Anda di halaman 1dari 9

(This is a sample cover image for this issue. The actual cover is not yet available at this time.

This article appeared in a journal published by Elsevier. The attached


copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elseviers archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Author's personal copy

Construction and Building Materials 29 (2012) 619626

Contents lists available at SciVerse ScienceDirect

Construction and Building Materials


journal homepage: www.elsevier.com/locate/conbuildmat

Thermalstress analysis of a three-dimensional end-plate steel joint


Andreas Kalogeropoulos a, Georgios A. Drosopoulos a,, Georgios E. Stavroulakis a,b
a
b

Department of Production Engineering and Management, Technical University of Crete, GR-73132 Chania, Greece
Department of Civil Engineering, Technical University of Braunschweig, Braunschweig, Germany

a r t i c l e

i n f o

Article history:
Received 7 June 2011
Received in revised form 19 October 2011
Accepted 24 November 2011

Keywords:
Thermalstress analysis
Steel connections
Contact mechanics

a b s t r a c t
In this article the behaviour of an extended end-plate steel connection under elevated temperatures is
studied using three-dimensional non-linear nite elements with unilateral contact and friction interfaces. The behaviour of the connection up to collapse has been studied. Within this study, the inuence
of the sequence between thermalmechanical loads, the opening of the interface as well as the representation of the shear forces transmitted through friction and bolts, are investigated. Useful results relevant
to re-structure interaction and the validation of simplied models, are presented.
2011 Elsevier Ltd. All rights reserved.

1. Introduction
Bolted steel connections have non-linear mechanical behaviour
due to unilateral contact and friction effects as well as the elastoplastic behaviour. Inuence of elevated temperatures on the overall response of the joints is very important and relevant to res,
during or after earthquakes.
In [1] an experimental research on several type of steel joints at
elevated temperatures showed that failure of the steel members
preceded failure of high strength bolts. A number of re tests on
both composite (steel to concrete) and non-composite steel joints
[2], indicated an improved moment-rotation behaviour, for the
composite ones. More recently, full scale re tests on top and seat
angle connections [3], demonstrated that there is a possibility of a
premature failure of bolts in re conditions, while re tests on
high-strength bolts presented in [4], showed that Eurocode 3 gives
a good prediction for the reduction of bolts strength, at elevated
temperatures. Fire tests on different structural materials have been
presented in [5,6].
From another point of view, in [7] three-dimensional nite element models simulating single plate shear connections were
developed, to show how bolts grade, bolts hole and several other
parameters, are inuencing the behaviour of the structure under
re. In [8], an explicit dynamic solver was used. Other research
projects involve study of the cooling phase of a re [9], as well as
implementation of an articial neural network for the description
of the stressstrain relations of steel under re [10].
Corresponding author. Tel.: +30 28210 37418; fax: +30 28210 69410.
E-mail addresses: ankaloger@gmail.com (A. Kalogeropoulos), gdrosopoulos@isc.
tuc.gr (G.A. Drosopoulos), gestavr@dpem.tuc.gr (G.E. Stavroulakis).
URL: http://www.comeco.tuc.gr/ (G.E. Stavroulakis).
0950-0618/$ - see front matter 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2011.11.012

In the present study a three-dimensional non-linear nite element model has been developed, for the simulation of the thermomechanical behaviour of an extended end-plate steel joint.
Parametric numerical investigation is used to study the inuence
of the sequence between thermalmechanical loads, the opening
of the interface as well as the shear forces transmitted through friction and through the bolts.
2. Framework of the non-linear model
Unilateral contact with Coulomb friction is considered at the
interface between the extended end-plate and the column ange,
and allows for possible separation of the connected parts. At each
point of the interface the basic unilateral contact mechanism can
be described by the no-penetration inequality, the no-tension
inequality, as well as a complementarity, either-or relation, indicating that either separation with zero contact force or compressive contact force with zero gap appears.
The arising non-smooth structural analysis problem has the
form of a non-linear complementarity problem. In the tangential
direction a similar either-or, variable structure behaviour concerning the stickslip effects appears. In particular, the behaviour in the
tangential direction is dened by a static version of the Coulomb
friction law. Two contacting surfaces start sliding when the shear
stress in the interface reaches a critical value equal to:

tt scr ljtn j

where tt, tn are the shear stress and the contact pressure at a given
point of the contacting surfaces respectively and l is the friction
coefcient. There are two possible directions of sliding along an
interface, so tt can be positive or negative depending on that

Author's personal copy

620

A. Kalogeropoulos et al. / Construction and Building Materials 29 (2012) 619626

direction. In principle, there is no sliding if jttj < ljtnj (stick conditions). However, for the numerical implementation of the present
computational scheme, the friction constraint is enforced with the
penalty method. Within this method a small sliding (approximately
0.5 mm) is permitted until the shear stress tt becomes critical and
equal to ltn. For the enforcement of the contact constraints (in
the normal direction of the corresponding interface), the Lagrange
multipliers method has been used. The aforementioned scheme
has been numerically implemented within commercial nite element packages like ABAQUS and MARC used here. More details
can be found, among others, in [11,12].

3. The nite element model


The extended end-plate connection was tested to failure at
ambient temperatures [13]. Eight high strength M20 bolts, of
grade-8.8, with average yield and ultimate stresses Fy = 600 N/
mm2 and Fu = 800 N/mm2 obtained from coupon tests (Fig. 1a),
were used. For each test, an IPE-360 beam section was attached
to an HEA-220 column section (Figs. 2 and 3) through an extended
end-plate (Fig. 4). The beam, the column, and the end-plate were
made of steel having average yield and ultimate stresses
Fy = 314 N/mm2 and Fu = 450 N/mm2, respectively, also obtained

from three different uniaxial coupon tests (Fig. 1b). In Fig. 2a the
geometry of the joint and the position of the load, are shown.
Three-dimensional 8-node brick nite elements have been
used. The mesh has 107,326 elements and it is denser around the
area of the connection, see Fig. 2b and 3. For the numerical solution
of the non-linear problem the NewtonRaphson incremental iterative procedure has been used. Within this, force control is considered. In particular, a relatively small initial load increment equal to
2 kN is chosen. For the force and displacement convergence criterion values at 0.005 and 0.01 are used, respectively. For the execution of each load case, two processors at 2.27 GHz have been used
and 1216 h have been needed to complete a single thermomechanical analysis.
The analysis has been performed in two phases. In the rst one,
a heat transfer model has been developed. The results are imported
into the mechanical nite element model and a coupled thermomechanical analysis takes place. It is possible to import the data
obtained from the thermal analysis, into any of the steps developed
within the thermomechanical model. Consequently, different
load sequences between the thermal and the mechanical loads
can be considered. The friction coefcient for the beam-column
interface is taken equal to 0.4. The Youngs moduli for the beam,
the column and the bolts are taken equal to 120 GPa, as it is obtained from relevant coupon tests. The Poissons ratio for the whole

Fig. 1. Coupon tests at ambient temperatures (a) bolts and (b) column, beam, end-plate.

(a)

(b)

Fig. 2. (a) Geometry and positioning of the mechanical load and (b) mesh.

Author's personal copy

A. Kalogeropoulos et al. / Construction and Building Materials 29 (2012) 619626

621

Fig. 3. Structural parts of the connection (a) column, (b) beam, and (c) bolts.

Fig. 4. Geometry of the end-plate and positioning of bolts.

structure is taken equal to 0.3. Moreover, large displacement analysis as well as v. Mises plasticity have been considered.
4. Concept of the thermomechanical analysis
The following aspects of the thermomechanical behaviour of
the steel joint will be studied:
The inuence of the sequence between the thermal and the
mechanical loading.
The contactfriction interface, and in particular different values
for the friction coefcient, the opening of the interface, and the
representation of the shear forces transmitted through the friction and through the bolts.
The consideration of the bolts at elevated temperatures and the
inuence they have on the overall response of the structure.

case, a small point load (50 kN) is followed by the thermal loading;
in the nal step the total mechanical load is applied to the connection. The self-weight has been applied before every one of the
aforementioned loads. It is noted that for the above load cases,
the rate of increase of both the mechanical and the thermal load
is linear, within each analysis step. In other words, values of the
mechanical and thermal load are linearly increased, as the corresponding time increments are increased, throughout each step.
The thermal properties are: thermal conductivity = 45 W/m C,
thermal expansion for the steel parts = 12  106/C, thermal
expansion for the bolts = 13  106/C. Concerning the mechanical
properties of the steel material, degradation of the Youngs Modulus has been considered according to Eurocode 3 [14]. In addition,
the stressstrain laws at elevated temperatures have been considered, as it is represented in Fig. 6b [14].
5. Results and discussion

4.1. Thermomechanical material properties


5.1. Inuence of the sequence of thermal and mechanical loads
Temperature boundary conditions have been applied as shown
in Fig. 5. In addition, heat ux equal to 2 kW/m2 has been applied
to the beams web and to the columns web and front ange, as it is
shown in Fig. 6a.
Three load cases have been considered. According to the rst
one, the thermal and a concentrated mechanical load are concurrently applied in the same analysis step. Within the second load
case the thermal load precedes the point loading. At the third load

The sequence of application of thermal and mechanical loads is


signicant for the overall behaviour and the collapse load of the connection. In case the thermal and the point load are concurrently applied to the structure, temperatures reach approximately 800 C at
the collapse of the joint, Fig. 7a. Temperatures reach similar values
for the third load case, where thermal loading follows a small
concentrated load of 50 kN. In both cases the connection shows a

Author's personal copy

622

A. Kalogeropoulos et al. / Construction and Building Materials 29 (2012) 619626

Fig. 5. Temperature boundary conditions.

Fig. 6. (a) Thermal loading: heat ux and (b) degradation of the stressstain laws of the steel due to elevated temperatures (C).

satisfactory strength to re although the limit load is signicantly


reduced in comparison with the structure at ambient temperatures.
From the loaddisplacement diagrams of Fig. 8 it is shown that
the behaviour of the joint is optimum in case no thermal load is applied to it. In addition, there is a satisfactory convergence between
the diagram obtained from the experimental research of the same
connection at ambient temperatures [13] and the one received
from the nite element model. The latter reaches higher deformation levels than the experimental one, however the yielding of the
structure has been expanded enough at load levels of 125 kN, thus
close to failure load obtained from the experiment. In case the
thermal load is applied in the same step with the point load, the
relevant forcedisplacement diagram indicates a signicant reduction of the ultimate strength of the structure.
The behaviour of the steel joint changes drastically when the
thermal load is applied rst and the mechanical point load follows

NT11
786
720
655
589
524
458
393
327
262
196
131
65
0

in the next step of the analysis. In this case the reduction of the diagram indicates a severe degradation of the strength of the connection. The elevated values of the temperatures developed at the end
of the thermal and before the application of the point load is a reason for this, Fig. 7b. Finally, if a small mechanical load equal to
50 kN is initially applied and the thermal load follows, the force
displacement diagram (bold line of Fig. 8) is almost identical with
the one obtained from the concurrent application of both thermal
and mechanical loads. However, when a mechanical load is applied
again in a third load step, no result is obtained indicating that the
joint has already reached its ultimate strength.
For the concurrent application of the thermal and the mechanical
load, plastic regions at failure have been expanded to the whole
beam and to the columns web and ange, near the connection.
In case the thermal load is initially applied together with the
self-weight and the point load follows, yielding of the joint is

NT11
2192
2010
1827
1644
1461
1279
1096
913
731
548
365
183
0

(a)

(b)

Fig. 7. Temperature distribution (a) concurrent application of point-thermal load-end of the analysis and (b) point load after thermal load-end of the thermal load (C).

Author's personal copy

623

A. Kalogeropoulos et al. / Construction and Building Materials 29 (2012) 619626

Fig. 8. Forcedisplacement diagrams.

expanded to the whole beam at the end of the thermal loading


step, before the application of the point load. This is a result of
the particular loading sequence which has been followed. Selfweight and thermal load cause yielding to the beam, before the
application of the concentrated load. As a result, when the point
load is applied only a small part of it can be supported by the structure. For the third load case where the thermal load is applied after
a small point load equal to 50 kN, failure occurs before the end of
the thermal loading step and expands to the whole beam. Consequently, the ultimate load is equal to 50 kN and it is smaller from
the one received from the rst load case (Fig. 8).

5.2. Behaviour of the contact-friction interface


At failure the structure collapses mainly due to yielding of the
columns ange in the region of the joint. Moreover, as the opening
of the interface connecting the columns ange with the beams
end-plate increases, yielding of the columnss ange tends to be
more severe. For this reason, the opening of the interface is very
important for the overall behaviour of the structure.
According to Fig. 9a, for the case that both thermal and mechanical loads are concurrently applied, opening of the interface
reaches 18.7 mm. Opening of the interface at the end of the second

COPEN
0.0187
0.0171
0.0156
0.0140
0.0125
0.0109
0.0094
0.0078
0.0062
0.0047
0.0031
0.0016
0.0000

load case (rst thermal then point load) is bigger than the one obtained from the rst load case, thus equal to 38.7 mm (Fig. 9b).
An interesting observation is that at the end of the thermal load
step and before the application of the point load, a small opening in
the interface is developed, see Fig. 9c.
For the third load case (rst small point load then thermal load),
opening at the end of the analysis becomes approximately three
times bigger than the one received before the application of the
thermal load (11.8 mm instead of 3.5 mm, see Fig. 10b and a). This
is due to re conditions, as in the nal step of the analysis the point
load is constant and equal to 50 kN.
It is worth noticing that the bigger the opening of the interface,
the smaller the strength of the joint (Figs. 8 and 9). This is attributed to the fact that connections failure is caused by the yielding
of the columns ange. As the opening is increased due to elevated
temperatures, yielding of the columns ange becomes more severe and the strength of the structure is reduced.
Furthermore we investigate the shear forces transmitted
through the interface by friction and through the bolts. In the
framework of nite element analysis the total friction force of
the contact interface is calculated as the integration of frictional
stresses at the interface. The shear forces transmitted through bolts
can be calculated similarly or, for higher accuracy, by using the free
body diagram of the beam.

COPEN
0.0387
0.0355
0.0323
0.0291
0.0258
0.0226
0.0194
0.0161
0.0129
0.0097
0.0065
0.0032
0.0000

(a)

COPEN
0.0010
0.0009
0.0008
0.0007
0.0006
0.0006
0.0005
0.0004
0.0003
0.0002
0.0002
0.0001
0.0000

(b)

(c)

Fig. 9. Opening (m); (a) concurrent application of point-thermal load, (b) point load after thermal load-end of the analysis, and (c) point load after thermal load-end of the
thermal load.

Author's personal copy

624

A. Kalogeropoulos et al. / Construction and Building Materials 29 (2012) 619626

COPEN
0.0035
0.0032

COPEN
0.0118
0.0108
0.0098

0.0029
0.0026
0.0023

0.0088
0.0079

0.0021
0.0018

0.0069
0.0059

0.0015
0.0012
0.0009
0.0006

0.0049
0.0039
0.0029

0.0003
0.0000

0.0010
0.0000

0.0020

(a)

(b)

Fig. 10. Opening (m) (a) point load before thermal load-end of the point load and (b) point load before thermal load-end of the analysis.

Fig. 11a shows the proportion of shear forces transmitted


through the bolts and through friction, in case no thermal load is
applied to the connection. For a value l = 0.4 of the friction coefcient, the shear force transmitted through bolts is greater than the

friction force almost for the whole range of the analysis (this difference is smaller, for low values of external load). Contrary to
Fig. 11a, Fig. 11b shows that for the concurrent application of both
the thermal and the mechanical load, the friction force is greater

Fig. 11. Shear forces at the interface (a) no re conditions and (b) rst load case of the thermomechanical model.

Fig. 12. Shear forces at the interface (a) second and (b) third load case of the thermomechanical model.

Author's personal copy

A. Kalogeropoulos et al. / Construction and Building Materials 29 (2012) 619626

625

Fig. 13. Forcedisplacement diagrams (a) degradation of the stressstain laws for the bolts and (b) variation of the thermal properties.

than the shear force of bolts, almost for the whole range of the
analysis. In addition, the friction force becomes bigger than the
one obtained from the model without any thermal loading. The rise
of the normal force in the interface, due to the expansion of the
steel at elevated temperatures, could be an explanation for this.
On the other hand, the shear force of bolts becomes lower than
the one received from the pure mechanical analysis.
A signicant alteration in the behaviour of the interface is obtained in case the second load case is considered, thus the thermal
load together with the self-weight are applied in a rst analysis

step and the mechanical load follows in a second step. Plots of


Fig. 12a are referred to the second step of the analysis, where the
application of the thermal load has been completed and the application of the point load begins.
Contrary to the previous diagrams, they are non-zero in the
beginning of the step. The initial non-zero friction force is attributed to the effect of the thermal load on the joint and it is in conjunction with the image of the interface of Fig. 9c, according to
which the interface becomes active at the end of the rst analysis
step.

Fig. 14. (a) Initiation of the yielding at the bolts and (b) yielding at the columns ange for the same load level.

Author's personal copy

626

A. Kalogeropoulos et al. / Construction and Building Materials 29 (2012) 619626

The diagrams of Fig. 12b represent the shear forces of bolts and
the friction force of the interface for the third load case, where a
small mechanical load equal to 50 kN is followed by the thermal
loading.
Finally, an analysis demonstrating the inuence of the rise of
the friction coefcient on the behaviour of the interface has been
conducted. In particular, the value l = 0.6 of it is considered, for
the rst load case where both the mechanical and the thermal load
are applied in the same analysis step. According to the results, the
overall strength of the joint remains the same with the one obtained when l = 0.4, thus the forcedisplacement diagram is identical with the one presented in Fig. 8. However, the friction force
transmitted through the interface is increased while the shear
force of bolts is reduced.
The present results show that the interaction between friction
and shear action at the bolts is a complicated mechanism and consequently it is difcult to replace a three-dimensional analysis
with some simplied model.

5.3. Alteration of the material properties of the joint due to re


In this section thermal-dependent material properties will be
considered for the bolts, according to Eurocode 3 [14]. Reduction
factors adopted in Eurocode 3 seem to suit well with the results
obtained from the experimental investigation on high strength
bolts under re, referred to [4]. The rst load case, thus the concurrent application of the thermal and the mechanical load has been
chosen.
According to the forcedisplacement diagrams of Fig. 13a, the
strength of the connection received from this consideration, is almost identical with the one obtained by the model without any
degradation in the stressstrain relations of bolts. A slightly modied model, where only the bolts stressstrain laws have been reduced due to elevated temperatures, has been developed. Within
this, the material law of the steel parts is not reduced due to re.
The relative forcedisplacement diagram demonstrates an increase
in the ultimate load of the connection, see Fig. 13a. Consequently,
the fact that the strength of the joint is increased if degradation of
the material law is considered only for the bolts and not for the
steel parts, indicates that re conditions on the bolts have a lower
impact on the connections behaviour than re conditions on the
steel parts.
In addition, analysis results show that failure of the steel parts
occurs rst and failure of the bolts follows. At Fig. 14 is shown that
at the value 46 kN of the external mechanical load, failure has just
arisen at the upper bolts while failure of the column ange has already started and spread at the interface. This result is in agreement with some relevant experimental and numerical research
studies referred to [1,2,15], according to which failure of the steel
members precedes failure of high strength bolts.
This section is completed with a third analysis in which the
thermal material properties of the structure, thus conductivity
and thermal expansion, are considered to be a function of the rise
of the temperature, according to Eurocode 3 [14]. Goal of this
investigation is to reveal how strongly the strength of the connection is altered at elevated temperatures. Fig. 13b shows that the
strength of the joint obtained from this model becomes lower than

the strength received from the model with thermal material properties independent of temperatures, after a load level of 55 kN.
6. Conclusions
A thermomechanical analysis of a bolted steel joint has been
performed. From the results, one concludes that the strength of
the connection is signicantly reduced in case the thermal load
precedes the application of the mechanical one. In addition, an
opening in the interface connecting the column ange with the
beams end-plate appears, when only the thermal load is present.
The behaviour of the joint is improved, if the mechanical and
thermal load are concurrently applied or if a small point load precedes the application of the thermal one. However, in both cases
the limit load of the connection still remains quite lower than
the one obtained from the pure mechanical analysis.
Moreover, the friction force of the interface is increased in comparison with the one obtained from the model without any thermal
loading, in case the thermal and the point load are concurrently applied. The rise of the normal force in the interface, due to the expansion of the steel at elevated temperatures, could be an explanation
for this. On the other hand, the shear force of bolts becomes lower
than the one received from the pure mechanical analysis.
Finally, results obtained from the developed models show that
failure of the steel parts seems to precede failure of the bolts in re
conditions. This conclusion is in agreement with some experimental and numerical research studies conducted in the past.
References
[1] Kruppa J. Rsistance en feu des assemblages par boulous. CTICM Report,
Document No. 1013-1, Centre Technique Industrial de la Construction
Metallique, St. Remy les Chevreuse, France; 1976.
[2] Lawson RM. Behaviour of steel beam-to-column connections in re. Struct Eng
1990;68:26371.
[3] Daryan AS, Yahyai M. Behavior of bolted top-seat angle connections in re. J
Constr Steel Res 2009;65:53141.
[4] Kuo-Chen Y, Re-Jia H, Yan-Jun C. Shear strength of high-strength bolts at
elevated temperature. Constr Build Mater 2011;25:365660.
[5] Khan MS, Prasad J, Abbas H. Shear strength of RC beams subjected to cyclic
thermal loading. Constr Build Mater 2010;24:186977.
[6] Heo YS, Sanjayan JG, Han CG, Han MC. Construction application of bre/mesh
method for protecting concrete columns in re. Constr Build Mater
2011;25:292838.
[7] Selamet S, Garlock ME. Robust re design of single plate shear connections.
Eng Struct 2010;32:236778.
[8] Yu H, Burgess IW, Davison JB, Plank RJ. Numerical simulation of bolted steel
connections in re using explicit dynamic analysis. J Constr Steel Res
2008;64:51525.
[9] Lien KH, Chiou YJ, Wang RZ, Hsiao PA. Nonlinear behavior of steel structures
considering the cooling phase of a re. J Constr Steel Res 2009;65:177686.
[10] Hozjan T, Turk G, Srpcic S. Fire analysis of steel frames with the use of articial
neural networks. J Constr Steel Res 2007;63:1396403.
[11] Panagiotopoulos PD. Inequality problems in mechanics and applications. In:
Convex and nonconvex energy functions. Boston, Basel, Stuttgart: Birkhauser
Verlag; 1985.
[12] Mistakidis ES, Stavroulakis GE. Nonconvex optimization in mechanics.
Algorithms, heuristics and engineering applications by the FEM. Dordrecht,
Boston: Springer/Kluwer; 1998.
[13] Abdalla KM, Abu-Farsakh GAR, Barakat SA. Experimental investigation of
force-distribution in high-strength bolts in extended end-plate connections.
Steel Comp Struct 2007:7.
[14] Eurocode 3. Design of steel structures part 1.2: general rules structural re
design ENV 1993-1-2. Brussels (Belgium): European Committee for
Standardization; 2001.
[15] Al-Jabri KS, Seibi A, Karrech A. Modelling of unstiffened ush end-plate bolted
connections in re. J Constr Steel Res 2006;62:1519.

Anda mungkin juga menyukai