Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

ROLE OF STRESS TRIAXIALITY ON THE SEISMIC DAMAGE

OF STEEL MOMENT FRAMES

Ayman A Shama 1, Rasko P Ojdrovic 2, Benjamin W Schafer3, and Mehdi S Zarghamee 4

ABSTRACT

This paper presents the results of our study into the role of stress triaxiality in the
fracture of steel moment connections. In pursuit of this objective, a building that
suffered considerable damage during Northridge earthquake is selected for
analysis. The rotational demands of typical connections of the structure during
this seismic event were determined through three dimensional finite element
analysis of the building model. The rotational capacities of the same connections
were determined using stress triaxiality analysis for an exterior connection.
Comparisons of the rotational demands of several connections in the building to
the rotational capacities determined by the stress triaxiality analysis showed that
fracture of steel moment connections of the building are governed by stress
triaxiality. Based on a number of moment connections analyzed, the paper
presents deign guidelines for moment connections taking into account the
fracture incidence due to stress triaxiality effect.

Introduction

The Northridge Earthquake of January 17, 1994, demonstrated that welded steel
moment frame connections are prone to damage under severe ground motions. An enormous
amount of new research under the sponsorship of the SAC joint venture has been conducted for
a better understanding of the seismic performance of moment frame connections and to develop,
guidelines for its seismic design. Early studies (Yang and Popov 1995; Chi et al 2000)
explicitly incorporated fracture mechanics methods with finite element analysis to simulate
fracture of steel moment connections. These approaches usually assume a pre-existing flaw of
critical size for an analysis to be performed and out of the scope of this paper. Alternatively,
stress triaxiality has been cited as a prospective reason for the indigent behavior of these
connections (SAC 1996; Ojdrovic et al 2000). The objective of this work is to determine to what
extent stress triaxiality contributed to the fracture of moment connections during an earthquake.
The study complements an earlier study by Schafer et al (2000) who developed a simple failure
1
Structural Engineer, Parsons Transportation Group Inc., 110 William Street, New York, NY, 10038; formerly, Sr.
Engr. Simpson Gumpertz & Heger Inc., 297 Broadway, Arlington, MA, 02474
2
Senior Project Manager, Simpson Gumpertz & Heger Inc., 297 Broadway, Arlington, MA, 02474
3
Assistant Professor, Department of Civil Engineering, The Johns Hopkins University
4
Principal, Simpson Gumpertz & Heger Inc., 297 Broadway, Arlington, MA, 02474
criterion for approximating fracture initiation in steel structures as a function of the stress
triaxiality and maximum principal stress. The validity of this approach is examined by
implementing it as an assessment of connections capacities for a building in the Los Angeles
area that suffered extreme damage during the Northridge Earthquake of 1994. The rotational
demand of typical connections of the building are calculated and compared to the rotational
capacity of the moment connections as determined by stress triaxiality analysis.

Review of Triaxiality Failure Criterion

Stress triaxiality (hereafter referred to as triaxiality) is defined as the ratio of the


maximum principal stress to the Von Mises stress at a point. This measure provides an
indication of the stress level that the material may undergo with respect to the stress level that
the same material is subject to under uniaxial loading conditions. Schafer et al (2000) conducted
a series of nonlinear finite element analyses on notched round bars, small scale tension tests and
full scale moment connections to assess stress levels at fracture. They found that fracture
locations are consistent with locations of high stress triaxiality in all the cases they considered.
They plotted the relationship of the stress triaxiality to the normalized maximum principal stress
in Fig. 1 for all the analyzed cases and were able to limit a range (two curves) in which fracture
is predicted by stress triaxiality and maximum principal stress. Therefore, when the combination
of maximum principal stress and stress triaxiality demands for a moment connection reaches this
range of limiting values fracture occurs.

1.80

1.70

1.60

1.50
fracture as predicted by
σmax 1.40 triaxiality and prin. stress

σult 1.30

1.20
Notch rd. bar (McAdam 1944)
1.10
Notch rd. bar (K&Y 1997) 1mm avg., 12mm dia.
Notch rd. bar (K&Y 1997) 3mm avg., 12mm dia.
1.00 T-Stub (Test 7)
T-stub (Test 13)
0.90
Moment Connection (SAC PN2)
0.80
1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40 2.60 2.80 3.00
σ
T2 = σmax
eff

Figure 1. Stress triaxiality and maximum principal stress at fracture after Schafer et al 2000

Case Study of a Building Damaged During Northridge Earthquake 1994

Fig. 2 shows a typical plan of the building subject this investigation with the locations of
the steel moment frames. A comprehensive survey of the building after the earthquake showed
that the building was leaning 6 inches to the north at roof level. Inspection of moment frame
connections indicated fracture damage at 29 locations in the north-south oriented frames. These
locations are shown in Fig. 3.

North Moment Frame


31’-4" 31’-4"

West Moment Frame

East Moment Frame


Stair

28’-0"

28’-0"
well
28’-0"

28’-0"
Stair
well Elevators
N
31’-4" 31’-4"
South Moment Frame

Figure 2. Typical building floor plan

Two dimensional inelastic analyses on the structure (Paret and Sasaki 1995) did not
predict precisely the locations of observed damages. Other 2-D studies (Anderson and Fillippou
1995) highlighted the significance of carrying out three dimensional analysis to account for the
building eccentricities. Later Chi et al (1998) conducted comprehensive analyses on the
structure to investigate the accuracy and reliability of a state-of-the-art static and dynamic
inelastic frame analyses to evaluate the seismic performance. They emphasized the effect of
several parameters such as modeling stiffness of structural elements other than the moment
frame, strength degradation and hysteric behavior of connections on the accuracy of results.
W36X135 W36X135
W36X150 W36X150
Analysis of Connection Capacity
W14x500 W14x455 W14x428 W14x398 W14x370 W14x342 W14x311

W14x500 W14x455 W14x428 W14x398 W14x370 W14x342 W14x311

W36X150 W36X150

W36X170
Examination of the damage by W36X170
W36X170
N
W36X170
floors in Fig. 3 indicates that the W36X194 W36X194

connection damage was concentrated in W36X194 W36X194

the upper floors of the west frame with W36X230 W36X230

less damage in the east frame. No do do

do do
damage was reported in the north and do do
south frames. Given the levels of do do

damage observed, the response of the do do

W36 beams connected to W14 columns do do

W36X230 W36X230
are of primary concern for better
W36X300 W36X300
understanding the rotational capacity of do do
W14x730

W14x730

the building connections. Therefore an W36X300 W36X300


exterior connection located in the 14th
floor in the building was selected for East Moment Frame West Moment Frame
analysis as shown in Fig. 4.
Figure 3. Locations of connections damage
The model consists of a beam: W36 x 170, A36 Steel, fy = 326 MPa , fu = 458 MPa , a
column: W14 x 342, A572 Grade 50 Steel, fy = 395 MPa , fu = 560 MPa . The initial modulus of
elasticity E is assumed equal to 200,000 MPa and strain hardening initiates at 1.5% strain. The
weld is assumed to have minimum specified properties for an E70 electrode: fy = 414 MPa and
fult = 496 MPa (Blodgett 1976) with an E = 200,000 MPa and strain hardening initiation at 0.8%
strain. The model was analyzed monotonically under displacement control directed upward and
applied at the beam mid span. The shear tab was modeled as connected to the beam web through
rigid bars and point wise constraints for the bolts. The analysis was conducted to a total drift
angle of 1.7%.

Detail A
Symmetry
Plane
Disp.

Figure 4. Finite element model and detail of beam bottom connection

Fig. 5 displays stress triaxiality through the beam bottom flange. It is shown that stress
triaxiality exceeds unity in the weld at the beam to column interface, which suggests that
fracture might start in this region. Fig. 6 details the maximum principal stresses at the column
flange. One can observe that regions of high principal stresses are concentrated in the interface
with the beam flange through the weld and stresses increase in the direction of plane of
symmetry around the cope hole. It is also observed that the backup bar has no significant effect
on the behavior of the connection.
On the basis of these observations, it can be concluded that fracture as indicated by high
triaxiality is presumably to start at the beam to column flange interface and emanate into the
column flange. Inspection of the damaged connections in the building after the Northridge
Earthquake agrees with this finding. According to Freeman and Nungent (1995), all weld
fractures in the building seemed to have initiated from the root of the full penetration welds and
extended through its thickness to the interface with the column flange. In other cases, the
fracture extended into the column flange thickness developing a flange tear-out or a divot.
Figure 5. Stress triaxiality at fracture. Illustrating beam bottom flange

Figure 6. Maximum principal stress at beam-to-column interface in the


column flange and through beam web

Two elements at the beam column interface were selected for evaluating the connection
capacity. As shown in figure 5, element 12133 is located on the weld at the beam flange
interface with the column flange near the center of the beam web. Element 72133 is located in
the same region but almost half way from the beam web. Comparisons of the maximum
principal stress and triaxiality for both elements are illustrated in Fig. 7. The results indicate that
onset of damage may start in the connection at rotations in the range of 0.6% and fracture due to
triaxiality and maximum principal stress may occur at rotations ranging from 1% to 1.67%.
1.8
θ = 0.0167 rad

Normalized Principal Stresssσ max/σ ult


1.6

1.4 θ = 0.011 rad


1.2
θ = 0.0067 rad
1.0 θ = 0.003 rad

0.8
Triaxiality Failure Envelope
0.6
θ = 0.0017 rad Element 12133 on beam
element 72133 on column
0.4 Lower bound
Upper bound
0.2 θ = connection rotation
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Triaxiality T2 σ max/σ eff

Figure 7. Capacity of an exterior connection located at 14th floor as dictated by stress


triaxiality

Analysis of Structural Demand

Building Model

The building has a regular geometry with 17


stories above ground, one story below grade,
and a mechanical penthouse on the roof.
The commercially code SAP 2000 was used
for the analysis. The structural model
developed for the analysis (Figure 8)
consists of all the moment resisting frames,
gravity columns, floor beams and floor slabs
of the original structure. Framing system
was modeled using beam elements and P-∆
effect was considered in the analysis. End
offsets were specified to account for the
finite size of beam and column intersections
at the joints. It is assumed that the
composite floor action for gravity columns
to floor beams connections is negligible.
Consequently, the rotational degrees of
freedom were released at these connections. Figure 8. 3-D FEM for the building
In SAP 2000 the flexibility of the panel zone and its contribution to joint deformation is
modeled by allowing reduction of the rigid lengths of the panel zone dimensions. Calculations
made for different combinations of beam-to-column connections (Krawinkler 1978)
representing joints at east, west, north, and south frames suggested average values of 0.7 and
0.90 for the stiffness reduction factors for both exterior and interior connections respectively.
Other Non-moment resisting frames connections were modeled with panel zone reduction factor
equals zero, i.e. center-line dimensions.
The seismic weight of the structure was calculated as the sum of the dead loads (structural
skeleton + partition weights + perimeter curtain wall weight) and live loads on each floor. The
unit weight of steel for all sections was taken as 76.8 kN/m3. The first floor, penthouse, and roof
were modeled as 0.15 m shell elements with 23.5 kN/m3 unit weight for concrete. The rest of the
floors were modeled as 0.12 m shell elements with 15.7kN/m3 unit weight for light weight
concrete. According to UBC 1979 edition (the latest edition when the structure was constructed)
a uniformly distributed dead load of 0.96 kN/m2 maximum was required for the building design.
For analysis, a value of 0.72 kN/m2 was added to the floors from first to seventeenth and
penthouse floor to account for partition loads. Additional value of 0.96kN/m2 was added to
account for other types of loads such as column cladding, ceiling, finishing, and electrical
equipment. Therefore a total value of 1.68 kN/m2 distributed load was used to represent partition
and other loads. The curtain wall is composed of structural glass (0.72 kN/m2) and Gypsum
blocks (0.45 kN/m2). An average value of 0.60 kN/m2 was taken for the curtain wall.

Building Response to the Seismic Event


The only instrumentation on the building is located at the 17th floor. The instrument
measures three components of acceleration. The digital records from this instrumentation were
obtained from the California Division of Mines and Geology (CDMG). Spectral analyses were
conducted on the recorded roof motion (Fourier spectra) for the two horizontal components and
results are displayed in Fig. 9 where the captured natural periods on both directions are shown.
W5: Smoothed Spectrum (w Hanning) W5: smoothed spectrum (Hanning)
5
0.61 HZ 1.64 sec 0.256 HZ 3.9 sec
.22 HZ 4.55 sec 0.732 HZ 1.37 sec
4 NS 4 EW
1.27 HZ 0.79 sec
A 1.07 HZ 0.93 sec A
3 3
m m
p p
2 2
l l
1.35 HZ 0.74 sec 1.76 HZ 0.57 sec
. .
1 1
2.69 HZ 0.37 sec
0.4 HZ 2.5 sec
0 0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 0.0 0.4 0.8 1.2 1.6 2.0 2.4 2.8
Frequency (Hertz) Frequency (Hertz)

Figure 9. Fourier Amplitude Spectra of the recorded roof motion


In order to verify the building structural model, modal analysis was carried out prior to running
time history analysis to identify the dynamic characteristics of the model and compare with
results from the analysis of the 17th floor record. Table 1 shows a comparison between the
Table 1. Theoretical and Measured Natural Periods
Mod # E-W Direction N-S Direction Torsion
1
SAP MPMR2
FAS FAS SAP MPMR FAS SAP MPMR
Mode % Model % Model %
l (Ux) Uy) (Ux) (Uy)
1 3.91 4.18 61 14.4 4.55 4.16 14.4 60 ----- 2.57 2.2 (Ux)
2 1.37 1.44 9.94 0.15 1.64 1.42 0.18 11.85 1.04 0.91 0.42 (Ux)
3 0.79 0.85 3.34 .00 0.93 0.84 0.01 4.10 0.76 0.54 0.38 (Ux)
4 0.57 0.59 1.3 .00 0.74 0.62 0.00 1.75 ---- ----
5 ------ 0.44 0.89 .00 ------ 0.50 0.00 0.90 ------ ----- .
1 = Fourier Amplitude Spectrum 2 = Modal Participation Mass Ratio
theoretical and the measured natural periods. As shown, there is a satisfactorily correlation
between the computational and the measured periods for most of the modes.
The Canoga Park (7769 Topanga Canyon Blvd) ground motion record located about 1.9
miles away from the building was used as the input ground motion (Sommerville et al 1995).
The three components of the earthquake ground motion were applied simultaneously to the
model structure. The recorded relative displacements for the roof in both the E-W and N-S
directions are plotted in Fig. 10. The computed relative displacements are also plotted in the
figures.
250 500
200 400
E-W N-S
DISPLACEMENT (mm)

DISPLACEMENT (mm)

150 300
100 200
50 100
0 0
-50 -100
-100 -200
-150 CALCULATED -300
RECORDED CALCULATED
-200 -400 RECORDED
-250 -500
0 10 20 30 40 50 60 0 10 20 30 40 50 60
TIME (SEC) TIME (SEC)

Figure 10. Recorded and Computed Relative Displacements at the Roof Level
20
One can observe that there is a
18
reasonable correlation between the computed
16
and recorded displacements. It is also shown 14
that the response of the structure in the north-
STORY #

12
south direction is quite higher than the east- 10
west direction. This agrees with the reported 8
damage of the structure after the earthquake 6
where the damage was only concentrated in the 4 EAST FRAME
2 WEST FRAME
east and west frames. The joint rotational
demands for the east and west frames are 0

plotted in Figure 12. As shown in the figure, 0.002 0.004 0.006 0.008 0.01
Total Rotation of Exterior Joints
rotational demands for the connections in the
14th floor are determined as 0095. Figure 12. Rotational demands for frames
This agrees with the connection rotational capacity dictated by stress triaxiality (.01-.0167). It is
also shown that the building model captured the damaged connections at floors starting from the
13th to the 17th on the west frame, which suffered high connection damage.

Application of the Procedure to Connections in other Buildings

The aforementioned procedure for 2.00

prediction of moment frame connections 1.80


damages was successfully applied to 1.60
connections in other buildings in the Los
1.40
Angeles area that experienced damage Failure envelope

ult
during the Northridge earthquake. 1.20 upper bound

max/
Performance characteristics of these 1.00 low er bound

connections are summarized in Table 2. Trillium


0.80 Trident (exterior)
Figure 13 displays the stress demands for
0.60 Trident interior
different connections. It can be observed
Westside
that onset of damage (flaw initiation) 0.40
SAC PN2
occurs at an average values of 1.65 and 0.20
1.42 for stress triaxiality and normalized 0.50 1.00 1.50 2.00 2.50 3.00
maximum stress respectively. These T2 = σ max/σ eff
average values, however, are 1.83 and
1.55 at fracture. On the basis of these Figure 13. Stress demands for different
results, the maximum stress in such analyzed connections
connections can be expressed as:
f max = 0.84 f u Ts (1)
in which, fmax and fu are the expected maximum and tensile stresses of the material in the
element where fracture initiates; and Ts is a coefficient to account for stress triaxiality. Adopting
a value of fmax = 1.4 fy (yield stress), and a value of 1.65 for the triaxiality coefficient the
expected moment for such connections is quantified as:
M pr = 1.94 Z p f y (2)
where, Mpr = the expected moment at which a flaw may initiate; Zp = the plastic modulus of the
connection; and fy = yield stress of the element material where a flaw starts. Equation (2) is
consistent with equation (3-1) in FEMA-350. It is, however, less conservative as it accounts for
fracture initiation.
Conclusions

This paper emphasized the significance of stress triaxiality in analyzing steel moment
frame structures under seismic loads. The study demonstrated that triaxiality of stresses was a
major factor for the fracture of moment connection buildings during the Northridge earthquake.
Based on a number of moment connections analyzed, the paper presented an equation for the
expected seismic moment for such connections taking into account triaxiality of stresses.
Table 2. Performance of Different analyzed moment frame connections
Onset of Damage Fracture
Structure Beam Column
θ% T σmax/σult θ% T σmax/σult
Trident Center LA W36x160 W14x283 0.79 1.65 1.35 2.55 1.80 1.55
1
Trident Center LA W36x160 W14x283 0.80 1.67 1.40 2.23 1.98 1.70
Westside Towers LA W36x150 W14x283 0.67 1.61 1.35 1.67 1.70 1.47
Trillium Building LA W36x170 W14x342 0.60 1.58 1.32 1.67 1.82 1.52
SAC PN2 W36x150 W14x257 0.57 1.75 1.44 1.14 1.87 1.58
1 interior connection

References

Anderson, J., and Fillippou, F. (1995). Dynamic Response Analyses of the 17-Story Canoga Building,
SAC report no. 95-04 Part 2, Applied Technology Council, Redwood City, CA., pp 12-1 to 12-53.
Blodgett, O.W., (1976), Design of Welded Structures, The James F. Lincoln Arc Welding Foundation,
Cleveland, Ohio.
Chi, W.M, El-Tawil, S., Deierlein, G., Abel, J.(1998). Inelastic Analysis of a 17-Story Steel Framed
Building Damaged During Northridge, Engineering Structures, Vol. 20 pp 481-495.
Chi, W.M., Deierlein, G.G., Ingraffea, A.R., (2000). Fracture Toughness Demands in Welded Beam-
Column Moment Connections, Journal of Structural Engineering, ASCE, 126(1), 88-97.
Freeman, S., and Nungent, W., (1995). Inspection and Repair of Welded Steel Moment Frame
Connections, The Trillium 6310, 6320 Canoga Avenue. Report No. 951783, Wiss, Janney, Elstner Inc.
Krawinkler, H., (1978). Shear in Beam-Column Joints in Seismic Design of Steel Frames, Engineering
Journal, AISC, Third Quarter, 82-91.
Ojdrovic, R.P., Schafer, B.W., Zarghamee, M.S., (2000). Fracture and the Role of Triaxiality for Steel
Structures, Structural Congress, Philadelphia, Pennsylvania.
Paret, T. and Sasaki, K. (1995). Analysis of a 17-Story Steel Moment Frame Building Damaged by the
Northridge Earthquake, SAC report no. 95-04 Part 2, ATC, Redwood City, CA., pp 10-1 to 10-52.
SAC Joint Venture (1996). Technical Report: Experimental Investigations of Beam-Column
Subassemblages, Report No. SAC 96-01, Sacramento, California.
Schafer, B.W., Ojdrovic, R.P. and Zarghamee, M.S. (2000) Triaxiality and Fracture of Steel Moment
Connections”. Journal of Structural Engineering, ASCE, Vol. 126, pp. 1111-1255.
Sommerville, P., graves, R. and Saikia,(1995). Characterization of Ground Motions During the Northridge
Earthquake of January 17 1994, SAC Report no. 95-02. ATC. Redwood City. California.
Yang, T.-S., and Popov, E. P., (1995). Behavior of Pre-Northridge Moment Resisting Steel Connections,
Report No. UCB/EERC-95/08, EERC, University of California at Berkeley, 48 pp.
Federal Emergency Management Agency (2000) Recommended Seismic Design Criteria For New Steel
Moment Frame Buildings, California

You might also like