Base isolation for retrofitting historic buildings: Evaluation of seismic performance through...

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICSEarthquake Engng Struct. Dyn. 2001; 30:1125–1145 (DOI: 10.1002/eqe.54)

Base isolation for retro+tting historic buildings: Evaluationof seismic performance through experimental investigation

Antonello De Luca1;∗;†, Elena Mele1, Javier Molina2, Guido Verzeletti2and Artur V. Pinto2

1Dipartimento di Analisi e Progettazione Strutturale; Universit�a di Napoli ‘Federico II’; P.le Tecchio 80;I-80125 Napoli; Italy

2ELSA Laboratory; ISIS; Joint Research Centre; Ispra (VA); Italy

SUMMARY

An experimental test program on a full-scale model representing a sub-assemblage of the cloister facadeof the Sao Vicente de Fora monastery, retro+tted through base isolation, has been recently carriedout at the European Laboratory for Structural Assessment of the Joint Research Centre of the EuropeanCommission. In this paper an overview of the laboratory model and the experimental results is provided.In particular, +rstly the test model is described, including the geometry and mechanical properties of themasonry specimen and the design of the isolation devices; then the testing method and the sub-structuringof the isolation system are described and the seismic inputs adopted for the pseudo-dynamic tests arede+ned. Finally, the experimental results are discussed and compared to the analogous results obtainedon the ‘as is’, +xed-base sub-assemblage model. The implications of the test outcomes are emphasizedand developments of this research line are presented. Copyright ? 2001 John Wiley & Sons, Ltd.

KEY WORDS: historic buildings; seismic retro+t; base isolation system

1. INTRODUCTION

The very recent earthquakes which have occurred in Italy have highlighted the particular vul-nerability of historic masonry buildings and the need of properly de+ning retro+tting measures.

Due to the inherent value of Cultural Heritage buildings, the de+nition of retro+tting inter-ventions for improving the seismic capacity of the structure is a point of major importanceand involves the research for an optimal balance between two opposite requirements, i.e. ar-chitectural preservation and the structural safety. With this regard, the base isolation system

∗ Correspondence to: Antonello De Luca, Dipartimento di Analisi e Progettazione Strutturale, UniversitEa di Napoli‘Federico II’, P.le Tecchio 80, I-80125 Napoli, Italy

† E-mail: [email protected]

Contract=grant sponsor: MURST PRIN ’97, COSISMO Project; Contract=grant number: CNR grant No. 98.3212.PF54

Received 17 July 2000Revised 10 October 2000

Copyright ? 2001 John Wiley & Sons, Ltd. Accepted 25 October 2000

1126 A. DE LUCA ET AL.

(BIS) has been recently suggested as an innovative retro+tting strategy, and has been adoptedfor the seismic upgrading of some major monumental buildings in the U.S.A., due to thepossibility of eLciently improving the seismic capacity of a building with minimal disruptionto its architectural features.

2. FRAMEWORK OF THE RESEARCH AND OBJECT OF THE PAPER

2.1. Progetto Finalizzato Beni Culturali

With the idea of checking the application of BIS to the retro+tting of Italian and, more bygeneral, European monumental heritage, a research program, developed in Italy by the +rst twoauthors in the context of the ‘Progetto Finalizzato Beni Culturali’ sponsored by the ItalianNational Research Council (CNR), is currently in progress with the following major tasks:

• study of the U.S. monument retro+tting projects through BIS and de+nition of the speci+cdesign issues and construction aspects involved in the application of BIS for monumentretro+tting [1; 2];

• selection of a representative historic building typology, the basilica plan churches, and ofsome speci+c case studies [3; 4];

• evaluating the possibilities of applying BIS to the seismic retro+tting of the selected casestudies [5] and numerical analyses for the assessment of performance improvements dueto the introduction of BIS [6].

2.2. COSISMO project

Furthermore, an European research program, the COSISMO project, has been recently carriedout by the Joint Research Centre (JRC) of the European Commission in collaboration withthe Portuguese General-Directorate for National Buildings and Monuments (DGEMN) and theNational Laboratory of Civil Engineering (LNEC), in Lisbon, through the following majorphases:

• dynamic characterization through numerical analysis and in situ tests of a representativemonumental building, the S. Vicente de Fora monastery in Lisbon [7];

• laboratory seismic testing of a full-scale model representing a sub-assemblage of thecloister facade of the S. Vicente de Fora monastery [8];

• linear and non-linear analyses of the building though numerical models calibrated on thebasis of the experimental data [9];

• assessment of the seismic vulnerability of the building and evaluation of retro+tting so-lutions and techniques.

2.3. S. Vicente model with base isolation

Starting from the above research experiences, an experimental test program on the S. Vicentede Fora sub-assemblage model, retro+tted through BIS, has been recently planned and carriedout at the European Laboratory for Structural Assessment (ELSA) Laboratory of the JRC.

Copyright ? 2001 John Wiley & Sons, Ltd. Earthquake Engng Struct. Dyn. 2001; 30:1125–1145

BASE ISOLATION FOR RETROFITTING HISTORIC BUILDINGS 1127

The tests on the base-isolated model have been planned with the following aims:

• assessing the seismic behaviour of the isolated masonry sub-assemblage;• evaluating the performance improvements due to the introduction of BIS, through the

comparison of the response obtained for the ‘as is’, +xed-base (FB) sub-assemblage andfor the retro+tted specimen;

• providing a benchmark for numerical modelling of isolated masonry sub-assemblage.

For this purpose, some pseudo-dynamic (PsD) tests have been carried out on the full scalemodel of the base isolated (BI) masonry sub-assemblage.

In this paper an overview of the laboratory model and of the experimental results is pro-vided. In particular, +rstly the test model is described, including the geometry and mechanicalproperties of the masonry specimen and the design of the isolation devices; then the testingmethod is presented, with the description of the sub-structuring procedure of the isolationsystem and the de+nition of the seismic inputs adopted for the PsD tests. Finally, the exper-imental results are discussed and compared to the analogous results obtained on the ‘as is’,FB sub-assemblage model.

3. THE TEST MODEL

3.1. The S. Vicente cloister facade sub-assemblage

The test model represents a sub-assemblage of the cloister facade of the S. Vicente de Foramonastery, a typical monument of Lisbon, erected from the end of the 16 century to thebeginning of the 18 century (Figure 1). The structural system of the monastery is composedby limestone block masonry columns and arches, together with stone masonry-bearing wallsand ceramic domes. More details on the historic, architectural and structural characteristics ofthe monument can be found in Reference [8].

The test model is a plane structure consisting of three stone block columns, two completearches, two external half arches, and an upper part of the arcade made of stone masonry. InFigure 2, a view of the full-scale specimen in the ELSA laboratory is shown, while in Figure 3the schematic layout of the specimen, with the main geometrical dimensions, is provided. Asreported in Reference [8], the specimen has been built in the ELSA Laboratory by usingmaterials and construction techniques as close as possible to those of the actual cloister.The structure displays a typical architectural=structural solution, given by the assemblage ofcolumns supporting arches and=or vaults, which frequently occurs in the masonry monumentalconstructions, particularly in basilica-type churches, where the longitudinal sections of thechurch along the nave arcade present the same structural con+guration and similar dimensionalratios as the test model.

The test model has been originally set-up for representing in a realistic manner a portion ofthe facade of the cloister. Thus, in order to simulate the appropriate boundary conditions ofa periodic structure, with relative vertical and horizontal displacements of the two ends equalto zero, horizontal post-tensioning was applied to the two ends of the model [8].

The presence of the vertical loads resulting from the upper part of the monument, notincluded in the test model, was simulated by vertical actuators located close to the columns.



Figure 1. The cloister facade of the Sao Vicente de Fora Monastery in Lisbon.

At the beginning of the test, at the base of each column, a value of vertical load equal to824 kN results from this load condition and from the self-weight of the specimen. During thetest, the vertical actuators were controlled to avoid the rotation of the upper masonry beam(periodicity condition) and, at the same time, maintain a constant total vertical load.

An ‘original’ loading apparatus was speci+cally designed for applying the horizontal loadson the specimen [8]: a system of rubber cushions +lled with water and communicated bytubes stet evenly distributed forces to be applied to the upper part of the masonry beam,while the horizontal displacement at the centre of it was controlled for the PsD integra-tion.

The total inertial mass moving with the masonry beam was estimated at 400Tons whichincluded masses of other parts of the monument which were not resting on the columns.The +rst period of the masonry sub-assemblage, identi+ed through numerical analyses andin situ dynamic tests, was originally set at TFB ini = 0:25 s. This corresponded to the initialstiPness of the specimen and the mentioned mass of 400Tons. From a cyclic test at 8mmamplitude carried out after the PsD tests on the FB specimen, a value of the secant stiPnessKFB =20:94 kN=mm has been derived. Thus, the period of the FB equivalent single-degree-of-freedom (SDOF) system, at the time of the tests described in this paper, has been obtainedfrom the value of KFB, providing: TFB =0:87 s. The signi+cant elongation of the period withrespect to TFB ini is related to the extensive cracking and local damage occurred in the previousFB tests.

The instrumentation of the specimen provided continuous recording of the forces applied inthe vertical and horizontal actuators and of the controlling displacements.



Figure 2. View of laboratory test model.

Figure 3. Layout of the masonry sub-assemblage.

3.2. The isolation system

The devices selected for the isolation system are high-damping rubber bearings (HDRBs). Inorder to introduce the isolation system, a rigid connection among the column bases has been



Figure 4. Schematic layout of the r.c. beam connecting the column bases.

provided through a reinforced concrete 0:40×0:80m beam, as reported in the schematic layoutin Figure 4. The mass of the beam is equal to 30Tons.

DiPerent working hypotheses on the isolation period, average compression stress and shearmodulus of the rubber have been considered in the isolators design phase. The design procedureconsisted in the main following steps:

• selection of the isolation period TI (2.5–3:5 s);• computation of the isolation stiPness KBIS required for shifting the total mass Mtot

(430Tons) of the test model (masonry sub-assemblage, r.c. connecting beam and iso-lation devices) to the selected value of TI:

KBIS =4�2Mtot=T 2I (1)

• computation of the single-device cross-sectional area A and then of the device diameterQ on the basis of an allowable level of average compression stress (3–10MPa) underthe vertical load acting at the base of each column Wcol (824 kN):

Q= (4(Wcol=)=�)1=2 (2)

• de+nition of the rubber height hrub of the single-device, which provides, for diPer-ent values of the shear modulus G (0.4–0:8MPa), the required lateral stiPness Kdev =KBIS=3:

hrub =GA=Kdev (3)

• check of the value of the aspect ratio R of the device (3–5), in order to guarantee globalstability

R=Q=hrub (4)



Figure 5. Cyclic response of the HDRB devices.

• de+nition of the single rubber layer thickness trub, on the basis of a selected value forthe shape factor S of the device (¿30)

trub =Q=4S (5)

• check of the outcomes of the design process and choice of the isolation devices.

The isolation system resulting from the above design process consists of three HDRBs, oneunder each column, with a diameter of 400 and 108mm of total rubber height. The rubberis characterized by a shear modulus G equal to 0:8MPa, and the working stress level undervertical load is 6:7MPa. This isolation system, at a shear deformation � equal to 100% ofrubber height, shifts the +rst period of the sub-assemblage to 2:5 s.

A sub-structuring method (brieTy described in the following paragraph) has been used inthe PsD testing procedure, thus the isolator devices have not been directly settled under themasonry sub-assemblage: the two parts of the BI specimen (masonry and isolators) have beenseparately tested and the interaction between them has been analytically accounted for. Owingto this sub-structuring arrangement, diPerent scales have been used for the two parts (masonryand isolators) of the test model: the monument sub-assemblage is full scale, while for theisolation system four HDRBs (rubber EN60) with a diameter of 250 and 66mm of total rub-ber height have been used instead of the designed three devices Q400 with 108mm of rubberheight. This sub-structuring strategy also allows for diPerent values of the shear modulus G,thus in the tests two diPerent rubber stiPnesses have been simulated: a ‘stiP’ rubber, charac-terized by G=0:8MPa, which shifts the period to 2:5 s, and a ‘soft’ rubber, with G=0:4MPa,which shifts the period to 3:5 s.

A typical cyclic response of the HDRB Q250 is provided in Figure 5, while in Figure 6the shear modulus and the damping ratio values, obtained in the device characterization tests,are reported as a function of the strain amplitude.



Figure 6. G modulus and damping ratio of the HDRB devices.

4. TESTING METHOD

4.1. Pseudodynamic method

The PsD testing technique is based on modelling the system by a discrete equation of motion

Ma+Cv+ r(d)= p(t) (6)

where M is the mass matrix, C is the viscous damping matrix, a, v, d and p are, respectively,the vectors of acceleration, velocity, displacement and external load, which are functions oftime t, and r is the vector of restoring forces, which is a non-linear function of the displace-ments. Within this model, M, C and p(t) are known data, while r(d) is directly measuredon line. Typically, the viscous damping matrix C is considered null in a PsD test. Usually,an explicit integration scheme is used and, at every step, the computed displacement is quasi-statically imposed to the specimen and the required forces are simultaneously measured. Byusing many actuators of the required capacity, the method can be applied to test large structureswith clear advantages with respect to a shaking table test [10; 11].

4.2. Fast continuous pseudo-dynamic testing

In a classic PsD test every integration time step takes typically at least 1 s of time whichallows to impose the ramp of incremental displacements, wait for the stabilisation of thesystem, before the forces are measured, and compute the next displacement. However, ina fast continuous PsD test, as currently implemented at the ELSA laboratory [12], everyintegration time step takes only 2ms, which is also the sampling period of the closed-loopcontrollers of the actuators. Within that time lapse, the same CPU which is in charge of thecontrol algorithm reads the force, integrates one step in the equation of motion and correctsthe target according to the new computed displacement. The ground acceleration time historyis sub-divided in very small time increments (for example 20 �s), so that the displacementincrements can be appropriately followed by the pistons in just 2ms. Thus, for a real-size



Figure 7. Sub-structuring method adopted in the PsD tests.

specimen, a test time scale of �=2ms=20 �s=100 can be reached. This could mean beingaround +ve to 10 times faster and still more accurate, comparing with a classic PsD testexecuted with the same hardware.

4.3. Sub-structuring of the isolation system

The ELSA laboratory team has developed a sub-structuring method for performing PsD testsof base-isolated structures [13]. In fact, the Texibility of the PsD technique allows to dividethe restoring force system in several substructures, which can be separately tested or numer-ically simulated. Thus, the experimental arrangement has not required the direct settling ofthe isolators under the masonry sub-assemblage, since the isolation system has been simulta-neously, but separately, tested (Figure 7). The interaction between the masonry specimen andthe isolators has been analytically considered at the level of the equation of motion. Then,only base-relative displacements are imposed to the masonry specimen and the restoring forcefor the base is obtained as the static reaction to the measured forces on the masonry specimenplus the force at the isolator specimens when submitted to the speci+ed displacement.

The described sub-structuring testing technique was particularly advisable in this speci+ccase since the S. Vicente de Fora sub-assemblage specimen, in the FB condition, had beenextensively tested in previous experimental programs [8] and therefore it was already clampedto the strong Toor of the laboratory.

4.4. Scaling of isolator forces and displacements

Although the rubber of the isolators used for the test was of the same type as the prototype(nominal G=0:8MPa), there was a diPerence of scale between the isolator devices of bothcases. Particularly, the test set-up used four isolator devices with a diameter of 250mm anda total rubber height of 66mm each, while the prototype isolation system consists of threedevices with a diameter of 400mm and a total rubber height of 108mm each. This means



that, in order to have the same strain as in the prototype, the displacement imposed to thespecimen has to be multiplied by a factor of

dSPEC=dPROT = hRUBBERSPEC =hRUBBER

PROT =66=108=0:611 (7)

In the same way, assuming that the stress is invariant, the measured load at the specimen hasto be multiplied by a factor of

rPROT=rSPEC =APROT=ASPEC =(3�44002

)/(4�42502

)=1:92 (8)

in order to translate it into the prototype scale.As an approximation, for two of the tests, the results for a case of a softer prototype rubber

with nominal G of 0:4MPa were also simulated. In this case, the considered force factor wasof

rPROT=rSPEC =GPROTAPROT=GSPECASPEC =(0:4 · 3�

44002

)/(0:8 · 4�

42502

)=0:96 (9)

From the results of other experiments with both types of rubber [11], we have seen that thesofter rubber with G of 0:4MPa diPers from the harder one with G of 0:8MPa not only in thestiPness, but also slightly in the strain-hardening characteristics. Moreover, in the tested case,the observed damping ratio for the softer rubber was an absolute 2 per cent smaller than forthe harder one. For this reason such simulation of the softer rubber based on a simple scalingof the forces can be considered only as a rough preliminary approximation.

Since, unfortunately, the shape factor of the rubber disks was not the same for the prototypeand the specimen isolators, applying the same vertical pressure on both cases would haveconducted to diPerent vertical-load-induced shear stress. For this reason, the quotient of verticalpressure and aspect ratio was settled as invariant instead. This meant that the adopted verticalpressure on the specimen rubber was around 60 per cent of the prototype one. On the otherhand, the ePect of changes of the vertical pressure on these specimens has also been tested:a decrease of 50 per cent in the vertical load meant a relative 2 per cent of increase in theshear stiPness accompanied by an absolute 1 per cent decrease in the damping ratio [11].

Of course, these problems can be avoided when specimens are available with rubber type,shape factor and number of rubber disks equal to the assumed ones for the prototype.

4.5. Compensation of the strain-rate e9ect

For materials with strain-rate ePect (SRE) as the high-damping rubber isolators used in thistest, the advantage of the continuous PsD technique is double. Firstly, the test is faster whichreduces the SRE and, secondly, the ramp and hold phases at every step become diPused ina continuous movement, which avoids the stress relaxation within each step and the diLcultmatter of the selection of the stabilization time before measuring the force.

Since the speed of testing is nevertheless limited by the required accuracy in the control,the duration of the test is usually from 30 to 300 times longer than the original durationof the event. The important SRE existing in the rubber devices at that testing speed, whichis still important, can be minimised by using the compensating strategy developed at the



Figure 8. Force correction factor as a function of the test time scale (Lambda) for the adopted rubber.

ELSA laboratory. If, for example, the current testing speed is �=100, the correspondingcorrection factor for the forces at the isolator level is 1.127 (see Figure 8). That is, instead ofsimply using in the integration algorithm the restoring force measured at the piston and scaled,this is modi+ed by a correction term of 1.127. This correction does not aPect the dampingcharacteristics and is independent of the deformation level. It can be shown that, for the typeof rubber used such a correction is well suited [14].

5. THE SEISMIC INPUT

The PsD tests have been carried out with two acceleration histories, reported in Figure 9,already utilised for the previous tests on the ‘as is’ (+xed base) masonry specimen, respectively,corresponding to a moderate (174 years return period) and high (975 years return period)earthquake intensity. In the tests, the high-intensity acceleration history has been applied tothe specimen scaled at 1.5 times.

The seismic inputs correspond to two earthquake scenarios envisaged for Lisbon, namelya near-+eld- and a far-+eld-type signal, characterized by major energy content in diPerentfrequency ranges and by diPerent duration, respectively, 10 and 30 s.

In the following, the moderate-intensity (174 years return period), near-+eld-type seismicinput is appointed as lip174, while the high-intensity (975 years return period), far-+eld-typeseismic input is labelled as lia975.



Figure 9. Acceleration histories.

Figure 10. Acceleration and displacement spectra.

In Figure 10(a) and (b) are provided the response spectra, respectively, in terms of acceler-ation and displacement. It is evident that the near-+eld seismic input has the highest spectralordinates in the low-period range, while for periods greater than 1 s a considerable reductionof the accelerations, with not signi+cant increase of the displacements, can be derived. Thefar-+eld signal presents less sharp reduction of acceleration in the long-period range than thenear-+eld signal and very large displacements (larger than 100mm) for periods greater than2 s.

Since the BIS has been designed for shifting the +rst period of the sub-assemblage to2:5 s (with HDRBs characterized by G=0:8MPa) and to 3:5 s (with HDRBs characterizedby G=0:4MPa), the maximum reduction of forces and deformation in the masonry part isexpected under the near-+eld-type input. The reduction of the forces in the elevation partis due to the ePects of the period shifting and of the damping in the isolators. The orderof magnitude of the structural response reduction can be estimated on the basis of the ra-tio between the spectral acceleration SA at TFB and TI. In Table I, the values of the ratio



Table I. Ratios between spectrum acceleration values at FB and BI periods.

Seismic input RSA =SA(TFB; �FB)=SA(TI; �I)

�I = 5 per cent �I = 10 per cent

TI = 2:5 s TI = 3:5 s TI = 2:5 s TI = 3:5 s

Lip174 5.5 10.2 7.5 11.7Lia975 2.9 6.3 4.2 7.6

RSA =SA(TFB; �FB)=SA(TI; �I) are reported for the two seismic inputs, assuming TI equal to2.5 and 3:5 s, �I equal to 5 and 10 per cent, TFB equal to 0.87 and �FB equal to 5 per cent.

6. SUMMARY OF EXPERIMENTAL RESULTS

The PsD test results concern displacement and restoring force histories of the specimen bothat the isolation level and at the top level of the masonry sub-assemblage.

In Figures 11(a–d), the histories of the displacement at the isolators level (level 0) and ofthe relative displacement of the masonry sub-assemblage (level 1) with respect to the top ofisolators are reported. The four charts refer to the diPerent PsD tests, namely:

(a) seismic input: 100 per cent lip174, rubber: G=0:8MPa,(b) seismic input: 150 per cent lia975, rubber: G=0:8MPa,(c) seismic input: 100 per cent lip174, rubber: G=0:4MPa and(d) seismic input: 150 per cent lia975, rubber: G=0:4MPa.

A large diPerence between the two inter-level displacement histories, at the isolation level(0) and at the masonry level (1), is evident in all the charts. Larger absolute displacements atthe level 0 as well as lower relative base-to-top displacements in the masonry sub-assemblagecan be observed in the tests with rubber devices characterized by G=0:4MPa (chart (c) and(d)) than in the tests with G=0:8MPa ((a) and (b)).

The maximum displacements experienced by the specimen in the four PsD tests are close tothe spectral displacements, respectively, at T =2:5 s for the tests with the stiP rubber devices(G=0:8MPa) and T =3:5 s for the tests with the soft rubber devices (G=0:4MPa). Theexperimental displacements and the spectral values (�I = 10 per cent) are compared in Table II.

In Figures 12(a–d), the histories of the restoring forces registered at the isolators level(level 0) and at the masonry sub-assemblage level (level 1) are reported. The four charts referto the diPerent PsD tests, as indicated with reference to the previous +gure.

From these charts it is possible to derive the cases in which the BIS has the maximum eL-cacy in reducing the forces in the elevation parts. As expected, the near-+eld (lip174) seismicinput (chart (a) and (c)) leads to the lower values of forces in the upper part (Fmax =58 kN(a), 28 kN (c)), particularly when ‘soft’ (G=0:4MPa) rubber devices are adopted. Signi+-cantly higher forces have been registered in the specimen under the far-+eld (lia975) seismicinput: the maximum value reaches 322 kN in the case (b) (rubber with G=0:8MPa) whichis reduced to 226 kN in the case (d) (rubber with G=0:4MPa).



Figure 11. PsD test results: inter-level displacement histories.

7. EVOLUTION OF EIGEN-FREQUENCIES AND DAMPINGS

Another way to characterise the obtained isolated system is by analysing the response nat-ural frequencies and damping. Figure 13 contains the histories of eigen-frequency (top) anddamping ratio (bottom) for the +rst mode, while Figure 14 refers to the second mode in thesame way. The frequency and viscous-equivalent damping ratio were obtained by applying anidenti+cation method to the experimental response. The method is based on a time-domainlinear +lter model for which the input is the accelerogram and the output are the displace-ments. This method is currently extensively applied to the results of PsD tests at the ELSA[15]. Each identi+cation of the parameters was based on a window of 5 s of data centred on



Table II. Displacement of isolation system: comparison betweenspectrum values and test results.

Seismic input Displacements (mm)

Spectrum Test

TI = 2:5 s TI = 3:5 s G=0:8MPa G=0:4MPa

10.8 (lev.0) 14.4 (lev.0)Lip174 11.5 14.2

11.6 (lev.1) 14.7 (lev.1)108 (lev.0) 138 (lev.0)

Lia975 113 136130 (lev.1) 149 (lev.1)

a time instant. The identi+cation was repeated for successive time instants in order to ob-tain the time history of the frequency and the damping ratio. Since the system is non-linearand the identi+cation is based on a linear model, some irregular results are detected at theplots. The isolated mode is the +rst one of the 2-DOF system and corresponds basically tothe deformation of the isolators while the superstructure remains almost undeformed as a rigidbody. As shown in Figure 13, its frequency changes depending on the time and on the test.Clearly, diPerent displacement amplitudes render diPerent frequencies even during each test.This is explained by the changes in the aspect of the force-displacement loops as a functionof the displacement amplitude. The diPerences from test to test are stronger and are due to thediPerence also in displacement amplitude or to the diPerent rubber-type. On the other hand,the associated damping ratio oscillates around 10 per cent throughout the test.

The second mode is characterized by signi+cative deformation of the superstructure as com-pared to the isolators. For an ePectively isolated structure, its response should is marginal.This fact makes its identi+cation harder. Particularly, for the far-+eld accelerogram with largernon-linearities, the method used fails very often as shown in Figure 14. Nevertheless, it is in-teresting to observe that apparently for the second mode either the frequency or the damping:

(1) does not depend on the rubber-type,(2) but does depend on the intensity of the earthquake.

This could be due to the addition of two factors:

(1) This mode is characterized by levels of drift which are in the same order of magnitudefor the monument and for the isolators. Of course, in that case, this frequency isgoverned by the stiPness of the masonry superstructure and does not depend much onthe stiPness of the rubber.

(2) DiPerent levels of pseudo-acceleration imply diPerent equivalent loads that determinediPerent amplitudes of deformation at the superstructure. These change the eigen-frequency.

The same reasons could be given for the damping.



Figure 12. PsD test results: restoring force histories.

8. COMPARISON TO THE FIXED-BASE EXPERIMENTAL RESPONSE

In Figures 15(a–d), the experimental cyclic response is plotted in terms of shear load vs. driftdisplacement in the masonry sub-assemblage. In the same charts, the analogous experimentalresponses obtained in the tests on the FB specimen, carried out by Pinto et al. [8], arereported. The four charts refer to the diPerent PsD tests, as already speci+ed with referenceto the previous +gures.

The comparison between the response of the BI model and the response of the FB modelallows to evaluate in quantitative terms the improvement achieved with the introduction ofBIS.



Figure 13. Histories of eigen-frequency (top) and damping ratio (bottom) for the +rst mode.One curve is plotted for each test.

The observations concerning the ePect of the seismic input and of the rubber shear mod-ulus on the response of the specimen are con+rmed with reference to the global force vs.displacement cyclic curves.

The comparison between the test cyclic response of the FB and BI specimen reported inFigure 15(a) provides reduction factors for the forces equal to 7.25 and for displacementsequal to 8.9; these reduction factors are remarkably increased considering the results obtainedin the test with soft rubber devices (G=0:4MPa), as can be observed from Figure 15(c): the



Figure 14. Histories of eigen-frequency (top) and damping ratio (bottom) for the second mode.One curve is plotted for each test.

ratio of FB to BI maximum force in the elevation is equal to 15, while the same ratio interms of maximum displacement reaches 24.5.

Lower reduction factors are obtained in the cases of Figure 15(b) and (d), which reportthe results of the tests with the far-+eld lia975 seismic input ampli+ed 1.5 times. The ratiosof the maximum displacements (FB=BI values) provide 2.8 and 5.8, respectively, for stiP (b)and soft (d) isolation devices, while the same ratio in terms of forces in the masonry part,respectively, provides: 1.5 (b) and 2.1 (d).



Figure 15. Force-displacement cyclic response: comparison of FB and BI specimens.

Table III. Restoring force values: comparison between FB and BI specimens.

Seismic input FB Force (kN)

BI FB=BI

G=0:8MPa G=0:4MPa G=0:8MPa G=0:4MPa

Lip174 420 58 28 7.25 15Lia975 481 322 226 1.5 2.1

In Tables III and IV, the above reduction factors are summarized together with the maxi-mum absolute values of forces and displacements experienced in the tests by the FB and theBI specimens.



Table IV. Displacement values in the masonry part: comparison between FB and BIspecimens.

Seismic input FB Displacements (mm)

BI FB=BI

G=0:8MPa G=0:4MPa G=0:8MPa G=0:4MPa

Lip174 7.55 0.85 0.3 8.9 24.5Lia975 64.1 23.1 11.0 2.8 5.8

9. CONCLUDING REMARKS AND FURTHER DEVELOPMENTS

In this paper the PsD tests carried out on the full-scale model of a BI masonry sub-assemblagehave been presented. An overview on the laboratory model and on the experimental resultshas been provided, assuming as term of comparison the analogous results obtained on the ‘asis’, FB sub-assemblage model.

The experimental results suggest a particularly improved behaviour of the BI specimen.Depending on the earthquake and on the rubber of the isolators, the masonry part above theisolators experiences displacements from 2.8 up to 24 times smaller than the ones registeredin the FB test model. The forces are reduced from 1.5 up to 15 times. No further crackingand damage in the masonry elements of the specimen, which extensively occurred during theprevious FB tests, have been observed following the BI tests.

These test results, even though con+rming the well known 2-DOF-type behaviour of BIstructures, are particularly signi+cant since have been obtained on a full-scale model, anddemonstrate that the BIS can eLciently reduce force and deformation in monumental buildings.In addition, such results are currently being used as reference for calibrating numerical models.

ACKNOWLEDGEMENTS

For the +rst two authors, the research presented in the paper is supported by CNR grant No. 98.3212.PF54and by MURST PRIN ’97.

The COSISMO Project was jointly developed by the Portuguese Directorate General for Monuments(DGEMN) and the National Laboratory for Civil Engineering (LNEC) as well as by the ELSA Labo-ratory of the JRC of the European Commission. The +nancing provided by the DGEMN is gratefullyacknowledged.

REFERENCES

1. De Luca A, Mele E. The seismic isolation in the retro+t of historic buildings. Proceedings of USA=ITALYSeminar Seismic Restoration of Historic Buildings, Los Angeles, USA, July 1996.

2. Mele E, De Luca A. Seismic strengthening of cultural heritage through base isolation: some case studies.Proceedings of International Conference New Technologies in Structural Engineering, Lisbon, Portugal, July1997.

3. Mele E, De Luca A. Behaviour and modelling of masonry church buildings in seismic regions. In EarthquakeResistant Engineering Structures, Oliveto G, Brebbia CA (eds). WIT: Southampton, UK, June 1999.



4. Mele E, Giordano A, De Luca A. Nonlinear analysis of some typical elements of a basilica plan church, InEarthquake Resistant Engineering Structures, Oliveto G, Brebbia CA (eds). WIT: Southampton, UK, June1999.

5. Mele E, Modano M, De Luca A. The seismic retro+t of historic masonry buildings through BIS: preliminaryanalysis for application to church typology. Proceedings of MONUMENT ’98 Workshop on SeismicPerformance of Monuments, Lisbon, Portugal, November 1998.

6. Mele E, De Luca A, Giagnuolo C. L’ePetto di alcuni interventi di adeguamento sismico sul comportamento diedi+ci in muratura. Proceedings of the 8th ANIDIS Italian National Conference on Earthquake Engineering“L’Ingegneria Sismica in Italia”, Torino, Italy, September 1999 (in Italian).

7. Dyngeland T, Pinto A. Linear dynamic analysis of the Sao Vicente de Fora Monastery in Lisbon, Portugal. JRCISIS ELSA Special Publication No.I.97.18, January 1997.

8. Pinto AV, Verzeletti G, Molina FJ, Plumier C. Seismic tests on the S. Vicente de Fora model. Proceedings ofMONUMENT ’98 Workshop on Seismic Performance of Monuments, Lisbon, Portugal, November 1998.

9. Pegon P, Pinto A. Numerical modelling in support of experimental model de+nition-The S. Vicente de Foramodel. Proceedings of MONUMENT ’98 Workshop on Seismic Performance of Monuments, Lisbon, Portugal,November 1998.

10. Donea J, Magonette G, Negro P, Pegon P, Pinto A, Verzeletti G. Pseudodynamic capabilities of the ELSAlaboratory for earthquake testing of large structures. Earthquake Spectra 1996; 12(1):163–180.

11. Molina FJ, Verzeletti G, Magonette G, Marazzi F, Renda V. Pseudodynamic tests on base isolators with numericalsubstructuring of the superstructure. JRC-ISIS-SSMU Internal Report, Contract no. 14908-1999-03, November1999.

12. Magonette G, Pegon P, Molina FJ, Buchet Ph. Development of fast continuous substructuring tests. Proceedingsof the 2nd World Conference on Structural Control, Kyoto, 1998.

13. Molina FJ, Verzeletti G, Magonette G, Tirelli D, Anthoine A, Renda V. Experimental pseudo-dynamic responseof a large-size model of the “Palazzo Geraci” with base isolation. Proceedings of the XI European Conferenceon Earthquake Engineering, Paris, 1998.

14. Molina FJ, Pegon P, Verzeletti G. Time-domain identi+cation from seismic pseudodynamic test results oncivil engineering specimens. Proceedings of the 2nd International Conference on Identi?cation in EngineeringSystems, The Cromwell Press Ltd., 1999.

15. Molina FJ, Verzeletti G, Magonette G, Buchet Ph, GYeradin M. Bi-directional pseudodynamic test of a full-sizethree-storey building. Earthquake Engineering & Structural Dynamics 1999; 28:1541–1566.


Base isolation for retrofitting historic buildings: Evaluation of seismic performance through...

Documents

Transcript of Base isolation for retrofitting historic buildings: Evaluation of seismic performance through...