beams.
To verify the ability of the proposed function in order
to improve the prediction, another set of 51 beam models
are analyzed for the stiffness. The beam configuration is
describes as follows:
- the rolled beam of H30020056.8, H600300150
and H900300243
- L/H ratio in range of 7 to 45
- do /d ratio in range of 0.8 to 1.2
- s /do ratio in range of 1.1 to 1.8
The FE results are compared with the bending
formulation (Eq. 1), the table method (Fabsec, 2006), the
coefficient method (ArcellarMittal, 2008) and the proposed
method as illustrated in Fig. 11. It can be seen that the
proposed method significantly improves the prediction of
the stiffness with the coefficient of determination R2 of
0.99.

5. Conclusion
The influence of cross-section dimension, beam
slenderness, opening size and opening spacing on the
deflection or the stiffness of the cellular beams is
investigated by using Finite Element program. The
stiffness of the analytical load-deflection curve is increased
with the smaller value of do /d and L/H but with the larger
value of s /do. Furthermore, the ability of the bending
formulation to predict the deflection of cellular beams is
also investigated through the normalized stiffness kFE /kTB.
The bending formulation can be used to predict the beam
with the large value of L /H. With the larger value of
s /do and the smaller value of Af /Aw, the value of kFE /kTB
is higher whereas the variation of do /d slightly affects the
value of kFE /kTB.
Based on the stress distribution, the shear effect in
terms of the strut stress is found to be the major reason of
the increasing deflection in addition to the bending effect.
As a result of the strut stress, the deformation of the webpost and the relative movement between the upper and

494

Pattamad Panedpojaman and Thaksin Thepchatri / International Journal of Steel Structures, 13(3), 487-494, 2013

lower tee-sections occur. The behavior accounts for the


additional deflection, especially for the short-span beams.
The shear effect, however, becomes lower for the longspan beams. The deflection obtained from the FEA of
such cellular beams is converted to that of the theoretical
bending deflection.
Since the shear force also affects the deflection of
cellular beams, the bending formulation cannot directly
be used to predict the deflection. The calibrating coefficient
function for the bending formulation is examined through
the correlation between the stiffness of the FE models and
the bending formulation. The function can approximately
be formulated only in terms of L/H, s /do and Af /Aw. The
predicted deflection is verified by another set of the
cellular beams. The accuracy of the deflection prediction
is significantly improved comparing to the previous
methods.

Acknowledgments
The authors would like to acknowledge Prince of
Songkla University, Thailand for providing the financial
support for this research project (contract no. ENG550149S).

References
ANSYS (2009). Users Manual Release 12 0 1. Canonsburg,
Pennsylvania, USA.
ArcellarMittal (2008). ACB Cellular Beams. Edition 2008.
ArcellarMittal, Luxembourg.
Benitez, M. A., Darwin, D., and Donahey, R. C. (1998).
Deflections of composite beams with web openings.
Journal of structural engineering, ASCE, 124(10), pp.
1139-1147.
Chung, K. F. and Lawson, R. M. (2001). Simplified design
of composite beams with large web openings to Eurocode

4. Journal of Constructional Steel Research, 57, pp. 135163.


Chung, K. F., Liu, T. C. H., and Ko, A. C. H. (2003). Steel
beams with large web openings of various shapes and
sizes: an empirical design method using a generalized
moment-shear interaction. Journal of Constructional
Steel Research, 59, pp. 117-1200.
Darwin, D. (1990). Design of steel and composite beams
with web openings. In: Steel design guide series 2.
American Institute of Steel Construction.
Dionisio, M. C., Hoffman, R. M., Yost, J. R., Dinehart, D.
W., and Gross, S. P. (2004). Determination of critical
location for service load bending stresses in noncomposite cellular beams. Proc. 17th ASCE Engineering
Mechanics Conference, University of Delaware.
Erdal, F., Doan, E., and Saka, M. P. (2011). Optimum
design of cellular beams using harmony search and
particle swarm optimizers. Journal of Constructional
Steel Research, 67, pp. 237-247.
Fabsec (2006). Design of FABSEC cellular beams in noncomposite and composite applications for both normal
temperature and fire engineering conditions, Edition
2006. Fabsec Ltd., UK.
Knowles, P. R. (1985). Design of castellated beams. The
Steel Construction Institute Publication.
Lawson, R. M., Mullett, D. L., and Rackham, J. W. (1997).
Design of asymmetric slim floor beams using deep
composite decking. The Steel Construction Institute, SCIP175.
Lawson, R. M. (1987). Design for openings in the webs of
composite beams. The Steel Construction Institute/CIRIA
joint publication, SCI-P068.
Lawson, R. M., Lim, J., Hicks, S. J., and Simms, W. I.
(2006). Design of composite asymmetric cellular beams
and beams with large openings. Journal of Constructional
Steel Research, 62, pp. 614-629.
Ward, J. K. (1990). Design of composite and non-composite
cellular beams. The Steel Construction Institute Publication.

Вам также может понравиться