Performance of Steel Structures During The 1994 Northridge Earthquake

Performance of steel structures during the
1994 Northridge earthquake
Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by University of Toronto on 12/02/14

For personal use only.
Robert Tremblay, Peter Timler, Michel Bruneau, and

Andre Filiatrault
Abstract: The performance of concentrically braced steel frames and moment resisting steel frames
during the January 17, 1994, Northridge, California, earthquake is examined. Most of the observations
made during the reconnaissance visits confirmed the current knowledge on the inelastic response of
these structural systems. This permits the anticipation of proper seismic behavior for buildings designed
according to the seismic provisions that have been recently introduced in the Canadian building code
and standard for steel structures. In some cases, however, the observed damage raised concerns that
should be addressed in future investigations or next editions of these codes. Preventing potentially
hazardous nonstructural damage, avoiding premature nonductile failures anywhere along the lateral load
paths, limiting structural and nonstructural damage due to brace buckling, and accounting for the
vertical ground motion are among those issues.
Key words: earthquake, seismic, steel, concentrically braced frames, moment resisting frames, weld.
R&um&: Dans cet article, on examine et commente le comportement de charpentes mCtalliques avec
contreventement en treillis ou cadres rigides lors du stisme qui a eu lieu le 17 janvier 1994 5
Northridge, en Californie. La majorit6 des observations faites lors des visites effectuCes sur le site
confirment les connaissances dCji acquises sur le comportement non lintaire de ces systbmes structuraux
lorsque soumis aux stismes. Ceci permet de croire que les bitiments conGus selon les dispositions
sCismiques nouvellement introduites dans le code du bitiment canadien et la norme canadienne pour la
calcul des structures d'acier seront adCquats dans 1'CventualitC de tremblements de terre importants.
Dans certains cas, cependant, les dommages subis soulbvent des interrogations qui devraient faire l'objet
d'Ctudes futures en vue dlamCliorer ces normes. Parmi les points qui devraient Ctre examinks, on
retrouve la prCvention des dommages non-structuraux 5 risque, la prevention de ruptures subites et
non-ductiles le long du cheminement des efforts induits par les charges latCrales, la rkduction des
dommages imposCes aux ClCments structuraux ou non-structuraux lors du flambement des membrures
diagonales des contreventements ainsi que la prise en compte des effets des mouvements verticaux du sol.
Mots c l b : tremblement de terre, sCisme, charpentes mttalliques, contreventement en treillis, cadre
rigide, soudure.
Introduction
The earthquake resistance of steel frames has been known to
be tremendously reliable overall, with steel building collapses so far being the rare occurrence worldwide (Yanev et al.
199 1). In the January, 1994, Northridge earthquake, steel
frames also sustained well the ground shaking, as no fatali-
ties were attributed to unsatisfactory performance of steel

structures and no collapses of steel bufidings were reported
(AISC 1 9 9 4 ~ ) .
However, evidences of significant inelastic response and
several structural deficiencies were observed on steel-framed
structures after the event. Most of these observations corroborate the current state-of-the-art in seismic design in Canada.
Nevertheless, some cases need to be brought to the attention
of the Canadian engineering community. The reflection on
these cases may lead to modifications to our present design
and construction practices in order to ensure that safer steel
structures be built in the future.
This paper reports and comments on the observations
made by reconnaissance team members of the Canadian Association for Earthquake Engineering (CAEE), which visited
the epicentral area of the Northridge earthquake. Contributions from local structural engineers were also included in
the paper for sake of completeness. A total of 14 cases are
presented, among which 12 are building structures. The
structures were either concentrically braced frames, moment
resisting frames, or a combination of the two.
u
Received June 16, 1994.

Revised manuscript accepted December 22, 1994.
R. Tremblay and A. Filiatrault. Department of Civil

Engineering, Ecole polytechnique, Montreal, QC H3C 3A7,
Canada.
P. Timler. Sandwell Inc., Engineering and Construction
Services Group, 1190 Hornby Street, Vancouver, BC
V6Z 2H6, Canada.
M. Bruneau. Ottawa-Carleton Earthquake Engineering
Research Center, Department of Civil Engineering,
Universitv of Ottawa. Ottawa. ON K1N 6N5. Canada.
Written discussion of this paper is welcomed and will be
received by the Editor until August 31, 1995 (address inside
front cover).
Can. J. Civ. Eng. 22: 338-360 (1995). Printed in Canada 1 Irnprirnt au Canada
Tremblay et al

The first part of the paper outlines the characteristics and

reviews the relevant seismic design fundamentals of steel
concentrically braced frames and moment resisting frames.
The observations from the reconnaissance site visits are presented in the second part of the paper. Finally, for each structural system, a comparison between the observed behavior
and the performance expected by the design provisions
included in the current edition of the National Building Code
of Canada (NBCC) (NRCC 1990) and standard for steel
structures (CSA 1989) is presented.
Seismic design fundamentals of steel

concentrically braced and moment
resisting frames
Concentrically braced frames
For low- and medium-rise structures, the concentrically
braced frame system has been extremely popular. It is simple
to design and fabricate and allows the required lateral strength
and stiffness to be easily provided at a low cost.
Lateral stability of concentrically braced frames is ensured
bv vertical cantilevered trusses anchored to the foundation.
This truss action is achieved by introducing diagonal bracing
members within the main beam and column framework. For
ease of design and construction, simple beam-column connections are most often used throughout the structure. With
this system, lateral loads primarily induce axial loads in the
members of the bracing bents. Therefore, inelastic action
under the design base earthquake is associated with axial
deformations in these members.
To maintain the gravity load carrying capacity of beams
and columns during the ground shaking, axial inelastic deformations in these members are not desirable. On the other
hand, bracing members carry little or no gravity loads. Thus,
these diagonal members are good candidates to act as critical
elements in concentrically braced frames to form the energy
absorbing and dissipating mechanism of the lateral load
resisting system.
Braces can dissipate energy through yielding in tension
and, additionally, through inelastic buckling in compression.
Numerous investigations (e.g., Popov and Black 1981) have
shown that the latter is achieved by using stockier braces.
Upon buckling, a plastic hinge forms at mid-length of the
brace and, possibly, at its ends depending on the fixity conditions. Energy is absorbed through inelastic bending of the
plastic hinges under compressive loads as well as upon
straightening up of the brace during subsequent tension loading. Width-to-thickness ratios of the bracing members must
be kept low to prevent local buckling and premature failure
at the hinge locations.
Ductile and stable behavior of concentrically braced frames
can be expected under severe ground shaking if inelastic
response is constrained to properly detailed bracing members, and if brittle failure is avoided in the other components
of the lateral load resisting system. Comprehensive seismic
design provisions have been introduced recently in the Canadian S16. l standard (CSA 1989) to meet these capacity design
objectives (Redwood and Channagiri 1991). The extent of
detailing prescribed for concentrically braced frames depends
on the magnitude of the anticipated inelastic deformations,
and is therefore related to the seismic load level adopted in
design. This is done by defining three different categories of

bracing systems for which a different design seismic load
level is prescribed by the NBCC. For each category, a different set of seismic detailing provisions is required by the
S 16.1 standard.
The first category is referred to as ductile braced frames
for which a ductile behavior is mandatory for its survival
during the design base earthquake. The storey shear must be
shared between tension- and compression-acting braces. The
slenderness ratio of the bracing members is limited and a
maximum width-to-thickness ratio is prescribed for the flat
elements of their cross section. Capacity design provisions
are also included to avoid overstressing, upon brace yielding,
of the brace connections and of the columns and beams
within the bracing bents.
The second bracing system category considered by the
code requires only nominal ductility and, therefore, is
assigned design seismic loads 50% higher than the ones for
the previous category. Braced frames of the third category,
referred to hereinis ordinary braced frames, rely only on the
inherent ductility of the steel material for resistance to earthquakes. No ductile detailing requirements, nor any capacity
design provisions, are specified for these frames. However,
they have to be designed for twice the loads prescribed for
ductile braced frames.
Chevron bracing is the concentrically braced frame configuration most often observed by the authors in the Northridge area. In this bracing scheme, two bracing members
form an inverted V shape at each storey, with the apex being
located at mid-span of the upper floor beam. his widely
used configuration exhibits high efficiency in the elastic
range (White and Salmon 1987) and allows more flexibility
for creating openings in the bracing bents. Once buckling of
both braces has occurred, however, the storey shear resistance and stiffness typically degrade rather rapidly (Khatib
et al. 1988). Therefore, during severe ground shaking,
chevron braced frames are likely to experience larger deformations. For multistorey structures, they are also more susceptible to concentration of large storey drifts within a single
storey (soft-storey mechanism) than other configurations. In
Canada, this system cannot classify under the ductile braced
frame category and must be designed for higher seismic loads.
Moment resisting frames

For low- and medium-rise structures, moment resisting frames
are generally less economical than braced frames for sustaining horizontal loads in the elastic range. They are more flexible and stiffness often governs the choice of the members.
In seismically active regions such as California, however,
moment resisting frames represent the essential lateral load
resisting system for multistorey structures. This system
exhibits high redundancy and high energy absorbing and
dissipating capabilities which overcome its less attractive
characteristics in the elastic range. Also, since no bracing
members are present, openings can be cut anywhere in the
structure.
In moment resisting frames, the beams are rigidly connected to the columns and lateral loads are resisted through
bending of these elements. Many types of beam-to-column
joints have been developed over the years, among which the
welded flange and bolted web type has become very popular
Can. J. Civ. Eng. Vol. 22, 1995
340
Table 1. Summary of observed structural damage.

Type*

Structure
Structural damage
Kaiser Permanente Hospital penthouse
CBF
Buckling of bracing members; excessive sway
First Interstate Bank Building at Northridge
CBF
Buckling of brace connecting plates; possible yielding of

anchor bolts
Student Union Building, California State University at

Northridge
CBF
No structural damage observed
Roof structure for the bleachers of the football field
CBF
Failure of anchor bolts (uplift)
Oviatt Library, California State University at Northridge
CBF
Failure of brace connecting plates; cracking of base

plates; yielding of anchor bolts
Three-storey building under construction in Van Nuys
CBF, MRF
Buckling of bracing members
NO. 2 Brewhouse, Anheuser-Busch Inc.
CBF
Buckling of bracing members
Asphalt and rock plant
CBF
Yielding and failure of anchor bolts
Department of Water and Power San Fernando Generating

Station
CBF
Four-storey commercial office structure
CBF
Buckling and failure of bracing members; failures of

brace welded connections; failure of a beam-column
moment connection
Two-storey fashion plaza
CBF, MRF
Cracking in floor slab; buckling of bracing members
Holy Cross Hospital administration building
Excessive sway; failure of anchor bolts
Van Nuys office building
MRF
MRF under construction
MRF
Failure of beam-column moment connections
*CBF: concentrically braced frame; MRF: moment resisting frame.
for earthquake resistance. This connection includes a bolted

web connection for carrying vertical shear whereas bending
moments are transmitted bycomplete penetration welding of
the beam flanges to the column (Fig. 43). For proper welding, weld backing or backing bars are used underneath the
beam flanges to support and retain the molten weld metal.
During ground shaking, energy can be dissipated through
any one, or any combination, of the following mechanisms:
plastic hinging in the beams, shear yielding in the panel zone
of the beam-to-column connections, or plastic hinging in the
columns. The latter is less desirable as it can lead more easily
to the formation of a single-storey collapse mechanism, and
because it is expected to result in larger storey drifts (Roeder
et al. 1989). Since allowing inelastic action in the panel zone
has been recognized only recently, most moment resisting
frames built in California have been designed according to
the weak beam strong column philosophy with flexural yielding limited to the beams. For such a mechanism, current
U.S. codes (AISC 1992; ICBO 1994) consider the welded
flange and bolted web beam-column joint adequate to develop
the flexural strength of the beam without tests or calculations, provided that the joint conforms to simple rules.
Similarly, in Canada, comprehensive seismic design provisions have also been introduced for moment resisting
frames in the S16.1 standard (Redwood et al. 1990). For the
most ductile system, stringent requirements for sizing the
critical elements are prescribed ti ensure a proper ductile
behavior of the three possible mechanisms (beam or column
hinging and panel zone shear yielding). Capacity design rules

are also specified in order to avoid brittle failures in the
remaining components. In particular, when beam hinging is
a critical element, welding of the beam flanges to the column
must develop the actual flexural strength of the beam.
Diaphragms and anchorage to foundations

For both concentrically braced frames and moment resisting
frames, the lateral load resisting system of the buildings
visited in Northridge also included floor and roof diaphragms
to collect and transfer horizontal loads to the vertical bracing
elements. Although it is implicitly assumed in design that no
brittle failure is to occur in the diaphragms, no explicit
capacity design provisions are included in current design
standards for these elements and their connections. These
provisions are also absent for the anchorage systems at the
base of the vertical bracing elements.
Observations after the Northridge
earthquake
A few days after the main shock of the Northridge earthquake, the authors visited the affected area for a period of
about a week. The third author returned for a second investigation three months after the event. During that time, the
authors stayed in contact with local engineers and authorities
to follow up on the latest developments in terms of new
observations, demolitions, retrofits, and repairs. The various
Tremblay et al.

Fig. 1. Kaiser Permanente Hospital: cladding loss on

penthouse.
Fig. 2. Kaiser Permanente Hospital: buckled double angle

chevron bracing in penthouse.
cases presented below represent up-to-date information at the

time of writing (May 1994) on the structures visited by the
authors. A summary of the observed structural damage is
presented in Table 1.
Kaiser Permanente Hospital penthouse, Panorama City

This existing hospital was a ten-storey reinforced concrete
building located only a few kilometres from the Northridge
epicenter. The facility suffered significant mechanical (HVAC)
losses due to severe damage to a steel framed penthouse.
This mechanical room was constructed from a light structural
steel frame with a cast-in-place concrete roof deck incorporating double angle chevron bracing. The penthouse was
severely racked in the north-south direction displaying near
complete cladding loss and extreme buckling (both in-plane
and out-of-plane) of the bracing system (Figs. 1 and 2).
Since the electrical, plumbing, and ventilation were rigidly
connected to both the roof soffit and walls, their associated
systems suffered either damage or service loss resulting in
the closure of the top three floors of the hospital until repairs
could resume their functionality (Fig. 3). It is estimated that
the hospital's capacity for patient care was reduced to
approximately the 60% level as a result of mechanical penthouse failures.
Fig. 3. Kaiser Permanente Hospital: racking of services

attached to penthouse roof.
First Interstate Bank Building at Northridge

This is a two-storey steel frame, 25 x 46 m in plan, located
on Nordhoff Street in Northridge, approximately 2 km north
of the epicenter. It was built in the mid-1970s and seismically
retrofitted in 1991. In the north-south direction, the frame
was concentrically braced along both end walls. As part of
the retrofit effort, four X-bracing bays were added at both
levels along the facade (two of them are shown in Fig. 4).
The performance of these four bracing assemblies is examined herein.
The X-bracing members were made from short legs backto-back L102 x 76 x 6.4 (4 x 3 x % in.) angles. At the
intersection of the braces, one brace was interrupted and continuity was provided by a connecting plate. All connections
were welded. The connecting plates showed evidence of
severe buckling and bending, whereas no indication of
inelastic action, nor buckling, could be observed along the
bracing members (Fig. 5). At the base of the columns of the

X-braced bays, evidence of uplift could be observed, as the
tiles on the ground and the stucco covering the columns were
damaged (Fig. 6). This suggests that neither the anchorage
nor the brace connecting plates could sustain the load that
developed in the braces.
Serious nonstructural damage, indicating that the structure had experienced significant deformations, could also be

Fig. 4. First Interstate Bank Building in Northridge: steel

X-bracing in the facade.
Fig. 6. First Interstate Bank Building in Northridge: column

uplift at the base of the X-bracing.
Fig. 5. First Interstate Bank Building in Northridge: buckling

of connecting plate at brace intersection.
Fig. 7. Student Union Building at the California State

University at Northridge: exterior view of building bracing
system.
observed. According to recent discussions with the owner's

representative, the extent of structural and nonstructural
damage was such that the building has to be demolished and
rebuilt. It is the owner's intention to replace it with a singlestorey timber construction.
Student Union Building, California State University at

Northridge
Of course, not all the steel structures visited had suffered
damage during the Northridge earthquake. For instance, on
the Northridge campus of the California State University,
which was located within 3 km from the epicenter of the
main shock, a four-storey concentrically braced structure
was under construction when hit by the earthquake (Fig. 7).
The structure was well braced and well detailed in the connection and panel zone regions of beam and column interactions (Fig. 8). No structural damage was observed, although
it is recognized that its intended full design dead loading was
not in place during the earthquake.
Roof structure for the bleachers of the football field,

California State University at Northridge
West of Zelzah Avenue, near Fullerfarm Street, stands the
football field of the California State University at Northridge, approximately 5 km north of the epicenter. A steel
structure built around 1970 sheltered the bleachers located
along the west side of the field. The roof structure included
corrugated steel panels bearing on purlins spanning between
18 main cantilevered steel trusses, approximately 10 m long,
that were spaced at 5.2 m on center along the length of the
field. The clear height below the trusses was 3.6 m.
Each cantilever truss was supported by a single W200 x36
(W8 x24) column located 7.4 m from the front edge of the
Tremblay et al

Fig. 8. Student Union Building at the California State

University at Northridge: interior view of building bracing
and details.
Fig. 10. Roof structure covering the bleachers of the football

field at the California State University at Northridge: uplift
of the back ties.
Fig. 11. Oviatt Library Building at the California State

University at Northridge: failure of the overhanging roof.
Fig. 9. Roof structure covering the bleachers of the football

field at the California State University at Northridge:
cantilevered steel frames.
roof and was tied down to the foundations at the back by

means of short legs back-to-back L63 x 63 x 6.4 mm (2 % X
2 % X 5/4 in.) angle ties (Fig. 9). Anchorage of the ties to the
foundation was achieved by a 6.4 mm thick base plate with
two 19 mm anchor bolts spaced 90 mm on center. Bracing
in the direction parallel to the field (north - south) was provided by three two-bay inverted V tension-only systems made
of 19 mm diagonal steel rods inserted between the vertical
back ties of the roof trusses.
The first four bays at the south end of the roof structure
exhibited approximately a 5" inclination towards the bleachers
(Fig. 9) as a result of the failure of the anchor bolts of the
back ties during the earthquake (Fig. 10). Complete collapse

of these four bays was inhibited only by the rotational
restraint offered by the column base plate. The fact that
anchor bolt failures occurred at the base of ties where longitudinal bracing diagonal members were attached as well as
at the base of ties with no such diagonal members indicates
that vertical accelerations undergone by the cantilevered roof
likely contributed to the observed damage. According to the
owner's representative, the structure has been dismantled
after the earthquake and was to be replaced with a new one.
Oviatt Library, California State University at Northridge

This is a four-storey building including a concrete framed
central part built in 1971. Two wings made of steel braced
frames were added in 1991 at the east and west ends. The
building had a steel framed overhanging roof along its
perimeter. At two locations along the north wall, the steel
beams of that projected roof running parallel to the exterior
wall, and spanning across an expansion joint in the main
framework, collapsed (Fig. 11). This failure has been attrib-


Fig. 12. Three-storey building under construction in Van Nuys:
Fig. 13. Three-storey building under construction in
chevron bracing on the north wall.
Van Nuys: out-of-plane buckling of the first-storey bracing

members on the north wall.
uted to the loss of bearing support of these beams and pounding damage due to the relative horizontal movement that took
place at the expansion joints. No other structural damage
could be noticed at the time of the visit.
According to the inspecting engineer, preliminary damage
assessment of the structure indicated the steel braced frames
had resisted very well the shaking with essentially no visible
damage. Later on, upon removal of the interior finishing, a
thorough inspection of the framework revealed a brittle
failure of the welded connection between brace gusset plates
and base plates as well as fracture of the base plates themselves at the bottom of approximately 75% of the columns
located within the bracing bents. Many of the 63 mm anchor
bolts used for these columns also suffered inelastic elongation up to 12 mm. In one case, the bolt failed in tension.
These failures have been reported elsewhere (EERI 1994).
So far, the investigation demonstrated that the capacity of the
welds was well below the actual strength of the bracing members and, thereby, below the forces that likely developed in
these members during the shaking. It is believed that such
failures "protected" the braced frames from undergoing
inelastic action during the shaking but, on the other hand,
likely resulted in larger horizontal deformations which could
have caused the collapse of the overhanging roof.
Fig. 14. Three-storey building under construction in
Van Nuys: undamaged beams at the second level on the

north wall.
Three-storey building under construction in Van Nuys

On Sepuvelda Boulevard near Victory Boulevard in Van Nuys,
8 km southeast of the epicenter, a three-storey building was
under construction when the earthquake struck. Only the first
floor slab was poured at that time. In the north-south direction, the frame was three bays wide with the center bay being
a moment resisting frame. Single-bay inverted V chevron
bracings were provided along the north and south exterior
walls for resisting lateral loads in the other principal direction (Fig. 12). The bracing bays were 10 m wide and the
storey height was approximately equal to 3.6 m.
The structure did not suffer any significant damage,
although the ground shaking was particularly strong in that
area. However, all bracing members at the first floor experienced significant inelastic out-of-plane buckling (Fig. 13).
As shown, both braces were permanently deformed in the
buckled shape, which indicates that both braces likely

yielded in tension during the ground shaking. These members were made from back-to-back channels, 152 mm in
depth and 76 mm in width, assembled by means of 10 mm
spacers at quarter span. Though an unbalanced vertical force
likely developed at the apex of the V at mid-span of the
beams, no signs of plastic deformation could be observed
along the beams (Fig. 14). The moment resisting frames in
the north -south direction did not suffer any visible damage.
No. 2 Brewhouse, Anheuser-Busch Inc.

The Anheuser-Busch brewing facility is located on Roscoe
Avenue in the Panorama City district, southeast of the North-
Tremblay et al.

Fig. 15. No. 2 Brewhouse, Anheuser-Busch, Inc.: cladding

failure on tower structure.
Fig. 17. No. 2 Brewhouse, Anheuser-Busch, Inc.: close-up

view of cladding failure on warehouse revealing steel braced
frame.
Fig. 18. No. 2 Brewhouse, Anheuser-Busch, Inc.: buckling

of double angle chevron bracing in the tower portion.
Fig. 16. No. 2 Brewhouse, Anheuser-Busch, Inc.: global

view of cladding failure on warehouse.
response of most of the plant structures, considering its close

proximity to the epicenter. The mechanical processing equipment housed within some of the buildings suffered significant
damage, resulting in down time for their repair, realignment,
or replacement.
ridge epicenter. Significant shaking at the plant was experienced by personnel during the main shock and aftershocks.
The No. 2 Brewhouse is a four-storey concentrically braced
structure incorporating wide flange shapes as bracing members. An additional five-storey mechanical tower structure
was located on the east end of the brewhouse roof. Significant cladding loss and damage occurred on the tower section
of the brewhouse, and other warehouse structures located at
the plant experienced similar problems (Figs. 15- 17). In
addition to the cladding failure, several double angle chevron
bracing in the tower portion had buckled (Fig. 18).
It was learned that the plant had been undergoing a seismic upgrade program for several years. This retrofit program coupled with the highly redundant structural lateral
load resisting system are responsible for the generally good
Asphalt and rock plant

An asphalt and rock planted located some 15 km from the
epicenter in the Sun Valley district was visited. The plant
contained numerous gallery and truss structures, some dating
back to the 1900s, with the latest major expansion of operations completed in 1964 (Fig. 19). Many of the steel frames
have undergone something of a cannibalization process during the course of years of regular maintenance and modification programs. Portions of the plant were originally designed
for approximately 0. l g lateral forces. The generally undamaged 1964 expansion was designed for 0.18g lateral loads
coupled with meticulous structural design which determined
the period and stiffness for each independent braced framed
structure. Because it is a gravel quarry and an active, on-site
production facility, the plant is founded on very deep, stiff,

Fig. 19. Asphalt and rock plant: typical conveyor and truss
system.
Fig. 20. Department of Water and Power San Fernando

Generating Station: boiler structures.
alluvium deposits with boulder inclusions.

There were no gallery, tower, or conveyor collapses at
this site even though the shaking was reported to have rattled
the structures intensely. Some evidence of elastic deformation in the bracing as well as minor foundation cracking at
column braces were also reported. Numerous anchor bolt
elongations of the order of 3 mm had been discovered and
one 19 mm diameter anchor displayed a tensile brittle fracture. Remains of this bolt were discovered some 60 m away.
Overall the plant sustained only nominal damage. After rectification, it was operational within four hours of power
resumption.
Department of Water and Power San Fernando

Generating Station
A 350 MW generating station, located approximately 15 krn
from the epicenter, again in the Sun Valley district, was also
visited. It had been reported to have experienced intense
shaking. This facility came on grid as the emergency power
source following disruption to the electrical service caused
by the main shock of the earthquake. The plant's boilers are
fuelled by natural gas and are each supported by separate sixstorey concentrically braced framed structures (Fig. 20).
Plant personnel reported essentially no structural damage. It
is interesting to note that the 70 m high stacks constructed of
redwood cedar with stucco finish and brick lining also
remained free of damage. A 24 m diameter by 15 m high
domestic water tank, which was nearly full at the time of the
main shock, exhibited nominal foundation movement as evidenced by the circumferential cracking in the soil (Fig. 21).
This movement resulted in a minor leak in a 50 mm diameter
discharge line.
Four-storey commercial office structure
An approximately 20 000 m2 four-storey office complex
(Fig. 22) suffered extreme bracing related fractures and
deformations. The building is located at the intersection of
Magnolia and Lankershim boulevards in North Hollywood,
some 16 km east of the epicenter. The concentrically braced
building was constructed in 1986 according to the 1980
Los Angeles Building Code, which is essentially the same as
the 1982 UBC code (ICBO 1982). The structure is believed

to be founded on deep stiff soils and there were no reported
geotechnical related failures in the area. The damage witnessed in the building is attributed only to the strong motion
shaking and not to any induced foundation settlement or

Tremblay et al.
Fig. 22. Four-storey commercial office structure: global
Fig. 24. Four-storey commercial office structure: local
view.
failure of hollow structural brace at center.
failures. Most of the severe damage was limited to the

north-south oriented braced bays at the second floor. This
potentially implies that the demand-to-capacity ratio of the
braces was somewhat higher at the second floor (Bonneville
1994) and that the existing chevron braced system was not
capable of redistributing forces to other levels after the
second-storey failures occurred. Significant ductility problems were exhibited in the HSS 305 x 305 x 9.5 mm (12 x
12 x % in.) tubular braces. Many of the failures found were
attributed to local plate instability of the section because of
the high width-to-thickness ratios of the walls of the steel
tubes. Local buckling was isolated to either the end connection or the center of the brace (Figs. 23 and 24). Ruptures
in the braces were caused by high stress, low cycle fatigue
resulting in the formation of a brittle crack at crimped wall
failures from local instability. One brace had ruptured completely through its cross section (Fig. 25).
One braced frame failed at its slotted welded connection
to the gusset (Fig. 26). At another location within the structure, where the bracing was discontinuous or offset and a
special moment connection was required, the resulting
deformations initiated the fracture of a full penetration weld

connecting the girder to column flange (Fig. 27). Intense outof-plane buckling of the braces initiated precast facia panel
damage. Even for this extreme loading condition, however,
the panels performed well and none became detached from
the structure. There was also some permanent deformation in
girders from in-plane bending due to frame behavior and torsion due to brace buckling during the earthquake.
The building remained plumb following the earthquake,
although there was strong evidence of a second-storey drift
of at least 5 mm. In retrospect, of the significant damage
found at this building, the initial assessment of the structure
by the owner's representative prior to the review engineer
arriving on site was that the structure had not sustained much
damage (only one window had been broken). Only after the
preliminary assessed cosmetic damage to the dry wall was
removed, at the order of the engineer, was the extent of
damage revealed and proper assessment of the structure could
begin.
Even though the failures did not jeopardize the vertical
load carrying capacity of the building, the extent of damage

Fig. 26. Four-storey commercial office structure: failure at

slotted gusset connection.
Fig. 28. Four-storey commercial office structure: newly

installed wide flange braces and gussets.
Fig. 27. Four-storey commercial office structure: fracture of

full penetration weld at girder flange.
Fig. 29. Four-storey commercial office structure: close-up

view of new gussets at brace end for corner columns.
to the lateral load resisting system fully justified the imposition of a "limited entry" restriction, not allowing continued
occupancy. Upon revisiting the building three months after
the main shock, the structure had been stabilized using temporary bracing and repair was under progress. The retrofit
scheme that was adopted features interesting aspects worth
discussing in the following.
A review of the structure by the inspecting consultant
indicated that the structure had conformed to the code intents
for its period of construction. In the 1980 Los Angeles code,
however, neither brace connection strength requirements nor
additional strength criteria were recognized. Typically, concentrically braced frames built at that time would classify
under the aforementioned ordinary braced frame category
for which little or no attention was paid to ductile detailing
or capacity design concepts. Therefore, repair solutions had
to recognize this potential for less ductile failures, and avoid
local overstrengthening without carefully considering the
severe impact this may have on the rest of the structure.
For example, introducing overly strong new braces into
this structure would have possibly ensured elastic brace
behavior, but risked buckling of the columns in the bracing
bents during future earthquakes. Such seismic structural

behavior would have been far less desirable than the actual
performance during the Northridge earthquake. Thus, all
braces throughout this building were replaced by new
350 mm deep wide flange members (Fig. 28). These elements have a global buckling strength close to that of the
original members but meet all the seismic design ductile
detailing requirements specified for special concentric braced
frames in the U.S. (e.g., AISC 1992). These requirements
are essentially identical to those for ductile braced frames in
the S16.1 Canadian standard. Attention has also been devoted
to brace connections. The sizeable gusset plates, typically
designed to be stronger than the brace members in special
concentric braced frames, are noteworthy (Figs. 28 and 29).
In order to better distribute the ductile behavior to all
stories of the building, vertical members were added to tie
all chevron braces together over the height of this building
(Fig. 30). As such, to the authors' knowledge, this represents
the first implementation of the "zipper" system developed a
few years ago by Khatib et al. (1988) to remedy the deficiency germane to chevron braced frames that have a tendency to concentrate seismically induced inelastic response
Tremblay et al.

Fig. 30. Four-storey commercial office structure: newly

installed braces in repaired inverted chevron-type
braced-frame with vertical tie-members added to implement
"zipper" system.
only in one or a few stories. During the retrofit analysis, the

effect of moment connections at the beam - column -brace
intersections was also investigated. It was discovered that
significant improvement to post-buckling performance of the
frames could be obtained through the transformation of
simple connections into moment joints in the repair areas.
Therefore this approach was also implemented in the retrofit.
Two-storey fashion plaza

A commercial redevelopment project incorporating the interesting features of an existing steel framed theater complex
was constructed circa 1970 to become a fashionable plaza.
This two-storey steel framed building, including underground
parking, is located on Ventura Boulevard in the Sherman
Oaks district. From a review of available drawings with the
owner's representative, it became apparent that the structural
system erected differed substantially from the reinforced
concrete and wood frame structure originally planned. The
lateral load resisting system consisted of several bracing systems in each orthogonal direction. The eastern section incorporated both chevron hollow structural section bracing and
wide flange concentric bracing in each bay orientation. For
the western section, the retail building contained moment
resisting frames in each structural direction.
Damage observed at this structure included longitudinal
cracking in the concrete slab cast on light gauge steel decking
likely related to diaphragm action. This was particularly evident in the rear parking area at street level along the beam
lines (Fig. 31), but was also witnessed at the second floor
level of the plaza. Tension and shear failures of two corbels
supporting beams were also discovered in the lower parking
level (Fig. 32). Nominal overall member buckling had
occurred in an exposed wide flange brace at the rear of the
structure (Fig. 33).
Very little nonstructural damage and chaos occurred in
the merchant area braced with the tubular chevron bracing
albeit these areas were void of suspended ceiling and other
architectural finishes, whereas severe nonstructural damage
occured in the moment resisting framed areas of the structure
Fig. 31. Two-storey fashion plaza: cracking along beam lines

in parking deck.
Fig. 32. Two-storey fashion plaza: beam anchorage failure at

corbel.
although the frames did not exhibit any damage themselves

from initial observations (Fig. 34). At the former theater section, a steel framed sign structure suffered bolt shear failures
at the column splice locations at the second floor, necessitating its subsequent removal for hazards mitigations (Fig. 35).
Holy Cross Hospital administration building

This facility is a three-storey reinforced masonry and steel

Fig. 33. Two-storey fashion plaza: slight overall member

buckling of wide flange brace.
Fig. 35. Two-storey fashion plaza: sign frame anchorage

shear failure in the former theater portion.
Fig. 34. Two-storey fashion plaza: nonstructural damage in

moment framed areas near expansion joints.
Fig. 36. Holy Cross Hospital administration building:

elevator and suspended walkway in background.
framed building located in the Mission Hills district. The

street bracing system was not identified because of building
finish cover but possibly contained moment resisting frames
in some areas. Entry to the structure was restricted to only
record retrieval following initial assessment and subsequent
decision for its future demolition. A steel framed (likely
diagonally braced) elevator shaft and suspended third floor
walkway (behind stair tower and walkway in background of
Fig. 36) had separated from the main building by some
200-250 mm due to column anchor bolt failures in a reinforced masonry wall (Fig. 37).
wythe of solid masonry brick with nominal reinforcing in the

cavity between the wythes. This end wall was very similar
to the one in the Kaiser Medical Building which suffered a
complete second storey collapse on Balboa Avenue (Mitchell
et al. 1995).
The separation of both the end wall and stair tower is
attributed to dowel shear failure between the floor slabs and
the end facade walls. The end wall's collapse mechanism is
surmised to have been initiated by shear failure at the base
followed by bearing loss, allowing the remaining wall to vertically shear and slide down the face of the structure. Concrete spalling was apparent at the top flange-slab locations as
well as in the beam-column joint regions. Physical separation
of the aluminum and glass south curtain wall was evident at
the second floor level.
Van Nuys office building

A five-storey office structure located in the Van Nuys district
suffered significant damage. Concrete-encased steel moment
resisting frames for the north-south and east-west orientations of the building layout provided the lateral load resisting
system. The building was essentially plumb following the
earthquake; however, facade failures were apparent, most
obvious at the east and west end walls of the structure
(Fig. 38). The east end wall was constructed of a double
Moment resisting frame under construction

The steel building shown in Fig. 39 is located in Santa Monica
and was under construction at the time of the main shock. It
had apparently survived the earthquake intact. However,
after reading a news story about steel failures discovered
Tremblay et al.

Fig. 37. Holy Cross Hospital administration building:

elevator shaft column anchorage failure.
351
Fig. 39. Moment resisting frame under construction: north

facade along which beam-column connections suffered
cracking and fractures.
Fig. 40. Moment resisting frame under construction: typical

beam-column connection damaged by column cracking and
fracture at the level of the beam's lower flange weld.
Fig. 38. Van Nuys office building: end wall fa~ilureon east
face of building.
elsewhere, the contractor decided to randomly inspect a few

beam-column connections by removing the fireproofing
installed prior to the earthquake. Nearly all joints of the
moment resisting frames located on the north side of the
building were found to have severe fractures.
Examples of the fractures in this building are shown in
Figs. 40-42. Typically, at a damaged connection, the
column fractured at the level of the full-penetration weld of
the beam's lower flange, and cracking propagated into the

column's web a short distance horizontally, then mostly verticallv and toward the other flange weld of the same beam.
he web of the column of Fig. 42 also fractured although
this is not visible on the photo. Clearly, since the building
was under construction at the time of the earthquake, this
damage cannot be attributed to causes other than the Northridge earthquake.
Although a single case study of beam-column connections
having suffered cracking or fractures is presented in this
paper, this type of brittle failure occurred frequently throughout the Los Angeles area. In May 1994, similar damage had

Fig. 41. Moment resisting frame under construction: close-up

view of the fractured column shown in Fig. 40.
Fig. 42. Moment resisting frame under construction: close-up

view of a beam-column connection damaged by column
flange fracture at the level of the column web stiffener (left)
and beam's lower flange (right).
been discovered in more than 60 buildings further to thorough postearthquake structural inspections (M. Engelhardt,
1994, private communication). As of November 1994, it was
believed that possibly hundreds of steel moment resisting
frames have suffered brittle joint failures in the Los Angeles
area (Ross and Mahin 1994).
Discussion
Steel ranks very high among structural materials suitable for
earthquake resistance. It exhibits high strength and stiffness
as well as good ductility and toughness. Its strength-toweight ratio is also remarkably high. When compared to that
of other common materials, the behavior of steel is rather
well defined and understood and a high quality can be achieved
by the in-shop prefabrication of all components of steel frames.
This makes the seismic performance of steel structures more
predictable than that of other construction systems. Moreover, after an extreme event, damaged components can be
easily repaired or replaced.
However, building with steel is not sufficient by itself to
warrant a proper performance during a strong earthquakeinduced ground shaking. The characteristics of future ground
motions are highly uncertain and the prescribed design seismic loads are considerably reduced by reliance on the ductility of material, which means that designing structures to code
does not preclude the possibility of damage to occur. Satisfactory behavior can only be achieved if a sound structural
arrangement is provided and if the structural elements and
their connections are sized in such a manner that appropriate
means of absorbing and dissipating energy exist and premature failures are avoided, especially within the gravity
load resisting system.
Except in case of collapse, earthquake-induced damage to
steel frames, even serious and potentially catastrophic, is
generally less apparent and thus more difficult to detect and
assess than in other types of construction. The framework is
most often hidden and because damage is usually constrained
to a very limited portion of the structure (connections, local
buckling, anchor bolts, etc.), thorough inspection is needed
to obtain an overall and realistic assessment of the damage.
For building structures, detailed inspection requires the

removal of architectural finish, fireproofing covering, and
even portions of concrete slabs, as noticed in some structures
visited in Northridge. This makes the process expensive and
lengthy.
Data obtained in the early stages after a seismic event are
then fragmentary. As owners proceed with examination of
their structures, additional information on inadequacies or
failures is continually reported. Unfortunately, not all cases
are made available to the engineering community, as some
information is kept confidential. (In fact, it is believed that
even the perception that steel constructions behave well in
earthquakes can be somewhat biased by this situation!) The
Northridge earthquake does not seem to be the exception and
it appears that the full extent of damage suffered by steel
structures will not be known before months and even years
have passed. Nevertheless, several lessons can already be
drawn from the observations reported so far. These are next
summarized for the concentrically braced frame and moment
resisting frame systems.
Inelastic response of bracing members in concentrically

braced frames
The numerous occurrences of severe brace overall buckling
and tension yielding which were observed during the reconnaissance visits stress the importance of properly detailing
bracing members in concentrically braced frames for which
a ductile behavior is required for earthquake resistance. In
particular, the need for preventing local buckling and subsequent premature failure of the braces at the plastic hinges has
proven to be mandatory. This problem is more acute for rectangular tubular bracing members and had already been rec-

Tremblay et al
ognized for some time (e.g., Liu and Goel 1988).

Observations after the Northridge earthquake also indicated
that engineers should be aware that inelastic buckling of
braces may create additional problems. For example, out-ofplant buckling of braces represents a serious damage potential for cladding, partitions, or mechanical equipment in
buildings. Even though such damage may not impair the
integrity of the structure, it may be highly hazardous for its
occupants (e.g., downfall of glass debris, masonry blocks, or
precast panels) and thereby should be considered in full as
an ultimate limit state in earthquake-resistant design (Knoll
1993). Breakage of mechanical equipment or collapse of
walls may also have serious impacts on the postearthquake
functionality of a building, which can be a serious matter of
concern for building owners or users, or business survival.
This problem is somewhat accentuated by the fact that
most designers favor out-of-plant, rather than in-plane, buckling for bracing members. Indeed, a plastic hinge can easily
be accommodated within the gusset plates in the out-of-plane
buckling mode when following the design recommendations
proposed by Astaneh et al. (1985). Moreover, out-of-plane
buckling can be easily achieved for any brace section. Hollow
structural section square bracing with standard connections
and W shapes with the web in the vertical plane tend to naturally buckle out-of-plane, whereas out-of-plane buckling of
double angle bracing can be readily obtained by properly sizing the section.
Damage can, however, be easily prevented by providing
sufficient free space along each side of the braces or, alternatively, by forcing the bracing members to buckle in-plane.
Bending moments that develop at the ends of bracing
members upon buckling can also create some problems by
overstressing the connections and the surrounding structural
elements. In the Northridge earthquake, the case of beams
being permanently distorted in torsion as a result of the outof-plane buckling of bracing members is an example of such
undesirable behavior. These secondary bending moments
must be resisted by adjacent connections and members, in
addition to the primarily axial loads likely to build up in the
braces, and must therefore be accounted for within an
appropriate capacity design procedure, as described in the
next section.
Alternative ways of dissipating energy in braced frames
have recently been proposed (Cheng et al. 1993; Tremblay
and Stiemer 1993) and could eventually represent promising
solutions for overcoming some of the shortcomings associated with inelastic buckling of bracing members.
Capacity design of concentrically braced frames

Numerous failures and evidences of inelastic response were
observed by the authors in elements other than the bracing
members along the lateral load path of concentrically braced
frames. These occurred as cracks in floor diaphragms,
failures of brace connections, torsional deformations of
beams, and tensile fracture of anchor bolts or base plates.
Many of the failures observed were brittle in nature.
Similar behavior has also been reported for many other
braced structures not visited by the Canadian reconnaissance
team. For instance, a riveted steel structure, part of a batch
plant constructed in the early 1950s and located within 68 km from the epicenter, exhibited various kinds of tension
failures in its braces (G. Hichborn, 1994, private communication). Failure modes included complete tensile fracture of
two braces at their connections, serious tearing and deformation in most angle clips, partial plate pull through of rivets,
and permanent elongation deformations of remaining braces.
Another nearby structure, also several decades old, was
constructed as stacked framing modules. Each module contained four columns with connection flanges at their ends and
incorporated seismic cross or knee bracing. This structure
experienced column flange bolt failure in single shear. The
four bolts failed completely in all of the first level column
splices. Shearing between the first and second level modules
was accompanied by severe horizontal translation. Luckily,
this movement was stopped by a heavy walled water supply
pipe running up one of the columns. Further motion would
have resulted in catastrophic failure of the fully loaded structure. In addition, many of this plant's foundation anchors
failed in combined tension and shear, while others deformed
significantly by elongation.
This broad range of failure types illustrates very well how
arbitrary and uncontrolled the seismic performance of concentrically braced frames can be when a comprehensive
capacity design has not been applied thoroughly. During a
strong ground shaking, the most overstressed element in the
as-built lateral load resisting system reaches invariably its
capacity first. Because braced frames inherently exhibit very
low redundancy, this overstressed element would have to
undergo substantial inelastic deformation prior to subsequent
redistribution, if any, of forces to other elements. Failure of
the element may then occur early in the earthquake if it does
not possess sufficient toughness to absorb the energy fed into
it by the ground motion.
Such behavior would be typical of ordinary braced frames.
Although higher seismic loads are prescribed for that system,
they still represent only a fraction of the loads that would be
expected in a structure responding elastically to the design
base earthquake (40% when comparing the elastic and design
1990 NBCC base shears). Thus, some degree of inelastic
response is still anticipated in ordinary braced frames and
premature failure is probable if the weakest element does not
exhibit enough ductility. This could be the case, for instance,
of concrete roof and floor diaphragms or welded connections. Anchor bolts are also prone to early failure as the
demand on these elements can easily exceed their capacity to
absorb energy.
Many of the braced frames damaged during the Northridge earthquake were rather old and most likely belonged to
the ordinary braced frame category. Despite the higher seismic design loads prescribed for these braced frames, they
still remain popular today in seismically active regions
because they require less stringent seismic detailing provisions. Braced frames with some degree of ductile detailing
were also found to have behaved in a similar uncontrolled
manner in Northridge; that is, exhibiting inelastic action or
failures away from the braces. This was the case, for example, for the Oviatt Library Building, built in 1991, or the
First Interstate Bank Building retrofitted the same year. In
these structures, most likely the capacity check on the
damaged elements had not been performed or had not
included all the effects that actually occurred during the
earthquake. In Canada, this undesirable behavior due to lack

of capacity check in the original design may occur in any

ductile braced frame, since, as mentioned earlier, the design
provisions included in the S16.1 standard only apply to the
steel framing members and not to other elements of the
lateral load resisting system. The secondary bending moments
developed at the ends of the bracing members upon buckling
is an example of effects that may occur during an earthquake.
Because of the low redundancy of concentrically braced
frames, any failure along the lateral load path can severely
degrade the stiffness and the strength of the lateral load
resisting system. The structure can then experience large
horizontal deformations and, ultimately, collapse. As mentioned, all damaged concentrically braced frames inspected
survived the Northridge earthquake and its aftershocks.
However, the potentially adverse effects that the damage
observed could have had on the stability of the structures,
had the earthquake lasted longer, strongly suggest that any
braced frame should be designed in such a manner that
inelastic response be constrained to the ductile elements,
namely the bracing members. This means that, desirably, a
capacity design philosophy should be implemented for (i) all
braced frame categories and (ii) the entire lateral load resisting system, including diaphragms, anchorage to the foundations, and the foundations themselves.
Even though anchorage and diaphragm failure have not
been observed in moment resisting frames, the same approach
should also be applied to this system for improved seismic
~erformance.
Implementing a comprehensive capacity design approach
to column base connections and foundations of concentrically braced frames and moment resisting frames still remains,
however, a matter of debate. Some recent investigations (e.g.,
Filiatrault et al. 1992) have suggested that intentionally
allowing rocking of the foundations or inelastic elongation of
anchor bolts at the base of vertical bracing bents may not
adversely affect the ability of a structure to sustain earthquake ground shaking. Such behavior may, on the other
hand, lead to an unacceptable level of horizontal deformations and should be thoroughly analyzed before construction.
Vertical ground motions in the design of concentrically

braced frames
Significant levels of vertical ground motions have been
recorded during the Northridge earthquake (Finn and Ventura
1994). Such vertical accelerations are known to be of significance for horizontal cantilevered structures (BSSC 1991)
such as the roof covering the bleachers of the football field
at the California State University at Northridge. The nurnerous anchorage failures observed at the base of the exterior
columns of bracing bents in concentrically braced frames
suggest that these structures could have been also affected by
vertical accelerations.
Substantial compressive and tensile axial loads develop in
the exterior columns of bracing bents of concentrically
braced frames to resist the overturning moment produced
by horizontal earthquake ground motion. Indeed, it is not
uncommon to have the combination of the gravity loads plus
these overturning induced forces governing the sizing of
these columns. Obviously, anchorage at the base of bracing
bents is mainly provided for resisting that overturning moment.
Therefore, it is very much likely that these columns and their
anchorage become overstressed under the combined action of

the horizontal and vertical ground motions, a situation that
probably occurred in Northridge.
Such a combined effect is not explicitly addressed in current Canadian codes. Neither is the effect of vertical accelerations on horizontal cantilevers. Thus, it is suggested that the
hazard potential of these two effects be assessed for structures located in Canada and, if deemed of significance, that
appropriate design provisions subsequently be implemented
in codes.
Nonstructural damage
Many of the structures visited experienced large interstorey
drifts and suffered extensive nonstructural damage. Moment
resisting frames, being more flexible than braced frames,
appeared to be involved in the more critical situations. In
some cases, the extent of damage was such that many injuries
and even fatalities could have occurred had the buildings
been occupied during the earthquake. Large portions of
exterior walls and interior partitions collapsed, glass broke,
suspended ceilings and mechanical equipment toppled, etc.
The objective of the NBCC earthquake-resistant design
provisions is to prevent not only building collapses but also
loss of life. Therefore, as mentioned earlier, hazard potential
associated with nonstructural damage should be explicitly
addressed at the design stage.
Controlling horizontal drift certainly is a possible avenue
in preventing damage in a structure. This can be achieved by
increasing the stiffness of the lateral load resisting system.
For short period structures, a more appropriate approach
would be to increase the strength to limit the drift to a target
value. This can be done by using inelastic displacement
spectra, as proposed by Priestley (1993).
Nonstmctural damage can also be limited by using better
construction techniques such as stronger wall attachments o r
improved suspended ceiling structures. Alternatively, proper
structural and architectural detailing can be provided, which
can allow the anticipated deformation to take place without
failure or collapse due to pounding, tearing, or loss of support. Without doubt, the existence of a good collaboration
and communication between the architect and the engineer is
also part of the solution to such problems.
Regardless of the approach used, engineers must humbly
recognize that a great deal of uncertainty is associated with
predicting drift in seismic design. Such incertitude should be
reflected in design and details by allowing a sufficient margin
of safety.
Connection failures in moment resisting frames
The brittle fracture of field welded beam-to-column connections in steel moment resisting frames is one of the most significant issue of the Northridge earthquake. The following is
a summary of the information available as of November 1994
on this consequential structural damage.
The first connection failures were reported only weeks
after the earthquake, mainly in buildings under construction,
where finishing and fireproofing materials had not been
installed yet, or in buildings exhibiting significant permanent
interstorey drifts which indicated that some structural damage
had likely taken place. In many cases, however, no apparent
sway nor architectural damage could suggest that failure had

Tremblay et al.
occurred, and the structure was declared safe for occupancy

after the initial inspection. Failures in connections were
found only after structural engineers, recognizing the significance of the problem, required random inspection of
joints in various steel structures, leading to the discovery of
more failures and prompting other engineers and owners to
act similarly. In other cases, buildings were reinspectd after
owners had noticed that damage caused by aftershocks was
more important than that produced by the main event.
Failures of beam-column connections were then detected
which had likelv diminished the stiffness of the structures.
Most of thed structures that have suffered fractures at
beam-column joints had been constructed post-1980 and had
been designed as special moment resisting frames, which is
the equivalent American designation for- the S 16.1 ductile
moment resisting frame category. It is worth mentioning that
frequent design practice is to incorporate these special
moment resisting frames for only portions of the available
building dimension, usually centrally located on the exterior
perimeters. In such an arrangement, one can take advantage
of the reduced res scribed seismic loads for ductile moment
resisting frames while minimizing the costs associated with
the construction of moment resisting frames. However, larger
members and connections are required in the bracing bays
when compared to structures with more evenly distributed
moment connected frames.
Occurrence of damage in Northridge does not appear to
have been limited to a given range of building height. A
survey presented by Ross and Mahin (1994) indicates that
75 buildings for which a permit has been issued for repair in
the city of Los Angeles vary from 1 to 22 stories in height.
Ghosh (1994) reports that buildings from 1 to 27 stories have
been damaged, the majority being less than six stories high.
Sabol (1994) indicates that connection failures in taller structures generally occurred in the upper half to two thirds of the
building whereas lower frames were more evenly damaged.
The observed failure rate in a building varied from less than
10% to nearly 100% of the connections. It was also noticed
that failures occurred in both types of construction: buildings
with nearly all frames moment connected and structures
including only a few moment resisting frames. However, it
appears that more buildings of the second type experienced
structural damage (Bertero et al. 1994).
All fractured joints were of the welded flange and bolted
web type, as shown in Fig. 43. Some had their vertical shear
connectors partially welded to the beam web. Each discovered failure had full-penetration weld fractures in the bottom
flange, very few had these fractures in the top flange. In all
cases, there was little or no evidence that plastic hinging had
developed in the beams prior to weld fracture. Some damaged
connections also exhibited vertical cracking through the
shear connector or shearing of the web bolts. It is believed
that this secondary mode of failure developed as the beam
rotated relative to the face of the column after fracture of the
beam flange weld had occurred.
Figure 44 shows some of the failure modes that have been
observed at the bottom flange of the beams. Reports indicate
that most cracking initiated in the root pass of the weld, near
the steel back-up bar. Type 1 and 2 cracks were very frequent. Failures of type 2 could be readily identified in the
field. Type 1 failures can also be detected through visual
Fig. 43. Typical welded flange and bolted web beam-column

connection in moment resisting frames.
CONTINUITY
.flm
NOTCH
CONDITION
inspection as the backing bar separated from the column face

(Miller 1994). Type 3 and 4 failures are similar to types 1
and 2, but crack propagation took place into the column
flange rather than at the interface between the weld and the
column flange material. Type 3 cracks can be detected only
by means of ultrasonic inspection, as these cracks do not exit
the column flange surface.
In other connections, cracks developed across the column
flange (type 5) and, in some cases, propagated further into
the column web (type 6) (see also Figs. 40-42). Miller
(1994) reports that some columns completely fractured horizontally. A few connections also exhibited a crack across the
beam flange (type 7) and, according to Bertero et al. (1994),
one case of column lamellar tearing (type 8) has been reported.
A review of the literature on the experimental research
performed in the U.S. and Japan on moment connections in
the past 25 years (Bertero et al. 1994) reveals that the seismic
performance of welded joint specimens varied from very good
to very poor. Indeed, many test programs (e.g., Bruneau and
Mahin 1990; Engelhardt and Hussain 1993) had demonstrated that fracture and brittle failures similar to those
observed after the Northridge earthquake could occur in
heavily welded steel beam-column connections, even when
these were constructed in-shop by competent commercial
structural steel fabricators, using certified welders and ultrasonic inspection.
For some reasons, these unfavorable results did not make
completely their way through the engineering community

Fig. 44. Typical failure modes in beam flange to column connections (adapted from Bertero
et al. 1994; Miller 1994; Ross and Mahin 1994).
TYPE 1
TYPE 2
TYPE 3
TYPE 4
TYPE 7
TYPE 5
TYPE 6
TYPE 8
and, as mentioned earlier, the full-penetration welded flange

connection was still recommended in U.S. codes and extensively used in California at the time of the earthquake. Thus,
the considerable number of brittle failures that occurred in
the Northridge area as a result of the earthquake was not surprising. However, the reasons for the failures observed in
laboratory had not yet been clearly identified by researchers.
Thus, only tentative explanations could be formulated for the
damage experienced by steel moment resisting frames in the
Northridge earthquake.
Cracking of the beam flange welds has been mainly attributed to overstressing of the welds and to weld defects. During
the welding process, significant stresses can be induced in
the welds because moment connections are highly restrained.
Upon loading, welds become overstressed because of the
inadequacy of the bolted web connection to transfer bending
moments; at the column face, increased axial stresses
develop in the beam flanges as the full plastic moment of the
beam must be entirely resisted by the flanges only. The fact
that actual beam material may exhibit lower tensile to yield
strength ratios could also have contributed in overloading the
welds of the beam flanges.
These high stresses make the connections sensitive to
weld imperfections. After the earthquake, weld flaws of
many types have been identified on failed connections: lack
of fusion and penetration in the root region of the welds,
inadequate fusion between multipath welds and adjacent base
metal, use of "end dams" instead of proper weld tabs, slag
inclusions left in the welds because of the nature of the access
to the area to be welded, etc. Inadequate preheat and cooling
rates are also seen as possible contributors to weld failures.

In addition, left-in-place steel backing bars create a notchlike
condition (Fig. 43) that may have led to stress concentration
in the welds and initiation of cracking.
The presence of beam flange continuity plates between the
flanges of the columns may also have played a role in the
weld failures. For instance, two 18-storey buildings displayed beam-column joint fractures; however, one structures' extent of weld failure was more than the others. The
main difference in construction between the two high-rises
was the use of continuity plates in columns. The building
with the beam flange continuity plates had less weld failures
than its counterpart, which suggests that the flexibility of the
column flanges could have resulted in local overstressing of
the welds.
The predominance of failures at the beam bottom flange,
rather than at the top flange, has been partly attributed to the
additional rotational restraint introduced by the composite
floor slab and the web connector, the latter being usually
located close to the top flange. Bottom flange welds are also
seen as better candidates for premature fracture. Weld execution at the bottom flange is more difficult because of the
beam web interference, which makes these welds more prone
to imperfections. Another possible reason why weld failures
occurred at the bottom flange is the fact that the notchlike
condition at the top flange exists on the side of the flange
opposite to the side subjected to the maximum beam flexural
stresses, which is less critical than the situation prevailing at
the bottom flange.
Other contributing factors to the observed weld failures

Tremblay et al.
have been suggested: inadequate field inspection, deficient

through-thickness strength of column flanges, insufficient
toughness of the material, high loading rates and strain rate
effects, large size of the connecting members, actual state of
stress in the welds and connected parts, etc. It is also
believed that tensile axial loads developed in columns during
the ground shaking, which would have been propitious to
crack propagation into the column flanges and webs.
Though an alarmed reaction is not warranted at this time,
since it is significant that none of the reported connection
failures led to collapse, the causes of such a deficiency must
promptly be identified and the current state-of-practice modified accordingly to ensure that building structures can survive much longer duration earthquakes, as expected to occur
in some areas of the country. This revision process will
undoubtedly affect the steel industry in Canada where this
type of connection is common in practice (CISC 1991). Also
of consequence is the fact that existing structures will also
need to be included in this review program.
Steel frames damaged in the Northridge area will obviously represent a potential hazard in future earthquakes and
must therefore be inspected and repaired. Thus, shortly after
the earthquake, the California Seismic Safety Commission
issued a public notice stating that steel buildings were potentially damaged and should be investigated. The City of
Los Angeles also issued a draft regulation which required all
owners of steel buildings to have their structures examined
by a structural engineer. Even moment resisting frames that
are found undamaged after the earthquake may need to be
retrofitted to withstand properly next earthquakes. It must be
realized, however, that the potential problem of existing
moment resisting frames, either damaged or undamaged, is
not limited to the Northridge region. Any steel moment
frame that has sustained ground shaking in the past may have
suffered some structural damage and can be vulnerable to
future earthquakes. Similarly, undamaged existing moment
resisting frames built in seismic regions may pose serious
risks in future seismic events.
In March 1994, a special task committee of the American
Iron Steel Construction (AISC) proposed interim recommendations for the repair of damaged connections and the design
of new moment resisting frames (AISC 1 9 9 4 ~ ) For
. repair,
it was suggested that any cracked material or weld be
removed and replaced by new steel material or welding. Of
course, fabrication and welding practices were to strictly
meet applicable code requirements. In addition, based on the
current state of knowledge at that time, three options were
proposed to enhance the performance of the joint. These are
illustrated in Fig. 45 for the connection shown in Fig. 43.
Firstly, it was suggested that the backing bars of the beam
flange welds be removed, the weld root region backgouged
and rewelded, and, subsequently, a reinforcing fillet weld be
applied. Second, the beam web was to be welded directly to
the column or, alternatively, to the vertical shear connectors
as shown in Fig. 45. The third correction consisted in adding
flange reinforcement plates, either vertical ribs or cover
plates, to help transferring to the column the axial loads acting in the beam flanges. The decision of using only one or
more of those three schemes was left to the engineer. Examples of these repair procedures are presented by Sabol(1994)
and by Tide (1994). For the design of new structures, AISC
recommended to implement these repair provisions when
Fig. 45. Proposed reinforced beam-column joints for retrofit

and new constructions: ( a ) vertical ribs; (b) cover plates
(adapted from Engelhardt and Sabol 1994).
1
PLATE M P )
0
0
/
.b-< 3 SIDES
BACKING BAR REMOVED

ROOT REGION BACKGOUGED
REINFORCED WELD M P )
COVER PLATE TYP
\'
--TYP
"F BACK GOUGE
making use of the standard welded flange and bolted web

connection specified in the 1992 AISC Seismic Provisions
for Structural Steel Buildings.
Meanwhile, research has been initiated in the United
States, as well as in other countries, to explain the observed
connection failures and to propose solutions. In particular, a
test program was started in April 1994 to develop guidelines
for improved moment connections (Engelhardt and Sabol
1994). A total of 16 full-scale specimens made of heavy
members were tested under cyclic loading, which permitted
to examine the performance of four different connection
details. Testing was completed in October 1994. Based on
the results and additional knowledge gained during that
period of time, AISC published a second set of interim guidelines for repair and design of steel moment resisting frames
(AISC 19946). These test results and guidelines are briefly
summarized in the following.
The first connection detail was of the standard welded
flange and bolted type (Fig. 43), but the beam flange welds
had been improved as shown in Fig. 45a. The second connection was similar except that the beam web was welded
directly to the column, making this connection an all-welded
detail. Upon loading, all specimens of these two connection
details experienced brittle fracture at the beam flange weld
early in the test, with no or little plastic rotation in the beam.
The third type of joint included reinforcement plates as
illustrated in Fig. 45. Cover plates were used in eight samples whereas two specimens included vertical ribs. More
effort was directed toward the testing of cover-plated specimens as this detail appears to be less expensive. In these samples, the section modulus at the face of the column ranged
between 1.6 and 2.0 times that of the unreinforced beam.

Standard bolted web connection was used in these connections. In all samples but two, the beam underwent large plastic rotation (between 0.025 and 0.035 rad) and exhibited high
energy dissipation. Both failures initiated in beam flange
weld of cover-plated connections. In one sample, failure took
place at the interface of the weld and the column flange,
whereas crack propagated into the column flange material in
the second failed specimen.
The fourth connection design examined in the test program included side straps connecting the outer edges of the
beam flanges to the outer edge of the column continuity plates.
This arrangement (not shown here) had been proposed to
avoid through-thickness failure of the column flange. In the
tests, both samples of this joint exhibited only fair performance with failure occurring at plastic rotations less than
0.02 rad (this value is generally considered as the minimum
level for acceptance). This detail was also the most costly to
fabricate.
Although limited in scope, this experiment clearly indicated that improving welding quality was not sufficient by
itself to achieve the desired performance. On the other hand,
shifting the plastic hinging in the beams away from the face
of the column by reducing the stress level in the welds through
reinforcement plates appeared to be promising. However,
brittle failures still occurred in the reinforced connections,
which suggests that other issues such as the throughthickness properties of the column flange material and the
presence of continuity plates (one of the reinforced joint
which failed early did not have continuity plates) need further investigation.
On the basis of these findings, AISC reiterated that the use
of the standard welded beam and bolted web connection
described in the 1992 Seismic Provisions should be suspended unless it can be shown that such connection is adequate or that proper reinforcement is provided. It must be
noted that in September 1994, the ICBO had deleted from the
1994 UBC code the section that allowed the use of this connection without testing and calculation. In view of the
difficulty experienced in achieving high quality welds and
detecting weld defects in the test program, AISC also recommended that all welding be performed in strict compliance
with comprehensive welding procedure specifications to be
developed for each project and suggested that ultrasonic testing of groove welds be conducted from the top and bottom
sides of the welds as well as from the back side of the column
flange.
American Institute of Steel Construction also developed
tentative provisions to assess the adequacy of existing connections and to design reinforced joints (Englekirk 1994).
The approach directly accounts for material overstrength,
lower reliability of through-thickness resistance of steel, and
actual stress -strain relationships of the different materials
involved (weld, base metal, etc.). It also includes ultimate
strain limit states for weld metal and steel material. Unfortunately, the testing program did not investigate all the possible contributing factors to the observed connection failures in
Northridge and the provisions could not be thoroughly validated. Moreover, data on the mechanical properties of the
material and on the initial state of stress and strain in connections are also insufficient. Even the expected demand on the
components of moment resisting frames in future earth-
quakes, including the effects of vertical accelerations, needs

to be established for design purposes.
Thus, not enough information is yet available to structural
engineers for implementing sound repair schemes or designing reinforced connections in moment resisting frames. The
Structural Engineers Association of California (SEAOC), the
Applied Technology Council (ATC), and the California
Universities for Research in Earthquake Engineering (CUREE)
have engaged in a joint effort, namely the SAC Joint Venture, to solve this problem. Their objectives are to understand what happened to steel moment resisting frames in the
Northridge earthquake, to provide guidance for the identification and assessment of damage in buildings, and, finally,
to prepare recommendations for the repair, retrofit, and
design of steel moment resisting frames.
The initial phase of this research effort started in November 1994 and is mainly devoted to the development of interim
guidelines for the repair and inspection of existing structures.
This $2.3 million U.S. program, which is funded by the
Federal Emergency Management Agency and the California
Office of Emergency Services, is expected to be completed
in the Spring of 1995 (A. Ross, 1994, private communication). Within the next 3 years, it is expected that an additional
$10 million U.S. research will be carried out by the SAC
Joint Venture to reach its objectives. Findings of this
research and other research programs that are conducted
elsewhere would need to be examined for possible inclusion
in Canadian codes.
Other avenues which have been proposed to enhance the
seismic performance of steel moment frames should also
draw the attention of researchers and designers in the coming
years. For example, intentionally weakening the beams at a
certain distance of the columns would force plastic hinging
in beams away from the connections as reinforcement plates
do. This can be easily achieved by reducing the width of the
flanges over a certain length to create a dog-bone shape, as
proposed by Chen and Yeh (1994), or by drilling holes in the
beam flanges. Similar behavior could also be obtained by
adding haunches to moment connections. As opposed to
other joints, such a reinforcement scheme would exhibit
higher redundancy in the connection, since failure of a single
weld would not lead to nearly losing all the bending strength
and stiffness of the connection (Bertero et al. 1994). Endplate beam-column moment connections, when properly
sized, can also be seen as a viable alternative for ductile
response under seismic loading (Tsai and Popov 1990).
Of course, means of dissipating energy other than beam
hinging can also be exploited. Shear yielding in the column
joint panel zone has been known as an efficient way of releasing the inelastic demand on beams and beam-to-column
connections (Popov et al. 1988). The implementation of
dissipating energy devices and base isolation systems in
moment resisting frames also represents a promising avenue,
both for retrofit projects and new structures.
Connections failures in moment resisting frames also
raised serious concerns on the overall reliability of structures
braced only by a few moment resisting bays. These structures can be very sensitive to failure of a single connection
and it is questionable whether such a nonredundant system
should classify under the ductile moment resisting frame
category for which superior redundancy is implied. This
Tremblay et al.
issue should be addressed in future editions of building codes

in Canada.

Conclusions
Some concentrically steel braced frames and steel moment
resisting frames experienced various structural and nonstructurd damage during the Northridge earthquake. None of them
collapsed as they generally maintained their gravity load
carrying capacity as well as some degree of lateral stability.
The observations made and the information presented in
this paper support for the most part the relevance of the current seismic design provisions included in the Canadian
building code and standard for steel structures. However,
Canadian researchers, code writing committees, and structural engineer designers will need to address a number of
additional issues which can have a significant impact on the
seismic resistant design of steel structures and which are not
(or deficiently) addressed by the current edition of the Canadian codes and standards. This includes the need to
- extend a capacity design approach to the whole lateral
load resisting system of the structure, as well as to all
categories of moment resisting frames and, particularly,
concentrically braced frames, because of their lower
inherent redundancy;
- account for secondary effects that occur upon buckling of
bracing members in concentrically braced frames;
- consider vertical ground accelerations in the design of
horizontal cantilever structures and exterior columns of
bracing bents in concentrically braced frames;
- account for the lower redundancy exhibited by moment
resisting frames having only a few moment resisting bays;
- recognize in the design process the hazard potential from
nonstructural damage occurring during earthquakes.
The first item is most important as the benefits of capacity
design in ensuring a stable ductile behavior positively impact
structural performance, particularly in view of the large
uncertainties involved in the definition of the design earthquake.
One of the main lessons learned from the Northridge
earthquake is the potential deficiency of welded flange and
bolted web beam-column joints in moment resisting frames.
The observation of the damage and the research performed
so far indicate that this type of connection is not adequate to
allow cyclic flexural hinging to take place in beams. Though
some interim guidelines have been proposed for the repair,
retrofit, and design of beam-column joints in moment resisting frames, further research is still needed to determine the
causes of this deficiency and to propose sound connection
details and design rules. The findings of this work may eventually justify the need for strengthening existing moment
resisting frames in seismically active regions of Canada,
even those designed under recent code provisions.
Finally, the Northridge earthquake has sent a strong signal that careful inspection of steel frames is mandatory after
a seismic event, even though only minor damage is apparent
from visual preliminary examination.
Acknowledgements
The authors gratefully acknowledge the Natural Sciences
and Engineering Research Council of Canada for providing
research funds for the reconnaissance visits and subsequent

investigations. The on-site cooperation of Susan K. Tubbesing
of the Earthquake Engineering Resarch Institute, John Hall
of Caltech, and Bob Harder, Manager of the Van Nuys
Office of the Los Angeles Buildings and Safety Division, is
greatly appreciated. The assistance of John Sherstobitoff of
Sandwell Inc., Denis Mitchell of the Department of Civil
Engineering of McGill University, and RenC Tinawi of the
Department of Civil Engineering of ~ c o l epolytechnique
during the reconnaissance, together with conveyance of
information and photographs from David Bonneville, Jay
Love, and James 0 . Malley of H.J. Degenkolb Associates of
San Francisco is gratefully acknowledged. Material extracted
from conversations with local Los Angeles engineers Charles
Thiel, Jock Scott, and Geoffrey Hichborn Sr. is similarly
appreciated. Michael Engelhardt of the Department of Civil
Engineering, University of Texas, Austin, and Arthur Ross
of the Structural Engineers Association of California, California, are also sincerely thanked for their contribution to the
paper. The information provided by CSIMP, OES, and
USGS was invaluable in interpreting the ground motion
characteristics.
References
AISC. 1992. Seismic provisions for structural steel
buildings. American Institute of Steel Construction,
Inc., Chicago, Ill.
AISC. 1994a. Executive summary - Interim observations
and recommendations of AISC Special Task Committee
on the Northridge Earthquake. Proceedings of AISC
Special Task Committee on the Northridge Earthquake
Meeting, American Institute of Steel Construction,
Chicago, Ill.
AISC. 19946. AISC Northridge Technical Bulletin No. 2 Interim observations and recommendations on steel
moment resisting frames. Northridge Steel Update I,
American Institute of Steel Construction, Chicago Ill.
Astaneh-Asl, A., Goel, S.C., and Hanson, R.D. 1985.
Cyclic out-of-plane buckling of double angle bracing.
ASCE Journal of Structural Engineering, 111:
1135- 1153.
Bertero, V.V., Anderson, J.C., and Krawinkler, H. 1994.
Performance of steel building structures during the
Northridge earthquake. Earthquake Engineering Research
Center, College of Engineering, University of California,
Berkeley, Calif., Report No. UCBIEERC-94/09.
Bonneville, D.R. 1994. Northridge earthquake damage to
a modern steel braced frame office building.
Proceedings of the AISC Special Task Committee on
the Northridge Earthquake Meeting, Chicago, Ill.
Bruneau, M., and Mahin, S.A. 1990. Full-scale tests of
butt-welded splices in heavy-rolled steel sections
subjected to primary tensile stresses. Engineering
Journal, American Institute of Steel Construction, 28:
1-17.
BSSC. 199 1 . National Earthquake Hazards Reduction
Program (NEHRP) recommended provisions for the
development of seismic regulations for new buildings.
Part I: provisions. Building Safety Seismic Council,
Washington, D.C.
Chen, S.-J., and Yeh, C.H. 1994. Enhancement of

ductility of steel beam-to-column connections for

seismic resistance. Proceedings of the Structural
Stability Research Council 1994 Task Group Meetings
and Technical Session, Bethlehem, Pa., pp. 327 -338.
Cheng, R.J.J., Rabinovitch, J.S., and Yam, M.C.H.
1993. Behavior of gusset plate connections under
compressive monotonic and cyclic loadings.
Proceedings of the 1993 Annual Technical Session of
the Structural Stability Research Council, Milwaukee,
Wis.
CISC. 1991. Handbook of steel construction. 5th ed.
Canadian Institute of Steel Construction, Willowdale,
Ont.
CSA. 1989. Limit state design of steel structures.
Canadian Standards Association, Rexdale, Ont.,
Standard CANICSA-S 16.1-M89.
EERI. 1994. Northridge earthquake January 17, 1994 Preliminary reconnaissance report. Earthquake
Engineering Research Institute, Oakland, Calif., Report
NO. 94-0 1.
Engelhardt, M.D., and Hussain, A.S. 1993. Cyclicloading performance of welded flange-bolted web
connections. ASCE Journal of Structural Engineering,
119: 3537-3549.
Engelhardt, M.D., and Sabol, T.A. 1994. Testing of
welded steel moment connections in response to the
Northridge earthquake. Progress report to the AISC
Advisory Subcommittee on Special Moment Resisting
Frames Research, Northridge Steel Update I, American
Institute of Steel Construction, Chicago, Ill.
Englekirk, R.E. 1994. Design implications derived from
the University of Texas, Austin, test program.
Northridge Steel Update I, American Institute of Steel
Construction, Chicago, Ill.
Filiatrault, A., Anderson, D.L., and DeVall, R.H. 1992.
Effect of weak foundation on the seismic response of
core wall type buildings. Canadian Journal of Civil
Engineering, 19: 530-539.
Finn, W.D., and Ventura, C.E. 1994. Ground motions.
Preliminary report on the Northridge, California,
earthquake of January 17, 1994. Canadian Association
for Earthquake Engineering, Vancouver, B.C.,
pp. 7-71.
Ghosh, S.K. 1994. Code implications of the Northridge
earthquake of January 17, 1994. Proceedings of the
Sixth ATC 15-5 U.S. - Japan Workshop on the
Improvement of Building Structural Design and
Construction Practices, Victoria, B.C., pp. 3.1 -3.16.
ICBO. 1982. Uniform building code. International
Conference of Building Officials, Whittier, Calif.
ICBO. 1994. Uniform building code. International
Conference of Building Officials, Whittier, Calif.
Khatib, I.F., Mahin, S.A., and Pister, K.S. 1988. Seismic
behavior of concentrically braced steel frames.
Earthquake Engineering Research Center, University
of California, Berkeley, Calif., Report
NO. UCBIEERC-88/01.
Knoll, F. 1993. Seismic performance of reinforced
concrete ductile moment-resisting frame buildings
located in different seismic regions: Discussion.
Canadian Journal of Civil Engineering, 20: 333.
Liu, Z., and Goel, S.C. 1988. Cyclic load behavior of
concrete-filled tubular braces. ASCE Journal of

Structural Engineering, 14: 1488- 1506.
Miller, D.K. 1994. Northridge: the role of welding
clarified. The Welding Innovation Quarterly, The
James F. Lincoln Arc Welding Foundation, IX: 4- 13.
Mitchell, D., DeVall, R.H., Saatcioglu, M., Simpson, R.,
Tinawi, R., and Tremblay, R. 1995. Damage to
concrete structures due to the 1994 Northridge earthquake.
Canadian Journal of Civil Engineering, 22: 361 -377.
NRCC. 1990. National building code of Canada, 1990.
10th ed. Associate Committee on the National Building
Code, National Research Council of Canada, Ottawa,
Ont.
Popov, E.P., and Black, R.G. 1981. Steel struts under
severe cyclic loadings. ASCE Journal of the Structural
Engineering Division, 107: 1857 - 1881.
Popov, E.P., Tsai, K.-C., and Engelhardt, M.D. 1988.
On seismic steel joints and connections. Proceedings of
the 57th Annual Convention, Structural Engineers
Association of California, Makahiki, Hawaii,
pp. 187-208.
Priestley, M.J.N. 1993. Myths and fallacies in earthquake
engineering - conflicts between design and reality.
Bulletin of the New Zealand National Society for
Earthquake Engineering, 26: 329 -348.
Redwood, R.G., and Channagiri, V.S. 1991. Earthquake
resistant design of concentrically braced steel frames.
Canadian Journal of Civil Engineering, 18: 839 - 850.
Redwood, R.G., Lefki, L., and Amar, G. 1990.
Earthquake resistant design of steel moment resistant
frames. Canadian Journal of Civil Engineering, 17:
659-667.
Roeder, G.C., Carpenter, J.E., and Taniguchi, H. 1989.
Predicted ductility demands for steel moment resisting
frames. Earthquake Spectra, 5: 409 -427.
Ross, A.E., and Mahin, S.H. 1994. Steel moment resisting
frames and the Northridge earthquake. Proceedings of
the Sixth ATC 15-5 U.S. - Japan Workshop on the
Improvement of Building Structural Design and
Construction Practices, Victoria, B.C., pp. 10.1- 10.10.
Sabol, T.A. 1994. Damage to ductile steel frames in the
Northridge earthquake. Proceedings of AISC Special
Task Committee on the Northridge Earthquake
Meeting, Chicago, Ill.
Tide, R.H.R. 1994. Fracture of welded beam-to-column
connections when subjected to seismic loadings.
Proceedings of AISC Special Task Committee on the
Northridge Earthquake Meeting, Chicago, Ill.
Tremblay, R., and Stiemer, S.F. 1993. A contribution to
friction bolted connections for concentrically braced
steel frames. Is your structure suitably braced? 1993
Structural Stability Research Council Conference,
Milwaukee, Wis., pp. 189-200.
Tsai, K.-C., and Popov, E.P. 1990. Cyclic behavior of
end-plate moment connections. ASCE Journal of
Structural Engineering, 116: 2917-2930.
White, R.N., and Salmon, C.G. 1987. Building structural
design handbook. John Wiley & Sons, New York, N.Y.
Yanvec, P.I., Gillengerten, J.D., and Hamburger, R.O.
1991. The performance of steel buildings in past
earthquakes. American Iron and Steel Institute and
EQE Engineering, Inc., Washington, D.C.

Performance of Steel Structures During The 1994 Northridge Earthquake

Uploaded by

Copyright:

Available Formats

Performance of Steel Structures During The 1994 Northridge Earthquake

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Performance of Steel Structures During The 1994 Northridge Earthquake

Uploaded by

Copyright:

Available Formats

Performance of steel structures during the

1994 Northridge earthquake

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by University of Toronto on 12/02/14

Robert Tremblay, Peter Timler, Michel Bruneau, and

ties were attributed to unsatisfactory performance of steel

Received June 16, 1994.

R. Tremblay and A. Filiatrault. Department of Civil

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by University of Toronto on 12/02/14

The first part of the paper outlines the characteristics and

Seismic design fundamentals of steel

design. This is done by defining three different categories of

Moment resisting frames

Can. J. Civ. Eng. Vol. 22, 1995

Table 1. Summary of observed structural damage.

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by University of Toronto on 12/02/14

Kaiser Permanente Hospital penthouse

Buckling of bracing members; excessive sway

First Interstate Bank Building at Northridge

Buckling of brace connecting plates; possible yielding of

Student Union Building, California State University at

No structural damage observed

Roof structure for the bleachers of the football field

Failure of anchor bolts (uplift)

Oviatt Library, California State University at Northridge

Failure of brace connecting plates; cracking of base

Three-storey building under construction in Van Nuys

Buckling of bracing members

NO. 2 Brewhouse, Anheuser-Busch Inc.

Buckling of bracing members

Asphalt and rock plant

Yielding and failure of anchor bolts

Department of Water and Power San Fernando Generating

No structural damage observed

Four-storey commercial office structure

Buckling and failure of bracing members; failures of

Two-storey fashion plaza

Cracking in floor slab; buckling of bracing members

Holy Cross Hospital administration building

Excessive sway; failure of anchor bolts

Van Nuys office building

No structural damage observed

MRF under construction

Failure of beam-column moment connections

*CBF: concentrically braced frame; MRF: moment resisting frame.

for earthquake resistance. This connection includes a bolted

hinging and panel zone shear yielding). Capacity design rules

Diaphragms and anchorage to foundations

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by University of Toronto on 12/02/14

Fig. 1. Kaiser Permanente Hospital: cladding loss on

Fig. 2. Kaiser Permanente Hospital: buckled double angle

cases presented below represent up-to-date information at the

Kaiser Permanente Hospital penthouse, Panorama City

Fig. 3. Kaiser Permanente Hospital: racking of services

First Interstate Bank Building at Northridge

bracing members (Fig. 5). At the base of the columns of the

Can. J. Civ. Eng. Downloaded from www.nrcresearchpress.com by University of Toronto on 12/02/14

Can. J. Civ. Eng. Vol. 22, 1995

Fig. 4. First Interstate Bank Building in Northridge: steel

Fig. 6. First Interstate Bank Building in Northridge: column