See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/245947909

Displacement spectra for seismic design

Article  in  Journal of Earthquake Engineering · January 1999

DOI: 10.1080/13632469909350338

CITATIONS READS
88 906

2 authors:

Julian Bommer Amr S. Elnashai

Imperial College London University of Houston
210 PUBLICATIONS   11,897 CITATIONS    306 PUBLICATIONS   7,103 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Croatian Science Foundation project IP-06-2016-5325: Seismic base isolation of a building by using natural materials - shake
table testing and numerical modeling View project

Thyspunt SSHAC Level 3 PSHA View project

All content following this page was uploaded by Julian Bommer on 14 June 2016.

The user has requested enhancement of the downloaded file.

December 1, 1998 10:23 WSPC/124-JEE 0023

Journal of Earthquake Engineering, Vol. 3, No. 1 (1999) 1–32

c Imperial College Press

DISPLACEMENT SPECTRA FOR SEISMIC DESIGN

JULIAN J. BOMMER and AMR S. ELNASHAI

Department of Civil and Environmental Engineering, Imperial College of Science,
Technology & Medicine, London SW7 2BU, UK

Received 17 August 1998

Revised 27 September 1998
Accepted 5 October 1998

Displacement-based seismic design and assessment of structures require the reliable def-
inition of displacement spectra for a wide range of periods and damping levels. The
displacement spectra derived from acceleration spectra in existing seismic codes do not
provide a suitable answer and there are no existing frequency-dependent attenuation
relationships derived specifically for this purpose. Using a carefully processed dataset of
European strong motion records, attenuation relationships have been derived for hori-
zontal displacement response spectral ordinates. The results have been simplified into
a parametric form that allows the straightforward construction of displacement design
spectra for rock, stiff soil and soft soil sites at distances of up to 50 km from earthquakes
with magnitudes between 5.5 and 7.5, for six damping levels and up to response peri-
ods of 3.0 seconds. Results from recent studies linking levels of ductility to equivalent
damping, using a complex hysteretic model and the same strong motion databank, are
also reported.

Keywords: displacement-based design, displacement spectra, attenuation relationships,

substitute structure.

1. Preamble
1.1. Force-based and displacement-based seismic design
Conventional seismic design, as employed in codes of practice, is entirely force-
based, with a final check on structural displacements. The reasons for this situation
are more historical than scientific. Force-based design is suited to design for ac-
tions that are permanently applied and where members are designed to resist the
effects of these actions at levels of stress constrained by their force resistance at the
plastic limit. The deformations corresponding to the plastic member capacity are
not normally excessive, and evaluating them is not an onerous task. Since seismic
design was developed as an extension to primary load design, it followed the same
procedure, noting though that inelastic deformations may be utilised to absorb
quantifiable levels of energy, leading to reduction in the forces for which structures
are designed. This led to the creation of the response modification (or behaviour)
factor; this all-embracing parameter purports to account for over-strength, ductility,
energy absorption and dissipation as well as the structural capacity to re-distribute

1
December 1, 1998 10:23 WSPC/124-JEE 0023

2 J. J. Bommer & A. S. Elnashai

actions from inelastic highly stressed regions to other less stressed locations in the
structure. Problems of evaluating behaviour factors that are generally applicable
to various structural systems, materials, configurations and input motions are well
documented and the inherent weakness in code-specified factors is widely accepted.
In force-based design, the primary input to the process is a set of forces, with a
check on the level of deformation corresponding to the attainment of values of be-
haviour factors equal to or higher than the design value. In contrast, displacement-
based design inverts the process. Here, the primary design quantity is a target
displacement. If the level of damping of an equivalent linear (substitute) system is
known, corresponding to the target displacement, then the period of vibration of the
required structure may be readily available from a displacement spectrum. Armed
with the knowledge of the required period of vibration, the designer can dimension
the structure with the stiffness, strength and ductility that ensure the realisation
of the target displacement. Moreover, control of the equivalent damping, through
inelastic deformations, is availed of. This process replaces the ‘elastic acceleration
spectrum’ and its derivatives with the ‘displacement spectrum’ as the centre-piece
of the design process. Since damage of structures subjected to earthquakes is cer-
tainly expressed in deformations (strains at fibres, curvatures at sections, rotations
at members and drift at storey levels), displacement-based approaches are concep-
tually more appealing.
The origins of displacement-based design may be traced to work published as
early as the 1960s, where comments on the displacements of inelastic systems and
their relationship to their elastic counterparts were made [e.g., Muto et al., 1960,
as reported by Moehle, 1992]. However, it was the work of Sozen and his associates
that developed the concept of a substitute structure [e.g., Gulkan and Sozen, 1974;
Shibata and Sozen, 1976]. The substitute structure is a single degree of freedom
elastic system, the characteristics of which represent the inelastic system (Fig. 1).

Fig. 1. Characteristics of substitute structure.

December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 3

The secant stiffness Ks is that at the deformational limit state (LS) under consider-
ation, and may be used to evaluate an equivalent period Te . The hysteretic energy
absorption characteristics of the substitute structure may be accounted for by eval-
uating the level of viscous damping that results in the same response displacement
amplitude as that of the inelastic system. The concept of substitute structure there-
fore enables the use of an elastic displacement spectrum in design, while availing of
the displacement capacity of an inelastic system.
Various contributions were made towards the development of displacement-
based seismic design since the early work mentioned above. However, it was in
the 1990s that formal proposals were made to implement the emerging ideas into a
design procedure, the earliest of which is that by Moehle and his co-workers [e.g., Qi
and Moehle, 1991; Moehle, 1992]. A complete and workable procedure for seismic
design of structures that sets aside forces and relies entirely on displacement as the
primary design quantity is that proposed by Kowalsky et al. [1995] for single de-
gree of freedom systems (such as bridge piers). A concurrent paper on multi-degree
of freedom systems is due to Calvi and Kingsley [1995]. The steps comprising the
design process for SDOF systems, for simplicity, are given below:

(i) A target displacement for the structure is selected, based on the type of struc-
ture and the governing limit states.
(ii) Knowing the yield and ultimate (or some other limit state) displacement, and
the material and structural system as well as the characteristics of site and
expected earthquake, a value of equivalent damping is determined.
(iii) Displacement spectra representative of the seismo-tectonic environment are
consulted, the input to which are the target displacement and the equivalent
damping. The output is an effective period of vibration.
(iv) The structure is dimensioned to give an effective period, taking into account re-
duced stiffness consistent with the level of deformation, equal to that obtained
from the displacement spectra.
(v) If the effective period is not sufficiently close to the required period, return to
step (ii) above and repeat until convergence.

It is clear from the above that whereas displacement-based design is certainly the
logical framework for seismic design, since the primary source of seismic energy dis-
sipation is inelastic deformations, it imposes new requirements for verifiable design,
as described below. It is important, however, to note that the procedure outlined
above is by no means the only framework for the application of displacement-based
design. Alternatives exist [e.g. Fajfar, 1998], where a proposal is made to develop
inelastic capacity spectra for use in displacement-based design. The capacity spec-
trum approach, proposed by Freeman et al. [1975], comprises a plot of accelera-
tion spectral response versus its displacement counterpart. The load-deformation
curve of the structure under consideration, obtained from push-over analysis, is
then super-posed on the combined force-versus-displacement response spectrum,
December 1, 1998 10:23 WSPC/124-JEE 0023

4 J. J. Bommer & A. S. Elnashai

to establish the point of supply-demand balance. This approach was extended by

Fajfar [1998] by plotting inelastic force-versus-displacement spectra instead of the
elastic approach used by Freeman et al. [1975].

1.2. Requirements for displacement-based assessment and design

Primarily, accurate, representative and parametrically described displacement spec-
tra are essential ingredients that are hitherto unavailable. Indeed, conversion of code
acceleration spectra to displacement spectra immediately emphasises this point,
since the ensuing spectral shape is rather unrealistic in most cases, as shown in sub-
sequent sections of this work. The underlying reason is that acceleration spectra are
not as sensitive to digitisation errors as their displacement counterparts. Following
the derivation of a suite of acceleration spectra with a range of properties, without
due consideration for the equally important range of properties of the displacement
records, smoothed acceleration spectra are derived for code application. Whereas
near-field records of small magnitude events may have high ground accelerations,
rendering them eligible for inclusion in regression studies of acceleration attenua-
tion, they are characteristically weak from a displacements point of view. Hence,
a well-balanced catalogue for acceleration studies may be totally unbalanced for
displacement purposes. It is therefore not surprising that idealised, perfectly valid,
acceleration spectra give rise to completely unrealistic displacement spectra. The
final outcome is that whilst sketching an acceleration spectrum given the ground
acceleration and corner periods is straightforward, there is no analogous approach
for displacement spectra. The application of displacement-based design therefore
hinges on developing attenuation relationships specifically for displacements from
datasets selected for this purpose.
The second essential requirement for application of displacement-based seismic
design, within the framework outlined in detail in Sec. 1.1, is the derivation of
verifiable relationships between inelasticity and equivalent damping. This requires
investigating comprehensively the various characteristics of both structural sys-
tems and input motion. For the former, hysteretic response models with various
hardening-softening characteristics are required, whilst for the latter, ensembles of
strong motion with an acceptable spread in magnitude, distance and site condition
are required.

1.3. Scope of work

In this study, the displacement spectra obtained from code acceleration spectra are
briefly examined, with particular attention to the spectra from EC8. Following this,
displacement spectra obtained from frequency-dependent attenuation relationships
are explored and the European dataset is used to investigate the effects of obtaining
spectra using pseudo-spectral conversions.
The derivation of new attenuation equations for the prediction of displace-
ment spectra ordinates in Europe is approached in two stages. The first step is
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 5

an exploration of the period ranges within which the spectral ordinates may be
considered reliable. In the second stage, a reduced dataset of European strong mo-
tion records, individually processed to obtain the most reliable information possible,
is assembled and new regression analyses are run. The spectra obtained from the
new attenuation equations are evaluated and conclusions are drawn regarding the
best methods through which to obtain displacement spectra for design. Finally, sim-
plified spectral shapes in a linearised form are derived and shown to be completely
described by a limited number of parameters. The proposed shapes and values are
suited to codified seismic design of new structures and assessment of existing ones.
To complement the above linearised spectra, simple relationships relating in-
elastic energy dissipation through ductile response and equivalent damping, from
the work of Borzi et al. [1998] are reported. These relationships have been de-
rived using the same dataset alluded to above, after applying the same procedure
for individual record filtering and correction. This completes the requirement of
application of displacement-based seismic design and assessment.

1.4. Response periods of ductile structures

It is instructive to relate ranges of equivalent periods used to describe displacement
spectra to actual response periods of structures. As discussed above, the pertinent
periods in a displacement-based seismic design scenario are the secant periods at
the target limit state displacement of the substitute structure (Fig. 1), as opposed
to cracked or uncracked periods referred to in force-based design.
Since measurements of the periods of vibration of RC structures at maximum
ductility (close to failure) has not been undertaken, two sources of information in
this respect are utilised. Firstly, detailed analysis of multi-storey RC buildings has
been undertaken by Elnashai et al. [1996] and Mwafy [1998]. The structural ideali-
sations employed in the above studies included stress-strain characterisation of the
concrete and steel materials using the most advanced models available. The frames
were analysed at the design, twice the design and the collapse ground accelera-
tion, and a number of local and global limit states were monitored. In particular,
the collapse analysis sheds light on the effective period of vibration that may be
employed in a displacement-based approach, since the response of the structures
was fundamental mode-dominated. Of the three structural configurations consid-
ered (regular frame, irregular frame, core-frame), the irregular frame was selected
for the purposes of this paper, due to its longer response period. There is a total
of four structures with this configuration, designed for pairs of ductility class (DC)
and ground acceleration (PGA). The DC-PGA pairs considered are DCL-0.15g,
DCM-0.15g, DCM-0.3g and DCH-0.3g, where L, M and H refer to low, medium
and high respectively. The structure is eight storeys and four bays in the ground
floor, with eight bays above, hence the irregularity.
The calculated elastic fundamental periods of the four structures varied between
0.67 and 0.72 seconds, whilst the first harmonic was in the range of 0.2 seconds.
December 1, 1998 10:23 WSPC/124-JEE 0023

6 J. J. Bommer & A. S. Elnashai

From push-over analysis, the stiffness of the buildings at the point of first yield (in
columns) varied between 35 and 45 kN/mm, corresponding to a displacement of
between 130 and 150 mm. The push-over values for the secant stiffness at maxi-
mum displacement (drift limit) varied between 14 and 22 kN/mm (about 40% of
the secant stiffness at first yield). Fourier analysis of the acceleration response was
undertaken at various levels of input. Prior to yield, the structures had dominant
response periods of 1.1–1.2 seconds. At yield, these were 1.3–1.4 seconds, whilst at
collapse the fundamental periods were 1.5–2.0 seconds, with the shorter period ob-
served for DCL-0.15g and longer value for DCH-0.3g. Considering the yield periods
from inelastic dynamic response alongside the secant stiffness values at yield and
collapse from push-over analysis, the secant periods at collapse (not available from
analysis) are in the range 2.0–2.2 seconds (1.58 times yield period, as a consequence
of the secant stiffness at collapse being 40% of that at yield). This correlates well
with the measured periods at twice the design acceleration given above, noting that
for three structures the collapse acceleration was close to twice the design value.
Similar observations were made for the other two structures, with the core-
frame system exhibiting lower drift at collapse (triggered by local criteria) and
hence shorter periods. Therefore, a wide range of low to medium rise RC buildings
have effective periods (at maximum ductility) below 3.0 seconds. High rise RC
structures will not necessarily have higher displacement response. This is because
they will exhibit higher mode response, thus reducing the amplitude of maximum
displacement. Moreover, drift limits will govern the design; thus they are likely to
have wall systems that will also delimit their displacements.
The response of structures hit by the Northridge earthquake of 17 January 1995
was studied by Naeim [1997]. Below are short notes on some of the characteris-
tics and response of the buildings alongside analysis of their response periods, as
determined from discrete Fourier transforms of the acceleration response:

(i) A 10-storey RC wall structure located about 20 km from the source, and
founded on alluvium, was subjected to 0.34 g ground acceleration. The ob-
served response period was about 0.6 seconds. This is typical of stiff RC wall
structures.
(ii) A 6-storey steel moment frame with concrete caissons founded on alluvium and
located about 20 km from the source exhibited a dominant response period
of 1.4 seconds when subjected to a ground acceleration of 0.13 g. Had the
acceleration been higher, it is likely that this period would have been above
2.0 seconds.
(iii) A 20-storey RC structure (frame in one direction, walls in the orthogonal
direction) was subjected to a ground acceleration of 0.32 g, measured at 19 km
from the source. The fundamental period was 2.5 seconds whilst the second
mode period was 0.8 seconds.
(iv) A 7-storey RC moment frame structure founded on piles in alluvium and lo-
cated at 7 km from the source was subjected to a ground acceleration of
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 7

about 0.49 g. The fundamental period of the structure was 2.2 seconds, with a
higher amplitude at 1.4 seconds. This structure was heavily damaged with most
columns on the fourth floor failing in shear. The periods measured therefore
represent highly inelastic response, probably near the deformation capacity of
the structure.
From the above brief treatment, it is reasonable to conclude that the period
range for derivation of robust displacement spectra up to 3.0 seconds covers the
majority of cases for RC and steel buildings. It would also be sufficient for a wide
range of RC bridges. However, long span bridges and bridges with isolated decks
may require spectra extending to periods longer than 3.0 seconds.

2. Displacement Spectra from Seismic Design Codes

The determination of the shape and amplitude of the displacement design spectra
is the objective of this paper. It is known that the spectral ordinates for all damping

Fig. 2. Acceleration (top) and displacement (bottom) spectra. New Zealand Code [1992] for
highest hazard/ordinary importance.
December 1, 1998 10:23 WSPC/124-JEE 0023

8 J. J. Bommer & A. S. Elnashai

levels increase with period from zero to some maximum value and then descend to
converge at the value of the peak ground displacement (PGD) at long periods.
The most simple and straightforward solution is to convert the acceleration
spectra (SA) from seismic design codes using the pseudo-spectral relationship:
 2
T
PSD = SA (2.1)
2π
where T is the response period. The displacement spectra obtained in this way from
22 seismic codes from around the world [IAEE, 1992; Paz, 1994] have been examined
and in nearly all cases the PSD ordinates increase indefinitely with period, either
linearly, as in the case of the codes of Japan and New Zealand (Fig. 2), or even
parabolically, as in the case of the USA code (Fig. 3). It is clear therefore that
none of these spectra are suitable for use in displacement-based design without
modification.

Fig. 3. Acceleration (top) and displacement (bottom) spectra. UBC [1992] for highest hazard
and ordinary importance.
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 9

Fig. 4. Acceleration (top) and displacement (bottom) spectra. French Code [1990] for highest
hazard and ordinary importance.

The two exceptions to this trend of continuously increasing ordinates are the
codes from France (Fig. 4) and Romania (Fig. 5). The PSD spectra obtained by
conversion of the SA spectra in EC8 are similar in shape to those obtained from
the French code except that for all three site classes the plateau begins at a period
of 3.0 seconds.
Another limitation of current code spectra for direct use in displacement-based
design is the fact that spectra are required for a wide range of damping values.
Although some codes present spectra for three or four damping ratios, such as the
Indian and Portuguese codes, most present only one elastic spectrum which usually
corresponds, implicitly or explicitly, to a damping ratio of 5% of critical. Many
codes incorporate other damping values implicitly into the behaviour factors, thus
December 1, 1998 10:23 WSPC/124-JEE 0023

10 J. J. Bommer & A. S. Elnashai

Fig. 5. Acceleration (top) and displacement (bottom) spectra. Romanian Code [1991] for highest
hazard and ordinary importance.

making it impossible to obtain elastic spectra for damping values other than 5%.
The French and Spanish codes both include the following equation to obtain a
correction factor η for different damping ratios ξ:
 0.4
5
η= (2.2)
ξ
In the French code the maximum value of the damping ratio is limited to 30%. In
EC8 the correction factor is defined by the equation:
r
7
η= (2.3)
2+ξ
In EC8 the minimum permissible value of η is 0.7, which means that the largest
damping ratio that can be accommodated is 12.3%.
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 11

A final point worthy of mention on the EC8 spectra is that the code provides a
formula for the prediction of the PGD as a function of the effective peak acceleration
ag . This implies that for any damping ratio higher than 9%, the displacement
spectra actually has to rise from the long-period plateau to converge with the PGD,
which would not be expected.
It is clear from the foregoing that none of the current code spectra are suitable
for use in displacement-based seismic design. The PSD from the Romanian code,
which corresponds to the particular and special tectonics of that country, is only
defined for a damping ratio of 5%. The EC8 spectra are currently limited to too
narrow a range of damping ratios and do not converge at longer periods. The PSD
from the French code for different damping levels also remain parallel at longer
periods and do not converge to PGD, although these seem to be the most suitable
of the existing spectra. Therefore, there is clearly scope to explore more suitable
criteria to define the shape and amplitude of displacement response spectra for
analysis, design and assessment.

3. Displacement Spectra from Attenuation Relationships

The first frequency-dependent attenuation relationships for response spectral ordi-
nates were published by Johnson [1973] and a large number of equations have since
appeared in the technical literature. The majority of the available equations em-
ploy the pseudo-velocity response (PSV) as the predicted variable although there
are also a number of attenuation relationships for acceleration response (SA) ordi-
nates. To date there have been no equations published in the technical literature
for the estimation of relative displacement (SD) response spectral ordinates, but
since PSV is actually determined directly from SD, relationships for the prediction
of PSV could be considered suitable for the prediction of SD ordinates.
Displacement response spectra can be determined from predicted values of SA
through Eq. (2.1), although this is an approximation. In order to explore the relia-
bility of PSD ordinates obtained in this way, an experiment was performed to obtain
predictions of SD directly and from conversion of SA ordinates. The attenuation re-
lationships derived for the prediction of SA ordinates in Europe and adjacent areas
by Ambraseys et al. [1996] were used to construct PSD spectra for different com-
binations of magnitude, source-to-site distance and site conditions. The dataset of
accelerograms used to derive these relationships were then employed, exactly as pro-
cessed for the SA regressions, to generate 5% damped SD spectral ordinates. Then,
using the same attenuation model as for SA, regressions were performed directly
on SD and the new equations used to construct displacement response spectra for
the same combinations of magnitude, distance and site classification [Chlimintzas,
1997]. The comparisons (Fig. 6) show that the displacement spectra obtained in
both ways are very similar, only deviating at periods greater than 1.3 seconds and
at very short source-to-site distances; in these cases, the PSD ordinates calculated
by conversion of SA ordinates slightly overestimate the SD values.
December 1, 1998 10:23 WSPC/124-JEE 0023

12 J. J. Bommer & A. S. Elnashai

In view of this observation, it could be argued that existing attenuation relation-

ships for response spectral ordinates, whether PSV or SA, could be employed to pre-
dict displacement spectra to be used in design. The displacement spectra predicted
in this way by a wide range of attenuation relationships have been explored and

Fig. 6(a) Displacement spectra for Ms = 6 from acceleration spectra (solid) and direct displace-
ments (dashed).
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 13

Fig. 6(b) Displacement spectra for Ms = 7 from acceleration spectra (solid) and direct displace-
ments (dashed).

compared [Bommer et al., 1998]. Direct comparison amongst attenuation equations

is always difficult because of different definitions of the independent variables (mag-
nitude, distance and site classification) and different treatment of the two horizontal
components of each accelerogram [Ambraseys and Bommer, 1995]. Nonetheless, it
can be observed that the predicted spectral shapes are generally much more realistic
December 1, 1998 10:23 WSPC/124-JEE 0023

14 J. J. Bommer & A. S. Elnashai

than those obtained from the code spectra. However, the various equations exhibit
very different levels of influence of magnitude, distance and soil classification.
A limitation on the use of current attenuation relationships for spectral ordinates
to provide the input for displacement-based design is the fact that the majority of
the available equations only provide spectral ordinates for 5% damping. A notable
exception to this are the equations presented by Boore et al. [1993, 1994] which
predict spectral ordinates for damping ratios of 2, 5, 10 and 20%. However, these
equations only predict spectral ordinates at response periods up to 2.0 seconds.
Mohammadioun [1994] also reports regressions on ordinates of PSV for damping
levels of 0, 2, 5, 10 and 20% up to periods of 5.0 seconds, but the coefficients for the
equations, which are a function of magnitude and distance only, are not presented.
Hence, there is a need to explore the development of attenuation relations for SD
ordinates for a range of damping levels and also for as wide a period range as
possible.

4. Displacement Spectra of European Strong Motion Records

In order to derive attenuation relationships for the prediction of response spectra
for use in displacement-based design, it is necessary to compile a dataset of high-
quality accelerograms for which the associated source, path and site parameters
are uniformly and accurately determined. High quality accelerograms, in this con-
text, are those with a high signal-to-noise ratio, particularly at long periods, which
means that it would be preferable to employ recordings from digital accelerographs
[Tolis and Faccioli, 1999]. However, the number of available digital accelerograms is
relatively low and hence even though these data may provide more accurate values
for the spectral ordinates, it would be difficult to find correlations between these
ordinates and the parameters characterising the earthquake source, the travel path
and the recording site. The lack of digital accelerograms is particularly apparent
for Europe and adjacent regions, where the majority of available accelerograms are
from SMA-1 and other analogue instruments. However, source, path and site pa-
rameters have been uniformly determined for the European strong motion database
and presented by Ambraseys et al. [1996]. For this study, it was decided to make
use of this dataset, taking the independent parameters directly from the published
paper but re-examining the strong motion accelerograms in order to obtain the
most suitable processing.
The dataset presented by Ambraseys et al. [1996] consists of 422 triaxial
accelerograms generated by 157 shallow earthquakes with surface wave magnitude
Ms between 4.0 and 7.9. The source-to-distance used in that study is the short-
est horizontal distance to the surface projection of the fault rupture, although
whenever this value cannot be determined, mainly for smaller earthquakes, the
epicentral distance is used. The recording site geology is classified according the
average shear wave velocity Vs over the upper 30 m [Boore et al., 1993]. Sites for
which Vs is greater than 750 m/s are classified as rock (R) and sites for which Vs is
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 15

less than 360 m/s are classified as soft (S); intermediate sites are classified as stiff
soil (A).
Two additional accelerograms were incorporated into the dataset, the first of
which was the record from the temporary Gemona station of the Ms 6.1 Friuli
(Italy) earthquake of 15 September 1976. According to Rovelli et al. [1991] the
station was located on massive Mesozoic limestones, suggesting that the appropriate
classification would be rock (R). However, information from a borehole 100 m from
the recording site confirms that the correct classification would be stiff soil (A)
[S. Tolis, personal communication, 1997]. The source distance for this record has

Fig. 7. Magnitude-distance distribution of dataset w.r.t. to site classes R (top), A (middle) and
S (bottom).
December 1, 1998 10:23 WSPC/124-JEE 0023

16 J. J. Bommer & A. S. Elnashai

been estimated as 6 km. The second new accelerogram to be incorporated into the
dataset was the recording of the Ms 6.1 Aegion (Greece) earthquake of 15 June
1995 obtained at the Telecommunications Building in Aegion. This recording site,
at a distance of 10 km from the source, has been classified as soft soil (S) according
to Ambraseys et al. [1996].
In view of the fact that this study is particularly focused on the long-period
response spectrum and that small magnitude earthquakes do not produce signif-
icant long-period radiation, it was decided to impose a lower magnitude limit to
the dataset. The removal of weak (low amplitude) records from the dataset, in or-
der to obtain better signal-to-noise ratios, would not be acceptable since it would
introduce a bias into the data, but the removal of all the earthquakes with magni-
tude below the chosen lower limit of Ms 5.5 does partially achieve this objective.
The reduced dataset consisted of 189 accelerograms, but a further six accelero-
grams were eliminated since they were only available as filtered by other agencies,
and hence could not be included in the uniform processing of the records. The fi-
nal dataset thus consisted of 183 accelerograms from 43 shallow earthquakes. For
three of the recording stations, each of which contributed only one record, the
site classification is unknown. For the remaining 180 accelerograms the distribu-
tion amongst the three site classifications R:A:S as percentages is 25:51:24, which
compares favourably to the distribution of the original dataset of Ambraseys et
al. [1996] which is 26:54:20. The distribution of the dataset in magnitude-distance
space is shown in Fig. 7.
In the study by Ambraseys et al. [1996] all of the accelerograms were processed
using an elliptical filter with a lower cut-off frequency of 0.20 Hz. A number of the
stronger accelerograms from the reduced dataset were processed with a straight
baseline and also filtered with cut-offs at periods of 15, 10, 7 and 5 seconds (frequen-
cies of 0.067, 0.10, 0.143 and 0.20 Hz) and both the acceleration and displacement
spectra generated. As can be seen from Fig. 8, the processing usually has very little
effect on the acceleration spectra and then only at periods of about 4.0 seconds and
greater. Indeed, for SA ordinates up to 2.0 seconds, it would seem that filtering is
not actually necessary. However, the SD ordinates show a very high sensitivity to
the applied filter at long periods. In general, the SD spectra are almost identical
for periods up to 2 seconds, regardless of the processing, but beyond 3 seconds the
divergence can be very considerable. Since the long-period response ordinates are of
particular interest, it is clear that special attention must be given to the processing
of each record.
It is difficult to establish a single optimum filter for the digitised accelerograms
because the variation of signal-to-noise ratios is very large. An interesting example
are the recordings from Procisa Nuova of an aftershock of the Irpinia (southern
Italy) earthquake of November 1980: the ground shaking at the site was recorded
simultaneously on an analogue SMA-1 accelerograph and a digital DSA-1 accelero-
graph. The SMA-1 record was digitised both automatically and manually. The
DSA-1 recording was filtered with a cut-off at 5 seconds (up to 3 seconds the SD
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 17

ordinates are almost unchanged by the filter limits) and the corresponding spectrum
was assumed to be free of noise. It was found that an almost identical displacement
spectrum could be obtained from the automatically digitised SMA-1 record if it
was filtered with a cut-off at 2.4 seconds. For the manually digitised record it was
not possible to obtain a displacement spectrum that matched that of the DSA-1
record beyond a period of about 0.6 seconds. However, these recordings were gen-
erated by a very small earthquake with very weak long-period radiation: for the
stronger records in the dataset, digitised in a similar fashion, the absolute noise
level is probably similar but the signal-to-noise ratio would be considerably better.

Fig. 8(a) Acceleration and displacement spectra for the Tabas [1978] earthquake (5% damping)
with different filtering and correction.
December 1, 1998 10:23 WSPC/124-JEE 0023

18 J. J. Bommer & A. S. Elnashai

Fig. 8(b) Acceleration and displacement spectra for the Corinth [1981] earthquake (5% damping)
with different filtering and correction.

One possibility for identifying the optimum filter cut-off for an accelerogram is
from the Fourier spectra of the record and the fixed trace, but the latter is not
available for the majority of the records. The procedure adopted for this study was
simply to filter each record starting with a cut-off at 10 seconds and then inspect
the velocity and displacement time-histories found by double integration. The long-
period cut-off was then successively decreased until the velocity and displacement
time-histories appeared to be physically reasonable and further decreases in the
filter cut-off did not significantly enhance them. Some examples of the filtered time-
histories and their associated displacement response spectra are shown in Fig. 9.
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 19

Fig. 9(a) Filtered time-histories and displacement spectrum for record IT50L (dotted line shows
filter cutoff).
December 1, 1998 10:23 WSPC/124-JEE 0023

20 J. J. Bommer & A. S. Elnashai

Fig. 9(b) Filtered time-histories and displacement spectrum for record KAL1T (dotted line shows
filter cutoff).
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 21

Fig. 9(c) Filtered time-histories and displacement spectrum for record TABAL (dotted line shows
filter cutoff).
December 1, 1998 10:23 WSPC/124-JEE 0023

22 J. J. Bommer & A. S. Elnashai

5. European Attenuation Relationships for Displacement

Response Spectra
Regression analyses were performed on the horizontal displacement spectral ordi-
nates for damping ratios of 5, 10, 15, 20, 25 and 30% of critical. The regression
model used for SD ordinates (cm) was the same as that employed by Ambraseys
et al. [1996] for acceleration spectral ordinates:
log[SD(T, ξ)] = C1 + C2 Ms + C4 log(r) + CA SA + CS SS + σP (5.1)
where
q
r= d2 + h20 , (5.2)
d is the source-to-site distance in km, σ is the standard deviation, and P is a variable
that takes a value of 0 for mean values of SD and 1 for 84-percentile values. SA is a
dummy variable that takes a value of 1 for stiff soil (A) sites and 0 for rock (R) and
soft soil (S) sites; SS is defined in the same way for soft soil sites. The coefficients
C1 , C2 , C4 , CA , CS and h0 are determined by regression analysis, performed in
three stages [Ambraseys et al., 1996]. The term in C3 corresponding to inelastic
attenuation was constrained to zero in all cases, since, as in the case of acceleration
response spectra, it was found that the distribution of the dataset did not permit the
simultaneous determination of inelastic and geometric attenuation. In some cases
the value of C4 was slightly less than −1.0 which sensu strictu is not physically
admissible but is acceptable since this term is accounting for both mechanisms of
energy dissipation.
At each period the larger spectral ordinate from the two horizontal components
of each accelerogram was used as the dependent variable. Each component record
was only used for regressions up to a period equal to 0.1 second less than the
long-period cut-off employed for that record. As a result, for periods greater than
1.8 seconds there was a reduction in the number of data points available for each
regression, as illustrated in Fig. 10. At a response period of 3.0 seconds, the dataset
was reduced from 183 accelerograms to 121. It was decided not to perform the
regressions for periods longer than 3.0 seconds since the number of usable spectral
ordinates becomes insufficient.
The regression coefficients for six damping levels and for periods between 0.05
and 3.0 seconds are presented by Bommer et al. [1998]. The standard deviations
of the regressions are consistently lower than those found for SA ordinates by Am-
braseys et al. [1996], nearly always being less than 0.3 for log(SD). However, as a
result of the relatively small number of data points used to perform the regressions,
there are fluctuations in the predicted spectra. Smoother spectral shapes could be
obtained by smoothing the values of each regression coefficient with respect to pe-
riod using an appropriate technique. However, two important observations are made
regarding the predicted displacement response spectra, which enable a more elegant
presentation: the first observation is that the fluctuations in the 30% damped re-
sponse spectra are considerably lower and hence these can be used to define the
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 23

Fig. 10. Number of records available for regression for each soil type at different periods.

average shape of the spectra; secondly, the apparent amplification of the average
response spectra for lower damping levels, with respect to the 30% damped spectra,
appear to be approximately constant regardless of the magnitude, distance and site
classification. These observations have been used to derive a simplified approach
to constructing displacement response spectra for design, presented in the next
section.
Regression analysis was also performed on the larger values of peak ground
displacement (in cm) from each record, using the same attenuation model as in
Eq. (5.1), resulting in the following equation:

log(PGD) = −1.757 + 0.526Ms − 1.135 log(r) + 0.114SA + 0.217SS + 0.32P (5.3)

where r is defined as in Eq. (5.2) with a value of h0 of 3.5. Although it is not

possible to make direct comparisons because of the use of different definitions for
the parameters, Eq. (5.3) predicts values of PGD very similar to those presented
by Bolt [1996] for the near field, but attenuating more rapidly with distance.

6. Recommended Displacement Spectra for Design

From inspection of a large number of displacement response spectra for the six
specified damping levels, particularly those for which the applied filter cut-off was
greater than 5.0 seconds, it was concluded that a general, idealised format would be
as shown in Fig. 11. The spectrum for each damping level is composed of six straight
line segments defined by four control periods and their corresponding amplitudes.
December 1, 1998 10:23 WSPC/124-JEE 0023

24 J. J. Bommer & A. S. Elnashai

Fig. 11. Linearised displacement spectra for design.

The amplitude corresponding to TE is the peak ground displacement. For this

study, only that part of the spectrum up to periods of 3.0 seconds is considered for
reasons explained previously. The approach followed was first to obtain the control
co-ordinates for 30% damped spectra for a representative range of magnitude and
distance. These results are presented in Tables 1 and 2.

Table 1. Control periods for design spectra as func-

tion of magnitude and soil type.

TA TB
Ms Rock Stiff Soft Rock Stiff Soft
5.5 0.75 0.75 0.75 2.00 2.30 2.70
6.0 0.85 0.85 0.85 2.15 2.30 2.80
6.5 1.00 1.00 1.00 2.30 2.50 2.90
7.0 1.40 1.40 1.40 2.50 2.70 3.00∗
7.5 1.90 1.90 1.90 3.00 3.00∗ 3.00∗

∗ Actual control period probably slightly greater

than 3.0 seconds.

Table 2. Control ordinates for design spectra as

function of magnitude and soil type.

SDA SDB
Ms Rock Stiff Soft Rock Stiff Soft
5.5 2.2 3.1 4.0 3.1 4.6 5.8
6.0 3.7 5.0 6.5 5.8 8.4 11.0
6.5 6.7 8.7 11.2 10.8 15.9 20.8
7.0 14.8 20.1 25.0 20.3 28.7 38.6
7.5 34.1 46.7 55.0 37.0 55.8 70.0
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 25

Table 3. Distance factors Fd for spectral ordinates.

d (km) 0 5 10 15 20 30 40 50
Fd 1.00 0.621 0.352 0.245 0.187 0.127 0.095 0.075

Table 4. Damping ratio factors Fξ for spectral

ordinates.

ξ (%) 5 10 15 20 25 30
Fξ 1.90 1.55 1.35 1.20 1.10 1.00

Inspection of the predicted spectral ordinates shows that the shape of the spec-
tra is strongly influenced by magnitude and site classification, but far less so by
distance. It was found that the decrease of the spectral ordinates with distance is
reasonably constant across the period range and similar for all three site categories,
hence simple reduction factors could de found. The 30% damped spectra for dis-
tances up to 50 km from the source can be obtained by simply multiplying the
ordinates by the appropriate factor Fd taken from Table 3.
The next stage was to establish the amplification factors to be applied to the
control ordinates in order to obtain the displacement spectra for damping levels
from 5 to 25% of critical. These factors Fξ are presented in Table 4. Therefore,
using the values presented in Tables 1–4, and interpolating where necessary, it is a

Fig. 12. Derived vs. linearised displacement spectra (rock, Ms = 6, d = 15 km).

December 1, 1998 10:23 WSPC/124-JEE 0023

26 J. J. Bommer & A. S. Elnashai

Fig. 13. Derived vs. linearised displacement spectra (stiff soil Ms = 6, d = 15 km).

Fig. 14. Derived vs. linearised displacement spectra (soft soil Ms = 6, d = 15 km).
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 27

Fig. 15. Derived vs. linearised displacement spectra (rock Ms = 7, d = 15 km).

Fig. 16. Derived vs. linearised displacement spectra (stiff soil, Ms = 7, d = 15 km).
December 1, 1998 10:23 WSPC/124-JEE 0023

28 J. J. Bommer & A. S. Elnashai

Fig. 17. Derived vs. linearised displacement spectra (soft soil, Ms = 7, d = 15 km).

simple matter to construct design displacement spectra for rock, stiff soil and soft
soil sites for magnitudes between 5.5 and 7.5 and distances up to 50 km.
Figures 12–17 compare the spectra obtained from the attenuation relationships,
smoothed by successive passes of a 14 – 21 – 14 running average, with the design spectra
for the corresponding situations. In some cases, particularly at short distances, the
design spectra are conservative at long periods for the lower damping ratios, but
the approximation is generally very good.
One limitation of these design spectra is that they are defined only up to pe-
riods of 3.0 seconds, although as it has been shown this covers the majority of
applications. Certain assumptions can be made in order to extrapolate the spectra
to longer periods knowing that the ordinates for all six damping levels will con-
verge to the value of PGD predicted by Eq. (5.3) at long periods. Inspection of the
spectra from the records filtered at longer cut-offs suggests that the spectra will
generally converge, even for large magnitudes and soft soil, at periods no greater
than 8–9 seconds. For smaller magnitudes and stiffer sites it is reasonable to assume
convergence between 5 and 6 seconds.

7. Ductility-Damping Relationships and Residual Inelastic

Displacements
The dataset used above was employed in a study by Borzi et al. [1998], alongside
the same attenuation model described above. Two structural response models were
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 29

employed in evaluating constant ductility spectra for all records. These were an
elastic perfectly plastic (EPP) and a hysteretic model (HHS). The latter is based
on the model implemented by Lee [1998] for representation of shear behaviour of
RC bridge piers. Whereas convergence to a target ductility was always achieved
for the EPP model, this was not the case for the HHS representation. However,
the maximum percentage of divergent solutions was 3% for elastic perfectly plastic
behaviour, and dropped to 0.8% for hardening response (post-peak load stiffness of
10% of the yield stiffness). For softening response, lack of convergence was observed
in 5%–8% of cases for µ = 3, increasing to 12%–30% for µ = 4. All non-converging
points were eliminated from the regression analysis. Table 5 summarises the median
values of equivalent damping for various ductility levels for both models employed.
The equivalent damping values given in Table 5 are recommended for use with
the elastic response spectra of Sec. 5 of this paper, for periods of up to 3.0 seconds.
Borzi et al. [1998] utilised a procedure based on spectral intensities to evaluate
median damping ratios for longer period structures. The resulting values are on
the whole lower than those given above. If a more accurate period-dependent value
of equivalent damping is sought, the complete set of attenuation relationships are
given in Borzi et al. [1998]. Nil entries in Table 5 indicate that structures with
highly degrading response (K3 = −20% and −30%Ky ) would not have ductility
capacity of four or more.
The design procedure based on displacement results in a structure with stiffness,
strength and ductility characteristics that satisfy the requirements of the displace-
ment spectrum used. It is important to note that the solution is not unique, and
other structural systems with different response characteristics may also satisfy the
design premise. Due to the asymmetric nature of natural earthquake records, there
is a possibility that the structure will have a residual irrecoverable inelastic dis-
placement. This issue was studied by Borzi et al. [1998] for subsets of the strong
motion records mentioned above. It was observed that for magnitude six at a dis-
tance of 10 km on soft ground, the residual displacement of a degrading system
(K3 = −20%Ky ) is 18%, 32% and 43% of the maximum displacement for ductility
factors 2, 3 and 4, respectively. This observation may lead to the requirement to
increase the strength of the structure, to reduce the ductility demand hence the
residual displacement. For non-degrading systems, this issue is of significantly less

Table 5. Equivalent damping ratios for substitute struc-

ture [Borzi et al., 1998].

µ=2 µ=3 µ=4 µ=6

EPP 8% 13% 16% 19%
HSS, K3 =0 11% 15% 18% 20%
HSS, K3 = 10%Ky 10% 13% 15% 17%
HSS, K3 = −20%Ky 13% 20% 28% –
HSS, K3 = −30%Ky 15% 27% – –
December 1, 1998 10:23 WSPC/124-JEE 0023

30 J. J. Bommer & A. S. Elnashai

importance for the earthquake magnitude studied. Further comprehensive work is

needed using degrading systems before a final proposal is made.

8. Discussion and Conclusions

In this study, attenuation equations for displacement response spectra ordinates
in Europe have been derived. Although the attenuation relationships can be used
directly to construct displacement spectra for design, a simple parametric presen-
tation has been formulated. Taking just six values from Tables 1–4, displacement
design spectra can be constructed for all sites (except exceptionally soft soils) for
the range of magnitudes and distances of greatest engineering interest. These spec-
tra cover response periods up to 3 seconds and damping ratios between 5% and 30%
of critical, which covers the majority of engineering applications of displacement-
based seismic design and assessment. To complete the requirements of the design
procedure, recent work on deriving attenuation relationships for inelastic displace-
ment response and its relationship with the damped elastic response of substitute
structures is briefly reported. Finally, the issue of irrecoverable displacements of
ductile structures is mentioned, and approximate limits are given for the median
value of residual displacement as a percentage of maximum response displacement.
The simplified design spectra and ductility-damping relationships are recom-
mended for use in displacement-based design. The spectra can be extrapolated
beyond 3 seconds, using the attenuation relationship for PGD, with caution. As
more digital accelerograms in Europe become available, such as those from the
1997 Umbria-Marche earthquakes in Italy, it will be possible to extend the atten-
uation relationships, and hence the design spectra, to longer periods, applicable
mainly to special structures.

Acknowledgements
The authors are indebted to a number of people who have assisted with different
aspects of the development of this work. Thanks are due to Professor D. Papasta-
matiou of the National Technical University of Athens, for supplying the accelero-
gram from Aegion, and Professor E. Faccioli and Dr. S. Tolis of the Politecnico di
Milano, for supplying the Gemona accelerogram. Dr. Tolis also went to consider-
able length to determine the site characteristics for the Gemona record. The work
on residual displacements was motivated by earlier work carried out by Professor
K. Kawashima of Tokyo Institute of Technology and discussed with one of the au-
thors in April 1998.
At Imperial College, George Chlimintzas undertook the painstaking task of
processing the strong motion accelerograms and also in performing the regres-
sions together with Dohyung Lee. Dr. K. Simpson provided very considerable as-
sistance with accessing the data and performing the regression analyses. Petros
Konstantakos, Stephen Scott, Alejandro Martı́nez and Barbara Borzi assisted with
December 1, 1998 10:23 WSPC/124-JEE 0023

Displacement Spectra for Seismic Design 31

parts of the analysis and the preparation of a number of illustrations and spectral
plots. The authors would also like to acknowledge the important contribution made
by Professor N. N. Ambraseys in the form of re-evaluated source, path and site
parameters for the European strong motion dataset, and Dr. S.K. Sarma for use of
his regression program.
This work is supported financially by the European Union project Innovative
Concepts for the Seismic Design of New and Existing Structures (ICONS).

References
Ambraseys, N. N., Simpson, K. A. and Bommer, J. J. [1996] “Prediction of horizontal
response spectra in Europe,” Earthq. Engrg. Struct. Dyns. 25, 371–400.
Ambraseys, N. N. and Bommer, J. J. [1995] “Attenuation relations for use in Europe:
An overview,” in European Seismic Design Practice: Research and Applications, ed.
A. S. Elnashai (Balkema, Rotterdam), pp. 67–74.
Bolt, B. A. [1996] “From earthquake acceleration to seismic displacement,” The Fifth
Mallet-Milne Lecture (Wiley).
Bommer, J. J., Elnashai, A. S., Chlimintzas, G. O. and Lee, D. [1998] “Review and devel-
opment of response spectra for displacement-based design,” ESEE Research Report
No. 98-3, Imperial College, London.
Boore, D. M., Joyner, W. B. and Fumal, T. E. [1993] “Estimation of response spectra and
peak accelerations from western North American earthquakes: An interim report,”
US Geological Survey Open-File Report 93-509.
Boore, D. M., Joyner, W. B. and Fumal, T. E. [1994] “Estimation of response spectra and
peak accelerations from western North American earthquakes: An interim report,”
US Geological Survey Open-File Report 94-127.
Borzi, B., Elnashai, A. S., Faccioli, E., Calvi, G. M. and Bommer, J. J. [1998] “Inelastic
spectra and ductility-damping relationships for displacement-based seismic design”
ESEE Research Report No. 98-4, Joint Imperial College-Politecnico di Milano, Impe-
rial College, London.
Calvi, G. M. and Kingsley, G. R. [1995] “Displacement-based seismic design of multi-
degree-of-freedom bridge structures,” Earthq. Engrg. & Struct. Dyns. 24, 1247–1266.
Chlimintzas, G. O. [1997] “Selection of response spectra for displacement-based seismic
design,” M.Sc. Dissertation, Imperial College, London.
Elnashai, A. S., Salvitti, L. M. and Broderick, B. M. [1996] “Assessment of EC8 behaviour
factors for RC, steel and composite frames,” Eleventh World Conference on Earth-
quake Engineering, June 23–27, Acapulco, Paper No. 2050.
Fajfar, P. [1998] “Capacity spectrum method based on inelastic demand spectra,” IKPIR
Report EE-3/98, Institute of Structural Engineering, University of Ljubljana.
Freeman, S. A., Nicoletti, J. P. and Tyrell, J. V. [1975] “Evaluation of existing buildings for
seismic risk — A case study of Puget Sound Naval Shipyard, Bremerton, Washington,”
First U.S. Conference on Earthquake Engineering.
Gulkan, P. and Sozen, M. [1974] “Inelastic response of reinforced concrete structures to
earthquake motions,” ACI Journal 71, 604–610.
IAEE [1992] International Association for Earthquake Engineering, Earthquake Resistant
Regulations: A World List — 1992, Tokyo, Japan.
Johnson, R. A. [1973] “An earthquake spectrum prediction technique,” Bull. Seis. Soc.
Am. 63, 1255–1274.
 December 1, 1998 10:23 WSPC/124-JEE 0023

32 J. J. Bommer & A. S. Elnashai

Kowalski, M. J., Priestley, M. J. N. and MacRae, G. A. [1995] “Displacement-based de-

sign of RC bridge columns in seismic regions,” Earthq. Engrg. & Struct. Dyns. 24,
1623–1643.
Lee, D. [1998] “Seismic analysis of RC bridge piers using an axial load-sensitive hysteretic
model,” Ph.D. transfer report, Imperial College, London, UK.
Moehle, J. P. [1992] “Displacement-based design of RC structures subjected to earth-
quakes,” Earthq. Spectra 8(3), 403–428.
Mohammadioun, G. [1994] “Calculation of site-adapted reference spectra from the statis-
tical analysis of an extensive strong motion databank,” Tenth European Conference
on Earthquake Engineering, Vienna, 1, pp. 177–181.
Mwafy, A. M. [1998] “Seismic performance of RC buildings under multi-axial earthquake
loading,” Ph.D. transfer report, Imperial College, London, UK.
Naeim, F., [1997] “Instrumented buildings information system; January 17, 1994
Northridge, California, earthquake,” CD-ROM, John A. Martin and Associates, Los
Angeles (for CDMG-SMIP), USA.
Paz, M. (ed.) [1994] International Handbook of Earthquake Engineering (Chapman & Hall).
Qi, X. and Moehle, J. P. [1991] “Displacement design approach for reinforced concrete
structures subjected to earthquakes,” Report No. UCB/ERC-91/02, University of
California at Berkeley, January, 186 pp.
Rovelli, A., Cocco, M., Console, R., Alessandrini, B. and Mazza, S. [1991] “Ground motion
waveforms and source spectral scaling from close-distance accelerograms in a compres-
sional regime area (Friuli, Northeastern Italy),” Bull. Seis. Soc. Am. 81, 57–80.
Shibata, A. and Sozen, M. [1976] “Substitute-structure method for seismic design in RC,”
J. Struct. Div. ASCE 102, 1–18.
Tolis, S. and Faccioli, E. [1999] “Displacement design spectra,” J. Earthq. Engrg. 3(1),
107–125.

View publication stats