Integrated shape memory alloy devices toward a high

Contents lists available at ScienceDirect

Engineering Structures

journal homepage: www.elsevier.com/locate/engstruct

Integrated shape memory alloy devices toward a high-performance glazedcurtain wall seismic retrofitLorenzo Casagrandea,b,⁎, Jessica Sisinnia, Antonio Bonatib, Antonio Occhiuzzib,c,Ferdinando AuricchioaaUniversity of Pavia, Civil Engineering and Architecture Department, Via Ferrata 3, 27100 Pavia, ItalybNational Research Council (CNR), Institute for Construction Technologies (ITC), Viale Lombardia 49, 20098 San Giuliano Milanese, ItalycUniversity Parthenope, Department of Engineering, Centro Direzionale Isola C4, 80143 Naples, Italy

A R T I C L E I N F O

Keywords:Glazed curtain wallsNon-structural elementsTechnological improvementsSeismic retrofittingShape memory alloysPassive seismic isolation

A B S T R A C T

Recent dynamic events, such as Alaska and San Fernando earthquakes, have shown shortcomings in con-temporary façade design processes, exacerbated by the occurrence of non-structural component damage andfailure when subjected to extreme wind loads or seismic events. Accordingly, in the present work the feasibilityof an original and advanced dissipation technology is pursued investigating on the façade performance by testingand modeling strategies.

Initially, based on the experimental activities performed at the Construction Technologies Institute (ITC)laboratories of the Italian National Research Council (CNR), numerical simulations are run to interpret andreproduce the experimentally tested response of two full-scale façade units. Thereafter, sophisticated 3D finiteelement models are calibrated acquiring information on the fundamental mechanisms accountable for the dy-namics in façades.

Subsequently, by virtue of the peculiar properties of Shape Memory Alloys (SMA), such as superelasticity andshape memory effect, innovative curtain wall joints are designed, focusing on the improvement of the façadeenergy dissipation performance and on the enhancement of the overall structural behavior.

Finally, the improvements are demonstrated both globally and locally: on one hand, traditional and novelcurtain wall force-drift capacity curves are compared; on the other hand, the effectiveness of the suggesteddevice is shown, when built on a reference structure.

1. Introduction

Structural element failures, originated by historical seismic events,have led engineers toward the harmonization of performance levelswithin the context of the performance-based earthquake design [1,2].However, although progresses in the technical expertise have beenachieved regarding the stability and the resistance of the frame, re-search advancements on the seismic vulnerability of non-structuralcomponents lagged, hindered by lack in information [3]. In this regard,the role of secondary components in lowering the functionality hasbeen accentuated after the Alaska earthquake (March 27, 1964): sinceeconomic losses due to non-structural components considerably ex-ceeded losses related to earthquake-induced structural damage, thefocus has gradually moved toward non-structural systems [4–7], al-though limited to theoretical methods, principally involving nuclear

plant construction [8–11].From the early 2000s, modern computational techniques have al-

lowed the validation of structural analysis algorithms to design stra-tegic facilities [12–15]. Accordingly, Miranda et al. [16] have high-lighted that the reparation cost of non-structural elements located inoffices, hotels and hospitals represents, respectively, the 82%, 87% and92% of the total initial investment.

In this context, façades assume an important role, both for the initialcost and the seismic risk associated [16]. Moreover, external glazedsurfaces can partially affect the overall stiffness/strength of strategicbuildings [17], although neither Italian nor European Codes [18,19]evaluated this occurrence yet.

Therefore, the main objective of this research is to discuss en-hancements in façade energy dissipation capacity, when integrated adhoc Shape Memory Alloys (SMAs) devices are applied to façades of mid-

https://doi.org/10.1016/j.engstruct.2018.11.023Received 21 April 2018; Received in revised form 29 September 2018; Accepted 9 November 2018

⁎ Corresponding author at: University of Pavia, Civil Engineering and Architecture Department, Via Ferrata 3, 27100 Pavia, Italy.E-mail address: [email protected] (L. Casagrande).URLs: http://dicar.unipv.eu, http://www.itc.cnr.it (L. Casagrande).

Engineering Structures 179 (2019) 540–555

Available online 15 November 20180141-0296/ © 2018 Elsevier Ltd. All rights reserved.

T

http://www.sciencedirect.com/science/journal/01410296

https://www.elsevier.com/locate/engstruct

https://doi.org/10.1016/j.engstruct.2018.11.023


mailto:[email protected]

http://dicar.unipv.eu

http://www.itc.cnr.it


http://crossmark.crossref.org/dialog/?doi=10.1016/j.engstruct.2018.11.023&domain=pdf

and high-rise structures.Initially, we investigate the seismic response of glazed curtain wall

stick systems (described in Fig. 1), both empirically (Section 2.1),through two full-scale experimental results, and numerically (Section2.2), developing 3D finite element (FE) models using the softwareAbaqus [20]. Subsequently, a sensitivity analysis campaign is proposedto calibrate modeling parameters responsible for the dynamic response

of façades (Section 3). Afterward, proposed SMA devices are described,highlighting monitoring and force–displacement control functions[21–23] (Section 4) and underlining local and global effects on theseismic performance, when applied in mid- and high-rise structure fa-çades (Section 5).

2. Material and methods

2.1. Reference experimental tests

According to American and Japanese Standards [24–26], the re-sponse of two full-scale façades, designated as Façade A and Façade B,is investigated. Quasi-static force control in-plane tests, performed inagreement with [27], are achieved pushing the intermediate façadebeam by an actuator, toward 1.6% as inter-storey drift demand, andmeasuring the forces by a 50 kN LeBow load cell.

The specimen characteristics (Table 1) and experimental results(Fig. 2) have already been presented in [28] and are briefly summarizedin the following.

Fig. 1. Curtain wall stick system: (a) Components definition, (b) T-Joint, (c) U-Joint.

Table 1Façade specimen description, from [28] [Dimensions in mm].

Façade A Façade B

Designed geometry High 7300 7200Wide 5650 5600Inter-storey 3100 3300

Connections Transom-to-mullion T-joint U-jointGlass-to-frame Silicone gaskets

Materials Alluminium EN-AW 6060-T6, E= 69 GPaTempered Glass E=70 GPa

Fig. 2. Force-drift cyclic results: Façade A (a) and Façade B (b), from [28].

L. Casagrande et al. Engineering Structures 179 (2019) 540–555

541

2.2. Finite element façade modeling

In this section, we suggest an exhaustive approach to develop reli-able and efficient three-dimensional FE models: in particular, accordingto previous experimental studies [27–30], the main components re-sponsible for the in-plane response are connections. Thereafter, weadopt the FE software ABAQUS [20] proceeding toward a preciseanalysis framework by employing ad hoc meshes in key members.

In order to investigate the local behavior of connections, we

recreate 3D brick-based models of transom-to-mullion (Fig. 3(a)) andglass-to-frame (Fig. 3(b)) joints (Section 2.2.1), adopting 8-node linearbricks (C3D8R) in an isoparametric framework. Geometrical and ma-terial non-linearities are accounted, as finite strain and rotation inlarge-displacement. The aluminium stress-strain relationship is mod-eled through the rate-independent Von Mises yielding principle formetal plasticity, while the glass is considered elastic. Moreover, thefollowed solution technique is Full Newton with iterative equationsolver. Finally, local contributions due to connections are included in a

Fig. 3. Façade elements: mullion and transom (a), mullion, gaskets and glass panel (b).

Fig. 4. Façade A - Three-dimensional model in transom-to-mullion connection: transom and mullion members (a), T-joint location (b), mesh (c).

Fig. 5. Façade B - Three-dimensional model in transom-to-mullion connection: transom and mullion members (a), U-joint location (b), mesh (c).


542

more global finite element scheme to reproduce full-size façade tests(Section 2.2.2). Although experimental tests are undertaken in cyclicfashion, this stage is conceived to calibrate the model, deciding to de-termine the façade peak response purely through a pushover loadingprotocol.

2.2.1. Connection FE modelingIn Figs. 4 and 5, transom-to-mullion steel connectors, respectively

named T-joint (in Façade A) and U-joint (in Façade B), are depicted inblue. In these elements, a higher mesh density is maintained through amaximum size control equal to 1% of global size, compared to 10% for

Fig. 6. Transom-to-mullion force-displacement curves: Façade A (a) and Façade B (b).

Fig. 7. Three-dimensional model of glass-to-frame connection: members (a), mesh (b).

Fig. 8. Glass-to-frame force-displacement curves: Façade A (a) and Façade B (b).

Table 2Calibration parameters of Façade A and Façade B modeling.

FAÇADE A FAÇADE B

Transom-to-mullion joint Fig. 6(a) Fig. 6(b)No. of transom-to-mullion connections 30 54No. of glass-to-frame connections 288 552Local contact stiffness 10 kN/mm 10 kN/mmGlass-to-frame gap 3mm 5mmGasket elastic stiffness 0.24 kN/mm 4.2 kN/mmGasket permanent deformation force 0.20 kN 0.58 kN


543

other members.In order to obtain the transom-to-mullion rotational stiffness, a

vertical displacement is applied at the end of the transom, reading thebending moment. As a result, transom-to-mullion stiffness curves arereported in Fig. 6: (1)-(2) are determined through Eurocode 3 [31]formula, and define the edges for Rigid, Semi-Rigid and NominallyPinned sub-regions; (3) is the upper boundary obtained assuming arigid joint on 3D FE models; (4) is the curve achieved considering T-joint and U-joint connectors (Figs. 4 and 5) on the same FE framework.Therefore, T-joint systems can be categorized as a Semi-Rigid connec-tion (since T-joint FEA results belong to the Semi-Rigid area), while U-joint systems can be classified as a Nominally-Pinned joint (since U-joint FEA results belong to the Nominally Pinned area).

In Fig. 7 glass-to-frame connection modeling is depicted: an hor-izontal displacement law is applied to the glass panel in order to recordforces in gaskets and characterize the glass-to-gasket force-displace-ment curves (Fig. 8). We use FEM parametric campaigns to validate themodeled interface effects and gasket mechanical deformations. Fur-thermore, we calibrate an equivalent elasto-plastic stress-strain con-stitutive law through the experimental investigations carried out at theInstitute of Construction Technology and in accordance with

[29,32–35]. According to Memari et al. [32], we model multilayeredglass panels through an equivalent full elastic section compatible withthe Enhanced Effective Thickness (EET) method, cited in [33].

2.2.2. Full-size test FE modeling: Façade A and Façade BToward a computational time-saving procedure, we assume 1D non-

linear link connectors [20] in order to replicate the force-displacementresponse achieved in transom-to-mullion and glass-to-frame (Section2.2.1) simulations (Figs. 6–8). Subsequently, equivalent connectors areemployed in shell-based (S4R) façade models, reducing the FE for-mulation and reproducing tests.

Corroborated by comparing input parameters in Table 2, analysesshow the correlation among connectors and FEA results: modifyingconnector input values, the lateral response of different curtain walltypologies can be reliably predicted. We want to underline that gasketelastic stiffness values are strictly correlated to in-opera clamping andglass-to-frame contact is modeled assuming high local stiffness values inorder to simulate the rigid interaction. Accordingly, we perform sen-sitivity analyses on the parameters, summarized in Section 3.

Fig. 9 depict the fine match between test results and FEA simula-tions. Moreover, numerical simulations confirm the expected stress

Fig. 9. Comparing FEA and experimental results: Façade A (a) and Façade B (b).

Fig. 10. Glass panel Mises stress distribution (MPa): Façade A (a) and Façade B (b).


544

Fig. 11. Sensitivity analysis: FAÇADE A (a) gasket elastic stiffness, (b) gasket plastic regime, (c) glass-to-frame gap, (d) glass-to-frame contact. FAÇADE B (e) gasketelastic stiffness, (f) gasket plastic regime, (g) glass-to-frame gap, (h) glass-to-frame contact.


545

distribution in glazed elements: corner glass are the most loaded, asthey first come in contact with the frame. Fig. 10 depicts a commonstress distribution due to glass-frame contact, predicting the typicaldiagonal glass crushing.

3. Sensitivity analysis campaign

Employing models described in Section 2.2, we conduct a numericalstep-by-step parametric campaign (Fig. 11) analyzing the sensitivity offaçade response to the parameters involved [30], i.e. varying, (i) thegasket elastic stiffness (Fig. 11(a) & (e)); (ii) the gasket permanentdeformation force regime (Fig. 11(b) & (f)); (iii) the glass-to-frame gap(Fig. 11(c) & (g)); (iv) the glass-to-frame contact stiffness (Fig. 11(d) &(h)). The transom-to-mullion rotational stiffness is considered fixed bygeometry, therefore, transom-to-mullion physical parameters are at-tributed by means of representative plastic curves (Fig. 6). Accordingly,

Fig. 12. Shape Memory Alloy device scheme (Façade B).

Fig. 13. Model of Shape Memory Alloy device application: Façade A (a) and Façade B (b).

Table 3Description and values of Shape Memory Alloy material and loading para-meters.

Iinput Description Value

EA Austenite elasticity 50,000 [MPa]EM Martensite elasticity 45,000 [MPa]

A M, Austenite/Martensite Poisson’s Ratio 0.33L Transformation strain 0.05

LS Start of transformation loading 380 [MPa]

LE End of transformation loading 490 [MPa]

US Start of transformation unloading 220 [MPa]

UE End of transformation unloading 120 [MPa]

Fig. 14. Resisting peak force increment at failure, after SMA device application: Façade A (a), Façade B (b).


546

we move forward the calibration campaign fixing, in each step, pre-vious adopted parameters and proceeding toward the definition of theexperimental curves. As a result:

(i) the gasket elastic stiffness influences the slope of the initial branchin the overall response (Fig. 11(a) & (e))

(ii) the gasket permanent deformation forces provoke the shift to non-linear plastic branch, causing the trigger of glazed panel sliding(Fig. 11(b) & (f))

(iii) before the gap closure occurs, glazed panels move as rigid bodiesallowing the in-plane façade drift; when the gap is closed, glass-to-frame contact provokes a stiffening increment [32,36]. Amplifyingthe clearance depth, the stiffness variation can be delayed(Fig. 11(c) & (g)).

(iv) fixing a gap value, the glass-to-frame impact stiffness influencesthe slope of the last branch (Fig. 11(d) & (h)).

In both façades, the lateral response is strongly influenced by glass-

Fig. 15. Correlations between the SMA device thickness and the first local failure resistance increment, achieved in: Façade A - glass panels (a), aluminum frame (b);Façade B - glass panels (c), aluminum frame (d).

Fig. 16. Performance improvement adopting SMA devices: Façade A (a) and Façade B (b).


547

to-frame connections: even if transom-to-mullion stiffness in Façade Ais considerably greater (Fig. 6), the global response assumes a morerigid initial branch in Façade B, due to the different gasket propertiesand clamping along the glass panels. To reproduce these results, pleaserefer to model assumptions highlighted in Section 2.2.

4. Calculation: smart-material integrated systems

Since 1960s, when Buchler and Wiley firstly developed NiTi alloys[37], the interest in termo-mechanical properties of shape memory al-loys (SMAs) has rapidly grown, reaching a compound annual growthrate of 12.8% in 2011–2016 [21]. SMAs represent an excellent candi-date in smart-material systems due to their peculiar properties: severelydeformed samples may recover the initial shape after a thermal cycle(shape-memory effect) or during mechanical loading-unloading cyclesexecuted at constant temperature (superelasticity) [37], as well as dis-tinctive energy dissipation and resistance to corrosion features.

Recently, the research has been extended to structural applications,aiming to reduce earthquake input by adding actuation and vibrationcontrol to structures [22,23]. Accordingly, the main goal of this work isthe original design of a new concept of passive devices, i.e., integratingSMA-based thin layers in non-structural element systems (Fig. 12) toprevent damages in primary and secondary components, when sub-jected to moderate seismic demands. In particular, the proposed glazedcurtain wall application targets a dual need: on one hand, minimize thecost/effect ratio due to the moderate request in NiTi alloy; on the otherhand, allow glass recentering capabilities, facing the limit state of vi-bration in serviceability conditions.

Moreover, as highlighted in Section 3, glass-to-frame connectionssignificantly affect the overall façade response: by virtue of specificSMA properties, such as shape memory effect, high temperature re-sistance, large elastic strain and stress (Fig. 13), this cutting-edge con-figuration is developed to improve the global façade damping. Table 3summarizes SMA attributes chosen to define the constitutive model.Through user defined material models (UMAT) [20], we perform aniterative analysis campaign in order to obtain the optimum devicegeometry, i.e., balancing on one hand, the dissipation increment, on the

Fig. 17. Case-study buildings: MF-01, thirty-storey frame (a), MF-02, sixty-storey frame (b), plan and section geometries (c).

Fig. 18. Equivalent 1D façade modeling in the OpenSEES environment.

Fig. 19. Static-to-seismic axial load ratio in braces and rightmost core columns, MF-01 (left) and MF-02 (right).


548

other hand, the stress rise on aluminum and glazed elements. There-after, we develop 3D brick-based glass-to-frame FE models explicitlyaccounting for SMA devices, extrapolating new glass-to-frame non-linear curves and the improvements in global façade response. Fig. 13shows achieved SMA device peak stresses.

4.1. Geometry definition of SMA devices

Initially, increments of resisting peak force at failure are calculated,comparing global resisting forces of traditional and SMA-integratedfaçades (Fig. 14). Due to technical details of mullion-to-transom joints,the height and the width of SMA appliances are fixed by aluminummember geometries. Accordingly, we may vary the SMA device thick-ness to influence façade response: the interpolation line shows an ex-ponential capacity increase for both specimens. However, the asso-ciated stress rises likewise. As a result: Façade A capacity increments by52% for 0.6mm, representing 30.4 kN at 50mm; Façade B capacityincrements by 93% for 0.7 mm, representing 48.3 kN at 38mm. Sub-sequently, Fig. 15 explicitly correlates the SMA device thickness (inmm) to the first local failure resistance increment, achieved either inglass or in aluminum members. Specifically: the maximum peak force inglass elements increments by 70% for 0.6 mm and 94% for 0.7mm,

respectively in Façade A and B; the maximum peak force in aluminummembers increments by 81% for 0.6mm and 71% for a device thickness0.7 mm, respectively in Façade A and B.

In conclusion, Fig. 16 shows the comparison between two curves:the black one is obtained in Section 2.2.2; the red one, is obtained in-tegrating SMA devices, selecting a thickness value of 0.4 and 0.6mmwith global resisting peak force increment at failure equal to 24.4% and63.4%, for façade A and B.

4.2. Dissipation performance on two reference buildings

This research addresses the dissipation properties of Façade Bequipped with SMA devices, toward its application in mid- and high-rise structures, to highlight local and global effects on structural seismicperformance.

We dedicate this section to examine two reference thirty- and sixty-storey buildings, respectively named MF-01 (Fig. 17(a)) and MF-02(Fig. 17(b)). In brief, we only indicate the outline of the frame con-ception, referring the numerical modeling and the design backgroundto Brunesi et al. [38]. Internally, lateral force resisting systems (LFRSs)are composed by a 11.2 × 11.2m concentric braced frame (CBF) core,connected with longitudinal and transversal outriggers to reduce inter-storey drifts and second order effects. Externally, belt trusses envelopthe structure enhancing the lateral stability (Fig. 17(c)).

We design frames in accordance with European codes [19,31], se-lecting ten natural records [39] on high seismicity area (PGA=0.40 g), soil class C (i.e., 180m/s Vs 360m/s), behavior factor(q) for V-bracing systems and medium ductility class (DCM).

Dead and live loads are assumed equal to 2 kN/m2 and 4 kN/m2,respectively, while the overcrowding weight contribution is con-servatively selected as 60% of live loads. Additionally, the horizontalwind pressure is calculated conforming to ASCE-7 05 [40], with a basespeed equal to 37m/s (84 mph). Consequently, we perform a series ofresponse spectrum analyses (RSAs) to achieve the first-stage design andnonlinear time history analyses (NLTHAs) for the definitive. HD taperedcolumn profiles, welded gusset plates, bolted beam-to-column connec-tions, HD belt trusses and HSS brace sections are designed to comply toserviceability limit states, i.e., inter-story drifts and displacementthresholds.

4.2.1. Structural and non-structural element couplingWe subject thirty- (Fig. 17(a)) and sixty-storey (Fig. 17(b)) moment

resisting frames (MRFs) to a set of ten natural records, scaled by Maleyet al. [39] and spectrum-compatible in displacement in accordance to

Table 5NLTHa Avg peak values in Façade B and Façade B/SMA, along with the relatedattenuation: outriggers.

Local Performance Façade B FaÇade B/SMA

30-storey 60-storey 30-storey 60-storey°15 Outrigger axial force 4968 kN 5200 kN 4239 kN 5107 kN°30 Outrigger axial force 2955 kN 5133 kN 2270 kN 4305 kN°15 Outrigger axial F. attenuation – – 14.67% 1.79%°30 Outrigger axial F. attenuation – – 23.18% 16.13%

Fig. 20. Peak force response of the most stressed first floor façade link, before and after SMA device application: Façade B, 60-storey (a) and Façade B/SMA, 60-storey(b).

Table 4NLTHa Avg peak values in Façade B and Façade B/SMA, along with the relatedattenuation: braces.

Local performance Façade B Façade B/SMA

30-storey 60-storey 30-storey 60-storeyRight brace axial force 4968 kN 5147 kN 4239 kN 4911 kNRight Brace Axial F. Attenuation – – 14.67% 4.59 %


549

Fig. 22. FAÇADE B – Axial force peak profile in the rightmost core column & compression variation, MF-01 (left) and MF-02 (right).

Fig. 21. FACADE B - Axial force profile in reference braces & compression variation, MF-01 (left) and MF-02 (right).


550

EC8 prescriptions [19], choosing a Type 1 spectrum with PGA=0.4 g,soil Type C and TD=8 s. A fiber-based framework is considered, per-forming NLTHAs with the Open System for Earthquake EngineeringSimulation (OpenSees) [41]. Accordingly, we consider to model theFaçade B as an equivalent 1D fiber-based zero-length link, as in Fig. 18,consistent with the OpenSEES environment. In particular, we decide toreproduce separately the mechanical phenomena emphasized duringsensitivity analyses, in Section 3, i.e.: (i) the gasket elastic stiffness; (ii)the gasket plastic regime; (iii) the glass-to-frame gap; (iv) the glass-to-

frame contact [30]. Consequently:

1. viscous GASKET links, account for the initial and the yielding branchin global façade response (Fig. 11(a)–(b)–(e)–(f));

2. GAP links, govern the slope change in the response (Fig. 11(c) and(d));

3. IMPACT links, act on the last branch of diagrams (Fig. 11(a)–(h));4. Bouc-Wen FRAME links [42], affect the overall response (Fig. 6);

Fig. 23. FAÇADE B – Axial force profile in outrigger braces & compression variation, MF-01 (left) and MF-02 (right).


551

5. Results and discussions

To investigate the response of the two reference structures, weconduct new NLTHAs considering, on one hand MRFs equipped withFaçade B, on the other hand, the change in MRF dynamics when SMAcomponents are integrated on Façade B. As a result, global and localperformance will be presented. In particular, MRF response to in-dividual earthquake (”MRF EQs”, in diagrams), NLTHA averages ob-tained considering Façade B (”TRAD. Façade”, in diagrams), andNLTHA averages referred to Façade B with SMA devices (”SMADEVICE”, in diagrams) will be shown. Afterwards, the percentagevariation between ”TRAD. Façade” and ”SMA DEVICE” results will besummarized. Specifically, positive variations (”Attenuation”, in dia-grams), mean that ”SMA DEVICE” results are lower than ”TRAD.Façade” values, emphasizing the effect of SMA devices in affecting bothglobal and local response.

5.1. Local performance

Outcomes exhibit that, if accurately designed, the proposed struc-tural system brings an attractive balance between strength and stiffness.In this regard, Fig. 19 displays the earthquake-induced axial overloadexperienced by critical components of lateral resisting systems (LRSs),i.e., emphasizing the importance of braces and core columns in ab-sorbing seismic actions.

Substantially, as reported by [38], the in-plane rotation of the MRFleads to a seismic-induced overburden in the leftmost and rightmostcore columns, with respect to the central one that persists relativelyunaffected by the dynamic excitation. Thereafter, as suggested by thestructural symmetry, results will be displayed only in rightmost bracesand columns. Besides, brace force peaks occur in outriggered floors:essentially, outriggers contribute to the LRS transmitting the seismicoverloads, from stiffened floors to core columns.

This is appreciable comparing Tables 4 and 5: maximum axial forces

in braces arise in correspondence to the 15th storey, both for MF-01 andMF-02 prototypes (496 kN for Façade B and 4239 kN for Façade B/SMA,Table 4 and Fig. 21). Compression forces are directly absorbed fromadjacent outriggers (15° Outrigger Axial Force, Table 5 and Fig. 23). Asa result, only brace, outrigger span and core column results will besummarized hereafter. In particular, brace and outrigger axial forcesare provided in Figs. 21–23, while column axial forces, bending mo-ments and shear peak profiles are introduced in Figs. 22, 24 and 25.Additionally, values are detailed for 30- and 60-storey buildings, withthe related percentage variation between NLTHAs Avg in Façade B andFaçade B-SMA. The planar rotation diminishes proportional to thestructural height increment, reflecting a compressive force attenuationalong the involved columns. Locally, the beneficial effect of SMA de-vices is clearly distinguishable in braces and outriggers (Table 4 and 5),although it is even more evident in columns where the mismatch levelbetween ”TRAD. Façade” and ”SMA DEVICE” is maximum (Table 6).Since outriggers induce a sudden discrepancy in column lateral stiff-ness, prominent stress discontinuities arise in columns.

5.2. Global performance

Fig. 26 displays the overall structural response in terms of flooracceleration and inter-storey drifts. Data are depicted for MF-01 (leftcolumn) and MF-02 (right column), respectively, while NLTHa Avgpeak values are summarized in Table 7. The acceleration curve adopt acoarse steady shape, while a remarkable inter-storey drift reduction isevident every 15 storey due to stiffening repercussions induced byoutriggers. As stated in Fig. 26, SMA devices in Façade B generallydecrease the structural acceleration (although one of the peak inTable 7 is increasing), while the global inter-storey drift experienced byfaçades slightly grows. In this case, since negative percentages inTable 7 reflect SMA DEVICE greater values, we report ”variation” in-stead of ”attenuation”. However, focusing on the local inter-storey drift(directly experienced by façades, Fig. 20), the lateral sway seems to

Fig. 24. FAÇADE B – Bending moment peak profile in the rightmost core column & percentage variation, MF-01 (left) and MF-02 (right).


552

decrease after the SMA device integration.

6. Conclusions

The survey presented proposes an extensive analysis oriented toassess the seismic response of façades and further suggest the design ofpioneering devices to enhance façade dynamic capabilities. In order toefficiently improve the non-structural dissipation performance, weemployed Shape Memory Alloys (SMAs) taking advantage from super-elasticity and shape memory peculiar properties. Therefore, we eval-uated the effect of transom-to-mullion and glass-to-frame joints infaçade response through dedicated FE strategies, addressing the mainmechanisms responsible for the response. As a result:

• the gasket elastic stiffness influences the slope of the initial branch;• gasket permanent deformation forces trigger the plastic branch and

the glass panel sliding;• when glass-to-frame contact occurs, a severe stiffening increment is

experienced, affecting the slope of the last branch.

Consequently, the match between experimental and numerical re-sults promises adequate prediction capabilities to simulate the SMA-based device response. Hence, we performed numerical FE campaignstoward the optimum design of SMA appliances, balancing the globalcapacity increment against the local stress growth. Accordingly, theintegrated device:

• produces results which suggest to consider a potential increase incurtain wall strength and energy dissipation;

• may leads to reduce stresses on glass panels and aluminum frame, ifadopting correctly designed miniaturized SMA devices;

• provide recentering capabilities to glass panels, due to SMA prop-erties.

Additionally, the above discussion induced us to evaluate to whatextent the novel dissipation properties of façades alter the dynamics ofmid- and high-rise structures. Therefore, we investigated local andglobal structural effectiveness of SMA devices through nonlinear dy-namic time history analyses, performed on fiber-based MomentResisting Frame models. Crucial results are outlined:

• locally, stresses achieved in key structural members (braces, out-riggers and columns) are significantly higher with traditional façadethan those stored equipping SMA-based curtain walls.

• globally, the device mainly reduce structural accelerations;

In detail, we want emphasize the effective weight of outer columnsin demand attenuation, underlining their leading role in earthquake-induced overload support. Further studies under assessment, pursuantwith this results, are:

• test façade dynamics when SMA devices are integrated;• inspect changes in device performance varying the temperature,

fatigue models and SMA parameters for this specific application;

Fig. 25. FAÇADE B – Shear force peak profile in the rightmost core column & percentage variation, MF-01 (left) and MF-02 (right).

Table 6NLTHa Avg peak values in Façade B and Façade B/SMA, along with the relatedattenuation: columns.

Local performance FaÇade B FaÇade B/SMA

30-storey 60-storey 30-storey 60-storeyColumn axial force 19886 kN 28162 kN 18691 kN 27800 kNColumn bending moment 427 kN m 1150 kN m 378 kN m 1048 kN mColumn shear force 606 kN 594 kN 221 kN 502 kNColumn axial F. attenuation – – 6.01% 1.29%Column bending M.

attenuation– – 11.48% 8.87%

Column shear F. attenuation – – 63.53% 15.49%


553

Fig. 26. FAÇADE B – Acceleration profile & Inter-storey drift with related percentage variations, MF-01 (left) and MF-02 (right).


554

• improve the device toward the glass fracture prevention when thestructure is subjected to dynamic loads in serviceability state;

• given the potential demand attenuation in structural members, in-vestigate the cost-saving balance optimizing structural and SMAmaterial price.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, in theonline version, at https://doi.org/10.1016/j.engstruct.2018.11.023.

References

[1] Welch DP, Sullivan TJ, Calvi GM. Developing direct displacement-based proceduresfor simplified loss assessment in performance-based earthquake engineering. JEarthq Eng 2014;18:290–322.

[2] Xue Q, Chen C. Performance-based seismic design of structures: a direct displace-ment-based approach. Eng Str 2003;25:1803–13.

[3] Filiatrault A, Sullivan T. Performance-based seismic design of nonstructuralbuilding components: The next frontier of earthquake engineering. Earthq Eng EngVib 2014;13:17–46.

[4] Ayres JM, Sun TY. Nonstructural damage to buildings, in the great Alaska earth-quake of 1964. Eng Div Earth Sci Nat Res Counc Nat Ac Sci Washington1973;1(B):346–456.

[5] Ayres JM, Sun TY. Nonstructural Damage, The San Fernando California Earthquakeof February 9, 1971. US Dep of Commun Nat Ocean Atm Adm 1973;1(B):736–42.

[6] Yuceoglu U. Seismic design of equipment, supports and connections in industrialinstallation, Fritz Eng Lab Report No. 424.3; 1978.

[7] Biggs JM, Roesset JM. Seismic analysis of equipment mounted on a massivestructure. Seism Des Nuc Plants 1970:319–43.

[8] Chen Y, Soong TT. State-of-the-art review: seismic response of secondary systems.Eng Struct 1988;10:218–28.

[9] Chowdhury AH. Seismic qualification of interior mechanical systems. Proc ThirdUSA Nat Conference on Earthq Eng 1986;11:1613–23.

[10] Northridge Earthquake, January 17, 1994: A Special Report by Dames & Moore,Dames & Moore; 1994.

[11] Villaverde R. Seismic design of secondary structures: state of the art. J Struct Engl997;123(8):1011–9.

[12] Singh MP, Moreschi LM, Suarez LE, Matheu EE. Seismic design forces. I: Rigidnonstructural components. J Struct Eng 2005;132(10):1524–32.

[13] Singh MP, Moreschi LM, Suarez LE, Matheu EE. Seismic design forces. II: Flexiblenonstructural components. J Struct Eng 2005;132(10):1533–42.

[14] Miranda E, Taghavi S. Approximate floor acceleration demands in multistorybuildings. I: Formulation. J Struct Eng 2005;131(2):203–11.

[15] Taghavi S, Miranda E. Approximate floor acceleration demands in multistorybuildings. II: Applications. J Struct Eng 2005;131(2):212–20.

[16] Miranda E, Mosqueda G, Retamales R, Pekcan G. Performance of nonstructuralcomponents during the 27 February 2010 Chile Earthquake. Earthq Spec

2012;28(1):453–71.[17] Casagrande L, Bonati A, Auricchio F, Occhiuzzi A. Dynamic high-rise moment re-

sisting frame dissipation performances adopting glazed curtain walls with super-elastic shape memory alloy joints. Int J Civ Environ Struct Constr Arch Eng2017;11(4):412–20.

[18] NTC08[2008] Norme tecniche per le costruzioni, Ministero delle Infrastrutture e deiTrasposti, Decreto Ministeriale del 14 gennaio; 2008.

[19] EN 1998-1[2005], Eurocode 8 (EC8) Design of structures for earthquake resistancePart 1: General rules, seismic actions and rules for buildings. UNI EN 1998-1.Directive 2004/18/EC.

[20] ABAQUS 6.14. Documentation, Dassault Systmes Simulia Corp. Providence, RI,USA; 2016.

[21] Menna C, Auricchio F, Asprone D. Applications of shape memory alloys in structuralengineering. Shape Memory Alloy Eng 2015;13:369–401.

[22] Attanasi G, Auricchio F, Fenves GL. Feasibility assessment of an innovative isolationbearing system with shape memory alloys. J Earthq Eng 2009;13:18–39.

[23] Auricchio F, Fugazza D, Desroches R. Earthquake performance of steel frames withnitinol braces. J Earthq Eng 2006;10(1):45–66.

[24] Building Seismic Safety Council, NEHRP Recommended provisions for seismicregulations for new buildings and other structures (FEMA 450), National Institute ofBuilding Sciences, BSSC; 2004.

[25] Architectural Institute of Japan Japanese Architectural Standard Specification JASS14: Curtain Wall, Japanese Standard Association, AIJ; 1996.

[26] American Architectural Manufacturers Association (AAMA), RecommendedDynamic Test Method for Determining the Seismic Drift Causing Glass Fallout froma Wall System, Pub. No. AAMA 501.6-09; 2009.

[27] Pantelides CP, Truman KZ, Behr RA, Belarbi A. Development of a loading history forseismic testing of architectural glass in a shop-front wall system. Eng Struct1996;18(12):917–35.

[28] Caterino N, Del Zoppo M, Maddaloni G, Bonati A, Cavanna G, Occhiuzzi A. Seismicassessment and finite element modelling of glazed curtain walls. Struct Eng Mech2017;61(1):77–90.

[29] Bouwkamp JG. Behavior of window panels under in-plane forces. B Seismol Soc Am1961;51(1):85–109.

[30] Casagrande L, Bonati A, Auricchio F, Occhiuzzi A. Dissipating effect of glazedcurtain wall stick system installed on high-rise mega-braced frame-core buildingsunder nonlinear seismic excitation. Proc. COMPDYN 2017, doi: 10.7712/120117.5677.17166.

[31] Eurocode 3 (EC3): Design of steel structures. General rules and rules for buildings,UNI ENV 1993-1-1; 1992.

[32] Memari AM, Shirazi A, Kremer PA. Static finite element analysis of architecturalglass curtain walls under in-plane loads and corresponding full-scale test. Struct EngMech 2007;25(4):365–82.

[33] Consiglio Nazionale delle Ricerche (CNR), Istruzioni per la Progettazione,l’Esecuzione ed il Controllo di Costruzioni con Elementi Strutturali di Vetro, CNR-DT 210/2013.

[34] Lu W, Huang B, Mosalam KM, Chen S. Experimental evaluation of a glass curtainwall of a tall building. Earthq Eng Struc Dyn 2016;45:1185–205.

[35] Behr RA, Belarbi A, Culp JH. Dynamic racking tests of curtain wall glass elementswith in-plane and out-of-plane motions. Earthq Eng Struct 1995;24(1):1–14.

[36] Sivanerupan S, Wilson JL, Gad EF, Lam NTK. Seismic Assessment of Glazed FaçadeSystems. Proc AEES; 2009.

[37] Auricchio F, Taylor L, Lubliner J. Shape-memory alloys: macromodelling and nu-merical simulations of the superelastic behavior. Comput Methods Appl Mech Eng1997;146:281–312.

[38] Brunesi E, Nascimbene R, Casagrande L. Seismic analysis of high-rise mega-bracedframe-core buildings. Eng Struct 2016;115:1–17.

[39] Maley TJ, Roldan R, Lago A, Sullivan TJ. Effects of response spectrum shape on theresponse of steel frame and frame-wall structures. IUSS Press. Pavia (Italy); 2012.

[40] ASCE 7-05, Minimum design loads for buildings and other structures, Reston (VA):American Society of Civil Engineers; 2006.

[41] Mazzoni S, McKenna F, Scott MH, Fenves GL, Jeremic B. Open System forEarthquake Engineering Simulation (OpenSees) 2003 Command Language Manual.

[42] Bouc R. Mathematical model for hysteresis. Report to the Centre de RecherchesPhysiques; 1971. p. 16–25.

Table 7NLTHa Avg peak values in Façade B and Façade B/SMA, along with the relatedattenuation: global performance.

Global performance FaÇade B FaÇade B/SMA

30-storey 60-storey 30-storey 60-storeyAcceleration 6.79 m/s2 4.38 m/s2 6.25 m/s2 4.99 m/s2

Inter-storey drift 0.59% 0.57% 0.58% 0.58%Acceleration variation – – 7.95% −13.93%Inter-storey drift variation – – 1.69% −1.75%


555


http://refhub.elsevier.com/S0141-0296(18)31264-1/h0005




























































Documents

Integrated shape memory alloy devices toward a high