c frames retrofitted with

This article was downloaded by: [Univ Studi Basilicata] On: 15 October 2012, At: 02:52 Publisher: Taylor & Francis Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Earthquake Engineering Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/ueqe20

EXPERIMENTAL BEHAVIOUR OF R/C FRAMES RETROFITTED WITH DISSIPATING AND RE-CENTRING BRACES a

a

a

DONATELLO GARDONE , MAURO DOLCE , FELICE CARLO PONZO & EMA COELHO

b

a

DiSGG—Department of Structures, Geotechnics, Engineering Geology, University of Basilicata, Contrada Macchia Romana, 85100, Potenza, Italy b

Laboratorio National de Engenharia Civil de Lisboa, Avenida do Brasil, Lisbon, Portugal Email: Version of record first published: 03 Jun 2008.

To cite this article: DONATELLO GARDONE, MAURO DOLCE, FELICE CARLO PONZO & EMA COELHO (2004): EXPERIMENTAL BEHAVIOUR OF R/C FRAMES RETROFITTED WITH DISSIPATING AND RE-CENTRING BRACES, Journal of Earthquake Engineering, 8:3, 361-396 To link to this article: http://dx.doi.org/10.1080/13632460409350493

PLEASE SCROLL DOWN FOR ARTICLE Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to anyone is expressly forbidden. The publisher does not give any warranty express or implied or make any representation that the contents will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in connection with or arising out of the use of this material.

Journal of Earthquake Engineering, Vol. 8, No. 3 (2004)361-396 @ Imperial College Press

@ Imperialwww.icprcss.co.uk College Press

EXPERIMENTAL BEHAVIOUR OF R/C FRAMES RETROFITTED WITH DISSIPATING AND RECENTRING BRACES

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

DiSGG - Department of Stmctures, Geotechnics, Engineering Geology, Unaversity of Basiiicata, Contrada Macchia Rornana, 85100, Potenzu, Italy 'donatello.~Qtiscalinet. it tdolcemmQlibem.it . [email protected] E M A COELHO Labomtorio National de Engenharia Civil de Lisbm, Avenida do Bmsil, Lisbon, Portvgal emu. we1hoQlnec.pt Received 6 M a y 2002 Revised 26 August 2002 Accepted 11 June 2003 An extensive program of shaking table tests on 1/4-scale three-dimensional R/C frames was jointly carried out by the Department of Structure, Soil Mechanics and Engineering Geology (DiSGG) of the University of Basihcata, Italy, and the National Laboratory of Civil Engineering (LNEC),Portugal. It was aimed a t evaluating the effectiveness of passive control bracing systems for the seismic retrofit of R/C frames designed for gravity loads only. Two different types of braces were considered, one based on the hysteretic behaviour of steel elements, the other on the superelastic properties of Shape Memory Alloys (SMA). Different protection strategies were pursued, in order to fully exploit the high energy dissipation capacity of steel-based devices, on one hand, and the supplemental re-centring capacity of SMA-based devices, on the other hand. The experimental results confirmed the great potentials of both strategies and of the associated devices in limiting structural damage. The retrofitted model was subjected to table accelerations as high as three times the acceleration leading the unprotected model to collapse, with no significant damage to structural elements. Moreover, the re-centring capability of the SMA-based bracing system was able to recover the undeforrned shape of the frame, when it was in a near-collapse condition. In this paper the experimenta1 behaviour of the non protected and of the protected structural models are described and compared.

Keywords: Reinforced concrete frames; retrofitting; seismic behaviour; shaking table tests; energy dissipating brace; re-centring brace; shape memory alloys.

1. Introduction

R/C framed structures designed for gravity loads only suffer heavy damage and often collapse under strong earthquakes. Actually, they have some inherent lateral strength, frequently improved by the collaboration of i d l e d masonry panels, which may be sufficient t o resist minor earthquakes. However, the lack of proper detailing in structural members and the disregard of basic seismic design rules lead to

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

inadequate seismic performance under strong earthquakes. The collapse of R/C framed structur& is often the major cause of loss of life and money during an earthquake and the need for seismic rehabilitation of these structures is a primary concern throughout the world. Conventional retrofitting techniques for R/C framed structures are based on the strengthening of structural elements e.g. by concrete jacketing [Bracci et al., 1992) and/or using steel angles connected by transverse steel plates. Alternatively, new strong structural members, such as R/C walls, are introduced into the existing structure, to absorb most of the seismic forces. A different approach, based on passive control concepts, introduces supplemental energy dissipation systems [Constantinou et al., 19981, which exploit special mechanical devices, incorporated in steel braces connecting two consecutive storeys. The interstorey drifts activate such devices before the main structural members are fully engaged in their inelastic behaviour. The two functions of supporting gravity loads and of resisting lateral forces while dissipating energy during an earthquake are kept separate, being carried out by two almost independent systems. -- -- - - - -- -- . --- -- - --- - -The advantages in using supplemental energy dissipating systems in the retrofit of existing structures are different. They attain higher levels of seismic protection than conventional ductile structures, maintain the functional and operational capability of a structure even after strong earthquakes, and reduce repair costs. Besides increasing the energy dissipation capacity of a structure, most of the energy dissipation systems also increase its strength and stiffness. Their use, therefore, results in a reduction of the interstorey drifts (and hence of structural and non-structural damage), but also in an increase of the total lateral force experienced by the structure [Constantinou et al., 19981. For structures designed to remain elastic during an earthquake, the addition of dissipating braces is generally favourable, in the sense that it produces a considerable reduction of the vibration amplitude [Chang et al., 1991; Constantinou and Symans, 1993; Whittaker et al., 1989). For ductile structures designed according to modern seismic criteria, hence characterised by a good energy dissipation capability through plastic hinges in beams, the addition of energy dissipating braces is not very effective in further reducing the amplitude of vibrations. On the contrary, structures designed for gravity loads often only develop plastic hinges in columns, whose hysteresis loops are characterised by low energy dissipation capacity and heavy degradation of strength and stiffness. In this case, the addition of dissipating braces results in a sigruficaut improvement of the dissipation capacity and, then, in a large reduction of the vibration amplitude [Lobo e t al.,1993; Reinhorn et al., 1995).The interest for these techniques is shown by the numerous theoretical and e,uperimental works of the last decade [Soong, D a r p h , 19971. Seismic devices can dissipate energy by exploiting the properties of different materials and mechanisms [Dolce, 19941, e.g. hysteresis of metals, sliding friction, viscosity of fluids,viscoelasticity of polymers. -

A

- - -

--

.

.

- -

.

Ezperimental Behaviour of

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

--

R/C Frames 363

Recently, some special alloys, referred to as Shape Memory Alloys (SMA) [Duerig et a!., 19901, have been regarded as possible candidates for their use in seismic isolation and energy dissipation systems [Dolce et nl., 2000; Hodgson and Krumme, 1994; Whittaker et al., 1995; Witting and Cozzarelli, 19921. SMA's show several characteristics which make them suited to seismic applications [Dolce and Cardone, 2001a and 2001bj. The most important one is their ability to dissipate energy through repeated cycling with neither significant degradation nor permanent deformations (superelasticity). Their usable strain range is also much larger than -ordinary metals. -Other -favourablefeatures-of SM-A's-are their considerable -fatigue resistance to large strain cycles and, in the case of Nickel-Titanium alloys, their exceptional corrosion resistance. The experimental work presented here investigates the effectiveness of passive control braces for the seismic retrofit of R/C frames designed for gravity loads, and follows two previous experimental research projects, both funded by the European Commission. These projects studied the experimental behaviour of R/C plane frames, respectively designed for gravity loads and retrofitted with steel-based energy dissipating braces [Braga et al., 20021, designed for low intensity seismic actions and retrofitted with steel-based and with SMA-based energy dissipating braces [Nicoletti et al., 19991. In the present research, two 1/4 scale three-dimensional frames are tested. Two different bracing systems are used to retrofit one of them, to compare their seismic response, as well as the response of the bare frame, in an extensive shaking table test investigation. A specific concern is devoted to the horizontal bi-directional action of earthquakes and to the capability of the bracing systems t o limit the torsion due to mass-stsness eccentricity. Two different devices for passive control braces were tested. They were based on the hysteretic dissipating capacity of steel and on the superelastic property of SMA's, respectively. The SMA-based device exhibits a remarkable (supplemental) re-centring capability, to restore the undefomed shape of the structure, even when the structural members yields 'extensively. This feature, carefully studied during the tests, opens new possibilities in the retrofit strategy of weak bames, allowing for even a large ductility demand on non-duct ile members. The shaking table tests were designed at the Department of Structures (DiSGG) of the University of Basilicata and carried out at the National Laboratory of Civil Engineering in Lisbon (LNEC), within the research program ECOEST 11. -

2. Experimental Model

The building prototype was designed for gravity loads only, according to the pre1971 Italian Regulation for R/C structures. R/ C buildings realised before 1971 constitute a remarkable part of the building stock, mot only in Italy, but all over the world. Most of them, though being built in seismic areas, were not designed to resist any horizontal action. Therefore, neither horizontal actions nor specific design criteria (capacity design) and rules for seismic detailing (minimum

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Fig. 1. (a) Axonometric view and (b) perspective views of the l/Pscale structural model tested on shaking table.

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Experimental Behaviour of

R/C Fmmes 365

amount of reinforcement, distance between ties a t the beam and column ends, etc.) were followed. The models being very poor in terms of both seismic resistance and ductility, represent an ideal test to verify the effectiveness of any retrofitting method. The R/C 1/4scale three-dimensional hame model, having four storeys, two longitudinal bays and one transverse bay, is shown in the axonometric view of Fig. l(a) and in the lateral and frontal views of Fig. l(b). The two lateral columns had constant cross section, corresponding to 25 x 30 cm in the prototype scale. On t h e - c o n t r m y , - t h e - cross -section-of the-central- column-reduced-from 35-x 30-c-m -to-25 x 30 cm, in the prototype scale, at the third storey. The cross section of the beams corresponds to 30 x 50 cm, in the prototype scale, a t the f i s t three storeys and to 30 x 40 cm at the fourth storey. A 30 cm thick R/C slab was made at the base of the model, to move and bolt it to the table. The models were the 3D extension of two similar plane models already tested at the University of Bristol [Braga et al., 19951, for the European research Program ECOEST. The same geometry of the plane model was kept, adding only one span. in the orthogond direction. The excellent performance of the energy dissipating steel-based braces tested with those models induced the authors to investigate the more complex behaviour of a similar 3D frame. In the present case the models were subjected to real earthquakes, instead of generated earthquakes, and different upgrading techniques were compared. Extreme care was taken in preparing the concrete for the model, as the choice of the microconcrete composition assumes great importance to reproduce the actual behaviour of the full-scale structure, especially its post-elastic behaviour. As a matter of fact, the mechanical properties of a microconcrete can significantly differ from those of a standard concrete [Woo et al., 19881. Typically, a microconcrete exhibits greater deformability in compression and higher tensile strength with respect to a standard concrete with the same compression strength. The greater deformability in compression results in a certain distortion in terms of displacement and strains between the full-scale structure and the reduced-scale model. The higher tensile strength, instead, can give rise to sensible delays in the cracking of the R/C members, with consequent improvement of the concrete-steel bond resistance. In order to attenuate the two aforesaid aspects, it was necessary to properly choose: (i) the gramdometry of the aggregate, (iij the aggregatelcernent ratio and (iii) the waterlcement ratio. Several mixes were manufactured and tested. The selected one (Table 1) was realised using sand with 50% of the weight having grain size between Table 1. Concrete weight proportions. Components

Ratio

Aggregate/Cernent I1 32.5

2.1

Water/Cement Additive/Cement (linosulfate modifyed)

0.6 0.01

Table 2.

Concrete compressive tests. - .-

Specimen

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

No.

uu (7 days) (N/mm2)

a, (28 days)

(N/mm2)

2 and 5 mm. The average compression strength was about 27 N / m 2 (Table 2), 25 N/mm2 being the corresponding design value. The amount of steel reinforcement designed for the prototype was scaled down for the model. Reference to FeB38k steel (375 N/rnmz characteristic yield strength) was made in the design. Threaded steel bars were used to simulate the bond with the concrete and achieve a good resemblance in cracking behaviour. The bars were subjected to annealing treatment for 1 hour at 650°C, followed by air cooling, to give them an -acceptatile~diuctili~~~d~rediice theiryieldstrengtkto -approximately 260-N/mm2 ;The strength distortion introduced by the actual strength of steel bars, after heat treatment, made it necessary to re-calculate the reinforcement amount obtained with the similitude ratio of the dimensional analysis. The calibration of the scale factor was based on the results of tensile tests carried out on two series of 3 to .

-..

Fig. 2. Design reinforcement of the 1/4-scaled model for gravity load condition.

Rettntorcement of the X dlrectlon tram8 *sect. n. Base Height Reinforc. a in-. un 1 7.5 304 625 lb 7.5 a 5 8.75 2 7.5 1+4+W 125 3 7.5 2+4+2$3 125 4 7.5 +2@ 125 5 7.5 6.25 3#4 7.5 8.75 345 7.5 125 1+2# 7 7.5 125 2+4+2@ ---a-7;5---12i5---w--9 7.5 8.25 394 9b 7.5 6.25 395 10 7.5 125 11 7.5 125 *2@ 12 7.5 125 3+4+2+3 13 7.5 8.25 W 2 + 3 13b 7.5 6.25 394 14 7.5 10 2+4+1+3 15 7.5 10 3+4 16 7.5 10 +2@

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

i

Reinforc. in compres, 394 395

Stimrps

.

@3/5

4318 293 9W.5 194+2g 9317.5 342 9312.5 394 6315 395 $36 293 93/2.5 1*2@ 9W.5 --2#2- --9=:53@4 034.5 345 (a315 293 9312.5 2+4+1@ #m.5 202 #3/2.5 3@+2@ 63/45 304 935 293 9W.5 204+1@ 937.5 2#2 63/25

.

Fig. 2. (Continued)

The prototype structure was designed to carry 2.00 kN/m2 live load on the first three floors and 0.80 kN/m2 on the top floor. Additional masses were put on each floor slab of the model, to account for the non structural dead loads, a fraction (1/3) of live loads and the increase d mass in the model due to the mass similitude scaling. Steel blocks were used to apply 24 kN additional weight at each storey.

3. Seismic Retrofit Two different types of braces were used, based respectively on the hysteretic behaviour of steel components and on the superelastic properties of shape memory alloys (SMA). Consequently, two alternative retrofitting strategies were pursued. The first strategy is aimed at limiting the ductility demand in R/C members and the residual displacement of the structure, by exploiting the high stiffness and energy dissipation capability of steel-based Energy Dissipating Braces (EDB's). The second strategy is aimed at restoring the undeformed shape of the model at the end of the earthquake (no residual displacement), even accepting significant inelastic deformations in R/C members, by exploiting the supplemental recentring capability of SMA-based Supplemental Re-centring Braces (SRB's) [Dolce et al.,

naool.

Figure 3 shows the arrangement of the steel-based EDB, together with a typical force-displacement diagram. The EDB was made up of a pair of steel C-section truss and two steel brackets, to connect the brace to the hame. The brackets were connected to beams and columns through tensioned bars (Fig. 4). Epoxy resin grout was used to assure a perfect bond and avoid any relative movements between brace and hame. The dissipating unit of the brace consist of four double fiag-shaped steel elements connecting the C-sections with the bottom steel bracket. During an

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

-.

-- --

_

-

- - - - _ .

-- - -

.

(b) .- -

-

-

..- .-

. .. .

.

-

Fig. 3. Steel-based EDB:(a) kerneI component, (b) arrangement of the device and (c) typical cyclic behaviour.

Fig. 4.

Bracing connections and dissipative unit with steel elements.

earthquake, the steel elements are plastically deformed in double bending, thus dissipating energy. They had dserent dimensions at the various storeys, in order to provide the device with the designed stiffness and yielding strength. Figure 5 shows the arrangement of the SMA-based SRB and a typical forcedisplacement diagram representing the cyclic mechanical behaviour of the device.

-

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Experimental Behaviour of R/C Fmmes

369

Fig. 5. SMA-based SRB: (a) arrangement of the SMA wires inside the device and (b) typical cyclic behaviour.

The SMA brace was made up of two concentric steel tubes moving relatively to each other, following the interstorey drifts. The kernel components were made of SMA austenite superelastic wires arranged in two different ways. They provided the device with two basic features, such as supplemental re-centring capacity and, compatibly with the first requirement, good energy dissipation capacity [Dolce et al., 20001. The re-centring group was made of pre-tensioned SMA wires wounded around two studs, and inserted into oval-shaped holes of the steel tubes. Due to this particular arrangement, the SMA wire loops are always subjected to tension loading- unloading cycles, independently of the direct ion of the relative movements between the steel tubes. This gives rise to a cyclic behaviour which can be modelled as non linear elastic. The dissipating group was again made of pre-tensioned SMA wires, wounded around three studs, resulting in two independent loops. The central stud was connected to the external tube, the lateral studs to the internal one. When the steel tubes move relatively to each other, one SMA wire loop elongates, while the other shortens. This gives rise to a highly dissipating cyclic behaviour, which can be modelled as elasto-plastic. By properly combining SMA re-centring and dissipating wires, double-flag-shap ed hysteresis loops are obtained. Moreover, depending on the number and pretensioning levels of the SMA wires of both goups, the device can exhibit a supplemental re-centring capacity, which can be exploited to counteract possible reacting forces external to the device, such as plastic forces in R/C members [Dolce et al., 20001. Figure 6 shows the arrangement of the braces in the structural model. In the longitudinal frames, the braces were arranged in echelon codiguration, to limit the variations of axial force in the columns. In order to avoid additional shear and

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Fig. 6 . Bracing arrangement.

Fig. 7. Experimental diagrams of the steel-based EDB's installed in the R/C model.

moment stresses in R/C members, the connection brackets were made in such a way that the axes of braces, beams and columns converged to the same point. Steel prestressed bars were also arranged along the beam ads,to generate a compression state that counteracts the tensile seismic stresses induced by the brace forces.

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Experimental Behauiour of

R/C

h m e s

371

The design of the steel hysteretic braces was carried out by following the procedure proposed in Braga and D'Anzi (19941, which optimises hysteretic or friction EDB's for seismic upgrading of R/C frames, to limit the ductility demands in R/C members. Starting from the characteristics of the frame, the method calibrates stiffness and strength distributions dong the height, using a linear optimisation process. The steel-based EDB's were preliminarily tested at the Laboratory of the University of Basilicata. Their mechanical behaviour was checked under repeated cyclic loading, at the amplitudes occurring in the structural model (Ck2-mm)~Differeit'fi3quFficiE--wWeazisiderd 7 0 . 2 T Hz)bXt T o LpiactiCil-iri-p fluence on the brace behaviour was detected. Figure 7 shows the experimental force-displacement relationships of the braces of each storey and each direction. In Table 3 the theoretical design mechanical properties of the EDB's are reported, while Fig. 8 shows the dimensions of the dissipating steel elements. The SMA-based SRB's were designed by assuming that the hysteretic behaviour of R/C members can be fully exploited to dissipate energy, conditioned upon the capability of the SRB'sto restore the frame to its undeformed shape at the end of the earthquake. To evaluate the minimum supplemental recentring forces of the Table 3. Theoretical mechanical properties of the EDB's.

X Direction Stiffness

Strength

Y Direction Stiffness

(kN/mm)

(kN)

(Wmm)

155.87 140.88

22.53

381.54

I1

253.68

22.69 20.78

111

75.09

163.66

10.25

IV

59.87

17.49 10.45 4.24

82.74

7.30

Storey

I

--

liv.

s

Dir. h

(m) (-1

*

X lb

1

.

,

.

.

Strength

(W

Dir.

£a

(nun) (mm)

thickness 5 mm -> yielding tension = 300 Nlmmq thickness 4 mm -ayielding tension = 350 N h m q thickness 3 mm --> yielding tension = 218 N / m q

Fig. 8. Dimensions of the energy dissipating steel element.

Y

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

04

(4

Fig. 9. Design procedure followed for SMA braces: (a) structural collapse mechanism considered to be assigned to each SMA brace, in the evaluation of the suppIementa1 re-centring capacity (Frs) (b) typicd mechanical behaviour of a R/C column under repeated cyclic deformations, (c) typical force displacement behaviour of a SMA brace.

SRB's {see Fig. 9(c)], reference was made to the worst condition that could take place during the tests, assuming the occurrence of plastic hinges at the ends of each column [see Fig. 9(a)]. Their ultimate moment was calculated by assuming the axial forces in columns produced by gravity loads only. The maximum storey shear was calculated and then reduced by about 5095, to take into account the actual shape of the hysteresis loops exhibited by R/C columns under repeated cyclic loading pounggon et al., 1992; Nakamura and Tanida, 1988; Pipa and Carvalho, 19901, as

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Experimental Behaviour of R/C Frames

373

shown in Fig. 9(b). The supplemental re-centring forces that resulted were equal to about 4.50 kN for each brace at the first and second storey, and 2.50 kN at the third and fourth storey. As a further design criterion, the reference forces of the SRB placed at different storeys were kept in the same ratio as those of the steel braces. This led to four different device configurations, differing in the number of §MA wire loops of the re-centring and dissipating groups, as shown in Fig. 10. The response of the structural model equipped with such devices was then analysed with a non linear finite element program [Prakash et al., 1993 and 19941, to check that the damage in the R/C members was acceptable. The same acceleration profile and seismic intensities selected for the shaking table tests were used. The ma.ximum interstorey drift was taken as the damage index, assuming 1.5% as the limit value. Before the shaking;table testing, all the SMA-based braces were tested at the Laboratory of the University of Basilicata, in order to check their cyclic mechanical behaviour. Figure 11 shows some typical force-displacement relationships relevant to the devices of each storey. As can be seen, in the examined displacement range, the force levels in the SR:B1sare lower than those of the steel braces. It is worthwhile to underline that the SFVIAwires used to realise such devices already underwent several hundreds of cycles at large strains, during previous tests in another research program [Dolce e t al., 2001c]. Nevertheless, none of them failed during the shaking table tests performed at LNEC, thus fully confirming the exceptional fatigue

;st

N. couples of ) : ; ; :N of re-centering wires Dissipating wires ($ lmm) 4% of deformation 4.5% of deformation 11 (44 wires) 6 (24 wires)

4

7 (28 wires )

4 (16 wires)

6 (24 wires )

3 (12 wires)

2 (8 wires )

1 (4 wires)

Fig. 10. Functional corifiguration of the SMA based SRD.

-

1" storev X direction

Fig. 11. Experimental diagrams of the SMA-based SRB's installed in the R/C model.

-

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

resistance under large strain cycles of SPvlA's. As only eight SRB's were available for the test, they were installed in the X-direction frames, while keeping the steel-based EDB's in the Y-direction frames. Due to the different design approaches, the comparison between SMA-based and steel-based braces cannot be just referred to the structural responses at the same seismic intensity, but also to the practical and economical consequences that the two Merent retrofitting strategies imply. The retrofitting design using EDB's is addressed to minimise, or better eliminate, plastic hinges in R/C structural members. This often calls for an extensive strengthening of beams and columns, especially when the R/C structure is designed without any seismic provision. On the contrary, a R/C structure upgraded with SRB's, even when totally hinged, recovers its undeformed shape at the end of the earthquake. As a consequence, larger ductility demands in structural members can be accepted. Therefore, the structure can be purposely exploited to dissipate energy, if a strong earthquake occurs. This, in turn, leads to two important advantages. The first is the lower stiffness - aad force levels in the-SRB's, resulting in significa& sayings in_thg_cogtof the bracing system and of the connections. The second is a great simplification of the seismic retrofitting, since R/C elements do not have to satisfy any flexural strength requirement. 4. Test Program and Sensor Set Up

Table 4 summarises the whole experimental program of shaking table tests. Nine different groups of tests have been carried out on two identical 1/4scde R/C frame models, with or without dissipating or re-centring braces. In Table 4 there are reported, for each group of tests: (i) the type of brace eventually installed, (ii) the arrangement of the additional masses, (iii) the characteristics of the input motion and (iv) the maximum peak table acceleration reached during the test. Each group of tests consisted of two alternate sets of tests, i.e. between one seismic test and another, the model was subjected to a low intensity (0.08g)pink noise signal, to evaluate the changes of its dynamic characteristics. Table 4. Test program. Test

Model 1 2

2 2 2 2 2 2 2

Device

-

Masses

Centred Steel-based Centred Steel-based Centred SMA-based Centred SMA-based (X), Eccentric Steel-based (Y) Steel-based Eccentric Eccentric SMA-based Centred Centred

Seismic Action

blax.

PGA

0.3229 Unidirectional (X) 0.9349 Unidirectional (X) Unidirectional (Y) 0.880g Unidirectional (X) 0.8819 Bidirectional (X Y ) 0.SOOg ( X ) , 0.8709 (Y)

+ Bidirectional (X+ Y ) Bidirectional (X+ Y ) Unidirectional (X) Unidirectional (X)

0.9199 (X), 0.9139 (Y) 0.3309 (X), 0.318g (Y) l.OlOg (X) 0.2959 (X)

-. .

0.3

Colfiorito N-S

:

4.3

o

s

10 -

--

I

!

(-1

15

20 -

I-)

4.3 Q

s

to

; 2o

15 -

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Fig. 12. Scaled acceleration-time histories used as input motion for the shaking table tests.

Besides the type of braces, two further parameters were considered in the experimental program: the position of the additional masses and the number of earthquake components acting at the same time. The aim was to study the effectiveness of the different types of braces in the simplest condition, i .e. unidirectional motion, mass and stiffness centres coincident (tests 1,2, 3 , 4 ) , and in the most complex condition, i.e. bidirectional motion, eccentric mass (tests 5, 6, 7). Test No. 7 was carried out to assess the behaviour of the bare frame with eccentric mass, to check its damage state after the previous very severe tests in the braced configuration and to take it to a near-collapse condition. Test No. 8 was aimed at verifying the effectiveness of the SMA-based braces in their re-centring function, when the hame joints have almost no moment resisting capacity. The acceleration profile recorded at Colfiorito (Italy) during the 1997 UmbriaMarche earthquake was assumed as the input motion of the table. The model scaling results in half the actual duration of the record [see Figs. 12(a) and (b)]. The peak ground acceleration was progressively increased, up to the operative limits of the table or to the collapse of the model. Thirty-six sensors were used to monitor the response of the non protected models, namely: (a) 16 Hamamatsu optical sensors, measuring absolute storey displacements (b) 8 wire-connected LVDT's , measuring interstorey drifts, (c) 12 Endevco piezoelectric post-amplified accelerometers, measuring horizontal floor accelerations (Fig. 13). Twenty-four further sensors were added to monitor the response of the braced models, namely: (d) 8 LVDT's, for the measurement of the relative displacements inside the device, and ( e ) 16 strain gauges, for the evaluation of strains, hence forces, in the device (Fig. 14). The instrumentation of the seismic platform is permanently incorporated and records vertical and horizontal table displacements, table accelerations, pitch, roll and twist accelerations. However, to better estimate the in-plane response of the table, two additional LVTD's and one accelerometer were used. This last was placed on the top of the table, nearby the model. Figure 15 shows a view of the three types of structures considered in the test program: non protected, equipped with steel-based EDB's, and equipped with SMA-based re-centring braces.

3'76

D. Cardone e t al.

South (+L)

...-

.\-

, , , , ,

West (-7

.

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

East (+T)

DISPL. AND ACCEL. AT EACH FLOOR

Pos

I), 11,2,3

Shaking table

Fig. 13. Instrumentation of the unbraced model.

.5. Experimental

In order to evaluate the response of the structure and its damage state at the different steps of the test program, the attention is focused on three.major aspects: (i) the evolution of the natural frequency of the structural model, as recorded in the characterisation tests and in the seismic tests, (ii) the evolution of the equivalent damping ratio and (iii) the maximum interstorey drifts, as recorded in the seismic tests. As a matter of fact, all these quantities are strictly related to the structural damage and are, therefore, useful to assess the effectiveness of the retrofit. When necessary, the records have been preliminarily filtered, to eliminate some electrical noise, and then elaborated and presented in diagrams.

Eqerimental Behaviour of

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

East

X-Frame NORTH (4)West

Y-Frame WEST (-T) south

R/C h m e s 377

Y-Frame EAST (+TI North

south

North

Fig. 14. Brace instrumentation.

To better analyse the experimental response of the structural model, it is useful t o consider the results of a pushover analysis carried out with the DRAIN-3DX program, by assuming a triangular distribution for the seismic forces and including gravity load effects. The actual experimentally evaluated mechanical properties of steel and concrete (concrete average compressive strength f,, = 29.5 MPa, steel yield stress f,, = 256 MPa, steel strain hardening ratio shrd = 0.135%) were

D. Cardone et ad.

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

378

Fig. 15. (a) Non protected structure, (b) structures equipped with steel-based EDB's, (c) structures equipped with SMA-based SRB's.

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Fig. 16. Pushover analysis on the non-protected model: base shear versus 1st interstorey drift

index.

Fig. 17. In plane Xdirection tests - natural frequency of the models versus PGA (MRF: non protected model, EDBsteel: model equipped with steel-based energy dissipating braces, SRBsma: model equipped with SMA-based supplemental recentring braces.

assumed. Figure 16 shows the base shear versus 1st storey drift curve in the X direction. When the first plastic hinge occurs, the interstorey drift is of the order of 0.5% and the force is about 4.00 kN (about 4% of W). The ultimate strength of the frame is about 7.80 kN (7.7% of the total weight W). The variation of the natural frequency of the bare frame (Model No. 1) and of the retrofitted hame with the two types of braces (Model No. 2, tests No. 2 and 4) are reported in Fig. 17, for the case of onedirectional action (X direction). The decay in the bare frame starts at the 0.099 PGA run, due to some cracks in the beams and joints of the first two storeys, caused by the slippage of the beam reinforcement at the joints. A similar behaviour was obtained in the tests carried out on the plane model tested for the ECOEST I project [D'Anzi et al., 19961. The

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

.

test on Model No. 1 attained 0.329 PGA, after which the model presented numerous deep cracks and some concrete crushing and spalling, prevalently at the ends of the columns and in the joints of the first storey. The evolution of the natural frequency of the model is in good accordance with the apparent damage observed directly on the model. At the end of the 0.329 test, the natural frequency that resulted was 58% of the initial one and, therefore, the stiffness was about 33% of the initial one. The results relevant to Model No. 2 equipped with steel-based and SMA-based braces show a great increase of frequency, and then of stiffness, due to the braces, and a much smaller decay, even for a much greater eearthquake intensity. In the case of the retrofit with steel-based braces, the natural frequency of the model at the end of the tests turned out to be 86% of the initial one. A significant part of this reduction occurred at seismic intensities (0.3-0.49) which were significantly less than the maximum one reached (0.93g), because of a defective connection between the hame and one steel bracket at the base of the model. In the case of the retrofit with SMA-baed braces, the model exhibits a natural -frequency-[about-8.-3--Hz) f-decidedly-greater-than that-measured..on.the .frame .with steel braces (about 7 Hz), in spite of the bigger cross section of the steel truss (27 cm2 versus 11 cm2). This occurred because the initial stiffness of the steel braces is strongly afTected by the flexibility of the dissipating components, while for the SMA braces it coincides with the axial stiffness of the steel tubes, due to wire pre-tensioning. At the end of the test sequence on the model with SMA braces, the natural frequency that resulted was just 92% of the initial one, although about 0.9g PGA was attained. In order to evaluate the change of dynamic properties of the building under strong action due to its non-linear behaviour, the results of the tests were also processed using the "Short Time Fourier Transform" (STFT) [Ewins, 1994, 1998; Gabor, 1946; Spina et al., 19961: STFT produces a signal representation both in the frequency and in the time domain, to observe the variation in time of the spectral properties relevant to the recorded accelerations (Fig. 18). The method uses the Fourier Transform for small sections of signal, using the "windowing" technique. In order to achieve well-fit results, a high number of points are necessary to pick the time-dependent frequency variations. At the same time a short temporal window is needed to get a good description of the variatiolls of the dynamic characteristics as a function of time. The right compromise was found to be a 5.12 sec windows (512 points), which slides 0.25 sec ahead at each step. Since the most important pieces of information are concentrated at the beginning of the considered portion of signal, a temporal filter was used to damp the points after the h t 0.75 sec, so that they quickly achieve negligible values. To this purpose, a temporal filter of the "generalised cosine window" type was implemented, where the "Winordinate, given "n" sampling steps, is expressed by the following relationship:

Experimental Behaviour of R / C Frames

381

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

window --.

L - 1

--"

I

Time

slide

Fig. 18. Shlort Time Fourier Transform - windowing technique.

Level n. 4

-

Fig. 19. Variations of the fundamental frequency of the bare frame model (Test no. 3 - 0.086 a/g PGA) computed with Short Time Fourier Transform.

The Short Time Fourier Transform was used to evaluate the frequency variations during the test time, and then the dependence of the natural frequency on the displacement amplitude. As an example, Fig. 19 shows the variations of the first frequency of the non retrofitted structure (Test No. 3 at 0.086 PGA) computed

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

382

D.

Cadone et al.

starting from the acceleration recorded during the test. This contour level plot was carried out considering the normalised value of the transfer function obtained from the table and the fourth level accelerations. Three frequencies vere considered for the characterisation of the non-linear behavior: the initial frequency, at low seismic level when cracks may close under self-weight, the minimum or apparent frequency, resulting from the STFT during the entire time-history,and the final frequency, at the end of the seismic application. This elaboration was extended to all in-plane seismic tests and configurations and compared with the hequencies obtained from the identification random tests carried after each seismic test. In Fig. 20, the frequencies evaluated with the above described procedure are compared for (a) the non-retrofitted model - X direction, (b) the model equipped with steel-based braces - X direction (c) the model equipped with steel-based braces - Y direction and (d) the model equipped with SMA-based braces - X direction. For the non retrofitted model the difference between the three curves of initial, minimum and final frequencies, shows that the non linear behavior of the -model-stas at -PGk great er-t han-0.19 and- a remarkable -damage .accumulation

2 -I . . . . : . . . . : . . . . . * . . . . :

0 -f

€ ~ % ; ~ t e eYl direction. -

. . - :

I

I

I

2:.

i

ol

..........................

EDB'SMA- X direction

(4 (4 Fig. 20. Frequency changes obtained from Short Time Fourier Transforms for (a) bare frame model - X direction; (b) model equipped with steel-based braces - X direction; (c) model equipped with steel-based braces - Y direction and (d) model equipped with SMA-based braces X direction.

-

-

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

Experimental Behawiour of R/C h m e s

Fig. 21. Equivalent damping: (a) .Peak-Picking Method, (b) Damping Variation as a function of PGA.

(difference between initial and final curves), is more evident for PGA greater than 0.159. In the case of the retrofitted model, all diagrams [20(b), (c) and (d)] show that its seismic behaviour is characterised by very low frequency values for high amplitude of vibration (PGA greater than 0.359) and a substantial coincidence of the initial and final frequency curves, resulting in a low damage accumulation. In these cases the evident non linear behavior of the model is mainly due to the activation of the dissipative units of the braces. All diagrams show a correspondence between the h a 1 frequency curve and the curve obtained from random tests. This confirms the reliability of the identification of damage through a random test after seismic tests. Besides the evolution of the natural frequency, the evolution of the equivalent damping ratio was considered important to evaluate the damage state. To estimate the equivalent damping, the peak-picking method [Ewins, 19941 was utilised. This method works adequately for structures that exhibit well-separated modes. For that reason the damping evaluation was carried out for the in-plane X direction tests only, to avoid the interference and superposition of the first modes in the two principal directions. The individual resonance peak corresponding to the first mode was detected and isolated with a FFT plot, computed through a window of 512 values in conespondence of the minimum d u e of frequency, as computed through the SFTF. The frequency of maximum response was taken as the natural frequency of that mode (w,) [Fig. 21(a)]. Then, the maximum value of the FFT was noted la'l and the frequency bandwidth of the function for a response level la'l / fi was determined

(Aw = w, - w b ) .The two points thus identified as (wb)and

(w,) are the half-power

points [Fig. 21(a)]. The structural damping of the mode can be estimated from the following fomulae: 0

rl

-0-

MRF Q 0.329

- *- EDB-steel Q 0.289

Downloaded by [Univ Studi Basilicata] at 02:52 15 October 2012

+

EDB-steel Q 0.939

-x-

SRB-sma O 0.379

-x-

SRB-sma @ 0.889

drifts at different PGA's for the non protected model (MRF),for the model equipped witkit&l