G. M. Calvi
Istituto Universitario di Studi Superiori (IUSS), Pavia
SUMMARY:
Risk assessment is affected by large uncertainties, depending on hazard, structure and damage analysis. Crucial
problems and choices may refer to: (i) hazard parameters, (ii) level of knowledge about materials, geometry,
detailing, (iii) assessed damage and failure modes (iv) resulting potential for step changes in performances.
The cost of attaining a high level of knowledge may significantly reduce the remaining resources, it is therefore
important to favour resilient solutions with a creative adoption of appropriate strengthening strategies.
In this framework, this paper discusses the possible criteria for the mitigation of seismic risk and some of the
alternative choices that may be adopted for strengthening, with reference to:
(a) the modification of damage and collapse modes strengthening individual elements or locally increasing
the deformation capacity;
(b) the insertion of additional systems resisting to horizontal actions;
(c) the introduction of base isolation, with the objective of capacity-protecting the existing structure;
(d) the reduction of displacement demand by added damping or introducing tuned mass systems.
Alternative strengthening choices lead to different protection levels and imply different performances, that are in
general represented by non linear or step functions. From these considerations, conceptual structure driven
strengthening criteria, based on a logical use of resources, are discussed.
The objective of this paper is to review the present criteria and effective tools to reduce seismic risk by
improving the structural response, i.e. by increasing constructions resilience, and shall thus face the
impossible problem of comparing needs and resources.
Resources are actually the absent guest in all tables where the issue is to define the safety level to be
met when designing and constructing new structures, and much more so when the issue is reducing
vulnerability in existing constructions.
Even today, several codes of practice are simply stating that existing constructions should attain the
same level of safety prescribed for new buildings, thus evidently affirming an impossible
philosophical concept, not applicable to reality. As often happens when common sense is lacking, this
actually results in relevant negative consequences, since those responsible for the safety of a structure
prefer not to know the risk level than being informed without having the resources to reach the
required performances and facing the problem of adopting less effective measures or dismissing the
use of the facility (that may be a hospital or a school).
Ignorance becomes a desirable state of mind.
Alternative, sounder approaches have been adopted in several countries, defining possible
prioritization schemes that somehow associate the target performances and time window to reach
them to some risk rating (e.g.: Grant et al. (2007)).
Any prioritization scheme obviously implies the possibility of correcting performance level and time
constraints to available resource, thus in principle giving an answer to the problem of comparing
resources and needs. However, it is still not considered that the cost to reach a given performance will
differ from case to case, even for the same risk rating, depending on possible strengthening measures.
In addition, since the correlation between cost of intervention and performance level is in general
represented by step functions, as discussed in the what follows, the obstinate pursuing of a determinate
level of performance may result in irrational cost benefit ratios.
Most approaches to the definition of design earthquakes and related performance levels are derived
from the scheme originally proposed for new constructions in the Vision 2000 document (SEAOC
(1995)). This document was first introducing the concept of varying the expected performance for
different return period earthquakes as a function of the relevance of the facility. The resulting matrix is
well known and not essential in this context, while the numerical definition of four intensity levels in
terms of their return period or probability of exceedence (frequent (50% in 30 years), occasional (50%
in 50 years), rare (10% in 50 years), and very rare (5% in 50 years) is still used as a possible
reference. Even more relevant is the attempt of numerically defining the expected performance as a
function of a single parameter, the inter-storey drift, as shown in table 1.
The concepts presented in the Vision 2000 document were elaborated and extended in the FEMA 356
(2000), but the next conceptual step was essentially due to the SAC project (Cornell (2002), where the
modelling of uncertainties was proposed), that inspired the development of a series of ATC documents
culminated with the ATC (2011), still in a draft form.
As pointed out, the inter-storey drift was identified as a potential fundamental parameter in a early
stage. However, in the first studies a clear distinction between the effects of structural and non
structural damage on performance was not discussed in detail. With reference to table 1 it appears that
for most buildings the first two performance levels refer to non structural problems, while the third
and fourth refer to structural problems. As such, the first two numerical values are likely to be more
reliable (or characterized by a lower dispersion) than the other two, since the prevention of brittle
failure modes and the structural design and detailing can affect very significantly these last two figure.
Note, too, that an increment of the level of knowledge (or a reduction of the epistemic uncertainties)
may have a much more relevant effect on them.
An in-depth study on the effects of different assessment parameters on the damage, and consequently
on the expected repair cost, of different construction elements is found in the ph. d. thesis of J.
Mitrani-Reiser (2007). In an extreme summary, it is there found that the structural damage for R.C.
frames can be correlated to a deformation damage index, which is essentially a function of inter-storey
drift demand versus capacity, while non structural damage can be related again to peak transient drift
(partitions, glazing), to peak floor diaphragm acceleration (acoustical ceilings, sprinklers), peak
ground acceleration (elevators), ratio of damaged area (paint).
The major relevance of displacement, drift and strain seems to be essentially confirmed.
1.4. Expected loss estimation
According to the Mitrani-Reiser approach, a loss estimate could be performed considering each
component-based fragility function and assuming a replacement cost equal to the total repair cost of
all damageable components considered.
A more complex approach has been proposed by Ramirez and Miranda (2009). Here, the component
inventory is still associated to damage fragility functions, but these are associated to a probability of
exceedence and to a probability of implying a repair cost level, according to a scheme developed at the
Pacific Earthquake Engineering Research Center (PEER, see Haselton et al. 2007).
It is interesting to note that the application of the two methods to the same sample reinforced concrete
frame building, designed confirming to the applicable codes, predict an expected annual loss
respectively equal to 0.55 and 0.9 % of the replacement cost.
A simplified approach based on direct displacement based assessment concepts, developed with the
aim of making it more applicable in the common practice, gave similar results (Sullivan and Calvi
(2011), Welch et al.(2012)).
It has to be underlined, however, that the main interest is here focused on buildings that do not
conform to applicable codes, where much larger uncertainties (and much larger annual expected loss)
have to be expected.
As anticipated, a commonly accepted framework to include uncertainties in risk assessment has been
derived from the SAC project and included in different forms in FEMA and ATC documents. The
basic formulation can be represented as a triple integral, as follows:
This formulation can appear quite difficult and inapplicable to practical problems, it is instead a
convenient formal expression of the need of combining four steps of analysis (hazard, structural,
damage and loss), each one characterized by epistemic and aleatoric uncertainties, as described in
figure 1.
Figure 1. PEER methodology to estimate losses in a probabilistic framework (Haselton et al. 2007)
2.2. Hazard reference parameters and analysis
The aim of hazard analysis is to estimate an intensity measure of an earthquake expected with a given
occurrence, i.e. with a given yearly probability of exceedence or average return period. The traditional
parameter to represent the seismic hazard has been the peak ground acceleration (PGA), possibly
associated with a spectral shape to immediately estimate a structure acceleration.
Once made clear that displacements are more relevant than accelerations, it may appear rational to
shift to a displacement response spectrum as key intensity parameter. Someone may state that there is
no difference, since a displacement spectrum can be derived from an acceleration spectrum as:
(T)=(T/2)2SA(T). This is a misleading observation, since peak displacement and corner period are
essentially affected by the long period part of the acceleration spectrum, where accelerations are low
and consequently little attention is paid.
The discussion becomes more complex when the problem to be addressed is how to define the seismic
intensity parameter based on seismic sources and their assessed potential to induce given magnitude
earthquakes with specific recurrence intervals.
The standard approach is a site specific probabilistic seismic hazard assessment (PSHA), based on
hazard disaggregation, i.e. on the separate consideration of all possible earthquakes that could occur
within a given region, their frequency of occurrence and the levels of ground motion they could
produce at a given site.
Often a PSHA is only carried out to calculate the PGA at a given return period and this is used to
anchor a spectral shape provided in a code. It is obvious that in this way the resulting spectrum is not
of uniform hazard, as the shape of a response spectrum in the code is fixed regardless of the return
period or location, instead of changing its shape with these two parameters, influenced by the
associated changing magnitude of the earthquakes which contribute most to the hazard.
Regardless of the procedure used to calculate PGA and spectral shape for a given return period at a
given site, structures will eventually be subjected to a single damaging earthquake event and thus,
considering the geographical variation of the ground shaking and the different response of each
building to its specific action, a variation of performance and damage has to be expected. However,
the magnitude and location of the next earthquake being unknown, it may seem rational to design all
structures to ground motions which should have the same probability of exceedence.
From a logical point of view, it appears that there are some potential flaws in the way of reasoning
briefly described above. Since low magnitude earthquakes, say, e.g., M5, which have perhaps a low
damage potential, are much more frequent than high magnitude ones, say M7, the uniform hazard
spectrum may be unrealistically dominated by small earthquakes. In general, close events are
dominating the hazard and a large magnitude earthquake is normally probabilistically relevant only as
a far distance event. However, it is unquestionable that in case of a M7 event it is likely that some
structure will be close to the epicenter. The case of a strong, close event seems in general to be of little
interest to a PSHA, unless extremely large return periods are considered. A structure is not likely to be
subjected to a large number of events during its life, thus responding with different performances to
each one of them. If a large magnitude earthquake should occur some structures close to the epicentre
are likely to be hit by actions much larger than those expected with a given return period and used for
design.
A possible (theoretical) solution to solve this problem has been proposed by Malhotra (2008), who
suggested to adopt an aggregate criterion, rather than a site specific one, thus imposing that the design
action should not be simultaneously exceeded over a given area more than a given number of times
per year.
Other, more radical, solutions are based on a combination of hazard and cost-benefit analyses to assign
directly the seismic strength to be provided to each building (Crowley et al. (2012)). This concept is
still at a development stage, but appears to be promising and presents some conceptual similarities to
the strengthening philosophy discussed later in this paper.
Since the main objective of this study is to propose alternative strengthening criteria, in the following
it will be assumed that the intensity parameters will be PGA and spectral displacement at 10 s period
(D10), their median will be obtained as a function of earthquake return period, and to each value will
be associated a spectral shape and a dispersion measure. The shape of the displacement spectra will be
assumed to be bilinear, being thus characterized by the value of the corner period only.
PGA (g) D10 (mm)
500
0.45
0.35 300
0.10
50
50y 475y 2475y TR 50y 475y 2475y TR
Figure 2. Example of schematic representation of the median PGA (left) and D10 (right) expected at a given site
as a function of the earthquake return period. Note that in general D10 varies more than the PGA.
0.00 0.00
0 1 2 3 4 0 1 2 3 4
Period (s) Period (s)
Figure 4. Demand assessment: samples of displacement spectra as a function of earthquake return period for
different values of equivalent damping (coherent with the current Italian code, note the different D10 values and
corresponding corner period).
The objective of damage analysis is to associate response parameters (e.g., as discussed above, inter
storey drift or floor acceleration) and expected damage. Detailed discussions on this issue can be find
in Mitrani-Reiser (2007) and Ramirez and Miranda (2009). Since the purpose of this paper is to
discuss strengthening choices and criteria, it will be simply assumed to separate non structural and
structural damage, to associate the first one to inter-storey drift, for a share of 80%, and to floor
accelerations, for a share of 20% and to associate structural damage to inter-storey drift alone,
including the assessment of elements brittle failures.
As shown in figure 5, the drift part of non structural damage will be assumed to start at 0.2% drift, to
reach 50 % at 0.5 % drift and to be complete (100%) at 1 % drift. The floor acceleration (Saf) part of
non structural damage will be assumed to start at a Saf of 0.4 g and to be complete at a Saf of 2 g.
The structural damage will be assumed to vary as shown in figure 5, up to 5% inter-storey if no failure
is assessed at lower drift values, at to reach 100% at any drift value at which a brittle failure is
expected.
These assumptions are essentially consistent with what available in the literature, it is however clear
that all of them can be easily modified without loosing generality and probabilistic evaluations can be
included. Note that a value of non structural damage equal to 1 does not necessarily imply the
complete collapse of all non structural elements, but rather that the cost of non structural repair will be
similar to that of a complete replacement (or that a complete replacement is necessary).
100% 100% 100%
2, 1 0.01, 1 0.05, 1
80% e 80% e 80%
e ga ga 0.02, 0.8
g
a
m m
a m
a
a D 60% D
D 60% f f 60%
f o o
o
e e
ga e
ga
g
at 40% t40% t 40%
n n 0.003, 0.4 n
e
cr
e
cr e
cr
NS Acceleration NS Drift Sensitive 0.015, 0.3 Structural Damage
e
P
e e 0.005, 0
Sensitive Damage P20% Damage P20%
20%
0.4, 0 0.002, 0.1
0.01, 0.1
0% 0% 0%
0 0.5 1 1.5 2 2.5 0 0.005 0.01 0.015 0.02 0 0.01 0.02 0.03 0.04 0.05 0.06
PeakFloorAcceleration(g) 0.001, 0 Inter-StoreyDriftRatio Inter-StoreyDriftRatio
Figure 5. Assumptions on floor accelerationassociated non structural damage (left, weight 20%), drift
associated non structural damage (centre, weight 80%), and driftassociated structural damage (right, note that
this plot can terminate at any drift value if a brittle failure is assessed).
In the framework of performance based assessment, the objective of loss analysis is to calculate the
probable repair cost for each level of damage state defined in the previous step. In its extreme
simplification, this could be done assuming that the cost of repair will be proportional to damage, for
example associating 75% of the value of the building to non structural content and 25 % to structures.
Again, this can modified and elaborated and probabilistic aspects can be included without loosing
generality.
In this framework the attention will be focused on cost of different strengthening measures and on
their comparison with a reduction in the expected repair costs, to obtain some cost-benefit evaluation
to drive the strengthening choices.
Similarly to what has been discussed with reference to assessment, strengthening interventions have
been traditionally designed verifying that the strength capacity of each structural element was larger
than the force resulting on that element applying the design earthquake.
The typical procedure was thus based on the following steps:
(1) define the design earthquake, often considering the same event used to design new buildings;
(2) analyse the structure, obtaining for each structural member demand and capacity in terms of
strength;
(3) design a strengthening intervention for each member for which the demand is larger than the
capacity;
(4) analyse again the structure, to check that now all members strengths are larger than the actions;
(5) if this is not the case go back to step (3).
It is evident that this may be a very ineffective procedure. Consider for example the case in which a
soft storey response is predicted at the ground floor, thus limiting the shear force to the ground floor
capacity. Strengthening all columns at the ground floor may induce a soft storey in the second storey,
that will have to be strengthened, and so on.
Clearly, there is no reason why a good designer should not conceive a more comprehensive
intervention based on the results of the assessment, after step (2), applying intuition and experience,
then simply showing that all members are now verified. This, however, cannot be considered a
procedure. Note as well that this procedure is oriented to structural response, and does not pay any
attention to the reduction of non structural damage, even if this is implicitly reduced increasing
structural strength. Since it is recognized that non structural damage is essentially related to drift
demand, a verification of performance will imply a check that a drift limit is met at all floors by the
strengthened structure, possibly requiring further strengthening measures and iteration checks.
At the present state of practice, the application of displacement based approach to seismic
strengthening will require as a first step the definition of a series of displacement spectra as shown in
figure 4, and the definition of some required performance in terms of accepted displacement (normally
derived from accepted drift values) for each earthquake return period of interest.
The building to be strengthened should have been preliminary assessed, and consequently a pushover
curve and the associated drift or displacement profiles will be known, as shown in figure 3.
At this stage, engineering judgement and ingenuity should be applied to define a strengthening
intervention that will then be assessed producing a new series of capacity curves.
Clearly this approach is more comprehensive and desirable, essentially because of the immediate
determination of potential damage and failure modes and the estimate of the global response and the
global effect of the strengthening choices. However, also in this case there are at least two aspects that
are not satisfying: (i) the objectives of the intervention are still based on conventional, code assigned
performances, that do not take into consideration any sort of cost/benefit analysis, and (ii) no guidance
is provided on the optimal combination of measures to modify the expected response.
4.1. Introduction
In this section a very brief overview of possible strengthening measures will be discussed.
The purpose is to highlight relative effects, with no attempt to depict a complete and thorough
presentation, impossible within the space constraints of a paper.
A second important issue is an attempt of roughly indicating the relative costs of each possible choice,
to have later a frame that will allow the conceptual discussion of a rational approach to seismic
strengthening. All costs indicated have to be regarded as a reasonable academic exercise, whose
purpose is simply to allow a discussion of the procedure.
It is also assumed that the average amount of linear elements in a reinforced concrete frame structure
is around 0.14 m per cubic meter of building (this derives from the assumption of 6 m spans beams in
both directions and 3 m tall columns: 6+6+3/663=0.14).
Clearly both values have to be considered as first estimate mean values, to which some measure of
dispersion could be associated to operate in a probabilistic environment.
If it is guessed or calculated the fraction of the total number of elements to be strengthened, it is thus
possible to estimate the total cost of the structural intervention:
Cost of Percentage of Cost of strengthening Element length per cubic Volume of building
intervention = elements to be per unit length [200 meter of building [0.14 [m3]
[] strengthened [%] /m] m/m3]
As shown in the photographs, heavy nonstructural interventions may have to be considered around
the elements to be strengthened. The related cost is assumed to be in the range of 100 /m3, weighted
by the per cent fraction of the elements to be strengthened.
Figure 8. Examples of strengthening existing masonry walls (left) or inserting new concrete walls (right).
To estimate the potential average cost of this kind of intervention, it will be assumed to start with a
very weak structure and to take all the loads on the new system. Assuming to need a total length of 6
m of shear walls (250/300 mm thick) for a building area of 100 m2 and that the total cost of concrete,
including reinforcement and scaffolding, will be 600 /m3, the total cost per cubic meter of building
will result to be around 10 /m3.
To this cost it may be necessary to add the sum required to strengthen the floor diaphragms, if needed,
that will be of the same order of magnitude (35 /m2, or 10 /m3) and the cost of removal and
replacement of non structural finishing material, that may be extensively required when the walls will
be added internally or when an extensive floor strengthening will be required. This cost may be in the
range of 100 /m2 or 30 /m3.
The structural cost of steel braced frames is similar to that of concrete walls.
Figure 9. Examples of introduction of an isolator: column cutting, temporary support, final configuration (cost
around 150 per square meter of building plan for the structural intervention, that could double to about 300 /
m2 considering the additionally required measures).
Figure 10. Building uplifting on new foundations and insertion of isolation devices (cost about 1,200 per
square meter of building plan).
The cost required for this type of strengthening measure may thus vary very significantly as a function
of the actual situation and obviously of the ingenuity of the solution.
Consider for example that cutting a column at the base, at the middle or at the top, will completely
change the bending moment diagram generated by the same shear force, and consequently induce the
need of strengthening the column itself or other parts of the structure. Consider as well that using a
single or double sliding surface device and positioning the concavity upward or downward may
change the elements on which the relative movement will induce a bending moment due to the
eccentricity of the vertical reaction.
In general, the global cost is proportional to the plan area of the building, rather than to its volume, and
may be as high as about 1,200 /m2 for the most complex cases, including uplifting.
A reasonable average estimate of cost per cubic meter of construction might be in the range of about
80 /m3, with a large coefficient of variation.
The design of this kind of strengthening should in general include the procedures for substituting a
device, to recentre the building after a strong event, possibly to test the response of the upgraded
system.
Figure 11. Examples of introduction of a damper, with additional piles to strengthen the foundation, in
combination with the creation of an isolation system.
As anticipated, the introduction of additional reaction systems, of any sort, may induce problems of
force transfer, typically in the horizontal diaphragm and at the foundation level. The consequent need
of additional intervention implies higher costs, in the structure and in the non structural elements.
Examples of floor strengthening and introduction of additional diaphragm systems are shown in figure
12. A mean cost estimate for intervention of this type is in the range of 35 /m2, which corresponds
roughly to a cost of 10 /m3 considering the building volume.
This obviously excludes the cost related to removal and replacement of non structural elements, that
could be roughly estimated in 100 /m2 or 30 /m3, when a heavy intervention on the floors is
required, which may be the case either when a new reaction wall system is inserted inside the building
or when the existing floors are in general not adequate to transmit the floor forces.
Figure 12. Examples of strengthening the diaphragm action capacity: application of carbon fibres on the lower
side of a floor (left), insertion of a steel braced truss (centre), concrete casting at the foundation level (right). The
cost of these kinds of intervention is around 35/m2, plus the replacement of non structural elements.
Foundation problems may arise when additional shear walls or large size damping devices are added.
This may be emphasized by the possibly low knowledge about the real capacity of the soil-foundation
system and by the general tendency of adopting large protection factors on the foundations.
While on one side this approach is today questioned, since it is felt that the soil-foundation system
may often provide a convenient post elastic response (e.g.: Gazetas (2012)), on another side this is a
typical situation in which the application of back analysis principles can prove to be quite effective,
particularly when the strengthening intervention comes after a known earthquake event.
This aspect is discussed further in the next section.
It is obvious that all the above indications have a conceptual nature and the separation between them is
artificial. In real practice it is common to combine different approaches to obtain a better intervention,
with greater benefits and smaller costs.
A first example to illustrate this statement is a strengthening project on a building in LAquila, Italy,
severely damaged, mainly in the two upper storeys, by the 2009 earthquake. It is a four storey
reinforced concrete frame building, with a plan area of about 4,500 m2 and a total volume of about
60,000 m3.
The repair and strengthening intervention included the demolition and reconstruction of the two upper
stories, the introduction of an isolation system at mid height of the lower, in ground, storey, the
addition of a damping system and the selected strengthening of a few weak columns (the photos in
figures 9 and 11 are related to this job).
The isolation devices adopted were double curvature friction pendulum ones, with radius of 6 m and
dynamic friction of 2.5 %. The additional damping system was constituted by eight viscous dampers,
with a constitutive law in terms of force (F, in kN) versus velocity (v, in m/s) of the type:
F = 1,900 v 0.1
The damping system was designed to adsorb a significant fraction (about 30 %) of the total base shear,
reducing the shear forces at the column foundations. The total maximum base shear was assessed as
about 7 % of the building weight. The fraction of the base shear taken by the damping system was
transmitted to new foundations, while the existing ones were verified applying back analysis
principles. The effects of isolation and damping system on force and displacement demand are shown
in figure 13: the total shear action is limited to less than 7.5 % of the weight of the building,
Again applying back analysis principles, it was decided that about 10 % of the existing beams and 20
% of the columns and 5 % of the walls had to be strengthened by carbon fibres wrapping, considering
the actual forces transmitted through the isolation system.
The total cost of the structural intervention is about 4 M, including the reconstruction of the upper
stories. The cost of isolation, member strengthening and additional damping is about 1 M (i.e. about
70 per cubic meter of building). The cost associated to non structural elements is irrelevant,
considering the damage level induced by the earthquake.
Figure 13. Effects of the isolation and damping system shown on a forcedisplacement diagram.
(a) (b)
(c) (d)
Figure 14. Base shear adsorbed during the 2009 earthquake ((a), estimated through back analysis, the grey lines
show a limit of 7.5 % of the building weight), base shear transmitted by the isolation devices (b), base shear
transmitted by the damping devices (c), total base shear after strengthening (d).
A second example is a seven storey reinforced concrete building, again in LAquila, completed in
1943 and considered to be relevant from an architectural point of view.
The building is L shaped, has an area of about 1,100 m2 per storey, with the exception of the last one,
whose area is 450 m2. The total volume is about 28,000 m3, two storeys are below the ground level.
The 2009 earthquake induced some significant damage, with the exclusion of the very stiff and strong
underground part, essentially limited to the non structural elements.
Considering the effect of the real earthquake, a possible strengthening strategy (suggested, but not
adopted by the owner) included the insertion of steel bracings endowed with dampers and the addition
of two tuned masses, each one located at the extremities of the L, constituted by a concrete slab resting
on low damping rubber bearings.
The total mass of the building was about 15,000 t, but only about one third of it was associated to the
first mode of vibration, that was essentially related to the part of the building elevated out of the
ground.
Considering the insertion of the damped bracings, the participating mass was estimated as 5,400 t,
with a period of vibration of about 0.4 s and a total damping of 6.5 % (1.5 % due to the dampers,
considering their contribution to the total shear strength, see Priestley et al., 2007).
Applying the concepts discussed in section 4.4.2 the added tuned mass was designed to be of
approximately 500 t, with a period of vibration TTM = 0.42 s and consequently a resulting stiffness
KTMeq = 112 MN/m.
The benefit obtained in terms of reduction of interstorey drift prediction for a 475 y return period
event is shown in figure 13 and ranges between 30 and 65 % . The obtained drift values are clearly
compatible with an essentially elastic response.
The total cost estimated for the structural intervention is about 2 M (i.e. about 70 per cubic meter of
building), while the cost associated to non structural elements, including the required after earthquake
repair, was estimated as about 3 M (i.e. about 100 per cubic meter of building, excluding
installations).
Figure 14. Predicted interstorey drift for a 475 year return period event with (black line) and without (grey line)
the addition of a tuned mass
5.1. Introduction
All the concepts, comments and data previously addressed will be now combined to try to develop a
rational approach to guide strengthening, making constant reference to a sample real building
structure, a six storey reinforced concrete residential building damaged by the LAquila earthquake
(Calvi, 2009) and later repaired and strengthened.
Though all the analysis have been performed in both directions and attempts of combination to assess
a 3D response have been made, here only one direction will be considered, for the sake of simplicity.
The site seismicity has been assessed and a complete adaptive pushover analysis has been run.
The spectral acceleration and displacement demand are reported in table 2 for different performances,
indicating for each of them the corresponding return period (RP) event (as pointed out, it is more
sensible to perform these predictions for different performances, rather than to predict the performance
for given, discrete return period, events). The assessed pushover response and the storey drift demands
for the same events are shown in figure 15, where for each performance it is also indicated the value of
equivalent displacement demand and viscous damping.
It may be noted that a response approximately corresponding to a fully operational performance (FO),
with maximum drifts of the order of 0.2 %, corresponds to a very short RP of 20 years RP; a damage
controlled (DC) performance, with drifts lower than 0.5 %, is well approximated by an event with 72
years RP, while the life safety (LS) and near collapse (NC) performances would be attained for RP
approximately corresponding to 975 and 2475 years, if a brittle failure should not have been assessed
for an RP of about 275 years, at an equivalent displacement of approximately 0.15 m.
Clearly, a basic strengthening intervention on the building could simply consider the possibility of
eliminating the potential for brittle failures, to attain the complete pushover curve shown in the figure.
Table 2. Spectral acceleration (Sa) and spectral displacement (Sd), calculated using secant stiffness
(corresponding period Te) and appropriately estimated damping for different return period (RP) events for the
sample building and the local seismicity in LAquila.
Performance FO DC LS NC
RP (y) 20 72 975 2475
Te (s) 1.55 1.55 2.35 2.86
eq 5% 5% 12% 18%
Sa(Teff, eq) (g) 0.041 0.093 0.138 0.132
Sd(Teff, eq) (m) 0.024 0.054 0.189 0.269
0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.005 0.01 0.015 0.02 0.025
Displacement (m) Inter-storey Drift Ratio
Figure 15. Capacity assessment: pushover curve (left) and associated storey drifts (right).
For each RP, drift and floor acceleration have been computed, considering each floor, and the
corresponding damage has been calculated with reference to figure 5. The damage value has been
attributed 25 % to the structure and 75 % to nonstructural elements; the non structural part has been
considered 80 % drift sensitive, 20 % acceleration sensitive, each storey has been assumed of equal
value.
The results are reported in table 3, as percentage of total replacement cost, assumed as 400 /m3 of
building. Conventionally, the table shows a 100 % damage assessment for all cases after the
assessment of a brittle failure.
The estimate of the expected annual loss has been performed approximating the curve that expresses
the annual probability of exceedence considering a multilinear functions, with straight segments
between the assessed performances (see figure 16, Sullivan et al. (2011), Welch et al. (2012)). A
zero loss point has been arbitrarily assumed at an annual probability equal to 10 % (this assumption
can have significant effects on the loss values). As anticipated, in this specific case a total loss is
assumed at a RP equal to 275 years, for which a brittle failure is expected.
Table 3. Assessed damage cost, expressed as a percentage of the cost of replacement of the building. The
predicted brittle failure before reaching a life safe performance is strongly conditioning the results
Limit State DMstruct (%RC) DMNS-Drift (%RC) DMstruct (%RC) LOSStot (%RC)
FO 0.00% 6.67% 0.00% 4.0%
DC 0.00% 41.21% 0.00% 24.7%
LS 100.0% 100.0% 100.0% 100.0%
NC 100.0% 100.0% 100.0% 100.0%
*Dashed line indicates that brittle collapse is expected before reaching the life safety limit state
Figure 16. Loss assessment: example of annual probability of exceedence versus estimated direct loss; the
expected annual loss can be estimated calculating the grey area.
0 0%
0.00 0.20 0.40 0.60 0 1000 2000 3000 4000 5000
Displacement (m) Return Period (years)
Figure 17. Modified pushover curve (left) and estimate of direct loss versus return period (right) for each
strengthening strategy considered.
Table 4. Assessed damage cost, expressed as a percentage of the cost of replacement of the building
Performance Interruption Time (days) Lossindirect (% of RC)
Fully Operational 1 week 7 1.2 %
Damage Control 4 weeks 30 5.0%
Life Safety 6 months 180 30.0%
Near Collapse 1.5 years 540 90.0%
Strengthening Strategy
LIMIT *
STATE Existing Element FPS Shear Added
(do-nothing ) (50%) Isolation** Walls Damping
RP (y) 20 20 20 30 30
Fully
LossDirect | O (%) 4.00% 4.00% 4.00% 2.90% 4.00%
Operational
LossIndirect | O (%) 1.17% 1.17% 1.17% 1.17% 1.17%
RP (y) 72 72 72 140 200
Damage
LossDirect | DC (%) 28.27% 28.27% 28.27% 24.73% 28.27%
Control
LossIndirect | DC (%) 5.00% 5.00% 5.00% 5.00% 5.00%
RP (y) 273 975 3400 2000 3200
Life
LossDirect | LS (%) 100% 66.50% 39.8% 62.53% 66.50%
Safety
LossIndirect | LS (%) 90.00% 30.00% 5.00% 30.00% 30.00%
RP (y) 2475 2475 3400 4400 4300
Near
LossDirect | NC (%) 100% 81.38% 39.8% 81.17% 81.38%
Collapse
LossIndirect | NC (%) 90.00% 90.00% 5.00% 90.00% 90.00%
Expected EALDirect (%) 1.70% 1.37% 1.20% 0.79% 0.84%
Annual Loss EALDirect+Indirect (%) 2.66% 1.81% 1.41% 1.06% 1.07%
*Brittle collapse is expected at a 273 year return period for the existing building and near collapse downtime is
assumed
**The FPS Isolation case assumes only damage control level of downtime beyond the stick-slip activation of
bearings
Expected Annual Loss
3.0%
EAL- Direct Loss
2.5% EAL Direct + Indirect
)t 2.0%
n
e
m
e
c
al 1.5%
p
e
r
f
o
%
( 1.0%
L
A
E % % % % %
6 1 1 6 7
6
. .8 .4 0
. .0
0.5% 2 1 1 1 1
% % % % %
0 7 0 9 4
7
. 3
. 2
. 7
. .8
1 1 1 0 0
0.0%
Existing
1 Elem-50%
2 FPS3Iso Walls
4 Dampers
5
It may be noted that in this exercise the break even point results to be in the range of 15 to 20 years if
indirect loss is taken into account.
Table 6. Benefit/cost ratios and break even times for each one of the strengthening strategies
Strengthening Strategy
Existing Element FPS Shear Added
(do-nothing ) (50%) Isolation Walls Damping
Cost (%RC) - 16.0% 22.0% 24.0% 22.2%
EALDirect (%RC) 1.70% 1.37% 1.20% 0.79% 0.84%
NPVEAL (%RC) 85.24% 68.55% 59.81% 39.32% 41.87%
Direct
Losses Benefit (%RC) - 16.69% 25.43% 45.92% 43.37%
Benefit-Cost Ratio - 1.043 1.157 1.913 1.955
tbreak even (y) - 47.9 43.2 26.1 25.6
EALDirect+Indirect (%RC) 2.66% 1.81% 1.41% 1.06% 1.07%
Direct + NPVEAL (%RC) 133.04% 90.52% 70.31% 52.95% 53.57%
Indirect Benefit (%RC) - 42.52% 62.73% 80.08% 79.46%
Losses Benefit-Cost Ratio - 2.658 2.853 3.336 3.582
tbreak even (y) - 18.8 17.5 15.0 14.0
Benefit-Cost Analysis for Considered Retrofit Options Break-Even Analysis for Considered Retrofit Options
4.0 60.0
B/C Direct Losses Direct Losses
3.5 B/C Direct + Indirect Direct + Indirect
50.0
3.0 s)r
io a
e
at y( 40.0
R 2.5 n
ts e
o vE
C
ti 2.0 ka 30.0
f e
e r
n B
e r
B 1.5
6 fo
5 4 8 e 20.0
.6 .8 .3 5
. m
2 2 3 3 iT
1.0
10.0 .9 2
. 1
0.5 7 3 . .6
1 5 4 4 5 6 0 5 0
4 6 .8 . 2 . .
0
. 1
. .9 9
. 8 7 5 2 4
1 1 1 1 1 1 1 1
0.0 0.0
Elem-50%
1 FPS2- Iso Walls
3 Dampers
4 Elem-50%
1 FPS -2Iso Walls
3 Dampers
4
Figure 19. Histogram showing the summary of benefit/cost ratios and break even times for each one of the
strengthening strategies
It is interesting to observe that the expected annual loss (EAL) should be appropriately used as the
base to calculate insurance premia, considering or not the indirect loss part depending on whether
indirect losses are or not covered in the contract.
As such, the reduction in EAL could be immediately transformed into an annual economic return
wherever an insurance contract is actually issued.
Note, as well, that the EAL could be appropriately used as a global evaluation parameter of the
seismic quality or, maybe better, of the seismic resilience of a building, much more than any
single protection factor towards collapse or a specific conventional performance.
If used in this context, it appears more reasonable to adopt the EAL referred to direct loss only, since
the downtime value may easily change whenever a building is used for a different purpose.
It may also be consider to impose by code that specific strategic functions (also in economical terms)
should only be possible in buildings that met a minimum threshold of seismic resilience.
Similarly to what is done to evaluate the energy and environmental performance of a building, discrete
classes of earthquake resilience might be defined, for example saying that a building has a seismic
resilience A (or SR-A) if the EALdirect is lower than, say, 0.5 %, a SP-B if EALdirect < 1.0 %, a SP-C if
EALdirect < 2.0 %, etc.
If the expected seismic resilience will be certified, insurance companies should have no reason not to
accept it as a base to define the annual premium, the value of a building may increase as a
consequence of a strengthening intervention, unsafe buildings will soon be difficult (or impossible) to
be sold and, in general, a positive spiral may be favoured.
Finally, governments may consider to subsidize strengthening intervention on sounder basis,
considering that there will be an actual and measurable benefit for the society if the damage in case of
earthquake should be reduced.
6. CONCLUSIONS
In this paper a quick state of the art/practice in seismic assessment is used as a sort of introduction to a
discussion on criteria and choices to actually reduce risk through seismic strengthening. A number of
conclusions are drawn throughout the paper and summarized here.
(a) The target performances to be met in strengthening are normally based on academic decisions and
conventional values, without considering that available resources and the capacity of using them
are the only reasonable parameter to scale relative evaluations.
(b) It is common to assess the expected performances for conventional return period earthquakes,
rather than to assess the expected return period event that will induce a given performance. This
second option has to be preferred for several reason, including the smooth nature of a hazard
curve versus the step nature of a performance function. While the definition of a given return
period or a given yearly probability of exceedence is totally conventional, and consequently
irrelevant, the association of a performance to physical events is possible and desirable.
(c) This concept applies to the choices related to a strengthening intervention to a much larger extent,
since the economical resources required to reach a given performance are normally changing with
finite steps.
(d) Studies to improve the level of knowledge about materials and geometry are often wasted to
assess useless parameters, while should be associated with progressive analysis and choices
oriented to the most efficient strengthening intervention.
(e) The potential of back analysis concepts in the evaluation of dependable levels of members and
structures capacity is often underestimated and neglected.
(f) Damage and loss analysis shows that performances are essentially related to drift and strain
demand and that the major part of the economical losses are associated to non structural elements
and to relatively low displacement demands.
(g) A number of strengthening approaches are discussed, showing that it is possible to associate a
cost to each one of them at a preliminary design stage, improving the reliability of the estimate
with the progress of design and analysis.
(h) A simple conceptual way of calculating indirect losses for residential buildings is presented and
discussed. This approach could be easily applied to loss estimation and possibly extended to other
kinds of use.
(i) It is shown that it is possible to associate a loss analysis to different strengthening strategies, in
turn associated to a cost of intervention. This allows the definition of a costbenefit parameter for
each strategy, possibly expressed in terms of time to breakeven.
(j) It is shown that for any designed, existing or strengthened building it is possible to calculate the
expected annual loss, a parameter that could be used to grade the seismic global performance or
seismic resilience of the building. This parameter may refer to direct loss only or include
indirect losses and could be used for several purposes, including the evaluation of insurance
premia.
(k) All the evaluations and numerical exercises presented in this paper have a conceptual nature,
should be refined and weighed in a more proper and extended way, uncertainties related to
epistemic and aleatory knowledge should be considered, in a general probabilistic framework.
AKCNOWLEDGEMENTS
The author wishes to recognize that without thorough discussions with Nigel Priestley and Tim Sullivan,
practical design experiences with Matteo Moratti and intelligent work and support by David Welch this paper
would not have been written.
REFERENCES
ATC (2011). Guidelines for seismic performance assessment of buildings, (uncompleted draft). Applied
Technology Council, Redwood City, CA.
Calvi, G. M. (1999). A displacement-based approach for vulnerability evaluation of classes of buildings. Journal
of Earthquake Engineering. 3:4. 411-438.
Calvi, G.M. Editor (2009). LAquila, April 6th 2009, 3.32 am. Progettazione Sismica. 03:03.
Calvi, G.M., Pietra D. and e Moratti M. (2010). Criteri per la progettazione di dispositivi di isolamento a pendolo
scorrevole. Progettazione Sismica. 04:03. 7-30
Cornell, C.A., Jaylayer, F., Hamburger, R.O., Foutch, D.A. (2002). Probabilistic basis for 2000 SAC Federal
Emergency Management Agency Steel Moment Frame Guidelines. Journal of Structural Engineering.
128:4. 526-553.
Crowley, H, Silva, V., Bal, I.E. and Pinho, R. (2012). Calibration of seismic design codes using loss estimation.
15 WCEE, Lisbon, Portugal.
FEMA 356 (2000). Prestandard and commentary for the seismic rehabilitation of buildings. Federal Emergency
Management Agency, Washington, D.C.
Gazetas, G. (2012). Nonlinear soil-foundation-structure interaction. 2nd Int. Conf. on performance-based design
in earthquake geotechnical engineering. Taormina, Italy.
Grant, D.N., Bommer, J.J., Pinho, R., Calvi, G.M., Goretti, A. and Meroni, M. (2007). A prioritization scheme
for seismic intervention in school buildings in Italy. Earthquake Spectra. 23:2. 291-314.
Haselton, C.B., Goulet, C. A., Beck, J. L., Deierlein, G. G., Porter, K.A., Stewart, J. P., Taciroglu, E. (2007). An
assessment to benchmark the seismic performance of a code-conforming reinforced concrete moment-frame
building. Pacific Earthquake Engineering Research Center (PEER), Berkeley, CA.
Malhotra, P.K. (2008). Seismic design loads from site-specific and aggregate hazard analyses. Bulletin of the
Seismological Society of America. 98:4. 1849-1862.
Mitrani-Reiser, J. (2007). An ounce of prevention: probabilistic loss estimation for performance-based
earthquake engineering. PhD Dissertation, California Institute of Technology, Pasadena, CA.
Priestley, M. J. N., Calvi, G. M., Kowalsky, M. J. (2007). Displacement-based seismic design of structures.
IUSS Press, Pavia, Italy.
Priestley, M.J.N. (1997). Displacement-based seismic assessment of reinforced concrete buildings. Journal of
Earthquake Engineering. 1:1. 201-241.
Priestley M.J.N. and Calvi, G.M. (1991). Towards a capacity design assessment procedure for reinforced
concrete frames. Earthquake Spectra. 7:3. 413-437.
Ramirez, C. M. and Miranda, E. (2009). Building specific loss estimation methods and tools for simplified
performance-based earthquake engineering. Report 171. John A. Blume Center, Stanford University, CA.
Sadek, F., Mohraz, B., Taylor, A. W. and Chung, R. M. (1997). Method of estimating the parameters of tuned
mass dampers for seismic applications. Earthquake Engineering and Structural Dynamics. 26. 617-635.
SEAOC (1995). Vision 2000, A framework for performance-based engineering. Structural Engineers
Association of California, Sacramento, CA.
Sullivan, T.J., Calvi, G.M. (2011). Considerations for the seismic assessment of buildings using the direct
displacement-based assessment approach. 2011 ANIDIS Conference, Bari, Italy.
Sullivan, T.J., Calvi, P.M. and Nascimbene R. (2012). Towards improved floor spectra estimates for seismic
design. Earthquake and Structures. Accepted for publication.
Welch, D.P., Sullivan, T.J and Calvi G.M. (2012). Towards a direct displacement-based loss assessment
methodology for RC frame buildings. 15 WCEE, Lisbon, Portugal.