Concerns on Seismic Moment-Shear Connections

using available Indian Hot-Rolled I-Sections

Rupen Goswami 1, Jaswant N. Arlekar 2 and C.V.R. Murty 3

Abstract
A rational capacity design method is presented for determining
connection design forces considering increased yield stress of steel, strainhardening of the beam /column cross-section, and local instability of the
beam /column flange and web due to slenderness. A procedure is
presented for arriving at shear-moment interaction boundaries at
different axial load levels for commonly available Indian steel I-sections.
The design procedure outlined in SP:6(6) needs to be upgraded and
available Indian sections needs to be remolded to have larger flange
widths.

1. Introduction
Satisfactory performance of steel structures during strong seismic shaking
depends on numerous factors, including the three significant factors namely stability,
strength and ductility of individual members. Apart from these, connections between
members play an important role in the overall seismic performance of steel structures;
inadequate connections can result in failure of structures even when structural
members are adequately designed. A rational method for moment-shear connection
design coupled with a preferred collapse mechanism is essential in achieving a ductile
response of the whole structure during strong earthquake shaking. This paper presents
one such connection design method and compares it with the current design method.

2. Connection Design Philosophy

Following the large number of connection failures occurred during the 1994
Northridge earthquake (USA) and 1995 Kobe earthquake (Japan), a fresh approach
emerged for the design of beam-to-column and column-to-base connections. Beam-tocolumn connections are designed now as per the Capacity Design Concept, discussed in
an earlier paper [Goswami et al., 2003]. By this design method, premature fracture of
welds or fasteners is avoided at the connection. Here, beams are allowed to undergo
ductile yielding, and connections are forced to remain elastic by designing them for the
maximum demand arising from the members (beam, column) under plastic condition
(Figures 1 and 2).
1

Ph.D. Scholar, Department of Civil Engineering, IIT Kanpur, Kanpur 208016; [email protected]
Formerly Ph.D. Scholar, Department of Civil Engineering, IIT Kanpur, Kanpur 208016;
[email protected]
3
Professor, Department of Civil Engineering, IIT Kanpur, Kanpur 208016; [email protected]
2

Beam Plastic
Hinge Region

Beam Plastic
Hinge Region

Column

Beam

Connection
Reinforcement Region

Shear Link

Beam-end forces
Plastic Moment
Hinge

Plastic Moment
Hinge

Lateral Seismic Load

M pr

M pr
P

P
Connection forces

V pr

V pr

Figure 1: Location of ductile plastic hinges adjacent to beam-to-column connections.

3. Connection Design Forces

The connection design forces generally accounts for the member nominal
strength, i.e., moment capacity M pr and the associated equilibrium compatible shear
Vpr , including effect of gravity load. Using M pr in the connection design indirectly
accounts for some strain-hardening of the beam and has the advantage of being simple
to apply. However, to formalize the actual behaviour, a realistic stress-strain
relationship for steel with strain-hardening must be used to assess the maximum
demand imposed by the beam and column members on the connection elements.

Column
P

V pr
Column Plastic
Hinge Region

M pr
Plastic Moment
Hinge

Connection
forces

Concrete Pedestal

Figure 2: Location of ductile plastic hinges adjacent to column-to-foundation connections.

3.1 Axial Load - Shear - Moment (PVM) Interaction

Under seismic action, all columns and beams of lower storeys in tall buildings
under seismic actions are subjected to combined action of axial force, bending moment
and shear force. To calculate the design forces on the connections, interaction between
the axial force-shear-moment capacities of the members needs to be considered. In the
shear-moment interaction for I-sections based on the maximum shear strength criterion
for yielding [Hodge and Brooklyn, 1957], the yield strength f y was assumed to be the
limiting strength, and strain-hardening of steel was not considered. Approximate
shear-moment (V-M) interaction curves proposed for deep beams (plate girders) were
based on the tension-field action of the web [Basler, 1962]. The shear capacity of the
web was reportedly not affected by the bending moment on the section so long the
flanges did not yield. The web shear capacity dropped quickly as yielding of the beam
flanges increased. Even this study does not consider the effect of strain-hardening in
steel. Based on another study [Cooper et al, 1978], a linear interaction between shear
and bending moment for plate girder design when the design shear was more than 60%
of the factored nominal shear capacity and the bending moment was more than 75% of
the factored nominal bending moment capacity of the section was given (Figure 3)
3

[AISC, 1994]. Such an interaction, however, is not considered in the IS code [IS 800,
1984].
Vu
Vn
1.0

0.6

Mu
0

0.75 1.0

Mn

Figure 3: AISC-LRFD shear-moment interaction. Shear-moment interaction is prescribed

only for I-shaped plate girders with slender webs.
Like in the V-M interaction, the existing axial force-moment (P-M) interaction
curves for steel sections do not consider strain-hardening of steel in design. Moreover,
the hysteretic behavior of the material is also not considered. Thus, while obtaining the
moment-curvature (M-) curves, the strain profile resulting from the simultaneous
application of axial load (P) and a specific curvature is imposed on the section in one
step starting with zero initial curvature and zero initial axial strain, irrespective of the
state of the section at the immediately preceding curvature value; the stresses in the
fibers are obtained directly from the virgin stress-strain curve.
In this study, a fiber model [Murty and Hall, 1994] (Figure 4) is used to develop
the P-V-M interaction curves for sections subjected to known compressive axial loads.
Due to the presence of the axial load, the section is already subjected to some initial
axial strain. Now, if this section is subjected to a specific curvature , to keep the axial
load P constant, the axial strain in the section also changes if the section goes into
inelasticity. A strain-hardened stress-strain curve of steel with the rules for hysteretic
behavior is used in this study (Figure 5. A stressed fiber returns along the virgin stressstrain curve only within the initial elastic range. Fibers that are subjected to increased
axial strain will continue along the virgin stress-strain curve, and those subjected to
reduced strain will return along (a) the virgin stress-strain curve if the fiber is in elastic
4


u
y
1mm thick
Fiber

sh

Figure 4: Fiber model showing the discretization of the beam section along with the
explicit form of stress-strain relationship for steel [Murty and Hall, 1994] used
in this study.
range, or (b) the new unloading stress-strain curve, which is parallel to the initial elastic
portion of the virgin stress-strain curve if the fiber is in the inelastic range. Thus, for
fibers already beyond the elastic limit, unloading takes place along a new unloading
curve. On further unloading, some fibers may reach the translated virgin stress-strain
curve in the other direction, and from then on they follow the same path [Arlekar and
Murty, 2002].

Virgin Curve
Hysteretic Curve

B
A

D
C

P
B3 B2

(a) Non-hysteretic

B4

B22

B1

D2

D3

(b) Hysteretic

Figure 5: Schematic representation of the loading and unloading paths for steel.

D
E

S
R

R2

B21

R1

B23

R3
B

D1

Normalized V-M interaction curves for typical ISMB 600 section [SP:6(1), 1964]
for various levels of the compressive axial load are obtained as discussed above
(Figure 6). The moment is normalized with the nominal plastic moment capacity

M p = f y Z and shear with the nominal shear capacity Vp = y t w d . The V-M interaction
curves obtained using a strain-hardened virgin stress-strain curve are also shown in
Figure 6. The V-M curves without hysteretic loading are marginally higher than the
corresponding curves obtained using the hysteretic loading, only when the axial load is
higher than the yield load Py (Figure 6). Thus, the V-M curves with non-hysteretic
loading, commonly used in codes, are acceptable in static design where the axial load
does not change or is below the member yield load. However, under earthquake
shaking, the axial load can swing by large amount and the V-M curves with hysteretic
loading better reflect the actual lower member capacity and should be considered for
the member design. Figure 7 shows the three-dimensional hysteretic P-V-M interaction
surface for ISMB 600. It is, however, noteworthy that the use of non-hysteretic curves
for connection demand estimate results in a conservative approach.
1.50

1.25

V / Vp

1.00

0.75

0.50

P/Py = 1.0
1.2

0.4

0.6

0.8

0.2
0

Non-hysteretic
Hysteretic

0.25
1.4
0.00
0.00

0.25

0.50

0.75

1.00

1.25

1.50

M / Mp
Figure 6: Shear-Moment interaction showing normalized V-M curves for a typical
ISMB 600 for different axial load levels with and without hysteretic stressstrain curve.
6

Normalised
Axial Force P/Py

Normalised
Shear Force V/Vp

Normalised
Bending Moment M/Mp

Figure 7: Strength interaction in ISMB 600: Normalized P-V-M interaction surface of

ISMB 600 generated using hysteretic stress-strain model.
While developing the P-V-M interaction curves, the following is considered. For
a given normal stress xx (due to axial load and bending moment) in a fiber, the
von-Mises yield criterion for steel represented by
2
2
xx
+ 3 xz
= Y2,

(1)

is used to calculate the available shear capacity xz . Here, Y is taken as the ultimate
stress f u . The curvature is increased from zero to a maximum value corresponding to
the maximum strain r at the extreme fiber, and at each level, the shear and normal
capacities are estimated. The uniaxial stress-strain curve of steel (Figure 5) has a drop in
the stress beyond the strain u corresponding to the ultimate stress. The limiting shear
stress from Eq. (1) when xx = f u , is zero. For strains greater than u , Eq. (1) suggests
that the shear stress xz in fibers is non-zero. However, in this study it is assumed that
all fibers having strains beyond u do not have shear capacity. Further, while obtaining
the limiting V-M boundary, it is assumed that beam flanges and webs do not undergo

buckling. The nominal shear strength V p = y t w d and the nominal bending moment

capacity M p = f y Z

of the section are used to normalize the shear and moment

capacities, respectively. The first yield shear stress corresponding to a state of pure
7

shear is used and defined as y = f y

3.

3.2 Axial Load - Moment (PM) Interaction at Zero Shear Force

The P-M interactions obtained in this study, using the fiber model with
hysteretic stress-strain curve for hot-rolled Indian I-sections are shown in Figure 8. An
upper bound of the normalized P-M interaction curves for zero shear can be expressed
by the following expression
fy P
f
M

= u 1
Mp
f y
f u Py

1.54

(2)

This upper bound limit is conservative for ascertaining the connection demand forces.
Using this for member design would result in an overestimate of the member capacity
and thus, a lower bond limit is required. This is also shown in dotted line in Figure 8.
The design codes generally give such lower bound limit for member design purpose.
However, using this lower bound limit for connection design would result in an
underestimate of the maximum demand on the connection components and may lead
to premature failure of the connection before the member capacity is reached.
The Indian Standard IS:800-1984 assumes a bilinear P-M interaction curve as also
shown in Figure 8. Since, the normalized P-V-M curves obtained in this study are for
fully strain-hardened condition, the IS interaction curved is scaled to the ultimate
strength capacity f u . The IS curve depicts an average member capacity for moments up
to about the nominal plastic moment M p . Thus, it underestimates the moment capacity
of some members, and hence, connections designed using this interaction would be
under-designed.

4. Section Capacity Modification Factors

The P-V-M curves developed in this study are for the full capacity of the section
without considering the effect of uncertainty in the estimation of yield strength,
compactness of the section, slenderness of the member, and the stability against
flexural-torsional buckling of the member. The first factor mentioned above is related to
the strength of the member, and the latter three are related to the stability of the
member. Taking into account all the section capacity modification factors, the
connection design moment is then given by
M pr = M p R y Rs Rc .

(3)

1.5

Computed
For connection design
For member design
IS:800-1984
IS:800 scaled

P / Py

1.0

0.5

0.0
0.0

0.5

M / My

1.0

1.5

Figure 8: P-M interaction curve along with the actual P-M points for V = 0 for Indian
Hot Rolled I-Sections.
Considering the plastic hinges at the ends of the beam with moments M pr , the
corresponding equilibrium compatible shear design force V pr on the connections is
then determined. These, together with the design axial load as appropriate, are the total
demand force on the connections. The effects of these factors and the method of
incorporating them in the member capacity obtained from the P-V-M curves developed
in this study are as follows.

4.1 Yield Strength of Material

The existing code procedures for the design of members are based on the
minimum specified yield strength fy of the steel. The uncertainty in material strength
can cause overstrength and this should be accounted [Goswami et al., 2003]. AISCSPSSB provisions recommend the use of higher yield strength while calculating the
member strength for the determination of the design forces for connection elements
[AISC, 2002]; the ratio R y of the expected yield strength to the minimum specified yield
strength of the connected member as suggested by AISC [AISC, 2002] varies from 1.1 to
1.3 for different grades of steel. In absence of such data of R y for the Indian sections, a
value of 1.0 is used for the P-V-M curves obtained previously, using a yield strength of
250MPa.
9

4.2 Strain Hardening of Steel

Strain-hardening of steel cause increase in member capacity, and hence demand
on the connections. Thus, a strain-hardening factor Rs is introduced given by the
following [Goswami et al., 2003]:

Rs = 1

2
3
4
0.81 + 2 2 + 0.3

100
100
100
100

for 0 y
for y < sh .

(4)

for sh < u

where is the curvature ductility imposed on the section. For steel of f y = 250 MPa
and the Indian sections, the value of Rs is in the range 1.0 to 1.24. The P-V-M curves
developed in this study are based on strain-hardening stress-strain curve for steel.
Thus, the use of these curves for calculating the maximum member capacities includes
the effect of strain-hardening.

4.3 Compactness of the Section

Local buckling of flanges and web of the column adversely affect its maximum
strength. Since the column capacity, M, as determined from the P-V-M interaction does
not consider the effect of the compactness of the section, a compactness factor Rc , is
introduced to account for the reduction in the maximum achievable member capacity
owing to premature local buckling given by

1.0

(b t ) p

Rc = 1.0 0.2

r p

0.8

for

b
p
t

for p <
for

b
r .
t

(5)

b
> r
t

Here, the minimum value of Rc is 0.8. The limiting values for r , p and pd
are prescribed in the codes. However, the limits prescribed in the codes are originally
for the purpose of beam design and thus, will tend to give a conservative
underestimate of the member strength. But, in connection design, the upper bound
strength is required. Moreover, these values are for prismatic flange and web. For
Indian hot rolled I-sections with tapered flanges, such limits of r , p and pd for
connection design purpose and dependence of Rc on these needs to be prescribed.
10

5. Connection Design Examples

To illustrate the current Indian connection design practice and its limitations,
consider a MRF with ISMB 300 beam and ISMB 600 columns with spans as shown in
Figure 9. The bay span considered is 4m. The uniformly distributed gravity load
(including dead load, fraction of live load on roof and floor as per IS 1893 (Part I), 2002,
and roof finish load) is 16kN/m on the roof beam and 15kN/m on the floor beam.
Considering the structure to in seismic zone V [IS 1893 (Part I), 2002], it is analysed for
the different load combinations. The maximum joint moment and shear forces for load
combination 1.3(DL+LL+EL) are respectively 43kNm and 48kN. The beam-to-column
connection is designed for this force.

5.1 Common Design Practice

In the common design practice, members and connections are designed based on
the linear static analysis results. The web is considered to carry the shear and the two
flanges carry the flexure, in the form of tension and compression. Accordingly, from the
static analysis results above, 6mm fillet welds of 100mm length on both sides of the web
are sufficient to carry the shear. The flanges can be connected to the column through a
10mm full penetration butt weld, or by 10mm fillet weld along the straight portions of
the flanges. As there are no particular recommendations for the type of connection
arrangement to be adopted in the existing Indian Standards, such simple form of
connection can be designed still adhering to the code provisions, if desired.
16kN/m
26.5kN

3.5m
15kN/m

7kN
ISMB 300
(TYP)

4m

ISMB 600
(TYP)

4m

4m

4m

Figure 9: Structural arrangement: Member sizes, boundary conditions with loadings.

11

5.2 Design Procedure in SP:6(6)

As a second step, the same beam-to-column connection is checked following the
design philosophy given in the ISI Handbook for Structural Engineers: Application of
Plastic Theory in Design of Steel Structures [SP:6(6), 1973]. SP:6(6) recommends that the
connections be designed for the nominal plastic moment that is to transmitted from one
member to another. Thus, under the condition of an extreme shaking, assuming that
plastic hinges are formed at the beam ends, the design forces for connection design is
the nominal plastic moment M p (161.6kNm) of the beam and a shear of 120.6kN,
considering the critical sections to lie at the beam-column interface. Thus, now the
design shear and moment are increased by 1.5 to 2.8 times over the structural analysis
results of the frame discussed earlier, and now, the connection designed earlier
becomes inadequate. Further, the code does not specifically say that plastic analysis
and design needs to be done for seismic conditions. Thus, it remains at the hand of the
designer to choose the type of analysis and design one wishes to do, and in the process,
the structural safety is put at stake.

5.3 A Rational Design Procedure

Supplementary to the above, as discussed previously, now considering the
overstrength factors R y = 1.3, Rs = 1.24 and Rc = 1.0, the design moment at the column
face becomes 261kNm plus the shear times the length of the connection reinforcement
region and plastic hinge length. Thus, there is an increase of more than 62% in the
design moment alone. This can cause premature failure of the connection even before
the beam reaches its full plastic capacity resulting in collapse of the structure. In
addition, such simple connection schemes discussed in Sections 5.1 and 5.2 do not
facilitate smooth flow of forces through the connection region; stress concentration at
the beam flange-column flange junction severely affects the functioning of the
connection. To address these difficulties, a rational method of design of moment-shear
connections is developed wherein a cover plated rib plated connection is designed for
the overstrength beam forces [Arlekar and Murty, 2003]. A step-wise procedure
presented for standard AISC sections, is extended here for the current design problem
with Indian sections. Following this, the associated shear force also increases to
186.5kN, an increase of about 55% compared to a value of 120.6kN in Section 5.2. The
resulting connection configuration is shown in Figure 10. However, this scheme works
best on sections with wide non-tapered flanges; Indian sections have very small flange
12

width [Goswami et al., 2003]. As such, the available width of cover plate is much less
and may be insufficient to transfer the forces in higher structures with higher forces; the
example frame is a nominal two-storey lightly loaded structure chosen only with the
intention to discuss the important issues. Also, due to tapering of these flanges,
additional inner rib plates cannot be provided efficiently to further reinforce the
connection, if required.

5.4 Discussion
In the above, a rational method for moment-shear connection design is
proposed. Moreover, the design procedure outlined in SP:6(6) is found to be inadequate
in that it does not account for the increase in the maximum demand that may be
mobilized due to the overstrength factors discussed in Section 4. Also, given the sizes of
the available hot-rolled section, it may not be possible to develop moment-shear
connections for tall structures in high seismic areas.

14mm Fillet
Weld

Cover Plate
1509014

ISMB 300
Beam
Web not Connected

600

300
Outer Vertical
Rib Plate
555510

Detail W

ISMB 600
Column

10mm Fillet
Weld

Complete Joint
Penetration Weld

A
150

Section A-A

Fillet Weld
(14mm)

Detail W

Figure 10: Beamto-column connection arrangement: Geometry, location of connection

elements and type of welds.

13

6. Conclusion
In the recent times, the design of connections in welded steel MRFs has seen
a major change. Most developed codes now recommend that the connections for
MRF should be designed using the capacity design concept. This means that
the

connections

should

be

able

to

resist

and

transfer

the

forces

and

deformations corresponding to the maximum capacity that is expected to be

mobilized in the connected members. In this regard, axial load-moment-shear
interaction plays an important role on deciding upon the maximum mobilized demand.
Further, material strain hardening and higher material strength over the nominal
specified values significantly increases the demand on the connections over the code
specified values. With this, although the basic perspective of moment-shear connection
design is in light, the idea can be effectively put to practice only with clearly laid out
code provisions and availability of appropriate raw materials, namely proper
wide-flange non-tapered hot-rolled sections.

References
AISC, (1994), Metric Load and Resistant Factor Design Specification for Structural Steel
Buildings, American Institute of Steel Construction, Inc., Illinois, USA, 1994.
AISC, (2002), Seismic Provisions for Structural Steel Buildings, American Institute of Steel
Construction, Inc., Illinois, USA, 2002.
Arlekar, J. N., and Murty, C. V. R., (2002), P-V-M Interaction Curves for Seismic
Design of Column Base Connections, Engineering Journal, AISC, 3rd Quarter, 2002.
Arlekar, J. N., and Murty, C. V. R., (2003), Capacity Design of Welded Steel MRF
Connections, Proceedings of the 4th International Conference on behaviour of Steel
structures in Seismic Areas, 9-12 June 2003, Naples, Italy.
Basler, K., (1962), Strength of Plate girders Under Combined Bending and Shear,
Journal of Structural Division, Proceedings of the American Society of Civil
Engineers, Vol.87, No.ST7, pp.181-197, 1962.
Cooper, P. B., Galambos, T. V., and Ravindra, M. K., (1978), LRFD Criteria for Plate
Girders, Journal of Structural Division, Proceedings of the American Society of
Civil Engineers, Vol.104, No.ST9, pp.1389-1407, 1978.
Goswami, R., Arlekar, J.N., and Murty, C.V.R., Limitations of available Indian HotRolled I-Sections for use in Seismic Steel MRFs, (another paper submitted with this
paper).
14

Hodge, P. G. Jr., and Brooklyn, I. N., (1957), Interaction Curves for Shear and Bending
of Plastic Beams, Journal of Applied Mechanics, American Society of Mechanical
Engineers, Paper No.57, APM-19., pp.453-456, 1957.
IS 800, (1984), Indian Standard Code of Practice for General Construction in Steel,
Bureau of Indian Standards, New Delhi, 1995.
IS 1893 (Part 1), 2002, Criteria for Earthquake Resistant Design of Structures, Part 1:
General Provisions and Buildings, Bureau of Indian Standards, New Delhi, 2002.
Murty, C. V. R. and Hall, J. F., (1994), Earthquake Collapse Analysis of Steel Frames,
Earthquake Engineering and Structural Dynamics, Vol.23, pp 1199-1218, 1994.
SP6(1), (1964), Indian Standard Handbook for Structural Engineers: Structural Steel
Sections, Indian Standards Institution, New Delhi, 1964.
SP6(6), (1973), ISI Handbook for Structural Engineers: Application of Plastic Theory in
Design of Steel Structures, Indian Standards Institution, New Delhi, 1973.

Notations

The following symbols are used in this paper:

b

Width of plate element; width of flange of section

d
f

=
=

Depth of member
Stress
Ultimate normal stress

Minimum specified normal yield stress of steel

=
=
=

Thickness of plate element

Bending moment
Nominal flexural strength of member

Section plastic moment capacity using minimum specified yield

=
=
=

Connection design moment

Factored moment
Axial load

=
=
=
=

Yield load

fu
fy
t
M
Mn

Mp
M pr
Mu

Py
R
Rc
Rs
Ry

Section capacity modification factor

Strength reduction factor due to compactness
Strength reduction factor due to strain hardening of steel

Strength reduction factor due to uncertainty in the estimation of

yield strength

V
Vn

Shear force

Nominal shear capacity of section

Vp
V pr

=
=

Section plastic shear capacity using minimum specified yield

Connection design shear
15

Vu

Factored shear strength of member

Failure stress in Von Mises criterion

=
=

Plastic section modulus of the member

Normal strain

Rupture strain

sh
u
y

Strain-hardening strain

Strain corresponding to ultimate stress

=
=

Yield strain
Resistant safety factor

Curvature

Slenderness parameter

=
=
=
=

Limiting slenderness parameter for compact section

Limiting slenderness parameter for compact section with minimum
guaranteed plastic rotation capacity
Limiting slenderness parameter for non-compact section
Curvature ductility of the section

=
=

Yield curvature ductility

Strain-hardening curvature ductility

p
pd

y
sh
u
, xx
y

Ultimate curvature ductility

=
=

Normal stress
Yield stress

Ultimate stress

xz
y

Shear stress

Minimum specified shear yield stress of steel

16