REXEL v 3.

2 beta
Computer aided code-based real record selection for seismic analysis of structures
© Iunio Iervolino, Carmine Galasso and Eugenio Chioccarelli 2008-2011
Dipartimento di Ingegneria Strutturale, Università degli Studi di Napoli Federico II, Naples, Italy

TUTORIAL
07/27/11 Version

For information: iunio.iervolino@unina.it

 Page |2

Summary
Preface ................................................................................................................................................... 3
Guide to installation ............................................................................................................................... 5
Guide to step by step selection............................................................................................................... 6
Definition of the target spectrum ...................................................................................................... 6
Looking at disaggregation ................................................................................................................. 6
Looking at ID conditional hazard ........................................................................................................ 7
Selection of the records to be considered in the compatibility analysis (Preliminary search) ............ 8
Spectral matching parameters and analysis options ......................................................................... 8
Output Management ........................................................................................................................... 11
EXAMPLES ............................................................................................................................................ 13
Case 1 : Cosenza, SLD ....................................................................................................................... 14
Case 1a: Selection of a set of 7-unscaled accelerograms decreasing ranges of magnitude and
distance. ....................................................................................................................................... 14
Case 1b: Selection of two sets of 7-unscaled accelerograms which don’t share events. .............. 15
Case 1c: Selection of three sets of 7-unscaled accelerograms which don’t share recordings. ...... 17
Case 2: Cosenza, SLC ........................................................................................................................ 19
Case 2a: Selection of a set of 7-scaled accelerograms according to the hazard disaggregation in
terms of PGA and Sa(T=1.0sec). ................................................................................................... 19
Case 3: Forlì, SLV .............................................................................................................................. 21
Case 3a: Selection of a set of 7-unscaled accelerograms by using the Italian Accelerometric
Archive ITACA. .............................................................................................................................. 21
Case 3b: Selection of a set of 7-unscaled accelerograms for spatial analysis (3 components). .... 21
Case 4: Sant’Angelo dei Lombardi, SLV ............................................................................................ 26
Case 4a: Selection of a set of 7-unscaled accelerograms by using disaggregation for Sa(T=1.0s).
..................................................................................................................................................... 26
Case 5: Napoli-Ponticelli, SLC ........................................................................................................... 29
Case 5a: Selection of a set of 7-unscaled accelerograms by using second disaggregation mode for
Sa(T=1.0s)......................................................................................................................................... 29
Case 5b: Selection of a set of 7-scaled accelerograms by using conditional hazard. ....................... 31
Case 5c: Selection of an individual record using PGV as parameter for preliminary database search.
......................................................................................................................................................... 32
Case 5d: Selection of a set of 7-unscaled accelerograms using PGA as parameter for preliminary
database search. .............................................................................................................................. 33
Case 6: Napoli-Fuorigrotta, SLV ....................................................................................................... 34
Caso 6a: Selection of a set of 30-scaled accelerograms. .............................................................. 34
REFERENCES ......................................................................................................................................... 35
 Page |3

Preface
REXEL 3.2 beta, available on the internet on the website of the Italian consortium of
earthquake engineering laboratories: Rete dei Laboratori Universitari di Ingegneria Sismica –
ReLUIS (http://www.reluis.it/), allows to define the design spectra according to the Eurocode
8 (EC8 – CEN, 2003), the new Italian Building Code (NIBC – CS.LL.PP., 2008) or completely
user-defined. Based on these spectra, the software allows to search for sets of 7 records
compatible, in the average, with them.

The datasets included in REXEL are the European Strong-motion Database (ESD) (last
updated on July 2007), whose URL is http://www.isesd.cv.ic.ac.uk, the Italian
Accelerometric Archive (ITACA) (last updated on October 2010) from Istituto Nazionale di
Geofisica e Vulcanologia (INGV), whose URL is http://itaca.mi.ingv.it and the database with
Selected Input Motions for displacement-Based Assessment and Design (SIMBAD) (last
updated on June 2011) developed by Smerzini and Paolucci (2011) in the framework of the
ReLUIS 2010-2013 project, task Displacement Based Approaches for Seismic Assessment of
Structures.

All the records contained in REXEL satisfies the free-field conditions and were produced by
earthquakes of moment magnitude larger than 4 (5 in the case of SIMBAD). Figures below
show distributions regarding ESD, ITACA and SIMBAD in REXEL; the records of SIMBAD are
classified by the countries they come from. In the case of ITACA the Eurocode 8 soil
classification is that from task 2 of the project S4 of INGV (http://esse4.mi.ingv.it/) and may
be updated in the future.

WARNING: ESD, ITACA and SIMBAD have records in common although with different
seismological processing. There was no attempt by the authors to homogenize/combine the
three databases, which are separated in the software. Moreover, the three databases cover
different magnitude and distance ranges, therefore the appropriate database to use in
searches may also depend on which range of magnitude and distance one is interested in,
see figures below.

Magnitude vs epicentral distance distribution for ESD (left) and ITACA (right) datasets on
which REXEL operates. The records are grouped by site class according to EC8
classification.
 Page |4

Magnitude vs epicentral distance distribution for SIMBAD dataset (left) on which REXEL
operates. The records are grouped by site class according to EC8 classification; distribution
of the records of SIMBAD by the countries they come from (right).

The sources of strong ground motion records in SIMBAD are summarized in Table 1.

Table 1 Sources of strong ground motion records of the SIMBAD database.

No.
Country/area Source Website
of records
§ K-NET http://www.k-net.bosai.go.jp/
Japan 136
KiK-net http://www.kik.bosai.go.jp/
ITalian ACcelerometric Archive:
Italy 66 http://itaca.mi.ingv.it/
ITACA
Center for Engineering Strong
http://strongmotioncenter.org/
Ground Motion Data: CESMD
http://peer.berkeley.edu/peer_ground_m
USA 54 PEER Strong Motion Database
otion_database
U.S. Geological Survey National
http://nsmp.wr.usgs.gov/
Strong Motion Project: NSMP
European Strong-Motion Data
Europe 18 http://www.isesd.hi.is/
Base: ESMD
Institute of Geological and
New Zealand 16 http://www.geonet.org.nz
Nuclear Sciences: GNS
Turkish National Strong Motion
Turkey 11 http://daphne.deprem.gov.tr
Project: T-NSMP
Iran Strong Motion Network
Iran 3 http://www.bhrc.ac.ir/
ISMN
§
Records from the 1995 Hyogo-ken Nanbu earthquake come from the ESG98 data distribution CD-ROM for the
Kobe Simultaneous Simulation

If you use REXEL, please cite as: Iervolino I., Galasso C., Cosenza E. (2010). REXEL: computer aided
record selection for code-based seismic structural analysis. Bulletin of Earthquake Engineering.
8:339-362, DOI 10.1007/s10518-009-9146

REXEL and this tutorial may be used and distributed for free while their modification and
commercialization are not authorized. The authors have made every effort in order to ensure the
accurate working of the software; however, any responsibility is declined for wrong results.
 Page |5

Guide to installation
1. Installation of MATLAB Component Runtime (MCR) v 7.11
REXEL was developed in MATLAB environment; for its use the MATLAB Component Runtime
(MCR) 7.11 is required. Please launch the MCRInstaller.exe file and follow the instructions of
the installation procedure. REXEL requires a specific version of MATLAB. If you have a
different version of MATLAB, it will not work with REXEL; for this reason, you must install the
MCR. The MATLAB Component Runtime is a free redistributable that allows you to run
programs written in a specific version of MATLAB without installing the MATLAB version
itself. There is no harm in having MATLAB and the MCR installed simultaneously, or in having
multiple versions of each one installed.

Image of the user interface of the software

2. Installation of REXEL v 3.2 beta

In order to install REXEL, please launch REXELInstaller.exe file and follows the instructions of
the installation procedure. In order to remove REXEL, please select Programs > REXEL v 3.2
beta > Uninstall REXEL 3.2 beta and follow the instructions to complete the procedure.
 Page |6

Guide to step by step selection

The steps needed to search real records sets compatible with a given target spectrum are
described below.

Definition of the target spectrum

First of all, it is necessary to build the acceleration elastic response spectrum for the site of
interest [Build code spectrum]. To this aim, if one wants to make a selection according to
the Italian Building Code spectra, it is necessary to enter the geographical coordinates of the
site, longitude and latitude in decimal degrees, and specify, through a drop-down menu,
developed in accordance with the requirements of code, the Site Class, the Topographic
category, Nominal Life, Functional Type and the Limit State of interest.
For the Eurocode 8 spectra, it is necessary to specify only the anchoring value of the
spectrum, ag, and the ground type (see also Iervolino et al., 2010a). The value of ag can be
defined manually by the user or, in the case of sites on the Italian territory, can be derived
automatically from the geographical coordinates of the site (ag values used in the software
are those of NIBC). A third alternative is the possibility of using a reference spectrum
completely defined by the user [User define spectrum].
In addition, it is necessary to specify the component of the earthquake that will be
considered. The two orthogonal independent components that describe the horizontal
motion (X and Y) are characterized by the same elastic response spectrum, while the
component that describes the vertical motion (Z) is characterized by a specific spectrum. It is
possible to select X, Y and Z for finding three dimensional combinations of motion.
If the specified coordinates do not fall into a node of the reference grid of NIBC, the values
of the parameters useful for the definition of the target spectrum are automatically
calculated as a weighted average of the values that the parameter of interest assumed in the
vertices of the elementary mesh containing the site under examination, using as weights the
inverse of the distances between the site and the four nodes, as specified in Annex A of
NIBC.

Looking at disaggregation
Disaggregation is a procedure which allows to identify, from a probabilistic point of view, the
contribution to the hazard of each couple of magnitude (M) and source to site distance (R).
Contributions are dependent on hazard assessment of the site, spectral ordinates and return
period. REXEL provides disaggregation results of each Italian site for two spectral ordinates,
i.e. 0 second (PGA) and 1.0 second (the latter is referred to medium-long period structures)
and for four return periods, i.e. 50 yrs, 475 yrs, 975 yrs and 2475 yrs. For all different return
periods, REXEL provides automatically results of the closer return period.
To visualize disaggregation distribution, the user must select the spectral ordinate of interest
and press the button [Disaggregation].
Disaggregation can provide indications about the intervals to be used in record selection
(see next section). Using disaggregation results following points have to be considered:
1. Official Italian hazard data and disaggegation results (only for PGA) are provided by INGV
and are available at the web site http://esse1-gis.mi.ingv.it. Disaggregation results
provided by REXEL refer to a specific independent study.
 Page |7

2. Parameters of disaggregation analyses have been fitted for the whole Italian region and
results have been considered generally reliable even if in site with low seismicity, hazard
evaluation (a preliminary step for disaggregation analysis) can be wrong because the
software is not able to capture hazard differences with return period. As consequence,
also disaggregation results do not vary with return period and a particular attention has
to be used. In these condition, approximation is lower for higher return periods. REXEL
shows a warning for all these cases.
3. Disaggegation plots suggest the best intervals of magnitude and distance for record
selection but records’ availability is not insured. Moreover the choice of the right intervals
is up to users.
For further details and any technical aspect about disaggegation analysis refer to Iervolino et
al., 2011 and Convertito et al., 2009.

Looking at ID conditional hazard

Acceleration-based intensity measure (IMs, e.g., spectral ordinates) have been shown to be
important and useful in the assessment of structural response of buildings. However, there
are cases in which it is desirable to account for other ground motion IMs, while selecting
records. For example, although it is generally believed that, under some hypotheses, integral
IMs associated to duration are less important for structural demand assessment with respect
to peak quantities of ground motion, there are cases in which the cumulative damage
potential of the earthquake is also of concern.
An easy yet hazard-consistent way of including secondary IMs in record selection is
represented by the conditional hazard curves (or maps); i.e., curves of secondary ground
motion intensity measures conditional, in a probabilistic sense, to the design hazard for the
primary parameter.
To illustrate the conditional hazard concept, in the study of Iervolino et al. (2010b) the joint
distribution of PGA and a parameter which may account for the cumulative damage
potential of ground motion, was investigated.
The chosen energy related measure is the so-called Cosenza and Manfredi index (ID),
Equation (1), the ratio of the integral of the acceleration squared to the PGA and peak
ground velocity (PGV).


   


 
 
(1)

REXEL 3.2 beta includes the ID conditional hazard results, suggesting to the user the
distribution (in terms of complementary cumulative density functions) of ID value (for Italy)
given the design PGA (i.e. the anchoring value of the target spectrum).
This should allow for improving record input selection for earthquake engineering
applications in a hazard consistent manner yet easily viable for practitioners.
For further details and any technical aspect about conditional hazard analysis refer to
Iervolino et al., 2010b
 Page |8

 Selection of the records to be considered in the compatibility

analysis (Preliminary search)
The user may select the records of database (ESD or ITACA, both contained in REXEL)
corresponding to a given range of:
1. M (moment magnitude for A-D site class records of ESD, local magnitude for E site class
records; moment magnitude for all ITACA records) and R (epicentral), in kilometers (e.g. this
choice may be guided by disaggregation of hazard);
2. peak ground acceleration (PGA) of horizontal components of motion, in fraction of g, the
acceleration of gravity;
3. peak ground velocity (PGV) of horizontal components of motion, in m/s;
4. Cosenza and Manfredi index (ID) (e.g. this choice may be guided by conditional hazard) of
horizontal components of motion;
5. Arias Intensity (IA) of horizontal components of motion, in m/s.
See for example Cosenza and Manfredi (2000) for a complete review of the ground motion
parameters that can be assumed as structural and non-structural damage measures.
To this aim, the user must specify the intervals [min, max] of the ground motion parameter
of interest in which he wants the accelerograms fall and the database of interest.
A parameter that is desirable to include in record selection is the site classification, affecting
both the amplitude and shape of response spectra. However, specifying a close match for
this parameter in record selection may not always be feasible, because for some soft soils,
only a few records are usually available. Moreover, if the spectral shape is assigned by the
code, the site class of real records may be of secondary importance.
In light of these considerations, there may be cases in which it may be useful to relax the
matching criteria for site classification.
Therefore, in REXEL it is possible to select records from Same as target spectrum soil or from
Any site class. This, as shown in the following, should help to overcome some of the
problems when for specific site conditions it is hard to find spectrum matching sets.
Then, the software returns the number of records (and the corresponding number of events)
available in these ranges and to be considered in the compatibility analysis.
After these bounds are defined and confirmed [Check database], the software returns the
number of records (and the corresponding number of originating events) available in the
intervals. This list constitutes the inventory of records in which to search for suites of seven
which are in the average compatible with the code spectra of step 1 above. These spectra
may also be plotted [Preliminary plot] along with the reference spectrum to have a picture
of the spectra REXEL will search among.

 Spectral matching parameters and analysis options

In this step the parameters related to the spectral compatibility are defined. In which period
range, [T1, T2], the target spectrum has to be matched has to be defined first. T1 and T2 can
be any pair in the 0s – 4s range. Secondly, the tolerances allowed in spectral matching are
required. This means the user has to specify the maximum deviations (lower and upper
tolerances) in percentage terms that the average spectrum of the combination can have
with respect to the target in the specified [T1, T2] range.
 Page |9

EC8, for example, explicitly states that the average elastic spectrum must not underestimate
the code spectrum, with a 10% tolerance (lower limit) but does not provide any indication
about the upper limit. It is economically viable to reduce as much as possible the
overestimation of the spectrum and this is why also an upper limit was introduced in REXEL.
In this phase is also possible to select the Scaled records option, which corresponds to
choose whether to search for unscaled or scaled record sets. In fact, REXEL allows to obtain
combinations of accelerograms compatible with the code spectrum which does not need to
be scaled, but it also allows choosing sets of accelerograms compatible with the reference
spectrum if linearly scaled. If this second option is chosen, the user have to check the Scaled
records box, which means the spectra of the list defined in step 2 are preliminarily
normalized dividing the spectral ordinates to their PGA. Combinations of these spectra are
compared to the non-dimensional code spectrum . If this option is selected it is also possible
to specify the maximum mean scale factor (SF) allowed; REXEL will discard combinations
with an average SF larger than what desired by the user.
The user can also select the option I’m feeling lucky in order to stop the analysis after the
first compatible combination is found. This option, in most cases, allows to immediately get
a combination compatible with the reference spectrum, otherwise, the search for
compatible combinations may take a very long time (as warned by the software).
Alternatively, the maximum number of compatible combinations to find after which the
search has to stop, can be specified.

 Combination search
At this point the initial list of records and the analysis’ parameters are set and it is possible to
decide which kind of search to perform. The software searches for:

(a) 7 1-component accelerograms [7 records (1 component)] whose average matches the

target spectrum in the specified range of periods and with the provided upper- and
lower-bound tolerances. The found combinations can be applied in one direction for
plane analysis of structures1;

(b) 7 pairs of accelerograms [7 records (2 component)]. This option allows to search for 14
records which are 7 2-components recordings (both X and Y components of 7 recording
stations only), which on average are compatible with the target spectrum; this kind of
search is for cases in which horizontal motion has to be applied in both directions of a
3D building;

(c) 7 triplets of accelerograms [7 records (3 component)] which include the two horizontal
and the vertical component of seven recording stations (i.e., 21 records which are the X,
Y and Z components of 7 recording stations only) for full 3D analysis. In this case, the
selection proceeds in two sub-steps: first, the combinations compatible with the
horizontal component of the code spectrum are found, exactly as case (b); then, the
software analyzes the vertical components of only those horizontal combinations which
have been found to be compatible with the code spectrum, and verifies if the set of

1
Note that this options applies alternatively to horizontal or vertical components of motion, although it is
unlikely one is looking for a suites of seven vertical accelerograms only.
 P a g e | 10

their seven vertical components are also compatible with the vertical code spectrum.
The tolerances and period ranges in which average spectral matching of vertical
components may be different from that regarding horizontal components and may
defined by the user after the horizontal analysis has finished.

(d) 30 1-component accelerograms [30 records (1 component)] whose average matches the
target spectrum in the specified range of periods and with the provided upper- and
lower-bound tolerances. The found combinations can be applied in one direction for
plane analysis of structures2;

(e) 30 pairs of accelerograms [7 records (2 component)]. This option allows to search for 60
records which are 30 2-components recordings (both X and Y components of 30
recording stations only), which on average are compatible with the target spectrum; this
kind of search is for cases in which horizontal motion has to be applied in both
directions of a 3D building;

(f) Individual (horizontal) records [Individual record search] matching the target spectrum
in the specified range of periods and with the provided upper- and lower-bound
tolerances.

An important feature of the code is that the list of records out of step 2, which is an input for
this phase, are ordered when the analysis is launched in ascending order of the parameter
defined in Equation 1, which gives a measure of how much the spectrum of an individual
record deviates from the spectrum of the code. In Equation (2), Saj(Ti) is the pseudo-
acceleration ordinate of the real spectrum j corresponding to the period Ti, while Satarget(Ti)
is the value of the spectral ordinate of the code spectrum at the same period, and N is the
number of values within the considered range of periods.

$
   !" 
   ∑

%& 
  !" 
# (2)

Preliminary ordering allows to analyze first the records which have a similar spectral shape
with respect to the target. This ensures the first combinations, e.g. the one found with the
I’m feeling lucky option, to be those with the smallest individual scattering in respect to the
target spectrum as shown in the following examples.

2
Note that this options applies alternatively to horizontal or vertical components of motion, although it is
unlikely one is looking for a suites of seven vertical accelerograms only.
 P a g e | 11

Output Management
After the analysis has finished, REXEL returns [Output > Results > Horizontal (Vertical)] a list
of combinations whose average spectrum respects compatibility with the target in the
chosen range of periods and with the assigned tolerance.
The results of the analysis are sorted so that record combinations with the smallest
deviation from the spectrum of the code are the first of the output list, due to the
preliminary ordering of records according to δj.
For all the combinations found, it is possible to calculate [Output > Deviation > Horizontal
(Vertical)] the deviation of each accelerograms of the combination compared to the target
spectrum, and the deviation of the average spectrum of the combination compared to the
elastic response spectrum again according to Equation (2), in which the average spectral
acceleration of the records replaces Saj(Ti).
Combinations returned are uniquely identified by a serial number; this code can be used
[Output > Results > Plot & get set > Horizontal (Vertical) for 7 record search or Output >
Results > Plot & get record for individual record search] to graphically display the spectra of
a specific combination of interest and obtain the spectra and acceleration time-histories
grouped in a compressed file by REXEL.
If a specific combination is chosen REXEL also returns the mean value of magnitude and
distance of combination and the information about the individual records as retrieved by
ESD and ITACA that is, recording station, event (time, date and country), magnitude,
distance, fault mechanism, etc [Output > Info records plot]. In the case of non-dimensional
sets, the scale factors of individual record and the mean scale factor of the combination are
given.

Different 2 and Different 3 options

Although both considered codes only require a minimum size for suites of 7 to consider
mean effects on the structure, it is known how this number may affect the confidence (i.e.,
the standard error) in the estimation of the structural response, which increases as the
variability of individual records with respect to the target increases. Therefore, in case one
wants to run the analyses with a larger number of records REXEL has two options which both
are applicable to the list of results of the analyses discussed above:

- Different 2 allows to search within the list of output, pairs of combinations, i.e., 2 sets of 7
records of the type (a), (b), or (c) above, which have accelerograms from earthquake
events which do not overlap. This allows to have a larger set of 14 one- or multi-
component records, spectrum matching in the average, in which there are no dominating
events. For each found pair of sets the software also computes the maximum deviation of
the two original combination and this may be a parameter for choosing one pair in
respect to another;

- Different 3 allows to search within the list of output, triplets of combinations (3 seven of 7
one- or multi-component records) having no accelerograms in common although may
have common events. This analysis provides sets of 21 record which, still, on match the
code-required average spectral compatibility with the given tolerances.
 P a g e | 12

Repeat search excluding a station

In some cases, the analyst may want to exclude a particular record appearing in a found
combination. To this aim, REXEL ncludes the option Repeat search excluding a station, which
allows to repeat the performed analysis by excluding from the list of records created in the
Preliminary database search one or more waveforms, in an iterative way.

Displacement spectra compatibility

REXEL allows to check the displacement spectra compatibility for a combination selected to
match a pseudo-acceleration spectrum. For each period T, the elastic spectral displacement
is computed as inverse transformation of elastic pseudo-spectral acceleration.
 P a g e | 13

EXAMPLES
In the following some illustrative examples of automatic selection of spectrum-matching
accelerogram, using REXEL v 3.2 beta, are shown as well as some “strategies” aiming at a
better selection, although for further details the reader should refer to Iervolino et al.
(2010a). In each example, the target spectra considered are built according to NIBC even if
the software itself permits to define any type of Eurocode 8 or user-defined spectrum. The
examples refer to different Italian sites, with different geographical location, site conditions
(according to EC8) and hazard level as expressed by the maximum value of acceleration on
rock (ag) with a 10% probability of exceedance in 50 years:
 Forlì (Central Italy: latitude 44.218°; longitude 12.054);
 Cosenza (Southern Italy: latitude 39,314°; longitude 16.215°);
 Sant’Angelo dei Lombardi (Southern Italy: latitude 40.8931°; longitude 15.1784°);
 Ponticelli (Southern Italy: latitude 40.8516°; longitude 14.3446°);

The considered cases are:

Case 1 : Cosenza, Site class A, selection for Damage Limit State (SLD)
 Case 1a: selection of a set of unscaled accelerograms decreasing magnitude and distance
ranges;
 Case 1b: selection of two sets of unscaled accelerograms which don’t share earthquake
events;
 Case 1c: selection of three sets of unscaled accelerograms which don’t share records.

Case 2: Cosenza, Site class A, selection for Collapse Limit State (SLC)
 Case 2: selection of a set of scaled accelerograms according to the hazard disaggregation
in terms of PGA and Sa(T=1.0s).

Case 3: Forlì, Site class B, selection for Operability Limit State (SLV)
 Case 3a: selection of a set of unscaled accelerograms according to the hazard
disaggregation in terms of PGA and using the Italian Accelerometric Archive (ITACA).
 Case 3b: Selection of a set of unscaled accelerograms for spatial analysis (3 components).

Case 4: Sant’Angelo dei Lombardi, Site Class A, selection for Collapse Limit State (SLV):
 Case 4: selection of a set of unscaled accelerograms according to the hazard
disaggregation in terms of Sa(T=1.0s).

Case 5: Ponticelli, Site Class B, selection for Collapse Limit State (SLC):
 Case 5a: selection of a set of unscaled accelerograms according to the hazard
disaggregation (second mode) in terms of Sa(T=1.0s).
 Case 5b: selection of a set of scaled accelerograms by using conditional hazard.
 Case 5c: selection of an individual record using PGV as parameter for preliminary
database search.
 Case 5d: selection of a set of unscaled accelerograms using PGA as parameter for
preliminary database search.

Case 6: Fuorigrotta, Site Class A, selection for Life Safety Limit State (SLV):
 Case 6a: selection of a set of 30-unscaled accelerograms
 P a g e | 14

Case 1 : Cosenza, SLD

In the following, a set of 7 unscaled accelerograms for Cosenza (Latitude 39.314, Longitude
16.215) considering the Damage Limit State (SLD), a nominal life of 50 years, functional type
II, topographic category T1 and soil type A, according to NIBC, is selected. The first step is to
define the above mentioned parameters to determine the elastic acceleration response
spectrum, that is the target spectrum (Figure 1).

Figure 1. Definition of the target spectrum for Cosenza, case 1

Case 1a: Selection of a set of 7-unscaled accelerograms decreasing ranges of

magnitude and distance.

The range of magnitude [4.8 – 7.3] and of distance [0 km– 50 km], obtained by the
disaggregation of seismic hazard in terms of Sa(T=1.0sec) with a probability of exceedance of
63% in 50 years (Figure 2a), allows to find 2x219 accelerograms from 99 different events in
the ESD (Figure 2b). Because of the large number of accelerograms found, the search may
take a very long time, so it is recommended to reduce the magnitude and distance ranges,
excluding the records with a magnitude value lower than 5.5 and higher than 6.5 and with a
distance values higher than 20. In this way intervals with higher hazard contribution are
considered only (another selection of records may be done considering higher values of
distance and magnitude). With new intervals the software finds 2x31 compatible records
from 15 different events. In order to search for 1-component record sets, assigning a
compatibility tolerance between 10% and 30% in the range of period [0.15s – 2s], in a few
seconds the software finds 100000 compatible combinations and, by default, it displays the
first combination (Figure 3).
 P a g e | 15

(a)

(b)
Figure 2. Disaggregation of the hazard in terms of Sa(T=1.0sec) for Cosenza for a return period of
50 years (a) and definition of magnitude and distance intervals (b).

Figure 3. The first combination found for Cosenza, case 1a

Case 1b: Selection of two sets of 7-unscaled accelerograms which don’t share
events.

In the following, it is supposed one wants to select two sets of 1-componets ground motions
starting from the case 1a. The user can use the option Different 2; this provides pairs of sets
with records which come from non-overlapping events. Using this option the software finds
in about five minutes 100000 compatible pairs and opens DIVERSI.txt file by default, where
 P a g e | 16

the combinations are listed (Figure 4). To decrease the time of the analysis it is suggested to
select a lower number of desired pairs. Referring to the first pair found (Figure 5 a and b),
the software provides 14 registrations from 12 different events.

Figure 4. Results by using the option Different 2.

C om binazione 17372

Epicentral EC8 Site

W aveform ID Earthquake ID Station ID Earthquake N am e Date Mw Fault M echanism Distance [km ] class
1243 473 ST 575 Izm it (aftershock) 13/09/1999 5.8 oblique 15 A
5655 1825 ST 2950 NE of Banja Luka 13/08/1981 5.7 oblique 10 A
359 174 ST 136 Um bria 29/04/1984 5.6 norm al 17 A
473 228 ST 40 Vrancea 31/05/1990 6.3 thrust 7 A
383 176 ST 153 Lazio Abruzzo (aftershock) 11/05/1984 5.5 norm al 14 A
4675 1635 ST 2487 South Iceland 17/06/2000 6.5 strike slip 13 A
7142 2309 ST 539 Bingol 01/05/2003 6.3 strike slip 14 A
m edie: 5.957143 12.85714286
a) Combination 17372 ( Output> Plot & get records >17372)
 P a g e | 17

Combinazione 54092
Epicentral EC8 Site
Waveform ID Earthquake ID Station ID Earthquake Name Date Mw Fault Mechanism Distance [km] class
365 175 ST140 Lazio Abruzzo 07/05/1984 5.9 normal 5A
6342 2142 ST2556 South Iceland (aftershock) 21/06/2000 6.4 strike slip 20 A
3802 1226 ST2368 SE of Tirana 09/01/1988 5.9 thrust 7A
365 175 ST140 Lazio Abruzzo 07/05/1984 5.9 normal 5A
149 65 ST26 Friuli (aftershock) 15/09/1976 6 thrust 12 A
652 292 ST236 Umbria Marche (aftershock) 14/10/1997 5.6 normal 12 A
6326 2142 ST2496 South Iceland (aftershock) 21/06/2000 6.4 strike slip 14 A
medie: 6.0143 10.71428571
b) Combination 54092 (Output > Plot & get records > 54092)

Figure 5. The first pair of sets (Different 2) found for Cosenza, case 1b

Case 1c: Selection of three sets of 7-unscaled accelerograms which don’t share
recordings.

In the following it is supposed one wants to select three sets of 1-componets ground
motions, for the case 1a, with no records in common. The user can use the option Different
3; in about ten minutes the software finds 100000 triplets of compatible combinations and
opens the DIVERSI.txt file by default where the triplets are listed (Figure 6). The first triplet is
shown in Figure 7.

Figure 6. Results by using the option Different 3.

C om binazione 419
Fault Epicentral EC 8 Site
W aveform ID Earthquake ID Station ID Earthquake N am e D ate Mw M echanism D istance [km ] class
1243 473 ST575 Izm it (aftershock) 13/09/1999 5.8 oblique 15 A
365 175 ST140 Lazio Abruzzo 07/05/1984 5.9 norm al 5 A
359 174 ST136 U m bria 29/04/1984 5.6 norm al 17 A
3802 1226 ST2368 SE of T irana 09/01/1988 5.9 thrust 7 A
149 65 ST26 Friuli (aftershock) 15/09/1976 6 thrust 12 A
359 174 ST136 U m bria 29/04/1984 5.6 norm al 17 A
6326 2142 ST2496 South Iceland (aftershock) 21/06/2000 6.4 strike slip 14 A
m edie: 5.885714286 12.42857143
a) Combination 419 (Output>Plot & get records > 419)
 P a g e | 18

Combinazione 89881
Fault Epicentral EC8 Site
Waveform ID Earthquake ID Station ID Earthquake Name Date Mw Mechanism Distance [km] class
5655 1825 ST2950 NE of Banja Luka 13/08/1981 5.7 oblique 10 A
5655 1825 ST2950 NE of Banja Luka 13/08/1981 5.7 oblique 10 A
385 176 ST155 Lazio Abruzzo (aftershock) 11/05/1984 5.5 normal 15 A
385 176 ST155 Lazio Abruzzo (aftershock) 11/05/1984 5.5 normal 15 A
6115 2029 ST1320 Kozani 13/05/1995 6.5 normal 17 A
4675 1635 ST2487 South Iceland 17/06/2000 6.5 strike slip 13 A
6335 2142 ST2557 South Iceland (aftershock) 21/06/2000 6.4 strike slip 15 A
medie: 5.971428571 13.57142857
b) Combination 89881 (Output> Plot & get records > 89881)

Combinazione 99444
Fault Epicentral EC8 Site
Waveform ID Earthquake ID Station ID Earthquake Name Date Mw Mechanism Distance [km] class
6342 2142 ST2556 South Iceland (aftershock) 21/06/2000 6.4 strike slip 20 A
6341 2142 ST2497 South Iceland (aftershock) 21/06/2000 6.4 strike slip 20 A
6341 2142 ST2497 South Iceland (aftershock) 21/06/2000 6.4 strike slip 20 A
670 291 ST238 Umbria Marche (aftershock) 06/10/1997 5.5 normal 20 A
766 292 ST266 Umbria Marche (aftershock) 14/10/1997 5.6 normal 12 A
382 176 ST140 Lazio Abruzzo (aftershock) 11/05/1984 5.5 normal 16 A
242 115 ST225 Valnerina 19/09/1979 5.8 normal 5 A
medie: 5.942857143 16.14285714
c) Combination 99444 (Output> Plot & get records >99444)

Figure 7. The first triplet found by using the option Different 3.

 P a g e | 19

Case 2: Cosenza, SLC

Case 2a: Selection of a set of 7-scaled accelerograms according to the hazard
disaggregation in terms of PGA and Sa(T=1.0sec).

In the following, a set of unscaled accelerograms for Cosenza (Latitude 39.314, Longitude
16.215) considering the Collapse Limit State (SLC) of a structure on a type A soil with T1
topographic category, 50 years nominal life and Functional Class II, is selected. The first step
is to define the above mentioned parameters to determine the elastic acceleration response
spectrum, that is the target spectrum.
Only for structure of low fundamental periods (e.g. for T minor than 0.5 sec), disaggregation
of PGA may be more indicated for addressing record selection. Anyway for this case,
disaggregation of PGA and Sa(T=1.0sec) with a probability of exceedance of 5% in 50 years
are both shown in Figure 83. The ranges of magnitude [5.5 – 7] and of distance [0 km – 20
km] have been used and 2x32 (soil class A) accelerograms from 6 different events in the ESD
have been found.

(a) (b)
Figure 8. Disaggregation of the hazard in terms of PGA (a) and Sa(T=1.0sec) (b) for Cosenza for a
return period of 975 years.

Assigning a compatibility tolerance between 10% and 30% in the range of period [0.15s – 2s],
the software doesn’t find compatible combinations. By clicking on the Preliminary plot
option (Figure 9), it is clear that the analysis won’t give back a positive response if the
choiche falls on an unscaled set. The average spectrum of the scaled records seems more
encouraging, hence we prefer this solution. By clicking on Non-Dimensional option and by
choising a maximum scale factor equal to 2, the software provides 2 compatible

3
In this case, disaggregation results do not change significantly between PGA and Sa(T=1.0sec). For other sites,
changes can be more relevant and it is not possible to indentify a general trend without considering further
details about hazard and disaggregation assessment.
 P a g e | 20

combinations, with the same tolerance limits, the first of which is indicated in Figure 10.
Note that REXEL only limits the average scale factor, which may be larger than prescribed for
individual records.

Figure 9. Preliminary plot for Cosenza SLC: Case 2

Figure 10. The first combination found for Cosenza, case 2

 P a g e | 21

Case 3: Forlì, SLV

Case 3a: Selection of a set of 7-unscaled accelerograms by using the Italian
Accelerometric Archive ITACA.

In the following, a set of unscaled accelerograms for Forlì (Latitude 44.218, Longitude
12.054) considering the Safe Life Limit State (SLV), a nominal life of 50 years, functional type
II, topographic category T1 and soil type B, according to NIBC, is selected. The first step is to
define the above mentioned parameters to determine the elastic acceleration response
spectrum, that is the target spectrum. Defining arbitrary the range of magnitude [6 – 7] and
of distance [0 km – 40 km], selecting the ITACA as database, considering the recordings of
any site class, REXEL finds 2x30 accelerograms from 5 different events .
Searching for 1-component ground motion sets, assigning a compatibility tolerance between
10% and 30% in the range of period [0.15s – 2s], the software finds 100000 compatible
combinations, the first of which is shown in Figure 11.

Figure 11. The first scaled combination found for Forlì, Case 3a

Case 3b: Selection of a set of 7-unscaled accelerograms for spatial analysis (3

components).

In the following, a set of unscaled accelerograms for Forlì (Latitude 44.218, Longitude
12.054) considering the safe life limit state (SLV), a nominal life of 50 years, functional type
II, topographic category T1 and soil type B, according to NIBC, is selected. In this case it is
searched for a set which includes all three components of ground motion (i.e., including the
vertical one). The first step is to define the above mentioned parameters to determine the
horizontal and vertical elastic acceleration response spectra (Figure 12).
 P a g e | 22

Figure 12. Definition of the target spectrum for Forlì, case 3

Considering intervals of magnitude and distance of respectively [4.5, 6] and [0, 30km]
obtained from the disaggregation of seismic hazard in terms of Sa(T=1.0sec) with a
probability of exceedance of 10% in 50 years (Figure 13), selecting ESD as database,
considering only the recordings on B class soil (Same as target spectrum), REXEL finds 3x237
accelerograms referred to 149 different events. Because of the large number of
accelerograms found, the search may take a very long time; moreover, clicking on the
Preliminary plot option (Figure 14), it is clear that the analysis won’t give back a positive
response.

Figure 13. Disaggregation of the hazard in terms of Sa(T=1.0sec) for Forlì for a return period of 475
years.
 P a g e | 23

Figure 14. Preliminary plot referring to M [4.5, 6] and R [0, 30Km]

Increasing the range of magnitude ([6, 8]) and decreasing the range of distance ([10, 20]),
the software finds 3x24 accelerograms from 12 different events. Assigning a compatibility
tolerance between 10% and 30% in the range of period [0.15s – 2s] for the horizontal
component and assigning a compatibility tolerance between 10% e 70% in the range of
period [0.15 s – 1 s] for the vertical component, REXEL finds 1005 compatible combinations
for the horizontal component, but doesn’t find any compatible set with spectrum-
compatible vertical components. At this point the program asks the user if he wants to
change the compatible limits for the vertical component and then the user can decide to
decrease the range of periods or to increase the tolerance limits.
It’s possible, for the same scenario of magnitude [6, 8] and distance [10, 20], to choose
records from all type of soil (Any site class) increasing the numbers of registrations to be
processed (Figure 15).
 P a g e | 24

Figure 15. Preliminary plot referring to M [6, 8] and R [10, 20Km]; Any site class

Processing 3x42 accelerograms from 19 events, assigning a compatibility tolerance between

10% and 30% in the range of period [0.15s – 2s] for the horizontal component and assigning
a compatibility tolerance between 10% e 70% in the range of period [0.15 s – 1 s] for the
vertical component, REXEL finds 100000 combinations for the horizontal component and 31
for the vertical component . The program shows by default the first compatible three
dimensional combination found, which is that featuring the ID (referred to horizontal
components) number 53644 (Figure16).
 P a g e | 25

Epicentral EC8 Site

Waveform ID Earthquake ID Station ID Earthquake Name Date Mw Fault Mechanism
Distance [km] class
333 157 ST121 Alkion 24/02/1981 6.6 normal 20 C
1313 474 ST1100 Ano Liosia 07/09/1999 6 normal 16 B
134 63 ST24 Friuli (aftershock) 15/09/1976 6 thrust 14 B
170 81 ST46 Basso Tirreno 15/04/1978 6 oblique 18 C
4673 1635 ST2482 South Iceland 17/06/2000 6.5 strike slip 15 B
6277 1635 ST2558 South Iceland 17/06/2000 6.5 strike slip 15 A
535 250 ST205 Erzincan 6.6 strike slip 13 B
medie: 6.314286 15.85714286

Figure 16. The first 3-components compatible combination found (Combination no. 53644) for
Forlì, case 3b.
 P a g e | 26

Case 4: Sant’Angelo dei Lombardi, SLV

Case 4a: Selection of a set of 7-unscaled accelerograms by using
disaggregation for Sa(T=1.0s).

In the following, a set of unscaled accelerograms for Sant’Angelo dei Lombardi (Avellino)
(Latitude 40.8931, Longitude 15.1784) considering the Life Safety Limit State (SLV), a
nominal life of 50 years, functional type II, topographic category T1 and soil type A,
according to NIBC, is selected. The first step is to define the above mentioned parameters to
determine the elastic acceleration response spectrum, that is the target spectrum (Figure
17).

Figure 17. Definition of the target spectrum for Sant’Angelo dei Lombardi, case 4

Starting from the disaggregation of seismic hazard in terms of the Sa(T=1.0sec) with a
probability of exceedance of 10% in 50 years (Figure 18), selecting the ESD as database,
considering only the recordings on A class soil (Same as target spectrum) with high
magnitude [6.4-7.2] and low distance [5-15 km], REXEL finds 2x8 accelerograms referred to
3 different events in the ESD database. Clicking on the Preliminary plot option (Figure 19), it
is possible to imagine that the analysis will give back a positive response.
 P a g e | 27

Figure 18. Disaggregation of the hazard in terms of Sa(T=1.0sec) for Sant’Angelo dei Lombardi for a
return period of 475 years.

Figure 19. Preliminary plot referring to M [6.4, 7.2] and R [5, 15Km]; A class soil

Assigning a compatibility tolerance between 10% and 30% in the range of period [0.15s – 2s]
for the horizontal component, REXEL finds, by I’m feeling lucky option, the compatible
combination of Figure 20.
 P a g e | 28

Figure 20. Set found for San’Angelo dei Lombardi, case 4

 P a g e | 29

Case 5: Napoli-Ponticelli, SLC

Case 5a: Selection of a set of 7-unscaled accelerograms by using second
disaggregation mode for Sa(T=1.0s).

In the following, a set of unscaled accelerograms for Ponticelli (Napoli) (Latitude 40.8516,
Longitude 14.3446) considering the Collapse Limit State (SLC), a nominal life of 100 years,
functional type IV, topographic category T1 and soil type B, according to NIBC, is selected.
The first step is to define the above mentioned parameters to determine the elastic
acceleration response spectrum, that is the target spectrum (Figure 21).

Figure 21. Definition of the target spectrum for Ponticelli, case 5

Starting from the second disaggregation mode of seismic hazard referred to the Sa(T=1.0sec)
with a probability of exceedance of 2% in 50 years (Figure 22), selecting the ESD as database,
considering only the recordings on B class soil (Same as target spectrum) with high
magnitude [6.3-7.6] and medium-high distance [15-100 km], REXEL finds 2x60
accelerograms referred to 20 different events in the database. Clicking on the Preliminary
plot option (Figure 23), it is possible to imagine that the analysis will give back a positive
response.
 P a g e | 30

Figure 22. Disaggregation of the hazard in terms of Sa(T=1.0sec) for Ponticelli for a return period of
475 years.

Figure 23. Preliminary plot referring to M [6.4, 7.2] and R [5, 15Km]; A class soil

Assigning a compatibility tolerance between 10% and 30% in the range of period [0.15s – 2s]
for the horizontal component, REXEL finds, by I’m feeling lucky option, the compatible
combination of Figure 24.
 P a g e | 31

Figure 24. Set found for Ponticelli, case 5a

Case 5b: Selection of a set of 7-scaled accelerograms by using conditional

hazard.

Considering again the same example in Ponticelli, let’s consider selection of horizontal scaled
accelerograms according to conditional hazard (Iervolino et al., 2010b) in terms of ID given
the design value of PGA (i.e. the value of PGA from code-based target spectrum). To this aim,
using the [ID conditional hazard] option of REXEL, the software returns the complementary
cumulative density functions of ID conditional on PGA for the site (Figure 25).

Figure 25. Probability of exceedance of ID given PGA for Ponticelli, case 5b

 P a g e | 32

Specifying the ID intervals to [5, 10], i.e., an intervals chosen on the basis of conditional
hazard in terms of ID with an exceedance probability greater than 50% given PGA (Figure 25),
and selecting the Same as target spectrum option, REXEL founds 105 x 2 components of
motion record from 94 different earthquakes (ESD database). Assigning, as tolerances for
the average spectral matching, 10% lower and 30% upper in the period range 0.15s ÷ 2s and
selecting the options to stop the search after the first combination is found, REXEL
immediately returns the combination of scaled accelerograms in Figure 26.

Figure 26. Set found for Ponticelli, case 5b

Case 5c: Selection of an individual record using PGV as parameter for

preliminary database search.

Considering again the same example in Ponticelli, let’s consider selection of an individual
horizontal record using PGV as parameter for preliminary database search.
Specifying the PGV intervals to [0.3m/s, 0.5m/s] and selecting the Same as target spectrum
option, REXEL founds 4 x 2 components of motion record from 4 different earthquakes from
ESD. Assigning, as tolerances for the average spectral matching, 50% lower and 50% upper
in the period range 0.15s ÷ 2s and selecting the options to stop the search after the first
record is found, REXEL immediately returns the record in Figure 27.
 P a g e | 33

Figure 27. Record found for Ponticelli, case 5c

Case 5d: Selection of a set of 7-unscaled accelerograms using PGA as

parameter for preliminary database search.

Considering again the same example in Ponticelli, let’s consider selection of an individual
horizontal record using PGA as parameter for preliminary database search.
Specifying the PGA intervals to [0.3g, 0.5g] and selecting the Any site class option (for
records from ESD), REXEL founds 14 x 2 components of motion record from 13 different
earthquakes. Assigning, as tolerances for the average spectral matching, 10% lower and 30%
upper in the period range 0.15s ÷ 2s and selecting the options to stop the search after the
first combination is found, REXEL immediately returns the set in Figure 28.

Figure 28. Set found for Ponticelli, case 5d

 P a g e | 34

Case 6: Napoli-Fuorigrotta, SLV

Caso 6a: Selection of a set of 30-scaled accelerograms.

In the following, a set of 30-scaled accelerograms for Fuorigrotta (Napoli) (Latitude 40.829,
Longitude 14.191) considering the Safe Life Limit State (SLV), a nominal life of 50 years,
functional type II, topographic category T1 and soil type A, according to NIBC, is selected.
Selecting SIMBAD as database, considering recordings on any site class with high magnitude
[6-7] and distance [0-30 km], REXEL finds 2x146 accelerograms referred to 81 different
events in the database.
Assigning a compatibility tolerance between 10% and 30% in the range of period [0.15s – 2s]
for the horizontal component, REXEL finds, by I’m feeling lucky option, the compatible
combinations of Figure 29 (1-component) and Figure 30 (2-components).

Figure 29. Set (1-component) found for Fuorigrotta, case 6a

Figure 30. Set (2-components) found for Fuorigrotta, case 6a

 P a g e | 35

REFERENCES
CEN, European Committee For Standardisation (2003). Eurocode 8: Design provisions for
earthquake resistance of structures, Part 1.1: general rules, seismic actions and rules for
buildings. Pren1998-1.

Iervolino I., Chioccarelli E., Convertito V. (2011). Engineering design earthquakes from
multimodal hazard disaggregation, Soil Dynamics and Earthquake Engineering, 31(9): 1212–
1231.

Convertito V., Iervolino I., Herrero A. (2009). The importance of mapping the design
earthquake: insights for southern Italy. Bulletin of the Seismological Society of America,
99(5), 2979–2991.

Cosenza E., Manfredi G. (2000). Damage Index and Damage Measures. Progress in Structural
Engineering and Materials, 2(1): 50-59.

CS.LL.PP (2008). DM 14 Gennaio 2008: Norme tecniche per le costruzioni. Gazzetta Ufficiale
della Repubblica Italiana, 29.

Iervolino I., Galasso C., Cosenza E. (2010a). REXEL: computer aided record selection for code-
based seismic structural analysis. Bulletin of Earthquake Engineering, 8:339-362.

Iervolino I., Giorgio M., Galasso C., Manfredi G. (2010b). Conditional hazard maps for
secondary intensity measures. Bulletin of the Seismological Society of America, 100(6): 3312–
3319.

Smerzini C., Paolucci R. (2011). SIMBAD: a database with Selected Input Motions for
displacement-Based Assessment and Design – 1st release. Report of DPC-ReLUIS 2010-2013
project (http://wpage.unina.it/iuniervo/SIMBAD_Database_Polimi.pdf).