17th World Conference on Earthquake Engineering, 17WCEE

Sendai, Japan - September 13th to 18th 2020

Paper N° C004022
Registration Code: S- A02988

SAFETY OF TALL PRE-NORTHRIDGE STEEL FRAME BUILDINGS AND

IMPLICATIONS ON CORDONING AND RECOVERY
G. G. Deierlein1 , W.Y. Yen2 , A.M. Hulsey3 , F. Galvis4, J.W. Baker5 and C. Molina Hutt6

(1)J.A.Blume Professor of Engineering, Stanford University, [email protected]

(2)PhD Student, Stanford University, [email protected]
(3)PhD Student, Stanford University, [email protected]
(4)PhD Student, Stanford University, [email protected]
(5)Professor, Stanford University, [email protected]
(6)Assistant Professor, University of British Columbia, [email protected]

Abstract
The downtown business districts of San Francisco, Los Angeles and other cities in the western United States have
significant inventories of older (pre-2000) tall steel buildings that have elevated risk, compared to new code-conforming
buildings, of damage and collapse from strong earthquakes. Many of these buildings have welded connections of the type
that experienced sudden brittle fractures during the 1994 Northridge earthquake, and they were designed under building
codes with less stringent requirements on seismic drifts, minimum column strengths, and seismic base shears.
Compounding these deficiencies are new seismic hazard models which indicate that long-period spectral accelerations
and durations of earthquake ground motions are larger than considered in older design codes. This study applies newly
developed models to assess the response of older steel buildings from the onset of damage up to collapse. The analysis
results are used to examine issues associated with post-earthquake inspection and safety, which have implications on
building reoccupancy and recovery. The study considers the implication of structural damage on recovery of both the tall
building itself, as well as implications of safety cordons around the tall building on the re-occupancy and recovery of
neighboring buildings. A method is proposed for quantifying the post-earthquake safety of damaged tall buildings to
inform criteria for establishing safety cordons. The building-specific performance-based assessment method of FEMA
P58 is extended and applied to quantify the post-earthquake recovery time for neighboring buildings. In addition to
providing more comprehensive estimates of community downtime, this framework can include metrics for whether certain
buildings are likely to induce inordinately long downtimes in surrounding buildings or how a cordoning strategy will
affect a community. This type of downtime assessment can inform a range of policy and planning decisions, such as
requiring seismic retrofit of vulnerable tall buildings, evaluating emergency response strategies, and developing post-
earthquake recovery plans.

Keywords: seismic assessment, tall buildings, steel frame, cordoning, collapse safety, recovery
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

1. Introduction
Tall buildings play an important role in the socio-economic activity of major metropolitan areas in the United
States, and their resilience is vital to ensuring an effective recovery after major disasters. Damage to tall
buildings has the potential to affect a large number of people and can have significant consequences on
surrounding areas. Events such as the Canterbury earthquake in 2011 have highlighted the impact of poorly
performing buildings on the business continuity of downtown districts, where tall buildings are clustered
together. Following the Christchurch earthquake, the 26-story Hotel Grand Chancellor sustained significant
damage and residual drifts, prompting authorities to cordon off an area with a radius that was roughly equal to
its height [1]. Thus, in addition to the direct economic losses in the Hotel Grand Chancellor, there were
significant indirect losses attributed to business disruption in surrounding buildings.

Since the 1906 San Francisco earthquake, structural engineers have generally regarded steel moment-
resisting frame (MRF) systems as being among the most ductile and reliable seismic force-resisting systems for
buildings [2]. The common expectation was that, when subjected to earthquake shaking, MRFs would
experience only localized damage due to ductile yielding of members and connections. This expectation led to
widespread construction of this system, particularly in the high seismic regions of the western US from the
1960s through the 1990s [2,3]. The 1994 Northridge earthquake dramatically changed perceptions of the
performance of such frames, after which post-earthquake inspections revealed cracking in the beam-to-column
joint welds in several dozens of low- and mid-rise, steel-frame buildings [2].

Until about 15 years ago, most tall buildings in high seismic regions were designed using conventional
building code approaches, following a prescriptive force-based approach, without an explicit assessment of
nonlinear dynamic performance during major earthquakes. More recently, several jurisdictions, including Los
Angeles, San Francisco and Seattle, have adopted performance-based approaches for the design of new tall
buildings, which employ nonlinear analyses of response. While these approaches generally provide more
reliability in terms of building safety in modern buildings, little is known about the seismic performance of older
existing tall buildings that were designed prior to the adoption of modern design approaches [4], many of which
employ seismically vulnerable steel MRFs.

Recent studies have highlighted the vulnerability of these existing tall steel MRF buildings in terms of
structural performance, direct economic losses and downtime. For example, Lai et al. [5] evaluated the
performance of an existing 35-story steel building, under the design basis ground motions in San Francisco,
and observed the tendency to form weak-story regions, connections failures of brittle pre-Northridge welded
beam-to-column connection details, and a high probability of brittle failures of column splices. Molina Hutt et
al. [4], conducted a seismic performance assessment of a similar vintage building under design basis motions
and determined an expected direct economic loss of over 30% of building replacement cost and downtime
estimates of more than 1-1/2 years. Under higher intensity maximum considered earthquake ground motions,
the building had a significant collapse risk. These studies demonstrated the capabilities of state-of-art methods,
such as FEMA P58 [6] and REDi [7], to perform detailed assessments of individual buildings. However, these
existing frameworks only consider individual building performance and neglect the impact of damage to these
buildings on neighboring buildings and overall community resilience.

This study applies newly developed models to assess the response of older steel buildings from the
onset of damage up to collapse. The analysis results are used to examine issues associated with post-earthquake
inspection and safety, which have implications on building reoccupancy and recovery. The study considers the
implication of structural damage on recovery of both the tall building itself, as well as implications of safety
cordons around the tall building on the re-occupancy and recovery of neighboring buildings. A method is
proposed for quantifying the post-earthquake safety of damaged tall buildings to inform criteria for establishing
safety cordons. In addition to providing more comprehensive estimates of community downtime, this
framework can include metrics for whether certain buildings are likely to induce inordinately long downtimes
in surrounding buildings or how a cordoning strategy will affect a community. This type of downtime

2
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

assessment can inform a range of policy and planning decisions, such as requiring seismic retrofit of vulnerable
tall buildings, evaluating emergency response strategies, and developing post-earthquake recovery plans.

While only two moderate size cities [15,16] have introduced requirements for assessment and retrofit
of existing steel MRF buildings, San Francisco, Los Angeles and other cities do have ordinances that require
assessments and retrofit during substantial building renovation. Looking towards the future, San Francisco’s
Earthquake Safety Implementation Program [8] has plans to develop assessment and retrofit requirements for
existing tall steel buildings, which would be modeled after ordinances that are currently being implemented
for soft-story and non-ductile concrete buildings. Just as with current programs for concrete buildings, the
development and implementation of an effective program for tall steel buildings will require guidelines for
screening and assessment of steel MRFs, along with practical and cost-effective retrofit solutions.

2. Characterizing the Risk to the Urban Built Environment

2.1 Influence of Tall Buildings on the Resilience of San Francisco
Since the late 1950s, San Francisco’s skyline has changed dramatically with the construction of tall
buildings in the downtown area. Among the multi-faceted earthquake risks facing the city, the concentration
of tall buildings and infrastructure in the densely populated downtown neighborhood raises questions about
the risks to life, property, and recovery from large earthquakes. As a first step towards addressing these
questions, San Francisco’s Tall Buildings Study [3] developed an inventory of the city’s buildings that are
taller than 240 ft. The inventory classifies tall buildings in terms of height, age, use, and structural system
characteristics. Although tall buildings are not the only structures at risk, they are of special concern due to
their large size and occupancies, where earthquake damage to one tall building can have disproportionate
effects on its occupants, its neighbors, and the community at large.

Figure 1 – Inventory of Tall Buildings in Downtown San Franciso [1]

The tall building inventory includes detailed information on 156 buildings over 240-feet tall, which currently
exist or have been permitted for construction in San Francisco. In terms of height, roughly 10% of the inventory
is below 20 stories, 70% of the inventory is in the 20- to 40-story range, 20% is above 40 stories. With regards
to occupancy, approximately 60% of the buildings are commercial and just under 40% are residential or hotel.
With regards to age of construction, over 55% of the tall building inventory was constructed between 1960
and 1990. Tall building construction slowed down in the 1990s, but has resurged since then, with almost 25%
of the inventory constructed since 2000. Illustrated in Figure 1 are the number of tall buildings that were
constructed in San Francisco over the past sixty years, differentiated by structural materials (steel or concrete)

3
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

and structural system types. In terms of construction material, structural steel systems account for about 65%
of the inventory, reinforced concrete systems account for about 20% of the buildings, and the remainder either
incorporate mixed steel-concrete systems or, in a few instances, the systems were unidentified.

With regards to structural configuration, about 50% of the San Francisco tall building inventory is comprised
of 65 steel MRF tall buildings constructed from the 1960s to the 1990s. These buildings are of interest due to:
(1) their prominence as one of the most common structural system types in the tall building inventory; (2) their
design, which followed an equivalent lateral force procedure based on the first-mode translation response,
without capacity design principles that protect against story mechanisms, and lower base-shear strengths than
those specified in modern building codes; and (3) concerns regarding the potential for fracture-prone welded
connections, which came to light following the 1994 Northridge earthquake.

2.2 Benchmarking Older vs. Modern Steel Moment Resisting Frames

It is well established that changes in commonly used weld processes during the mid-1960s led to welds
with low toughness, as evidenced by weld fractures observed in the 1994 Northridge earthquake [2]. Therefore,
it is believed that fracture-prone pre-Northridge moment connections are present in many of the steel MRFs
within the inventory. As outlined in [4], a comparison of pre-1990s Uniform Building Code seismic design
requirements against modern design standards highlights a number of design considerations not present in
older standards, including: (1) use of response spectrum analysis method as opposed to equivalent lateral force
procedures; (2) consideration of lateral forces acting simultaneously in both building directions; (3) minimum
base shear requirements; (4) consideration of p-delta effects; (5) consideration of accidental torsion and vertical
and horizontal irregularities; (6) strong column weak beam requirements; (7) panel zone design checks; (8)
capacity design principles; and (9) prequalified seismic connection details.

(a) (b)
Figure 2 – Comparison of Loss Functions (a) 1973 Steel Moment Frame, (b) 2015 Steel Moment Frame [4]

While steel MRFs are clearly more vulnerable than originally envisioned, the significance of the risk in
these buildings is likely to be highly variable, depending on the specific characteristics of the building. A recent
study by Molina Hutt et al. [4] suggests that the older (pre-1994) steel MRFs with fracture prone connections
could have a mean annual frequency of risk of collapse 28 times higher than new code conforming buildings,
or approximately 13% versus 0.5% probability of collapse in 50 years. The study also found a 65% increase
in average annual risk of economic loss, as illustrated in Figure 2, and a twofold increase in downtime to re-
occupancy and functional recovery. The modest differences in annualized loss and downtime between older
and modern steel MRFs are in rather stark contrast to the significant changes in collapse risk. The likelihood
of observing collapse in older steel MRF buildings is comparable to the likelihood of demolition due to

4
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

excessive residual drifts in modern ones. While these outcomes have distinct impacts on life safety, they have
similar implications on cost and downtime (i.e. total replacement cost and time).

Building performance evaluations at the design basis earthquake (DBE in Figure 2) and the maximum
considered earthquake (MCE in Figure 2) shaking intensities further suggest that pre-1994 tall steel MRFs
have much higher risks of collapse under extreme ground motions and of damage and building closure in
moderate earthquakes. Furthermore, while modern building code requirements provide acceptable seismic
collapse safety (negligible collapse risk), they do not necessarily ensure a level of damage control to assure a
swift recovery after a damaging earthquake due to extensive downtime.

3. Nonlinear Response Simulation and Retrofit of Steel Buildings

3.1 Simulation of Fracture Critical Beam-Column Connections
A key challenge in assessing the performance of pre-Northridge type steel frames is being able to
accurately simulate the degrading response of fracture-critical welded connections. As illustrated in Figure 3,
one approach for modeling the connections is through a fiber-hinge, where the fibers are defined to simulate
(1) yielding and fracture in the welded beam-flange to column joint, and (2) yielding, bolt shear and/or tear
out of the bolted shear-tab web connection. An overview of this approach is described in guidelines for
nonlinear analysis of steel moment frames [9]. The equivalent fracture stress (σcr, ksi) in the welded flange
connection (Figure 3) is described by the following fracture mechanics-based equation:
𝜎𝜎𝑐𝑐𝑐𝑐 = 𝐾𝐾𝐼𝐼𝐼𝐼 ⁄(𝐶𝐶 + 2𝑎𝑎0 ) (1)
where ao (inches) is equal to the depth of the weld root flaw, C is equal to 1.2 or 0.5 for the lower and upper
beam flange weld, respectively, and 𝐾𝐾𝐼𝐼𝐼𝐼 (ksi√in) is the fracture toughness, which can be determined from

Figure 3 – Fiber Hinge Model of Steel Connection with Weld Fracture and Bolt Failure [9]

5
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

Charpy V-Notch (CVN) measurements of the weld metal. Equation 1 applies to welded flange joints, where
the weld backing bars are left in place and the weld root flaw ao is limited to the lesser of one half of the flange
thickness or 0.5 inch. For pre-Northridge connections, typical values of ao are on the order of 0.1 to 0.2 inches
and 𝐾𝐾𝐼𝐼𝐼𝐼 ranges from 50 to 100 ksi√in.

To account for the cumulative effects of low-cycle fatigue behavior, fracture of the flange fiber is evaluated
using the following damage index (DI), which is an adaptation of Miner’s rule [10]:
1�
𝜎𝜎𝑖𝑖 𝑚𝑚
𝐷𝐷𝐷𝐷 = ∑𝑖𝑖 � � (2)
𝜎𝜎𝑐𝑐𝑐𝑐

where m is a parameter that is calibrated to a parametric equation that accounts for the material toughness
(CVN) and flaw size (ao). For random loading cycles, the damage index is evaluated using rain-flow counting
[11]. The cyclic fatigue-fracture model is implemented in OpenSees [12] in series with the Steel02 Menegotto-
Pinto material model to simulate the inelastic response of beam flange welds. Shown in Figure 4 is an example
of the resulting connection model, where the plot on the upper left shows the overall moment versus rotation
response of the connection (superimposed over data from a connection test), and the plots on the right side
show the equivalent uniaxial stress-strain response of the beam flanges, accounting for weld fracture in the
upper flange.

Figure 4 – Simulation of Steel Connection With Cyclic Damage Model

3.2 Case Study Building: Numerical Simulations vs Observations following Northridge Earthquake
The proposed connection model was incorporated in the analysis of a 17-story steel moment frame building
that experienced connection fractures in the Northridge earthquake, to further evaluate and illustrate the
model’s capabilities. Details of the building and the observed connection fractures are described in a previous
analysis study of the frame by Chi et al. [13]. In contrast to the analyses that Chi et al. conducted twenty years
ago, which were limited in their ability to directly simulate weld fractures, the proposed connection model
allows direct simulation of weld fractures, based on the assumed weld toughness and initial flaw sizes. Shown
in Figure 5 is a comparison of the observed connection damage and the simulated connection fractures and
story drift ratios, where the simulation was run using a ground motion that was recorded nearby to the building.

6
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

Overall, the results agree well, where the peak story drifts were consistent with previous frame studies and the
locations of observed connection weld fractures agree well with the observed damage.
Among many questions that one can ask about the building performance, an important consideration is whether
the building is safe to reoccupy prior to and during repairs. Whereas many of the steel moment frame buildings
that were damaged in the Northridge earthquake remained occupied, the same decision may not be made today,
considering (1) what we know about the widespread extent of the fractured connections, (2) greater
appreciation about the risk of large aftershock motions, and (3) overall greater risk aversion. The FEMA 352
guidelines for post-earthquake evaluation and repair of welded steel moment frames [2] include a damage
index for connections, which, when applied to the frame shown in Figure 5, would suggest that the frame is
on the border of being safe to occupy during repairs. Whereas the FEMA 352 criteria for occupancy were
based largely on expert judgment, it is now possible to better quantify the collapse safety, based on detailed
nonlinear analyses coupled with rigorous earthquake hazard and risk assessment.

18

2 GMs

median

16

14

12

10

0 0.01 0.02

SDR
max

Figure 5 – Comparison of observed connection damage (left) and simulated response and damage (right) for
a seventeen story steel moment frame

3.3 Seismic Retrofit

The large population of pre-Northridge steel moment frames in San Francisco, Los Angeles, Seattle and other
west coast cities, combined with (1) experiences of recent devastating earthquakes in New Zealand, Chile, and
Japan and (2) community interest in earthquake resilience, has prompted discussions of whether these buildings
should be retrofit [14]. The Cities of Santa Monica and West Hollywood have already adopted seismic retrofit
ordinances for older steel moment frames [15,16], the City of San Francisco is evaluating development of an
ordinance [8], and other cities are considering retrofit programs. Moreover, recognizing the significant
investment in their buildings and valuing the safety of their employees or tenants, some building owners have
already implemented voluntary seismic retrofits [14].
Seismic retrofit of older steel moment frame buildings can cost tens of millions of dollars in direct construction
costs, in addition to the associated project costs due to disruption of operations, relocation of tenants, etc. As
such, seismic retrofit should be undertaken carefully Seismic retrofit solutions can vary significantly,
depending on the specific configuration of the building, the goals of the seismic retrofit (basic life safety versus
higher performance), and other factors, such as construction logistics and whether the building is undergoing

7
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

other major renovations or will be occupied during construction. In tall buildings, the seismic retrofit work
may be staged over several years to minimize disruption to tenants by timing the work around lease renewals.
Design of seismic retrofits generally includes detailed nonlinear dynamic analyses to evaluate the seismic
performance under the existing conditions and the effectiveness of the proposed retrofit. For pre-Northridge
moment frames, areas of specific concern in the analyses include (a) modeling of the fracture-critical beam-
to-column and column splice connections, (b) identification of weak stories or vertical irregularities, due to
extra tall stories at ground floor lobbies or other locations, (c) in-plan irregularities introduced by large atria
or other irregular floor plans, and (d) degradation due to non-compact shapes or members with large out-of-
plane slenderness.
There are typically many possible retrofit strategies, the feasibility and effectiveness of which will depend on
the specific circumstances. Some of the common strategies that have either been implemented on real projects
or explored in research studies include:
• Adding viscous dampers or buckling restrained braces to stiffen the frame and reduce deformation
demands
• Adding reinforced concrete shear walls or steel bracing (usually in the building core) to either
supplement or replace the lateral stiffness of the moment frame
• Adding pivoting or rocking concrete walls or steel braced spines to regularize and reduce excessive
story drifts
• Rewelding existing welded beam-column and/or column splice connections with toughness-rated
materials to improve their fracture resistance
• Reinforcing or reconfiguring existing welded connections to improve their strength and fracture
resistance

4. Risk and Policy Implications for Cities

4.1 Collapse Risk and Reoccupancy
Looking beyond the primary concern for adequate collapse safety of buildings in earthquakes, a further
question that can affect retrofit policy decisions is whether buildings that are damaged by earthquakes are safe
to reoccupy. Answering this question requires (1) evaluation of the collapse safety of damaged building to
future earthquake ground shaking or other loading effects, and (2) criteria to judge the minimum acceptable
level of required safety, recognizing that even an undamaged existing building is likely to have lower collapse
safety than is generally deemed acceptable for new buildings. Thus, with respect to safety from aftershocks,
one can approach the evaluation in either an absolute sense (i.e., what is the risk of collapse due to earthquake
aftershocks) or in a relative sense (i.e., how much less safe is the damaged building to aftershocks as compared
to the undamaged building).

Several prior studies to examine collapse safety of damaged buildings [17,18,19] have proposed to evaluate
post-earthquake safety in terms of changes to the collapse fragility as a function of damage introduced into the
structure from an earthquake. Following an approach proposed by Burton et al. [18], this study used nonlinear
analyses of a 20-story reinforced concrete moment frame as a trial study to examine the effect of earthquake
damage on collapse safety. These consisted of, first, subjecting the model to a damaging input ground motion
(referred to herein as a ‘foreshock’), and then, performing incremental dynamic analyses to evaluate the
collapse fragility. Shown in the left plot of Figure 6 are collapse fragility curves for the (1) undamaged frame,
and (2) frames that had experienced foreshock ground motions that induced specific levels in peak story drifts.
As expected, the collapse fragility curves (Figure 6a) shift to the left with increasing foreshock intensities, as
measured by the story drift demands induced by the foreshocks. Note that the specific levels of peak foreshock
story drift demands for each of the collapse fragilities in Figure 6a are identified by the corresponding points
in Figure 6b, which relate the reduction in median values of collapse intensity to the foreshock story drift
demands.

8
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

As shown in Figure 6, the median values of collapse capacity reduce to about 60% of the value of the intact
frame due to damage induced by foreshocks with peak story drift demands of 5%. This shift in median
translates to a large increase in the probability of collapse (reading vertical statistics from the collapse
fragilities in Figure 6a). For this illustration, peak story drift is considered as a fairly robust metric of induced
damage, which is easy to measure from the analyses. Ongoing efforts will look at other measures, such as
residual drifts, fractured connections, etc. that would be physically observable damage measures. With this
type of data, one could then make a determination regarding the minimum threshold of observed damage that
would reduce the collapse safety below a minimum acceptable level. This analysis would provide a more
quantitative basis to support existing criteria for building reoccupancy, which up to this point has been based
largely on engineering judgement.

(a) (b)
Figure 6 – Assessment of Post-Earthquake Building Collapse Safety (a) Normalized Collapse Fragilities
and (b) Relationship Between Peak Foreshock Story Drifts and Reduction in Median Collapse Capacity

4.2 Safety Cordons

Along with reoccupancy of a building, an associated question is the risk that a damaged building poses to its
neighbors. Where the risk from a damaged building is deemed too large, safety cordons are established to
restrict access around the damaged building. While the cordons may be necessary for life-safety protection,
the they can significantly impede the recovery of neighboring buildings and the community in general. This
was seen most dramatically in Christchurch, NZ following the 2011 Canterbury Earthquake, where entire
regions of the downtown area were cordoned off for many months. As illustrated in Figure 7 for downtown
San Francisco, the potential impact of cordons can be quite dramatic in dense urban downtown areas with
many tall buildings.

This study developed a new community recovery assessment framework that includes the effect that safety
cordons around damaged buildings may have on the accessibility of neighboring buildings. This concept is in
contrast to current downtime models that consider communities as the sum of isolated buildings. The
framework incorporates FEMA P-58’s high resolution, building-level performance assessment methodology
within a spatially distributed recovery model. Each building is evaluated at a range of shaking intensities, using
FEMA P-58 to get the distribution of potential building consequences, including the level of damage and the
time required to stabilize the building (in case it requires a cordon) and the time required for repairing it to a
functional status. The regional hazard is based on earthquake scenarios (e.g. 7.2Mw on the San Andreas Fault),
simulated through many possible ground motion realization maps to capture the uncertainty in the shaking
intensity associated with a specific rupture magnitude and location. These hazard maps provide the shaking
intensities at each building location, which are then used to sample from the distribution of building
consequences for that shaking intensity. After sampling building damages consequences across the
community, the framework uses REDi’s [7] impeding factor model to consider the logistical delays prior to
initiating each building’s repairs. The cordon assessment is incorporated into this impeding factor model as
follows. A cordon is established around any tall building (240ft+) that exceeds the cordon trigger criteria. (For
the current example, the triggering threshold is set as 2% peak story drift demand). The duration of the cordon
is equal to the total time required to stabilize the building. Just as all the buildings must wait until their logistical

9
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

Figure 7 – Potential Extent of Cordoning Around Damaged Tall Buildings in Downtown San Francisco

impeding factors are resolved, any building that is located within a cordon must also wait until the cordon is
removed before initiating repairs (or, in the case of an undamaged building, before the building can be
reoccupied). Every building in the affected area is evaluated in this way to determine the time at which they
are functional again. This recovery simulation is repeated for all realizations of the ground motion maps. Figure
8 shows the expected recovery over all the realizations in downtown San Francisco, where functionality
measured as the percentage of available office space in the community. As illustrated in the figure, the
presence of cordons (shaded area in orange) can have a significant effect on displacing long-time residents.
This type of assessment can provide a direct comparison to recovery targets, such as the time until a certain
level of functionality is restored. There will often be a gap between the status quo and the recovery targets. In
such cases, the recovery can be re-assessed under various policy options in order to evaluate the policies’
efficacy in reducing the gap. The policies may include pre-earthquake planning for mitigating the logistical
delays before the damaged tall buildings can be stabilized, accepting a higher level of risk when establishing
the cordon criteria, or requiring retrofits to limit the potential for significant damage in tall buildings. The
framework could also be used to in conjunction with a map of emergency routes to determine which routes are
least likely to be inaccessible due to the location of the safety cordons (or which buildings need to be addressed
in order to keep the emergency routes open).

Figure 8 – Effect of Repairs, Impeding Factors and Cordons on Post-Earthquake Recovery of

Commercial Office Function

10
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

4.2 Policy Interventions

As noted previously, two smaller cities in California have recently implemented policies for seismically
retrofitting pre-Northridge steel moment frames to mitigate life safety risks from collapse, and San Francisco
and other cities are continuing to evaluate policies to address the risk posed both to life safety and to recovery.
Proposed policies for retrofit range from (1) mandatory requirements for screening and evaluation of all
buildings, (2) mandatory requirements that are triggered by building code provisions related to major
renovations, change in building use, and post-earthquake repairs, and (3) voluntary provisions with appropriate
incentives (e.g., tax credits, zoning variances, etc.). Apart from structural retrofitting, there are other programs
and interventions that could be employed to mitigate the effects of building damage on post-earthquake
recovery. These include requirements for (1) building owners to develop earthquake recovery plans with the
goal toward reducing impeding factors for inspection and repairs to stabilize and reoccupy the building, (2)
installation of seismic instrumentation to facilitate post-earthquake evaluation of structural building damage.

5. Concluding Remarks
While the risks posed by older steel frame buildings are generally not as serious as those associated with
unreinforced masonry or non-ductile concrete buildings, they nevertheless are a significant concern both for
life-safety and post-earthquake recovery. The risks are particularly significant in cities with large numbers of
tall pre-Northridge steel frame buildings, where damage to multiple buildings in close proximity can cause
disproportionate disrupting effects, exacerbated by safety cordons and long recovery times. While the
Northridge and Kobe earthquakes led to major building code changes to improve the reliably of new buildings,
remarkably little has been done over the past quarter century to (1) quantify the risk posed by large inventories
of existing steel frame buildings and (2) develop appropriate policy interventions to mitigate the risks. Recent
initiatives to more proactively address disaster recovery and community resilience suggest that the situation
may be changing. The research described in this paper is one effort to develop quantitative models and tools
to help engineer effective policy interventions.

Acknowledgements
This research is supported by the National Institute of Standards and Technology (NIST Award #
70NANB17H245). Also acknowledged are HBrisk for providing the SP3 software used in the cordon
downtime modeling and the City of San Francisco and the Applied Technology Council for supporting
development of the tall building inventory in San Francisco.

References
[1] NZPA (2011). “Christchurch earthquake: Cordon around Grand Chancellor narrows.”
[http://www.stuff.co.nz/national/christchurch-earthquake/4716748/Christchurch-earthquake-Cordon-
around-Grand-Chancellor-narrows].
[2] FEMA (2000). “Recommended post-earthquake evaluation and repair criteria for welded steel moment-
frame buildings, FEMA 352, Washington, DC.
[3] Applied Technology Council (2018), San Francisco Tall Buildings Study,
https://www.onesanfrancisco.org/sites/default/files/inline-files/ATC-119-
1_SF_TallBuildingsStudy_FINAL.pdf
[4] Molina Hutt, C., Rossetto, T., and Deierlein, G. (2019), “Comparative risk-based seismic performance
assessment of 1970s vs modern tall steel moment resisting frames”, JCSR, 159, pp. 598-610.
[5] Lai, J.W., Wang., S., Schoettler, M.J., Mahin, S. (2015). “Seismic Evaluation and Retrofit of Existing
Tall Buildings in California: Case Study of a 35-Story Steel Moment-Resisting Frame Building in San
Francisco.” PEER Report 2015/14, Univ. of Calif. @ Berkeley.

11
 17th World Conference on Earthquake Engineering, 17WCEE
Sendai, Japan - September 13th to 18th 2020

[6] FEMA (2012). “Seismic performance assessment of buildings.” Rep. FEMA P-58, Washington, DC.
[7] Almufti, I., Willford, M. (2013). “Resilience-based earthquake design initiative (REDi) for the next
generation of buildings.” [http://publications.arup.com/Publications/R/REDi_Rating_System.aspx].
[8] Kornfield, L. (2011), CAPPS Earthquake Safety Implementation Program: Workplan 2012-2042, City
and County of SF, http://sfgov.org/esip/sites/default/files/FileCenter/Documents/9765-esipplan.pdf
[9] Applied Technology Council (2017), Guidelines for Nonlinear Structural Analysis for Design of
Buildings: Part IIa – Steel Moment Frames, https://doi.org/10.6028/NIST.GCR.17-917-46v2
[10] Miner M A. (1945). Cumulative damage in fatigue journal of applied mechanics 12 (1945) No. 3,
12pp[J]. A159-A164
[11] Downing S D, Socie D F. (1982). Simple rainflow counting algorithms[J]. International journal of
11fatigue, 4(1): 31-40.
[12] Mazzoni S, McKenna F, Scott M H, et al. (2006). OpenSeescommand language manual[J]. Pacific
10Earthquake Engineering Research (PEER) Center, 264.
[13] Chi, W-M, El-Tawil, S., Deierlein, G.G., Abel, J.F. (1998), “Inelastic Analyses of a 17-Story Steel
Framed Building Damaged During Northridge,” Engineering Structures, 20(4-6), pp. 481-495.
[14] Fuller, T., 2018, At risk in a big quake—39 of San Francisco’s top high rises—The New York Times,
June 14, 2018, accessed Sept. 10, 2018, at https://www.nytimes.com/2018/06/14/us/california-
earthquakes-high-rises.html.
[15] City of Santa Monica, Seismic Retrofit Program,
https://www.smgov.net/Departments/PCD/Programs/Seismic-Retrofit/
[16] City of West Hollywood, WEHO Seismic Retrofit Program, https://www.weho.org/city-
government/city-departments/planning-and-development-services/building-and-safety/seismic-retrofit
[17] Raghunandan M., Liel A., Luco N., (2015) “Aftershock collapse vulnerability assessment of
reinforced concrete frame structures”, Earthquake Engineering and Structural Dynamics, vol 44
[18] Burton, H.V., Deierlein, G.G. (2017), “Integrating Visual Damage Simulation, Virtual Inspection, and
Collapse Capacity to Evaluate Post-Earthquake Structural Safety of Buildings,” Earthquake Engineering
and Structural Dynamics, 47(2), pp. 294-310.
[19] Burton H., Sreekumar S., Sharma M., Sun H., (2017) “Estimating aftershock collapse vulnerability
using mainshock intensity, structural response and physical damage indicators”, Structural Safety, vol 68

12