Chapter 1

Surrealism in Facing the Earthquake Risk

Mete A. Sözen

Abstract The possibility of a violent ground motion in a population center poses

multiple threats to the safety and continuity of society. It has to be met on
multiple planes including the political and the economical. It is not an exaggera-
tion to claim that the construction trades in many parts of the world understood
the threat only in twentieth century. In many towns and cities in seismic zones, a
strong ground motion can destroy 10 % or more of the existing buildings. To
locate the vulnerable, it is necessary to investigate all. If this study is to be done
in detail using codified criteria, it may involve an investment that would be
unacceptable to a political system that seldom appreciates the risk. That condition
constrains the engineering effort to the minimal and requires a procedure that has
to be simple and transparent. Above all, data acquisition needs to be within the
reach of workers without technical degrees. The paper investigates such a proce-
dure, the Hassan Index, using available information from five earthquake events
and finds it imperfect but useful.

1.1 Introductory Remarks

In planning for a battle in 1921 in which his chances of success were slim, Mustafa
Kemal Atatürk made a declaration that may be translated freely as, “Our defense is
not to be limited to a line. It is to include the entire country.”1 In two short sentences
he was able to define the multi-dimensional nature of his defense strategy.
Defense of a region against the earthquake threat also involves more than one
dimension. The threat may include any or even all of (a) unpredictable ground

1
In Turkish, his statement was limited to six words.
M.A. Sözen (*)
School of Civil Engineering, Purdue University, 550 Stadium Mall Drive,
West Lafayette, IN 47907, USA
e-mail: [email protected]

A. Ilki and M.N. Fardis (eds.), Seismic Evaluation and Rehabilitation 1

of Structures, Geotechnical, Geological and Earthquake Engineering 26,
DOI 10.1007/978-3-319-00458-7_1, © Springer International Publishing Switzerland 2014
2 M.A. Sözen

motion, (b) disastrous landslides, (c) soil liquefaction, (d) an overwhelming tsunami
followed by fire, and (e) social unrest. To boot, that what needs to be defended
includes building structures of all types and ages, bridges, tunnels, roads, food-
distribution system, health resources, and the communication infrastructure, all
constructed under different requirements at different times. For a planner to focus
on only one of these is, in a sense, declaration of defeat before engaging the threat.
Having said that, we continue with a limited objective, one aimed at safety of
reinforced concrete building structures having seven or fewer stories. The focus is
on planning the structural-risk assessment to be done before the earthquake occurs.
Proper assessment of the structural risk is not easy to define, because it is subject
to the judgments of many professional groups. It has been said that the first material
that needs to be purchased in starting construction is money. Thorstein Veblen has
placed engineering between physics and banking. From the viewpoint of pure
engineering concerns, “proper” is not difficult to define. But if the criteria are
also controlled by banking interests, it becomes very difficult to define them,
especially with respect to a threat that, unlike gravity, occurs infrequently.
In intellectual activities related to earthquake-risk assessment, the needed
actions go against the grain of conventional engineering thought processes that
begin with needs interpreted in terms of forces and space to produce a safe,
serviceable, and durable artifact. The engineer’s concern is to build the right
thing in the right way. That, unfortunately, is not necessarily the way to success
in earthquake-risk assessment. If the goal is to reduce the risk of severe damage or
the time of recovery, there has to be a vision shared by all parties involved in the
decision process. An unforgiving requirement of minimum safety, or an upper-
bound estimate of the ground-motion demand is very likely to lead to inaction.
Because assessment involves large inventories, it is better if it is rule-based and
best if the rules are simple. In the following section, we discuss a simple approach
to assessment that is flexible enough to be modified to fit local engineering and
banking goals.
For assessment of vulnerability to strong ground motion of individual structures,
engineering seeks safety. The evaluation process requires robust data and detailed
analysis. Only those structures that are unquestionably safe according to the driving
criteria pass muster. The process may be called a high-pass filter. Its goal is to
separate the safe from the questionable.
For risk assessment of earthquake vulnerability of building inventories, the
constraints demand a different process. Large-scope decisions must be made
based on inexpensively acquired data and evaluation algorithms that can be
implemented rapidly. The goal is to separate the definitely vulnerable from the
rest. This process may be deemed to be a low-pass filter. It is driven by the fear of
failure rather than desire for success.
In regions with construction built without earthquake-related codes and
construction control, the low-pass process is likely to be safer because explicit
recognition of substantial need tends to lead to inaction.
The low-pass filter needs to be as simple as it can be. And it needs to have
transparent rules for data acquisition that do not require an advanced degree in
1 Surrealism in Facing the Earthquake Risk 3

engineering to implement them. In fact, acquisition of the input data should be easy
enough to be obtained by high school graduates after a week’s training. It should be
to engineering analysis what surrealist painting is to classical art. Even if the design
of its overall concept requires talent, training and keen perception, it should be
reproducible by almost any interested person.
Such a procedure is available in the Hassan Index (Hassan and Sözen 1994), a
simple method that was inspired by the wall index proposed by Shiga et al. (1968).
The Hassan index uses two indices to rank the relative safety of the structure. To
organize data on damage states from earthquake events in different regions built at
different times subjected to different earthquakes, and evaluated by different
engineers using only two numbers, the column index and the wall index, in order
to estimate vulnerability of existing buildings to earthquakes that have not yet
happened, does appear to be something that cannot be done by anyone born of
woman.2 Within the scope of this paper, that hypothesis will be tested.

1.2 The Hassan Index

The Hassan Index has its roots in a brilliant insight contained in (Shiga et al. 1968).
After studying the aftermath of the Tokachi-Oki Earthquake of 1968 in Northern
Japan, they observed that buildings with reinforced concrete walls satisfying the
following two simple criteria had survived the demands of the ground motion with
light damage,

W
P P  1:2 MPa (1.1)
Ac þ Aw

and
P
A
P w  0:3 % (1.2)
Af

where:
W: Total weight of the building above base
∑Ac: Total cross-sectional area of columns at base
∑Aw: Total cross-sectional area of reinforced concrete walls in a given plan
direction, continued over the building height, at base
∑Af: Sum of tributary floor areas above base
This simple and direct procedure turned out to be difficult to apply for evaluating
the effects on buildings of the Erzincan 1992 earthquake, primarily because most of

2
William Shakespeare, “Macbeth,” Act 5, Scene 3.
4 M.A. Sözen

the walls were masonry infill, not reinforced concrete, and the primary element of
resistance was the column. A different set of measures were proposed: the Column
Index (CI) and the Wall Index (WI), defined as:
P
A
CI ¼ P c  100 (1.3)
2  Af

and
P
P Amw
Awc þ
WI ¼ P 10  100 (1.4)
Af

where:
∑Aw: Sum of cross-sectional areas of reinforced concrete walls in a given plan
direction
∑Amw: Sum of cross-sectional areas of masonry infill walls in a given plan direction
Before getting into the data it is important to mention that the reported damage
state of a building is not exactly a measurement that can be obtained with
reasonable confidence, such as the weight of a concrete cylinder. The reported
damage state depends usually on three factors: The first is the ground motion.
Even if two identical buildings are subjected to the same earthquake event, and
even if they are in the same district of the same town, they may be subjected to
ground motions that may be different. The second factor is the toughness and
strength of the structure. Geometrically identical structures, that would have
similar CI and WI indices, may have different defects. An oversight in the
detailing, or the presence of the well known captive column, in one and not in
the other may make all the difference between success and failure. The third
factor is likely not to affect evaluations of no damage, light damage, and collapse.
But depending on the experiences and attitudes of two different observers, a
damaged building may be classified as moderately damaged when it is severely
damaged and vice versa.
In the next section CI and WI indices obtained after five different
earthquakes are reported. The earthquakes occurred in Erzincan 1992, Düzce
1999, Bingöl 2003, Wenchuan 2008, and Haiti 2010. It must be emphasized that
(1) the damage statistics included only a fraction of the buildings at each site,
and (2) structural and architectural properties of the buildings varied from site
to site, as did the earthquake demand. The Wenchuan data were limited to a
university campus. No collapses occurred. Collapses occurred in Haiti, but they
were not reported, because of lack of information on the buildings that
collapsed.
Earthquake magnitudes, Mw, and their approximate epicentral distances to the
site of the damage are listed in Table 1.1.
1 Surrealism in Facing the Earthquake Risk 5

Table 1.1 Earthquake magnitudes and their approximate epicentral distances to the site of the
damage
Approx. distance
Earthquake Magnitude, Mw to epicenter, km
Erzincan 1992 6.7 17
Düzce 1999 7.2 10
Bingöl 2003 6.4 15
Wenchuan 2008 7.9 70
Haiti 2010 7.0 25

Fig. 1.1 Data from Erzincan 1992

1.3 Evaluation of Recorded CI and WI Indices

Figure 1.1 contains data obtained by investigation teams from Middle East Technical
University, Ankara, after the Erzincan earthquake of 1992 reported in http://www.
anatolianquake.org. The diagonal lines in the plot, Boundaries 1 and 2, represent
priority indices, PI, determined after the data were plotted. Index PI is defined as

PI ¼ CI þ WI (1.5)

such that Boundary 1 defines the locus of points with CI + WI ¼ 0.25 and
Boundary 2 defines the locus of points with CI + WI ¼ 0.5.
Data available from five earthquake events are summarized in terms of the
column and wall indices in Figs. 1.1, 1.2, 1.3, 1.4 and 1.5. Before discussing the
data from the earthquakes, we should recognize that not all the data from a single
6 M.A. Sözen

Fig. 1.2 Data from Düzce 1999

Fig. 1.3 Data from Bingöl 2001

1 Surrealism in Facing the Earthquake Risk 7

Fig. 1.4 Data from Wenchuan 2008

Fig. 1.5 Data from Haiti 2010

8 M.A. Sözen

Table 1.2 Cases of light, moderate, severe damage and collapse within Bound 1
Erzincan Düzce Bingöl Wenchuan Haiti
1992 2003 1999 2008 2010
Total cases documented with Light 28 125 11 66 130
Damage
Cases with light damage included within 2 30 7 43 43
bound 1
Total cases documented with moderate 16 40 21 25 33
damage
Cases with moderate damage within 6 21 8 7 13
Bound 1
Total cases documented with severe 2 55 6 24 67
damage
Cases with severe damage within 1 36 5 6 49
Bound 1
Total collapse cases 4 8 12 None None reported
Collapse cases within Bound 1 4 7 6 None None reported
Total cases for each event 50 228 50 115 230
Grand Total 673

event were reported in a format suitable for evaluating the WI and CI indices. We
are not dealing with complete data but only with a partial view. In fact, it may be
readily apparent that, if a building collapsed, one could not always obtain the
needed dimensions unless plans were available.
The data are categorized in four broad damage groups: Light, moderate, severe
and collapse. The category “light” includes apparently undamaged buildings, as
well as those with flexural cracks not exceeding approximately 0.3 mm in thickness
at the level of the tensile reinforcement. “Moderate” refers to those with flexural
and inclined (combined bending and shear) cracks and without permanent drift
exceeding approximately 0.25 %. “Severe” refers to a standing building with one
or more structural elements having localized failure.
Figure 1.1 contains data from the Erzincan earthquake of 1992. It also includes
two additional data items from the Skopje, 1963, and San Fernando, 1971, examples
that had been used as a spot check of the overall bounds. The dominant trend to be
inferred from the plot is that the closer the CI and WI coordinates place a building
with respect to the origin, the more vulnerable it is to earthquake demand. Trajectory
of Bound 1 was selected to contain the four collapse cases. Bound 2 was selected to
contain the cases with moderate damage. That the bound trajectories connect to the
same values on each axis is not dictated by mechanics. It is a choice made to enable
convenient comparison of the sum of the CI and WI indices with the particular bound
that is being used as a criterion. Otherwise, the bounds can be set in any shape or
distance from the origin depending on the local conditions and demands. As
summarized in Table 1.2, two cases of light and six cases of moderate damage
were also within Bound 1, showing that the bound is not as good as a litmus test.
1 Surrealism in Facing the Earthquake Risk 9

All cases with moderate damage were within Bound 2. That was the reason for setting
Bound 2 with PI ¼ 0.25.
The trajectories of the bounds refer to constant values of the Priority Index,
PI ¼ CI + WI. For Bound 1 PI is 0.25 and for Bound 2 it is equal to 0.5.
It goes without saying that if decisions about strengthening had been made
before the earthquake using Bound 1, resources would be wasted on two buildings
that did not require strengthening. On the other hand, lives in four buildings would
have been saved. Admittedly, this is a light conjecture because the bounds were
determined after the data were obtained.
In Fig. 1.2, the data acquired in Düzce, Bolu, and Kaynaşl{ after the 1999 Düzce
earthquake are compared with the two bounds established after the Erzincan 1992
earthquake (Boğazici University and other university teams 1999, Dönmez and
Pujol 2005). On the positive side, we note five collapses that would have been
identified. On the negative side, the data indicate that the CI/WI coordinates of
30 out of 125 cases sustaining no or light damage (about 25 % of the total) placed
the buildings within Bound 1. Furthermore, 21 or approximately one half of the
40 buildings that sustained moderate damage, also would have been identified as
vulnerable. On the positive side, it was observed that two-thirds of the severely
damaged buildings would have been identified (36 buildings out of 55).
Figure 1.3 refers to Bingöl (Ozcebe et al. 2004). Five cases of collapse would
have been identified by Bound 1. Unfortunately there are seven cases of collapse
that would not have been identified by Bound 1. The recurring cause of collapse in
Bingöl was the “captive column.” It could be that an experienced observer would
have identified these buildings as vulnerable before the earthquake; but it has to be
accepted that such a person might not have been available for the inspection. It has
to be accepted that these tragic collapses would have not been identified by a
routine application of the CI/WI indices with Bound 1 defining the limit for serious
vulnerability.
Figure 1.4 is based on studies on buildings belonging to a university in Chengdu,
Sichuan, after the 2008 Wenchuan earthquake (Zhou et al. 2011). Bound 1 would
cover all cases of severe damage. Out of 66 cases with light or no damage, 43 or
two-thirds were within Bound 1 and would have been diagnosed as vulnerable.
Only one-fourth (6 out of 24) of the severe damage cases reported were within
Bound 1.
In Haiti (Fig. 1.5), of the 132 buildings reported to have light damage, 43 of
approximately 1/3 had CI/WI coordinates within Bound 1. Approximately 40 % or
13 of the 33 buildings reported to be moderately damaged were within Bound 1.
A total of 67 buildings were reported to have severe damage. Of these, 49 or nearly
¾ of total had indices that placed them within Bound 1. Even though many
buildings collapsed in Port au Prince, there was no relevant report because infor-
mation on dimensions of the structural elements was lacking (O’Brien et al. 2011).
All of the 673 observed damage states are plotted in Figs. 1.6, 1.7, 1.8, and 1.9.
Each figure contains observations for a single level of reported damage. It is to be
noted that the data combined were obtained from sites in three different countries
and five different cities that experienced different earthquakes. Uniformity is not to
be expected.
10 M.A. Sözen

Fig. 1.6 Cases with light or no damage

Fig. 1.7 Cases with moderate damage

1 Surrealism in Facing the Earthquake Risk 11

Fig. 1.8 Cases with severe damage

Fig. 1.9 Reported collapses

12 M.A. Sözen

One definite conclusion from the four figures is that the CI/WI coordinates of no
building that collapsed or had severe damage was observed to be out of Bound 2.
Strong as this conclusion is, it is of little practical value, because identifying every
building that is within Bound 2 would make remedial action practically impossible.
There is at least one earthquake-threatened city with large districts in which
over 95 % of the low-rise reinforced concrete building inventory would fall within
Bound 2.
“Prediction” of collapses is certainly flawed. The index does not by itself
identify factors such as captive columns or inadequate splices.

1.4 Concluding Discussion

The Hassan Index has its flaws, but it can be used as a first filter in making decisions
about which low-rise reinforced concrete buildings of a city inventory ought to be
subjected to a detailed vulnerability assessment.
If the building inventory of the city is comparable to that included in the
combined set, using Boundary 2 for selection of the buildings to be investigated
would be appropriately conservative, but would mean that approximately 95 % of
the building inventory would be involved in evaluation. It is highly unlikely that
such a plan would be accepted by either the political or the banking communities.
If Bound 1 is used as the criterion for selecting the buildings to be evaluated,
approximately 40 % of the inventory would be included in the study. It is probably
doable, provided that economic and engineering resources are available and the
political leadership is willing to assign those resources to the cause of limiting
human and economic losses.
If the evaluations are made properly in this sample of building quality, 3 % of the
total inventory in low-rise reinforced concrete might be identified as susceptible to
collapse and would be strengthened or demolished. Strengthening or demolishing
would also demand sizeable expense.
Approximately 1 % of the same inventory would be susceptible to collapse, but
would not be identified unless the study was expanded beyond Bound 1. Consider-
ing that not all collapse cases were included in the statistical sample, 1 % may be
optimistic.
The above statements, besides being arguable, are easy to make. But the
conclusions take on a sinister form if they are projected to a particular city (Griffiths
et al. 2007).
Consider a major city that is likely to be subjected to a strong ground motion in
the near future, although it is difficult to state how near. Assume that the reinforced
concrete low-rise building inventory within reach of the energy from the fault is
approximately 500,000 and is similar in quality to those included in the five sites
considered. If 3 % is found to need strengthening or demolition, the target may be
approximately 15,000 buildings. The task is expensive, but within reach of a rich
community. If 1 % “miss” of buildings susceptible to collapse is correct, we should
1 Surrealism in Facing the Earthquake Risk 13

expect approximately 5,000 or even more collapses; this disastrous result will occur
after having strengthened or demolished approximately 15,000 buildings.
The vision that is based on observed damage is gloomy if its projection, in terms
of the quality of building stock in the five sites, is plausible.
Whatever the conditions, it is not mandatory but wise to assess earthquake risk
of the existing building stock in a city threatened by earthquake. As an accurate
indicator of vulnerability, the Hassan Index is a failure. It is, after all, surrealistic to
expect success using two simple indices to identify every failure in an environment
where details such as height or strength of an architectural wall or an oversight in
proper spacing of a few ties can lead to disaster. However, a timely pre-selection for
assessment using the Hassan Index or a similar vetted and simple criterion is likely
to help the cause of reducing human and economic losses in an earthquake.

References

Boğazici University and other university teams (1999) Building data. http://www.anatolianquake.org
Dönmez C, Pujol S (2005) Spatial distribution of damage caused by the 1999 earthquakes in
Turkey. Earthq Spectra 21:53–69
Griffiths JHP, Irfanoglu A, Pujol S (2007) Istanbul at the threshold: an evaluation of the seismic
risk in Istanbul. Earthq Spectra 23(1):63–75
Hassan AF, Sözen MA (1994) Seismic vulnerability assessment of low-rise buildings in regions
with infrequent earthquakes. ACI Struct J 94:31–39
O’Brien P et al (2011) Measures of the seismic vulnerability of reinforced concrete buildings in
Haiti. Earthq Spectra 27(S1):373–386
Ozcebe G et al (2004) 1 May 2003 Bingöl earthquake engineering report. Tübitak Structural
Engineering Research Unit. Report no 2004/1, Ankara
Shiga T, Shibata A, Takahashi T (1968) Earthquake damage and wall index of reinforced concrete
buildings. In: Proceedings of Tohoku district symposium. Architectural Institute of Japan, vol
12, pp 29–32
Zhou W et al (2011) Seismic vulnerability of reinforced concrete structures affected by the 2008
Wenchuan earthquake. https://nees.org/warehouse. p 21