CHAPTER-1 INTRODUCTION Countries

CHAPTER-1
INTRODUCTION
Countries, which are developing more rapidly are unprotected for hazards because of their increasing rate of development and urban growth. There is a poor proper disaster management which leads to increase in risk in most densely populated cities. Most of the growth in terms of building construction structures and infrastructure will concentrate in the developing countries for the next few decades on which are already loaded with various urban problems like population growth, urban sprawl, building density and lack of financial strength. The risk is continuously increasing in these countries at an alarming rate. The main purpose of all vulnerable processes in the world is to save human lives and their property from the impact of natural disasters and earthquakes. It is impossible to live in a disaster free environment but it is possible to reduce the impact of disasters by proper risk management strategies.

Indian subcontinent is among the world’s most earthquake prone areas because it is the more rapidly growing country. Geology predisposes sixty percent of the country’s area vulnerable to earthquake disaster which are needed a proper risk management strategies, among which twelve percent of its land is liable to severe earthquake intensity. The highest vulnerable area is concentrated in the north, near the border with Pakistan, Bangladesh, Bhutan, China and Nepal. This region of high seismic risk is home to 610 million people, 60% of the nation’s population containing cities with populations over 14 million inhabitant. Seven major earthquakes have struck different parts of India over a span of last 25 years in which the approximate deaths, affected people and injured people in last 20 years are 32 thousand, 25 million and 200 million respectively. On 26 January 2001, a very severe earthquake struck Bhuj and shook most parts of Gujarat, causing widespread damage and devastation. Among which 13,805 persons lost their lives, 167,000 persons were injured, over a million homes were destroyed, some were damaged and there was large-scale damage to social and physical infrastructure.

The word “seismic” means the enormous proportions or effect and the term “vulnerability” means the quality or state of being exposed to the possibility of being at a risk, either physically or socially. Hence, the term ”seismic vulnerability” can be defined as the susceptibility of a population of buildings to undergo damage due to seismic ground motion or earthquake. There are two types of seismic vulnerability assessment, one is regional seismic vulnerability assessment framework, which is an essential tool for governments and decision makers to optimally allocate resources and mitigate consequences of earthquakes and another is existing vulnerability assessment methods, which varies with different assumptions, for example, quantification of seismic hazard, building vulnerability assessment and building type. There is an increasing research in the development of seismic vulnerability assessment techniques on earthquake prone areas. Vulnerability assessment methods, proposed by Hill and Rosette are FEMA 154, Euro Code 8, New Zealand Guidelines, Modified Turkish method, NRC Guidelines and Hybrid method and the vulnerability factors, which are utilized in different seismic vulnerability assessment techniques, are soft story, Heavy overhangs, Short column, and Pounding possibility, age of the structure, building height, visible ground settlement, and uncertainty within the model.

Table1. Major vulnerability factors considered in different vulnerability assessment Methods
Vulnerability
assessment
method Soft
story Heavy
overhang Short
column Pounding
possibility Age of the structure Building
height Visible
ground
settlement Uncertainty of the
model
FEMA 154 N N N N Y Y N N
FEMA 310 Y Y Y Y Y Y Y N
IITK- GSDMA Y Y Y Y Y Y N N
Euro Code 8 – – – N Y Y – Y
New Zealand Code Y Y Y Y Y Y – Y
NRC Y N Y Y Y – N N
Modified Turkish Method Y Y Y Y Y Y N N
Here, N = Not considered, Y = Considered, -= Not clearly mentioned
Due to the poorly designed and constructed buildings in the urban areas, they are in vulnerable condition. Until the 2001 Bhuj earthquake, our country was fortunate not to experience a large earthquake in an urban area in which the very high vulnerability of urban India was starkly demonstrated the urban centers of Bhuj, Anjar and Bhachau experienced extensive damage and losses to both new and old constructions. During this earthquake, a large number of recently constructed concrete buildings in Ahmedabad, Gujurat were also badly damaged even though the city is located over 200 km from the epicenter i.e, very far from epicenter and these buildings should have suffered only minor damage if properly designed and constructed. The damage of an earthquake is not only physical, but also psychological and social. By providing well-designed and safe structures, most of these hazards can be prevented. But this is the case for the new structures; the older ones still need to be evaluated from the point of view of the seismic performance. The importance of seismic vulnerability assessment is to take preventive measures to reduce the number of losses of lives and physical damage during possible future earthquakes.

Table 2 provides the details of some past earthquakes in India.

Date Event Time Magnitude Max. Intensity Deaths
12 June 1897 Assam 16:25 8.7 XII 1500
8 Feb,1900 Coimbatore 03:11 6.0 X Nil
4 Apr 1905 Kangara,Himachal Pradesh 06:20 8.6 X 19,000
15 Jan 1934 Bihar-Nepal 14:13 8.4 X 11,000
31 May 1935 Quetta 03:03 7.6 X 30,000
15 Aug 1950 Assam 19:31 8.5 X 1,530
21 Jul 1956 Anjar 21:02 7.0 IX 115
10 Dec 1967 Koyna 04:30 6.5 VIII 200
23 Mar 1970 Bharuch 20:56 5.4 VII 30
21 Aug 1988 Bihar-Nepal 04:39 6.6 IX 1,004
20 Oct 1991 Uttarkashi, Uttaranchal 02:53 6.6 IX 768
30 Sep 1993 Killari(Latur) 03:53 6.4 IX 7,928
22 May 1997 Jabalpur,Madhyapradesh 04:22 6.0 VIII 38
29 Mar 1999 Chamoli,Uttaranchal 12:35 6.8 VIII 63
26 Jan 2001 Bhuj,Gujurat 08:46 7.7 X 13,805
08 Oct 2005 India-Pakistan 09:20 7.4 X 20,600

In general, current seismic vulnerability assessment approaches can be categorized under three main stages of evaluation.

Rapid screening (walk-down evaluation)
Preliminary evaluation
Detailed evaluation
Rapid visual screening
This is the first stage which is used to assign an evaluation index to be used to determine the relative vulnerability of the structure. It is called as Walk-down evaluation which is a simple, inexpensive and short method of evaluation. It can be completed at most in ten minutes per building. It doesn’t require a qualified technician; it just relies on the past performance of similar buildings. It separates the worst and best conditioned buildings based on a ranking procedure. The region with uncertainty may require additional more advanced evaluation methods. So the structure is further evaluated, if it is found inadequate at this stage.

Preliminary evaluation
This is the second stage which is applied for those buildings found inadequate at the first stage. In this stage, a simplified analysis based on many different methods is performed. This stage requires sufficient information needed to perform the method chosen and the time needed for a preliminary evaluation of one building can range from a few hours to a few days.
Detailed evaluation
If a structure is found out to be inadequate in the first two stages, it is evaluated in the third and last stage. So it is known as the final stage. In this stage, a final decision is made by an experienced structural engineer through some advanced engineering studies. The time period of this stage for a particular building can be several weeks.

There is an urgent need to assess the seismic vulnerability of buildings in urban areas of India as an essential component of a comprehensive earthquake disaster risk management policy. Detailed seismic vulnerability evaluation is a technically complex and expensive procedure and can only be performed on a limited number of buildings. It is therefore very important to use simpler procedures that can help to rapidly evaluate the vulnerability profile of different types of buildings, so that the more complex evaluation procedures can be limited to the most critical buildings. Therefore, rapid visual screening (RVS) procedure is used.

Rapid visual screening procedures are not so good at identifying buildings at risk and there appears to be little room for improvement. To obtain the improved results from screening procedures, an alternative method of interpretation based on fuzzy logic and artificial neural network are presented which are the part of the soft computing techniques.

1.1 SOFT COMPUTING
Soft computing, proposed by L.A. Zadeh in 1990s for constructing a new generation of computational intelligent system, is a technique characterized by the use of inexact solution for problems that has no known method to compute the exact solution. These techniques are based on the human type information processing which involves both logical and intuitive information processing. It yields rich knowledge representation (symbol and pattern), flexible knowledge acquisition (by machine learning from data and by interviewing experts), and flexible knowledge processing (inference by interfacing between symbolic and pattern knowledge), which enable intelligent systems to be constructed at low cost. It has been used in many applications such as time series forecasting, supply chain management, motion control and maximum power point tracking. Soft computing methodologies mimic consciousness and cognition in several important respects: they can learn from experience; they can universalize into domains where direct experience is absent; and, through parallel computer architectures that simulate biological processes, they can perform mapping from inputs to the outputs faster than inherently serial analytical representations.

1.2.3. Artificial Neural Network
The development of artificial neural network (ANN) began approximately 50 years ago inspired by a desire to understand the human brain. With the advent of modern electronics, it was only natural to try to harness the thinking process. The first step toward artificial neural networks came in 1943 when Warren McCulloch, a neurophysiologist, and a young mathematician, Walter Pits, wrote a paper on how neurons might work. They modeled a simple neural network with electrical circuits. The evolution of neural networks had been facilitated by the rapid development of architectures and algorithms. In the later years, the discovery of the neural net resulted in the implementation of optical neural nets, Boltzmann machine, spatiotemporal nets, pulsed neural networks and support vector machines. ANNs have found application in such diverse areas as neuro-physiology, physics, biomedical engineering, electrical engineering, acoustics, computer science, robotics, image processing etc. Since early 90’s ANNs have been successfully used in hydrology related areas such as rainfall –runoff modelling, stream flow modelling, water quality etc. An artificial neural network (or simply a neural network) is a biologically inspired computational model which consists of processing elements (called neurons) and connections between them with coefficients (weights) bound to the connections, which constitute the neuronal structure, and training and recall algorithms attached to the structure.

What basically a neural networks do is to classify similar data or to make interpolation or extrapolation in an n-dimensional space, after a process called “training” which is done using an available input-output set.

1.2 Motivation
The motivation for this project work is to obtain the future development of the rapid visual screening procedure based on soft computing techniques. Different types of methods are included in soft computing, among them artificial neural network and fuzzy logic methods are useful. Here, by using artificial neural network toolbox we have analyzed the performance of different buildings at different places. So soft computing is the important mathematical technique for the future development of rapid visual screening procedure.

1.3 Thesis organization
The present chapter with chapter-1 of the thesis provides general information on the purpose of the study and a brief overview on vulnerability assessment using artificial neural network and the problems associated with them.
The chapter-2 provides a state of art review of literatures available on vulnerability assessment and processes associated with them and also about the rapid assessment using artificial neural network. Based on which some critical observations are drawn and the objective and scope of present study have defined.

Chapter-3 details the fundamentals of the neural network toolbox, types, application areas, the algorithm and also briefly describes the methodology of artificial neural network which is based on rapid vulnerability assessment.
Chapter-4 presents results and discussions of artificial neural network and graphs are plotted according to the results.
Several conclusions are drawn and presented in chapter-5. Moreover, some scopes for carrying out future research work also highlighted.

Finally, the references cited in the present work are provided.

Chapter-2
Literature review
This chapter presents the brief review of published literature on rapid improvement in seismic design standards in different earthquake-prone cities. The detailed review of literature on rapid assessment of buildings due to earthquake using artificial neural network has been presented in this chapter.

2.1 Review on Rapid Visual Screening
Though number of significant results have been reported on the seismic vulnerability assessment using rapid visual screening process.
Rojahn (1973) demonstrated the need for rapid improvement in seismic design standards and hazard mitigation strategies. He published different books for seismic design of new buildings and bridges; rapid screening of buildings for potential seismic hazards; detailed seismic evaluation of buildings; seismic upgrade (rehabilitation/retrofit) of buildings; earthquake damage prediction for buildings and bridges (and other structures); safety evaluation of buildings after earthquakes; and detailed evaluation and repair of earthquake damaged buildings. ATC was succeeded due to defining earthquake engineering practice in the United States relate to a variety of factors; including the process by which the documents are developed; the funding made available by government agencies and other sources; the availability of technically qualified specialists; strict quality control measures; and careful editing and attention to report format and content attributes that make the reports easy to use and follow by intended users.

ATC 13 (1982) enumerated a methodology for evaluating earthquake damages in buildings for California City. The ATC-13 report included damage probability matrices for 78 types of representative California facilities, estimates of the time required to restore damaged facilities to their pre-earthquake usability, and estimates of expected death and injuries, expressed as a function of percent damage (repair cost divided by replacement value) and building mass. Prediction of damage probability matrices as per ATC- 13 solely based on expert opinion obtained from direct damage from ground shaking and collateral hazard and collateral losses. It suggested a scale of 13 damage probability matrices well suited for Rapid damage evaluation based on Modified Mercalli Intensity (MMI) scale.
Hassan et al (1995) presented the method of seismic vulnerability assessment of low-rise buildings in regions with infrequent earthquakes. They presented a simplified method of ranking reinforced concrete, low-rise, monolithic buildings according to their vulnerability to seismic damage. They showed that the ranking process required only the dimensions of the structure. They concluded that the process was tested using a group of buildings that suffered various levels of damage during the earthquake of 1992.
FEMA 310 (1998) outlined a detail procedure for seismic vulnerability evaluation of masonry and reinforced buildings. The evaluation procedure is based on rigorous approach to determine existing structural conditions. Buildings are evaluated for certain extent of structural damage that is expected in the building when subjected to earthquake. This level of damage (or Performance Level) is determined a priori by the design professional considering the importance of building and consequences of damage. Two levels of performance defined as Life Safety and Immediate Occupancy during design earthquake. For life safety performance, the building can sustain significant damage to both structural and nonstructural components with some margin against either partial or total structural collapse such that level of risk for life-threatening injury and getting trapped is low. Immediate occupancy building performance means very limited damage to both structural and nonstructural components during the design earthquake. The primary vertical and lateral-force-resisting systems retain nearly all of their original strength and stiffness; however, there could be some minor injuries and damage, which could be easily repaired while the building is occupied. This document prescribes a three-tiered process of increasing detail and reducing margin of safety for the seismic evaluation of existing buildings, as described below: Screening Phase (the design professional find out the potential deficiencies and expected behavior of the buildings to quickly decide whether it complies with the provisions of the FEMA 310 ); Evaluation Phase (The design professional conduct a complete analysis of the building with all of the deficiencies identified in Tier 1 or a deficiency only analysis) and the Detailed Evaluation Phase (conducts a detailed evaluation of buildings which takes a long time).

FEMA 356 (2000) described a prestandard and commentary for the seismic rehabilitation of buildings. It obtained a target performance level of buildings through the rehabilitation method. This procedure included simplified rehabilitation by identifying building model type and considering deficiencies, we determined rehabilitation measures to meet applicable FEMA 310 requirements. The other procedure is Systematic Rehabilitation. It Considered deficiencies, Selected rehabilitation strategy and analysis procedure and Performed mathematical model and force and deformation response evaluation. The rehabilitation design for the two buildings required careful consideration of the owner’s objectives including costs, aesthetics, and continued building operation by minimizing disruptions to building occupants.

FEMA 154 (2002) described about the Rapid Visual Screening of Buildings for Potential Seismic Hazards. The technical basis for the methodology, including the scoring system and its development, is contained in the companion. It provides a “sidewalk survey” approach that enables classification of buildings into two categories: those that appear to be adequately safe and those that may be seismically hazardous and should be evaluated in more detail. It includes improvements in the screening form; an added optional more detailed level of screening; updates of the scoring values; new reference guides for vertical and plan irregularities; additional building types; improved consideration of additions, adjacent structures, and retrofits; an optional electronic scoring methodology.

Sinha et al (2002) formulated a national policy document on Seismic Vulnerability Assessment of Buildings for India. Therein he emphasized the importance of rapid visual screening for all buildings in urban areas as the only viable mean for assessment of seismic vulnerability. The detailed vulnerability assessment is time consuming and costlier as it requires sophisticated equipment and technical manpower. Therefore they recommended of limiting the detailed evaluation in all important and lifeline buildings. For other buildings with significant damage and high concentration of people they suggested for conducting simplified vulnerability assessment using simple information gathered from visual observations and structural drawings or on-site measurements.
Fischer et al (2002) observed that this study is devoted to the formulation and construction of an integrated model for earthquake risk assessment of buildings in seismic regions. This study is devoted to the formulation and construction of an integrated model for earthquake risk assessment of buildings in seismic regions. The model developed has five stages: characterization of ground motion; construction of the building model; evaluation of the inelastic building response; structural damage assessment and risk evaluation. Examples included a large building inventory and two individual structures are developed to show the potential use of the model. They concluded that the model is capable of discriminating different foundation soils, earthquake performance of shear-wall and frame buildings, asymmetries in height and plan, and between conventional and seismically isolated structures. Such features may be useful to engineers working in city planning, emergency and risk management, and the insurance industry.

Tyagunov et al (2004) presented a report on vulnerability and risk assessment for earthquake prone cities. They considered the countries of low and moderate seismicity depending not only on the hazard level, but also on the aggregate elements at risk. They conducted an interdisciplinary study aimed at assessment and mapping of different kinds of risks for the territory of Germany which were related to seismic prone zones, where earthquakes are probable, producing shaking intensity up to VIII. They presented mainly to consideration of vulnerability analysis, including methodological aspects of the approach at the country level. Vulnerability composition models were constructed for the building stock of communities of different population classes, which can serve prototypes for the risk analysis. They concluded that using these models on GIS platform we computed and mapped specific damage distribution and estimated distribution of seismic risk potential over the territory of Germany.

Calvi et al (2006) proposed the development of seismic vulnerability assessment methodologies over the past 30years. They presented models, which were capable of estimating losses in future earthquakes and the main ingredients in a loss model is an accurate, transparent and conceptually sound algorithm to assess the seismic vulnerability of the building stock. They had taken the most significant contributions in the field of vulnerability assessment and identified the key advantages and disadvantages of these procedures in order to distinguish the main characteristics of an ideal methodology over the past 30 years.

Guéguen et al (2007) developed a simplified approach for vulnerability assessment in moderate-to-low seismic hazard regions in France. They proposed a light vulnerability assessment method and tested in Grenoble (France), based on classes and scores provided in the GNDT method and Compared with the Risk UE method, the damage obtained by our approach showed that 90% of buildings have residuals smaller than 0.2. They found that non-experts reducing substantially the estimate cost can rapidly collect the information useful for the light method of vulnerability assessment and the average damage was then computed using the GNDT formula considering the probable intensities, which could be observed in Grenoble and the average damage reaches 0.4. They showed that the results of relative vulnerability assessment provided useful initial information for the urban zones of Grenoble where the vulnerability is higher.

Tesfamariam et al (2008) made a comparative study among several seismic vulnerability assessment techniques i.e. ”Hybrid” vulnerability assessment method, FEMA 154 (Rapid Visual Screening), Euro Code 8, New Zealand Guidelines, Modified Turkish method and NRC Guidelines for evaluating their suitability for use in seismic risk assessment. With three case studies comprising three different seismicity and geological zones (Dhaka and Rangamati city in Bangladesh and Kelowna in Canada) are made for the purpose. A scoring system based on general description of vulnerability, building response factors, variance in output, applicability and ease of use is employed for evaluating the suitability of the technique. It is observed from the study that, hybrid method performed well in all these cases. Furthermore, this methods could easily integrated into GIS frame work to visualize the building vulnerabilities in a spatial manner, which will facilitate the authority to manage effective seismic hazard risk reduction measures, including upgrading, repairing and retrofitting of structures.

Kaplan et al (2008) developed a new rapid seismic vulnerability assessment method for turkey. Rapid evaluation methods had been developed to estimate seismic performance of buildings in tectonically active areas for seismic hazard mitigation. They presented a HAZUS based rapid evaluation method using building census data to find validation of the method and building grading system. They collected the important parameters such as properties of the structural system, seismic quality of construction and geotechnical properties of soil by costly field investigations. They found that proposed method can be used for seismic assessment of mid to large size cities.

Rastogi et al (2010) considered two cities of Gujurat that is Gandhidham and Adipur cities for a comprehensive study of seismic risk assessment after the seismic code was usually revised in 2002. They were conducted rapid visual screening procedure on 16000 buildings in Gandhidham and Adipur cities. Performance score can be calculated by adding the score modifiers. It can be said that the buildings with higher performance scores perform better compared to lower performance scores. He concluded that Gujarat lies in one of the most seismically active zones of the world and also possibility of future earthquakes of moderate to great nature cannot be ruled out. In this regard, a comprehensive study of seismic risk assessment of Gujarat was carried out and based on low scores; it can said that the area is potentially vulnerable to future earthquakes.
Ferreira et al (2010) presented the seismic vulnerability assessment strategies for facade walls of traditional masonry buildings. They developed a methodology and its subsequent application to the old building stock of the historical city center of Coimbra over 600 buildings. They found that from the post-earthquake damage assessment of masonry buildings in Aquila, Italy, it was developed and calibrated an analytical function to estimate the mean damage grade for masonry façade walls. They showed that from vulnerability function for facade walls, the calculation of damage scenarios was carried out and was subsequently used in the development of an emergency planning tool and in the elaboration of an accessibility routing proposal for the case study – Old city center of Coimbra.

Vicente et al (2011) explored about seismic vulnerability and risk assessment of the historic city center of Coimbra, Portugal. They found that Seismic risk evaluation of built-up areas involved analysis of the level of earthquake hazard of the region. They devoted that rigorous vulnerability assessment of existing buildings and the implementation of appropriate retrofitting solutions can help to reduce the levels of physical damage, loss of life and the economic impact of future seismic events. They developed this study with the aim of identifying building fragilities and reducing seismic risk. The main purpose of this research was to discuss vulnerability assessment methodologies, particularly those of the first level, through the proposal and development of a method previously used to determine the level of vulnerability, in the assessment of physical damage and its relationship with seismic intensity. They discussed about the strategy and proposed methodology adopted for the vulnerability assessment, damage and loss scenarios for the city centre of Coimbra, Portugal, using a GIS mapping application.

Achs et al (2012) addressed rapid seismic evaluation of historic brick-masonry buildings in Vienna (Austria) by visual screening. They adopted RVS methodologies for this specific type of buildings considering the validity and quality of the seismic assessment. The derived score was calculated by taking the parameters such as regularity of the inspected building, its state of preservation and geometry. They predicted the damage potential of a seismic event comparable with the L’Aquila 2009 earthquake based on the derived score. They showed the correlating results of the RVS methodology and damage grades. They found that the resulting maps of damage scenarios gave useful information for emergency and evacuation planning as well as for identification of critical objects vulnerable to seismic loading.

Nanda et al (2014) stated Rapid Seismic Vulnerability Assessment of Building Stocks for developing Countries. They showed that rapid visual screening procedures needed to identify buildings susceptible to earthquake damage. Relevant structural characteristic information was collected and used to determine a structural score which were related with damage grades I to V and structural score ;0.7 indicated high vulnerability requiring detail evaluation and retrofitting of the building. They presented a procedure for rapid visual screenings for building stocks constructed in developing countries. They prepared score sheets for three-seismicity viz. low, moderate and high.

Ozmen et al (2014) proposed the evaluation of the main parameters affecting seismic performance of the low and mid-rise reinforced concrete (RC) buildings. They studied an important portion of the building stock in many earthquake prone countries for understanding their seismic behavior. They evaluated seismic code, number of stories, concrete strength, amount of transverse reinforcement and infill-wall contribution parameters. They determined seismic performances of the models for different performance levels and seismic loading conditions. They concluded that the specifications and higher transverse reinforcement up to 50%, the concrete strength up to 66%, infill-walls 15% and number of story 55% increased the seismic performance for life safety level.

FEMA 155 (2015) can be defined as the supporting documentation, which provides a methodology for rapid visual screening of buildings. It included improvements in the screening form, an added optional more detailed level of screening, updates of the scoring values, new reference guides for vertical and plan irregularities, additional building types, improved consideration of additions, adjacent structures, and retrofits, an optional electronic scoring methodology, new insight into risk and more detailed discussion on how to run an effective screening program. It supported the FEMA 154 code, which is actually based on rapid visual screening.

Perrone et al (2015) proposed a rapid visual screening method to determine a Safety Index for hospital buildings. The newly developed procedure provides a risk index by evaluating the main parameters that affect the vulnerability of buildings during a systematic sidewalk survey. The procedure had been applied to two Italian hospitals located in different seismic areas, and the results are compared with a similar index obtained from a pushover analysis. Surveys performed that in recent earthquakes have shown that the performance of hospitals and their functionality after an earthquake are related not only to structural damage but also to damage that occurs to nonstructural elements and medical equipment, but for a large-scale mapping of the seismic risk to hospitals, it is impossible to proceed with advanced methodologies; simplified methodologies are thus required. These methodologies must consider the performance of nonstructural elements and equipment.

Albayrak et al (2015) proposed a rapid seismic risk assessment method for existing building stock in urban areas. They had taken 1643 buildings of Eskisehir city for identifying and strengthening of the deficient buildings in northern part of Turkey. The risk assessment considered criteria as the age of building, number of stories, existence of soft story, short column, heavy overhangs, pounding affect, topographic effects, visual building construction quality and earthquake zone where the building was located and performance score (Earthquake Risk Score–E.R.S) of each building. They classified the buildings as high risk, moderate risk and low risk. They concluded that total 218 among 1643 buildings were classified as high risk and more detailed evaluations of these buildings were recommended before confirming the building as earthquake risk.

Wahyuni (2015) studied the evaluation of the vulnerability of reinforced concrete buildings against earthquake loads. He found that the previous studies had been conducted a testing of a case study by providing lateral loads of a building to simulate the earthquake loads and it was conducted to find out the behavior of damage and deviation that occurred in the case study based on a numerical analysis. Then, He compared the testing and numerical analysis of the buildings in a state of immediate occupancy, life safety, or collapse prevention. He concluded that a proven of assessing visually of the building was agree with the numerical results.

Dey et al (2015) proposed a method of rapid visual screening (RVS) of seismically vulnerable Reinforced Concrete buildings in Guwahati city. They showed that The city lies in zone “V” and due to this seismic risk in Guwahati is increasing with population growth and the encroachment of vulnerable built-in environment into areas susceptible to seismic hazard.

They investigated that majority of deaths and injuries in earthquakes occured because of the disintegration and collapse of buildings, and much of the economic loss and social disruption caused by earthquakes. So they presented a Rapid Visual Screening (RVS) survey of multi storied buildings of Guwahati was carried out to rectify the defects in the existing buildings thus making it safe against earthquake.

2.2 Literature review on Artificial Neural Network:
There are not many studies on ANNs in the field of seismic vulnerability assessment. In addition, the present ones are not beyond the first steps of a child. Still the neural networks give satisfactory results for specific cases, which imply a promising future in the subject area.
A probabilistic neural network model for seismic vulnerability assessment is developed by Aoki et al. The aim of the authors was to outline a general procedure to be followed in order to assign a seismic vulnerability assessment estimation, while the classical neural networks give a univocal resulting vector with the greatest possibility of occurrence for a given input vector, probabilistic networks supply several answers each of which associated with an estimate of the respective probability.

Zang et al (2001) investigated structural damage detection using measured frequency response functions (FRFs) as input data to artificial neural networks (ANNs) and the output is a prediction for the actual state of the specimen, i.e., healthy or damaged. A further advantage of this particular approach was found to be the ability to deal with relatively high measurement noise, which is of common occurrence when dealing with industrial structures. Here, three different networks were trained and twenty compressed FRFs, obtained from further measurements, were used for the actual damage detection tests. The results showed that, in all cases considered, it was possible to distinguish between the healthy and damaged states with very good accuracy and repeatability.

Demartinos et al (2006) emphasized the performance of a fuzzy logic–based rapid visual screening procedure that results in the categorization of buildings into five different types of possible damage with respect to the potential occurrence of a major seismic event. In order to provide results representing expected damage, adaptive neural networks were used to train the method according to information obtained from the vulnerability of 102 buildings stricken by the Athens earthquake of 1999. They concluded that Fuzzy logic rapid visual screening procedure is more efficient in terms of leading to the reliable formation of a high-priority set of buildings.

Moseley et al (2007) investigated an alternative screening procedure based on fuzzy logic and artificial neural networks. Two databases of buildings damaged during the Athens earthquake of 1999 are used for training purposes. Then they concluded that the trained fuzzy logic based rapid visual screening procedure represents a marked improvement when identifying buildings at risk. Here, they explained that the proposed procedure has a significant optimization potential, is worth pursuing and, to this end, a strategy that outlines the future development of the fuzzy logic based rapid visual screening procedure is proposed. RVSPs are well-accepted methods that generally give good results.
Sen (2010) presented a soft computation methodology based on the fuzzy logic model (FLM) and system principles for the classification of buildings into five distinctive but mutually inclusive classes in terms of fuzzy sets as ”without”, ”slight”, ”moderate”, ”heavy”, and ”complete” hazard categories. Visually assessable variables, namely, story number, cantilever extension, soft story, weak story, building quality, pounding effect, hill-slope effect, and peak ground velocity were considered as inputs with a single output variable as earthquake hazard category. He also discussed about the application of the FLM for building hazard categorization is performed for 1249 existing buildings in Zeytinburnu quarter of Istanbul City.
Adnan et al (2012) developed an effective, convenient and reliable intelligent seismic evaluation system for buildings in Malaysia by using Back-Propagation Artificial Neural Network (ANN) algorithm. They considered forty-one buildings with 164 sets of input data spreading throughout Peninsular and East Malaysia were chosen under seismic loading at peak ground accelerations of 0.05g, 0.10g, 0.15g and 0.20g respectively. He concluded that the ANN system is suitable to be used for predicting the seismic behavior of their buildings at any given time. This study focused on feasibility of adopting ANN in predicting structural damages due to seismic ground motions, as well as parametric investigation to determine the effect of different combinations of input parameters in affecting the prediction accuracy.
2.3 Critical observation and shortcomings:
Rapid visual screening (RVS) is a very quick way of assessing the building vulnerability based on visual screening. This procedure is requiring only for visual evaluation and it is recommended for all buildings.

Simplified vulnerability assessment is a complete analysis of the building that addresses all of the deficiencies identified. This procedure is required limited engineering analysis based on information from visual observations and structural drawings and it is recommended for all buildings with high concentration of people.

Detailed vulnerability assessment gives a detailed picture of building and thoroughly evaluation can be occurred. It required a detailed computer analysis that required for design of a new building and it is recommended for all important and lifeline buildings.

The input parameters of RVS are apparent quality, wall openings, orientation of openings, diaphragm action, horizontal bands, arches, soil condition, basement and random rubble stone masonry wall. By adding these score modifiers, we get the performance score.

Rapid visual screening procedures are not so good at identifying buildings at risk and there appears to be little need for improvement. To obtain the improved results from screening procedures, an alternative method of interpretation based on fuzzy logic and artificial neural network are presented which are the part of the soft computing techniques.

Fuzzy logic, ANFIS and neural network are used through the toolbox presented in MATLAB and used in rapid visual screening procedure through the training procedure.

Artificial neural network can handle large amount of data sets, it has the ability to implicitly detect complex nonlinear relationships between dependent and independent variables. So we will use ANN in RVS procedure.

2.4 OBJECTIVE OF STUDY
The main objective of present work is to develop a methodology for rapid visual screening (RVS) procedure of a building using artificial neural network (ANN).

Most significant engineering parameters are selected based on engineering judgement and used to feed ANN models which kind of models are inspired from the neurological system of humans .

2.5 SCOPE OF PRESENT WORK
To fulfill the above objective the scopes for the present work are defined as follows
To conduct rapid visual screening procedure of masonry and Reinforced Concrete (RC) buildings.

To calculate the damage condition of building using artificial neural network (ANN).

To predict the co-relation between performance score calculated from rapid visual screening and the damage condition of buildings.

Chapter-3
THEORY AND FORMULATION
This chapter describes the best methods applied in this research work for characterization of materials used in manufacturing the concrete and evaluation of the effect of coal bottom ash as replacement of river sand on properties of concrete.

3.1 PRINCIPLES OF RAPID VISUAL SCREENING
Rapid visual screening of buildings is the process of assessing their seismic vulnerability by defining, via inspection, their structural integrity and safety. It involves the identification of the load-resisting system, the materials of the structure, the respective seismic region, and the inventory of various characteristics that arguably affect a structure’s seismic response. The collected data is then processed in order to compute a numeric score that relates to seismic vulnerability.
In this respect, the first comprehensive methodology was the procedure presented by FEMA-154, which is actually a probabilistic approach to vulnerability assessment via rapid visual inspections. The principal concept of this methodology is that given the information about the state of various structural characteristics, it is possible to evaluate a numeric score associated with the probability of collapse occurrence, should a major seismic event occur. This evaluation is based on a frequency analysis of recorded data relevant to the influence of the various structural characteristics on seismic performance.

The object of performing a RSVP is to quickly inspect a set of buildings in order to assess their susceptibility to earthquake damage. Structural analysis calculations are not carried out during this first assessment. Structural characteristic parameters in the form of yes/no answers are identified during a RVSP with the object of determining a final structural score. This final structural score should indicate if a further investigation is required. Structural scores for a set of buildings can be used to rank the buildings in an order of importance concerning susceptibility to earthquakes. Score modifiers for parameters that are considered to affect the structural performance are then added to or subtracted from the basic score. These parameters are the seismic region, the age of the building (as in the seismic code used for the design), the soil type, the condition of the building (as in a lack of maintenance) and if the building is high-rise. Additional parameters are if the building has a soft story, short columns, a regular layout of infill walls, any previous damage, vertical or plan irregularities and the possibility of torsion or pounding. Summing the initial score and the score modifiers gives a final structural score.

The evaluation is based on few parameters of building. The parameters of the buildings are building height, frame action, pounding effect, structural irregularity, short columns, heavy overhang, soil conditions, falling hazard, apparent building quality, diaphragm action etc. On the basis of above mentioned parameters, performance score of the buildings has been calculated. The formula of the performance score is given as
PS=BS-VSM×VS
Where VSM represents the Vulnerability Score Modifiers and VS represents the Vulnerability score that is multiplied with VSM to obtain the actual modifier to be applied to the BS or Basic score.

In detail, the earthquake planning and protection has introduced a rapid visual screening procedure (OASP RVSP) evaluates a score according to judgments about the truth or false-hood of the states of various structural characteristics. These are as follows: structural type, seismic region, existence of soft story, short columns, regular layout of infill walls, aseismic code used in the design, poor condition of the building exterior, previous damage, high rise, irregularities with height, irregularities in layout, torsion effect, pounding effect, and soil type. In order to evaluate the final structural score, OASP provides a preliminary score that corresponds to each state that describes the structural characteristics. Let Pi be the frequency of collapse occurrence when the state of a particular characteristic is true, then the preliminary score can be expressed by Equation1:
Si=-log?Pidamage=collapse;state=true (1)
It should be noted that all variables are dependent upon the structural type and are generally independent of each other with the exception of high-rise buildings (more than five stories) constructed on a poor quality soil. That is, a base score is first assigned according to the state of the variable ‘structural type’, and afterwards each performance attribute is associated with a modifying score, with reference to the structural type. The final structural score, S, is evaluated as the sum of the base and modifying scores and can be expressed by Equation 2. Consequently, the physical meaning of the structural score is that it represents a probability of collapse equal to 10-S. At this point, it should be noted that the OASP RVSP may result in negative values, which evidently have no physical meaning. In practice, it is assumed that the higher the building’s structural score, the better its seismic performance is expected to be.

S=iSi, i=1,2,…………………n n:number of structural characteristics (2)
The basic disadvantage of this approach to RVSP is the fact that most performance attributes are considered to affect a structure’s seismic response independently. Thus, the interaction between them is not taken into account and the structure is not treated as a whole. This contradicts the actual behavior of a structure and interprets seismic response as the sum of all characteristics’ partial contributions. In addition, during the visual inspection of buildings, it is rather difficult to describe the state of various structural characteristics by assigning true or false values. The qualitative judgment of the engineer is thus limited to two extreme values, while very often the description of a structure’s identity has many uncertainties.
The FL-RVSP is an alternative treatment process of information gathered during the rapid visual inspection of buildings. It constitutes a typical computational subroutine, even though it includes complicated mathematical functions. One might argue that the use of such an advanced computational subroutine is not justified for the purposes of RVSP when compared with simpler subroutines required for the evaluation of the structural score. The primary objective, however, is the formation of a reliable and precise methodology of seismic vulnerability assessment via visual observation. Moreover, the evolution of computer technology has eliminated the differences between executing simple and complicated computational subroutines.

3.2 THEORETICAL BACKGROUND OF ANN
Full form of ANN is Artificial Neural Network which is a part of the Artificial Intelligence (AI). Artificial Intelligence is a very potential technology in the field of computer application, which helps computer users in various fields to solve problems where algorithm approach can’t be formulated, so it requires human expertise and intelligence.

Neural network technology came from the mammalian brain, mainly from the cerebral part. So, neural network understands all the incomplete and confusing problems as human brain solves. Each neuron of the brain is composed of a body which consists one axon and a number of dendrites.

The dendrites of the brain receives signal from other neurons and the axon of the body divides into branches which is terminating in little end bulbs. The small gap between an end bulb and a dendrite is called a synapse. One axon of the neuron forms a synaptic connection with another neurons. The network must learn the connection weights from available training patterns. A neural network responds –in parallel- to the inputs given to it and the final result consists of a overall state of the network after the network has reached to a steady condition, which co-relates to the input data sets and corresponding output sets.

Neural networks can learn complex nonlinear relationships even when the input is noisy and imprecise. ANNs have been applied to a number of problems with variable success: speech recognition, image processing, pattern recognition, classification, optimization problems, robotics and control, and medical and commercial applications. An artificial neural network is performed as a biological neural networks. This neural network is connected with many mathematical functions based on the assumptions that at first information is received by simple elements called neurons, signals are passed between neurons through connection links, each connection link has associated with a weight function and at last each neuron applies an activation function to its net input to determine its output signal. So a neural network is characterized by its architecture, its method of determining the weights on the connection and its activation function.

Types of ANNs
Based on the connection pattern, artificial neural network can be classified into two categories.

They are:
Feed-forward networks, in which network have no looped connections and
Feedback network which is called recurrent network, in which loops occur because of feedback connections.

Feed-forward network is called multilayer perceptron in which neurons are organized into layers that have unidirectional value which are connected between them. Feed-forward networks are static, that is, they produce only one set of variables rather than a sequence of values from input. Here, the inputs of each neuron is the weighted sum of the outputs from the previous layer. If the weight is zero, there is no connection between two nodes.

Feedback networks are called recurrent network in which the inputs of each layer is affected by the output of the previous layer. The inputs of the network consist of both external inputs and outputs. It is called the dynamic process in which inputs of each layer is connected with the previous layer.

3.2.2 Backpropagation rule
Backpropagation is the most popular default training algorithm for ANN, and it has been used by many researchers for daily flow forecasting. ANNs trained using backpropagation are also known as feedforward multi-layered networks trained using the backpropagation algorithm. The backpropagation algorithm was originally introduced in the 1970s, but its importance wasn’t fully appreciated until a famous 1986 paper by David Rumelhart. The mechanism and process of the standard 3-layer BPNN is described as below.

Firstly, the given data are stored in the input neurons. The input neurons then transmit these values across the links to the hidden neurons. On each link there is a weight used to multiply transmitted values. The weighted sum associated with the neurons bias is then put through a simple function (transfer function or activation function) to generate a level of activity for the hidden neuron. The activation levels of hidden neurons are then transmitted through their outgoings links to the neurons in the output layer. As before, the values are weighted and summed during transmission. Then, the summed value is put through an activation function to get activation level of the output neurons, which is the final solution of the network. It provides an efficient computational procedure to evaluate the performance of the network. After variables are loaded into a neural network, they must be scaled from their numeric range into the numeric range that the neural network, they must be scaled from their numeric range into the numeric range that the neural network can deal with efficiently. The common numeric ranges for the networks to operate are from 0 to 1 denoted (0,1) and minus one to one denoted (-1,1). The activation function usually used for backpropagation is a sigmoid function.

f(I)=11+e-1Where, Ii=j=1nwij xjThe sigmoid function is found to be useful for most neural network applications. It maps values into the (0,1) range. The weight updates are based on a variation of the generalized data rule :
?wij = ?EfI+??wijpreviousWhere, E is the error.

In BPNN, errors of the current layer are calculated based on the errors of the former layer. For example, the error of output layer is Eijoutput=yjdesired-yjactual, then the error of the hidden layer can be calculated according to
Eihidden=df(Iihidden)dIj=1nwijEjoutputThis is an operation that is propagated backwards across the network. Hence it is named as backpropagation neural network.

30727142010541
001
341187354970Get error for each output neuron in network and add to total error
00Get error for each output neuron in network and add to total error
66057511430Start
00Start

464614118699954034901705231079770118677
116205168893Apparent quality, pounding, soil condition, irregularities, arches, diaphragm action as input and performance score as output
00Apparent quality, pounding, soil condition, irregularities, arches, diaphragm action as input and performance score as output

402830392573
YES
401139833020If last pattern has trained
00If last pattern has trained
1060315166424
5049795239687116732253730Divide data into 3sets: Training data, Test data and Validation
00Divide data into 3sets: Training data, Test data and Validation

NO
40942051787101079770214333
313518429296If Total error<final target error
00If Total error<final target error
116732165451Create feed forward, Back Propagation 3 network layers
00Create feed forward, Back Propagation 3 network layers
YES

2092411161256107977087144 NO
11533087836Train the network
0Train the network

107977096385
1070937534Make total error = 0
00Make total error = 0
4102443172143394865100828Save the data as a MATLAB file
00Save the data as a MATLAB file

10797708350
2388973395931
001
9885425451Apply first pattern and train
00Apply first pattern and train
41189198976
367809211207End
End

Fig.4. Flow chart for backpropagation procedure
CHAPTER-4
RESULTS AND DISCUSSION
Here, we had taken different places such as Bharatnagar, sapananagar1, sapananagar2, apananagar1, apananagar2, apananagar3 and sectors into consideration. The parameters taken for the places are structural irregularity, apparent quality, soil condition, pounding, wall openings, orientation of openings, diaphragm action, horizontal bands, arches, random rubble stone masonry, water tank at roof capacity, location of water tank and basement. These parameters are called the score modifiers.

In Artificial Neural Network Toolbox, we had taken the parameters such as structural irregularity, apparent quality, soil condition, arches, diaphragm action, basement as input data and the performance score was taken as output data or target data. At first, Bharatnagar data were taken for the training process. According to the results getting from the training process, assess the other buildings and also calculate results for other buildings.

Number of data points taken for Bharatnagar is 401. Out of total data, 70% data points i.e. 281 data points are taken for training, 20% data i.e. 81 data points are taken for validation and 10% data i.e. 40 data points are taken for testing process. Number of hidden layers taken is 25. The training for Bharatnagar buildings continued until the validation error failed to decrease for six iterations (validation stop). From these results, the optimum network was selected for each problem in two stages. The first stage consisted of choosing the ideal training method and network architecture by minimizing the weighted error for each method and architecture. The ideal step for stepwise training was selected. Then, results are produced using backpropagation were noticed and saved as a MATLAB program.
The following regression plots display the network outputs with respect to targets for training, validation, and test sets. For a perfect fit, the data should fall along a 45-degree line, where the network outputs are equal to the targets. If even more results that are accurate were required, you could retrain the network by clicking Retrain in nftool. This will change the initial weights and biases of the network, and may produce an improved network after retraining. Table 3 shows the training, validation and testing value of Bharatnagar.

Table 3. Values for Bharatnagar
Procedure Values
Training 0.8780
Validation 0.9222
Testing 0.8821
Here, the performance of bharatnagar is presented. The regression value for the training, validation and the testing phase are 0.87808, 0.92222, 0.88215 respectively and the total regression value was found as 0.88631 which is shown in the below graph.

Fig.5. Fitting curve of Bharatnagar for training

Fig.6. Fitting curve of Bharatnagar for validation

Fig.7. Fitting curve of Bharatnagar for Testing

Fig.8. Total fitting curve value of Bharatnagar
The trained model is used for predicting the performance score at six other locations namely apananagar1, apananagar2, sapananagar1, sapananagar2, sapananagar3 and sector with values are found as 0.8298, 0.7661, 0.7417, 0.8567, 0.7567 and 0.6823 respectively which is shown in the below graph. It is found the regression value curve lies between 0.68-0.82. This means a good co-relation exists between the trained model and the actual results obtained at the site.

Therefore, it can be concluded that the trained model can be utilized for conducting the Rapid visual screening of other masonry buildings in similar seismic zones. Although the present study is conducted with limited parameters, but it can be compute a complete set of parameters and also for Reinforced Concrete framed buildings , easily which is included in the scopes for the work to be done in the next semester.

Fig.9. Fitting curve of apananagar1

Fig.10 Fitting curve of apanagar2

Fig.11. Fitting curve of sapanagar1

Fig.12. Fitting curve of Sapanagar2

Fig.13. Fitting curve of Sapanagar3

Fig.14. Fitting curve of Sector
MATLAB Neural Network toolbox and scripting
MATLAB is a source which consists of various toolboxes which are the part of the soft-computing.
CHAPTER-5
5. CONCLUSION
An Artificial Neural Network (ANN) was trained for the masonry buildings, which gives a good co-relation between the trained model and the actual results obtained at the site.

It was found out that the ANN gave a highest accuracy to the selected six input parameters among all the parameters. It also gives accuracy when any of a parameter added or eliminated.

Increasing number of hidden neurons give highest accuracy.

The system managed to produce accurately 352 number of outputs among 401 building samples i.e. 88% accuracy.

5.1. WORK DONE SO FAR
The rapid visual screening procedure using FEMA 154 handbook calculates the performance score of the buildings.

Data are taken into the neural network toolbox and the target data are the output data.

The values are trained and validating, testing values are saved, and the performance graph can be plotted from that values.

For a perfect fit, the data should fall along a 45-degree line, where the network outputs are equal to the targets. If results that are even more accurate were required, you could retrain the network by clicking Retrain in nftool.

5.2. FUTURE SCOPE
Calculation of the damage condition of masonry building and Reinforced Concrete buildings is needed through neural network toolbox.

Calculation of damage condition of masonry building and reinforced concrete buildings is needed by using Fuzzy logic toolbox and ANFIS toolbox.

CHAPTER-6
6. REFERENCES
BSSC (Building Seismic Safety Council) (2000), Pre-standard and Commentary for the Seismic Rehabilitation of Buildings, FEMA-356, FEMA, Washington D.C..

Demartinos, K. and Dritsos, S. (2006), “First level pre-earthquake assessment of buildings using fuzzy logic”, Earthquake Spectra, Volume 22, No. 4, pages 865–885,
FEMA 154 (1988), Rapid Visual Screening of Buildings for Potential Seismic Hazards: A Handbook, ASCE, California.

FEMA 154 (2001), Rapid Visual Screening of Buildings for Potential Seismic Hazards: A Handbook, 2nd edition, ATC, Redwood City, California.

FEMA 155 (1988), Rapid Visual Screening of Buildings for Potential Seismic Hazards: Supporting Documentation, ASCE, 3rd edition, Washington, D.C.

Mandas, A. and Dritsos, S. (2004), “Vulnerability assessment of RC structures using fuzzy logic”, Proc. of the Int. Conf. on Risk Analysis IV, Rhodes, Greece, 51-60, Transactions of the Wessex Institute.

Mathworks Inc. (2002), Matlab® Release 14, Apple Hill Drive 3, Natick, Massachussetts.

Srikanth.T and Kumar.S (2010), “Earthquake Vulnerability Assessment of Existing Buildings in Gandhidham and Adipur Cities Kachchh, Gujarat (India)”, EJSR, Vol.41, pp.336-353
Sen.Z.(2010),” Rapid visual earthquake hazard evaluation of existing buildings by fuzzy logic modeling”, Expert systems with applications, vol-37, pp- 5653–5660.

Moseley.J and Dritsos.S(2008), “Pre-earthquake fuzzy logic and neural network based rapid visual screening of buildings”, journal of structural engineering, Vol. 27, No. 1 pp- 77-97.

Sinha.R and Goyal.A(2002), “Seismic Vulnerability Assessment of Buildings and Procedure for Rapid Visual Screening of Buildings for Potential Seismic Vulnerability”, A national policy, Department of Civil Engineering, Indian Institute of Technology Bombay.

T. Fischer, M. Alvarez, J.C. De la Llera(2002), “An integrated model for earthquake risk assessment of buildings”, Engineering Structures, vol-24, pp- 979–998.

Adnan.A(2012), “Artificial Neural Network Application for Predicting Seismic Damage Index of Buildings in Malaysia”, Journal of Structural Engineering, vol-12, pp-46-57.

C. H. Juang and D. J. Eltont(2013), “Fuzzy logic for estimation of earthquake intensity based on building damage records”, Fuzzy sets and systems, vol-3, pp-33-45.

N. Alam • M. Shahria Alam • S. Tesfamariam(2012), “Buildings’ seismic vulnerability assessment methods: a comparative study”, Research Gate, vol-62, pp-405–424.

M. H. Arslan?, M. Ceylan†, T. Koyuncu(2010), “An ANN Approaches on Estimating Earthquake performances of existing RC buildings”, Neural Network World.

Applied Technology Council (1987). Evaluating the seismic resistance of existing buildings. Report ATC-13. Washington, DC, USA: Building Seismic Safety Council.

D. Perrone, M.A. Aiello, M. Pecce, F. Rossi (2015), “Rapid visual screening for seismic evaluation of RC hospital buildings”, Structures, Volume 3, 2015, pp. 57-70.

Achs, G. and Adam, C. (2011). Risk assessment of historic residential brick-masonry buildings in Vienna by Rapid-Visual-Screening. III ECCOMAS Thematic Conference on Computational Methods in Structural Dynamics and Earthquake Engineering (COMPDYN 2011), May 25-28, 2011, Corfu, Greece (Papadrakakis M., Fragiadakis, M., Plevris, V., eds), CD-ROM paper, paper no. 419.

U.Albayraka, M.Canbaz, G.Albayrak (2015), “A rapid seismic risk assessment method for existing building stock in urban areas”, Procedia Engineering 118 ( 2015 ) pp-1242 – 1249.

E. Wahyunia (2015), “Vulnerability Assessment of Reinforced Concrete Building Post-Earthquake”, Procedia Earth and Planetary Science 14 ( 2015 ) pp-76 – 82.

Kaplan. H, Yilmaz. S, Akyol. E and Sen. G, “A new rapid seismic vulnerability assessment method for Turkey”, The 14th World Conference on Earthquake Engineering October 12-17, 2008, Beijing, China, pp-44-46.

G.M. Calvi, R. Pinho, G. Magenes, J.J. Bommer, L.F. Restrepo-Vélez and H. Crowley (2006), “Development of seismic vulnerability assessment methodologies over the past 30years”, ISET Journal of Earthquake Technology, Paper No. 472, Vol. 43, No. 3, September 2006, pp. 75-104.

R.Vicente, S. Parodi ,S. Lagomarsino, H. Varum , J. A. R. M. Silva (2011), “Seismic vulnerability and risk assessment: case study of the historic city centre of Coimbra, Portugal”, Bull Earthquake Eng (2011) 9:pp-1067–1096 .

R. P. Nanda and D. R. Majhi (2014), “Rapid Seismic Vulnerability Assessment of Building Stocks for Developing Countries”, KSCE Journal of Civil Engineering (2014) 18(7):2218-2226.

S. Tyagunov, L. Stempniewski, G.Grunthal, R. Wahlstrom, J. Zschau, “Vulnerability and risk assessment for earthquake prone cities”, 13th World Conference on Earthquake Engineering Vancouver, B.C., Canada August 1-6, 2004,Paper No. 868.

H. B. Ozmen, M. Inel and E. Meral, “Evaluation of the main parameters affecting seismic performance of the RC buildings”, Vol. 39, Part 2, April 2014, pp. 437–450.

A.F.Hassan and M.A.Sozen, “Seismic vulnerability assessment of low rise buildings in regions with infrequent earthquakes”, 94-S4.

Ferreira, T., Vicente, R. and Varum, H. (2010). Seismic vulnerability assessment of masonry facade walls. 14th European Conference on Earthquake Engineering (14th ECEE), Ohrid, Macedonia.

Y. Dey, N. Das, F. F. Islam (2015), “rapid visual screening (RVS) of seismically vulnerable R.C.C buildings in Guwahati city”, pp-1328-1420.

P. Guéguen, C. Michel, L. LeCorre (2007), “A simplified approach for vulnerability assessment in moderate-to-low seismic hazard regions: application to Grenoble (France)”, Bull Earthquake Eng (2007) 5:467–490.

32. Alam. N, Shahria Alam. M, Tesfamariam. S (2012), “Buildings’ seismic vulnerability assessment methods: a comparative study”, Nat Hazards, 62, pp-405–424.