Nat Hazards (2016) 83:177–192

DOI 10.1007/s11069-016-2307-z

ORIGINAL PAPER

Application of the FEMA-P58 methodology for regional

earthquake loss prediction

Xiang Zeng1 • Xinzheng Lu1 • T. Y. Yang2,3 •

Zhen Xu4

Received: 9 November 2015 / Accepted: 18 March 2016 / Published online: 26 March 2016
Ó Springer Science+Business Media Dordrecht 2016

Abstract Earthquake-induced building collapses and casualties have been effectively con-
trolled in the last two decades. However, earthquake-induced economic losses have continued to
rise. Following the objective and procedure of next-generation performance-based seismic
design, the economic loss prediction method proposed by FEMA-P58 is extended to regional
earthquake loss prediction in this study. The engineering demand parameters for a large number
of buildings within a region are efficiently obtained through nonlinear time history analysis using
multi-story concentrated-mass shear models. The building data, including structural and non-
structural components, are obtained through field investigation, structural and architectural
drawings, and default database published in the FEMA-P58 document. A case study of Tsinghua
University campus in Beijing is performed to demonstrate the implementation and advantage
using proposed FEMA-P58 method for regional earthquake loss prediction. The results show the
advancement in loss simulation for a region, and in identifying the influence of the different
ground motion characteristics (e.g., velocity pulse) on the regional loss.

Keywords Earthquake engineering  FEMA-P58  Earthquake economic loss  Regional

seismic damage simulation  Next-generation performance-based seismic design

& Xinzheng Lu
luxz@tsinghua.edu.cn
1
Department of Civil Engineering, Tsinghua University, Beijing 100084,
People’s Republic of China
2
International Joint Research Laboratory of Earthquake Engineering, Shanghai,
People’s Republic of China
3
Department of Civil Engineering, University of British Columbia, Vancouver, Canada
4
School of Civil and Environmental Engineering, University of Science and Technology Beijing,
Beijing, People’s Republic of China

123
178 Nat Hazards (2016) 83:177–192

1 Introduction

Earthquakes are one of the most destructive natural disasters, especially when they occur in
an urban area with dense population and high volume of buildings and other civil
infrastructures. The seismic resistance of buildings has been improved significantly over
the last three decades, due to the continuous advancement in earthquake engineering
research. As a consequence, earthquake-induced building collapses and casualties on new
constructions have been effectively controlled. By contrast, the economic loss due to
earthquake is still very high. For example, after the devastating 1960 M9.5 Valdivia
earthquake in Chile, a strict building code was implemented in Chile (Guha-Sapir et al.
2011). As a result, during the 2010 M8.8 Maule earthquake in Chile, only four buildings
constructed after 1985 were collapsed (MAE Center 2010). But this earthquake still caused
a direct economic loss of US$ 30.9 billion1 which represented 24.2 % of the global
economic damages from all natural disasters in 2010 (US$ 127.8 billion) (Guha-Sapir et al.
2011). The economic loss could be even higher if the earthquake strikes other highly
urbanized regions, as it has occurred in the 2011 M9.0 Tohoku earthquake which caused
US$ 210 billion direct loss (Ponserre et al. 2012). Hence, it is crucial to develop a robust
earthquake loss prediction model for an urban area, where the information can be used by
the decision makers to make informed risk management decisions.
HAZUS (FEMA 1999, 2012a) is one of the most widely used methods for regional
earthquake loss prediction (e.g., Peterson and Small 2012; Remo and Pinter 2012). HAZUS
calculates the building response using capacity spectrum method (CSM) (Kircher et al.
2006), in which buildings are treated as a single-degree-of-freedom (SDOF) system when
subjected to a pushover load. As a result, there are three major limitations when using
HAZUS to predict regional earthquake loss: (1) Because the SDOF model used within
HAZUS cannot accurately differentiate the response at different stories and the financial
loss may vary significantly at different stories, the economic loss cannot be accurately
predicted using the HAZUS method; (2) As stated in the FEMA-445 report, the non-
structural components in the HAZUS method are rather general (FEMA 2006). The
financial loss for the nonstructural components cannot be characterized well using the
HAZUS method. (3) The influence of the ground motion characteristics (e.g., near-field
velocity pulses) on the building damage and economic loss cannot be easily considered
using the CSM (Lu et al. 2014).
Since 2002, the Federal Emergency Management Agency (FEMA) funded a 10-year
project to define the objective and procedure of next-generation performance-based seis-
mic design which eventually became the FEMA-P58 report: ‘‘Seismic Performance
Assessment of Buildings, Methodology and Implementation’’ (referred to as ‘‘the FEMA-
P58 method’’ hereafter) (FEMA 2012b, c). This report provides a solution for the above
limitations of the HAZUS method. The fragility of every structural and nonstructural
component in a building is directly considered in the FEMA-P58 method during the
seismic assessment. The FEMA-P58 method has been successfully applied to many
individual buildings (Yang et al. 2012; Shoraka et al. 2013; Yang and Murphy 2015; Yang
et al. 2014; Shome et al. 2015). The Global Earthquake Model (GEM) vulnerability
assessment guidelines (Meslem and D’Ayala 2012; D’Ayala et al. 2015) further proposed

1
In this work, the economic losses of different years are adjusted into 2011 US$ considering inflation. The
adjustment factors are calculated according to the Consumer Price Index (CPI) statistics provided by U.S.
Department of Labor Bureau of Labor Statistics or by referring to Coin News (2015). The adjustment factor
from 2010 to 2011 is 1.03.

123
Nat Hazards (2016) 83:177–192 179

the methods to develop the fragility curves and consequence functions for different
regions. However, currently there is no such application for regional predictions. The
primary challenge when using the FEMA-P58 method for regional earthquake loss pre-
dictions is the difficulty of obtaining detailed seismic responses (i.e., engineering demand
parameters, or EDPs) and collecting the structural and nonstructural component data for
every building in the region.
To address this limitation, a practicable approach for regional earthquake loss prediction
based on the FEMA-P58 method is proposed in this study. This approach allows detailed
prediction of the economic loss at each story of each building in a region. To obtain the
EDPs of each building, a series of multi-story concentrated-mass shear (MCS) models
were developed and used in the nonlinear time history analyses (THA). The building data
and the structural and nonstructural component information were obtained through field
investigation and design drawings. The values of the fragility curves were adopted from
FEMA-P58 study. It should be noted that the default values from the FEMA-P58 for
structural and nonstructural components should be modified if they are used in China.
However, the purpose of this work is to illustrate the process of applying the FEMA-P58
methodology, not to redefine the fragility data information for all buildings in China
(which will be implemented in the future study). In addition, such default values are
convenient for readers to understand the results of this study. Hence, the default values as
presented in the FEMA-P58 document were used. Two buildings in the Tsinghua
University campus located in Beijing, China, were chosen as the demonstration buildings
for the detailed implementation of the proposed earthquake loss assessment of a region.
Finally, an intensity-based earthquake loss prediction model of the entire Tsinghua
University (including 619 buildings) was performed. The outcomes of this work can be
used as a reference for future earthquake loss prediction model for large urban areas.

2 Loss prediction methodology

The fundamental principle of the FEMA-P58 loss assessment methodology is the perfor-
mance-based earthquake engineering framework supported by the Pacific Earthquake
Engineering Research Center (Cornell and Krawinkler 2000; Moehle and Deierlein 2004;
Yang et al. 2009). Three loss prediction methodologies have been proposed in FEMA-P58
document (FEMA 2012b). The intensity-based method requires an user-specified earthquake
intensity, while the scenario-based and the time-based method are based on the intensity-
based method, except that these two methods require more earthquake hazards information,
such as a specific fault relative to the building site, the ground motion attenuation relationship,
and the seismic hazard curve for the site. To obtain such information, extra seismology study
is needed, which is not the focus of this work. Therefore, the simplest method, i.e., intensity-
based assessment method, was adopted. The procedure relies on the use of conditional
probability, where the probability of the exceedance of an EDP demand is conditioned on the
probability of the earthquake intensity occurred. Similarly, the probability of incurring a
damage at different damage states is conditioned based on the occurrence of a given EDP.
Once the damage state is identified, the repair action and repair cost are calculated from a look
up table (FEMA 2012b). Because there is lack of available and suitable ground motion and the
THA is very time consuming, Yang et al. (2009) developed a practical implementation of the
above framework by utilizing a Monte Carlo approach. The proposed Monte Carlo approach
is adopted by FEMA-P58 document (FEMA 2012b) and has been implemented into many

123
180 Nat Hazards (2016) 83:177–192

earthquake loss predictions for individual buildings (Yang et al. 2012; Shoraka et al. 2013;
Yang and Murphy 2015; Yang et al. 2014). This research further expanded the above-
mentioned method to quantify the earthquake loss of an urban region.

2.1 Flowchart of the FEMA-P58 method

Figure 1 shows the flowchart of applying the FEMA-P58 method for an urban area. The
process consists of three parts: (1) assemble the building performance model; (2) analyze
the building response to determine EDPs; and (3) calculate the economic loss.
During the assembly of the building performance models, basic building data (e.g., the
number of buildings, building type, number of stories, floor area and occupancy), fragility,
and repair cost of vulnerable structural and nonstructural components on each building story
shall be collected. The components that are vulnerable to the same EDP can be categorized
into one group, referred to as the ‘‘performance group (PG).’’ The probability of the com-
ponent in each damage state is presented using fragility curves, which are developed from a
large number of statistical data, experimental data and expert opinions. Given the value of an
EDP, the probability of incurring a particular damage state of each PG can be calculated from
its fragility curves. Then, the repair cost of a PG is calculated according to the consequence
function corresponding to each damage state. Figure 2 shows the procedure to identify the
repair costs: (a) The EDP is calculated using THA with the MCS model using the ground
motion records selected; (b) The damage state is determined using the fragility curves; (c) The
repair cost is then determined according to the total quantity and unit repair cost of each PG.
Details are described in Chapter 3 and Chapter 7 of FEMA (2012b).
The EDPs including the peak floor acceleration (PFA), peak floor velocity (PFV), peak inter-
story drift ratio (IDR), and residual drift at each story are recorded from a series of THA. Using
the procedure presented in Yang et al. (2009), a statistical correlated distribution of the EDPs is
generated. Using the synthetically generated EDP matrix, the repair cost is then identified from
a look up table. It should be noted, as seen in Fig. 1, if the building is considered collapsed or
irreparable according to its collapse fragility and repair fragility curves (FEMA 2012b), the
repair cost is considered as the replacement cost of the building. Otherwise the earthquake loss
is calculated by adding up the repair cost of every component under different damage states. The
process is executed repeatedly to quantify the distribution of the repair cost of the building.

Begin The flowchart of a realization

Assemble building
performance No Yes Calculate
Is building Is building
models repair cost
collapsed repairable
(Figure 2)

No
Construct
MCS models
Yes Repair cost = i = number of Yes Fit a lognormal
Replacement cost realizations? distribution
Generate EDPs

No
i ++ End
Loop counter i = 0

Fig. 1 Flowchart of the FEMA-P58 methodology for building earthquake loss prediction

123
Nat Hazards (2016) 83:177–192 181

Fragility curves Consequence Functions

Unit Cost
Randomly Randomly
determines DS3 determines
Input Input Repair cost
Obtain Damage State
DS1 DS2 of the PG
the EDP DS2
(DS) Uncertainty, c
DS3 DS1

EDP Quantity

Fig. 2 Calculation of repair cost for a PG

Challenges Recommended procedure

The assembly of performance models Field investigation, design

of a building group drawings, etc. (Figure 4)

Rapid calculation of EDPs

of a building group
MCS models and
non-linear THA
Prediction of the collapse fragility of a
building group

Fig. 3 Challenges and recommended procedure of extending the FEMA-P58 method to a region

Assemble building performance models

Basic
• GIS investigation design drawings other information
building data

• Method A: according to building design drawings

Structural PGs • Method B: according to design drawings of neighborhood buildings with similar building data
• Method C: making estimation according to field investigation

• Buildings of major occupancies (e.g. classroom, resident, office, etc.): obtaining distribution of PGs by investigation
Nonstructural PGs
• Other buildings: according to normative quantity information provided by FEMA-P58

Fig. 4 Approach to assembling the performance models of a building group

2.2 Challenges and recommended procedure to extend the FEMA-P58

method to a region

The key challenges of using the FEMA-P58 method to a region are: (1) the assembly of
performance models, (2) rapid calculation of EDPs, and (3) prediction of the collapse
fragility of a building group. Figure 3 shows the recommended procedure.
(1) Assembly of the performance models.
The performance model of buildings contains the basic information with both the
structural and nonstructural PGs. The information can be collected through the procedure
as shown in Fig. 4. The basic building data can be obtained from a geographic information

123
182 Nat Hazards (2016) 83:177–192

system (GIS) database, field investigation, or design drawings. Generally, such information
is easy to collect if a GIS database of the target area is available.
If the exact replacement value of the building is missing, the replacement value of the
building can be determined by summing the average repair cost of each PG at its most
severe damage state. Such procedure might be a simple yet practical approximation. To
validate the reliability of the predicted replacement value, the replacement value of the
case study region in this work is also estimated using the method proposed in the HAZUS
technical manual (FEMA 2013). The replacement values predicted by the HAZUS method
and the sum of the median replacement costs of components are US$ 6.905 billion and
7.476 billion, respectively. The HAZUS method predicted a slightly smaller total
replacement value, but the difference is not significant and both predictions are accept-
able in this case study considering the random nature of the replacement value. In addition,
the sum of the median replacement costs of components is more adaptive because it
explicitly considers the values of individual components and contents.
Three recommended procedures have been proposed to determine the structural PGs.
(a) Method A: If the structural and architectural drawings of the building are available,
the type and quantity of each PG can be obtained from these drawings directly. In some
urban region, where similar buildings are constructed at the same period, using the same
structural design codes, the information obtained from one building can be duplicated to
the other buildings in the region.
Once the PG information has been identified, the repair cost for the PG can be defined
using the fragility data published by the FEMA-P58 document. It should be noted that as a
part of the 10-year effort for the FEMA-P58 project, more than 700 fragility curves and
associated consequence functions have been identified using a large number of statistical
data, experimental data and expert opinions. The structural performance groups are defined
based on the material, size and behavior of structural components. Each PG is used to
describe one type of structural components. For example, ‘‘B1041.041b’’ (one of the PGs
used in this study) is applicable to the reinforced concrete moment frames with 24 in. 9 24
in. beam size (FEMA 2012c). Details of PG clarification can be found in Appendix A of
FEMA (2012c).
(b) Method B: If structural and architectural drawings of the target building are not
available, but there is a similar building (with similar year of construction, structural type
and occupancy) in the neighborhood, the quantity of the structural PGs of the target
building can be estimated by
Q0
Q¼A ð1Þ
A0
where A, A0 and Q, Q0 are the floor area and quantity of the PG of the target building and
the neighborhood building with actual structural and architectural data, respectively.
(c) Method C: For buildings where neither the design information nor similar building
information is available, the PG information can be estimated according to the field
investigation of the buildings.
To determine the nonstructural PGs information, the buildings are categorized into
groups according to building occupancies. For example, the building inventory for a region
can be characterized as office buildings (e.g., 30 %), residential buildings (e.g., 50 %) and
industrial buildings (e.g., 20 %). In each group of occupancy (e.g., office buildings), the
distribution of the nonstructural PGs can be identified from a detailed site visit of several
typical buildings. Other PGs which are difficult to estimate (such as pipelines) can be

123
Nat Hazards (2016) 83:177–192 183

identified according to the normative quantity information provided by Appendix F of

FEMA (2012b). For example, the sanitary waste piping in an office building is assumed to
be 0.057 feet per 1 gross square foot according to Appendix F of FEMA (2012b), so the
length of sanitary waste piping in an office building with a floor area of 1000 m2 is 187 m
at each floor. For other specialized buildings, the nonstructural PGs can be estimated
according to Appendix F of FEMA (2012b).
Note that the above approaches of assembling performance models require some efforts.
However, it can be implemented in parallel by groups of people with basic knowledge of
architectural and structural engineering. For the case study of Tsinghua University campus
(as presented in this work), the data collection of the performance model of all 619
buildings were established by 11 students in 4 weeks. As for the assessment of an entire
city, building information model (BIM) can be used to systematically store and identify the
building PGs.
(2) Rapid calculation of the EDPs of a building group.
Another difficulty in extending the FEMA-P58 method from an individual building to a
region is to obtain the EDP distribution in an efficient manner. For an urban area consisting
of thousands of buildings, the establishment of refined computational models is extremely
time consuming. It requires a professional understanding of structural engineering, such as
the finite element modeling of beams, columns and shear walls, to develop the refined
computational model. Even if the refined computational model of a building group can be
efficiently established, the resulting computational workload is still extremely expensive
(Sobhaninejad et al. 2011). To overcome these difficulties, the multi-story concentrated-
mass shear (MCS) models (Fig. 5a) proposed by Lu et al. (2014) were adopted in this
study. MCS model is a common numerical model to simulate the nonlinear response of
multi-story buildings. MCS model assumes that:
(a) A multi-story building can be simplified to a multiple-degree-of-freedom (MDOF)
model, where each story has its nonlinearity;
(b) The nonlinear response of the building is assumed to be dominated by the shear
mode;
(c) Depending on the structural system, the nonlinearity of the structure at each story
can be modeled using either modified-clough, bilinear elasto-plastic or pinching
model.

(a) (b)

Fig. 5 Multi-story concentrated-mass shear (MCS) models proposed by Lu et al. (2014): a illustration,
b validation (Xu et al. 2014)

123
184 Nat Hazards (2016) 83:177–192

MCS models are suitable for regional seismic response analysis (Lu et al. 2014; Xu
et al. 2016). Lu et al. (2014) proposed a modeling approach, by which all of the parameters
of MCS models can be determined by either refined finite element (FE) models or the basic
information of the buildings (i.e., structural type, number of stories, occupancy, con-
struction year, and floor area). The MCS model was compared with a refined FE model by
Xu et al. (2014). For a six-story reinforced concrete (RC) frame, the top displacement
versus time histories predicted by the FE model and the MCS model agree well with each
other (Fig. 5b). Based on the MCS models and nonlinear THA, Xu et al. (2014) performed
a regional seismic response analysis for the city of Shantou, China. It took \10 min to
complete the nonlinear THA of one ground motion for the region with 7449 buildings on a
desktop computer (Intel 2.8 GHz i5 CPU; 4-GB memory), which demonstrated the effi-
ciency of using the MCS model. Other simplified MDOF models, for example the fishbone
model (Nakashima et al. 2002; Luco et al. 2003), can also be used if the model parameters
of a large number of buildings can be conveniently determined.
(3) Prediction of the collapse fragility of a building group.
As indicated in the FEMA-P58 methodology, the collapse fragility curves of buildings
are needed for the loss calculation. The response analysis of buildings adopted in this work
based on the MCS model and nonlinear THA can estimate whether a building will collapse
when subjected to a particular ground motion (Xu et al. 2014). Thus, the collapse fragility
curves can be obtained directly by conducting an incremental dynamic analysis (IDA)
using the proposed MCS model for each building.

3 Case study: regional earthquake loss prediction of the Tsinghua

University campus

3.1 Introduction of the case study region

The campus of Tsinghua University, consisting of 619 buildings, has a gross area of
approximately 4 km2. According to the Chinese Code for Seismic Design of Buildings (CMC
2010), the buildings in Tsinghua University have a seismic design intensity of VIII (with the

(a) (b)
Shear Others
Steel Office
wall RC 9%
0% 18%
12% frame
Research
15% Retail 9%
1%
Frame- Classroom
shear 6%
wall
18%
Masonry
55%
Residential
57%

Fig. 6 Percentages of a different structural types and b building occupancies

123
Nat Hazards (2016) 83:177–192 185

peak ground acceleration (PGA) of 0.2 g at the 10 % probability of exceedance in 50 years

hazard level). According to the geological investigation, the site condition of the campus is
class II in Chinese code, which approximately corresponds to site class C and D as presented
in the ASCE 7-10 document in the United States (Luo and Wang 2012; ASCE 2010).
Figure 6 shows the percentage of different structural types and building occupancies. In
terms of structural type, more than half of the buildings are masonry structures. Other
major structural types are RC frames, RC shear walls, and RC frame-shear wall dual-
systems. In terms of occupancy, residential buildings comprise more than 50 %, and other
major occupancies are offices, research facilities and classrooms.

3.2 Assembly of the building performance models

Eleven students were assigned to collect the structural drawings and perform the field
investigations. Table 1 shows the building inventory collected using the three methods
outlined in Fig. 4.
A RC frame office building (named RC_Office) and a masonry residential building
(named Mas_Residence) were selected as the two demonstration examples for the seismic
performance assessment using the proposed FEMA-P58 approach. Table 2 shows the basic
facts about the two example buildings selected.

3.3 Ground motion selection and scaling

The following steps are proposed in FEMA-P58 for ground motion selection and scaling:
(1) Select a target spectrum that is suitable for the site (typically use site specific hazard
analysis); (2) Select ground motions with spectral shapes that are similar to the target
spectrum over a period range of interest; (3) Amplitude-scale the selected ground motions
such that the spectra will match the target spectrum. In this work, some simplifications
were adopted for the ground motion selection and scaling:
1. Because the latest strong earthquake happened in Beijing is M8.0 Sanhe-Pinggu
earthquake in 1679, more than 200 years before strong motion seismograph was
developed, there are few strong motion records in Beijing. As a result, 50 pairs of the
horizontal components of the ground motion records for site classes C and D proposed
by FEMA P695 (FEMA 2009) are adopted in this work. This includes the 22 pairs of

Table 1 Results of the building investigation according to the completeness of the structural data
Classification Building Percentage of the Percentage of the total Comments
quantity total number of replacement value of
campus buildings campus buildings (%)
(%)

Method A 497 80.3 88.8 Design drawings are

accessible for the
majority of the important
buildings
Method B 64 10.3 9.8 Mainly dormitory, office or
for temporary use
Method C 58 9.4 1.4 Mainly old buildings for
temporary use

123
186 Nat Hazards (2016) 83:177–192

Table 2 Basic data for the two example buildings

Label RC_Office Mas_Residence

Name Dept. Civil Engineering Residential quarter building 5

Number of stories 4 6
Floor area/m2 630 434
Structural type and occupancy RC frame, office Masonry, residential
Replacement value/$ million 6.85 2.51
Construction year 1995 1991

ground motions from the far-field records, 14 pairs of ground motions from the near-
field records (with pulse), and 14 pairs of ground motions from the near-field records
(without pulse). These ground motions are adopted because (a) their site conditions are
similar to those of the case study region (i.e., site classes C and D); (b) these ground
motions have been widely used in related studies (Lu et al. 2013, 2014; Shi et al.
2014), so these results can be easily compared to other studies; and (c) the main
purpose of this work is to demonstrate the feasibility of using the FEMA-P58 method
to assess the seismic performance of a region, rather than an actual engineering
application, so these ground motions are selected.
2. FEMA-P58 suggests using Sa(T1) as the intensity measure. However, in this work,
hundreds of buildings are analyzed simultaneously in the THA. T1 varies between
buildings, it will be inappropriate to analyze the entire region using one Sa(T1) value.
Using various Sa(T1) may result in significantly different ground motion scaling factors
among neighborhood buildings, which is also unreasonable. Moreover, PGA is the
seismic intensity measure adopted in the Chinese Code for Seismic Design of
Buildings (CMC 2010) for the design of all kinds of structures. Hence, PGA is selected
to quantify the shaking intensity.
3. Because the case study (Tsinghua University) is not overly a large area where the site
condition is significantly different, the same set of ground motion is used for each
building. For other study, where the area is large or the site condition is complex, more
detailed sets of ground motions should be used for each building.
Three earthquake shaking intensities namely the service-level earthquake (SLE), design
basis earthquake (DBE), and maximum considered earthquake (MCE) were included in
this study. The corresponding PGAs and return periods are shown in Table 3 according to
Section 5.1.2 and Table 5.1.2-2 in CMC (2010).

Table 3 Return periods and the corresponding PGAs

Earthquake level Return period PGA (g)
(year)

Service-level earthquake (SLE) 50 0.07

Design basis earthquake (DBE) 475 0.2
Maximum considered earthquake (MCE) 2475 0.4

123
Nat Hazards (2016) 83:177–192 187

Fig. 7 Comparison of the loss 1

SLE
results of RC_Office predicted by
the proposed method and by 0.9 DBE
PACT software at the hazard

Probability of non-exceedance
0.8
levels of SLE, DBE, and MCE MCE
0.7

0.6

0.5 This work

PACT
0.4

0.3

0.2

0.1

0
0.0% 20.0% 40.0% 60.0% 80.0% 100.0%
Economic loss ratio

4 Results and discussion on loss prediction

4.1 Validations

Due to the lack of credible and accurate studies of regional seismic loss in the target area,
the loss results in this study are not able to be compared with other studies. Instead, the
earthquake loss results of typical buildings are compared with those calculated using
PACT [an earthquake loss simulation software developed by FEMA to implement FEMA-
P58 method (FEMA 2012b)] for validation.
Taking RC_Office as an example, the loss was normalized based on the building
replacement value. Figure 7 shows the comparison of the earthquake loss from current
study and the results obtained from the PACT software. Each curve in Fig. 7 consists of
1000 points generated by 1000 times of the Monte Carlo approach. For clarity, only a few
points on the curves are marked. The points on the vertical line at the right end of the
curves stand for the realizations where the building is predicted to collapse or damage
irreparably, viz., the loss ratio is 100 %. As shown in this figure, the loss simulation results
are very similar.

4.2 Earthquake loss predictions of the two example buildings

Figure 8 shows the results of the loss simulation of the two example buildings. At the SLE
hazard level, the RC_Office building has minor losses. There is nearly no loss contributed
from the structural (Str) PGs as all components remain elastic. By contrast, there are some
damages related to the drift-sensitive nonstructural (NSD) PGs. This is mainly due to the
repair cost of the partitions and the wall finishes. When the shaking intensity reaches the
DBE hazard level, the loss associated with the structural and acceleration-sensitive non-
structural (NSA) PGs increases. By contrast, the loss for the Mas_Residence is higher than
the RC_Office building, this shows the masonry structures are more sensitive to defor-
mation. The proposed approach also allows the users to identify the distribution of the
structural response among different stories. For example, Fig. 9 shows the distribution of
peak IDR and PFA of RC_Office at the MCE hazard level.

123
188 Nat Hazards (2016) 83:177–192

(a) (b)
4.5% 70.0%
4.0% NSA NSA
60.0%

Economic loss ratio

Economic loss ratio

3.5% NSD NSD

50.0%
3.0% Str Str
2.5% 40.0%
2.0% 30.0%
1.5%
20.0%
1.0%
0.5% 10.0%

0.0% 0.0%
SLE DBE MCE SLE DBE MCE
Hazard level Hazard level

Fig. 8 Loss result of a RC_Office and b Mas_Residence under the three hazard levels considered

(a) (b)
4 4
Median Median

Individual Individual
Ground Motion Ground Motion

3 3
Story #

Story #

2 2

1 1
0 0.005 0.01 0.015 0.02 0.025 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Peak IDR PFA

Fig. 9 Distribution of a peak IDR and b PFA of RC_Office at the MCE hazard level

Figure 10 shows the detailed loss of nonstructural PGs at each story of the RC_Office
building. The result shows that a high percentage of the repair costs come from the
nonstructural walls (including exterior walls, partitions, and wall finishes). The walls start
to crack and contribute to the repair costs even when the peak IDR is small. The repair
costs of the teleconference equipment on the third story and the computers on the fourth
story are noticeable because the teleconference equipment is expensive and the quantity of
computers on the fourth story is considerably larger than on other stories. Therefore, the
loss prediction clearly represents the property distribution characteristics within the
RC_Office building.

4.3 Earthquake loss prediction of the entire campus

Figure 11a shows the earthquake loss prediction of the entire campus under three levels of
earthquake shaking intensities. The total earthquake loss ratio is defined as the ratio of total

123
Nat Hazards (2016) 83:177–192 189

(a) (b)
0.05% 0.6% 4th story
4th story
0.5% 3rd story
0.04% 3rd story
Loss ratio

2nd story

Loss ratio
0.03% 2nd story 0.4%
1st story
1st story 0.3%
0.02%
0.01% 0.2%

0.00% 0.1%
0.0%

(c)
0.6% 4th story
0.5% 3rd story
Loss ratio

0.4% 2nd story

0.3% 1st story
0.2%
0.1%
0.0%

Fig. 10 Loss details of nonstructural PGs at each story of RC_Office under the hazard levels of a SLE,
b DBE and c MCE

(a) (b)
Median total economic loss ratio

40%
Probability of non-exceedance

1
Repair cost
Irreparable deformation
0.75 30% Collapse

SLE
0.5 DBE 20%
MCE

0.25
10%

0
0% 10% 20% 30% 40% 50% 0%
SLE DBE MCE
Total earthquake loss ratio
PGA

Fig. 11 Loss results of the entire campus under the three hazard levels considered: a cumulative probability
of the total earthquake loss ratio, b distribution of the total earthquake loss ratio

earthquake loss to the total replacement value (US$ 7.476 billion) of the region. The
median total loss ratio are 1.3, 13.7 and 34.9 % for the SLE, DBE and MCE hazard level,
respectively. The median loss is further classified into the losses due to building collapse,
irreparable deformation, and repair cost. Figure 11b shows the distribution of the median
repair cost. At the SLE hazard level, the economy loss is mainly concentrated from repair

123
190 Nat Hazards (2016) 83:177–192

SLE DBE MCE

Median total economic loss ratio

50%

40%

30%
47.6%
20%
29.6%
10% 18.8%
1.5% 1.9% 12.8%
0%
No Pulse Pulse No Pulse Pulse No Pulse Pulse

Fig. 12 Influence of velocity pulse on the economic loss of the case study region detected by the proposed
method

costs. When the earthquake intensity increases, the loss starts to contribute from both repair
costs and irreparable deformation. As the shaking intensity reaches the MCE hazard level,
the contribution from the collapse-induced loss is introduced but the percentage is very low
(\2 %), which is similar to the loss results observed in the 2010 Chile earthquake (MAE
Center 2010) and 2011 Christchurch earthquake (Smyrou et al. 2011).
In addition to the loss assessment from the 50 pairs of ground motions, a side study
about the influence of the ground motion velocity pulse on the building performance and
earthquake loss is also studied in this work. The results from the loss simulation were
compared between the 14 pairs of near-field records with pulse and the 14 pairs of near-
field records without pulse. As shown in Fig. 12, the economic loss increases when the
region is subjected to pulse type motion. Therefore, the proposed approach is able to
account for the influence of ground motion characteristics, such as velocity pulse, on
building earthquake response and loss prediction. By contrast, in the HAZUS method,
building response is determined by the intersection of the building capacity curve and the
earthquake response spectrum (Kircher et al. 2006; Xiong et al. 2016), known as the
capacity spectrum method (CSM) (ATC 1996). According to the discussion by Krawinkler
and Seneviratna (1998), due to the fact that CSM is based on static loading, it cannot
represent dynamic phenomena with high accuracy. As a result, the proposed approach
improves one of the limitations from the HAZUS simulation approach.

5 Conclusions

With the increasing improvement in the building codes, earthquake-induced building col-
lapses and casualties have been well controlled. However, hefty financial losses due to
earthquake are still impacting many major cities in high seismic zones. To effective quantify
the earthquake loss of a major city center, the FEMA-P58 method was used. To efficiently
analyze the seismic performance of a region, simplified MCS models were developed and
used with nonlinear THA to quantify the seismic response. The intensity-based earthquake
loss prediction was implemented in Tsinghua University to demonstrate the implementation
and advantage of the proposed approach. The following conclusions are obtained:
1. MCS models can be used to calculate the EDPs for each building at each story with
high efficiency, which solves one of the key challenges of the extension of FEMA-P58
method from an individual building to a region.

123
Nat Hazards (2016) 83:177–192 191

2. For the case study, when the shaking intensity reaches the SLE hazard level, the total
earthquake loss mainly contributes from the repair costs. On the other hand, when the
hazard level increases to the MCE, the total earthquake loss mainly contributes from
the repair cost and the cost of the demolition and reconstruction of irreparable
buildings. The percentage of loss caused by building collapse is low.
3. The proposed approach can be used to study the influence of ground motion
characteristics, such as velocity pulse, on building earthquake response and loss, which
improves one of the limitations of the HAZUS method.
Note that the purpose of this work is to propose a practical approach for regional
earthquake loss prediction to take advantage of the benefits of next-generation perfor-
mance-based seismic design instead of conducting a precise prediction for Tsinghua
University campus. The accuracy of the loss result relies on the quality of the data. The
fragility curves, consequence functions and normative quantities used in this work are the
proposed default values in FEMA-P58. Due to the differences between China and the USA,
such default data may not be sufficiently accurate if they are used for Chinese buildings. In
the future, these data will be further carefully determined to represent the characteristics of
the region to be analyzed. Thus, the proposed loss prediction approach will have a higher
accuracy and a more significant contribution to disaster prevention in cities.

Acknowledgments The authors are grateful for the help from Runhua Gong, Qiuhan Huang, Huiping Li,
Jian Liu, Shixuan Liu, Yizhe Meng, Yao Ming, Jian Yang, and Zhebiao Yang in the investigation and
collection of basic building data, building design drawings, and property distribution, which forms the data
basis of this work. The authors are also grateful for the financial support received from the National Natural
Science Foundation of China (Nos. 51578320, 51378299), the National Key Technology R&D Program
(No. 2015BAK14B02), and the National Non-profit Institute Research Grant of IGP-CEA (Grant No:
DQJB14C01).

References
American Society of Civil Engineers (ASCE) (2010) Minimum design loads for buildings and other
structures, ASCE/SEI 7-10, Reston, VA
ATC (1996) Seismic evaluation and retrofit of concrete buildings (ATC-40). Applied Technology Council,
Redwood City
China Ministry of Construction (CMC) (2010) Code for seismic design of buildings, GB50011-2010. China
Architecture and Building Press, Beijing (in Chinese)
Coin News (2015) US Inflation Calculator. http://www.usinflationcalculator.com/
Cornell CA, Krawinkler H (2000) Progress and challenges in seismic performance assessment. PEER Center
News 3(2):1–3
D’Ayala D, Meslem A, Vamvatsikos D, Porter K, Rossetto T, Crowley H, Silva V (2015) Guidelines for
analytical vulnerability assessment of low/mid-rise buildings, GEM technical report 2015-08 V1.0.0
Federal Emergency Management Agency (FEMA) (1999) Earthquake loss estimation methodology—
Hazus99, Technical Manual, Washington, DC
Federal Emergency Management Agency (FEMA) (2006) Next-generation performance-based seismic
design guidelines program plan for new and existing buildings. Technical report FEMA-445, Wash-
ington DC
Federal Emergency Management Agency (FEMA) (2009) Quantification of building seismic performance
factors. Technical report FEMA-P695, Washington DC
Federal Emergency Management Agency (FEMA) (2012a) Multi-Hazard loss estimation methodology
Hazus-MH 2.1 Advanced Engineering Building Module (AEBM) Technical and User’s Manual.
Washington, DC
Federal Emergency Management Agency (FEMA) (2012b) Seismic performance assessment of buildings
volume 1—methodology, Technical report FEMA-P58, Washington, DC

123
192 Nat Hazards (2016) 83:177–192

Federal Emergency Management Agency (FEMA) (2012c) Seismic performance assessment of buildings
volume 2—implementation guide, Technical report FEMA-P58, Washington, DC
Federal Emergency Management Agency (FEMA) (2013) Multi-hazard loss estimation methodology—
earthquake model, Hazus-MH 2.1, Technical Manual, Washington, DC
Guha-Sapir D, Vos F, Below R, Ponserre S (2011) Annual disaster statistical review 2010: the numbers and
trends. Centre for Research on the Epidemiology of Disasters (CRED), Brussels
Kircher CA, Whitman RV, Holmes WT (2006) Hazus earthquake loss estimation methods. Nat Hazards Rev
7(2):45–59
Krawinkler H, Seneviratna GDPK (1998) Pros and cons of a pushover analysis of seismic performance
evaluation. Eng Struct 20(4):452–464
Lu X, Ye LP, Lu XZ, Li MK, Ma XW (2013) An improved ground motion intensity measure for super high-
rise buildings. Sci China Technol Sci 56(6):1525–1533
Lu XZ, Han B, Hori M, Xiong C, Xu Z (2014) A coarse-grained parallel approach for seismic damage
simulations of urban areas based on refined models and GPU/CPU cooperative computing. Adv Eng
Softw 70:90–103
Luco N, Mori Y, Funahashi Y, Cornell CA, Nakashima M (2003) Evaluation of predictors of non-linear
seismic demands using ‘fishbone’ models of SMRF buildings. Earthq Eng Struct D 32(14):2267–2288
Luo KH, Wang YY (2012) Researches about the conversion relationships among the parameters of ground
motions in the seismic design codes of China, America and Europe. In: Proceedings of the 15th world
conference on earthquake engineering, Lisbon, Portugal
Meslem A, D’Ayala D (2012) Toward worldwide guidelines for the development of analytical vulnerability
functions and fragility curves at regional level. In: Proceedings of the 15th world conference on
earthquake engineering, Lisbon, Portugal
Mid-America Earthquake Center (MAE Center) (2010) The Maule (Chile) earthquake of February 27, 2010:
consequence assessment and case studies. Report No. 10-04, Urbana, IL
Moehle J, Deierlein GG (2004) A framework methodology for performance-based earthquake engineering,
paper no. 679. In: Proceedings of the 13th world conference on earthquake engineering, Vancouver,
BC, Canada
Nakashima M, Ogawa K, Inoue K (2002) Generic frame model for simulation of earthquake responses of
steel moment frames. Earthq Eng Struct D 31(3):671–692
Peterson J, Small MJ (2012) Methodology for benefit-cost analysis of seismic codes. Nat Hazards
63:1039–1053
Ponserre S, Guha-Sapir D, Vos F, Below R (2012) Annual disaster statistical review 2011: the numbers and
trends. Centre for Research on the Epidemiology of Disasters (CRED), Brussels
Remo JW, Pinter N (2012) Hazus-MH earthquake modeling in the central USA. Nat Hazards 63:1055–1081
Shi W, Lu XZ, Guan H, Ye LP (2014) Development of seismic collapse capacity spectra and parametric
study. Adv Struct Eng 17(9):1241–1256
Shome N, Jayaram N, Krawinkler H, Rahnama M (2015) Loss estimation of tall buildings designed for the
peer tall building initiative project. Earthq Spectra 31(3):1309–1336
Shoraka MB, Yang TY, Elwood KJ (2013) Seismic loss estimation of non-ductile reinforced concrete
buildings. Earthq Eng Struct D 42(2):297–310
Smyrou E, Tasiopoulou P, Bal HE, Gazetas G (2011) Ground motions versus geotechnical and structural
damage in the February 2011 Christchurch earthquake. Seismol Res Lett 82(6):882–892
Sobhaninejad G, Hori M, Kabeyasawa T (2011) Enhancing integrated earthquake simulation with high
performance computing. Adv Eng Softw 42(5SI):286–292
Xiong C, Lu XZ, Guan H, Xu Z (2016) A nonlinear computational model for regional seismic simulation of
tall buildings. Bull Earthq Eng 14(4):1047–1069
Xu Z, Lu XZ, Guan H, Han B, Ren AZ (2014) Seismic damage simulation in urban areas based on a high-
fidelity structural model and a physics engine. Nat Hazards 71(3):1679–1693
Xu Z, Lu XZ, Guan H, Tian Y, Ren AZ (2016) Simulation of earthquake-induced hazards of falling exterior
non-structural components and its application to emergency shelter design. Nat Hazards 80(2):935–950
Yang TY, Murphy M (2015) Performance evaluation of seismic force resisting systems for low-rise steel
buildings in Canada. Earthq Spectra 31(4):1969–1990
Yang TY, Moehle JP, Stojadinovic B, Der Kiureghian A (2009) Seismic performance evaluation of facil-
ities: methodology and implementation. J Struct Eng-ASCE 135(10):1146–1154
Yang TY, Moehle JP, Bozorgnia Y, Zareian F, Wallace JW (2012) Performance assessment of tall concrete
core-wall building designed using two alternative approaches. Earthq Eng Struct D 41(11):1515–1531
Yang TY, Atkinson JC, Tobber L (2014) Detailed seismic performance assessment of high-value contents
laboratory facility. Earthq Spectra. doi:10.1193/092313EQS259M

123