The Second International Conference on

Numerical analyses of improved beam-to-column connections

Marco Valente1) and Maria Gabriella Mulas2)

1),2)
Department of Structural Engineering, Politecnico di Milano, Milano, Italy
1)
valente@stru.polimi.it, 2)mulas@stru.polimi.it

ABSTRACT

This paper presents the results of non-linear finite element analyses on two types of improved
steel moment-resisting connections, either reinforced with cover plates or weakened with
RBS detail: the aim is to protect the potentially vulnerable beam-to-column groove welded
joint by relocating the plastic hinge away from the face of the column. The investigation is
based on results from experimental tests carried out on a two-storey steel frame tested at Ispra
and involves the modeling of the 3-way beam-to-column joints. Numerical analyses on
reinforced joints show that fracture indices decrease in the weld region and near the weld
access hole respect to the unreinforced connection, but high values of the Triaxiality Index are
registered in the weld at the column interface. In order to eliminate the high shear stress in the
beam flange and to reduce the high triaxial stress condition at the beam flange-column flange
interface, the beam flange was disconnected from the column flange. The reduced beam
section detail provided both in the top and bottom beam flanges reduces the vulnerability of
the connection to the occurrence of a fracture near the weld: yielding is concentrated within
the reduced section of the beam where large inelastic strains are registered.

INTRODUCTION

Brittle fractures in beam-to-column connections of welded steel moment frame structures

were observed during recent earthquakes in California and Japan. Damage consisted in
fractures crossing the beam bottom flange, generally arisen from the beam-to-column
connection welds. In an experimental study several beam-to-column joint specimens were
tested up to failure at the Department of Structural Engineering (DIS) of Politecnico di
Milano, within the European Steelquake Research Project. The specimen failure was
generally due to fractures originating from the beam flange-column flange weld. Excessive
ductility demand on welded connections was deemed one of the primary causes of the damage,
both on field and in the laboratory. The presence of welds does not allow for the development
of large inelastic flexural strain into plastic hinges, limiting the local ductility of the
connection in terms of rotational capacity. The region of the connection near the face of the
column may be vulnerable to fracture for low toughness weld metal, presence of notches
caused by weld defects, high levels of stress in the vicinity of the beam flange weld and at the
toe of the weld access hole, conditions of restraint which inhibit ductile deformation.
Improvements in details are required in order to reduce stress levels in the beam flange
near the groove weld. The use of high toughness materials alone does not provide sufficient

234
fracture resistance to ensure the levels of ductility required for seismic design. New configu-
rations of connection details are developed to improve the available rotation capacity of the
beam-to-column connections. Two main approaches are adopted to minimize the potential for
fracture in the weld region: strengthening the connection in order to reduce the beam flange
stress or limiting the beam moment at the column face. The first method employs the
approach of strengthening the connection and thereby forcing inelastic action to take place in
the beam section away from the face of the column and the flange welds. The second method
seeks to limit the moment at the column face by reducing the beam section, and hence the
plastic moment capacity, at some distance from the column. These procedures are used for
new constructions and can be also applied to the retrofit of connections in existing buildings.
The general configuration of the cover-plate and the RBS connections is shown in Fig. 1.
By adding cover plates at the top and bottom beam flanges, the level of axial stresses in the
beam flanges is reduced. Thus, the overload of the beam flanges near the connection is
reduced and the location of the beam plastic hinge is shifted to the end of the cover-plate.
Two plate-reinforced connections are analyzed in this study: the cover-plate and the flange
plate connections. The main difference between them is in the way of joining the beam flange
to the column flange. For the cover-plate connection, both the cover plate and the beam flange
are welded to the column flange, whereas the plate only is welded to the column in the flange-
plate connection. Finite element analyses are carried out to identify zones of high stress and
strain in different connections, to understand the likely sequence of yielding and determine
whether the stress and strain conditions in reinforced connections are substantially better than
in unreinforced connections.
Of the many alternative connections that have been proposed since the Northridge
earthquake, one of the most reliable and economical is the reduced beam section (RBS)
connection, which is also schematically illustrated in Fig. 1. In a RBS connection a portion of
the beam flange is selectively trimmed in the region adjacent to the beam-to-column
connection in order to force plastic hinging to occur within the reduced section. The stress
state near the beam-to-column connection is conveniently reduced and plastic deformations
are forced in the beam, away from the column face.

Fig. 1. Details of cover-plate and RBS connections

The investigation described in this paper is based on previous experimental tests carried
out on unreinforced connections at the Politecnico di Milano, DIS and at the ELSA
Laboratory of the JRC in Ispra and on experimental results obtained in recent research
programs performed on cover-plate and RBS connections in the USA. Within the

235
“Steelquake” European Research Project, an extensive experimental campaign was carried
out at the ELSA Laboratory of the JRC in Ispra. It included pseudo-dynamic tests and cyclic
tests up to failure on a real scale steel moment resisting frame prototype. The test specimen
was a two-storey, one bay, moment resisting steel frame, as shown in Fig.2. The cyclic test
was performed by imposing a displacement time history at the second storey of the structure;
displacements were amplified up to failure of the specimen. The cyclic history consisted of a
set of cycles at the reference ductility level, alternated with cycles of increasing ductility.
Three of the eight beam-to-column connections failed during the cyclic tests, showing similar
failure modes, with fractures in the plane of the beam flange weld. No evidence was found of
tearing of the column flange material, cracking or substantial buckling of the beam flange; the
fracture propagated through the web of the beam, fig. 2. The panel plate contributed in some
measure to the plastic deformation of the connection.

Fig. 2. A view of the frame tested at Ispra and a typical brittle fracture observed in the beam bottom
flange during experimental tests.

A finite element model of an exterior beam-to-column joint was created using the finite
element computer code Abaqus. The sub-assembly, shown in Fig. 3, was defined by theore-
tical inflection points in the beam and column when the frame is subjected to horizontal loads.
The column is vertical and subjected to an imposed horizontal monotonic displacement at the
upper free edge. The main purpose of this work is to study the stress and strain distribution in
standard beam-to-column connections and to compare the results to the case of alternative
connections.

 22




Fig. 3. Scheme of the exterior connection sub-assembly

236
FAILURE INDICES

The study is based on the definition of the ductile rupture strain:

ε cr = a × e (−1.5×TI ) (1)
where a is a material constant and TI is the Triaxiality Index, defined as the ratio between the
Hydrostatic stress and the Mises stress:
σ hyd
TI = (2)
σ mis
The ductile rupture strain derives from an experimental study on steel specimen subjected to
multi-axial stress states (Hancock and Mackanzie 1976). It depends on the state of stress and
on the material properties and it is a measure of potential for ductile fracture. When the tensile
hydrostatic stress increases, the rupture strain decreases.
The Rupture Index (RI) is the ratio between the PEEQ Index and the rupture strain:
PEEQI
RI = −1.5×TI (3)
e
The equivalent plastic strain index (PEEQI) is a measure of the local plastic strain demand:
PEEQ
PEEQI = (4)
εy
where εy is the yield strain and PEEQ is defined as:
2 p p
PEEQ = ε ij ε ij (5)
3

STANDARD MODEL

Three-dimensional models of the beam-to-column joints were developed using the non
linear finite element analysis program ABAQUS. The models are single-sided steel beam-
column sub-assemblies that are representative of the exterior beam-column connections of the
frame tested at Ispra; the column section is HEB300 and the beam section is IPE400. The
connection is modeled with 8-node solid elements and finer mesh is arranged close to the
joint area. The analyses account for material non linearity, through the classical metal
plasticity theory based on the Von Mises yield criterion, and for the effects of strain hardening.
Isotropic hardening is assumed for monotonic analyses whereas kinematic hardening is
assumed for cyclic analyses. Data from tests coupon extracted from the beam and column
were used to establish the stress-strain relationships for the beam and column elements: a
multi-linear model with strain hardening was considered.
The beam, column and plates are discretized using three-dimensional brick elements.
These components are joined by constraining the nodes at common locations to have identical
displacements. Weld access holes, doubler plates, continuity plates are included in the model,
as shown in fig. 4. The weld access hole was present only in the bottom beam flange. The
beam K-line region was modeled with 2 layers of elements to provide a smooth transition
between beam web and beam flange elements. The presence of the K-stiffeners in the panel
zone is accounted for by increasing the column thickness (15 mm), so that the stiffness of the
uniform thickness panel zone is equal to the stiffness of the real joint.

237
 Fig. 4. Finite element model of the standard connection and detail of the beam bottom flange

The results of non linear analyses here reported are related to the value of displacement of
150 mm at column tip. The PEEQ distribution of the traditionally designed welded beam-to-
column connections is investigated near the weld access hole and at the beam-column
interface. Fig. 5 shows a concentration of plastic strain at the toe of the weld access hole and
on the bottom surface of the beam flange. At the column interface PEEQ is maximized at the
edge of the beam flange. High values of PEEQ are recorded in the beam web around the weld
access hole. Stress and strain concentrations, resulting from the analyses, at the root of the
access hole are correlated with the locations and types of low-cycle fatigue connection
failures seen in the tests where some fracture originated in the beam flange at the root of the
weld access hole. The abrupt change in geometry at the toe of the access hole creates a stress
and strain concentration in the beam flange: the combination of stress concentration, high
strain demand and low fracture toughness can lead to premature fracture initiation.

Fig. 5. PEEQ Index distribution along the beam bottom flange and contour of PEEQ in the model.

The analyses show a high triaxial tensile stress state at the beam flange near the column
interface, which may lead to premature fracture prior to significant ductile yielding and local
buckling. The maximum value of the Triaxiality Index in the inelastic range is recorded at the
beam web centerline. The maximum Triaxiality Index is about 0.7: this value may cause a
reduction in the rupture strain. Triaxiality Index is greater at the column interface than on the
line passing through the toe of weld access hole. The values of the Triaxiality Index indicate
that the initiation of brittle fracture is likely in the beam flange near the beam web at the face
of the column. The value of the Rupture Index near the toe of the weld access hole is higher
than the value in the beam flange at the face of the column: much smaller values are recorded
in the beam flange far from the beam web centerline. At the face of the column, the maximum
values of the RI occur at the edge and at the beam web centerline. The peak values of the
considered failure indices are shown in Table 1.

238
Table 1. Peak values of the failure indices in the standard model
At the column interface At the weld access hole
PEEQI TI RI PEEQI TI RI
47 0.7 84 62 0.45 135

COVER-PLATE MODEL

Different finite element models of cover-plate connections are generated from the
unreinforced model, where high stress concentration regions that may lead to potential
fracture are identified. The connection is reinforced by cover plates both at the top and bottom
beam flanges with approximate length of h/2, where h is the beam depth, and with thickness
equal to the beam flange thickness. The reinforcing plates are designed to relocate the plastic
hinge beyond the edge of the cover plate and away from the face of the column: they are
modeled with two elements through the thickness. Fillet welds at cover plates are modeled;
the beam flange and the cover plate edge are joined by constraining the nodes at common
locations to have identical displacements: the plates are joined to the beam flanges on three
sides. The longitudinal and transverse fillet welds that join the plate to the flange are modeled
using two layers of six-node and eight-node solid elements. Fig. 6 shows a view of the
longitudinal fillet weld between the beam flange and the cover-plate. Surface contact between
the beam flange and the reinforcing plate is accounted for in this study. The continuity plate
thickness is equal to the sum of the beam flange and cover-plate thickness. These models are
not used to capture the beam flange and web local buckling; also in this section the results are
calculated at 150 mm tip column displacement.

Fig. 6. Finite element model of the cover-plate connection and detail of the beam bottom flange.

In the model with cover-plates the maximum value of the PEEQ Index is 16 and it is
registered at the edge of the cover-plate, near the end of the longitudinal fillet weld. The peak
value of the PEEQ Index (16) in the cover-plate model is smaller than the peak value (62) in
the unreinforced connection. In particular the values of the PEEQ Index decrease significantly
near the weld with respect to the unreinforced model. At the face of the column the value of
the PEEQ Index is maximum at the bottom of the cover-plate and it is less than 4; near the
weld access hole the value of the PEEQ Index is 1. The critical sites for the two models are
different: in the cover-plate model the maximum value is registered at the edge of the cover
plate in the beam flange, while in the unreinforced model the maximum value is at the root of
the weld access hole. In the presence of cover-plates the heat affected zone and the weld
access hole region are less sensitive to low-cycle fatigue failures.
Fig. 7 shows a contour plot of the PEEQ in the cover plate model. The sequence of
yielding in the cover-plate model is different from the sequence in the unreinforced model.
The region near the edge of the cover-plate yields first followed by the panel zone and then by
the beam flange-column flange interface.

239
Fig. 7. PEEQ Index distribution at the edge of the cover-plate and contour of PEEQ in the model.

The maximum values of the Triaxiality Index are registered at the column interface. In the
cover-plate model the peak value is higher (0.86) than in the unreinforced connection (0.65)
and it is registered at the bottom surface of the beam flange above the cover-plate. In the
beam flange at the edge of the cover-plate, the Triaxiality Index is small and the maximum
value is registered at the center of the transversal weld. The maximum value of the Rupture
Index is registered at the middle of the line at the beam flange-column flange interface near
the end of the longitudinal weld, far from the connection region. The peak values of the
Rupture Index in the model with cover-plates are smaller than in the unreinforced model, in
particular at the weld-beam interface and in the region near the weld access hole, as it is
shown in the following Table 2.

Table 2. Peak values of the fracture indices in the cover-plate model

At the column interface At the weld access hole At the edge of the
cover-plate
PEEQI TI RI PEEQI TI RI PEEQI TI RI
4 0.86 8 1 0.55 2 16 0.62 32

FLANGE-PLATE MODEL

The analyses on cover-plate models have indicated the weld connecting the beam flange to
the column flange as critical sites for fracture potential. In order to eliminate the high shear
force in the beam flange and to reduce the high triaxial stress condition at the column
interface, a joint arrangement having the beam flange disconnected from the column flange
has been analyzed. The general configuration of the flange-plate connection is shown in Fig. 8.
The flanges of the beam in the connection are not welded directly to the column and the
stresses are transferred through the flange plates. Such configuration substantially reduces the
deformation restraints in the beam; in addition, the requirements of welding for the flange
plates are not restrictive. The flange plate thickness is assumed of 18 mm.
A few models of flange plate connections were developed from the basic model of the
cover plate connection. In the numerical model they were obtained by deleting the elements
connecting the beam flange to the column flange. Finite element analyses were carried out to
determine whether the changes in plate geometry and details substantially modify the values
of the selected fracture indices.

240
Fig. 8. Finite element model of the flange-plate connection and detail of the beam bottom flange.

The maximum value of the Peeq Index is recorded at the bottom surface of the beam flange,
at the edge of the transverse fillet weld and at the end of the longitudinal fillet weld (Fig. 9).
The maximum value of the Peeq Index at the column-interface is significantly smaller than
the maximum value in the unreinforced connection and slightly greater than the maximum
value in the similar cover-plate connection.

Fig. 9. Peeq Index distribution at the edge of the flange-plate and contour of PEEQ in the model.

The maximum value of the Triaxiality Index is registered at the column interface, at the
edge of the reinforcing plate in the beam flange. The peak value (0.77) is higher than the
maximum value in the unreinforced model, but smaller than in the cover-plates model (0.86).
The maximum value of the Rupture Index is found at the end of the longitudinal beam flange-
flange plate weld. This result is desirable because the objective of the flange-plate connection
is to force inelastic action in the beam away from the column face. The maximum value of the
Rupture Index is much smaller than the maximum value in the unreinforced connection and
slightly smaller than the maximum value in the similar cover-plate connection. The peak
values for all the afore mentioned indices are shown in Table 3.

Table 3. Peak values of the fracture indices in the flange-plate model

At the column interface At the weld access hole At the edge of the
flange-plate
PEEQI TI RI PEEQI TI RI PEEQI TI RI
7 0.77 11 1 0.3 1 13 0.64 26

In the unreinforced connections the welding zone and the region near the weld access hole
in the beam flange are critical for fracture potential; the PEEQ Index and the Rupture Index
assume high values and the likelihood of failure is high if some flaws are present. When
cover-plates are used, in the weld access hole zone and near the weld the PEEQ Index and the
Rupture Index decrease, whereas the Triaxiality Index increases above all at the column
interface. The critical region shifts to the edge of the cover plates where the PEEQ Index and

241
the Rupture Index are maximum. Moreover some problems concern the high values of the
Triaxiality Index in the weld, the high shear stresses in the beam flange and the modality of
welding. In the flange-plate model the hydrostatic stress and the Rupture Index in the weld
zone are smaller than in the cover-plate model and the welding is less problematic.

RBS MODEL

In the standard connection model, Rbs cuts are provided both in the top and bottom beam
flanges. Various profiles have been tried for the reduced beam section. The geometry of the
removed flange section may be a constant cut, a tapered cut or a radius cut. In this study the
RBS profile is curve shaped. Compared to the constant cut or tapered cut, which have
reentrant corners, the radius cut tends to minimize stress concentrations within the reduced
section. Previous experimental tests showed that constant and tapered cut RBS specimens
may fracture in the RBS while the curved RBS behaves with the highest rotational capacity.
The geometry of curve shaped dog-bone is shown in Fig. 10.

Fig. 10. Geometry of the RBS detail

The RBS design methodology based on FEMA procedure was used in this study. The length
of the dog-bone “b”, the distance of the start of the RBS cut from the column face “a”, the
depth of the RBS cut at its minimum section “c” are the key dimensions that must be chosen
by the designer. Based on successful past tests the following suggestions are made for these
dimensions:
a = (0.5 − 0.75)b f (6)

b = (0.65 − 0.85)d (7)

where bf and d are the beam flange width and depth respectively. The value of c should be
chosen to be less than 0.3 bf. The radius of the cut R is related to dimensions b and c by the
following equation:
4c 2 + b 2
R= (8)
8c
In this study the maximum practical cutout corresponding to 55% flange width reduction is
employed in the RBS design procedure and the finite element model is shown in Fig. 11.
Flange reduction larger than 55% may risk impairing the stability of the beam and for this
reason it is not recommended without experimental tests.

242
 Fig. 11. Finite element model of the RBS connection and detail of the beam bottom flange

The PEEQ Index presents much smaller values near the weld in the RBS model than in the
standard model. Very low values of the PEEQ Index are registered in the panel zone of the
RBS model. The peak value (39) is registered in the weakened section far from the region of
the weld, Fig.12.

Fig. 12. PEEQ Index distribution in the reduced beam section and contour of PEEQ in the model

In the RBS model the value of the Triaxiality Index is maximum (0.55) along the beam
flange-column flange interface and the peak value is smaller than in the standard model. The
values of the Rupture Index registered at the edge and at the middle of the column-weld
interface in the RBS model are much lower than the standard model. In the reduced section
the value of the Rupture Index is maximum (66). Peak values for failure indices are shown in
Table 4.
Table 4. Peak values of the fracture indices in the RBS model
At the column interface At the weld access hole At the RBS section
PEEQI TI RI PEEQI TI RI PEEQI TI RI
1 0.55 1 1 0.38 1 39 0.38 66

DISCUSSION OF THE NUMERICAL RESULTS

Cover-plate model
Experimental tests and numerical analyses demonstrated that the presence of the weld
access hole in standard beam-to-column connections can cause stress and strain concentration
and improved geometry details of the hole were investigated (Valente, 2002). In reinforced
connections the presence and the geometry of the weld access hole is unimportant because the
value of the stress and strain indices are negligible in this region because of the presence of
the reinforcing plates. However, the weld configuration of the beam flange and the cover-
plate to the column flange may affect the potential for fracture, as evidenced in some
experimental tests carried out at the University of Texas (Engelhardt, 1998). In the cover-

243
plate model the connection geometry may produce a stress increase near the face of the
column flange at the interface of the reinforcing plate and the beam flange. This stress
increase, along with a triaxial state of tension at the face of the column, could lead to
dangerous fractures, especially if the weld is composed of two groove welds, each with a root
pass. The numerical analyses show maximum values of the brittle fracture index, the
Triaxiality Index, at the bottom surface of the beam flange above the cover-plate, in the root
pass of the cover plate-column flange groove weld. The cracking may be caused by the notch
provided by the contact surface of the cover-plate and the beam flange; at this interface there
is significant stress concentration both in the beam flange and the cover-plate.
In the cover-plate connections high shear stresses are registered in the beam flange at the
column interface. The maximum value of the shear stress in the cover-plate model is higher
than in the unreinforced model. The addition of the plates to the unreinforced connection
reduces the shear transferred through the beam web to the column flange. When the thickness
of the cover-plate increases, the transferred shear force through the beam flange and the
cover-plate increases.
Different values of the thickness of the continuity plates were used in the cover-plate
model. When the stiffener thickness is equal to the beam flange thickness, numerical results
evidenced a stress and strain concentration in the column web, beneath the continuity plates.
If the thickness of the continuity plate is equal to the sum of the beam flange and the cover-
plate, the stress and the strain indices diminish in the column web.

Flange-plate model
The flange-plate connection is introduced in order to eliminate the high shear stress in the
beam flange and to reduce the value of the Triaxiality Index at the column interface registered
in the cover-plate model. In the flange-plate connections the region near the weld access hole
is not critical because the value of the fracture indices are low. The plate thickness is selected
in order to move the assumed location of the plastic hinge to the edge of the flange plate.
Different values of the flange plate thickness are analyzed. When the flange plate thickness is
reduced to promote yield penetration into the flange plate, both the Rupture Index and the
PEEQ Index increase near the weld. However the use of thick reinforcing plates to reduce the
stress and strain state at the column face is not recommended because large fillet welds will
be needed to join the plate to the flange and much of the beam shear will be transferred to the
column through the plates. When the thickness of the flange plate increases, the Triaxiality
Index at the column interface becomes critical and the values of the indices at the edge of the
plate grow.
When the doubler plates reinforcing the panel zone are removed, the values of the fracture
indices in the beam flange decrease because the plastic deformations in the beam are
substantially reduced due to yielding of the panel zone. The large contribution of the panel
zone deformation to the drift of the specimen reduces the plastic deformation demand on the
beam, and consequently reduces the plastic strain demands on the beam at the edge of the
cover plate. The likelihood of fracture of the beam flange at the edge of the flange plate is
remote. In the absence of doubler plates, the critical section is the flange plate at the face of
the column where the maximum value of the Rupture Index is recorded.

RBS model
The stress and strain conditions at the beam flange-column flange interface are less severe
in the RBS model than in the standard model because the Mises stress and the Peeq Index
present smaller values in the model including the RBS detail. The Triaxiality Index is
maximum along the beam flange-column flange interface, but the peak value is smaller than
the one measured in the standard model. The likelihood of fracture, represented by the

244
Rupture Index, occurring at the beam flange groove welds and in the surrounding base metal
region is reduced. However the Rupture Index presents high values in the reduced zone of the
beam flange. The use of the doubler plates strengthens the column web and the concentration
of the plastic deformation occurs mainly in the reduced beam section. Moreover, in this study
a flange width reduction of 55% was employed in the RBS design procedure, representing the
maximum limit to the amount of beam flange material which can be removed; in this case the
plastic deformation concentrates in the reduced beam section and the panel zone remains in
the elastic range. In the RBS models smaller concentration of plastic deformation occurs in
the panel zone than in the standard model. The reduced beam section appears attractive
because the plastic demands on the column and panel zone diminish. When the beam flange
section is reduced weakening the beam in bending with the aim to move the formation of the
plastic hinge in the reduced zone, the weak beam-strong column requirement is easier to
satisfy.

CONCLUSIONS

Experimental tests and numerical analyses indicate the beam flange weld near the column
flange as a critical site for fractures in standard beam-to-column connections. The use of
reinforcing plates placed on the beam flange reduces the likelihood of ductile and brittle
fracture of steel moment resisting connections with respect to the unreinforced connections.
The cover-plates decrease the stress and strain states in the region of the weld and near the
weld access hole and shift the formation of the plastic hinge to the end of the cover-plates.
However a critical zone is the middle of the weld which connects the cover plate and the
beam flange to the column flange where numerical analyses indicate high values of the
Triaxiality Index.
In order to eliminate the high shear stress in the beam flange and to reduce the high triaxial
stress condition at the column interface, the beam flange was disconnected from the column
flange. In the flange-plate model the Triaxiality Index in the weld zone and the Rupture Index
at the edge of the plate are smaller than in the cover-plate model. The plastic deformation of
the connection comes from yielding near the edge of the steel plate and avoids stress and
strain concentration in the weld.
The adoption of the reduced beam section (RBS) detail to protect the beam-to-column
connection region has been investigated. This detail moves the location of the plastic hinge
away from the connection to the reduced section and limits the flange force that can be
transmitted into the connection. Numerical results show that the RBS connection detail
reduces the vulnerability of the connection to the occurrence of a fracture near the weld. Less
severe stress and strain conditions were observed at the beam flange-column flange interface
and the likelihood of fracture occurring at the beam flange groove welds and in the
surrounding base metal region is reduced. Adopting the maximum practical cutout
corresponding to 55% flange width reduction, yielding is concentrated within the reduced
section of the beam where large inelastic strains are registered. With the RBS detail there is
less concentration of plastic deformation in the panel zone than in the standard model; the
reduced beam section appears attractive because the plastic demands on the column and panel
zone decrease.

ACKNOWLEDGEMENTS

The financial contribution of Italian MIUR is gratefully acknowledged.

245
REFERENCES
Blackman B., Popov E. P. 1995. Studies in steel moment resisting beam-to-column connections for
seismic design. EERC 1995/11 Berkeley.
Chi, B., Uang C., 2002. Cyclic response and design recommendations of reduced beam section
moment connections with deep columns. Journal of Structural Engineering.
Engelhardt M., Sabol T. 1998. Reinforcing of steel moment connections with cover-plates: benefits
and limitations. Engineering Structures, vol. 20, 510-520.
Engelhardt, M., Winnerberger T., Zekany A., Potyraj T., 1998. Experimental investigation of dogbone
moment connections. Engineering Journal, Fourth Quarter.
Hancock, J. W., and MacKenzie, A. C. 1976. On the mechanisms of ductile failure in high-strength
steel subjected to multi-axial stress states. Journal of the Mechanics and Physics of Solids, 24,
147–169.
Hibbit, Karlsson and Sorensen, 1999. Abaqus/Standard, Version 6.2.
Popov, E.P., Blondet, M., Stepanov, L. 1996. Application of dog-bones for improvement of seismic
behaviour of steel connections. Report no. UCB/EERC-96-05, University of California,
Berkeley, USA.
Srivanich W., Shen J., Kitjsateanphun T. 1999. Seismic performance of cover-plate strengthened
frames. The Structural Design of Tall Buildings, 8, 215-246.
Valente M., Chesi C. 2002. Panel zone deformation of steel welded beam-to-column connections,
Proc. 3rd Eurosteel Conference, Lisbon.
Valente M., Castellani A. 2002. Cyclic behaviour and weld access hole detailing of steel welded
beam-to-column joints. Proc. 3rd Eurosteel Conference, Lisbon.
Valente M., Castiglioni C. A. 2003. Effects of concrete slabs on the behaviour of steel beam-to-
column joints: experimental and numerical study. Proc. 4th International Conference STESSA
Behavior of steel structures in seismic areas, Naples.
Whittaker A., Gilani A., Takhirov S., Ostertag C. 2002. Forensic studies of a large cover-plate steel
moment resisting connection. The Structural Design of Tall Buildings, 11, 265-283.

246