Articles Giandomenico Toniolo* Antonella Colombo

DOI: 10.1002/suco.201100052

Precast concrete structures: the lessons learned from the LAquila earthquake
This paper examines the problems of the seismic design of precast concrete structures as indicated by the effects of the LAquila earthquake of 2009. The behaviour of such structures on that occasion is analysed on the basis of detailed surveys performed on site on a relevant number of buildings just after the earthquake. Following this analysis, which points out the inadequacy of the cladding panel connections, the paper presents a systematic definition of the problem, examining the behaviour of the whole system of structure + panels. Proposals are made for the typology of one-storey buildings for industrial uses, some possible alternative solutions that ensure the stability of all the construction elements. Specific calculations for a typical precast building give the order of magnitude of the forces and/or displacements for which the connections are to be designed.
Keywords: precast structures, seismic behaviour, structural connections

Introduction

The violent shake that struck the town of LAquila and its neighbourhood on the morning of 6 April 2009 had a destructive power of 5.8 on the Richter scale and a moment magnitude of 6.3 [1]. From the data of the National Seismic Network in different zones, there were horizontal peak ground accelerations between 0.35 and 0.40 g (0.60 g near the fault) which, from the seismic design point of view, correspond to the maximum seismicity expected in Italy. There were also particularly high vertical peak ground accelerations, of the order of magnitude of 0.50 g in some zones. This violent earthquake struck a stock of hundreds of different precast constructions spread over some 10 zones around LAquila. Most of them were industrial onestorey buildings and in a few cases commercial buildings of two or three storeys. The territory struck by the earthquake was classified for a seismic action of about 0.25 g. Some buildings were designed with the 1996 Italian Code, which already gave some specific detailing rules for seismic behaviour. Many others were designed with an-

* Corresponding author: toniolo@stru.polimi.it Submitted for review: 25 October 2011 Revised: 31 January 2012 Accepted for publication: 14 March 2012

tecedent codes which lacked such rules. Obviously, the latest Italian Code updated to 2008 did not have any influence on the earthquake effects and we can wonder whether an anticipated issue of this code would have had a positive influence. The first thing to be noted is the general good behaviour of the structures on their own, together with the notso-good behaviour of wall claddings, with the failure of many panels. On the one hand, these results confirm the reliability of the seismic design of precast frame structures with hinged beams, as codified after more than 15 years of theoretical and experimental research. On the other hand, they show the inadequacy of the present design approach to fastening systems for perimeter wall panels. As on preceding occasions, the earthquake has provided a dramatic lesson that has to be learned properly. We remember, for example, the lesson learned from the Friuli earthquake of 1976: the fall of a beam from its bearing had shown that under seismic action it is not possible to rely on friction for the transfer of horizontal forces. The combination of contemporary horizontal and vertical shakes can reduce the gravity action on which friction is based. This fact had been learned by the National Italian Association of Precast Concrete Producers, which promptly circulated a specific recommendation to its members. From that initiative, a general correct design practice was derived for mechanical connections between beams and columns which later led to the specific rule in the latest version of Eurocode 8. For the LAquila earthquake of 2009, the effects should again be examined and indications derived for updating design criteria and execution rules. The work should confirm the things revealed as adequate and lead to improvements to things revealed as inadequate, proposing also propose possible new alternative design rules based on innovative approaches. Very important to this end will be the results of an European research on the seismic behaviour of connections in precast structures involving 16 partners from the European countries mostly subjected to seismic risks, including the principal centres of seismic experimentation. The work started in 2009, will last three years and lead to the synthesis of the experiences gathered over more than 15 years of experimental and theoretical research on the subject of precast construction by the different national and international schools [2], [3] involved.

2012 Ernst & Sohn Verlag fr Architektur und technische Wissenschaften GmbH & Co. KG, Berlin Structural Concrete 13 (2012), No. 2

73

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

Observations after the earthquake

During the month following the earthquake, the authors made visits to the area of LAquila, including the industrial zones of Poggio Picenze, Monticchio, Ocre, Varranoni, Bazzano, Pila, Pettino and Popoli in particular. Some 50 buildings were examined with different degrees of depth depending on access possibilities, the availability of specific information and the interest in the case: from simple external visual inspections to detailed inspections of the internal and external parts of the building. The visits have been addressed on the basis of the indications of some local operators in the sector who had experienced the earthquake personally and knew the situation in the area well. The same operators guided us on many visits, obtained permission for access and provided a great deal of technical information about the constructions of which often they were the designers and producers. This allowed us to cover all the situations where relevant damage had been noted. Fig. 1 shows an emblematic image: a recently constructed building where the structure (columns, beams, roof elements) has essentially retained its integrity, but a whole wall of vertical panels has collapsed. The cause of

the collapse is shown in Fig. 2: the panel fastenings failed under forces not considered in the design. A list of the principal serious damage to the structures is presented first: a shed beam that has fallen from its bearing (together with the adjacent roof span) a long-span roof element that has fallen from its bearing the overturning of some TT floor elements (at erection stage) some spalling of column edges some spalling of beam-to-column bearings In addition, there are different minor types of damage, such as cracking in columns, local spalling of the edges at the bearings of floors on beams and of beams on columns. The large displacements during the earthquake also led to different damages to finishings such as internal partitions.

Structural damage

Fig. 1. Cladding panel collapse in Ocre

Fig. 3 shows the shed beam that has fallen from its bearing. The beam dragged down the adjacent span of the roof with it, which at the time of the picture had already been removed. The building is a cowshed in Fossa. The part of the structure where the collapse occurred was built in 1997. The shed beams were placed in pocket supports at the tops of the columns. Supports and beams were connected with a through-dowel. Under the seismic shakes, this connection did not work, probably because of the absence of a mortar infilling to fix the dowel. Fig. 4 shows the roof from which a long-span element fell (the other elements were removed after the earthquake because they were considered unsafe). The building is a vehicle workshop in Pile built in 1991. The ends of the roof elements were placed on specially shaped supporting elements and fixed with steel dowels. Under the seismic shakes, the edges of the supporting elements broke around the dowels (Fig. 5) because of insufficient anchorage to cope with the forces. Again, friction did not contribute adequately. These are the only two instances of structural elements in finished precast buildings collapsing around LAquila. But in the Fossa cowshed, serious damage oc-

Fig 2. Detail of fastening failure

Fig. 3. Shed beam failure in a cowshed in Fossa

74

Structural Concrete 13 (2012), No. 2

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

Fig. 4. Collapse of roof element in Pile

Fig. 7. Buckling of longitudinal bars

Fig. 5. Edge spalling of supporting element

Fig. 8. Failure of wall panels and floor elements Monticchio

Fig. 6. Failure of pocket support

curred in a part of the structure built in 1994. At the top of the columns, the lateral walls of some pocket supports broke (Fig. 6) pointing out the importance of adequate lateral restraint against the overturning of beams. Some

spalling occurred at mid-height (where the longitudinal reinforcement is curtailed), with buckling of the longitudinal bars (Fig. 7) because of an excessive spacing of the stirrups with respect to the flexural ductility demand under earthquake loads. This type of damage occurred in some other buildings, too. Concerning the collapses of structural elements, Fig. 8 shows the case of a building under construction in Monticchio where the cladding panels also dragged down two TT floor elements to which they were connected. The structure was not finished: if the planned concrete topping had been added to the floor, the collapse would not have occurred. Floor elements not yet fixed to the structure also fell in a few other buildings under construction. The two-storey building in Monticchio used as a gymnasium received a particularly strong shake with a very high vertical component. Fig. 9 shows a close-up of a seriously damaged bearing for a beam. No collapses occurred, but the building had much other relevant damage. Local failures of bearings also occurred in some other buildings, e.g. the factory in Varranoni shown in Fig. 10. Minor damage not affecting the resistance of the buildings was more numerous. Small cracks can be seen at the bases of several columns, but also at mid-height, pre-

Structural Concrete 13 (2012), No. 2

75

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

Fig. 9. Serious damage at a bearing in Monticchio

Fig. 11. Detachment of a vertical panel in Bazzano

Fig. 10. Spalling at a bearing in Varranoni

Fig. 12. Failure of a horizontal panel in Monticchio

sumably corresponding with the curtailment of the longitudinal reinforcement. Also frequent was spalling at the corners of supporting elements under concentrated bearing actions due to floor elements. Finally, signs of large displacements have been noticed, up to 150 mm at the top of some buildings. This has a relevant impact on the considerations presented in the following section regarding the design of wall cladding systems.

Damage to wall claddings

Collapses of wall panels caused by the failure of their fastenings were more numerous. These collapses concerned a relevant number of existing buildings (around 15 %). The forces received under the earthquake were much higher than those calculated in the design stage on the basis of local behaviour. The effects are those shown in Fig. 1 mentioned earlier, which refers to a recently constructed building in Ocre. The collapsed wall was in the NW-SE direction of the strongest acceleration component. The fastenings, made from channel bars, have been forced in the tangential transverse direction for which they were not designed.

The anchorage head worked as a lever, pulling out the edges of the channel (Fig. 2). In the orthogonal walls, the force, normal to the connection plane and related to the local mass of the panels and not to the whole mass of the roof, had sufficient resistance. Such collapses occurred in different buildings, involving both vertical and horizontal panels, as shown in Figs. 11 and 12 respectively (buildings in Bazzano and Monticchio). Failures occurred not only with channel bars, but also with other types of fastenings (Figs. 13 and 14). It is not a question of product inadequacy, but of inadequate design of the connection. Fig. 15 shows the interior of a factory in Poggio Picenze: the structure of columns, beams and roof elements suffered no damage, but a whole row of horizontal cladding panels fell down. The cause of the collapse is shown in Fig. 16, i.e. the failure of the fastenings. Fig. 17 shows the strengthening promptly applied to the remaining panels after the earthquake in the form of strong bolted steel angles. At the conclusion of this brief survey made in the field, it should be pointed out that the heavy damages pre-

76

Structural Concrete 13 (2012), No. 2

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

Fig. 13. Broken fastening

Fig. 15. Factory in Poggio Picenze with failed panels

Fig. 14. Wrenched bracket

Fig. 16. Failure of channel bars in the factory

sented above concern only a minority of buildings. The large majority of precast structures passed the earthquake check without relevant damage.

Design criteria

Present design practice for the precast structures under consideration is based on a bare frame model where the peripheral cladding panels are entered as masses only without any stiffness. The panels are then connected to the structure with fixed fastenings designed with a local calculation on the basis of their mass for anchorage forces orthogonal to the plane of the panels. This approach is not sound the LAquila earthquake has demonstrated this. Panels fixed to the structure in this way come to be an integral part of the resisting system, conditioning its seismic response as that of a dual wall-frame system with lower energy dissipation capacity. The high stiffness of this resisting system leads to much higher forces than those calculated from the frame model. These forces are related to the global mass of the floors and are primarily directed in the plane of the walls. The unforeseen intensity and direction of the forces caused

Fig. 17. Connection strengthened with steel angles

many fastenings to fail, leaving the frame of columns and beams practically undamaged. The old criterion leaving the walls to break since after their failure the remaining structure will resist by itself

Structural Concrete 13 (2012), No. 2

77

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

anyway does not work when failure implies the fall of panels with weights of up to 10 t. The mortal danger of these collapses requires a different approach. And these considerations hold true for all precast and cast-in-situ concrete structures: the fall of a large masonry cladding represents an equally serious danger. From these observations, three possible solutions for the cladding system can be presented, the principles of which are described below. A first solution, here called isostatic, is based on wall panels connected to the structure with supports that allow the free development of the large displacements expected for the frame structure under seismic action. This solution allows us to pursue the traditional design approach of the analysis applied to a bare frame model. A second solution, here called integrated, is based on wall panels connected to the structure with an iperstatic arrangement of fixed supports that make them part of the overall resisting system. This solution requires a completely new design approach with an analysis applied to a dual frame-wall model that involves unexplored participations of the roof diaphragms. Both these solutions require the specific design of new connectors: these must allow the large free displacements of the isostatic solution, or transfer the high forces expected in the integrated solution. A third solution can be mentioned as proposed by some operators in the sector, based on the present practice of under-proportioned fixed connectors, expecting them to fail under seismic actions, but providing the wall panels with a back-up system made of elementary devices that intervene in the case of incipient collapse. This solution, here called secured, could be mainly applied to the seismic retrofitting of many existing structures.

is oriented vertically, parallel to the gravity force. The origin is placed in a corner at the base side of the panel. Four connections are placed at the corners of the panel, indicated by A, B, C and D. These connections are intended to provide only translational restraint without any rotational restraint. The possible joint connections with the adjacent panels are indicated by E and F. Usually, connections A and B are attached to the foundation beam, connections C and D to the top beam. The same reference system is applied in Fig.18b to a horizontal panel, for which connections A, B, C and D are usually attached to the columns and E and F refer to the possible joint connections with adjacent panels, where the uncertain friction effect due to the weight of the superimposed panels may act. Tables 1 and 2 indicate the effect of the supports in the three directions x, y and z for the vertical and horizontal panels respectively, where the symbols are: f = fixed i = indifferent s = sliding 0 = absent Strictly speaking, for an isostatic arrangement in the plane of the panel, the indifferent supports should be defined as sliding supports but, in expectation of the negligible effects of the small linear deformations of the panel, they can be executed with the simpler fixed connections. However, this choice has to be verified for connections C and D of the vertical panel because it leads the panel to collaborates in the vertical bearing of the loads applied after completing the connections. In the plane of the vertical panel (in the x direction), the connection system defined in Table1 ensures horizontal displacements of the structural frame independent of the panel, which remains supported at the base, on the foundation beam, providing its stability by itself (Fig. 19). In the orthogonal y direction, the panel remains simply supported at its upper and lower ends, following, without reactions, the vibratory motion of the structure, to which it transfers the inertia force due to part of its mass. Choosing the simpler type of connection for the indifferent supports, i.e. the fixed one, the system requires two types of connectors, one with full support and one with partial support that allows one of the three translations; this is in expectation of large displacements up to
Table 1. Supports for the vertical panel A f f f B i f f C s f i D s f i E 0 0 0 F 0 0 0

5.1

Isostatic wall system

Different panel-to-structure connection systems can ensure an isostatic arrangement of the panels. Examples of some of these systems are presented below with reference to both vertical and horizontal panels, pointing out some problems of practical application. Fig. 18a shows a vertical panel referred to an orthogonal axis system, where x is oriented horizontally in the panel plane, y is oriented orthogonally to that plane and z

x y z

Table 2. Supports for the horizontal panel A f f f B i f f C s f i D s f i E s 0 0 F s 0 0

Fig. 18. Schemes for vertical (a) and horizontal (b) panels

x y z

78

Structural Concrete 13 (2012), No. 2

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

Fig. 22. Iperstatic system with interconnected panels Fig.19. Isostatic supports for vertical panels

Fig. 20. Isostatic supports for horizontal panels

Fig. 23. Iperstatic connection system for vertical panels

Fig. 21. Isostatic pendulum system of connections

15 cm. It should be noted that the base connections A and B at the foundation beam cannot be simple bearings, but must be mechanical connections with a bilateral effect. In the plane of the horizontal panel (in the x direction) the connection system defined in Table2 binds the displacements of the panels to those of the joints with the columns on which the inertia forces due to their mass are transferred (Fig.20). For a free motion without reactions, it is essential that no relevant friction arises at the joint with the superimposed panels, and this requires the interposition of a free space or adequate unusual deformable seals. Due to the friction, it is therefore more difficult with the horizontal panels to create an isostatic system that does not react with the structure. In the y direction, the horizontal panels follow the vibratory motion of the columns without reactions, transferring the inertia forces due to their mass to the four joints. Fig. 21 shows another possible solution with an isostatic pendulum arrangement of vertical panels, connected at the base and the top with connections placed in the

middle of the horizontal sides. In this arrangement, the cladding system can freely follow the motion of the roof, displaying relative sliding db/h at the joints of the adjacent panels, where d = top horizontal displacement, b = panel width and h = height of upper connection. Only fixed connections are required, unless the z support is released to avoid vertical actions on the panels due to the gravity loads transmitted from the connected roof element. It should be noticed that, in their rocking motion around the base support, the adjacent panels, besides sliding tangentially with respect to each other, get closer by an amount that, for the expected small values of the top storey drift, is lower by one order of magnitude than the top displacement and can be easily offset by ordinary joint play. Equally negligible is the lowering of the top connection. The connections of the corner panels, which had different shortcomings during the LAquila earthquake, have to be investigated in more depth.

5.2

Integrated wall system

The same pendulum support system can be transformed into an integrated system if the E and F connections are added to prevent the relative sliding of the panels at the side joints (Fig. 22). Particular fasteners have been designed and tested for these joint connections in order to dissipate energy under a violent earthquake [4], [5]. The ordinary arrangement of the connections placed in the four corners of the panels, which make them collaborate with the structure, is based on full translational supports and the columns A, B, C and D of Tables 1 and 2 all

Structural Concrete 13 (2012), No. 2

79

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

become filled with f. Fig. 23 shows this arrangement for vertical panels, where the distortion effects at the top beam can be seen. These effects may be avoided if only one connection is placed on the top beam, as indicated in Fig. 24. In this solution, the panels behave in their plane like cantilevers fixed at the base and placed side by side. Similar solutions can be used for horizontal panels: one with four connections as described in Fig. 25 leading to distortion effects at the columns; another without distortion effects where connections A and B at the columns are replaced by the connections E on the underlying panel (Fig.26). The technological solution of fully fixed connections is simplified, but the integrated participation of the panels, with their stiffness, in the vibratory motion of the structural system leads to very high forces at the connections. In

terms of present production, new products with improved capacities are needed. Of course, it is not only the steel connector that needs improvements, but also the nearby concrete part of the panel that has to resist without anticipated failures. Furthermore, not only the calculation of the connections has to be updated, but also the calculation of the roof diaphragm through which the inertia forces are transmitted to the resisting lateral walls. Anyway, sizing the connections shall be carried out case by case, with a global analysis of the structural system and quantifying the force and displacement distributions. With a much higher structural stiffness, small displacements are expected and the serviceability limit state requirements (limitation of damage) are easily satisfied. However, more complex resistance and ductility verifications at the no-collapse limit state are expected, addressed more to the walls than to the columns. The resistance and plastic deformation capacities of the panels are to be quantified for in-plane actions. Assuming low-ductility connections, a capacity design calculation should be set up and tested for the over-proportioning of the connections with respect to the ultimate strength of the panels in their structural arrangement. Following these investigations, the value of the behaviour factor q (force reduction factor) must also be defined for this structural system, which does not seem comparable with what is defined as a wall system in the design codes because of the great difference in the detailing and the supports.

Fig. 24. Alternative Iperstatic system for vertical panels

5.3

Secured wall systems

Finally, we shall introduce as a small seed to germinate Fig. 27, which refers to the third solution, i.e. a back-up line of elementary fasteners securing against the fall of panels added to a structural system designed following the traditional practice of under-proportioned connections. The theme belongs to the wide scope of the assessment and possible retrofitting of the building stock: a very complex problem that is difficult to solve and for which only a few technical suggestions can be advanced here.

Fig. 25. Iperstatic connection system for horizontal panels

Fig. 26. Alternative iperstatic system for horizontal panels

Fig. 27. Device securing against falling

80

Structural Concrete 13 (2012), No. 2

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

Analysis of a typical building

The order of magnitude of the displacements and forces that are demanded by the isostatic and the integrated solutions respectively as described above is deduced below. As a typical building to be examined, the one-storey precast industrial building illustrated in Figs. 28, 29 and 30 (50 41 m on plan) has been chosen. The columns have a square section of 60 60 cm, are 5.2 m high and are placed at a spacing of 10m along three longitudinal lines. The prestressed concrete beams (with a constant

depth of 1.2 m) are positioned at the top of the columns along these lines. Resting on the beams are the TT prestressed concrete roof elements, each 70 cm deep, 2 m wide and 20.5 m long. A strip 1m wide is left open for a skylight every two elements. Vertical wall panels 2.5 m wide are placed along the perimeter to complete the building. Concrete C45/55 is used for beams and roof elements, C30/37 for columns and cladding panels. The cladding panels are sandwich wall elements with an intermediate insulating layer. The amount of reinforcement is

Fig. 28. Plan of the typical building

Fig. 29. Longitudinal section

Fig. 30. Transverse section

Structural Concrete 13 (2012), No. 2

81

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

not relevant for the present analysis, which is aimed only at quantifying displacements and forces in the panel-tostructure connections. In a first case, this building is assumed to be standing by itself with cladding panels on the four perimeter sides in a doubly symmetrical arrangement. In a second case, it is intended to be one of the two parts of a double length building divided by a seismic joint and in this case the wall panels are on three sides only, with an asymmetric arrangement in one direction. The analyses are applied to a three-dimensional model that reproduces closely the geometry of the structure together with the distribution of the masses and of the elastic stiffnesses of the construction elements, including the in-plane stiffness of the roof elements. The arrangement of the supports consists of fully fixed supports at the column bases, hinged beam-to-column connections in the horizontal and vertical planes of the beam, fixed beam-tocolumn connections in the orthogonal vertical plane, and spherical hinges between the beams and the two ribs of the roof elements. In the calculation model, the wall panels are represented by plate elements connected to the foundation by two spherical hinges, whereas at the top they are connected to the beam by two joint elements, the stiffness of which is assumed to be very low or very high each time in order to simulate the two quoted solutions of isostatic and integrated connections. These arrangements are those described in Figs. 18a and 19 and Table 1. The masses involved in the vibratory structural response are those of the seismic action combination with only permanent loads at their characteristic or nominal values (with G = 1.0). These loads are evaluated automatically by the computer program: 4.0kN/m2 distributed on the roof and 3.0 kN/m2 distributed on the peripheral wall claddings. No unintended eccentricities of the masses have been introduced; obviously, the systematic eccentricities due to the asymmetric wall arrangement remain. The seismic action corresponds to a reference peak ground acceleration ag = 0.25 g, with the response spectrum of Eurocode 8 for a soil category B (with S = 1.2). The value q= 3.5 has been assumed for the behaviour factor. This assumption seems to be consistent with the bare frame structure (isostatic connection system) for column proportioning that takes into account the storey drift limitation at the serviceability limit state and with the dual wall-frame structure (integrated connection system) to represent its reduced dissipative capacities. Dynamic modal analyses have been performed with the help of a computer program for defining the principal vibration modes, subsequently evaluating forces and displacements separately for longitudinal (x direction) and transverse action (y direction). Figs. 31 and 32 show the three-dimensional representation of the model for the symmetric and the asymmetric arrangements respectively.

Fig. 31. 3D model for symmetric arrangement

Fig. 32. 3D model for asymmetric arrangement

Results of the analysis

For the symmetrical wall arrangement, the first two modes correspond to pure translations along y and x, with natural vibration periods for the isostatic connection system of 1.03 and 0.80 s respectively (common for this type of

frame structure). A third significant vibration mode corresponds to a torsional rotation with a natural vibration period of 0.74 s. These periods are much lower for the integrated connection system (0.65, 0.42 and 0.41s) because of the much higher stiffness of the dual frame-wall structural system. For the asymmetric wall arrangement with isostatic connection system, the first three modes remain practically the same as for the symmetric arrangement since the frame structure, disconnected from the walls, remains symmetrical. The different mass distribution modifies the natural vibration periods a little, to 1.03, 0.79 and 0.74 s. For the integrated system, the asymmetry of the wall substantially modifies the response, with a first translationalrotational mode along y that keeps a high natural vibration period of 0.86 s because of the absence of the stiffening wall on one side, a torsional mode that becomes the second mode with a low natural vibration period of 0.46 s and a symmetrical translational mode along x with a natural vibration period of 0.41 s. For connection design purposes, we shall consider the maximum values of the behaviour parameters of the response that are for the isostatic support system the relative horizontal displacements between panels and structure at their connections, and for the integrated support system the horizontal components of the forces transmitted between panels and structure at their connections. The structure analysed can be assumed to be representative of the typology, so the maximum values quoted can be taken as indicative of the order of magnitude of the parameters. These parameters are given in Table 3 together with some other data regarding the overall behaviour of the structures. The terms max slide and max force indicate the displacement and force in the connection quoted above, max drift indicates the maximum top displacement of the

82

Structural Concrete 13 (2012), No. 2

G. Toniolo/A. Colombo Precast concrete structures: the lessons learned from the LAquila earthquake

Table 3. Maximum values of behaviour parameters

Symmetric arrangement along x isostatic max slide (mm) max force (kN) max drift (mm) max shear (kN) max mom (kNm) 56 59 62 362 integrated 104 31 31 191 along y isostatic 70 80 47 331 integrated 83 59 38 256

Asymmetric arrangement along x isostatic 55 67 62 358 integrated 100 35 31 189 along y isostatic 70 73 47 331 integrated 73 49 44 306

structure, and max shear and max mom indicate the components of the internal force at the column base. It must be remembered that the calculation of the displacements and forces has been performed using a modal dynamic analysis with linear elastic behaviour of the elements and a design response spectrum reduced by a factor q = 3.5 for a peak ground acceleration equal to ag = 1.2 0.25 g = 0.30 g, which corresponds to the no-collapse limit state. The computed values of the displacements have therefore been amplified by the same factor q = 3.5 to account for the elasto-plastic non-linear behaviour of the structure. As can be seen in Table 3, the maximum relative displacement between panels and structure for the isostatic support arrangement is 7.0 cm. The maximum force transmitted between panels and structure for the integrated arrangement is 104 kN. For the symmetrical structure in the longitudinal direction, the integrated support system practically halves the internal forces at the column base, transferring most of the action to the lateral walls because of their much higher stiffness.

search. The way to reach a complete knowledge of the seismic behaviour of the constructions is still long. And this is valid not only for precast structures. The LAquila earthquake in particular was a dramatic reminder of the inadequacy of cladding and partition walls, both masonry and precast, for all types of construction. This is a topic of primary importance for safeguarding lives when earthquakes occur.

Acknowledgements
The present work was performed within the scope of the SAFECAST research project supported by the contribution of the European Commission in the FP7-SME-2007-2 Programme with Grant agreement No. 218417 of 2009.
References
1. Menegotto, M.: Experiences from LAquila 2009 earthquake, Proceedings of the 3rd fib Congress, Washington D.C., 2009. 2. Colombo, A., Negro, P.: Tailoring experimental strategy and set-up: the long story of the seismic behaviour of precast structures, 3AESE, San Francisco, 2009. 3. Biondini, F., Toniolo, G.: Experimental research on seismic behaviour of precast structures, Industria Italiana del Cemento No. 3, 2009. 4. Iqbal, A., Pampanin, S., Buchanan, A., Palermo, A.: Improved seismic performance of LVL post-tensioned walls coupled with UFP devices, 8th Pacific Conference on Earthquake Engineering, Singapore, 2007. 5. Shultz, A. E., Magana, R. A., Tadros, M. K., Huo, X.: Experimental study of joint connections in precast concrete walls, Proceedings of 5th national Conference of Earthquake Engineering, 1994. 6. Biondini, F., Dal Lago, B., Toniolo, G.: Seismic behaviour of precast structures with dissipative connections of cladding wall panels, ANIDIS Congress, Bari, 2011.

Conclusions

The survey and calculations reported above give the first indications for the necessary updating of the design criteria for the connection system for the wall cladding panels of precast structures. Introducing a proper margin with respect to the only example treated in the text (e.g. for higher seismic actions), 1.75 7.0 12 cm indicates the free drift capacity of the panel fastenings for the isostatic solution to the connection system, and 1.75 10.4 18t indicates the force transmission capacity of the panel fastenings for the integrated solution. This last strength capacity relates to the single connection as a whole not only the steel fastener, but also the adjacent concrete parts of the connected elements. A reduction in the required strength is possible if joined by an adequate plastic deformation capacity or with an additional system of connections between the panels [6]. Some problems are still pending with reference to the seismic design of the structural assembly in the case of an integrated panel connection system that involves the interaction between the spatial frame of beams and columns, the box wall cladding system and the roof diaphragm. This presentation is a long list of subjects to be investigated in greater depth and will require some years of re-

Prof. Giandomenico Toniolo Politecnico di Milano Ingegneria strutturale p.za Leonardo da Vinci 32 Milan 20133, Italy Email: toniolo@stru.polimi.it

Dr. Antonella Colombo Assobeton Milan, Italy Email: a.colombo@assobeton.it

Structural Concrete 13 (2012), No. 2

83