You are on page 1of 13

Engineering Structures 27 (2005) 1125–1137

www.elsevier.com/locate/engstruct

Development of practical design methods for steel structures with


semi-rigid connections
J.M. Cabrero∗, E. Bayo
Department of Structural Analysis and Design, School of Architecture, University of Navarra, 31080 Pamplona, Spain

Received 26 January 2005; received in revised form 23 February 2005; accepted 23 February 2005
Available online 10 May 2005

Abstract

The use of semi-rigid joints is among the novelties incorporated in Eurocode 3. Although the benefits of semi-rigid joints are extensively
documented, they are not really used much in practice. There is a lack of appropriate design methods, models and tools. In this article a design
method suitable for semi-rigid joints assumption is introduced. The proposed method allows to optimize not only the size of the structural
profiles, but also the joint design to make it fit to the optimal theoretical values. Pre-design methods for semi-rigid extended end-plate joints
are also provided to easily check the feasibility and suitability of a connection design. Two design examples are proposed to demonstrate the
application of the proposed semi-rigid design methods, and their results compared to pinned and rigid alternatives. The semi-rigid approach
results in more economical solutions.
© 2005 Elsevier Ltd. All rights reserved.
Keywords: Design methods; Semi-rigid joints; Eurocode 3; Design of steel structures; Limit States method; Elastic design; Plastic design

1. Introduction or pinned. This assumption simplified much the structural


analysis and design. Modern standards, as Eurocode,
Future adoption of Eurocode 3 [6] brings in several classify joints according to their actual behaviour. Real
new features the designer can take advantage of. The use joint behaviour consideration constitutes a major advance in
of semi-rigid joints is among the novelties incorporated in the current codes, however it requires the development of
Eurocode 3. The benefits of semi-rigid joints are extensively models for the joints and structural design methods that are
documented: they can provide sufficient lateral bracing for adequate to that assumption. As future Eurocode binds to
usual wind loads in low-rise buildings, making bracing take into account the actual behaviour of the joint and its
elements unnecessary; their rotational stiffness allows to influence in the whole structure, it becomes suitable for the
optimize the distribution of the bending moments in the designer to profit from it. The main difficulty is how to bring
structure, thus leading to optimal weight and moment semi-rigid into the everyday practice.
distribution [34]; semi-rigid structures also achieve an extra Codes and standards provide methods of analysis for the
redundancy [24], meaning a more favourable situation for joint behaviour (the so-called component method presented
fire and vibration analyses. As a consequence, semi-rigid in the Eurocode 3, part 1.8 [7]), but no tools exist which
structures result in more efficiency, lightness and economy. would allow to introduce and design them, from the early
Although all joints are in fact semi-rigid, previous stages to the final design, in an adequate and simple
standards have always accepted joints as either totally rigid way. Tools and methods to facilitate their incorporation to
structural design process are required.
Several proposals have been made to promote the
∗ Corresponding author. Tel.: +34 948 425 600x2704; fax: +34 948 425
approach to semi-rigid structures. Because of the different
629.
E-mail addresses: jcabbal@alumni.unav.es (J.M. Cabrero), normative framework in which they have been developed,
ebayo@unav.es (E. Bayo). the American [9,11,17,21–23,28] (based on LRFD [1])

0141-0296/$ - see front matter © 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.engstruct.2005.02.017
1126 J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137

and the European [13,16,18,19,24,31,33] (mainly based on


Eurocode 3 [6,7]) approaches can be distinguished.
Within the American framework, Dhillon and O’Malley
[11] developed a computerized analysis method. The model
adopted for the joint was based on the Frye–Morris
equation. Joints were chosen so as to stand the 90% of
the bending moment present at the joint if a clamped
beam were assumed. Chen et al. [9,21–23] proposed (with
slight changes for each of the references) to get the joint
characteristics from a power-model or a database. Presizing
was obtained applying the wind-moment method or similar.
A non-linear second order analysis of the resulting structure
was then fulfilled. After the analysis, stability, serviceability
(the most unfavourable checking for semi-rigid structures, Fig. 1. Design method for traditional joints.
according to their experience [23]), resistance and ductility
checkouts had to be satisfied. Trying to also optimize the joint design, a methodology
Goverdhan et al. [17,28] suggested a series of rather for the design of braced structures with semi-rigid extended
different and ambiguous analyses; for each one of them, end plate joints was developed by Faella et al. [13]. Design
the load assumptions (first just gravitational loads and abaci were provided in order to design the joint in an
afterwards all the possible load combinations) and the model interactive manner with the structural analysis.
for the joint were changed. One of the proposed analyses, In this article a design method suitable for semi-rigid
following an ASCE recommendation [3], fixed the stiffness joints assumption is introduced. The proposed method
of the joint as the secant stiffness at a rotation of 2.5 mrad. allows to optimize not only the structural profiles sizing, but
A similar assumption was considered by Leon and also the joint design to meet optimal theoretical values. The
Hoffman [24], who determined the joint characteristics proposed design algorithm is found to be quite similar to the
according to the secant stiffness at a rotation of 2 mrad, prior one currently used for pinned and rigid joints assumption.
to a second-order plastic analysis. It is hoped that this similarity will simplify and help the
The previous approach is not common in European introduction of the semi-rigid concept into the everyday
reports, among which the excellent works done by structural design practice.
Jaspart [19] and Weynand and Feldman [33] must be pointed Also pre-design methods for semi-rigid extended end-
out. These authors tried to bring to the structural designer plate joints are provided to easily check the joint design
some helpful tools that applied Eurocode 3 [6,7]: design and its suitability. The design of two frames applying the
tables, data sheet and structural software. In quite a similar proposed methods is presented, and their results compared
approach, Steenhuis et al. [31] proposed a joint pre-design to those of pinned and rigidly-connected frames. The cost of
method for a set of standardized semi-rigid joints, so their the different designs is also analysed.
incorporation into the structural analysis could be simplified.
The British Steel Construction Institute published some 2. Design philosophies and methodologies
guides [2,10,29] to promote the wind-moment method in
the design of low-rise buildings. Capacity tables for semi- Let us first analyse the usual structural design flowchart,
rigid joints design were provided. Although an extremely shown in Fig. 1. With the traditional joints assumption as
complicated approach for daily structural design practice, pinned or rigid, joint behaviour is introduced in the structural
contributions based on the implementation of a genetic analysis as a previous morphological assumption in a simple
algorithm [20,25] must be also pointed out. way. This first design decision of the type of connections
None of the previous references deal in an appropriate used allows to fulfil the overall structural analysis without
way with the joint design. Connection design is taken from the need of dealing with the joint design till the structural
a database or a table, so it constitutes a kind of fixed starting design is complete. The joint design is accomplished as a
point. Joint re-design and optimization are not taken into final step, detailing the joint according to the firstly assumed
account, unless a structural section should be modified, what rotational behaviour: rigid or pinned.
would make necessary the choosing of a different joint. No The previous flowchart becomes more complicated in
reference theoretical parameters for the joint behaviour— the case of semi-rigid joints. With the available analysis
stiffness and resistance—have been developed. It must be methods (among them, the component method shown in
pointed out that, in spite of optimizing the structural profiles, Eurocode 3 part 1.8 [7], should be highlighted) the joint
if the joints are not optimized, the resulting structure cannot must be completely detailed in order to be able to obtain
be considered optimal and even adequate, as its main its stiffness and resistance characteristics. However, to detail
differential characteristic, the semi-rigid joint, has not been the joint (whose structural characteristics, due to the absence
yet fully exploited. of stiffeners, are highly dependant on the structural sizing)
J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137 1127

it is necessary to know in advance the size of the attached


structural profiles. Conversely, to obtain the structural profile
sizing it is necessary to know the joint characteristics in
advance. This comes to be a point of no-return: without
sizing there is no joint, without joint there is no sizing.
The designer must stand out to an increasingly iterative and
tedious process as everything influences everything.
Fig. 2. Semi-rigid beam with optimum bending moment distribution.
A design method suitable for semi-rigid joints is
proposed. Apart from minimizing the need for iteration, the
main advantage is its similarity to the design method used at
present for the traditional types of joints (pinned and rigid):
the structure is sized according to design assumptions, and
the joint is detailed afterwards.

2.1. Optimum initial stiffness

The joint stiffness cannot be defined in a proper way


just by its quantitative value, as it does not provide a clear
estimation of it. For a clear understanding it should be
defined in relation to the attached structural elements.
Eurocode 3 part 1.8 [7] in its clause 5.2.2 defines the Fig. 3. Beam-line diagram: optimum rotational stiffness.
rotational stiffness in the following way
E Ib line M + corresponding to the hogging moment can be drawn
Sj = α . (1) for a uniform load in an isolated beam
Lb
The α parameter relates the joint stiffness to the attached q L 2b 2E Ib
M+ = − ϕ. (3)
beam. This factor is used by Eurocode to classify the joint 12 Lb
according to its rotational stiffness. Just as a reminder, joints Another line M − , corresponding to the sagging moment, can
are classified as nominally pinned if α < 0.5; nominally be traced
rigid in a braced frame if α > 8, and in an unbraced frame if
q L 2b 2E Ib
α > 25. Intermediate values of α between pinned and rigid M− = + ϕ. (4)
are the ones for which the joint is considered to be semi- 24 Lb
rigid. But this α parameter does not provide a clear idea of The intersection point A indicates the optimum bending
the rotational behaviour of the structure, because it is not moment distribution, so that M + = M − = q L 2b /16. This
related linearly to the resulting structural behaviour, which intersection point corresponds to a stiffness degree K opt of
makes its understanding somehow difficult. r ≈ 0.66 (or α ≈ 6, expressed in the Eurocode manner).
Chen [8] proposed the adoption of the end-fixity factor r
as 2.2. Resistance
1
r= . (2)
1 + 3E Ib If plastic analysis is performed and plastic hinges should
Sj L b
arise, it is advisable to favour their location in either the joint
Its main virtue is that it provides a clear understanding of or the beam, not in the column. This is a recommendation
the joint rotational behaviour, as it does relate linearly to the widely applied for plastic design in seismic zones. This
rotational attached beam behaviour. This parameter takes as assumption leads to the use of partial-resistant joints.
limits: null value for a theoretically pinned joint, and unity Dhillon and O’Malley [11] proposed to choose the
for a theoretically rigid one. required joints according to the resistance. The chosen joint
Xu [34] showed how the optimum bending moment must resist 90% of the hogging moment for the clamped
distribution shown in Fig. 2, for an isolated beam with beam. This is not an optimized approach for the joint
uniform distributed load, and characterized as having the resistance, since it may result in oversized joints in most of
same hogging and sagging moments, leads to the least the cases. Besides, it may become quite difficult to obtain
weight beam design. It is the most suitable behaviour for such a moment resistance degree without column stiffeners.
common steel profiles. The absence of stiffeners is one of the strong points of semi-
The optimum joint rotational stiffness that leads to the rigid joints, in order to obtain economically competitive
previously shown optimum bending moment distribution results.
can be easily found [13]. In the beam-line diagram shown in The method introduced in this article proposes to perform
Fig. 3, which relates the bending moment to the rotation, a a previous elastic structural analysis which leads to a more
1128 J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137

optimized moment requirement: the minimum required Ultimate Limit State, the rotation value is fixed at 20 mrad.
moment resistance should be the bending moment at the Rex and Goverdhan [28] state the inadequacy of this model
joint in the elastic analysis. for SLS (fixing the joint stiffness at a fixed rotation) as
deflections are very sensitive to the chosen stiffness.
2.3. Rotational ductility
2.5. Equivalent beam element
In order to provide structural safety when plastic hinges
are present, it is important to guarantee they possess In the usual structural software there is a lack of a
enough rotational capacity. This required ductile rotational rotational spring element which could model semi-rigid
behaviour is being heavily studied at present [14,15,27,30]. joints in an appropriate way, as there has been no need of
Eurocode 3 part 1.8 [7] in its clause 6.4 provides a it with the traditional type of joints.
few simple recommendations about rotational behaviour. There is a simple way to model the joint rotational
When the joint rotational behaviour is ruled by the column behaviour, accounting also for its resistance, by means of
web in shear, the column flange in bending or the end- an equivalent beam element [12] whose structural properties
plate in bending, the joint behaviour is supposed to be relate to the modelled joint characteristics:
ductile enough. Taking into account the statement regarding Sj.Rd L eq
resistance in the previous section (see Section 2.2), and Ieq = (6)
E
in order to obtain an adequate control of the rotation in Mj.Rd
the plastic hinge, it can be advisable to consider the joint Wpl.eq = . (7)
behaviour commonly named as “thin end-plate”. fy
This equivalent element constitutes an approximation to
2.4. Joint modelling the actual result: the shorter it is, the more exact is the result.
However, it is recommended to define it with a length equal
Due to the highly non-linear rotational behaviour of to half the column height, so that, the joint eccentricities are
the semi-rigid joints, it is quite complicated to adapt already introduced in the structural model.
their rotational behaviour to a simple and adequate
approach. In the previously referred design methods (see 2.6. Shear interaction
Section 1) several models were adopted: polynomials (as
the Frye–Morris model), power-models . . . Such models, So to obtain an accurate model of the behaviour and
although they can finely reproduce the joint behaviour, are shear deformation of the connection, it would be desirable
quite complex and only adequate for certain types of joints, to model each of the joints and the web panel in shear in an
which reduces their scope mainly to research purposes. In independent way. Although it would be the right way, it is
addition, their implementation into a conventional structural not feasible with currently available structural software.
software is not an easy task. Eurocode 3 part 1.8 [7] in its clause 5.3, allows
A bilinear model would be a much more appropriate to simplify the model by having just one rotational
approach for everyday practice, as it only requires defining spring, representing the behaviour of the connection
the rotational stiffness and the moment resistance. Eurocode (joint + corresponding web panel in shear). Consequently, a
3 part 1.8 [7] allows such a simplified bilinear model to double beam–column joint would be modelled by means of
simulate the joint rotational behaviour. In order to carry this two springs at both sides of the column interacting to some
model out, a secant stiffness Sj.Rd is defined as degree, each one with different moment–rotation curves.
Sj.ini Because the shear deformation at the column web is taken
Sj.Rd = . (5) into account independently for each of the springs, it is
η necessary to develop a coefficient that takes into account
The reduction factor η becomes two in the case of the shear interaction in the web panel. This factor, called βi ,
beam–column joints [7, Table 5.2]. Because of this factor, modifies the spring stiffness and resistance.
the Eurocode approach (whose validity has already been  
 Mj.bj.Ed 
proved [32]) is called the half initial secant stiffness 
β i = 1 − ≤ 2. (8)
approach. The joint resistance, Mj.Rd , is taken as the model Mj.bi.Ed 
resistance. Eurocode 3 admits not to reduce the initial joint βi becomes one for one-sided joints, zero for double-
rotational stiffness by the η factor, if the required moment sided joints with different sign and same magnitude bending
resistance is found to be less than 2/3Mj.Rd . moments, two is considered as the maximum possible value
The model recommended by the ASCE [3] varies from for double-sided joints with the same sign and the same
that of Eurocode. The proposed secant joint stiffness magnitude bending moments at both sides. As this factor
changes depending on the Limit State to be fulfilled. For βi is included in the joint model, it is necessary to know
the Serviceability Limit State the secant stiffness is that its value, so that the final joint design can be adjusted and
corresponding to a fixed rotation of 2.5 mrad; for the optimized according to the actual stresses.
J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137 1129

2.7. Joint pre-design

One of the major problems when trying to make a semi-


rigid structural design is getting to know in advance the ade-
quacy of the chosen type of joint to the required values. Sim-
ple methods of evaluating the joint stiffness and resistance
are required prior to undertaking a more detailed analysis.
Previous works have been done in Europe by Steenhuis,
Gresnigt and Weynand [4,31]. Their work can be applied to a
number of different joint configurations with some pre-fixed
features: mainly, joints should have two bolt-rows in the
tension zone and the end-plate thickness should be similar to
that of the column flange. Resulting expressions for stiffness
and moment resistance were obtained from simplifying the
component method results as a function of the lever arm h t Fig. 4. Yield-line models for resistance computation of the extended end-
(which can be approximated as the beam height) and the plate (left) and the unstiffened column flange (right).
thickness of the column flange tcf .
This approximation provides the structural characteristics Their assembling results in the joint initial stiffness
just for that standardized joint design. In addition, it is a 1
rough measure of the possible joint stiffness and resistance, Sj.Pred = . (12)
1
Sepb + 1
Scfb + 1
Scws
and it cannot provide a fair estimation if any parameter is
modified. This is its main drawback, as it makes impossible 2.7.2. Resistance pre-design
the optimization of the joint. Following the results of the parametric study, the joint
In this section, a pre-design method for unstiffened moment resistance of unstiffened extended end-plate joints
extended end-plate joints is proposed. It allows to obtain and can be accounted for as the minimum resistance of three
check the stiffness and the resistance of the joint, so that the main components: the column web in shear, the column
designer can change the main geometrical parameters of the flange in bending and the end-plate in bending.
joint: the end-plate thickness tcf and the bolt vertical pitch p. The resistance of the column web in shear can be
evaluated by means of
2.7.1. Stiffness pre-design
0.9 f y Avc h b f y Avc h b
The proposed pre-design method is based on conclusions Mj.cws = √ ≈ 0.52 (13)
from an extensive parametric study for unstiffened end-plate 3 β β
joints. It consists of the development of a series of three formula obtained from the one provided by Eurocode 3 part
springs. 1.8 [7].
Two of the springs (Scws and Scfb ) are related to the The column flange and the end-plate resistance com-
column, and indeed, their value is independent of the joint putation are based on the developments by Murray and
geometry. They can be obtained just knowing in advance the Shoemaker [26] and recommended by the AISC [5]. The
sizes of the attached column and beam. The first spring, Scws , proposed approach is based on the establishment of yield-
corresponds to the column web in shear, and its formula is line models for the extended end-plate and the unstiffened
taken from Eurocode 3 part 1.8 [7] column flange, shown in Fig. 4.
h b Avc The main resistance formula for both components is
Scws = 0.38E . (9) expressed as
β
Mj.i = 0.9 f y Yi ti2 (14)
The second, Scfb , relates to the column flange in bending,
3 where Yi is a geometrical parameter whose values
h 2b tcf
Scfb = 60E (10) (simplifying those proposed in the AISC publication [5] as
bc2 a function of the bolt pitch p, the height of the beam h b and
and, the third spring, Sepb , is obtained by means of a curve- the column width bc ) are
fitting approach to data from various joints with different 0.8 p 2.5bc h b
geometrical parameters (see Appendix A). Yep = 2.8h b − 1.4 p + (2h b − p) + (15)
bc p
(h b − p/2)2 tep
3 for the end-plate, and
Sepb = (60 − 2tep )E . (11)
p2 p( p + 4h b )
Ycf = 5.66h b + 0.25bc + (16)
The last spring (Sepb ) is related to the joint geometry, bc
accounting for the two main geometrical joint parameters: for the unstiffened column flange; while ti is the thickness
the end-plate thickness and the bolt vertical pitch. of the corresponding component.
1130 J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137

Fig. 5. Proposed design method for semi-rigid elastic analysis.

2.7.3. Pre-design rules of thumb Table 1


A few rules of thumb can be assumed to easily compute Design example frames: applied loads
the joint resistance and stiffness: Load Frame A (kN/m) Frame B (kN/m)

• The parameter β can be initially estimated as β = 0.5 for qsw−0 7.80 16.60
double-sided joints and as β = 1 for single-sided joints. qsw−1 6.50 13.90
qil−0 11.20 34.00
• The pitch p may be simplified, as the maximum stiffness
qil−1 3.20 6.80
values are obtained when the bolts are closer to the beam qw 3.80 5.40
flange (due to constructive limitations, the minimum
distance to the column flange is about 20 mm, which
gives a minimum pitch of about 80 mm).
• If the end-plate is thinner than the column flange, the end-
plate in bending is probably the weakest component. analysis, the joint initial stiffness is determined. Structural
• If the end-plate is thicker than the column flange, the analyses are done afterwards, accounting for the joint
weak component will be the minimum between the stiffness. The required moment resistance of the joints is
column flange and the column web. derived from the analysis results. The joints are designed in
• With high β values, the weak component is the column detail at the final step of the design.
web in shear. The required change of design philosophy (compared
to pinned and rigid joints) is minimum, as the designer
The consideration of two limit (thin and thick)
deals always with simple structural models. The only added
thicknesses for the end-plate may allow to establish a range
complication is the introduction of the joint mechanical
of values for the joint initial stiffness and resistance (please
properties in the model, which can be easily done by
refer to tables corresponding to stiffness—Tables 2 and 7—
means of rotational springs or the equivalent beam model.
and resistance—Tables 3 and 8—in the design examples
The practicability of the required joint parameters can be
detailed in Section 4). These limit values help to check the
checked with the pre-design methods previously explained
feasibility of the obtained required values.
(see Section 2.7).
The structural model is that of Eurocode, consisting
3. Analysis design procedure of a rotational spring accounting for both the joint and
the web panel in shear deformation. Second-order analysis
Two different procedures are proposed for elastic with a buckling length factor of 1 is recommended, so
(Fig. 5) and plastic (Fig. 7) analysis. Both share the same that structural analysis remains simple, and buckling is
philosophy: after a pre-sizing estimation, and prior to any conservatively considered.
J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137 1131

Table 2
Frame A: joint required stiffness and pre-design results

Joint Stiffness Pre-design


(kN m/rad) βPred Sj.Pred.8 r8 Sj.Pred.20 r20 rPred > 0.6

A 8,442.00 1.00 7,681.01 0.79 13,470.93 0.87 OK


B 8,442.00 0.50 8,066.41 0.80 21,624.20 0.91 OK
C 25,156.00 1.00 11,361.29 0.65 19,839.75 0.77 OK
D 25,156.00 0.50 15,136.69 0.71 35,148.94 0.85 OK

Table 3
Frame A: joint required resistance and pre-design results

Joint Resistance Pre-design


(kN m) βPred Mj.Pred.8 Mj.Pred.20 Mj.Pred > Mj.Req

A 22.00 1.00 23.27 epb 37.41 cws OK


B 35.00 0.50 24.96 epb 67.43 cfb OK
C 40.00 1.00 32.24 epb 50.50 cws OK
D 60.00 0.50 34.50 epb 89.85 cfb OK

3.1. Elastic design method 3.1.4. Joint design


Because the joint feasibility has already been checked by
The corresponding flowchart for elastic analysis design means of the pre-design formulas, the only thing left in this
method is shown in Fig. 5. The different design phases are last stage is the detailed joint design and checking.
explained in detail below. As fitting the exact stiffness value for semi-rigid joints
may be complicated, following a similar philosophy to
3.1.1. Pre-sizing that of modern standards (such as Eurocode), a range
A previous pre-sizing for the structure should be made. of acceptable values is proposed. The stiffness variations
The designer’s self-experience is an important factor at cause a change in the range of the structural behaviour
this stage. As a rule of thumb, though it leads to an lower than 5%. This admissible variation, impossible to
underestimation (as serviceability tends to be the critical determine by just accounting for the stiffness value, can be
factor), the beams can be sized according to a bending easily quantified by means of the end-fixity factor. As it
moment equal to the optimum value: q L 2b /16. relates linearly to the attached beam rotational behaviour,
the proposed range can be assumed as a variation for
3.1.2. Rotational stiffness this index of ±0.05. This assumption leads to a range of
The optimum theoretical stiffness value, fixed as an end- acceptable stiffness limits of 0.6 ≤ r ≤ 0.7. A detailed
fixity factor r = 0.65, has been previously derived (see flowchart for the checking of the joints is presented in
Section 2.1). This optimum stiffness value can be easily Fig. 6.
obtained from Eq. (2) after the first pre-sizing of the As an elastic analysis has been made, the minimum joint
structure, required moment resistance is already fixed as the moment
E Ib value present at the joints. Actual joint moment resistance
Sj.Req = 5.6 (17) should always be higher than this required value.
L
and introduced into the structural model by means of a
rotational spring model or the equivalent beam element
3.2. Plastic design method
previously reported in Section 2.5. The joint stiffness pre-
design method in Section 2.7.1 allows to easily check the
feasibility of the required value. The plastic analysis design method flowchart is shown in
Fig. 7. It can be observed how the design process is quite
3.1.3. Elastic analysis similar to the elastic one, as the main concepts remain the
Having introduced the joint stiffness in the structural same: after having the joint required stiffness checked and
model, a second-order elastic analysis is performed for both introduced in the model, a second-order elastic analysis for
Limit States. Results from this analysis allow to fix the the Serviceability Limit State is accomplished. With the
required moment resistance for each joint, Mj.Req . As for obtained results, the joint required moment resistance can
the stiffness, the joint resistance pre-design method (see be determined and checked by means of the pre-design
Section 2.7.2) allows to check its feasibility. formulas.
1132 J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137

Fig. 6. Flowchart for the joint validation.

Fig. 7. Proposed design method for semi-rigid plastic analysis.

3.2.1. ULS plastic analysis dealing with plastic design methods, it may be desirable to
This step constitutes the main difference in comparison to let the joint resistance undergo the required resistance, up
the previously explained elastic method. As a second-order to a value of 90%Mj.req . In this case, a new second-order
plastic analysis must be fulfilled, the joint model must also plastic analysis, with updated joint resistance values should
take into account the required moment resistance. The joint be fulfilled.
required moment resistance can be increased by a safety
coefficient γ to reduce the number of resultant plastic hinges
4. Design examples
in the structure. As explained before, the joint behaviour
can be introduced by means of a rotational spring or an
equivalent beam model. Two design examples are presented to demonstrate the
application of the design procedure. The proposed plastic
design method with a γ factor of 1 for resistance is applied.
3.2.2. Joint design The material is steel S275, with a modulus of elasticity
Once this last step of the design process is reached, and of 210,000 MPa and yield stress of 275 MPa. European
knowing already by virtue of this method the feasibility of sections (i.e. HEB and IPE sections) are used. Bolts are
the joint, the only thing left is just a more detailed design. steel 8.8. Maximum drift is restricted to H /150, while the
The joint stiffness validating process is the same as maximum floor beam deflection is restricted to L b /400, and
discussed in the elastic design method (see Section 3.1.4 the maximum deflection for roof beams is limited to L b /250.
and Fig. 6). Resistance checking changes a little bit: as The considered load factors are 1.3 for self-weight loads
J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137 1133

Table 4
Frame A: final design of the joints (sizes in mm)

Joint Beam Column Bolt h ep bep tep p w e

A IPE200 HEB140 T20 285 140 10 90 80 30


B IPE200 HEB160 T16 295 140 12 110 80 30
C IPE270 HEB140 T22 345 140 15 70 80 30
D IPE270 HEB160 T22 345 140 14 70 80 30

Fig. 8. Frame A: configuration.

Fig. 9. Typical unstiffened extended end-plate joint layout.


(qsw ), and 1.5 for imposed (qil ) and wind (qw ) loads. Both
wind directions are considered.
The columns are supposed to be continuous, and the in Table 6. Semi-rigid design shows to be a cost competitive
beams are grouped so to share the same section. The joints alternative for this kind of low-rise frames.
are also grouped in terms of same characteristics and design.
Semi-rigid design results are compared to those of pinned Table 5
and rigidly-connected frames. A cost estimation among Frame A: resulting profiles for different types of connections
the three resulting structures is done for the two frames Member Profile
presented below. Prices are obtained from major Spanish Semi-rigid Pinned Rigid
steel fabricators. It must be pointed out that, of course, the 1, 2, 3 IPE 200 IPE 240 IPE 200
given cost estimations may not apply for other countries, 4, 5, 6 IPE 270 IPE 330 IPE 270
steel contractors or situations. They are presented just as a 7, 10 HEB 140 HEB 120 HEB 140
demonstration of semi-rigid structures competitivity. 8, 9 HEB 160 HEB 140 HEB 160

4.1. Frame A: Two storey, three-bay regular frame Table 6


Frame A: cost comparison
The configuration, dimensions, and numbering of
Connection Cost estimation (e)
members and joints are shown in Fig. 8. Loading is shown type Connections Steel Total
in Table 1.
The connection required parameters and their feasibility Semi-rigid 505.08 3,619.50 4,124.58 100%
Pinned 249.36 4,296.00 4,545.36 110%
according to the proposed pre-design method are shown in Rigid 1,821.36 3,619.50 5,440.86 132%
Table 2 for stiffness and in Table 3 for resistance. In both
Tables 2 and 3, pre-design values for two different “limit”
thicknesses (8 and 20 mm) are shown: pre-design stiffness
Sj.Pred.i and its corresponding end-fixity factor ri ; pre-design 4.2. Frame B: Two storey, four-bay irregular frame
resistance Mj.Pred.i and the corresponding weak component.
Shown pre-design values are obtained for a bolt vertical Fig. 10 shows the configuration, dimensions and number-
pitch of 80 mm. Final joint design parameters are shown ing of the members. Loading values are given in Table 1.
in Table 4 referred to the typical joint layout presented in The joint stiffness parameters are given in Table 7. the
Fig. 9. joint resistance parameters are presented in Table 8. In both
The results of the design sections for the design of semi- cases, required values practicability is checked by means
rigid connections are shown in Table 5. Resulting profiles for of the pre-design methods. Final joint design values are
the same frame in the case of pinned and rigid joints are also presented in Table 9, referring to the typical joint layout
presented. A cost estimation for the three joint types is given drawn in Fig. 9. Although joint D does not satisfy the
1134 J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137

Fig. 10. Frame B: configuration.

Table 7
Frame B: joint required stiffness and pre-design results

Joint Stiffness Pre-design


(kN m/rad) βPred Sj.Pred.8 r8 Sj.Pred.20 r20 rPred > 0.6

A 5,764 1.00 4,120.80 0.75 9,570.22 0.87 OK


B 5,764 0.50 4,912.02 0.78 15,290.03 0.92 OK
C 12,044 0.50 10,096.35 0.78 26,572.94 0.90 OK
D 18,382 1.00 7,827.19 0.64 15,072.42 0.78 OK
E 18,382 0.50 10,096.35 0.70 26,572.94 0.86 OK
F 70,690 0.50 27,329.75 0.62 58,279.94 0.77 OK
G 70,690 1.00 22,129.14 0.56 38,823.37 0.69 OK

Table 8
Frame B: joint required resistance and pre-design results

Joint Resistance Pre-design


(kN m) βPred Mj.Pred.8 Mj.Pred.20 Mj.Pred > Mj.Req

A 19 1.00 18.15 epb 29.93 cws OK


B 32 0.50 20.93 epb 61.64 cfb OK
C 62 0.50 29.61 epb 83.29 cfb OK
D 47 1.00 25.84 epb 41.15 cws NO
E 80 0.50 29.61 epb 83.29 cfb OK
F 133 0.50 49.87 epb 133.80 cfb OK
G 93 1.00 49.87 epb 104.20 cws OK

Table 9
Frame B: final design of the joints (sizes in mm)

Joint Beam Column Bolt h ep bep tep p w e

A IPE160 HEB140 T16 245 140 10 90 80 30


B IPE160 HEB180 T20 240 140 12 80 80 30
C IPE220 HEB180 T20 295 140 14 70 80 30
D IPE220 HEB140 T22 295 140 16 70 80 30
E IPE220 HEB180 T22 295 140 18 70 80 30
F IPE360 HEB180 T22 435 140 16 70 80 30
G IPE360 HEB160 T22 435 140 16 70 80 30
J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137 1135

pre-design requirements, as seen in Table 8, the design The association Alumni Navarrenses and the Spanish
procedure explained in 6 leads to a suitable final design as Science and Education Ministry are also gratefully
shown in Table 9. acknowledged for providing sponsorship to the first author.
The design sections of the frame, also compared to
pinned and rigidly-connected frames, are given in Table 10.
Appendix A. Stiffness pre-design formula: Curve-fitting
Cost comparison is shown in Table 11. Semi-rigid design
approach
qualifies, as in the previous design example frame, as the
lowest cost structural solution.
Partial results for the curve-fitting process that allowed
to develop the stiffness pre-design method presented in
Table 10
Frame B: resulting profiles for different types of connections
Section 2.7.1 are shown here. Eqs. (A.1) and (A.2), based on
the T-stub approach, constitute the formulas to be used in the
Member Profile design procedure. The parameter A1 is fixed to 60, and A2
Semi-rigid Pinned Rigid
is obtained from Eq. (A.3) in terms of B1 and B2 . These two
1, 4 IPE 160 IPE 240 IPE 200 parameters are defined by means of a curve-fitting approach
2, 3 IPE 220 IPE 360 IPE 270 from the values shown in Tables A.1 and A.2. The resulting
5 IPE 240 IPE 300 IPE 220
6 IPE 360 IPE 450 IPE 330
final mean values are B1 = 60 and B2 = −2.
3
7, 11 HEB 140 HEB 120 HEB 120 h 2b tcf
8, 9, 10 HEB 180 HEB 140 HEB 160 Scfb = A1 E (A.1)
bc2
(h b − p/2)2 tep
3
Sepb = A2 E (A.2)
Table 11 p2
Frame B: cost comparison A2 = B1 − B2 tep . (A.3)
Connection Cost estimation (e)
type Connections Steel Total Appendix B. Notation
Semi-rigid 505.08 3,820.50 4,325.58 100%
Pinned 249.36 4,617.00 4,866.36 113% B.1. Upper cases
Rigid 1,821.36 3,717.00 5,538.36 128%
Avc shear resistant area of the column
A i , Bi curve-fitting constants
E Young modulus
5. Conclusions
H total height of the frame
I moment of inertia
Two semi-rigid design methods for elastic and plastic
K opt optimum stiffness
analysis have been presented. Incorporated design examples
L beam length
demonstrate the applicability of the proposed methods. Cost
Mj joint moment resistance
estimations for the design examples have been also carried
M− hogging beam-moment line
out, and semi-rigid design has proved as the most cost
effective solution in comparison to traditional types of joints M+ sagging beam-moment line
(pinned and rigid). S rotational stiffness
Proposed design methods can be easily introduced in Sj joint rotational stiffness
everyday practice, facilitating semi-rigid joints assumption Sj.ini joint initial rotational stiffness
and use. The added complication of rotational stiffness Wpl plastic modulus
consideration is reduced due to the application of optimal Yi resistance geometrical parameter corresponding to
theoretical values. By means of these methods, advantages component i
of semi-rigid construction can be incorporated into structural
design practice in a competitive and efficient manner. B.2. Lower cases
Methods have also been proposed to easily obtain a joint
pre-design adequate to the required stiffness and resistance. b width
e bolt distance to edge
Acknowledgements fy yield stress
h height
The support of this work provided by the European ht lever arm of the joint
Research Fund for Coal and Steel under contract number p bolt vertical pitch
7215-PP-070 and the Chair Arcelor—University of Navarra q uniform load
is greatly acknowledged. r end-fixity factor
1136 J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137

Table A.1
Curve-fitting: IPE 270-HEB 160 beam–column joints with 90 and 120 mm bolt pitch

Beam Column Bolt Beam Column Bolt


IPE270 HEB160 T22 IPE270 HEB160 T22
h ep bep p w e h ep bep p w e
355 150 90 90 30 370 150 120 90 30
tep Sj.Rd A1 A2 tep Sj.Rd A1 A2
8 13,096.33 60.00 39.91 8 9,679.75 60.00 47.28
12 19,149.59 60.00 33.53 12 15,087.22 60.00 33.07
15 20,692.98 60.00 24.39 15 17,754.69 60.00 26.71
20 23,055.42 60.00 22.14 20 20,838.20 60.00 21.80
B1 = 51.3650 B2 = −1.5544 B1 = 60.8798 B2 = −2.0964

Table A.2
Curve-fitting: IPE 450-HEB 160 beam–column joints with 90 mm and 120 mm bolt pitch

Beam Column Bolt Beam Column Bolt


IPE450 HEB160 T22 IPE450 HEB160 T22
h ep bep p w e h ep bep p w e
535 150 90 90 30 550 150 120 90 30
tep Sj.Rd A1 A2 tep Sj.Rd A1 A2
8 32,160.05 60.00 42.94 8 26,419.30 60.00 50.45
12 39,988.74 60.00 29.53 12 34,621.25 60.00 30.75
15 42,675.58 60.00 22.96 15 38,741.26 60.00 24.67
20 45,151.37 60.00 16.80 20 42,237.83 60.00 17.20
B1 = 57.5967 B2 = −2.1483 B1 = 67.5947 B2 = −2.6783

ri end-fixity factor of the joint with extended end- cws column web in shear
plate of thickness i epb end-plate in bending
r (S) end-fixity factor corresponding to stiffness value S il imposed loads
t plate thickness sw self-weight loads
w bolt horizontal distance w wind loads

B.3. Greek letters


References
α Eurocode stiffness factor
β Eurocode shear interaction factor [1] Manual of steel construction—load and resistance factor design
(LRFD). 2nd ed. Chicago (IL, USA): American Institute of Steel
βi β factor for the joint on side i of the column
Construction (AISC); 1994.
ϕ rotation [2] Joints in steel construction: moment connections. Ascot: The Steel
η Eurocode stiffness reduction factor Construction Institute; 1995.
[3] Design guide for partially restrained composite connections. Journal
of Structural Engineering, ASCE 1998;124(10):1099–113.
B.4. Subscripts [4] CeStruCo (Continuing Education in Structural Connections). 2003.
[5] AISC. Prequalified connections for special and intermediate steel
Ed design action moment frames for seismic applications. Draft 24 December 2004
Rd design resistance edition; 2004.
[6] CEN. Eurocode 3: Design of steel structures. Part 1-1: General rules
Req required
and rules for buildings (prEN 1993-1-1:2003). CEN, stage 49 draft
Pred pre-design value edition; 2003.
Pred.i pre-design value of the joint with extended end- [7] CEN. Eurocode 3: Design of steel structures. Part 1.8: Design of joints
plate of thickness i (prEN 1993-1-8:2003). CEN, stage 49 draft edition; 2003.
b beam [8] Chen WF. Practical analysis for semi-rigid frame design. Singapore:
c column World Scientific; 2000.
[9] Chen WF, Goto Y, Richard Liew JY. Stability design of semi-rigid
cf column flange
frames. John Wiley & Sons, Inc.; 1996.
ep end-plate [10] Couchman GH. Design of semi-continuous braced frames. In:
eq equivalent beam element Specialist design guides, Ascot (UK): The Steel Construction
cfb column flange in bending Institute; 1997.
J.M. Cabrero, E. Bayo / Engineering Structures 27 (2005) 1125–1137 1137

[11] Dhillon BS, O’Malley III JW. Interactive design of semirigid [23] Kim Y, Chen WF. LRFD frame design with PR connections.
steel frames. Journal of Structural Engineering, ASCE 1999;125(5): In: Chen WF, editor. Practical analysis for semi-rigid frame design.
556–64. Singapore: World Scientific; 2000. p. 17–94.
[12] Dubina D, Greccea D, Zaharia R. Evaluation on static and dynamic [24] Leon RT, Hoffman JJ. Plastic design of semi-rigid frames.
structural coefficient of steel frames with semi-rigid joints via In: Bjorhovde R, Colson A, Zandonini R, editors. Connections in
numerical simulations. In: Bjorhovde A, Colson A, Zandonini R, steel structures III: Behaviour, strength and design. Trento (Italy):
editors. Connections in steel structures III, Behaviour, strength and Pergamon; 1996. p. 211–22.
design. Trento (Italy): Pergamon; 1996. p. 349–60. [25] Mahfouz SY. Design optimization of structural framework: design
[13] Faella C, Piluso V, Rizzano G. Structural steel semirigid connections: optimization of steel frame structures according to the British codes
theory, design and software. Boca Ratón: CRC Publishers; of practice using a genetic algorithm. Ph.D. thesis, University of
2000. Bradford; 1999.
[14] Gervasio H, Simoes da Silva L, Borges L. Reliability assessment [26] Murray TM, Lee Shoemaker W. Flush and extended multiple row:
of the post-limit stiffness and ductility of steel joints. Journal of moment end-plate connections. In: Steel design guide series. Chicago:
Constructional Steel Research 2004;60:635–48. American Institute of Steel Construction; 2002.
[15] Girao Coelho AM, Bijlaard FSK, Simoes da Silva L. Experimental [27] Piluso V, Faella C, Rizzano G. Ultimate behavior of bolted T-Stubs.
assessment of the ductility of extended end plate connections. I: Theoretical model. Journal of Structural Engineering, ASCE 2001;
Engineering Structures 2004;26:1185–206. 127(6):686–93.
[16] Gizejowski MA, Papangelis JP, Parameswar HC. Stability design [28] Rex CO, Goverdhan AV. Design and behavior of a real PR building.
of semi-continuous steel frame structures. Journal of Constructional In: Fourth international workshop on connections in steel structures.
Steel Research 1998;46(1–3):99–101. 2000. p. 94–105.
[17] Goverdhan AV, Lindsey SD. PR connections in design practice. [29] Salter PR, Couchman GH, Anderson D. Wind-moment design of
In: Bjorhovde R, Colson A, Zandonini R, editors. Connections in low rise frames. In: Specialist design guides. Ascot: The Steel
steel structures III: Behaviour, strength and design. Trento (Italy): Construction Institute; 1999.
Pergamon; 1996. p. 505–14. [30] Simoes da Silva L, Santiago A, Vila Real P. Post-limit stiffness
[18] Ivanyi M. Direct design method of steel frames with semi-rigid and ductility of end-plate beam-to-column steel joints. Computers &
connections. In: Ivanyi M, Baniotopoulos CC, editors. Semi-rigid Structures 2002;80:515–31.
connections in structural steelwork. In: CISM courses and lectures, [31] Steenhuis M, Weynand K, Gresnigt AM. Strategies for economic
Udine: Springer Verlag; 2000. p. 87–98. design of unbraced steel frames. Journal of Constructional Steel
[19] Jaspart JP. Integration of the joint actual behaviour into the frame Research 1998;46(1–3):88–9.
analysis and design process. In: Ivanyi M, Baniotopoulos CC, editors. [32] van Keulen DC, Nethercot DA, Snijder HH, Bakker MCM. Frame
Semi-rigid connections in structural steelwork. In: CISM courses and analysis incorporating semi-rigid joint action: applicability of the
lectures, Udine: Springer Verlag; 2000. p. 103–66. half initial secant stiffness approach. Journal of Constructional Steel
[20] Kameshki ES, Saka MP. Optimum design of nonlinear steel frames Research 2003;59:1083–100.
with semi-rigid connections using a genetic algorithm. Computers & [33] Weynand K, Feldmann M. Quick and easy design of joints in practice
Structures 2001;79:1593–604. using new tools for designers. In: Fourth international workshop on
[21] Kim WS, Chen WF. Practical advanced analysis for semi-rigid frame connections in steel structures. 2000. p. 106–16.
design. Engineering Journal 1996;35(4):129–41. [34] Xu L. On the minimum–maximum bending moment and the least-
[22] Kim Y, Chen WF. Practical analysis for partially restrained weight design of semi-rigid beams. Journal of the International
frame design. Journal of Structural Engineering, ASCE 1998;124 Society for Structural and Multidisciplinary Optimization 2001;21:
(7):736–49. 316–21.

You might also like