user manual for the earthquake loss estimation …selena.sourceforge.net/selenamanual.pdfuser manual...

DR

AFT

User Manual for the Earthquake Loss Estimation

Tool: SELENA

Sergio Molina

NORSAR and Universidad de Alicante

Dominik H. Lang, Conrad D. Lindholm and Fredrik Lingvall

NORSAR

October 1, 2010

DR

AFT

Contents

1 Introduction 1

1.1 Scope and methodology of SELENA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Copyright 5

2.1 Disclaimer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3 Technical Description of SELENA 6

3.1 Basic Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

3.2 Provision of Seismic Demand . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3.2.1 Probabilistic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3.2.2 Deterministic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

3.2.3 Analysis with Real-time Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3.3 Site-dependent Seismic Demand — Amplification of Ground Motion . . . . . . . . . . . . 8

3.3.1 IBC-2006 (International Code Council, 2006) . . . . . . . . . . . . . . . . . . . . . 10

3.3.2 Eurocode 8 (European Committee for Standardization CEN, 2002) . . . . . . . . . 14

3.3.3 Indian Standard IS 1893 (Part 1) : 2002 (Bureau of Indian Standards, 2002) . . . 14

3.4 Structural Performance Under Seismic Action . . . . . . . . . . . . . . . . . . . . . . . . . 17

3.4.1 The Capacity Spectrum Method (CSM) as Proposed in ATC-40 . . . . . . . . . . 17

3.4.2 The Modified Capacity Spectrum Method (MADRS) . . . . . . . . . . . . . . . . . 23

3.4.3 Improved Displacement Coefficient Method (I-DCM) . . . . . . . . . . . . . . . . . 27

3.5 Fragility Curves and Damage State Probability . . . . . . . . . . . . . . . . . . . . . . . . 28

3.6 Economic Losses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29

3.7 Humanloss — Casualties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

3.7.1 The Basic Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

3.7.2 The HAZUS Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32

4 Installation 34

4.1 System Requirements and Resent Code Changes . . . . . . . . . . . . . . . . . . . . . . . 34

4.1.1 Installing the GNU Scientific Library . . . . . . . . . . . . . . . . . . . . . . . . . . 34

4.2 The Directory Structure of SELENA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.3 The SELENA m-files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

4.3.1 The SELENA mex-files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.3.2 The SELENA oct-files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38

5 Running SELENA 38

5.1 Preparation of Input Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

5.1.1 Input Files for Deterministic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . 39

5.1.2 Input Files for Probabilistic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . 40

5.1.3 Input Files for Analysis with Real-time Data . . . . . . . . . . . . . . . . . . . . . 41

5.1.4 Common Input Files for all Analysis Types . . . . . . . . . . . . . . . . . . . . . . 41

2

DR

AFT

5.1.5 Input Files for the Calculation of Economic Losses . . . . . . . . . . . . . . . . . . 44

5.1.6 Input Files for the Calculation of Human Losses — Casualties . . . . . . . . . . . 44

5.1.7 Mandatory Input Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

5.2 Mean Damage Ratio Computation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

5.2.1 Median Values and Confidence Levels . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.3 The SELENA Program Sequence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.3.1 The Stand-alone SELENA Application . . . . . . . . . . . . . . . . . . . . . . . . . 51

5.3.2 The Matlab and Octave Command-line Interface . . . . . . . . . . . . . . . . . . . 51

5.3.3 The Matlab Graphical User Interface . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5.4 Dealing with Uncertainties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53

5.5 Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.5.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.5.2 Format of the Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

5.5.3 Mean Damage Ratio Output Files . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

6 Examples 62

6.1 The Bucharest Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

6.2 Determistic Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

6.3 Probabilistic Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

6.4 Realtime Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

7 Plotting results in Geographic Information Systems (GIS) 63

8 Known Issues 63

9 Summary 64

Bibliography 64

A Tables 68

B Compiling the C-code 74

B.1 Tools and Libraries for Windows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

B.2 Building the Stand-alone Application on Linux/Unix . . . . . . . . . . . . . . . . . . . . . 75

B.3 Building the Stand-alone Application on Windows . . . . . . . . . . . . . . . . . . . . . . 75

B.4 Building the Stand-alone GUI Application on Linux/Unix . . . . . . . . . . . . . . . . . . 75

B.5 Building the Stand-alone GUI Application on Windows . . . . . . . . . . . . . . . . . . . 76

B.6 Building the Linux/Unix mex-files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

B.6.1 Building the Windows mex-files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

B.7 Building the oct-Files . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

DR

AFT

Selena 1 INTRODUCTION

1 Introduction

THE earthquake loss estimation tool SELENA, which is described herein, provides local, state andregional officials with a state-of-the-art decision support tool for estimating possible losses from

future earthquakes. This forecasting capability enables users to anticipate the consequences of futureearthquakes and to develop plans and strategies for reducing risk. GIS-based software (e.g., ArcView [1])can be utilized at multiple levels of resolution to graphically show loss results and to prepare responsestrategies.

Some of the first earthquake loss estimation studies were performed in the early 1970’s following the1971 San Fernando earthquake. These studies put a heavy emphasis on loss of life, injuries, and theability to provide emergency health care. More recent studies have focused on the disruption of roads,telecommunications and other lifeline systems. The present loss estimation tool computes analytically,based on ground shaking estimates, the degree of damage on specific construction groups and detailed aswell as gross economic losses.

Earlier the National Institute of Building Sciences (NIBS) has developed the tool HAZUS-MH [2–4]for the federal emergency management agency (FEMA) in order to provide a powerful technique fordeveloping earthquake loss estimates. This to be used in:

• anticipating the possible nature of an earthquake disaster and the scope of the emergency responseneeded to cope with an earthquake disaster,

• developing plans for recovery and reconstruction following a disaster, and

• mitigating the possible consequences of earthquakes.

The methodology generates an estimate of the damage consequences for a city or a region based ona ‘scenario earthquake’, i.e., an earthquake with a specified magnitude and location. The resulting lossestimate will generally describe the scale and the extent of damage and disruption that may result fromsuch an earthquake. Using such computations the following information can principally be obtained by:

• Quantitative estimates of losses in terms of direct costs for repair and replacement of damagedbuildings and lifeline system components; direct costs associated with loss of function (e.g., loss ofbusiness revenue, relocation costs); casualties; people displaced from residence; quantity of debris;and regional economic impacts.

• Functionality losses in terms of loss-of-function and restoration times for critical facilities such ashospitals, and components of transportation and utility lifeline systems and simplified analyses ofloss-of-system-function for electrical distribution and potable water systems.

• Extent of induced hazards in terms of fire ignitions and fire spread, exposed population and buildingvalue due to potential flooding and locations of hazardous materials.

All the system, methods, and data have been coded into a user-friendly software that operates througha geographical information system (GIS) which is called HAZUS-MH [2]; the ESRI GIS system is usedby HAZUS-MH.

In a simplified form, the steps followed by the HAZUS methodology are:

1. Select the area to be studied. This may be a city, a county or a group of municipalities.

2. Specify the magnitude and location of the scenario earthquake. In developing the scenario earth-quake, considerations should be given to the potential fault locations.

3. Provide additional information describing local soil and geological conditions, if available.

1

DR

AFT


4. Using formulas embedded in HAZUS, probability distributions are computed for damage to differentclasses of buildings, facilities, and lifeline system components and loss-of-function estimates aremade.

5. The damage and functionality information is used to compute estimates of direct economic loss,casualties, and shelter needs. In addition, the indirect economic impacts on the regional economyare estimated for the years following the earthquake.

6. An estimate of the number of ignitions and the extent of fire spread is computed. The amount andtype of debris are estimated. If an inundation map is provided, exposure to flooding can also beestimated.

The earthquake-related hazards considered by the methodology in evaluating casualties, and resultinglosses are collectively referred to as potential earth science hazards (PESH). Most damage and loss causedby an earthquake is directly or indirectly the result of ground shaking, but there are also other featuresof an earthquake (such as fault rupture, liquefaction, land sliding etc.) that can cause permanent grounddisplacements and have an adverse effect upon structures, roads, pipelines, and other lifeline structureswhich are also considered.

Soil type can have a significant effect on the intensity of ground motion at a particular site. Thesoftware contains several options for determining the effect of soil type on ground motions for a givenmagnitude and location.

Tsunamis and seiches are also earthquake-caused phenomena that can result in inundation or wa-terfront damage. In the methodology, potential sites of these hazards may be identified, but they areevaluated only if special supplemental studies are performed.

The type of buildings and facilities considered in HAZUS-MH are as follows:

General Building Stock: The commercial, industrial and residential buildings in the studied regionare not considered individually when calculating losses. Instead, they are grouped together into 36model building types and 28 occupancy classes and degrees of damage are computed for groups ofbuildings.

Essential Facilities: These include medical care facilities, emergency response facilities and schools.Specific information is compiled for each building so the loss-of-function is evaluated in a building-by-building basis.

Transportation lifeline systems: These include highways, railways, light rail, bus systems, ports,ferry system and airports and they are broken in components such as bridges, stretches of roadwayor track, terminal, and port warehouses. The damage and losses are computed for each componentof each lifeline.

Utility lifeline systems: These include potable water, electric power, waste water, communications,and liquid fuels (oil and gas) and are treated in a manner similar to transportation lifelines.

High-potential loss facilities: These include dams, nuclear power plants, or military installationswhich need supplementary specific studies to be evaluated.

All results from HAZUS-MH are provided as “best estimates”, and no uncertainty in the results isprovided for.

The downside of these fascinating developments implemented in HAZUS-MH is that it has been sointimately connected to the U.S. environments that it is practically impossible to apply it to the rest ofthe world.

We have in the present study developed and adapted the core of the HAZUS methodology to greaterflexibility compared to non-free tools, such as, ArcGIS [5].

2

DR

AFT


A more important extension is that a logic tree scheme with weighted input of uncertain parametershas been incorporated, and an example of seismic damage scenarios for the city of Oslo have beenconducted and published.

1.1 Scope and methodology of SELENA

While the HAZUS approach is attractive from a scientific/technical perspective, the fact that it is tai-lored so intimately to U.S. situations and to a specific GIS software makes it difficult to apply in otherenvironments and geographical regions.

Aware of the importance of a proper seismic risk estimation, the international centre for geohazards(ICG), through NORSAR (Norway) and the University of Alicante (Spain), has developed a free softwaretool in order to compute the seismic risk in urban areas using the capacity spectrum method namedSELENA (SEimic Loss EstimatioN using a logic tree Approach). The user will supply built area or numberof buildings in the different model building types, earthquake sources, empirical ground-motion predictionrelationships, soil maps and corresponding ground-motion amplification factors, capacity curves andfragility curves corresponding to each of the model building types and finally cost models for buildingrepair or replacement. This tool will compute the probability of damage in each one of the four damagestates (slight, moderate, extensive, and complete) for the given building types. This probability issubsequently used with the built area or the number of buildings to express the results in terms ofdamaged area (square meters) or number of damaged buildings. Finally, using a simplified economicmodel, the damage is converted to economic losses in the respective input currency and human casualtiesin terms of different injury types are computed [6].

The algorithm is transparent in writing and loading the input files and getting the final results. Themain innovation of this tool is the implementation of the computation under a logic tree scheme, allowingthe consideration of epistemic uncertainties related with the different input parameters to be properlyincluded, and the final results are provided with corresponding confidence levels. Until now the methodhas been successfully applied to the city of Oslo and Naples [6, 7].

The basic approach is often called the capacity-spectrum method, because it combines the groundmotion input in terms of response spectra (see, for example, the spectral acceleration versus spectraldisplacement illustrated in Figure 1) with the building’s specific capacity curve (see the example shownin Figure 2. The philosophy is that any building is structurally damaged by its permanent displace-ment (and not by the acceleration by itself). For each building and building type the inter-story drift(relative drift of the stories within a multistory structure) is a function of the applied lateral force thatcan be analytically determined and transformed into building capacity curves (capacity to withstandaccelerations without permanent displacements). Building capacity curves naturally vary from buildingtype to building type, and also from region to region reflecting local building regulations as well as localconstruction practice. Under the HAZUS-umbrella FEMA developed capacity curves for 36 U.S. buildingtypes for four earthquake code regimes (reflecting the variation in building regulations as a function oftime across the U.S.). These 144 capacity curves are developed analytically, but adjusted so that empir-ical knowledge is incorporated in the curves whenever possible. The building capacity curve is definedthrough three control points: Design, Yield and Ultimate capacity (Figure 2). Up to the yield point,the building capacity curve is assumed to behave elastically linear. From the yield point to the ultimatepoint, the capacity curve changes from an elastic to a fully plastic state (curved form), and the curveis assumed to remain fully plastic past the ultimate point (linear form). A bi-linear representation (twolinear parts) is sometimes used to simplify the model shown in Figure 2. The vulnerability curves (alsocalled fragility curves) are developed as log-normal probability distributions of damage from the capacitycurves (see the illustration in Figure 3). The structural damage states are (as in most other proposedschemes and neglecting the state no damage) divided into four damage states: slight, moderate, extensive,and complete. A detailed description of these damage states are found many places. For example, thedescription for light frame wood buildings are:

3

DR

AFT


0

0.25

0.5

0.75

1

0 0.05 0.1 0.15 0.2 0.25 0.3 0.35

Sp

ectr

al a

ccel

erat

ion

Sa

[g]

Spectral displacement Sd [m]

Figure 1: The methodology is based on presenting the ground-motion response spectral ordinates (atgiven damping levels) of spectral acceleration versus spectral displacement.

Sp

ectr

al a

ccel

erat

ion

Sa

[g]

Spectral displacement Sd [cm]

dd dy du

ad

ay

au

demand curve

(dd , ad) − design capacity

(dy , ay) − yield capacity

(du , au) − ultimate capacity

median

median +1σ

median −1σ

Figure 2: The principle of the building specific capacity curve intersected by the load curve representingthe seismic demand.

4

DR

AFT

Selena 2 COPYRIGHT

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50 60 70 80 90 100

Dam

age

pro

bab

ilit

y P

(d

s |

Sd)


slig

ht

mod

erat

eex

tens

ive

complete

Figure 3: Example fragility curves showing the probabilities P (ds|Sd) of being in or exceeding the differentdamage states, ds, for building type C1M as given in HAZUS99.

slight: Small plaster cracks at corners of door and window openings and wall-ceiling intersections; smallcracks in masonry chimneys and masonry veneers. Small cracks are assumed to be visible with amaximum width of less than 1/8 inch (cracks wider than 1/8 inch are referred to as “large” cracks).

moderate: Large plaster or gypsum-board cracks at corners of door and window openings; small diagonalcracks across shear-wall panels exhibited by small cracks in stucco and gypsum wall panels; largecracks in brick chimneys; toppling of tall masonry chimneys.

extensive: Large diagonal cracks across shear-wall panels or large cracks at plywood joints; permanentlateral movement of floors and roof; toppling of most brick chimneys; cracks in foundations; splittingof wood sill plates and/or slippage of structure over foundations.

complete: Structure may have large permanent lateral displacement or be in imminent danger of collapsedue to cripple wall failure or failure of the lateral load resisting system; some structures may slipand fall off the foundation; large foundation cracks. Three percent of the total area of buildingswith Complete Damage is expected to be collapsed, on average.

2 Copyright

THE SELENA program is an open source software and the source code for the program is freelyredistributable under the terms of the GNU General Public License (GPL) as published by the Free

Software Foundation (http://www.gnu.org). See also the file COPYING which is distributed with theSELENA program.

The SELENA program can be downloaded at: http://selena.sourceforge.net At this website youcan also find information how to contact the authors and report bugs etc.

5

http://www.gnu.org

http://selena.sourceforge.net

DR

AFT

Selena 3 TECHNICAL DESCRIPTION OF SELENA

2.1 Disclaimer

The SELENA program is distributed in the hope that it will be useful but WITHOUT ANY WAR-RANTY. More specifically:

THE PROGRAM IS PROVIDED “AS-IS” WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSOR IMPLIED, INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OR CONDITIONSOF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. IN NO EVENT SHALL ANYOF THE AUTHORS OF THE SELENA PROGRAM AND/OR NORSAR, NORWAY, UNIVERSIDAD DEALICANTE, SPAIN, BE LIABLE FOR ANY SPECIAL, INCIDENTAL, INDIRECT, OR CONSEQUEN-TIAL DAMAGES OF ANY KIND, OR DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,DATA, OR PROFITS, WHETHER OR NOT THE AUTHORS OF THE SELENA PROGRAM HAVE BEENADVISED OF THE POSSIBILITY OF SUCH DAMAGES, AND/OR ON ANY THEORY OF LIABILITYARISING OUT OF OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

3 Technical Description of SELENA

IN this section the main features of SELENA is discussed.

3.1 Basic Procedure

The HAZUS methodology covers a wide range of different damages and losses to buildings, lifelines,people etc.; however, in the present version of SELENA we have only implemented the first part of themethodology, the estimation of damage to the general building stock, the economic and human lossesrelated to these physical damages. All results are provided with ranges of uncertainty facilitating theeasy computation of, e.g., median value and 16%- respectively 84%-fractiles of damage.

It has to be noted that SELENA requires quite extensive basis information within a number of inputfiles. These can be easily generated as tables in, for example, a spreadsheet program (e.g., MS-Excel,OpenOffice, MS-Access etc.) and exported as ASCII-table files with all required information given in thematrices.

Since a resolution of the damage outputs on the level of individual buildings would require huge compu-tation efforts, SELENA as most other risk estimation software tools considers the minimum geographicalunit (GEOUNIT), i.e., the census tract, as the smallest area unit. In practice, this unit is related tobuilding blocks or smaller city districts. The decision on the extent of each geographical unit has to bemade considering different aspects such as having equal soil conditions, constant surface topography or ahomogeneous level of building quality within the demarcated area. The main basis information consists inthe building inventory database (which is also somewhat difficult to generate). This type of informationsometimes is provided by local agencies or governmental institutions. In any case, a thorough investiga-tion of the local building stock by walk-downs and on-site inspections should be conducted in order toallow a representative classification of the prevalent building typologies. The building inventory databaseshould contain a maximum of details about building materials, building techniques, built area, floors ofthe building, height, foundations, seismic regulations used in the construction, use of the building, num-ber of occupants, year of construction, etc. The building information is classified according to buildingtype, built square meters in each one of the geographical units which form the region under study or asan individual building if a site-specific study is going to be done. The classification of the building typecan be either done according to the HAZUS methodology (see http://www.fema.gov/hazus documentsor previous HAZUS reports) or following a user-defined classification scheme being more specific for theavailable building stock.

6

http://www.fema.gov/hazus

DR

AFT


3.2 Provision of Seismic Demand

A key point in any seismic risk assessment is the provision of seismic ground motion (level and spectralcharacteristics of earthquake shaking). In order to carry out a seismic risk and loss assessment withSELENA, the user can provide the seismic ground-motion amplitudes on three different ways:

• provision of spectral ordinates (taken out from probabilistic shaking maps) for each geographicalunit (probabilistic analysis),

• definition of deterministic earthquake scenarios (e.g., historical or user-defined events) and ap-propriate ground-motion prediction equations in order to compute the spectral ordinates in eachgeographical unit (deterministic analysis),

• provision of recorded ground-motion amplitudes at the locations of seismic (strong-motion) stations(analysis with real-time data).

Following the provisions of the international building code 2006 (IBC-2006) [8], spectral accelerationsat the three periods T = 0.01 [s] peak ground acceleration (PGA), T = 0.30 [s] (Sa0.3) and T = 1.00[s] (Sa1.0) have to be provided in order to describe the elastic design spectrum. Most other earthquakecodes (e.g., Eurocode 8 [9]) define the shape of the design spectrum such that only a design accelerationvalue, generally the PGA, is required to scale the amplitudes of the spectrum. Consequently, in case thatEurocode 8 design spectra are chosen for the analysis, spectral accelerations values Sa0.3 and Sa1.0 arenot regarded.

3.2.1 Probabilistic Analysis

The probabilistic analysis procedure denotes the use of spectral ordinates which are taken from probabilis-tic shake maps. In addition to the acceleration values (PGA, Sa0.3, Sa1.0) for each minimum geographicalunit, the geographical coordinates of the centroid have to be provided. Probabilistic shake maps are gen-erally developed for rock conditions such that soil amplification is not included in the spectral ordinates.

3.2.2 Deterministic Analysis

For the deterministic analysis the ground-motion parameters (PGA, Sa0.3, Sa1.0) produced by the sce-nario earthquake are calculated by selectable ground-motion prediction relations (attenuation relations).Since all geographical units are located in different distances to the assumed epicenter of the scenarioearthquake, this process is done separately for each geographical unit. A considerable number of well-established ground-motion prediction relations is already incorporated in the SELENA code (Appendix A,Table 19) but any user-provided relation can be easily implemented. It should be considered that allprovided prediction relations refer to rock site conditions and thus compute ground-motion amplitudeswithout soil amplification since this is covered in a separate (subsequent) calculation step. Even thoughthe respective soil terms are provided in the code they are not considered during the analysis.

Depending on the type of design spectrum chosen for the analysis, predicted ground-motion amplitudesare either used to define the shape of the elastic design spectrum (e.g., IBC-2006) or only to scale theamplitudes of the spectrum (e.g., Eurocode 8) which later represents the seismic demand for the capacityspectrum method (CSM) procedure. As Table 1 illustrates, all predefined ground-motion predictionequations can be used to derive mean values of ground-motion amplitudes as well as their ±1σ (standarddeviation) values in order to account for aleatoric uncertainty.

Since each ground-motion prediction equation is dependent on a particular distance, SELENA au-tomatically computes four different types of distances: epicentral distance Repi , hypocentral distanceRhypo, “Joyner-Boore” distance Rjb (shortest distance to the vertical surface projection of the faultrupture plane), and the shortest distance to the subsurface fault rupture plane Rrup (see Figure 4).

7

DR

AFT


Figure 4: Schematic illustration of the different distance types.

Thereby, the expected value of the surface fault rupture length, L, is based on the relationship byWells and Coppersmith [10]:

log10(L) = −3.55 + 0.74M for strike-slip faults (1)

log10(L) = −2.86 + 0.63M for reverse faults (2)

log10(L) = −3.22 + 0.69M for all other fault types (3)

where L is the rupture length in [km] and M is the moment magnitude of the earthquake.

3.2.3 Analysis with Real-time Data

In case of an analysis with real-time data, a major problem consists in the fact that the locations forwhich ground-motion data is available will certainly not comply with the center points of the definedgeographical units (i.e., centroids). Consequently, the provided spectral ordinates at these locations haveto be assigned to the centroids in a somehow reliable way. The procedure applied here is schematicallyillustrated in Figure 5. Basically it is checked which of the available points (here the nodes of an equally-spaced grid pattern) are within a 5 km-radius around each centroid. If at least 5 points meet this criterionthe mean value and corresponding 16%- resp. 84%-fractiles of the spectral ordinates of all stations arecomputed and assigned to the respective centroid. If less than 5 points are within this 5 km-radius, anew circle of 10 km diameter is drawn. Given that the location of a centroid is more or less identicalwith the location of one recording station, its spectral ordinates are directly assigned to the centroidwithout further processing. It should be regarded, that the provided spectral ordinates already cover soilamplification effects as they are realistic ground-motion data recorded at the ground surface. Thereforeany additional consideration of soil amplification has to be avoided. [In practice, this means that theSELENA soil input files (i.e., soilcenteri.txt) do not contain soil class indecies other than those forrock, i.e., 2 for IBC-2006, 1 for Eurocode 8].

3.3 Site-dependent Seismic Demand — Amplification of Ground Motion

In case that sedimentary soil materials are present at a site, the seismic ground motion at the groundsurface is modified both in amplitude and frequency content. Respective amplification factors and/orcorner periods which basically describe shape of the design spectra for the different soil classes are givenin the corresponding code provisions. Currently, the procedures of IBC-2006, Eurocode 8 (Type 1 and

8

DR

AFT


Author(s) (year)Target ground-motion parameter

mean value (mv) mv+1σ mv - 1σ

Boore et al. [11], Boore et al. [12], Boore et al. [13] * * *Ambraseys et al. [14] * * *Toro et al. [15] * * *Campbell and Bozorgnia [16]), Campbell [17] * * *Campbell and Bozorgnia [18] * * *Abrahamson and Silva [19] * * *Sabetta and Pugliese [20] * * *Ambraseys et al. [21] * * *Akkar and Bommer [22] * * *Sadigh et al. [23] * * *zbey et al. (2003) * * *Spudich et al. [24] * * *Bommer et al. [25] * * *Atkinson and Boore [26] * * *Zonno and Montaldo [27] * * *Schwarz et al. [28], Ende and Schwarz [29] * * *Ambraseys and Douglas [30], Douglas [31],Ambraseys and Douglas [32] * * *Chapman [33] * * *Crouse and McGuire [34] * * *Gulkan and Kalkan [35] * * *Lussou et al. [36] * * *Dahle et al. [37] * * *Bommer et al. [38] * * *Marmureanu et al. [39] * * *Sharma et al. [40] * * *

Table 1: Selection of the empirical ground-motion prediction methods which are implemented in thecurrent SELENA-version (see the m-file att sub.m or the C-file att sub.c distributed with SELENA).

Figure 5: Spectral ordinates at the sites of randomly-distributed recording stations are assigned to thecenters of the geographical units (centroids).

9

DR

AFT


0

Sp

ectr

al a

ccel

erat

ion

Sa

[m/

s2 ]

Period T [s]

TA TAV TVD

PGA

Sa0.3

Sa1.0

damping ξ = 5 %

response spectrum

Figure 6: Standard shape of the response spectrum.

Type 2), and Indian standard IS 1893 [41] are incorporated while more will follow in upcoming SELENA-versions.

3.3.1 IBC-2006 (International Code Council, 2006)

The methodology characterizes ground shaking using a standardized response spectrum shape as givenin IBC-2006 [8], which consists of four parts: PGA, a region of constant spectral acceleration at periodsfrom zero seconds to Tav, a region of constant spectral velocity between periods from Tav to Tvd, and aregion of constant spectral displacement for periods of Tvd and beyond (see Figure 6).

The region of constant spectral acceleration is defined by the constant Sa at 0.3 s (Sa1.0). The regionof constant spectral velocity has Sa proportional to 1/T and is anchored to the constant Sa at 1.0 s(Sa1.0). In general, the elastic design spectrum Sa(T ) is defined by the following equations:

Sa(T ) = Sa0.3(0.4 + T/TA) for T < TA (4)

Sa(T ) = Sa0.3 for T < T < TAV (5)

Sa(T ) = Sa1.0/T for TAV < T < Tvd (6)

Sa(T ) = Sa1.0Tvd/T 2 for Tvd < T < 10 s (7)

The period Tav is based on the intersection of the region of constant spectral acceleration and constantspectral velocity and its value varies depending on the values of spectral acceleration that define thesetwo intersecting regions:

Tav = Sa1.0/Sa0.3 (8)

10

DR

AFT


The period Ta representing the left corner period of the spectral plateau can be determined as follows:

Ta = 0.2TAV = 0.2(Sa1.0/Sa0.3) (9)

The constant spectral displacement region has spectral acceleration proportional to 1/T 2 and is anchoredto the spectral acceleration value at the period Tav, where constant spectral velocity transitions to constantspectral displacement.

The period Tvd is based on the reciprocal of the corner frequency fc, which is proportional to stressdrop and seismic moment. This frequency is estimated from the Joyner and Boore [42] relationship as afunction of moment magnitude:

Tvd =1

fc= 10(M−5)/2 (10)

where fc is the corner frequency and M is the moment magnitude. When the moment magnitude is notknown (probabilistic earthquake scenario), the period Tvd is assumed to be 10 seconds (M = 7.0).

In order to be able to describe the elastic design spectra (for rock: site class B) in case that the PGAis given, the following expressions have to be regarded:

Sa0.3 = Saas = 2.5apga (11)

Sa1.0 = Sasl = apga (12)

Amplification of ground shaking to account for local site conditions is based on the site classes (seeTable 2) and soil amplification factors as given by the IBC-2006 provisions. These code provisions do not

Sitesite class description

shear-wave velocityclass vs,30 [m/s]

A Hard rock, eastern U.S. sites only > 1500B Rock 760–1500C Very dense soil and soft rock 360–760D Stiff soil 180–360E Soft soil, profile with > 3 m of soft clay defined as soil

with plasticity index PI> 20, moisture content w > 40% < 180F Soils requiring site-specific evaluations –

Table 2: “NEHRP” site classification [43] as applied by IBC-2006 [8].

provide specific soil amplification factors for PGA or PGV. The methodology amplifies rock (site classB) PGA by the same factor as that specified in Table 3 for short period (0.3 s) spectral acceleration, as

Site Class B Site Class

Spectral Acceleration A B C D E

Short-Period, Sas [g] Short-Period Amplification Factor, FA

≤ 0.25 0.8 1.0 1.2 1.6 2.5(0.25, 0.50] 0.8 1.0 1.2 1.4 1.7(0.50, 0.75] 0.8 1.0 1.1 1.2 1.2(0.75, 1.0] 0.8 1.0 1.0 1.1 0.9

> 1.0 0.8 1.0 1.0 1.0 0.91-Second Period,Sal [g] 1-Second Period Amplification Factor, FV

≤ 0.1 0.8 1.0 1.7 2.4 3.5(0.1, 0.2] 0.8 1.0 1.6 2.0 3.2(0.2, 0.3] 0.8 1.0 1.5 1.8 2.8(0.3, 0.4] 0.8 1.0 1.4 1.6 2.4

> 0.4 0.8 1.0 1.3 1.5 2.4

Table 3: Site amplification factors as given in IBC-2006 [8].

expressed in the following expression:apga

i = apgaFAi (13)

11

DR

AFT


where apgai is the PGA for site class i (in units of [g]); apga is that for site class B (in units of [g]) and FAi

is the short period amplification factor for site class i, for spectral acceleration Sas. The construction ofdemand spectra including soil effects is done using the following equation for short periods:

Sasi = SasFai (14)

and for long periods:Sali = SalFvi (15)

while the period TAVi, which defines the transition period from constant spectral acceleration to constantspectral velocity is a function of the site class. It can be determined by the following equation:

TAVi =Sasi

Sas

Fvi

Fai(16)

where:

Sasi: short-period spectral acceleration for site class i (in units of [g])

Sas: short-period spectral acceleration for site class B (in units of [g])

FAi: short-period amplification factor for site class i and for spectral acceleration Sas

Sali: 1-second (long) period spectral acceleration for site class i (in units of [g])

Sal: 1-second (long) period spectral acceleration for site class B (in units of [g])

Fvi: short-period amplification factor for site class i and for spectral acceleration Sal

Tavi: transition period between constant spectral acceleration and constant spectral velocity for siteclass i (in [s]).

Note that the period Tvd, which defines the transition period from constant spectral velocity to constantspectral displacement, is not a function of site class [see Eq (10)].

For the evaluation of structural damage it is more convenient to plot the acceleration response spec-trum as a function of the spectral displacement (rather than the period). This could be achieved due tothe relation between the different spectral parameters:

Sa

ω= Sv = Sdω (17)

where ω is the angular (natural) frequency of the oscillator (i.e., ω = 2πf , where f is the frequency in[Hz]).

The final result of this process is the computation of a 5% damped response spectrum at the centerof each geographical unit (where values of ground motion were computed) or at the specific site understudy. In the following, this will be done exemplary for selected site classes according to the IBC 2006provisions, i.e., NEHRP site classes A–E.

Example 3.1 Generation of elastic demand spectra for NEHRP site classes B, C and D

Given parameters:

• PGA for rock site conditions (site class B): aPGAb = 0.20 g

12

DR

AFT


• Sa (0.3 s) for rock site conditions (site class B): Sa0.3b = 0.50 g

• Sa (1.0 s) for rock site conditions (site class B): Sa1.0b = 0.20 g

Steps:

1. Calculation of spectral parameters for soil demand spectra: In case that Sa0.3B and Sa1.0B cannot be derived by spectral attenuation equations, both can be provided by: Sa0.3 = Sas = 0.50 g[Eq. (11)] and Sa1.0 = Sal = 0.20 g [Eq. (12)].

2. Determination of site amplification factors for site classes (according to Table 3).

Site amplification factors for Site ClassSas = 0.50 g resp. Sal = 0.20 g B C D

Fa 1.0 1.2 1.4Fv 1.0 1.6 2.0

3. Calculation of short-period and long-period spectral accelerations as well as transition period Tavi

ParameterSite Class

B C D

apgai = apgaFai 0.20 g 0.24 g 0.28 g

Sasi = SasFai 0.50 g 0.60 g 0.70 gSali = SalFvi 0.20 g 0.32 g 0.40 g

0.40 s 0.53 s 0.57s0.08 s 0.106 s 0.114 s3.16 s (for M 6.0), 5.62 s (for M 6.5), 10.0 s (for M 7.0)

4. Generation of elastic demand spectrum (damping ξ = 5%):

0

0.2

0.4

0.6

0.8

1

0 0.5 1 1.5 2 2.5

Sp

ectr

al a

ccel

erat

ion

Sa

[g]

Period T [s]

Site Class: BCD

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50 60

Sp

ectr

al a

ccel

erat

ion

Sa

[g]


Site Class: BCD

13

DR

AFT


3.3.2 Eurocode 8 (European Committee for Standardization CEN, 2002)

For the description of seismic action, two different types of design spectra are provided within Eurocode8 [9]. This mainly in order to account for the differing level of seismic hazard in Europe and the differentearthquakes susceptible to occur. In case that earthquakes with a surface-wave magnitude Ms > 5.5 areexpected it is suggested to use Spectrum Type 1, else (Ms ≤ 5.54) Type 2. The question which spectrumtype to choose for a specific region should be based upon “(...) the magnitude of earthquakes that areactually expected to occur rather than conservative upper limits defined for the purpose of probabilistichazard assessment”.

For the sake of completeness both spectrum types are incorporated in the current version of SELENAeven though scenario earthquakes with magnitudes smaller than 5.5 are not expected to cause considerablestructural damages to the general building stock.

Both types of the horizontal design spectrum are defined by the following expressions:

Sa(T ) = agS[

1 + TTB

(η2.5 − 1)]

for T < TB (18)

Sa(T ) = agSη2.5 for TB < T < TC (19)

Sa(T ) = agSη2.5[

1 + TC

T

]

for TC < T < TD (20)

Sa(T ) = agSη2.5[

1 + TCTD

T 2

]

for TD < T < 4 s (21)

where:

ag: design ground acceleration (here: PGA) on soil type A ground,

TB, TC: corner periods of the constant spectral acceleration branch (plateau),

TD: corner period defining the beginning of the constant displacement range,

S: soil factor (see Table 4),

η: damping correction factor (η = 1 for 5% viscous damping).

The shape of the design spectrum is thus determined by the corner periods, soil factor, and the levelof input ground motion. Both, corner periods TB , TC and TD as well as soil factor S are dependenton ‘ground type’ which is mainly distinguished by the average shear-wave velocity of the uppermost 30m vs,30 into 5 different soil classes (Table 4). Both, soil factor and corner periods for the different soilclasses are given in Table 5 and Table 6 for Type 1 and Type 2 design spectra, respectively. Figure 7illustrates the corresponding sets of normalized elastic design spectra.

3.3.3 Indian Standard IS 1893 (Part 1) : 2002 (Bureau of Indian Standards, 2002)

The construction of the horizontal design spectra following the provisions of the Indian standard IS1893 (Part 1) : 2002 [41] can be compared with the procedure of Eurocode 8. The amplitude level ofthe spectrum solely is dependent on the value for peak ground acceleration (PGA). The shape of thehorizontal design spectrum is thus defined by the following expressions:

Sa(T ) = agS[

1 + TTB

(η2.5 − 1)]

for 0 ≤ T < TB (22)

Sa(T ) = agSη2.5 for TB < T < TC (23)

Sa(T ) = agSη2.5[

1 + TC

T

]

for TC < T < 4.0s (24)

(25)

where:

14

DR

AFT


Ground type Description of stratigraphic profileShear wave velocity

vs,30 [m/s]

ARock or rock-like geological formation,

> 800incl. at most 5 m of weaker material at the surface

BDeposits of very dense sands, gravel, or very stiff clay

360-800(at least several tens of m in thickness) characterized bya gradual increase of mechanical properties with depth

CDeep deposits of dense or medium-dense sand, gravel or

180-360stiff clay with thickness from several tensto many hundreds of m

DDeposits of loose-to-medium cohesionless soil

< 180(with or without some soft cohesive layers), or of predominantlysoft-to-firm cohesive soil

ESoil profile consisting of a surface alluvium layer with vs,30

n.avalues of type C or D and thickness H varying between5-20 m underlain by stiffer material with vs,30 > 800 m/s.

Table 4: Ground types.

Ground type Soil factor S TB [s] TC [s] TD [s]

A 1.00 0.15 0.40 2.00B 1.20 0.15 0.50 2.00C 1.15 0.20 0.60 2.00D 1.35 0.20 0.80 2.00E 1.40 0.15 0.50 2.00

Table 5: Values of the parameters describing Eurocode 8 Type 1 spectra.

Ground type Soil factor S TB [s] TC [s] TD [s]

A 1.00 0.05 0.25 1.20B 1.35 0.05 0.25 1.20C 1.50 0.10 0.25 1.20D 1.80 0.10 0.30 1.20E 1.60 0.05 0.25 1.20

Table 6: Values of the parameters describing Eurocode 8 Type 2 spectra.

15

DR

AFT


0

1

2

3

4

5

6

0 0.5 1 1.5 2 2.5 3 3.5 4

No

rmal

ized

sp

ectr

al a

ccel

erat

ion

Period T [s]

Type 1Ground type: A

BCDE

0

1

2

3

4

5

6

0 0.5 1 1.5 2 2.5 3 3.5 4

No

rmal

ized

sp

ectr

al a

ccel

erat

ion

Period T [s]

Type 1Ground type: A

BCDE

Figure 7: Elastic design spectra of Type 1 and Type 2 for ground types AE (prEN 1998-1:200x).

ag: design ground acceleration (here: PGA),

TB, TC: corner periods of the constant spectral acceleration branch (plateau),

TD: corner period defining the beginning of the constant displacement range,

η: damping correction factor (η = 1 for 5% viscous damping).

The periods TB and TC are the only parameters which depend on the soil conditions. An explicit soilamplification factor is not defined. Table 7 describes the three different soil types and assigns theircontrol periods. Unfortunately, no characteristic values of shear wave velocities vs,30 are assigned to the

Spoil type Description of stratigraphic profileShear wave velocity

TB [s] TC [s]vs,30 [m/s]

I

Rock or Hard Soil:

> 400 0.10 0.40well graded gravel and sand gravel mixtures with or withoutclay binder, and clayey sands poorly graded or sand claymixtures (GB, CW, SB, SW, and SC) having N > 30)

II

Medium Soils:

200-400 0.10 0.55a) all soils with 10 < N < 30b) poorly graded sands or gravelly sands with littleor no fines (SP) with N > 15

IIISoft Soils:

< 200 0.10 0.67all soils other than SP with N < 10

Table 7: Soil types, deduced ranges of shear wave velocities as well as corner periods of the horizontaldesign spectra. N is the standard penetration value and values of shear wave velocities, vs,30, are notprovided by the Indian standard IS 1893 (Part 1) [41]; ranges of vs,30 are derived by comparing thestandard penetration values with soil classification schemes of other earthquake codes (e.g., IBC-2006;Turkish code TMPS [44]) providing both values of N and vs,30.

soil classes thus standard penetration test (SPT) values N are the only tangible parameters which allowa classification of the soil conditions. A comparison with different soil classification schemes (e.g., IBC-2006; Turkish seismic code TMPS [44]) providing both SPT-values N and ranges of shear wave velocitiesvs,30 enabled a coarse allocation of vs,30 ranges to the three soil classes. Normalized elastic (R = 1.0)design spectra for soil types I–III are reproduced in Figure 8.

16

DR

AFT


0

1

2

3

4

0 0.5 1 1.5 2 2.5 3 3.5 4

No

rmal

ized

sp

ectr

al a

ccel

erat

ion

Period T [s]

IS 1893 (Part 1) : 2002Soil type: I

IIIII

Figure 8: Elastic design spectra for Soil Types I-III [IS 1893 (Part 1) : 2002] [41].

Note: Since most earthquake codes adopt a comparable procedure in order to generate the designspectra as the one described in Eurocode 8, any new set of site-specific design spectra can be easilyimplemented by the user itself.

3.4 Structural Performance Under Seismic Action

In order to determine the seismic performance of a building, the spectral displacement along its capacitycurve must be determined that is consistent with the seismic demand and at the same time being reducedfor nonlinear effects. Currently a number of different methodologies are available in order to identify theso-called performance point on the capacity spectrum. In the following the CSM as proposed by ATC-40 [45] and FEMA 273 [46] , a recent modification of this procedure, the MADRS method, and thedisplacement coefficient method (DCM) of FEMA-356 [47] with the improvements proposed in FEMA-440 [48] [referred henceforth as improved displacement coefficient method(I-DCM)] will be described todetermine the performance point and thus to establish the basis in order to estimate the structuraldamage state under an estimated seismic demand. All three procedures are implemented in SELENA,such that the user can choose which one to use.

3.4.1 The Capacity Spectrum Method (CSM) as Proposed in ATC-40

The building response (e.g., peak displacement) is determined by the intersection of the seismic demandspectrum and the building capacity curve. The demand spectrum is based on the PESH input spectrumreduced for effective damping (when effective damping exceeds the 5% damping level of the PESH inputspectrum).

The elastic response spectra provided as a PESH input applies only to buildings that remain elastic

17

DR

AFT


during the entire ground shaking time history and have elastic damping values equal to 5%. This isgenerally not true on both accounts. Therefore, elastic response spectra are modified in case of:

a) buildings with elastic damping not equal to 5%, and

b) buildings pushed beyond their elastic limits and thus dissipating hysteretic energy.

Modifications are represented by reduction factors through which the spectral ordinates are dividedto obtain the damped demand spectra. The methodology reduces demand spectra for effective dampinggreater than 5% based on statistically-based formulas of Newmark and Hall [49]. These relationshipsestimate elastic response spectra at different damping ratios B (expressed as a percentage) and representall site classes (soil types) distinguishing between domains of constant acceleration and constant velocity.Ratios of these formulas are used to develop an acceleration-domain (short-period) reduction factor RAand a velocity-domain (1-second spectral acceleration) reduction factor RV, in order to modify the 5%-damped elastic response spectra. These reduction factors are based on effective damping Beff:

Ra(Beff) =2.12

3.21 − 0.68 log(Beff)(26)

Rv(Beff) =1.65

2.31 − 0.41 log(Beff)(27)

where Beff is the effective damping given by the expression:

Beff = Be + Bh (28)

and where Be is the elastic damping and Bh is the hysteretic damping, which is a function of the yieldand ultimate capacity points (see Figure 2 in ATC-40 [45]) as follows:

Bh = 63.7κ

(

Ayi

Au−

Dyi

Du

)

(29)

where κ is a degradation factor that defines the effective amount of hysteretic damping as a function ofearthquake duration and energy-absorption capacity of the structure during cyclic earthquake load (seeHAZUS documents, Table 5.18 in [2]), and Ayi and Dyi are obtained through an iterative process as apart of the capacity curve bilinearization.

Following the recommendations of Newmark and Hall [49], Be is the elastic (pre-yield) damping ofthe model building type, which is:

5% for mobile homes (MH),

5–7% for steel buildings (S),

7% for concrete (C) and pre-cast concrete buildings (P),

7–10% for reinforced masonry buildings (RM),

10% for un-reinforced masonry (URM) and masonry buildings (M),

10–15% for wood buildings (W).

The methodology recognizes the importance of the duration of ground shaking on building response byreducing the effective damping (i.e., κ-factors) as a function of shaking duration. Dependent on themagnitude of the scenario earthquake, the effective damping is based on the assumption of differentground shaking durations:

magnitude M ≤ 5.5: short duration

18

DR

AFT


magnitude 5.5 < M < 7.5: moderate duration

magnitude M ≥ 7.5: long duration

The new demand spectral acceleration Sa(T ) in units of gravity [g] is defined at short periods (accelerationdomain), long periods (velocity domain), and very long periods (displacement domain) using the 5%damped response spectrum and dividing by the before mentioned factors following the expressions:

Sa(T ) = Saasi(0.4 + T/TA)/Ra(Beff) for 0 < T < Ta (30)

Sa(T ) = Saasi/Ra(Beff) for Ta < T < Tavb (31)

Sa(T ) = (Saali/T )/Rv(Beff) for TAVb < T < Tvd (32)

Sa(T ) = (SaaliTvd/T 2)/Ra(Beff) for T > Tvd (33)

where:

SASi: 5% damped, short-period spectral acceleration for site class i (in [g])

Ssli: 5% damped, 1-second (long) period spectral acceleration for site class i (in [g])

Btvd: value of effective damping at the transition period Tvd

Tavb: transition period between acceleration and velocity domains as a function of the effective dampingat this period which is defined by the equation:

Tavb = TaviRA(Btavb)

RB(Btavb)(34)

where:

Tavi: transition period between 5%-damped constant spectral acceleration and 5%-damped constantspectral velocity for site class i

Btavb: value of effective damping at the transition period Tavb.

The transition period Tvd is independent of effective damping and only depends on the moment magni-tude, as previously said.

Example 3.2 Performance point calculation between inelastic demand spectra for site classes B, C andD and a given capacity curve by the Capacity spectrum method (CSM) described in ATC-40 Chapter2.4.1 [45]

Steps:

1. Determination of the structure’s yield capacity and ultimate capacity:

19

DR

AFT


0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50 60

Sp

ectr

al a

ccel

erat

ion

Sa

[g]


(Dy, Ay)

(Du, Au)

capacity curve

e.g.,: capacity curves as given in HAZUS, here: C1Mfor moderate-code design

• yield capacity point: Ay = 0.104 g, Dy = 0.58in. = 1.47 cm.

• ultimate capacity point: Au = 0.312 g, Du =6.91 in. = 17.55 cm

2. Determination of effective damping Beff by calculating hysteretic damping Bh according to Ta-ble 5.18 in HAZUS99 [50]: degradation factor, κ, depending on earthquake duration and energy-absorption capacity of the structure during cyclic earthquake load

moderate durationmoderate-code designbuilding type C1M

⇒ κ = 0.4

Bh = 63.7κ

(

Ay

Au−

Dy

Du

)

= 4%

Be = 7.0%

Beff = Bh + Be = 11.1%

3. Calculation of reduction factors Ra and Rv, short-period and long-period spectral accelerations aswell as transition period Tavi

ParameterSite Class

B C D

Ra(Beff) = 2.123.21−0.68 log(Beff)

[Eq. (26)] 1.35

Rv(Beff) = 1.652.31−0.41 log(Beff)

[Eq. (27)] 1.35

Sasbi = Sasi/Ra [Eq. (31)] 0.37 g 0.44 g 0.52 gSalbi = (Sali/T )/RV [Eq. (32)] 0.16 g 0.26 g 0.32 g

Savb = SaviRa

Rv[Eq. (34)] 0.43 g 0.57 g 0.62 g

4. Generation of reduced (inelastic) demand spectrum (damping ξ = 11.1%):

20

DR

AFT


0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50 60

Sp

ectr

al a

ccel

erat

ion

Sa

[g]


Site Class: Bdemand spectra:

Be = 5 %Beff = 11.1 %

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50 60

Sp

ectr

al a

ccel

erat

ion

Sa

[g]


Site Class: Cdemand spectra:

Be = 5 %Beff = 11.1 %

0

0.2

0.4

0.6

0.8

1

0 10 20 30 40 50 60

Sp

ectr

al a

ccel

erat

ion

Sa

[g]


Site Class: Ddemand spectra:

Be = 5 %Beff = 11.1 %

1. Calculation of the spectral accelerations and spectral displacements at the site in question takinginto account soil response, so that the elastic response spectrum can be generated.

2. Creation or selection of a capacity curve for the respective building type reflecting the building’sperformance under an increasing, laterally applied (earthquake) load.

3. Determination of effective damping Beff by specifying elastic damping Be and by computing thehysteretic damping Bh. Based on this the calculation of both reduction factors RA and RV can berealized.

4. Reduction of the elastic response spectrum by reduction factors RA and RV to account for theincreased damping that occurs at higher levels of ground motion and consequently building response(non-linear behavior).

5. Superposition of the building capacity curve with the modified (inelastic) response spectrum (de-mand curve). The resulting building displacement is estimated from the intersection of the buildingcapacity curve and the response spectrum (performance point; see also Figure 9).

6. The estimated building displacement is later used to define the damage degree at the intercept ofthe fragility curve and the damage probability curve (see Figure 10).

21

DR

AFT


Sp

ectr

al a

ccel

erat

ion

Sa

[g]


performance point

(dp , ap)

PESH spectrum (elastic)reduced spectrum (inelastic)

capacity curve

Figure 9: Estimation of building displacement from a given PESH input.

0

0.2

0.4

0.6

0.8

1

Dam

age

pro

bab

ilit

y P

(d

s |

Sd)


no damage

slight

moderate

extensive

complete

dp

dp − expected spectraldisplacement

damage state: slightmoderateextensivecomplete

Figure 10: The expected displacement (obtained from the performance point) is overlaid with the fragilitycurves in order to compute the damage probability in each one of the different damage states.

22

DR

AFT


3.4.2 The Modified Capacity Spectrum Method (MADRS)

The conventional capacity spectrum method (ATC-40 [45]) uses the secant period as the effective linearperiod in determining the maximum displacement (performance point). This assumption results in themaximum displacement occurring at the intersection of the capacity curve for the structure with thedemand curve for the effective damping in ADRS format. However it has been shown in several studiesthat this method can not be used with a non-IBC response spectrum and that it does not provide anaccurate performance point in some cases. Later, some improvements of the method have been publishedin FEMA 440 [48]. This revised methodology has some advantages. First, it provides the engineer witha visualization tool by facilitating a direct graphical comparison of capacity and demand. Second, thereare very effective solution strategies for equivalent linearization that rely on a modified ADRS demandcurve (MADRS) that intersects the capacity curve at the maximum displacement. As it is also explicitlystressed in FEMA 440 “the user must recognize that the results are an estimate of median response andimply no factor of safety for structures that may exhibit poor performance and/or large uncertainty inbehavior”. Furthermore it should be noted that the results of the MADRS method as described in thefollowing may not be reliable for extremely high ductility values, e.g., greater than 10 to 12.

The MADRS method basically relies on the determination of effective damping, βeff, and effectiveperiod, Teff, with which a maximum spectral displacement can be derived. This in turn matches with theintersection point of the radial effective period (radiating line from the origin in the Sa-Sd-domain) and theADRS demand for the effective damping (Figure 11). The effective period of the improved procedure Teff

Figure 11: Modified acceleration-displacement response spectrum (MADRS) for use with secant periodTsec. Figure taken from FEMA 440 [48].

is generally shorter than the secant period Tsec defined by the point on the capacity curve correspondingto the maximum displacement dmax. The effective acceleration aeff is not meaningful since the actualmaximum acceleration amax must lie on the capacity curve and coincide with the maximum displacementdmax. Multiplying the ordinates of the ADRS demand corresponding to the effective damping βeff by themodification factor:

M =amax

aeff(35)

results in the modified ADRS demand curve (MADRS) that may now intersects the capacity curve atthe performance point. Since the acceleration values are directly related to the corresponding periods,the modification factor can be calculated as:

M =

(

Teff

Tsec

)2

=

(

Teff

T0

)2 (

T0

Tsec

)2

(36)

23

DR

AFT


where(

T0

Tsec

)2

=

√

1 − α(µ − 1)

µ(37)

and where the post-elastic stiffness, α, and the ductility demand, µ, are:

α =

api−ay

dpi−dy

ay

dy

(38)

and

µ =dpi

dy, (39)

respectively.

Equivalent linearization procedures applied in practice normally require the use of spectral reductionfactors to adjust an initial response spectrum to the appropriate level of effective damping βeff. Thesefactors are a function of the effective damping and are termed damping coefficients B(βeff). They areused to adjust spectral acceleration ordinates as follows:

(Sa)β =(Sa)5%B(βeff)

(40)

where

B(βeff) =4

5.6 − log(βeff)(41)

with βeff given in [%].

Since the effective period Teff and the effective damping βeff are both functions of ductility demand,the calculation of a maximum displacement using equivalent linearization is not direct and requires aniterative procedure (Figure 11).

Both, effective damping and period are strongly dependent on the building’s inelastic behavior. WithinFEMA 440 [48] three different inelastic hysteretic systems have been studied including bilinear hysteretic,stiffness degrading and strength degrading behavior. The procedure incorporated in SELENA and thefollowing description is based on the stiffness degrading hysteretic model. The effective viscous dampingeff can be calculated by the following equations dependent on ductility demand. Values of the coefficientsA to F can be found in Table 8.

βeff = A(µ − 1)2 + B(µ − 1)3 + β0 for µ < 4.0 (42)

βeff = C + D(µ − 1) + β0 for 4.0 ≤ µ ≤ 4.0 (43)

βeff = E[

F (µ−1)−1F (µ−1)2

Teff

T0

]

for µ > 6.5 (44)

The effective period values, Teff, are to be computed using the following equations. Values of the

α [%] A B C D E F

0 5.1 -1.1 12 1.4 20 0.622 5.3 -1.2 11 1.6 20 0.515 5.6 -1.3 10 1.8 20 0.3810 5.3 -1.2 9.2 1.9 21 0.3720 4.6 -1.0 9.6 1.3 23 0.34

Table 8: Coefficients in order to calculate effective damping values βeff.

24

DR

AFT


coefficients G to L can be found in Table 9.

Teff = [G(µ − 1)2 + H(µ − 1)3 + 1]T0 for µ < 4.0 (45)

Teff = [I + J(µ − 1) + 1]T0 for 4.0 ≤ µ ≤ 4.0 (46)

Teff ={

K[√

µ−11+L(µ−2) − 1

]

+ 1}

T0 for µ > 6.5 (47)

α [%] G H I J K L

0 0.17 -0.032 0.10 0.19 0.85 0.002 0.18 -0.034 0.22 0.16 0.88 0.025 0.18 -0.037 0.15 0.16 0.92 0.0510 0.17 -0.034 0.26 0.12 0.97 0.1020 0.13 -0.027 0.11 0.11 1.00 0.20

Table 9: Coefficients in order to calculate effective damping values Teff.

In order to find the performance point (di, ai), FEMA 440 [48] provides three alternative procedures,which all are based on reducing the initial ADRS demand spectrum by the effective viscous dampingβeff. One of these procedures consists in the automated derivation of a ‘locus’ of possible performancepoints. This by generating a number of modified ADRS (MADRS) demand spectra for different valuesof ductility demand µ. As Figure 12 illustrates, the performance point is defined by the intersection ofthe capacity spectrum and the line being described by all respective performance points on the differentMADRS curves. In the following, the single steps of the method are subsequently described taking upthe previous example with the given capacity curve and NEHRP site class C.

Figure 12: Finding the performance point (red star) using the modified acceleration-displacement responsespectrum (MADRS). Figure taken from FEMA 440 [48].

Example 3.3 Performance point calculation following the MADRS method (Chapter 2.4.2 in [48])

Steps:

25

DR

AFT


1. Selection of a spectral representation of the ground motion of interest with an initial damping βi

(i.e., normally 5%) and conversion into an ADRS ⇒ elastic design spectrum for NEHRP site classC (see Example 3.2).

2. Generation or selection of a capacity curve ⇒ capacity curve described by yield (ay, dy) and ulti-mate capacity points (au, du); in case of a generated capacity curve, the development of a bilinearrepresentation has to be conducted (ATC-40 [45]) ⇒ here: capacity curve as given in HAZUS formodel building type C1M and moderate-code seismic design level (ses HAZUS [2], Table 5.7 b):ay = 0.1044 g, dy = 0.58 in. = 1.47 cm, au = 0.312 g, and du = 6.91 in. = 17.55 cm.

3. Calculation of effective damping, βeff, and modification factor, M for integer increments of ductilityµ (µ = 2, 3, 4, . . .) ⇒ initial (elastic) period T0 (µ = 1): T0 = 2π

√

dy/ay0.7542 s ⇒ post-elastic

stiffness parameter: α =

api−ay

dpi−dyay

dy

= 0.1828 [Eq. (38)]

ParameterDuctility µ

2 3 4 5 6 7dpi = µdy [m] [Eq. (39)] 0.0294 0.0441 0.0588 0.0735 0.0882 0.1029βeff [%] [Eqs. (42)-(44)] 8.686 15.606 18.740 20.143 21.546 22.654Teff [sec] [Eqs. (45)-(47)] 0.836 0.997 1.109 1.194 1.278 1.332Tsec [sec] [Eq. (37)] 0.981 1.118 1.212 1.282 1.335 1.378B(βeff) [Eq. (41)] 1.163 1.402 1.499 1.540 1.581 1.613

M =“

Teff

Tsec

”2

[Eq. (36)] 0.727 0.795 0.838 0.867 0.916 0.935

4. Adjustment of the initial ADRS to the effective damping βeff ⇒ reduction of spectral acceleration

ordinates (Sa)5% by damping coefficients B for all considered ductility values µ: (Sa)β = (Sa)5%B(βeff)

[Eq. (40)]

5. Multiplication of the ADRS for βeff by the modification factor M reduction of spectral accelerationordinates (Sa)β by damping coefficients M for all considered ductility values µ.

6. Generation of possible performance point by the intersection of radial secant period Tsec withthe MADRS for all considered ductility values µ ⇒ Determination of performance point by theintersection of the locus line with the capacity spectrum:

26

DR

AFT


0

0.2

0.4

0.6

0.8

0 10 20 30

Sp

ectr

al a

ccel

erat

ion

Sa

[g]


ap

dp

T0 (µ=1)

Tsec (µ=2)

Tsec (µ=3)...

Tsec (µ=7)

ADRS: µ=1

MADRS: µ=2...

µ=7

locus of possibleperformance points

3.4.3 Improved Displacement Coefficient Method (I-DCM)

The displacement coefficient method modifies the displacement demand of the equivalent linear singledegree of freedom (SDOF) system by multiplying it by a series of coefficients in order to generate anestimate of the maximum displacement demand of the nonlinear oscillator. The process begins with thegeneration of the capacity curve of the nonlinear oscillator. The effective period of the system is thencomputed as (Figure 13):

Te = 2π

√

Dy

Ay(48)

When plotted on an elastic response spectrum representing the seismic ground motion, as peak spectralacceleration, Sa, vs. period, T, the spectral acceleration demand of the equivalent linear SDOF system,Sel

a , can be computed (Figure 13). The peak elastic spectral displacement demand, Seld is then directly

related to the Sela by:

Seld =

T 2e

4π2Sel

a (49)

The target displacement, δt, is then computed as:

δt = C1C2Seld (50)

where:

C1 = Modification factor to relate the expected maximum displacement demand of a nonlinear oscillatorwith elastic-perfectly-plastic (EPP) hysteretic properties to the peak displacement demand of thelinear oscillator,

27

DR

AFT


Dy

Sa

(D , A )yy

T = 2πe

Demand Curve

TePeriod, T

Sael

δ = C C St 1 2 ael

4π2

T e

2

Capacity CurveAy

Sd

Figure 13: Schematic illustraion of process I-DCM which is used to compute the target displacementdemand of a nonlinear oscillator for a given capacity curve and response (demand) spectrum

.

C2 = Modification factor to represent the effect of pinched hysteretic shape and stiffness degradationon the maximum displacement response.

The coefficeints C1and C2 can be computed using the approximation relationships given in FEMA-440 [48]:

C1 = 1 + R−1aT 2

e(51)

C2 = 1 + 1800 (R−1

Te)2 (52)

where

R = Ratio of elastic strength demand to the calculated strength capacity; R=Sela /Ay,

a = Equation constant; a is equal to 130, 90, and 60 for NEHRP site classes B, C, and D, respectively.Table 10 summarizes the assumed values for the parameter a for different site classes and spectralshapes.

3.5 Fragility Curves and Damage State Probability

The conditional probability of being in, or exceeding a particular damage state, ds, given by the spectraldisplacement Sd (or other seismic demand parameter) is defined by the following equation:

P (ds|Sd) = Φ

[

1

βdsln

(

Sd

Sd,ds

)]

(53)

where

Sd,ds: median value of spectral displacement at which the building reaches the threshold of damagestate ds,

28

DR

AFT


Spectral shape Soil type a

IBC A 130B 130C 90D 60E 60

Eurocode I and II A 130B 90C 60D 60E 60

Indian Code I 90II 60III 60

Table 10: Assumed values for the parameter a for different site classes and spectral shapes.

βds: is the standard deviation of the natural logarithm of spectral displacement for damage state ds,

Φ(·): is the standard normal cumulative distribution function.

In HAZUS99, for instance, both mean displacement threshold of damage state and its correspondingstandard deviation βds are table values (see Table 5.9) which depend on the model building type and itsseismic design level. However, it should be regarded that the parameters defining the fragility functionsfor a certain building type are closely connected to its respective capacity curve.

Cumulative probabilities are defined to obtain discrete probabilities of being in each of the five differentdamage states (Figure 14).

The final damage results are given as absolute square meters of the respective damaged building type,so that users are able to present and further process these results using a spreadsheet program (MS Excel,OpenOffice, etc.) or any other software applications in any desired format (e.g., as percentage of builtarea [m2] normalized by the total built area in each geographical unit or by the total built area in thestudied region, i.e., summed all geographical units). Nonetheless, results can also be given as absolutenumbers of damaged buildings.

3.6 Economic Losses

The SELENA software can also estimate the total amount of economic losses (in any input currency) inany geographical unit caused by the structural damage.

The economic losses for building repair (and in case of complete damage for replacement) are computedin the following way:

Leco = Cr

Not∑

i=1

Nbt∑

j=1

Nds∑

k=1

Ai,jPj,kCi,j,k (54)

where Not is the number of occupation types, Nbt is the number of building types, Nds is the number ofdamage states, and where

Cr: regional cost multiplier (currently is set to 1.0, but can have different values for each geographicalunit in order to take into account the geographic cost variations),

Ai,j : built area of the model building type j in the occupancy type i (in [m2])

29

DR

AFT


0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

0.45

0.5

Dis

cret

e d

amag

e p

rob

abil

ity

P

Damage state ds

none slight moderate extensive complete

Figure 14: Discrete damage probabilities derived from the cumulative damage probabilities for an ex-pected displacement as illustrated in Figure ??.

Pj,k: damage probability of a structural damage k (slight, moderate, extensive or complete) for themodel building type j,

Ci,j,k: cost of repair or replacement (by [m2]) in the input currency of structural damage k for occupancytype i and model building type j (provided by input files eloss .txt).

In the current version of SELENA only the direct economic losses caused by structural damageare computed. Those being caused by non-structural damage (acceleration sensitive damage) are notconsidered. The absolute values for building repair costs in the different damage states will be determinedby the user. HAZUS expresses the cost of damage for damage states slight, moderate and extensive as apercentage of the complete damage state:

slight damage: 2% of complete damage,

moderate damage: 10% of complete damage,

extensive damage: 50% of complete damage.

These relationships are consistent with those damage ratios given in ATC-13 [51]. However, more reliableor suitable values can be defined by the user.

It should be regarded that economic losses will only be calculated by SELENA if the user chooses theanalysis type dependent on damaged building area. In case of an analysis dependent of the number ofdamaged buildings, no economic losses are computed.

30

DR

AFT


3.7 Humanloss — Casualties

The methodology applied in order to calculate the number of human casualties follows basically theHAZUS approach but is somewhat simplified using the formulas given by Coburn and Spence [52]:

K = Ks + K ′ + K2 (55)

where

Ks: number of casualties due to structural damage

K ′: number of casualties due to non-structural damage

K2: number of casualties due to follow-on hazards, such as landslides, fires, etc.

The above equation can also be modified such that the level of injury (severity) is considered:

Ki = Ks

i + K ′

i + K2i (56)

where i is representing the level of severity, ranging from light injuries (i = 1), moderate injuries (i = 2),heavy injuries (i = 3), to death (i = 4). A more detailed description of the severity levels is givenin Table 11. However, the loss model applied in the current version of SELENA is only consideringthe direct human losses caused by structural damage not due to non-structural damage or follow-onhazards. By using SELENA, the number of human losses (casualties) can be computed using two different

Injury Level Description Examples

Severity 1

Injuries requiring basic medical aid that - sprainscould be administered by paraprofessionals. - severe cuts requiring stitchesThese types of injuries would require bandages or - minor burns (first or second degree on aobservation.* small part of the body)

- bumps on the head without loss of consciousness

Severity 2

Injuries requiring a greater degree of - bump on the head that causes loss of consciousnessmedical care and use of medical - fractured bonestechnology such as x-rays or surgery, but not expected - dehydration or exposureto progress to a life threatening status.

Severity 3

Injuries that pose an immediate life - punctured organsthreatening condition if not treated - other internal injuriesadequately and expeditiously. - spinal column injuries

- crush syndrome

Severity 4instantaneously killed ormortally injured.

Table 11: Injury Classification Scale according to HAZUS. *Injuries of lesser severity which can be selftreated are not covered by HAZUS.

methodologies:

1. Basic methodology → in case that no detailed information on population distribution is availableor can not be inferred from available data,

2. ‘HAZUS’ methodology → in case that detailed information on population distribution is available.

In order to also cover extreme cases of occupancy which are strongly dependent on the time of theday (i.e., school occupancy only during daytime), the number of casualties will be computed for threedifferent times of the day:

1. nighttime scenario (called 02:00 am): i.e., earthquake striking during nighttime.

31

DR

AFT


2. daytime scenario (called 10:00 am): i.e., earthquake striking during day time.

3. commuting time scenario (called 05:00 pm): i.e., earthquake striking during the commuting time(rush hour).

These scenarios are expected to generate the highest casualty numbers for the population at home (night-time), the population at work/education (daytime), and the population during rush hour, respectively.

3.7.1 The Basic Methodology

The number of casualties due to direct structural damage for any given structural type, level of buildingdamage, and injury severity can be calculated by:

Ks

i = {Injuries (severity i)} =

Nbt∑

j=1

Nds∑

k=1

Ccsr

i,j Pj,kNpop

j (57)

where:

Ccsri,j : casualty rate of severity i for damage state j as provided by input files injury1.txt to injury4.txt

(these statistical values have to be provided by local authorities),

Pj,k: structural damage probability for the kth damage type (k = 1 slight, k = 2 moderate, k = 3extensive, k = 4 complete or complete with collapse) for the jth model building type.

Npopj : number of people in the jth model building type.

The total number of people in all buildings of the j:th model building type (MBT), for one geographicalunit (i.e., census tract) at a specific time period (time of the day), is computed in a simplified way:

Npop

j = NtpCpoCombt

j (58)

where Ntp is the total number of people living in the respective geographical unit provided by input filepopulation.txt, Cpo is the percentage of people staying indoors or outdoors dependent on the timeof the day provided by input file poptime.txt (compare to Table 12), and Combt

j is the percentage ofoccupancy class for the jth model building type (MBT) provided by input file ocupmbtp.txt.

Occupancy type night (2:00 am) day (10:00 am) commuting (5:00 pm)

INDOOR 98% 90% 36%OUTDOOR 2% 10% 64%

Sum∑

100% 100% 100%

Table 12: Population percentages indoors and outdoors dependent on the time of the day. Note thatthese values are strongly dependent on the country and its cultural peculiarities and consequently mayvary considerably.

3.7.2 The HAZUS Methodology

The total population of each census tract is classified into five different groups:

1. residential population,

2. commercial population,

32

DR

AFT


3. education population,

4. industrial population,

5. hotel population.

The default population distribution is calculated for the three times of the day for each census tract.Table 13 provides the relationships used to determine the default distribution. Each element of the table

Distribution of people in census tractOccupancy 2:00 am 2:00 pm 5:00 pm

Indoors

residential (0.999) 0.99 (NRES) (0.70) 0.75 (DRES) (0.70) 0.50 (NRES)commercial (0.999) 0.02 (COMW) (0.99) 0.98 (COMW) + 0.98 [0.50(COMW) +

(0.80) 0.20 (DRES) + 0.10 (NRES) + 0.70 (HOTEL)]0.80 (HOTEL) + 0.80 (VISIT)

educational – (0.90) 0.80 (AGE 16) + 0.80 (COLLEGE)(0.80) 0.50 (COLLEGE)

industrial (0.999) 0.10 (INDW) (0.90) 0.80 (INDW) (0.90) 0.50 (INDW)hotels 0.999 (HOTEL) 0.19 (HOTEL) 0.299 (HOTEL)

Outdoors

residential (0.001) 0.99 (NRES) (0.30) 0.75 (DRES) (0.30) 0.50 (NRES)commercial (0.001) 0.02 (COMW) (0.01) 0.98 (COMW) + 0.02 [0.50 (COMW) +

(0.20) 0.20 (DRES) + 0.10 (NRES) + 0.70 (HOTEL)] +0.20 (VISIT) + 0.50(1-PRFIL)

0.50 (1-PRFIL) 0.05 (POP) [0.05 (POP) + 1.0 (COMM)]educational – (0.10)0.80 (AGE 16) + (0.20) 0.50 (COLLEGE)

0.20 (COLLEGE)industrial (0.001) 0.10 (INDW) (0.10) 0.80 (INDW) (0.10) 0.50 (INDW)

hotels 0.001 (HOTEL) 0.01 (HOTEL) 0.001 (HOTEL)

Table 13: Default relationships for estimating population distribution (taken from HAZUS) where POP

is the census tract population taken from HAZUS, DRES is daytime residential population inferred fromcensus, NRES is nighttime residential population inferred from census data, COMM is the number of peoplecommuting inferred from census data COMW is the number of people employed in the commercial sectorINDW is the number of people employed in the industrial sector GRADE is the number of students in gradeschool (usually under 17 years old) COLLEGE is the number of students on college and university campusesin the census tract (over 17 years old), HOTEL is the number of people staying in hotels in the censustract, PRFIL is a factor representing the proportion of commuters using automobiles, inferred from profileof the community (0.60 for dense urban areas, 0.80 for less dense urban or suburban areas and 0.85 forrural). Default value is 0.80. VISIT is the number of regional residents who do not live in the study area,visiting the census tract for shopping and entertainment. Default is set to zero.

contains two multipliers of which the second one indicates the fraction of a population component (e.g.NRES) present in an occupancy type at a particular scenario time. The first multiplier (given in brackets)divides this population component into indoor and outdoor occupancy. For example: At 02:00 am, thedefault is that 99% of the nighttime residential population (NRES) will be in residential occupancy while99.9% of those will be indoors. If more detailed information is available on these issues, these factors canbe changed in the m-file humanlosshz.m. The methodology takes into account a wider range of causalrelationships in the casualty modeling. It is an extension of the model proposed by Stojanovski andDong [53].

By using an event tree as shown in Figure 15 and multiplying the population in each of the occupancytypes and model building types by the damage probabilities and casualty rates, the total number ofcasualties for each severity level can be estimated. Figure 15 is illustrating an event tree for indoorcasualties (but also outdoor casualties are contemplated within this methodology). The outdoor casualtymodel tries to quantify the number of casualties outside of buildings due to falling materials with respect

33

DR

AFT

Selena 4 INSTALLATION

Figure 15: Indoor casualty event tree model (taken from HAZUS). Bridge casualties are currently notimplemented in SELENA.

to areas where people congregate such as sidewalks. To accomplish this, the number of people on sidewalksor similar exterior areas is estimated from Table 13. The table is designed to prevent double counting ofcasualties from outdoor falling hazards with building occupancy casualties.

4 Installation

IN this section the details for intalling SELENA is treated, both for Windows stystems and POSIX(Linux/Unix) systems.

4.1 System Requirements and Resent Code Changes

There has been a change in the system requirements from version 4.x to version 5.x (and above) ofSELENA. Starting with version 5.0.0 of SELENA, the Matlab m-code has been translated into C-codewhich allows SELENA to run without using Matlab; it is, however, still possible to use SELENA fromMatlab. Furthermore, the m-code has been changed in such way that it now can run without any specialMatlab Toolboxes (which was required before) and it now also now runs using the free (open source)Matlab clone Octave (http://www.gnu.org/software/octave/).

In order to avoid Matlab toolbox dependencies, SELENA now uses the open source GNU ScientificLibrary (GSL) which is available for most systems, such as, Linux/Unix, MacOS, and Windows. Anotherchange is that SELENA previously used some input files in Matlab’s binary mat-format. This has nowbeen changed and SELENA now only uses input files in a plain (ascii) text format.

4.1.1 Installing the GNU Scientific Library

The GNU Scientific Library can be found here: http://www.gnu.org/software/gsl/

The GSL for Windows must be installed in a directory where Matlab/Octave or the stand-alone bi-nary selena.exe can find it. The easiest way is just to copy the libgsl.dll and libgslcblas.dll files toC:\WINDOWS\System32\ These files can be found at: http://gnuwin32.sourceforge.net/packages/gsl.htmor in the dll/bin folder (see also Section B.1).

For Linux, the GSL is probably included in the package system for your Linux distribution so that

34

http://www.gnu.org/software/octave/

http://www.gnu.org/software/gsl/

http://gnuwin32.sourceforge.net/packages/gsl.htm

DR

AFT


you can use your package installer (yast, emerge etc.) to install it. For example, on a Gentoo Linuxdistribution just type,

# emerge gsl

If GSL is not included in your Linux distribution’s package manager system then you can install it fromsource which is described at the GSL webpage.

4.2 The Directory Structure of SELENA

Unpacking the compressed (zip) file onto your computer automatically creates the main folder SELENAunder which a number of sub-folders can be found:

examples src

gnumex userman

include m_files

dll mexopts

The dll folder contain Windows specific files (for GSL) which is needed for building the Windowsbinaries (both stand-alone application and the mex/oct-files). The examples folder contain some exampleinput files which can be used to test SELENA, the m files folder contain the m-files and the mex/oct-files Matlab/Octave (e.g., the mex/oct files must be copied from the src after compilation). The src

contain the C-source files (and Makefiles) for both the stand-alone application and the mex/oct-files, andthe include folder contain the header files for C-code. The mexopts folder contain two bat-files whichis used when building the mex-files using the MinGW compiler, and finally, the userman folder containvarious files for the user manual.

The main folder also contain the four text:

COPYING build_mexfiles.m

Make.inc build_mexfiles_mingw.m

README.txt build_mexfiles_win.m

README_INSTALL_MinGW_Windows.txt build_oct_files.m

README_SVN.txt

where the two m-files, build mexfiles mingw.m and build mexfiles win.m, are for building themex-files using Windows, build oct files.m is for building the oct-files (on all systems) and Make.inc

is a file for setting compiler options for Linux/Unix. Furthermore the README.txt is a quick guide onhow to install and run SELENA (and the README SVN.txt contains some information for developers onsetting svn keywords). Also for license information see the COPYING file. For more build and compilationinstructions, see Appendix B.

4.3 The SELENA m-files

The SELENA comprises 34 different m-files files (*.m) which are consecutively accessed during the pro-gram sequence. Their respective functions and tasks are briefly described below:

selena gui.m core file that starts the graphical user interface (calling for startwin.m).

selena.m core file (the command line interface of SELENA).

startwin.m initialization of the window environment in order to choose between a probabilistic, a de-terministic or an analysis based on real-time data.

dettool.m Function (called by startwin.m) to initialize the window environment for a deterministicanalysis.

35

DR

AFT


probtool.m Function (called by startwin.m) to initialize the window environment for a probabilisticanalysis.

realtool.m Function (called by startwin.m) to initialize the window environment for an analysis withreal-time data (grid pattern shaking scenario).

computetool.m Function (called by dettool.m) which starts the main processes of a seismic risk com-putation for a deterministic earthquake.

computetoolp.m Function (called by probtool.m) which starts the main processes of a seismic riskcomputation based on a probabilistic shake map.

computetoolr.m Function (called by realtool.m) which starts the main processes of a seismic riskcomputation based on real-time data.

gmotion.m Function which gets the ground motion at the center of each geographical unit from a de-terministic earthquake (numerous attenuation relationships are provided, while new attenuationrelations can be easily implemented) and computes ground-motion amplification using the factorsas e.g., given in IBC-2006 [8] (called by computetool.m)

att sub.m Function with attenuation relationships from different authors which provides the ground-motion values (units of [g]) for PGA, Sa at 0.3s and Sa at 1.0s (called by gmotion.m).

dtorry.m Function used to compute the closest distance from a point (latitude, longitude) to a segment(lat1, lon1)-(lat2, lon2) (called by gmotion.m).

gmotionp.m Function which amplifies the ground motion at the center of each geographical unit fromprobabilistic shaking maps using the factors as e.g., given in IBC-2006.

gridtogeounit.m Function selecting the concerning nodes of the grid pattern, making a statistical eval-uation of their ground-motion ordinates (mean value, standard deviation), and assigning them tothe centers of the geographical units (called by computetoolr.m).

damagep.m Function which computes the probability of damage for the building stock using the capacityspectrum method (called by computetool.m, computetoolp.m, and computetoolr.m)

spectralshape.m Computation of spectral ground-motion ordinates following different code provisions,e.g., IBC-2006 (called by damagep.m)

csm.m Performance point calculation based on the “traditional” capacity spectrum method (CSM) fol-lowing ATC-40 [45] (Procedure A) (called by damagep.m).

madrs.m Performance point calculation by using the modified capacity spectrum method (MADRS)following FEMA 440/ATC-55 (called by damagep.m).

curveintersect.m Function which finds the intersection points of two curves in the X-Y plane (calledby csm.m and madrs.m).

local parseinputs.m Script used by curveintersect.m.

mminvinterp.m 1-D inverse interpolation (called by curveintersect.m)

squaredam.m fFnction which computes the absolute square meters of damaged built area for eachmodel building type in each geographical unit (called by computetool.m, computetoolp.m, andcomputetoolr.m)

numdam.m Function which computes the absolute number of damaged buildings for each model buildingtype in each geographical unit (called by computetool.m, computetoolp.m, and computetoolr.m)

36

DR

AFT


losssqm.m function which computes the total economic losses due to structural damage (called bycomputetool.m, computetoolp.m, and computetoolr.m).

tree.m Function used to fit the damage estimation results coming from each branch of the logic tree toa normal distribution function; computes median (mean) value and 16% and 84% fractiles (calledby computetool.m, computetoolp.m, and computetoolr.m).

treeloss.m Function used to fit the economic loss results coming from each branch of the logic tree toa normal distribution function; computes median (mean) value and 16% and 84% fractiles (calledby computetool.m, computetoolp.m, computetoolr.m).

meanest.m Function to compute (estimate) the mean and the variance from a set of data when the vari-ance is unknown. The confidence intervals are therefore obtained using the Student t-distributionfor a chosen alpha (0.16 and 0.84) (16% and 84% fractiles)1 This function is used by tree.m,treemdr.m and treeloss.m.

wfigmngr1.m Function to manage SELENA windows.

wfighelp1.m Function to manage SELENA help in main window.

wfigobj1.m Function to manage objects in windows.

humanloss.m Function to compute the number of human casualties according to the basic methodology(called by computetool.m , computetoolp.m , and computetoolr.m).

humanlosshz.m Function to compute the number of human casualties according to the HAZUS-methodology(called by computetool.m , computetoolp.m , and computetoolr.m).

distance.m Function which computes the distance between points on a sphere.

deg2km.m Function that converts distance from degrees to kilometers.

km2deg.m Function that converts distance from kilometers to degrees.

4.3.1 The SELENA mex-files

The mex-files currently implemented in SELENA are:

att_sub.mex* gsl_interpolate.mex* madrs.mex* tinv.mex*

csm.mex* humanloss.mex* meanest.mex* tree.mex*

curveintersect.mex* humanlosshz.mex* normcdf.mex* treeloss.mex*

damagep.mex* imp_dcm.mex* numdam.mex* treemdr.mex*

gmotion.mex* interp1.mex* spectralshape.mex*

gmotionp.mex* losssqm.mex* squaredam.mex*

These files have the same functionality as the corresponding m-files but runs much faster. The fileextension of the mex-files depends of which architecture you are using; Windows 32-bit has the extension∗.mexw32, Linux x86 ∗.mexlx, Linux x86 64 ∗.mexa64, Intel MacOS X ∗.mexmaci etc. The biniarypackages for SELENA comes with pre-compiled mex-files for Windows 32-bit and Linux 32 and 64 bit.These files can be found in the m files folder and to use them one just need to set the Matlab pathusing the menu in the Matlab GUI or with (add your path),

>> addpath(’/home/<the user>/selena/m_files’)

1For different alpha please change the value in the function callback.

37

DR

AFT

Selena 5 RUNNING SELENA

which also can be added to your startup.m file to permanently add the path.

There are also m-files available in the same folder which can be useful for evaluating and testing thecode (easy to add plots, print intermediate results etc.) If Matlab finds a mex-files with the name as anm-file then the mex-file will take precedece so to use the m-files one need to remove the mex-files fromthe m files folder. Note, however, that the gsl interpolate.mex∗ file is mandatory to use since it callsthe interpolation routines in the GNU Scientific Library (which is used by SELENA).

4.3.2 The SELENA oct-files

The oct-files currently implemented in SELENA are:

att_sub.oct gmotion.oct imp_dcm.oct tree.oct

csm.oct gsl_interpolate.oct madrs.oct treemdr.oct

curveintersect.oct humanloss.oct spectralshape.oct

damagep.oct humanlosshz.oct squaredam.oct

Note that there are no binary oct-file distributed with SELENA since they often need to be build forthe particular version of Octave that is used. See Appendix B for build instructions.

5 Running SELENA

TO run SELENA, a certain number of files containing the input data have to be prepared. Theseinput files have to be available in the “input” folder. Note that a selection of necessary input files

already exist in the examples folder and preferably can be modified for any new analysis run. The inputfiles needs to be prepared in ASCII-format and provided as plain text files (*.txt). In addition, a numberof input files are required which either contain fixed parameter values (variables) or which include thespectral acceleration and displacement values of the single capacity curves. Input files containing thefixed parameter values (e.g., ec8t1.txt) will normally not be modified by the user and should be keptas they are. In the following, the format of the different input files will be explained separately in moredetail. Their order conforms to SELENA’s sequence of prompting the different inputs.

Note that the results of any new analysis with SELENA will be written into a sub-folder called output.The user has to be careful when running SELENA a second time, since the sub-folder output is auto-matically recreated and so all files of previous runs will be overwritten.

The contents of the input folder can, for example, look like this:

attenuation.txt elosscd1.txt indcasratec.txt injury4.txt population.txt

builtarea.txt elossed1.txt indcasratecc.txt newconstruction.txt soilcenter1.txt

capacity1.txt elossmd1.txt indcasratee.txt ocupmbt_files/ soilcenter2.txt

capcurves/ elosssd1.txt indcasratem.txt ocupmbtp.txt soilcenter3.txt

collapserate.txt fragility1.txt indcasrates.txt outcasratec.txt soilfiles.txt

cpfile.txt header.txt injury1.txt outcasratee.txt ubcampfact.txt

earthquake.txt headermdr.txt injury2.txt outcasratem.txt vulnerfiles.txt

ecfiles.txt headerocc.txt injury3.txt poptime.txt

where the sub-folder capcurves contain the capacity curve files, such as,

capc_C1M-pre.txt capc_C2M-pre.txt capc_C3M-pre.txt

and the sub-folder ocupmbt files contain, for example,

ocupmbt1.txt ocupmbt2.txt ocupmbt3.txt

The input files are described in Section 5.1.

38

DR

AFT


5.1 Preparation of Input Files

5.1.1 Input Files for Deterministic Analysis

For a deterministic analysis 5 different input files are required.

cpfile.txt: see Section 5.1.4: Common input files for all analysis types.

earthquake.txt: Input file containing the information about the earthquake to be used in the seismicrisk study. This file includes different earthquakes with corresponding weights to be run by the logic treemethodology.

Format:

%Earthquake scenarios information

%1st column is the weight for the logic tree scheme:weight

%2nd column is latitude in degrees:lat

%3rd column is longitude in degrees:lon

%4th column is focal depth in km:depth

%5th column is Ms magnitude:Ms

%6th column is Mw magnitude:Mw

%7th column is fault orientation in degrees from North:strike

%8th column is dip angle in degrees:dip

%9th column is fault mechanism:strike-slip/normal(1);reverse(2);all(3)

%10th column is the numerical code for the spectral shape, e.g. 1 for IBC-2006

0.33 59.90 10.90 20.00 5.50 5.50 0.00 90.00 2 1

0.34 59.90 10.90 20.00 6.00 6.00 0.00 90.00 2 1

0.33 59.90 10.90 20.00 6.50 6.50 0.00 90.00 2 1

where weight is the weight for the logic tree scheme, lat is the latitude in degrees, lon is th longitudein degrees, depth is the focal depth in [km], Ms is the surface wave magnitude Ms, Mw is the momentmagnitude Mw, strike is the fault orientation in degrees from North, dip is the dip angle in degrees,fault is the fault mechanism: 1 - strike-slip/normal; 2 - reverse; 3 all, sshape is the numerical code forthe spectral shape as given in spectralshape.m and spectralshape.c (see Table 14)

soilfiles.txt: see Section 5.1.4: Common input files for all analysis types

attenuation.txt: Input file containing the labels of the different empirical ground-motion predictionequations (short: att. rel.) to be used in the study and its corresponding weights for the logic treemethodology.

Format:

%Ground motion information.

%1st column is weight: weight

%2nd column is the label of PGA att.rel.: PGA

%3rd column is the label of Sa at 0.3 s att.rel.: Sa03

%4th column is the label of Sa at 1.0 s att.rel.: Sa10

0.6 22 322 1022

0.2 23 323 1023

0.2 24 324 102

where: weight is the weight for the logic tree scheme, PGA is the label of the applied att.rel. in orderto determine PGA (see Table 19), Sa03 is the label of the applied att.rel. in order to determine Sa at 0.3s (see Table 19), Sa10 is the label of the applied att.rel. in order to determine Sa at 1.0 s (see Table 19),

In the sample file above, the ground-motion prediction equations by Ambraseys et al. [21] are used inorder to derive:

• the mean values of PGA (no=22 in the att sub function), Sa at 0.3 s (no=322), and Sa at 1.0 s(no=1022) weighted 0.6,

• the mean values plus standard deviation σ of PGA (23), Sa at 0.3 s (323), and Sa at 1.0 s (1023)weighted 0.2, and

39

DR

AFT


• the mean values minus standard deviation σ of PGA (24), Sa at 0.3 s (324), and Sa at 1.0 s (1024)weighted 0.2.

Note: For each attenuation relationship periods at 0 s (PGA), 0.3 s and 1.0 s should be given with thesame weights since each computation will need the ground-motion values at these three periods.

vulnerfiles.txt: see Section 5.1.4: Common input files for all analysis types.

ecfiles.txt: see Section 5.1.4: Common input files for all analysis types.

5.1.2 Input Files for Probabilistic Analysis

For a probabilistic analysis 4 different input files are required.


shakefiles.txt: Input file referring to the sub-files of the probabilistic ground-motion informationshakecenter(i).txt (2nd column), indicating their corresponding weight for the logic tree methodology(1st column) and referring to the numerical code for the spectral shape (3rd column) as given in Table 14and implemented in the files spectralshape.m and spectralshape.c.

Index Site classification scheme Site classes

1 United States: IBC-2006 [8] A-E (Table 15)2 Eurocode 8: Type 1 (CEN, 2002) A–E (Table 16)3 Eurocode 8: Type 2 (CEN, 2002) A–E (Table 16)4 India: IS 1893 (Part 1) : 2002 [41] I–III (Table 17)

Table 14: Indexing of the incorporated site classification schemes.

Format:

0.60 shakecenter1.txt 1



Each shakecenter(i).txt file contains the spectral ground-motion values PGA, spectral accelerationSa at 0.3 seconds, and spectral acceleration Sa at 1.0 seconds for each geographical unit (center coordi-nates) while i indicates the number of the different regarded probabilistic shakecenter. All values are forrock and in % of g. Note that the coordinates of the centroids have to be externally assigned in a GISprogram.

Format of the sub-file shakecenter(i).txt:

%GEOUNIT Lat Lon Soil PGA Sa_03 Sa_10

0100100001 59.914 10.719 2 0.2548 0.3160 0.0866

0100100002 59.916 10.711 2 0.2548 0.3160 0.0866

0100100003 59.919 10.707 2 0.2548 0.3160 0.0866

0100100004 59.916 10.698 2 0.2548 0.3160 0.0866

01001...

in which:

GEOUNIT is the label for the identification of the geographical unit, Lat is the label for the geographicalcoordinates (degree of latitude), Lon is the label for the geographical coordinates (degree of longitude),Soil is the label for the for the soil type in each geographical unit.

soilfiles.txt: see Section 5.1.4: Common input files for all analysis types.

40

DR

AFT




5.1.3 Input Files for Analysis with Real-time Data

For an analysis with real-time data 4 different input files are required.


realtimefile.txt: Input file referring to the names of the sub-files realtimegrid.txt (1st col-umn), the moment magnitude, Mw, of the respective earthquake (2nd column) and referring to thenumerical code for the spectral shape (3rd column) as given in Table 14 and described in the file(s)spectralshape.m/spectralshape.c (here: 1 for IBC-2006).

Format:

realtimegrid.txt 6.0 1

The sub-file realtimegrid.txt contains the information of Peak Ground Acceleration (PGA), spectralacceleration Sa at 0.3 s, and spectral acceleration Sa at 1.0 s at the equally-spaced points of a grid pattern.These have to be provided by the user for example by real-time shake maps.

Format of the sub-file realtimegrid.txt:

%Lat Lon PGA Sa03 Sa10

59.80 10.60 0.0819 0.1617 0.0394

59.80 10.61 0.0832 0.1642 0.0399

59.80 10.62 0.0846 0.1669 0.0405

59.80 10.63 0.0861 0.1697 0.0411

...

Note: Given that soil effects are already included in the available real-time shake maps then the soiltypes in input files soilcenter(i).txt have to be set as rock (2) so that the ground-motion ordinatesare not amplified a second time. In case that the available real-time shake maps are confined to rockthen the soil types in input files soilcenter(i).txt have to be set as normal.

soilfiles.txt: see Section 5.1.4: Common input files for all analysis types.



5.1.4 Common Input Files for all Analysis Types

cpfile.txt: Input file which decides on the type of analysis method to be used in order to calculate theperformance point and whether the damage results are dependent on number of damaged buildings ordamaged building area.

Format:

% 1st column is the type of analysis method used (1-CSM; 2-MADRS; 3-I-DCM)

% 2nd column is the type of damage results (1-square meters; 2-nr of buildings)

% 3rd column is the human losses method (1-basic method; 2-HAZUS method)

2 1 2

soilfiles.txt: Input file referring to the names of the sub-files soilcenter(i).txt and indicatingtheir corresponding weight for the logic tree methodology.

Format:

41

DR

AFT


0.33 soilcenter1.txt



with i indicating the number of the different regarded soil columns.

Each soilcenter(i).txt file contains the information about the geographical coordinates of thecenter of each geographical unit in which the studied region had been divided as well as a column withthe soil type associated to that specific geographical unit.

The soil column will be labeled with a code following the soil classification scheme applied by the user.Currently, four different classification schemes are available which have to be appointed in the respectiveinput file earthquake.txt, shakefiles.txt, or realtimefile.txt by an index (Table 14).

Code NEHRP site class Site class description Shear-wave velocity vs,30 [m/s]

1 A hard rock > 15002 B Rock 760-15003 C very dense soil and soft rock 360-7604 D stiff soil 180-3605 E soft soil < 180

Table 15: Site classification according to the NEHRP provisions [43] used by the International BuildingCode IBC-2006 [8].

Code Ground type Site class descriptionShear-wave velocity

vs,30 [m/s]

1 A rock or rock-like formation > 8002 B very dense sands, gravel or very stiff clay 360-8003 C deep deposits of dense or medium-dense sand, 180-360

gravel of stiff clay4 D deposits of loose-to-medium cohesionless soil, < 180

or of soft-to-firm cohesive soil5 E soil profile of a surface alluvium layer of C or D and n.a.

H = 5-20 m underlain by A

Table 16: Site classification according to Eurocode 8 (CEN, 2002).

Code Ground type Site class description Shear-wave velocity vs,30 [m/s]

1 I rock or hard soil > 4002 II medium soils 200-4003 III soft soils < 200

Table 17: Site classification according to the Indian standard IS 1893 (Part 1) : 2002 [41].

The center coordinates must be defined independently of SELENA. [If unknown: when the shapefilesof all the geographical units have been created in ArcView, the center of each geographical unit is easilyobtained using the internal scripts of ArcView.]

Format of the sub-file soilcenter(i).txt:

%GEOUNIT Lat Lon Soil

0100100001 59.91401 10.71870 2

0100100002 59.91562 10.71144 2

01001...

42

DR

AFT


where GEOUNIT is the label for the identification of the geographical unit, Lat is the label for thegeographical coordinates (degree of latitude), Lon is the label for the geographical coordinates (degree oflongitude), Soil is the label for the for the soil type in each geographical unit.

In the sample file the GEOUNIT column has a string format of 10 characters since it was created to fitthe HAZUS input files, however the SELENA code accepts any integer numerical format. Coordinatesmust be provided in geographical coordinates for the representation in ArcView and can be used later toprepare maps in other coordinate systems.

The respective soil amplification factors (and corner control periods) in order to construct the responsespectra for the different soil types are provided by input files ubcampfact.txt (for NEHRP site classesas given in IBC-2006, Table 15) and ec8t1.txt (for site classes of Eurocode 8 Type 1, Table 16),ec8t2.txt (for site classes of Eurocode 8 Type 2, Table 15) and IS1893.txt (for site classes of Indiancode, Table 17).

vulnerfiles.txt: Input file referring to the sub-files capacity(i).txt and fragility(i).txt, aswell as indicating their corresponding weight for the logic tree methodology.

Format:

0.33 capacity1.txt fragility1.txt



with i indicating the number of different sets of capacity curves capacity(i).txt respectively fragilitycurves fragility(i).txt for the logic tree computation scheme.

In turn, each of the capacity(i).txt files refers to one particular set of building capacity curveswhich have to be provided as text (ASCII) files. Very often there may be three or more sets (andthus capacity(i).txt files) representing the variability of the capacity curve (e.g., median, 84%- and16%-fractile).

Format of the sub-files capacity(i).txt:

capc_C1M-pre.txt 7 0.0052 0.40 0.20 0.00 %C1M

capc_C2M-pre.txt 7 0.0046 0.40 0.20 0.00 %C2M

capc_C3M-pre.txt 10 0.0046 0.40 0.20 0.00 %C3M

where

1st column: filename of the respective capacity curve,

2nd column: elastic damping in percentage, for each one of the model building types mbt accordingto the recommendations of Newmark and Hall [49] for materials at or just below their yield point(explained in the technical description of this report),

3rd column: spectral displacement corresponding to the elastic limit (in [m]),

4th column: kappa value for short duration earthquake (Table 5.18 in HAZUS [2]),

5th column: kappa value for moderate duration earthquake (Table 5.18 in HAZUS [2]),

6th column: kappa value for long duration earthquake (Table 5.18 in HAZUS [2]),

7th column: comment on the denomination of the respective (model) building type.

The files containing the spectral displacement and spectral acceleration values of the actual capacitycurve (e.g., capc C1M-pre.txt) in general are provided as plain ASCII text files in the following format:

0 0

0.0005 0.0049

43

DR

AFT


0.001 0.0098

0.0015 0.0147

0.002 0.0196

...

where the first column is the spectral displacement (in [m]) and the secone column is the spectral

acceleration (in [m/s2])

Each fragility(i).txt file contains the information of the fragility curve which has to be used incombination with its corresponding capacity curve.

Format of the sub-file fragility(i).txt:

%mbt smedian sbeta mmedian mbeta emedian ebeta cmedian cbeta Pre-Code

1 0.0305 0.73 0.0488 0.77 0.1219 0.83 0.3048 0.98 %C1M

2 0.0244 0.86 0.0465 0.83 0.1204 0.80 0.3048 0.98 %C2M

3 0.0183 0.90 0.0366 0.86 0.0914 0.90 0.2134 0.96 %C3M

where mbt is the index of model building type, xmedian is the median value of spectral displacementin unites of [m] at which the building reaches the threshold of the damage state, x, which can be one of:s(slight), m (moderate), e (extensive), and c (complete), and where xbeta is the standard deviation ofthe natural logarithm of spectral displacement of damage state, x, which also can be one of: s (slight),m (moderate), e (extensive), and c (complete).

5.1.5 Input Files for the Calculation of Economic Losses

In order to calculate the economic losses the user has to provide the monetary values per [m2] dependenton model building type, occupancy, and structural damage state.

ecfiles.txt: Input file referring to the sub-files elosssd(i).txt (slight damage), elossmd(i).txt(moderate damage), elossed(i).txt (extensive damage), and elosscd(i).txt (complete damage) aswell as indicating their corresponding weight for the logic tree methodology.

Format:

0.50 elosssd1.txt elossmd1.txt elossed1.txt elosscd1.txt

0.50 elosssd2.txt elossmd2.txt elossed2.txt elosscd2.txt

with i indicating the number of different loss models reflecting the monetary loss in a predefined cur-rency per square meters. Consequently, the sub-files elosssd(i).txt, elossmd(i).txt, elossed(i).txtand elosscd(i).txt contain the economic information in order to compute economic (monetary) lossesdue to slight, moderate, extensive and complete structural damage to a specific model building typedependent on occupancy type. The quantities are given in the user-desired predefined currency.

Format of the sub-file elossxd(i).txt:

%oct W1 W2 S1 S2 S3 S4 S5 C1 C2 C3 PC1 PC2 RM1 RM2 URM MH LABEL

1 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 3.0 0.0 %RES1

2 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 2.0 0.0 %RES3

3 ...

where x can be one of: s (slight), m (moderate), e (extensive), and c (complete).

5.1.6 Input Files for the Calculation of Human Losses — Casualties

population.txt: Input file containing the population distribution in the studied area (compare alsowith Table 13 for more detailed information). In case that the ‘Basic methodology’ is used to calculatehuman losses only the numbers of total census tract population (2nd column) are necessary to provide.

Format:

44

DR

AFT


%GEOUNIT POP DRES NRES COMW INDW COMM GRADE COLLEGE HOTEL PRFIL VISIT

0102500001 4125 1701 3929 716 153 278 275 400 50 0.80 0.0

0102500002 7874 3004 7368 694 47 377 350 600 80 0.80 0.0

0102500003 6777 2510 6415 918 57 252 300 500 20 0.80 0.0

0102500004 6534 1879 6333 631 102 337 280 450 20 0.80 0.0

where POP is the total census tract population, DRES is the daytime residential population inferred fromcensus data, NRES is the nighttime residential population inferred from census data, COMW is the numberof people employed in the commercial sector, INDW is the number of people employed in the industrialsector, COMM is the number of people commuting inferred from census data, GRADE is the number ofstudents in grade school (usually under 17 years old), COLLEGE is the number of students on college anduniversity campuses in the census tract (over 17 years old), HOTEL is the number of people staying inhotels in the census tract. Furthermore, PRFIL is the factor representing the proportion of commutersusing automobiles inferred from profile of the community (0.60 for dense urban areas, 0.80 for less denseurban or suburban areas and 0.85 for rural) where the default value is 0.80, and where, VISIT is thenumber of regional residents who do not live in the study area, visiting the census tract for shopping andentertainment (default is set to zero).

poptime.txt: Input file reflecting the population percentages (in decimal numbers) staying indoorsor outdoors dependent on the time of the day. This file is only needed if the human losses are going tobe computed using the ‘Basic methodology’.

Format:

%HOUR INDOOR OUTDOOR Label

1 0.99 0.01 %night 02:00 am

2 0.10 0.90 %day 10:00 am

3 0.15 0.85 %commuting 17:00 pm

ocupmbtp.txt: Input file indicating the share of each model building type (MBT) and its occupancyat the entire building stock. This file is only needed if the human losses are going to be computed usingthe ‘Basic methodology’.

Format:

%mbt RES COM EDU Label

1 0.0042 0.0006 0.0 %C1M

2 0.1342 0.0081 0.0 %C2M

3 0.8042 0.0487 0.0 %C3M

4 0.0 0.0 0.0 %NONE

injury(i).txt: Input files containing the casualty rates of severity i in percentages (i = 1, 2, 3, 4)for the different damage states (i = 1 slight, i = 2 moderate, i = 3 extensive, i = 4 complete or completewith collapse). These files are only needed if the human losses are going to be computed using the ‘Basicmethodology’.

Format:

% Slight Moderate Extensive Complete CompleteCollapse Label

1 0.05 0.20 1.00 10 50 %W1

2 0.05 0.20 1.00 10 50 %W2

3 0.05 0.20 1.00 10 50 %S1L

...

collapserate.txt: Input files containing the percentage of collapsing buildings when they reach thecomplete damage state according to each model building type. This file is only needed is human lossesare going to be computed using the ‘HAZUS methodology’.

Format:

%No COLLAPSE_RATE LABEL

45

DR

AFT


1 10.00 %C1M

2 10.00 %C2M

3 13.00 %C3M

indcasrates.txt: Indoor casualty rate for slight damage. This file is only needed if the human lossesare going to be computed using the ‘HAZUS methodology’.

Format:

%No SEVERITY1 SEVERITY2 SEVERITY3 SEVERITY4 LABEL

1 0.05 0.00 0.00 0.00 %C1M

2 0.05 0.00 0.00 0.00 %C2M

3 0.05 0.00 0.00 0.00 %C3M

indcasratem.txt: Indoor casualty rate for moderate damage. This file is only needed if the humanlosses are going to be computed using the ‘HAZUS methodology’.

Format:


1 0.25 0.030 0.00 0.00 %C1M

2 0.25 0.030 0.00 0.00 %C2M

3 0.20 0.025 0.00 0.00 %C3M

indcasratee.txt: Indoor casualty rate for extensive damage. This file is only needed if the humanlosses are going to be computed using the ‘HAZUS methodology’.

Format:


1 1.00 0.10 0.001 0.001 %C1M

2 1.00 0.10 0.001 0.001 %C2M

3 1.00 0.10 0.001 0.001 %C3M

indcasratec.txt: Indoor casualty rate for complete damage without collapse. This file is only neededif the human losses are going to be computed using the ‘HAZUS methodology’.

Format:


1 5.00 1.00 0.01 0.01 %C1M

2 5.00 1.00 0.01 0.01 %C2M

3 5.00 1.00 0.01 0.01 %C3M

indcasratecc.txt: Indoor casualty rate for complete damage with collapse. This file is only neededif the human losses are going to be computed using the ‘HAZUS methodology’.

Format:


1 40.00 20.00 5.00 10.00 %C1M

2 40.00 20.00 5.00 10.00 %C2M

3 40.00 20.00 5.00 10.00 %C3M

outcasratem.txt: Outdoor casualty rate for moderate damage. This file is only needed if the humanlosses are going to be computed using the ‘HAZUS methodology’.

Format:


1 0.05 0.005 0.00 0.00 %C1M

2 0.05 0.005 0.00 0.00 %C2M

3 0.05 0.005 0.00 0.00 %C3M

46

DR

AFT


outcasratee.txt: Outdoor casualty rate for extensive damage. This file is only needed if the humanlosses are going to be computed using the ‘HAZUS methodology’.

Format:


1 0.20 0.02 0.0002 0.0002 %C1M

2 0.20 0.02 0.0002 0.0002 %C2M

3 0.40 0.04 0.0004 0.0004 %C3M

outcasratec.txt: Outdoor casualty rate for complete damage. This file is only needed if the humanlosses are going to be computed using the ‘HAZUS methodology’.

Format:


1 2.20 0.70 0.20 0.20 %C1M

2 2.20 0.70 0.20 0.20 %C2M

3 3.00 1.20 0.30 0.40 %C3M

occmbtp1.txt: Represents the distribution of population in each census tract and model buildingtype for RESIDENTIAL occupancy (percentage). The percentages of each line summed up have to be1.0 (100%). This file is only needed if the human losses are going to be computed using the ‘HAZUSmethodology’.

Format:

%MBT C1M C2M C3M NONE RESIDENTIAL

0102500001 0.20 0.30 0.50 0

0102500002 0.00 0.00 1.00 0

0102500003 0.00 0.00 1.00 0

0102500004 0.00 0.40 0.60 0

occmbtp2.txt: Represents the distribution of population in each census tract and model building typefor COMMERCIAL occupancy (percentage). The percentages of each line summed up have to be 1.0 (100%).This file is only needed if the human losses are going to be computed using the ‘HAZUS methodology’.

Format:

%MBT C1M C2M C3M NONE COMMERCIAL

0102500001 0.25 0.25 0.50 0

0102500002 0.00 0.00 1.00 0

0102500003 0.00 0.00 1.00 0

0102500004 0.00 0.50 0.50 0

occmbtp3.txt: Represents the distribution of population in each census tract and model buildingtype for EDUCATIONAL occupancy (percentage). The percentages of each line summed up have to be1.0 (100%). This file is only needed if the human losses are going to be computed using the ‘HAZUSmethodology’.

Format:

%MBT C1M C2M C3M NONE EDUCATIONAL

0102500001 0.00 0.00 0.00 0.00

0102500002 0.00 0.00 0.00 0.00

0102500003 0.00 0.00 0.00 0.00

0102500004 0.00 0.00 0.00 0.00

occmbtp4.txt: Represents the distribution of population in each census tract and model buildingtype for INDUSTRIAL occupancy (percentage). The percentages of each line summed up have to be1.0 (100%). This file is only needed if the human losses are going to be computed using the ‘HAZUSmethodology’.

Format:

47

DR

AFT


%MBT C1M C2M C3M NONE INDUSTRIAL

0102500001 0.00 0.00 0.00 0.00

0102500002 0.00 0.00 0.00 0.00

0102500003 0.00 0.00 0.00 0.00

0102500004 0.00 0.00 0.00 0.00

occmbtp5.txt: Represents the distribution of population in each census tract and model building typefor HOTEL occupancy (percentage). The percentages of each line summed up have to be 1.0 (100%).This file is only needed if the human losses are going to be computed using the ‘HAZUS methodology’.

Format:

%MBT C1M C2M C3M NONE HOTEL

0102500001 0.00 0.00 0.00 0.00

0102500002 0.00 0.00 0.00 0.00

0102500003 0.00 0.00 0.00 0.00

0102500004 0.00 0.00 0.00 0.00

5.1.7 Mandatory Input Files

In addition to the input files described above in Section 5.1 there are a number of additional input fileswhich are required in order to run SELENA which need to be located in the same folder as the otherinput files. These files are described in this section.

header.txt: a file providing the necessary header data in order to create the damage output files(which then, for example, can be plotted with ArcView). The first four columns in the header.txt file(GEOUNIT, lon, lat, and soil) remain always the same. All other columns standing for the damageextent of each model building type can be modified or extended subject to the number of consideredmodel building types. Note that the number of additional columns has to be always a multiple of 5 (here:15 model building types × 5 damage states = 75 additional columns in the header). If, e.g., a new modelbuilding type called NB is to be considered, then five columns have to be added in the header: NBN, NBS,NBM, NBE, NBC (for the damage states: Nnone, Sslight, Mmedian, Eextensive and Ccomplete). Finally, thelast label NUMB stands for a column with an order number.Format:

%GEOUNIT Lat Lon Soil W1N W1S W1M W1E W1C ...

where GEOUNIT is the label for the identification of the geographical unit, Lat is the label for the geo-graphical coordinates (degree of latitude), Lon is the label for the geographical coordinates (degree oflongitude), Soil is the label for the for the soil type in each geographical unit, and Numb is the ordernumber.

The herein predefined model building types are those given in HAZUS. More details on these typesincluding ranges of typical building heights and number of stories are given in Appendix A, Table 19.

headerocc.txt: file providing the necessary header for the input files ocupmbt(i).txt allocating thebuilt area (in square meters) according to their occupancy in each geographical unit for the differentmodel building types i.

Format:

%GEOUNIT RES1 RES3 RES4 RES5 COM1 COM2 ...

The herein predefined occupancy classes are those given in HAZUS. A more detailed description ofthese classes is given in Appendix A, Table 20.

headermdr.txt: file providing the necessary header for the output files mdr(i).txt and mdrtot(i).txtfor allocating the mean damage ratio computations i.

Format:

%GEOUNIT MDRT W1 W2 S1L S1M S2L ...

48

DR

AFT


builtarea.txt: input file containing the total built area of each model building type (in squaremeters) for each geographical unit. This type of input file can be easily obtained through the databasesprovided by the local agencies using MS Access, Matlab, or Octave. This file is only needed if riskscenarios are going to be computed in terms of damaged built area with economic losses.

Format:

%GEOUNIT W1 W2 S1L S1M S2L ...

0100100001 8964 0.0 0.0 0.0 0.0 ...

0100100002 5549 0.0 0.0 0.0 0.0 ...

...

where GEOUNIT is the census tract identifier (has to be always in the same order), W1... is the builtarea information for each of the model building types (if more building types are included then morecolumns have to be added) NONE is the built area which can not be assigned to any of the model buildingtypesbecause of the lack of information.

Note: The NONE ‘building type area’ is excluded from the computations (since it covers the area ofall unknown model building types). This means that if a large percentage of the building mass is ofunknown building type, then all cumulative end-results will be wrong due to the fact that a large partof the building mass is excluded from the computations. Anyway, if the input file builtarea.txt ismodified e.g. by adding or removing model building types, the column NONE always has to remain. Ifthere are no ‘unknown’ buildings, then a 0 has to be inserted in column NONE.

numbuild.txt: input file containing the total number of buildings of each model building type foreach geographical unit. This type of input file can be easily obtained through the databases provided bythe local agencies using MS Access, Matlab, or Octave. This file is only needed if risk scenarios are goingto be computed in terms of damaged number of buildings without economic losses.

Format:

%GEOUNIT W1 W2 S1L S1M S2L ...

0100100001 10 0 0 0 0 0 ...

33 0 49

01001...

where GEOUNIT is the census tract identifier (has to be always in the same order), W1... is thenumber of buildings for each one of the model building types (if more building types are included thenmore columns have to be added) NONE is the number of buildings which can not be assigned to any ofthe model building types because of the lack of information.

Note: Input files builtarea.txt and numbuild.txt are not needed at the same time if only one of thedamage results is desired.

ocupmbtj.txt: input files containing the built area (in square meters) according to their occupancyin each geographical unit for the different model building types i. They are needed for the computationof economic losses.

Format:

%GEOUNIT RES1 RES3 RES4 RES5 COM1 ...

0100100001 130.000 690.000 0.000 0.000 0.000 ...

0.000 0.000 0.000 20.000 0.000 0.000 0.000 ...

0.000 0.000 7454.000 0.000 350.000 320.000

...

49

DR

AFT


5.2 Mean Damage Ratio Computation

According to FEMA (2003) [2] a useful parameter in order to be able to compare the risk estimation forthe different geounits within a city or between different cities or countries is named the mean damageratio (MDR); the MDR is defined the cost ratio corresponding to each damage state expressed as a ratioof the cost of new construction.

The MDR computation needs to read a definition of the damage ratio (DR). This information is givenin the input file named newconstruction.txt which has the following format:

%NO C3L C3M C3H RM2L RM2M S1M S5L URML PDC CC LABEL

1 770.00 733.92 693.00 770.00 733.92 694.89 660.00 611.60 990.00 805.20 %RES

2 895.27 853.30 805.75 895.27 853.30 807.95 709.50 657.47 1064.25 865.59 %COM

3 699.59 666.80 629.63 699.59 666.80 631.36 524.95 486.45 787.44 640.44 %IND

4 1375.04 1310.58 1237.53 1375.03 1310.58 1240.92 1031.79 956.12 1547.70 1258.80 %REL

5 987.80 941.50 889.02 987.80 941.50 891.45 741.22 686.85 1111.84 904.29 %GOV

6 987.80 941.50 889.02 987.80 941.50 891.45 741.22 686.85 1111.84 904.29 %EDU

7 1023.00 975.05 920.70 1023.00 975.05 923.22 767.64 711.33 1151.46 936.51 %HOTEL

Several MDRs can be defined, and here, the following definitions are used:

MDR for each model building type and for each geounit : This factor can be computed usingthe following formula

MDRki =

DRkSNk

Si + DRkMNk

Mi + DRkENk

Ei + DRkCNk

Ci

NkTi

(59)

where DRkj is the damage ratio of the model building type, k, corresponding the damage state,

j, where j=S for slight, M for moderate, E for extensive and C for complete. Njik is the damaged

built area corresponding to the damage state j (S,M,E,C) for the model building type, k at theith geounit. NTi

k is the total built area corresponding to the kth model building type at the ithgeounit.

MDR for each geounit and all model building types :

MDRi =

∑mbtk=1(DRk

SNkSi + DRk

MNkMi + DRk

ENkEi + DRk

CNkCi)

NTi(60)

where DRkj : is the damage ratio of the model building type k corresponding the damage state j

where j=S for slight, M for moderate, E for extensive and C for complete. Njik: is the damaged built

area corresponding to the damage state j (S,M,E,C) for the model building type k at the geouniti. NTi:is the total built area at the geounit i. for all the model building types i = 1, . . . ,mbt.

MDR for each model building type and all geounits :

MDRk =

∑geouniti=1 (DRk

SNkSi + DRk

MNkMi + DRk

ENkEi + DRk

CNkCi)

NkT

(61)


where j=S for slight, M for moderate, E for extensive and C for complete. Njik: is the damaged

built area corresponding to the damage state j (S,M,E,C) for the model building type k at thegeounit i. Nk

T:is the total built area for the model building type k and added to all the geounitsi = 1, . . . , geounits.

MDR for all model building type and all geounits :

MDR =

∑geouniti=1

∑mbtk=1(DRk

SNkSi + DRk

MNkMi + DRk

ENkEi + DRk

CNkCi)

NT(62)

50

DR

AFT



where j=S for slight, M for moderate, E for extensive and C for complete. Njik: is the damaged

built area corresponding to the damage state j (S,M,E,C) for the model building type k at thegeounit i. NT:is the total built area for all the model building types and for all the geounits.

5.2.1 Median Values and Confidence Levels

Similarly to the other SELENA results, data from the mdri.txt files will be used to obtain the ex-pected value and confidence values (based on a normal distribution assumption) given in the output filesmdrmedian.txt, mdr16prctile.txt, mdr84prctile.txt, respectively, Also, expected value and confi-dence values, based on data in the mdrtoti.txt files is given in the output files mdrtotmedian.txt,mdrtot16prctile.txt, and mdrtot84prctile.txt, respectively.

5.3 The SELENA Program Sequence

In Figure 16 a flowchart of the SELENA is shown. As noted above, SELENA can be used as a stand-aloneapplication, from Matlab, or from Octave. The stand-alone and Octave versions currently only has acomand line interface, wheras, the Matlab version both have a simple graphical user interface (GUI) anda command line interface; the Matlab SELENA GUI is only used for selecting input files. The differentinterfaces are discribed in Sections 5.3.1, 5.3.2, and 5.3.3, below.

Also, the MDR computation will be performed only if SELENA can find the newcontruction.txt

file in the input file folder.

5.3.1 The Stand-alone SELENA Application

To use the stand-alone version you first need to at it the the path (or use the full path to the binary)then start a start a shell (csh, bash, Windows cmd etc. and type

$ selena -p (or selena --probabilistic)

for probabilistic analysis,

$ selena -d (or selena --deterministic)

for deterministic analysis, or

$ selena -r (or selena --realtime)

for “real-time” (based on aquired real data) analysis. The stand-alone biniary expects that it can findthe input files for respective analysis method as described in Section 5.1.

5.3.2 The Matlab and Octave Command-line Interface

The command-line interface for Matlab/Octave is similar to the stand-alone version. Type

51

DR

AFT


Figure 16: Flowchart of SELENA.

52

DR

AFT


$ selena(’p’); [or selena(’probabilistic)’;]

for probabilistic analysis,

$ selena(’d’); [or selena(’deterministic’);]

for deterministic analysis, or

$ selena(’r’) [or selena(’realtime);]

for real-time analysis. Also here it is expected that the input files, described in Section 5.1, can be found.

5.3.3 The Matlab Graphical User Interface

After having started Matlab, the user has to switch with the Matlab environment to the folder where theinput files are located. After prompting

>> selena_gui

the main menu window, shown in Figure 17, will appear where the user can choose whether to carry outa probabilistic analysis, a deterministic analysis or an analysis based on “real-time” (aquired real) databy clicking the corresponding button.

By choosing one of the three possibilities, either the logic tree scheme window for the probabilisticanalysis (Figure 18), the deterministic analysis (Figure 19) or the analysis with real-time data (Figure 20)is opened. By clicking onto the single buttons, the user will be requested to specify the single input filesas defined in Section 5.1). After defining the five respectively six different input files the user needs toclick onto the Run Analysis button in order to launch the analysis process.

The computation time basically depends on the size of the studied region (number of geographicalunits GEOUNIT), the details of the building information [the number of model building types (MBT) andthe number of building occupancy types (OCT)], and the number of branches used in the logic treemethodology.

5.4 Dealing with Uncertainties

Currently, SELENA computes median values as well as 16% and 84% fractiles of the risk results. This tobe done by means of a logic tree methodology in which the different branches of the tree can be weightedso that at the end of the computation, the risk results are multiplied by their corresponding weights andthen are fitted to a normal distribution in order to get the median values as well as the fractiles. Thesingle branches of the logic tree (Figure 21) currently represent uncertainties in:

• earthquake source, attenuation relationship, soil type, vulnerability curves and economic values ofdamaged built area in case of a deterministic analysis, or

• ground shaking probabilistic maps, soil type, vulnerability curves and economic values of damagedbuilt area in case of a probabilistic analysis.

53

DR

AFT


Figure 17: Main menu of the SELENA Matlab graphical user interface.

Figure 18: Logic tree scheme windows and requested input files for the probabilistic analysis.

54

DR

AFT


Figure 19: Logic tree scheme windows and requested input files for the deterministic analysis.

Figure 20: Logic tree scheme windows and requested input files for the real-time data analysis.

55

DR

AFT


��

��

��

��

��

��

��

��

��

��

��

�� ! "��#�� !

��

�� "�� $%��%&!

� �� !

Figure 21: Logic tree structure. Each branch will be weighted in order to compute the expected valuesand confidence levels.

5.5 Output Files

5.5.1 Overview

All output files being generated during the analysis will be written in the sub-folder output. Table 18lists and describes these output files.

5.5.2 Format of the Output Files

gmotionsceni.txt: Output files for each logic tree branch i which contain the respective ground-motionordinates in the geographical units (GEOUNITS) without (rock) and with (soil) soil amplification as wellas the soil amplification factors (AF) according to the applied earthquake code.

Format:

Column:

1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th 11th 12th 13th

%GEOUNIT Lat Lon Soil PGA

(rock) Sa_0.3

(rock) Sa_1.0

(rock) AF

PGA AF

Sa_0.3 AF

Sa_1.0 PGA

(soil) Sa_0.3

(soil) Sa_1.0

56

DR

AFT


Output file DescriptionUser-requested output type*

damaged area damaged buildings

gmotionsceni.txt files containing ground-motion ordinates Sas and Sal [g],with/without soil amplification, Sasi and SAli [g],and amplification factors FA and FV

douti.txt damage probability for each branch ofprobabilities

the logic tree (number of the file i)

sqmctdouti.txt damage results corresp. to the branch [m2] –nobctdouti.txt of the logic tree (number of the file i) – numbers16prctilect.txt** 16% fractiles of damage

[m2] numbersmedianct.txt** mean value (median) of damage84prctilect.txt** 84% fractiles of damage

eclossesi.txt results corresponding to the branch of economic loss in user-definedthe logic tree (number of the file i) currency (e.g. US-$, e, NOK)containing the economic losses

loss16prctile.txt 16% fractiles of economic loss economic loss in user-definedlossmedian.txt mean value of economic loss currency (e.g. US-$, e, NOK)loss84prctile.txt 84% fractiles of economic loss

totalinjuri.txt results corresponding to the branch of injured persons (cumulative)hlbyinjuri.txt the logic tree (number of the file i) injured persons (disaggregated

containing the human casualties by injury type)hlbyinjur16pr.txt 16% fractiles of injured persons number of injured personshlbyinjurmean.txt mean value of injured persons (disaggregated by injury type)hlbyinjur84pr.txt 84% fractiles of injured personstotalinjur16.txt 16% fractiles of injured persons cumulative number of injuredtotalinjurmean.txt mean value of injured persons persons (from slight to dead)totalinjur84.txt 84% fractiles of injured persons

ltreewgth.txt weight of the damage results (excludingweights

the branches of economic losses)endwgth.txt final weight of the economic results (including

weightsall possible branches)

Table 18: Description of output files and units of results dependent on user-requested output type.*output type is decided in the input file cpfile.txt. **some columns may carry the term ‘-1’ whichindicates that no inventory data for this respective building type was available in order to computedamage; the term ‘0’ would mean that no building or square meters of this building type will undergothis particular damage state.

57

DR

AFT


(soil)

102500003 59.921 10.660 5 0.2050 0.4298 0.1194 2.5 1.7 3.2 0.5125 0.7307 0.3821

102500004 ... ...

Units: All ground-motion ordinates (PGA, spectral accelerations) are provided in [g]-units.

douti.txt: Output files for each logic tree branch i carrying the damage probabilities of each modelbuilding type for the five different damage grades (no, slight, moderate, extensive, complete). The file isstructured according to background file header.txt.

Format:

Column:

1st 2nd 3rd 4th 5th 6th 7th 8th 9th ... last

%GEOUNIT Lat Lon Soil C1MN C1MS C1MM C1ME C1MC ... NUMB

102500001 59.933 10.682 2 0.1446 0.2020 0.4657 0.1282 0.0595 ... 1

102500002 ... ...

where GEOUNIT is the label for the identification of the geographical unit, Lat is the label for the geo-graphical coordinates (degree of latitude), Lon is the label for the geographical coordinates (degree oflongitude), Soil is the soil type according to the chosen soil classification scheme, C1MN is the damageprobability of model building type C1M for state ‘no damage’, C1MS is the damage probability of modelbuilding type C1M for state ‘slight damage’, C1MM is the damage probability of model building type C1Mfor state ‘moderate damage’, C1ME is the damage probability of model building type C1M for state ‘exten-sive damage’, C1MC is the damage probability of model building type C1M for state ‘complete damage’,and NUMB is the order number. Units: Damage probabilities are given in decimal numbers with fourdecimal places (e.g., 0.1446).

Note: the summed up damage probabilities for one model building type must yield to 1.

sqmctdouti.txt respectively nobctdouti.txt: Output files for each logic tree branch i containing thedamage results (either damaged building area or number of damaged buildings) of each model buildingtype for the five different damage grades (no, slight, moderate, extensive, complete). The file is structuredaccording to background file header.txt.

Format (e.g., sqmctdouti.txt):

Column:

1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th last


102500001 59.933 10.682 2 40.63 56.76 130.86 36.02 16.72 ... 1

102500002 ... ...

Units: damage results are either given in square meters (sqmctdouti.txt) or number of buildings(nobctdouti.txt).

medianct.txt , 16prctilect.txt, and 84prctilect.txt: Output files with total damage resultsafter statistical analysis of the logic tree branches (mean value, mean value standard deviation). De-pendent on the type of output results chosen, the output files either contain damaged building area ornumber of damaged buildings of each model building type for the five different damage grades (no, slight,moderate, extensive, complete). The file is structured according to background file header.txt.

Format (here in dependence on square meters):

Column:

1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th last

58

DR

AFT



102500001 59.933 10.682 2 92 63 100 17 9 ... 1

102500002 ... ...

Units: damage results are either given in square meters (m2) or number of buildings.

eclossesi.txt: Output files for each logic tree branch i containing the total economic loss (in auser-defined currency) in each geographical unit.

Format:

Column:

1st 2nd

%GEOUNIT EURO

102500001 179118.4

102500002 469430.2

102500003 ...

Units: economic losses are given in the user-defined currency [here: Euro (e)].

lossmedian.txt, loss16prctile.txt, and loss84prctile.txt: Output files with total economiclosses after statistical analysis of the logic tree branches (mean value, mean value standard deviation) ineach geographical unit.

Format:

Column:

1st 2nd 3rd

%GEOUNIT EURO Order

102500001 517265.729 1

102500002 245537.129 2

102500003 ... ...

Units: economic losses are given in the user-defined currency [here: Euro (e)].

totalinjuri.txt: Output files for each logic tree branch i with cumulative numbers of human losses(from slightly injured to dead) for the three different daytime scenarios in each geographical unit.

Format:

Column:

1st 2nd 3rd 4th

%GEOUNIT INJUR_2:00 INJUR_10:00 INJUR_17:00

102500001 5.0 0.5 0.8

102500002 9.6 1.0 1.5

102500003 8.3 ... ...

Units: human losses are given in numbers of casualties (injured or dead persons).

hlbyinjuri.txt: Output files for each logic tree branch i with distinct numbers of human losses(from slightly injured to dead) for the three different daytime scenarios in each geographical unit.

Format:

Column:


%GEOUNIT INJUR

LOW_

59

DR

AFT


2:00 INJUR

LOW_

10:00 INJURLOW_

17:00 INJUR

MED_

2:00 INJUR

MED_

10:00 INJUR

MED_

17:00 INJUR

HEAV_

2:00 INJUR

HEAV_

10:00 INJUR

HEAV_

7:00 INJURV

DEATH_

2:00 INJURV

DEATH_

10:00 INJURV

DEATH_

17:00

102500001 3.0 0.3 0.5 1.2 0.1 0.2 0.4 0.0 0.1 0.4 0.0 0.1

102500002 ... ...


hlbyinjurmean.txt, hlbyinjur16pr.txt, and hlbyinjur84pr.txt: Output files with numbers ofhuman losses (from slightly injured to dead) for the three different daytime scenarios after statisticalanalysis of the logic tree branches (mean value, mean value standard deviation) in each geographicalunit.

Format:

Column:


%GEOUNIT INJUR

LOW_

2:00 INJUR

LOW_

10:00 INJURLOW_

17:00 INJUR

MED_

2:00 INJUR

MED_

10:00 INJUR

MED_

17:00 INJUR

HEAV_

2:00 INJUR

HEAV_

10:00 INJUR

HEAV_

7:00 INJURV

60

DR

AFT


DEATH_

2:00 INJURV

DEATH_

1 0:00 INJURV

DEATH_

17:00

102500001 15.6 1.6 2.4 1.6 2.4 8.1 2.4 8.1 0.8 8.1 0.8 1.2

102500002 ... ...


totalinjurmean.txt, totalinjur16.txt, and totalinjur84.txt: Output files with cumulativenumbers of human losses (from slightly injured to dead) for the three different daytime scenarios afterstatistical analysis of the logic tree branches (mean value, mean value standard deviation) in eachgeographical unit.

Format:

Column:

1st 2nd 3rd 4th

%GEOUNIT INJUR_2:00 INJUR_10:00 INJUR_17:00

102500001 30.8 3.1 4.6

102500002 58.7 5.9 8.9

102500003 50.5 5.1 ...


5.5.3 Mean Damage Ratio Output Files

For the damage ratio (MDR) computation there will be a file mdri.txt for each branch, i, of the logictree (see Section 5.2). These files will contain the results of MDR for each model building type and foreach geounit and the second column of the output file will contain the MDR for each geounit and allmodel building types.

Format:

%GEOUNIT MDR C3L C3M C3H RM2L RM2M S1M S5L URML PDC CC

101 0.19218600 0.28462806 0.52991545 -1 0.05215911 0.27170182 -1 -1 -1 -1 0.05066373

102 0.29652423 0.38219628 0.52629182 0.63284273 0.05155185 0.25478273 -1 -1 0.13397214 ...

104 0.39414670 0.39333182 0.53828636 -1 0.05332634 0.30766455 -1 -1 -1 0.04183736 ...

105 0.10640911 0.37688768 0.52042182 -1 0.05079273 0.22691088 -1 -1 0.13015636 -1 0.04914456

106 0.33274322 0.38893441 0.53353182 -1 0.05260550 0.28760177 -1 0.35689455 -1 0.04127933 ...

107 0.35963197 0.39865541 0.54384000 -1 0.05418364 0.33119364 0.45359613 -1 0.14599178 ...

108 0.37836659 0.39362086 0.53862544 0.64316636 0.05343548 0.30881818 -1 -1 0.14218388 -1 ...

When -1s are found in these output files it means that no built area of the corresponding modelbuilding type exists so the MDR can not be computed.

There will be a mdrtoti.txt file for each branch, i, of the logic tree. This file will contain the MDRresults for all model building types and all geounits and for each geounit; the second column of the outputfile will contain the MDR for all model building type and all geounits.

Format:

%GEOUNIT MDR C3L C3M C3H RM2L RM2M S1M S5L URML PDC CC

1 0.32518213 0.38878658 0.53521644 0.63821245 0.05231050 0.29422597 ...

61

DR

AFT

Selena 6 EXAMPLES

6 Examples

SELENA comes with one set of example input files (for Bucharest) which are located the examples

folder.

6.1 The Bucharest Example

The Bucharest example is for an earthquake at long 26.76 lat 45.77 (see Figure 22). There are 6 geo-graphical units and ?? building types.

Figure 22: Map (from http://www.openstreetmap.org) of location of the geounits (red markers) andthe earthquake (red asterisk) for the Bucharest example.

The earthquake.txt file specifies the 10 parameters for each earthquake scenarios (magnitude, loca-tion, etc.):

0.12 45.77 26.76 60.00 7.40 7.40 45.00 90.00 2 2

0.16 45.77 26.76 90.00 7.40 7.40 45.00 90.00 2 2

0.12 45.77 26.76 180.00 7.40 7.40 45.00 90.00 2 2

0.09 45.77 26.76 60.00 7.30 7.30 45.00 90.00 2 2

0.12 45.77 26.76 90.00 7.30 7.30 45.00 90.00 2 2

0.09 45.77 26.76 180.00 7.30 7.30 45.00 90.00 2 2

0.09 45.77 26.76 60.00 7.20 7.20 45.00 90.00 2 2

0.12 45.77 26.76 90.00 7.20 7.20 45.00 90.00 2 2

0.09 45.77 26.76 180.00 7.20 7.20 45.00 90.00 2 2

For this example there are 9 earthquake scenarios (for various depths) and the probabilies (“weights”)for each scenario is given in the first column. The depths . . .

6.2 Determistic Data

Nothing here yet. . .

62

http://www.openstreetmap.org

DR

AFT

Selena 8 KNOWN ISSUES

6.3 Probabilistic Data


6.4 Realtime Data


7 Plotting results in Geographic Information Systems (GIS)

SINCE most of the SELENA results are provided in ‘geo-referenced’ output files, the illustration in aGeographic Information System is recommended. This of course can be also done with those input or

inventory files whose data is connected to the geographical units (geounit) or the geographical coordinatesof its respective center (centroid).

In the following it is briefly described how the GIS package ArcView [1] can be applied to plot thedamage and loss results derived by SELENA. SELENA’s output files have been prepared such that theycan be easily imported into a spreadsheet (MS Excel, OpenOffice, etc.) in terms of a delimited formatwith tab and comma or space as delimiters. The user can then use the spreadsheet program to sumall columns with moderate damage (for example) and get a column with moderate damage for all thebuilding types or sum all columns with moderate, extensive and complete damage obtaining a columnwith at least moderate damage for all the building types. In the sum the user has to be carefully withthe dummy value of ‘1’ in some cells, so we suggest to change this dummy value to 0 before summingup. Finally, the files to be plotted in ArcView must be exported to *.dbf (Dbase 4) formatted files andthey must contain at least the following columns:

% GEOUNIT LONGI LATI other-columns-to-be-plotted NUMB

The user can run ArcView, create a new project, add a new view, include a theme in the view (e.g.,with the Oslo census tracts) and add a table (the *.dbf file which is going to be plotted). The user mustthen click into the view window, in THEME+TABLE in order to open the main table e.g., of the Oslo censustracts. Now, it is possible in the table window to see two different tables (Attributes of. . . , which isthe main table, and output.dbf which is the output file going to be plotted). Both tables should have acommon column (maybe with different headers but with the same data structure, e.g., GEOUNIT or NUMB.Then, the user has to click first in the common column of output.dbf (make a click in the header ofthe column) and then in the common column of Attributes of . . . (also in the header of the column). Inthat way it is possible to go to ‘Join both tables’ using (CTRL+J), and the theme which is currently inthe view window will contain all the information of the output results. The user can then make a click inTHEME+EDIT LEGEND and choose a Legend Type: Graduated Color and as Classification Field the columnwhich is going to be plotted. Then, the user can make a click in START EDITING and SAVE the EDITS AS

a new theme file which will be added to the view. Then the process can be repeated for other columnskeeping always the original theme without changes. The same methodology of plotting can be used toplot results from ground motion.

8 Known Issues

• The MADRS C-function produces a slightly different result compared to the corresponding m-codefor very large amplitudes. This is due to numerical issues when there is an infinite gradient in the

63

DR

AFT

Selena REFERENCES

Sa-Sd curve for such input files. Note that the m-file version is not necessarily more accurate thanthe C-version.

• On Windows platforms the MEX-files can not be aborted by pressing CTRL-C since Windows lacksreal asynchronous signals, see: http://www.mathworks.com/support/solutions/data/1-188VX.htmlTherefore, when CTRL-C is pressed, using Windows, the operation is interrupted after the MEX-filehas finished.

• All input (*.txt) files must end with a carrage return (CR), otherwise the stand-alone (commandline and GUI) versions of SELENA will crash.

9 Summary

THE herein described software tool SELENA can be used to provide damage results, economic andhuman losses for the general building stock and population of a city or country on the level of

minimum geographical units (geounit or census tracts). The level of resolution of the damage and losspredictions basically depends on the size of the geographical units which can be defined by the user. Thecode has been developed such that the user can introduce most of the needed inputs using a simple texteditor independ of which computer platform that is used. The Matlab/Octave-code, the C-code, and theASCII input files are fully transparent allowing the user to apply own modifications and adjustments.Furthermore, it was an aim to include as many comments as possible into the code such that the usercan go through the lines and easily change them when necessary. It should be noticed that the presentedtool for seismic risk and loss assessment, SELENA, is an ongoing development and thus will undergoa number of changes and extensions. Consequently, the authors depend on the users’ feedback andsuggestions which are very much appreciated.

Acknowledgments

THIS work has been developed thanks to the agreement between NORSAR and the University ofAlicante under the umbrella of the International Centre for Geohazards (ICG) [54]. The funding

through the SAFER [55] project and through ICG has facilitated major developments of SELENA fromits first version in 2004.

References

[1] http://www.esri.com/software/arcview/, .

[2] Multi-hazard Loss Estimation Methodology, Technical manual. Federal Emergency ManagementAgency, Washington DC, USA, 2003.

[3] R.V. Whitman, T. Anagnos, C.A. Kircher, H.J. Lagorio, R.S. Lawson, and P. Schneider. Devel-opment of a national earthquake loss estimation methodology. Earthquake Spectra, 13(4):643–661,1997.

[4] C.A. Kircher, A.A. Nassar, O. Kustu, and W.T. Holmes. Development of building damage functionsfor earthquake loss estimation. Earthquake Spectra, 13(4):663–682, 1997.

[5] http://www.esri.com/software/arcgis/, .

[6] S. Molina and C.D. Lindholm. A logic tree extension of the capacity spectrum method developed toestimate seismic risk in oslo, norway. Journal of Earthquake Engineering, 9(6):877–897, 2005.

64

http://www.mathworks.com/support/solutions/data/1-188VX.html

http://www.esri.com/software/arcview/

http://www.esri.com/software/arcgis/

DR

AFT

Selena REFERENCES

[7] D.H. Lang, S. Molina, and C.D. Lindholm. Towards near-real-time damage estimation using a csm-based tool for seismic risk assessment. Journal of Earthquake Engineering, 12, 2008. Special Issue2.

[8] 2006 international building code (ibc-2006). Technical report, International Code Council, UnitedStates, January .

[9] Design for structures for earthquake resistance, Part 1: General rules, seismic actions and rulesfor buildings. Technical report, European Committee for Standardization CEN, May 2002. prEN1998-1:200X, Eurocode 8.

[10] D.L. Wells and K.J. Coppersmith. New empirical relationships among magnitude, rupture length,rupture width and surface displacement. Bulletin of the Seismological Society of America, 84:974–1002, 1994.

[11] D.M. Boore, W.B. Joyner, and T.E. Fumal. Estimation of response spectra and peak accelerationsfrom western north american earthquakes: An interim report. Technical report, U.S. GeologicalSurvey, 1993. Open-File Report 93-509.

[12] D.M. Boore, W.B. Joyner, Fumal, and T.E. Estimation of response spectra and peak accelera-tions from western north american earthquakes: An interim report. Part 2. Technical report, U.S.Geological Survey, 1994. Open-File Report 94-127.

[13] D.M. Boore, W.B. Joyner, and T.E. Fumal. Equations for estimating horizontal response spec-tra and peak acceleration from western north american earthquakes: A summary of recent work.Seismological Research Letters, 68(1):128–153, 1997.

[14] N.N. Ambraseys, K.A. Simpson, and J.J. Bommer. Prediction of horizontal response spectra ineurope. Earthquake Engineering and Structural Dynamics, pages 371–400, 1996.

[15] G.R. Toro, N.A. Abrahamson, and J.F. Schneider. Model of strong ground motions from earthquakesin central and eastern north america: Best estimates and uncertainties. Seismological ResearchLetters, 68:41–57, 1997.

[16] K.W. Campbell and Y. Bozorgnia. Near-source attenuation of peak horizontal acceleration fromworldwide accelerograms recorded from 1957 to 1993. In Proceedings of the Fifth U.S. NationalConference on Earthquake Engineering, volume III, pages 283–292.

[17] K.W. Campbell. Empirical near-source attenuation relationships for horizontal and vertical compo-nents of peak ground acceleration, peak ground velocity, and pseudo-absolute acceleration responsespectra. Seismological Research Letters, 68(1):154–179, 1997.

[18] K.W. Campbell and Y. Bozorgnia. Updated near-source ground-motion (attenuation) relations forthe horizontal and vertical components of peak ground acceleration and acceleration response spectra.Bulletin of the Seismological Society of America, pages 314–331, 2003.

[19] N.A. Abrahamson and W.J. Silva. Empirical response spectral attenuation relations for shallowcrustal earthquakes. Seismological Research Letters, 68(1):94–127, 1997.

[20] F. Sabetta and A. Pugliese. Estimation of response spectra and simulation of nonstationary earth-quake ground motions. Bulletin of the Seismological Society of America, 86(2):337–352, 1996.

[21] N.N. Ambraseys, J. Douglas, S.K. Sarma, and P.M. Smit. Equations for the estimation of strongground motions from shallow crustal earthquakes using data from europe and the middle east:Horizontal peak ground acceleration and spectral acceleration. Bulletin of Earthquake Engineering,3:1–53, 2005.

65

DR

AFT

Selena REFERENCES

[22] S. Akkar and J.J. Bommer. Prediction of elastic displacement response spectra in europe and themiddle east. Earthquake Engineering and Structural Dynamics, 36(10), 2007.

[23] K. Sadigh, C.-Y. Chang, J.A. Egan, F. Makdisi, , and R.R. Youngs. Attenuation relationships forshallow crustal earthquakes based on california strong motion data. Seismological Research Letters,68(1):180–189, 1997.

[24] P. Spudich, W.B. Joyner, A.G. Lindh, D.M. Boore, B.M. Margaris, and J.B. Fletcher. A revisedground motion prediction relation for use in extensional tectonic regimes. Seismological Society ofAmerica, 89(5), 1999.

[25] Style-of-faulting in ground-motion prediction equations. Bulletin of Earthquake Engineering, 1.

[26] G.M. Atkinson and D.M. Boore. Empirical ground-motion relations for subduction zone earthquakesand their application to cascadia and other regions. Bulletin of the Seismological Society of America,93(4):1703–1729, 2003.

[27] Analysis of strong ground motions to evaluate regional attenuation relationships. Annals of geo-physics, (3–4):439–454, 2002.

[28] J. Schwarz, C. Ende, J. Habenberger, D.H. Lang, M. Baumbach, H. Grosser, C. Milkereit, S. Karak-isa, and S. Zunbul. Horizontal and vertical response spectra on the basis of strong-motion recordingsfrom the 1999 turkey earthquakes. In Proceedings of the XXVIII General Assembly of the EuropeanSeismological Commission (ESC), 2002.

[29] C. Ende and J. Schwarz. Einfluss von analysemethoden auf spektrale abnahmebeziehungen derbodenbewegung. Technical report, Bauhaus-Universitat Weimar, 2004. Schriften der Bauhaus-Universitat Weimar 116: 105–115.

[30] N.N. Ambraseys and J. Douglas. Reappraisal of the effect of vertical ground motions on response.Technical report, Department of Civil and Environmental Engineering, Imperial College, London,2000. ESEE Report 00-4.

[31] J. Douglas. A critical reappraisal of some problems in engineering seismology. PhD thesis, Universityof London, 2001.

[32] N.N. Ambraseys and J. Douglas. Near-field horizontal and vertical earthquake ground motions. SoilDynamics and Earthquake Engineering, 23(1):1–18, 2003.

[33] M.C. Chapman. On the use of elastic input energy for seismic hazard analysis. Earthquake Spectra,15(4):607–635, 1999.

[34] C.B. Crouse and J.W. McGuire. Site response studies for purpose of revising nehrp seismic provisions.Earthquake Spectra, pages 407–439.

[35] P. Gulkan and E. Kalkan. Attenuation modeling of recent earthquakes in turkey. Journal of Seis-mology, 6(3):397–409.

[36] P. Lussou, P.Y. Bard, F. Cotton, and Y. Fukushima. Seismic design regulation codes: Contributionof k-net data to site effect evaluation. Journal of Earthquake Engineering, 5(1):13–33, 2001.

[37] A. Dahle, A. Climent, W. Taylor, H. Bungum, P. Santos, C. Ciudad Real, M.and Lindholm,W. Strauch, and F. Segura. New spectral strong motion attenuation models for central america.In Proceedings of the Fifth International Conference on Seismic Zonation, volume II, pages 1005–1012, 1995.

[38] J.J. Bommer, D.A. Herna ndez, J.A. Navarette, and W.M. Salazar. Seismic hazard assessments forel salvador. Geofiısica Internacional, 35(3):227–244.

66

DR

AFT

Selena REFERENCES

[39] Marmureanu G., M. Androne, M. Radulian, E. Popescu, C.O. Cioflan, A.O. Placinta, I.A. Moldovan,and V. Serban. Attenuation of the peak ground motion for the special case of vrancea intermediate-depth earthquakes and seismic hazard assessment at npp cernavoda. Acta Geod. Geoph. Hung.,(3–4):433–440.

[40] M. L. Sharma, J. Douglas, H. Bungum, and J. Kotadia. Ground-motion prediction equations basedon data from the himalayan and zagros regions. Journal of Earthquake Engineering, 13(8):1191–1210,November 2009.

[41] Indian standard—criteria for earthquake resistant design of structures, part 1—general provisionsand buildings. Technical report, Bureau of Indian Standards (BIS), 2002. ICS 91.120.25.

[42] W.B. Joyner and D.M. Boore. Measurement characterization and prediction of strong ground motion.In Proc. of Earthquake Engineering and Soil Dynamics II, pages 43–102, Park City, Utah, June 1988.New York: Geotechnical Division of the American Society of Civil Engineers.

[43] Nehrp recommended provisions for seismic regulations for new buildings. Technical report, FederalEmergency Management Agency (FEMA), Washington DC, 1997. FEMA 222A.

[44] Specification for structures to be built in disaster areas. Part III — Earthquake disaster prevention(chapter 5–13). Technical report, Turkish Ministry of Public Works and Settlement (TMPS), 1998.

[45] Seismic evaluation and retrofit of concrete buildings. Technical report, Applied Technology Council(ATC), Redwood City, California, 1996. Report ATC-40.

[46] Nehrp guidelines for the seismic rehabilitation of buildings. Technical report, Federal EmergencyManagement Agency (FEMA), Washington DC, October 1997. FEMA 273.

[47] Prestandard and commentary for the seismic rehabilitation of buildings. Technical report, FederalEmergency Management Agency (FEMA), Washington DC, 2000. FEMA 356.

[48] Improvement of nonlinear static seismic analysis procedures. Technical report, Applied TechnologyCouncil (ATC, California, USA, 2005. FEMA-440.

[49] N.M. Newmark and W.J. Hall. Earthquake spectra and design. Technical report, EarthquakeEngineering Research Institute (EERI), Oakland, CA: EERI, 1982.

[50] Earthquake Loss Estimation Methodology: User’s Manual. Federal Emergency Management Agency,Washington DC, USA.

[51] Earthquake damage evaluation data for california. Technical report, Applied Technology Council(ATC), Redwood City, California, 1985. Report ATC-13.

[52] A. Coburn and R. Spence. Earthquake Protection. J. Wiley and Sons Ltd, 2002.

[53] P. Stojanovski and W. Dong. Simulation model for earthquake casualty estimation. In Proc. FifthU.S. National Conference on Earthquake Engineering, Paper No. 00592, Chicago, Illinois, July 10–14, 1994.

[54] http://www.geohazards.no/.

[55] http://www.saferproject.net/.

67

http://www.geohazards.no/

http://www.saferproject.net/

DR

AFT

Selena A TABLES

A Tables

Author(s) (year)Index

mean value (mv) mv+σ mv−σ

Boore et al. [11], Boore et al. [12],Boore et al. [13] 01 02 03Ambraseys et al. [14] 04 05 06Toro et al. [15] 07 08 09Campbell and Bozorgnia [16], Campbell [17] 10 11 12Campbell and Bozorgnia [18] 13 14 15Abrahamson and Silva [19] 16 17 18Sabetta and Pugliese [20] 19 20 21Ambraseys et al. [21] 22 23 24Akkar and Bommer [22] 25 26 27Sadigh et al. [23]* 28 29 30zbey et al. (2003) 31 32 33Spudich et al. [24] 34 35 36Bommer et al. [25] 37 38 39Atkinson and Boore [26] 40 41 42Zonno and Montaldo [27] 43 44 45Schwarz et al. [28], Ende and Schwarz [29] 46 47 48Ambraseys and Douglas [30], Douglas [31],Ambraseys and Douglas [32] 49 50 51Chapman [33] 52 53 54Crouse and McGuireciteCrouse1996 55 56 57Gulkan and Kalkan [35] 58 59 60Lussou et al. [36] 61 62 63Dahle et al. [37] 64 65 66Bommer et al. [38] 67 68 69Marmureanu et al.; *for hypocentral distance, Eq. (17) in [39] 77 78 79Marmureanu et al.; *for epicentral distance, Eq. (3) in [39] 80 81 82

Table 19: Empirical ground-motion prediction equations which are implemented in the current SELENA-version. *Note: prediction equations for spectral accelerations, Sa, are not provided.

68

DR

AFT

Selena A TABLES

No. Label Description Heightrange typical

name stories stories ft.

1 W1 Wood, Light Frame (≤ 5,000 sq.ft. / 465 m2) – all 1 142 W2 Wood, Commercial and Industrial (> 5,000 sq.ft. / 465 m2) – all 2 243 S1L Steel Moment Frame Low-Rise 1–3 2 244 S1M Mid-Rise 4-7 5 605 S1H High-Rise 8+ 13 1566 S2L Steel Braced Frame Low-Rise 1-3 2 247 S2M Mid-Rise 4–7 5 608 S2H High-Rise 8+ 13 1569 S3 Steel Light Frame – all 1 1510 S4L Steel Frame with Cast-in-Place Concrete Shear Walls Low-Rise 1-3 2 2411 S4M Mid-Rise 4-7 5 6012 S4H High-Rise 8+ 13 15613 S5L Steel Frame with Unreinforced Masonry Infill Walls Low-Rise 1-3 2 2414 S5M Mid-Rise 4-7 5 6015 S5H High-Rise 8+ 13 15616 C1L Concrete Moment Frame Low-Rise 1-3 2 2017 C1M Mid-Rise 4–7 5 5018 C1H High-Rise 8+ 12 12019 C2L Concrete Shear Walls Low-Rise 1-3 2 2020 C2M Mid-Rise 4–7 5 6021 C2H High-Rise 8+ 12 12022 C3L Concrete Frame with Unreinforced

Masonry Infill Walls Low-Rise 1-3 2 2023 C3M Mid-Rise 4–7 5 5024 C3H High-Rise 8+ 12 12025 PC1 Pre-cast Concrete Tilt-Up Walls – all 1 1526 PC2L Pre-cast Concrete Frames

with Concrete Shear Walls Low-Rise 1-3 2 2027 PC2M Mid-Rise 4–7 5 5028 PC2H High-Rise 8+ 12 12029 RM1L Reinforced Masonry Bearing Walls

with Wood or Metal Deck Diaphragms Low-Rise 1-3 2 2030 RM1M Mid-Rise 4+ 5 5031 RM2L Reinforced Masonry Bearing Walls

with Pre-cast Concrete Diaphragms Low-Rise 1-3 2 2032 RM2M Mid-Rise 4–7 5 5033 RM2H High-Rise 8+ 12 12034 URML Unreinforced Masonry Bearing Walls Low-Rise 1–2 1 1535 URMM Mid-Rise 3+ 3 3936 MH Mobile Homes – all 1 12

Table 20: Model building types as defined by HAZUS.

69

DR

AFT

Selena A TABLES

No. Label Occupancy class Description

Residential:1 RES1 single family dwelling detached house2 RES2 mobile home mobile home3 RES3 multi family dwelling apartment/condominium4 RES4 temporary lodging hotel/motel5 RES5 institutional dormitory group housing (military, college), jails6 RES6 nursing homeCommercial:7 COM1 retail trade store8 COM2 wholesale trade warehouse9 COM3 personal and repair service service station/shop10 COM4 professional/technical services offices11 COM5 banks/financial institutions12 COM6 hospital13 COM7 medical office/clinics office14 COM8 entertainment and recreation restaurants/bars15 COM9 theatres theatres16 COM10 parking garageIndustrial:17 IND1 heavy factory18 IND2 light factory19 IND4 food/drug/chemicals factory20 IND3 metals/mineral processing factory21 IND5 high technology factory22 IND6 construction officeAgriculture:23 AGR agricultureReligion/Non-Profit:24 REL churchGovernment:25 GOV1 general services office26 GOV2 emergency response police/fire stationEducation:27 EDU1 schools/libraries28 EDU2 universities/colleges does not include group housing

Table 21: Occupancy types as defined in HAZUS.

70

DR

AFT

Selena A TABLES

mbtdy (ord.) [m] ay (ord.) du (ord.) au (ord.) de (ord.) k k k

be fraction pre-code[m] [m/s2] [m] [m/s2] [m] (ord/sh) (ord/md) ord(lg)

1 0.0061 1.9620 0.1097 5.8860 0.0043 0.5 0.3 0.1 15 0 W12 0.0041 0.9810 0.0597 2.4525 0.0028 0.4 0.2 0 10 0 W23 0.0038 0.6082 0.0699 1.8345 0.0027 0.4 0.2 0 5 0 S1L4 0.0112 0.3826 0.1354 1.1478 0.0078 0.4 0.2 0 5 0 S1M5 0.0295 0.2354 0.2662 0.7161 0.0206 0.4 0.2 0 5 0 S1H6 0.0041 0.9810 0.0478 1.9620 0.0028 0.4 0.2 0 5 0 S2L7 0.0155 0.8142 0.1232 1.6383 0.0108 0.4 0.2 0 5 0 S2M8 0.0493 0.6180 0.2951 1.2459 0.0345 0.4 0.2 0 5 0 S2H9 0.0041 0.9810 0.0478 1.9620 0.0028 0.4 0.2 0 7 0 S310 0.0025 0.7848 0.0330 1.7658 0.0018 0.4 0.2 0 7 0 S4L11 0.0069 0.6573 0.0625 1.4715 0.0048 0.4 0.2 0 7 0 S4M12 0.0221 0.5003 0.1494 1.1183 0.0155 0.4 0.2 0 7 0 S4H13 0.0030 0.9810 0.0305 1.9620 0.0021 0.4 0.2 0 10 0 S5L14 0.0086 0.8142 0.0577 1.6383 0.0060 0.4 0.2 0 10 0 S5M15 0.0277 0.6180 0.1384 1.2459 0.0194 0.4 0.2 0 10 0 S5H16 0.0025 0.6082 0.0447 1.8345 0.0018 0.4 0.2 0 7 0 C1L17 0.0074 0.5101 0.0879 1.5304 0.0052 0.4 0.2 0 7 0 C1M18 0.0127 0.2354 0.1148 0.7161 0.0089 0.4 0.2 0 7 0 C1H19 0.0030 0.9810 0.0457 2.4525 0.0021 0.4 0.2 0 7 0 C2L20 0.0066 0.8142 0.0660 2.0405 0.0046 0.4 0.2 0 7 0 C2M21 0.0188 0.6180 0.1400 1.5598 0.0132 0.4 0.2 0 7 0 C2H22 0.0030 0.9810 0.0343 2.2073 0.0021 0.4 0.2 0 10 0 C3L23 0.0066 0.8142 0.0495 1.8443 0.0046 0.4 0.2 0 10 0 C3M24 0.0188 0.6180 0.1049 1.4028 0.0132 0.4 0.2 0 10 0 C3H25 0.0046 1.4715 0.0549 2.9430 0.0032 0.4 0.2 0 7 0 PC126 0.0030 0.9810 0.0366 1.9620 0.0021 0.4 0.2 0 7 0 PC2L27 0.0066 0.8142 0.0528 1.6383 0.0046 0.4 0.2 0 7 0 PC2M28 0.0188 0.6180 0.1120 1.2459 0.0132 0.4 0.2 0 7 0 PC2H29 0.0041 1.3047 0.0488 2.6193 0.0028 0.4 0.2 0 10 0 RM1L30 0.0089 1.0889 0.0704 2.1778 0.0062 0.4 0.2 0 10 0 RM1M31 0.0041 1.3047 0.0488 2.6193 0.0028 0.4 0.2 0 7 0 RM2L32 0.0089 1.0889 0.0704 2.1778 0.0062 0.4 0.2 0 7 0 RM2M33 0.0249 0.8339 0.1494 1.6579 0.0174 0.4 0.2 0 7 0 RM2H34 0.0061 1.9620 0.0610 3.9240 0.0043 0.4 0.2 0 10 0 URML35 0.0069 1.0889 0.0460 2.1778 0.0048 0.4 0.2 0 10 0 URMM36 0.0046 1.4715 0.0549 2.9430 0.0032 0.6 0.3 0.1 5 0 MH

Table 22: Parameters of capacity curves as provided by HAZUS for Pre-code seismic design.

71

DR

AFT

Selena A TABLES


be fraction low-code[m] [m/s2] [m] [m/s2] [m] (ord/sh) (ord/md) ord(lg)

1 0.0061 1.9620 0.1097 5.8860 0.0043 0.7 0.4 0.2 15 0 W12 0.0041 0.9810 0.0597 2.4525 0.0028 0.6 0.3 0.1 10 0 W23 0.0038 0.5886 0.0582 1.8639 0.0027 0.6 0.3 0.1 5 0 S1L4 0.0112 0.3924 0.1128 1.1772 0.0078 0.6 0.3 0.1 5 0 S1M5 0.0295 0.1962 0.2217 0.6867 0.0206 0.6 0.3 0.1 5 0 S1H6 0.0041 0.9810 0.0399 1.9620 0.0028 0.5 0.3 0.1 5 0 S2L7 0.0155 0.7848 0.1026 1.6677 0.0108 0.5 0.3 0.1 5 0 S2M8 0.0493 0.5886 0.2459 1.2753 0.0345 0.5 0.3 0.1 5 0 S2H9 0.0041 0.9810 0.0399 1.9620 0.0028 0.5 0.3 0.1 7 0 S310 0.0025 0.7848 0.0274 1.7658 0.0018 0.5 0.3 0.1 7 0 S4L11 0.0069 0.6867 0.0521 1.4715 0.0048 0.5 0.3 0.1 7 0 S4M12 0.0221 0.4905 0.1245 1.0791 0.0155 0.5 0.3 0.1 7 0 S4H13 0.0030 0.9810 0.0305 1.9620 0.0021 0.5 0.3 0.1 10 0 S5L14 0.0086 0.7848 0.0577 1.6677 0.0060 0.5 0.3 0.1 10 0 S5M15 0.0277 0.5886 0.1384 1.2753 0.0194 0.5 0.3 0.1 10 0 S5H16 0.0025 0.5886 0.0373 1.8639 0.0018 0.6 0.3 0.1 7 0 C1L17 0.0074 0.4905 0.0732 1.5696 0.0052 0.6 0.3 0.1 7 0 C1M18 0.0127 0.1962 0.0958 0.6867 0.0089 0.6 0.3 0.1 7 0 C1H19 0.0030 0.9810 0.0381 2.4525 0.0021 0.6 0.3 0.1 7 0 C2L20 0.0066 0.7848 0.0549 2.0601 0.0046 0.6 0.3 0.1 7 0 C2M21 0.0185 0.5886 0.1166 1.5696 0.0130 0.6 0.3 0.1 7 0 C2H22 0.0030 0.9810 0.0343 2.2563 0.0021 0.5 0.3 0.1 10 0 C3L23 0.0066 0.7848 0.0495 1.8639 0.0046 0.5 0.3 0.1 10 0 C3M24 0.0185 0.5886 0.1049 1.3734 0.0130 0.5 0.3 0.1 10 0 C3H25 0.0046 1.4715 0.0457 2.9430 0.0032 0.5 0.3 0.1 7 0 PC126 0.0030 0.9810 0.0305 1.9620 0.0021 0.5 0.3 0.1 7 0 PC2L27 0.0066 0.7848 0.0439 1.6677 0.0046 0.5 0.3 0.1 7 0 PC2M28 0.0185 0.5886 0.0932 1.2753 0.0130 0.5 0.3 0.1 7 0 PC2H29 0.0041 1.2753 0.0406 2.6487 0.0028 0.6 0.3 0.1 10 0 RM1L30 0.0089 1.0791 0.0587 2.1582 0.0062 0.6 0.3 0.1 10 0 RM1M31 0.0041 1.2753 0.0406 2.6487 0.0028 0.6 0.3 0.1 7 0 RM2L32 0.0089 1.0791 0.0587 2.1582 0.0062 0.6 0.3 0.1 7 0 RM2M33 0.0249 0.8829 0.1245 1.6677 0.0174 0.6 0.3 0.1 7 0 RM2H34 0.0061 1.9620 0.0610 3.9240 0.0043 0.5 0.3 0.1 10 0 URML35 0.0069 1.0791 0.0460 2.1582 0.0048 0.5 0.3 0.1 10 0 URMM36 0.0046 1.4715 0.0549 2.9430 0.0032 0.6 0.4 0.2 5 0 MH

Table 23: Parameters of capacity curves as provided by HAZUS for Low-code seismic design.

72

DR

AFT

Selena A TABLES


be fraction mod-code[m] [m/s2] [m] [m/s2] [m] (ord/sh) (ord/md) ord(lg)

1 0.0091 2.9430 0.1643 8.8290 0.0064 0.9 0.6 0.3 15 0 W12 0.0079 1.9620 0.1194 4.9050 0.0055 0.8 0.4 0.2 10 0 W23 0.0079 1.1772 0.1397 3.7278 0.0055 0.8 0.4 0.2 5 0 S1L4 0.0226 0.7848 0.2705 2.2563 0.0158 0.8 0.4 0.2 5 0 S1M5 0.0592 0.4905 0.5324 1.4715 0.0414 0.8 0.4 0.2 5 0 S1H6 0.0079 1.9620 0.0955 3.9240 0.0055 0.6 0.4 0.2 5 0 S2L7 0.0307 1.6677 0.2464 3.2373 0.0215 0.6 0.4 0.2 5 0 S2M8 0.0983 1.2753 0.5903 2.4525 0.0688 0.6 0.4 0.2 5 0 S2H9 0.0079 1.9620 0.0955 3.9240 0.0055 0.6 0.4 0.2 7 0 S310 0.0048 1.5696 0.0658 3.5316 0.0034 0.6 0.4 0.2 7 0 S4L11 0.0140 1.2753 0.1247 2.9430 0.0098 0.6 0.4 0.2 7 0 S4M12 0.0442 0.9810 0.2987 2.2563 0.0309 0.6 0.4 0.2 7 0 S4H13 S5L14 S5M15 S5H16 0.0051 1.1772 0.0894 3.7278 0.0036 0.8 0.4 0.2 7 0 C1L17 0.0147 0.9810 0.1755 3.0411 0.0103 0.8 0.4 0.2 7 0 C1M18 0.0254 0.4905 0.2299 1.4715 0.0178 0.8 0.4 0.2 7 0 C1H19 0.0061 1.9620 0.0914 4.9050 0.0043 0.8 0.4 0.2 7 0 C2L20 0.0132 1.6677 0.1318 4.1202 0.0092 0.8 0.4 0.2 7 0 C2M21 0.0373 1.2753 0.2799 3.1392 0.0261 0.8 0.4 0.2 7 0 C2H22 C3L23 C3M24 C3H25 0.0091 2.9430 0.1097 5.8860 0.0064 0.6 0.4 0.2 7 0 PC126 0.0061 1.9620 0.0732 3.9240 0.0043 0.6 0.4 0.2 7 0 PC2L27 0.0132 1.6677 0.1054 3.2373 0.0092 0.6 0.4 0.2 7 0 PC2M28 0.0373 1.2753 0.2240 2.4525 0.0261 0.6 0.4 0.2 7 0 PC2H29 0.0081 2.6487 0.0975 5.1993 0.0057 0.8 0.4 0.2 10 0 RM1L30 0.0175 2.1582 0.1407 4.3164 0.0123 0.8 0.4 0.2 10 0 RM1M31 0.0081 2.6487 0.0975 5.1993 0.0057 0.8 0.4 0.2 7 0 RM2L32 0.0175 2.1582 0.1407 4.3164 0.0123 0.8 0.4 0.2 7 0 RM2M33 0.0498 1.6677 0.2987 3.3354 0.0348 0.8 0.4 0.2 7 0 RM2H34 URML35 URMM36 0.0046 1.4715 0.0549 2.9430 0.0032 0.6 0.4 0.2 5 0 MH

Table 24: Parameters of capacity curves as provided by HAZUS for Moderate-code seismic design.

73

DR

AFT

Selena B COMPILING THE C-CODE


be fraction high-code[m] [m/s2] [m] [m/s2] [m] (ord/sh) (ord/md) ord(lg)

1 0.0122 3.9240 0.2924 11.7720 0.0085 1 0.8 0.5 15 0 W12 0.0160 3.9240 0.3183 9.8100 0.0112 0.9 0.6 0.4 10 0 W23 0.0155 2.4525 0.3726 7.3575 0.0108 0.9 0.6 0.4 5 0 S1L4 0.0450 1.5696 0.7214 4.6107 0.0315 0.9 0.6 0.4 5 0 S1M5 0.1184 0.9810 1.4194 2.8449 0.0829 0.9 0.6 0.4 5 0 S1H6 0.0160 3.9240 0.2545 7.8480 0.0112 0.7 0.5 0.3 5 0 S2L7 0.0617 3.2373 0.6574 6.5727 0.0432 0.7 0.5 0.3 5 0 S2M8 0.1969 2.4525 1.5740 5.0031 0.1378 0.7 0.5 0.3 5 0 S2H9 0.0160 3.9240 0.2545 7.8480 0.0112 0.7 0.5 0.3 7 0 S310 0.0097 3.1392 0.1755 7.0632 0.0068 0.7 0.5 0.3 7 0 S4L11 0.0277 2.6487 0.3327 5.8860 0.0194 0.7 0.5 0.3 7 0 S4M12 0.0886 1.9620 0.7968 4.5126 0.0621 0.7 0.5 0.3 7 0 S4H13 S5L14 S5M15 S5H16 0.0099 2.4525 0.2385 7.3575 0.0069 0.9 0.6 0.4 7 0 C1L17 0.0292 2.0601 0.4684 6.0822 0.0204 0.9 0.6 0.4 7 0 C1M18 0.0511 0.9810 0.6129 2.8449 0.0357 0.9 0.6 0.4 7 0 C1H19 0.0122 3.9240 0.2436 9.8100 0.0085 0.9 0.6 0.4 7 0 C2L20 0.0264 3.2373 0.3515 8.1423 0.0185 0.9 0.6 0.4 7 0 C2M21 0.0747 2.4525 0.7465 6.2784 0.0523 0.9 0.6 0.4 7 0 C2H22 C3L23 C3M24 C3H25 0.0183 5.8860 0.2924 11.7720 0.0128 0.7 0.5 0.3 7 0 PC126 0.0122 3.9240 0.1948 7.8480 0.0085 0.7 0.5 0.3 7 0 PC2L27 0.0264 3.2373 0.2812 6.5727 0.0185 0.7 0.5 0.3 7 0 PC2M28 0.0747 2.4525 0.5974 5.0031 0.0523 0.7 0.5 0.3 7 0 PC2H29 0.0163 5.1993 0.2598 10.4967 0.0114 0.9 0.6 0.4 10 0 RM1L30 0.0351 4.3164 0.3749 8.7309 0.0245 0.9 0.6 0.4 10 0 RM1M31 0.0163 5.1993 0.2598 10.4967 0.0114 0.9 0.6 0.4 7 0 RM2L32 0.0351 4.3164 0.3749 8.7309 0.0245 0.9 0.6 0.4 7 0 RM2M33 0.0996 3.3354 0.7963 6.6708 0.0697 0.9 0.6 0.4 7 0 RM2H34 URML35 URMM36 0.0046 1.4715 0.0549 2.9430 0.0032 0.6 0.4 0.2 5 0 MH

Table 25: Parameters of capacity curves as provided by HAZUS for Moderate-code seismic design.

B Compiling the C-code

THIS section describes how to build SELENA from the C/C++ source code and which tools that areneeded for the different platforms. First download the source code for SELELA. The source code can

be found in the download section on the SELENA/RISe web page at: http://selena.sourceforge.net

If you want to compile the development code from the SVN repository you first need to install asubversion (svn) client. On windows one can, for example, use TortoiseSVN which integrates into theWindows file browser, or a command line client from here http://www.sliksvn.com/en/download. OnLinux there exist several GUI based front ends to svn like, for example, KDESvn, RapidSVN or eSvn.One can also use the command line client. To download the development code one need to do a svn

checkout. First create a folder for the SELENA code, say selena sourceforge and go to that folder.Then, using the command line client, type

$ svn checkout https://selena.svn.sourceforge.net/svnroot/selena/selana/trunk

at the shell command line. This will download all the C# source code, user manual files etc. If you areusing a GUI client then choose the checkout menu/button in the particular GUI client that you are using.

74

http://selena.sourceforge.net

http://tortoisesvn.tigris.org

http://www.sliksvn.com/en/download

DR

AFT


For example, if you are using TortoiseSVN on Windows then right-click on the selena sourceforge folderthat you just created, select the SVN Checkout menu item, and add the path

https://selena.svn.sourceforge.net/svnroot/selena/selena/trunk

in the URL or Repository line. Then click on the OK -button and the files should start downloading.The C/C++ source code can be found in the trunk/src folder.

B.1 Tools and Libraries for Windows

1. Install the GNU Scientific Library (GSL): Binary packages for GSL can be found here:http://gnuwin32.sourceforge.net/packages/gsl.htm but you can also find these files in thedll folder.

2. Copy the libgsl.dll and libgslcblas.dll files to C:\WINDOWS\System32\

3. Install a compiler and mex-tools:

The mex-files, oct-files, and the stand-alone binaries should build with most standard C-compilerssuch MSVC or MinGW. Here we only describe building with MinGW [but there is an untestedbuild mexfiles win.m script which should (may) work with MSVC].

4. Install the MinGW compiler that comes with Qt, from: http://qt.nokia.com/products (recom-mended), or get directly from: http://www.mingw.org. Set the path to the build tools (mingw32-make.exe,mingw32-gcc.exe etc.) by right-clicking on ”My Computer” and edit (append) Advanced− >Enviroment

Variables− >Path variable. For example, add C:\Qt\2010.04\mingw\bin. Avoid ”ProgramFiles” or any folder with a space (blank) in the folder name.

5. Install the GNUmex tools from: http://gnumex.sourceforge.net and run the setup script (followthe instructions on the GNUmex website).

B.2 Building the Stand-alone Application on Linux/Unix

Here we assume that the gcc tools are installed on the system. To build the SELENA application justcd to the src folder and type make which will build the binary selena. It is recommended that you putthe selena binary in a directory which is in your PATH varaiable, such as, /usr/local/bin/, otherwise,you have to use the full path to the binary to run it.

B.3 Building the Stand-alone Application on Windows

Given that you have installed the MinGW tools (described in Section B.1) you can build the Windowsstand-alone binary by cd to the src folder and type mingw32-make -f Makefile mingw32 clean all

(in a cmd window) which will create the file selena.exe

B.4 Building the Stand-alone GUI Application on Linux/Unix

The SELENA GUI is using the QT toolkit which must be installed to build the selena gui application.

1. First you need to install the QT toolkit from http://qt.nokia.com/products for your platformor use your particular Linux distribution’s package manager (emerge, yast etc.).

2. Run qmake (in the src/gui folder).

75

http://gnuwin32.sourceforge.net/packages/gsl.htm

http://qt.nokia.com/products

http://www.mingw.org

http://gnumex.sourceforge.net


DR

AFT


3. Run make

The executable is named selena gui. You can clean everything with make clean (make distclean

which removes the Makefile and binaries as well).

B.5 Building the Stand-alone GUI Application on Windows

Building the stand-alone SELENA GUI is similar in Windows and Linux (UNIX).

1. First you need to install the QT toolkit from http://qt.nokia.com/products for your platform.

On Windows you also need to add the path to the QT tools (qmake etc) and the MinGW compilerthat comes with QT to the Windows PATH (right-click on My Computer − > Advanced − > env.variables. Typically you append something like:

;C:\Qt\2010.04\mingw\bin;C:\Qt\2010.04\qt\bin

to the PATH variable.

2. Run qmake (in the src/gui folder).

3. Run mingw32-make

You can also run the bat-file make gui.bat to build the gui.

The executable is named selena gui.exe. You can clean everything with mingw32-make clean

(mingw32-make distclean which removes the Makefile and binaries as well).

B.6 Building the Linux/Unix mex-files

1. Open the Make.inc file and check that the paths to Matlab fits your installation.

2. Go ro the src folder:

#cd src

3. Build the mex-files:

# make -f Makefile_matlab clean all install

The mex-files will be installed in the m files folder. Note you can edit the Make.inc file to fine-tunethe build for your particular hardware (set CFLAGS etc.).

Alternatively you can also use the build mexfiles.m script to build and install the mex-files (justrun the script in Matlab in the main SELENA folder).

B.6.1 Building the Windows mex-files

1. First look (open in a text editor) at the mexopt.bat file which the setup script generates. SELENAhas two mexopt-files in the mexopt folder which may need to be edited to fit the your installation.That is, you need to set the paths to Matlab, GNUmex tools, and GSL in the two files:

mexopts/mexopts_mingw.bat.inc

mexopts/mexopts_gsl_mingw.bat.inc

76


DR

AFT


Copy (or rename) these files to mexopts mingw.bat and mexopts gsl mingw.bat, respectively.Then edit the paths in these files to fit your installation (eg., set compiler and Matlab paths).

2. Start Matlab

3. Run the build script:

>> build_mexfiles_mingw

This should build the following files:

att_sub.mexw32 gsl_interpolate.mexw32 meanest.mexw32 tree.mexw32

csm.mexw32 humanloss.mexw32 normcdf.mexw32 treeloss.mexw32

curveintersect.mexw32 humanlosshz.mexw32 numdam.mexw32 treemdr.mexw32

damagep.mexw32 imp_dcm.mexw32 spectralshape.mexw32

gmotion.mexw32 losssqm.mexw32 squaredam.mexw32

gmotionp.mexw32 madrs.mexw32 tinv.mexw32

and install them in the m files folder.

B.7 Building the oct-Files

1. Start Octave

2. Run the build script

octave:1> build_oct_files

This should build the following files:

att_sub.oct gmotion.oct imp_dcm.oct tree.oct

csm.oct gsl_interpolate.oct madrs.oct treemdr.oct

curveintersect.oct humanloss.oct spectralshape.oct

damagep.oct humanlosshz.oct squaredam.oct

and install them in the m files folder.

Alternatively (on Linux), one can build the oct-files with:

# make -f Makefile_octave clean all install

77

user manual for the earthquake loss estimation …selena.sourceforge.net/selenamanual.pdfuser manual...

Documents