Composite modular floor prototype for emergency housing applications: Experimental and analytical approach

Composite modular floor prototype for emergency housing applications: experimental and analytical approach Hassan Abdolpoura, 1 , Julio Garzón-Rocab, Gonçalo Escusaa, José M. Sena-Cruza, , Joaquim A.O. Barrosa, Isabel B. Valentea a ISISE, University of Minho, Guimarães, Portugal b Department of Geotechnical and Geological Engineering, Universitat Poliècnica de València, Valencia, Spain Abstract The present paper explores a new modular floor prototype to be used in emergency houses. The prototype is composed of a frame structure made of glass-fiber-reinforced polymer (GFRP) tubular pultruded profiles, a slab made of sandwich panels with a polyurethane (PU) foam core and GFRP skins, and a tailored connection system that provides integrity between assembled components. A series of experimental tests are carried out including flexural tests on a single panel, on two and three connected panels, and on the assembled floor prototype. The behaviour of the panels is analysed when they are not considered part of the GFRP framed structure, namely the failure mechanisms and the efficiency of the proposed connection system between the panels. The performance of the floor prototype to support typical load conditions of residential houses is also assessed. Additionally, an analytical model was used to deeper study the behaviour of the developed sandwich panels, connection system and the modular floor prototype. Keywords: emergency house; composite materials; GFRP pultruded profiles; sandwich panels; GFRP skins; PU foam core. 1. Introduction One of the major concerns after a natural disaster is settling down surviving communities in shelters or temporary houses. This issue remains difficult to manage despite decades of experience. Availability of temporary housing is crucial since it allows people to quickly commence their daily activities such as school, working and cooking [1-3]. Even though there are different sorts of temporary buildings made of steel, wood 1 Corresponding author. Tel.: +351-253-510-200; fax: +351-253-510-217 E-mail address: Hassan.abdo[email protected]m (H. Abdolpour) and plastic [4-6], many of these temporary dwellings do not offer a basic level of security and protection for its occupants, and/or result in very complex and expensive solutions. Nowadays, a clear trend is observed in the industrial manufacturing and prefabrication of temporary building towards. This modern method of construction leads to achive tangible benefits in terms of faster construction, improved quaility and reduced wasting resource material[7]. Lightness is a key factor when proposing a material/structural system for a temporary building because, after a natural disaster, accessibility to roads is usually limited. Thus, low weight prefabricated components are very convenient for packing, shipping, unpacking and assembling [8]. Taking this into account, sandwich panels made fundamentally by GFRP pultruded profiles and sandwich panels may constitute excellent options in the field of temporary buildings. Sandwich panels and pultruded profiles are lightweight elements with very good mechanical performance, being able to be manufactured and rapidly assembled in modular sections. Typically sandwich panels are composed of two thin and stiff outer faces separated by a thick and low density core material. These panels offer various advantages, such as high strength and stiffness to weight ratio, immunity to corrosion, and a low thermal and acoustic conductivity [9-13]. Likewise, pultruded GFRP profiles show a series of promising advantages, such as relatively low production costs, low maintenance, high durability, immunity to corrosion and high strength [14-17]. Accordingly, sandwich panels have been efficiently used in several structural applications, such as cladding [18], facades [19, 20], roofing [21] and walls [22]. In the scope of this work the mechanical behaviour of a modular floor prototype, developed to be used in temporary emergency houses, is assessed. The proposed basic unit system is capable of covering an area of approximately 9 m2. Three sandwich panels made of GFRP skins and PU foam cores were used for configuring the enclosed floor surface, while GFRP tubular pultruded profiles formed the skeleton of the system. Elements were adequately interconnected upon assembly, constituting a floor system and providing a diaphragm behaviour [23]. In addition, the fitting connection system was designed for an easy and fast assembling and disassembling of the connected elements. The prototype is analysed through a series of experimental and analytical studies. In a first stage, the flexural behaviour of a single panel, and the one of systems formed by two and by three connected panels are experimentally assessed under serviceability limit state (SLS) and ultimate limit state (ULS) conditions. The tests with the systems of panels have also the purposes of analysing the efficacy and contribution of the connection between panels, by studying the transference of loads from one panel to the adjacent ones. Furthermore, ultimate capacity of a single floor panel and its corresponding failure mechanism system is experimentally assessed. In a second stage, an experimental program is conducted to evaluate the performance of the developed basic floor unit prototype as a structure designed to support serviceability and ultimate load conditions in residential houses. Finally, analytical studies are carried out to contribute for a deeper assessment of the failure mode of sandwich panels, the influence of the ribs placed inside the panels, the efficiency of the connections, and the behaviour of the floor prototype. 2. Prototype description 2.1 Concept and geometry The designed temporary building house was composed of a single-story building with a rectangular area of about 6.0 × 3.0 m2, formed by connecting two blocks of about 3.0 × 3.0 m2 and a height of about 3.0 m. Fig. 1 shows a plan and three lateral views, as well as a photo of the built prototype. The floor module of the building is depicted in Fig. 2 and was composed of two main components: (i) a frame formed by tubular GFRP pultruded profiles with a cross section of 120×120 mm2 and a wall thickness of 8 mm (Fig. 2a), and (ii) the pavement constituted by three sandwich panels formed by two outer GFRP skins of 5 mm thickness and a core of PU foam (Fig. 2b). For the sake of decreasing segment’s variation in the manufacturing process, the same profile was used in both beams and columns of the frame. The panels presented an overall thickness of 70 mm, a width of 1000 mm and a length of 3000 mm. A U-shape GFRP pultruded profile with a cross section of 60×55 mm2 and a wall thickness of 5 mm (U60×55×5) was adhesively bonded to the PU foam core during the manufacturing process, on the outer side of each panel, enabling the connection of each panel to the other elements of the prototype, such as beams and other panels (see Fig. 2c). For increasing the flexural stiffness of the panel, two additional U60×55×5 profiles were installed in the interior of each panel. PU foam blocks with a thickness of 60 mm and nominal density of 48 kg/m3 were used to form the sandwich panel core, providing the required thermal isolation. These blocks were bonded to the GFRP skins with a polyester resin. GFRP material was selected for the skins over traditional materials, such as steel or concrete, mainly due to its impunity to corrosion and higher strength/lightness ratio. The GFRP skins were produced with the hand-layup technique, using dry glass fibres impregnated with an isophthalic polyester adhesive. Multiple plies of glass fabrics were used in the process, comprising two different types of mat: chopped strand mat (CSM) and bidirectional woven fabric mats (BWFM). Each skin had five layers disposed symmetrically towards its middle surface (CSM-300 gr/m2, CSM-450 gr/m2, CSM-450 gr/m2 + BWFM-500 gr/m2, CSM-450 gr/m2, CSM-300 gr/m2) impregnated with adhesive. The weight of the sandwich panel was around 70 kg, which facilitates its transportation and on-site installation. The connections were designed for an easy and fast assembling / disassembling of the prototype, and assure continuity as much as possible between connected elements in order to mobilize efficiently their strength capacity (see Fig. 2c). For beam-panel connections, the aforementioned U-shape GFRP profiles placed on the edges of the sandwich panels were attached to a GFRP squared tubular profile of 50 mm edge and 5 mm of thickness that was mechanically and adhesively bonded to the GFRP beam. Finally, for panel to panel connection, a similar approach as that followed for beam-panel connection was used, by attaching the U-shape GFRP profiles to two GFRP squared tubular profiles (also of 50 mm edge and 5 mm of thickness) that were mechanically and adhesively connected together. 2.2. Assembly process The floor prototype was developed for disaster areas where special tools and equipment, as well as experimented workers, are scarce. Consequently, these issues were also considered in the design process of the prototype. Fig. 3 shows a general view of the process needed to assemble the floor of the developed prototype. The process starts by placing the four columns in their positions (note that for the floor test proposed in of this paper, short columns with approximately 1/3 of the real height were used), and then connecting three of them by beams (Fig. 3a and Fig. 3b). Afterwards, the three sandwich panels were installed. Panels were handled and mounted along the beam-panel connections (Fig. 3c), placing the panel to panel connectors after positioning the first and the second panels (Fig. 3d). After assembling the third panel, the last beam of the frame was installed and connected (Fig. 3e). Fixing ropes were used along the process for facilitating the adjustment of the panels. Fig. 3f shows the floor prototype after has been assembled, which required less than 2 hours and three people without any special equipment. 3. Experimental programme The following subsections provide details on the experimental programme conducted in terms of material characterization, test specimen, setup and procedures. All tests were carried out in the Laboratory of the Structural Division of the Civil Engineering Department at University of Minho, Portugal (LEST). 3.1. Material characterization Both GFRP profiles and sandwich panel GFRP skins were characterized by performing tensile tests according to ASTM D3039 recommendations [24]. Five tensile specimens with dimensions of 250255 mm3 were extracted from the profiles, and from the sandwich panel skins in both longitudinal and transversal directions. Tensile specimens were conducted in a universal testing machine, with a grip distance of 150 mm, and monotonically loaded up to failure, with a head displacement rate of 2 mm/min (Fig. 4a). Mechanical properties of the PU foam core were evaluated under compression, tension and shear tests. Flatwise compression properties of the PU foam were determined according to the ASTM C365-03 recommendations [25], by testing five prism-shape coupons of 70×70×50 mm3(Fig. 4b). Tensile properties of the PU foam were evaluated according to ASTM C297/C 297 M-04 prescriptions [26], by testing five coupons of 70×70×50 mm3 adhesively bonded to the flange of steel T-section (Fig. 4c).s. For the shear testing, five cubic specimens with a dimension of 120×120×120 mm3 were prepared. The specimens were bonded to four metallic plates and mounted in the universal testing machine where a tension load was applied along one side of the setup (Fig. 4d). The applied load was transferred into the plates that were encasing the specimen. Comprehensive information about this test can be found elsewhere [27]. Tensile bond strength of the adhesive joint between GFRP skin and PU foam core was measured by pull-off tests based on ASTM 1583-04 recommendations [28]. Five cores were drilled on GFRP skins with a diameter of 50 mm and a core depth of about 10 mm. Aluminium disks were adhesively glued to the GFRP skin. Tensile force was applied to the disks with a head displacement rate of 0.2 mm/min (Fig. 4c). Finally, the tensile behaviour of the polyester resin (adhesive material used to glue GFRP skin to PU foam core) was assessed by performing direct tensile tests, according to ASTM D638 prescriptions [29]. The samples were cast in dog-bone mould shapes and cured following the standard recommendations. Specimens were tested in a universal testing machine at a displacement rate of 2 mm/min (Fig. 4f). 3.2. Single sandwich floor panel: tests under uniformly distributed loading Four full-scale floor sandwich panels (hereafter designated by FP1 to FP4) were tested under a uniform load to evaluate their structural performance as a single panel. Following the United Nations (UN) recommendation for an emergency house, a uniform load of 1.6 kN/m2 was selected to be load in SLS. That load was increased 1.5 times to evaluate their response under ULS conditions, as traditionally defined in the Eurocodes. The panels were tested with a clear span of 2700 mm, and supports were materialized by placing a steel roller with a diameter of 50 mm under both panel ends. Both supports allowed free rotation and one of them also permitted longitudinal sliding (roller support), while the other was fixed in the longitudinal direction (pinned support). Load was manually applied by using cement bags of 20 kg each. In a first step (SLS loading configuration) 16 cement bags were disposed in two layers, representing a uniform distributed load of 1.6 kN/m2. In a second step eight extra bags were added to attain a loading level corresponding to ULS conditions. Loading operations were performed as fast as possible to avoid any potential creep effect. The cement bags were distributed as uniformly as possible on the panels, but gaps between bags were assured to avoid any arch effect. Vertical displacement was measured by means of a LVDT placed at the intersection of the specimen’s midspan section with its longitudinal axis. The panels were also instrumented in the tension skin (bottom skin) with a strain gauge bonded at a distance of 10 mm from the centre of the panel to avoid any interference with the LVDT. 3.3. Single floor sandwich panel: tests up to failure The static behaviour of one-way full scale sandwich panel up to its failure was investigated by executing a fourpoint bending test according to the ASTM C393 recommendations [30]. The panel was tested with a shear span of 850 mm, and supports were materialized as already described for the single panels submitted to a uniformly distributed load. Hydraulic jack was used to apply a monotonic load up to the failure of the specimen. The load was transferred to the panel by means of a longitudinal spreader HEB 200 with a length of 2000 mm, and two IPE 100 profiles with steel rollers of 20 mm of diameter welded at their bottom flange. A load cell of 300 kN was used to register the load applied. Rubber pads were placed between the panel and the steel rollers to avoid any indentation failure [31-34]. Fig. 5a shows the test setup configuration. Vertical displacements were recorded by five Linear Variable Differential Transformers (LVDTs) with a stroke ranging from 25 mm to 50 mm, placed under loaded sections (D4-D5) and at mid-span (D1-D3). Moreover, six strain gauges were bonded on the bottom skin (S1-S3) and on the top skin (S4-S6) at the midspan section of the specimen (see Fig. 5b). 3.4. Flexural response of connected sandwich panels After have been submitted to uniformly distributed load, the three floor panels (FP) described in Section 3.2 (FP1 to FP3) were also tested (Fig. 6a) in a two-by-two connection configuration (FP1 with FP2 and FP2 with FP3). Each pair was connected together by two GFRP tubular profiles of 50×50×5 mm3 cross section, in order to resemble the arrangement of the floor prototype (see Fig. 2c). A test with the three panels (Fig. 6b) connected together (FP1, FP2 and FP3) was also carried out. All these tests were undertaken under a four-point bending configuration with a shear span, a flexural span and a clear span of 850 mm, 1000 mm and 2700 mm, respectively, and according to ASTM C393 recommendations [30]. The support conditions were similar to those adopted previously in the single panel tests. The load was transferred to the panels by means of a frame formed by a 2000 mm long metallic HEB 200 profile, to which was attached (welded) two transversal HEB 200 profiles with a length equal to the width of the connected panels (i.e. 2000 mm for the case of two panels and 3000 mm for the case of three panels). Two cylinder steel bars of 50 mm diameter were placed between the panel and the load transfer frame in order to apply a line load. A monotonically increase load was applied by a hydraulic jack on the panels until reaching a magnitude of 10.29 kN and 15.43 kN for the case of two and three connected panels, respectively. These load levels correspond, in terms of load (maximum bending moment), to an equivalent uniform load for ULS (2.4 kN/m2) in a four-point bending test configuration. A load cell of 300 kN with a precision of 0.05% was used to measure the load. To assess the effectiveness of the connection in distributing the load amongst the connected panels, an additional test with three connected panels was conducted by applying the load only on the central panel (see Fig. 6c). This test configuration followed exactly the setup previously indicated, but in this case, the length of the steel cylinder placed under the HEB profiles was only 1000 mm, therefore the load is exclusively applied on the central panel. The instrumentation used for monitoring these tests is depicted in Fig. 6e. Ten and fifteen LVDTs (with a stroke ranging from 25 mm to 50 mm) were used in the two and three connected panels, respectively, for measuring the vertical deflection of the panels in their loaded and mid-span sections. Strain gauges positioned on both skins (top and bottom), at the midspan of the specimen were used to measure the strains developed in GFRP skins. 3.5. Loading test on the modular prototype After the prototype has been built by assembling its components according to the description in Section 2.2, its structural behaviour was evaluated under a uniform load of 1.6 kN/m2 in SLS and 2.4 kN/m2 in ULS, since these values, as already indicated, correspond to UN recommendations for temporary buildings. The loading arrangement was materialized by using a swimming pool of circular area and 6.25 m2 as illustrated in Fig. 7a. The monitoring system adopted for this test is displayed in Fig. 7b. Eleven LVDTs (D1 to D11) with a stroke ranging from 25 mm to 50 mm were placed at the bottom of the prototype to measure vertical deflection, while 15 strain gauges (S1 to S15) were positioned on the bottom surface of the beams and panels to register the strains during the loading process. 4. Results and analysis 4.1. Material characterization Table 1 summarizes the results of the material characterization tests conducted on the different components of the floor prototype, listing the values obtained for the maximum tensile stress in the longitudinal ( max,L  ) and transversal ( max,T  ) directions, elastic modulus in the longitudinal ( L E ) and transversal ( T E ) directions for both the GFRP profiles and skin. The maximum compression, tension and shear stress for the PU foam, and its longitudinal (E) and transversal (G) elasticity modulus in compression are also provided in this table. The specimens of GFRP profiles and skins presented tensile linear-elastic behaviour up to failure, which took place in a brittle manner in their middle part, having the rupture surface progressed perpendicularly to the specimen axis. The PU foam under compression developed a linear elastic stress-strain response, followed by a plastic plateau with nearly constant stress, and a final strain-hardening stage at relatively large strain level due to the increase of stiffness caused by the large experienced deformation, which is a typical behaviour for this type of foams [35, 36]. The response of the PU foam in shear was linear-elastic until failure, with the formation of failure surfaces at an angle of nearly 45. The PU foam coupons submitted to tension also presented a linear elastic behaviour up to failure, with a slight strain hardening in the last loading stage. In the pull-off tests, a tensile strength of 0.5 MPa (CoV = 18.7%) was obtained. The failure has occurred in the PU foam core. No failure was detected in the interface between GFRP and the PU foam core. Since the ultimate tensile strength of the PU foam core was quite close to the tensile strength obtained in the pull-off tests (0.5 MPa), it is confirmed the rupture was caused by the attainment of the tensile capacity of PU foam. 4.2. Single sandwich floor panel: tests under uniformly distributed loading The registered midspan deflections and midspan strains for tested panels FP1-FP4 under uniformly distributed load are shown in Fig. 8a and Fig. 8b, respectively. The application of the load in two steps, corresponding to SLS (1.6 kN/m2) and ULS (2.4 kN/m2) load conditions is fully recognizable in the graphs by the abrupt increase of midspan deflection (and strain) after the stabilization stage at the end of the SLS loading process (SLS_L). As can be seen, a full recovery of the deflections and strains took place after the total unloading phase, which evidences that for the considered load levels the panels have presented an elastic behaviour. Moreover, Fig. 8 shows that, the values for both deflections and strains registered were very similar in the four tested panels, revealing a manufacturing process of high repeatability. The Italian standard CNR can be used to verify the performance of sandwich panels under SLS [37]. According to this standard, the maximum long-term deflection registered for the quasi-permanent loading, Quasi  , (equal to 30% of the service load) should be less than L/250. The Quasi  considering creep effects is determined from the following equation: Quasi SLS Creep        (1) where SLS  is the deflection in SLS loading conditions,  is the proportion of quasi-permanent loading in respect to the SLS loading (i.e. 30%), and Creep  is the estimated coefficient due to creep effects. Based on creep study [46]. For the present sandwich panel, the equivalent flexural stiffness is obtained by the following equation:   2 33 3 ( ) 2 ( ) 2 6 2 2 12 cU ff eq f f u c u u c u t nE tt EI bE t b t b t t t                 (9) where b is the width of the panel, n is the number of the U-shape GFRP profiles (both located in the interior of the panel and at the edges), and U E , u t , u b are the Young’s modulus, thickness and width of those profiles, respectively. In Table 3 is compared the midspan deflection obtained from Eq. (8) and measured experimentally. A very good agreement between experimental and analytical values is observed, showing an adequate precision of FSDT in estimating the total deflection of the sandwich panels. Fig. 14 shows the influence of the number of U-shape GFRP profiles (n) on the midspan deflection of the sandwich panel, having been adopted values of n varying from 0 to 4. As can be observed, when deflection is computed for a panel without U-shape GFRP profile (n = 0) and for a panel with one U-shape GFRP profile (n = 1), a sudden decrease in the deflection of nearly 42% takes place. By using more than one U profile, the total deflection tends to decrease almost linearly with the increase of the number of GFRP profiles. Moreover, the relative contribution of bending (M) and shear (V) on the total deflection (indicated on the top right corner of Fig. 14) shows that the contribution of the shear deformation decreases with the increase of the number of GFRP profiles applied. When GFRP are not applied (n=0) the contribution of bending and shear for the total deformation is 60% and 40%, respectively, while when 4 GFRP profiles are adopted this relative contribution is 97% and 3%. Thus, for the panel designed, the dominate deformation is flexural and not shear, since four Ushape GFRP profiles were used in each panel (two interior and two at edges) that are working as ribs, providing high shear stiffness (GA) to the panel. The load-deflection behaviour of the tested single sandwich panel under four-point bending test configuration was also analytically determined by using FSDT and considering the same assumptions. Eq. (10) gives the total deflection for a four-point load configuration,   22 486 s Total PBT eq L Pa a Pa EI KGA         (10) where P is the applied line load, s L is the span (equal to 2700 mm), a is the shear span (equal to 850 mm),   eq EI is the equivalent flexural stiffness determined according to Eq. (9) (159.30 kN.m2), and GA is the shear stiffness (6147.4 kN). By adopting these values, a midspan deflection of 62.01 mm was calculated, which is in good agreement with the experimental result, since the difference is 1.60 %. 5.3. Efficiency of the proposed connection system between panels The Eqs. (9) and (10) were also applied to predict the midspan deflection of the connected panels (see Sections 3.4 and 4.4), and the results are presents in Table 4, which also include the flexural and shear stiffness values. Additionally, analytical results of a unique slab (a slab formed by the same number of U-shape GFRP profiles as the connected panels but assuming continuity between the panels) with the same dimensions and configuration of the connected panels were obtained in order to evaluate the efficiency of the proposed connection. The results of Table 4 indicate an acceptable predictive performance for the analytical expressions, since a relative difference of 2.15 % and 2.40 % for the, respectively, two and three connected sandwich panels, was obtained. Comparison of analytical values between connected panels and a unique slab shows that using the proposed connection system results in decreasing the midspan deflection in 7.17% and 9.48% for two and three panels, respectively. This is the consequence of the influence of using the connector profile (see Fig. 2c), which increases the flexural stiffness and shear stiffness of the connected panels. 5.4. Residential modular prototype As demonstrated in Section 5.2, for the present solution of sandwich panels, the contribution of shear deformation for the total deflection is marginal. Hence, neglecting shear effects in the evaluation of the total deflection of the modular prototype seems perfectly acceptable. Accordingly, the classical laminate plate theory (CLPT) can be used to analytically predicte the floor prototype deformational behaviour. The prototype was considered to be subjected to an uniform distributed load. In addition, it was considered that the sandwich panels were subjected to two kinds of support conditions: (i) the exterior panels (FP1 and FP3) were considered as simply supported panels along three edges, while the other edge of these panels was assumed free of any displacement restriction, (ii) middle panel (FP2) was considered as a panel supported along only two edges, being the other two edges free of any displacement restriction. Accordingly, Eqs. (11) and (12) are proposed for calculating the midspan deflection in panels FP1-FP3 (exterior panel) and FP2 (interior panel), respectively. Comprehensive information of the derivation methods can be consulted elsewhere [40, 42, 47]. 4 max 0.01302() eq qL wEI  (11) 4 max 0.0071() eq qL wEI  (12) where q is the uniform pressure load, L is the length of the panel. From Eq. (11) a deflection at midspan of the middle panel (maximum deflection) of 18.7 mm was computed for a uniform load of 2.4 kN/m2 and a flexural stiffness determined from Eq. (9). This value is close to the maximum slab deflection of 19.57 mm which experimentally was obtained (19.57 mm). In the case of exterior panels, the maximum deflection was occurred at the middle span of the free edge, and resulted in a value of 10.20 mm, calculated based on Eq. (12). Experimental observations showed a deflection of 12.58 mm. Therefore, experimental results obtained in the exterior panels are also coherent with the analytical ones. Finally, load distribution factor ( LDF  ) was evaluated to compute the working proportionality of the floor prototype in each longitudinal and transversal direction. This factor was assessed on the longitudinal and transversal GFRP beams according to Eq. (13). maxExp LDF     (13) where Exp  is the experimentally measured beam deflection in ULS condition ,     4 max 5 384( ) eb qL EI   is the maximum expected midspan deflection of the beam, 2 eB L L e   is the effective length of the beam, () b EI is the flexural stiffness of the beam, L is the length of the beam and B e is the distance between centre of the bolt (bolts were used to connect beam element to the column element as indicated in section 2.1) and end of the beam, equal to 55 mm. Considering q , B e L equal to 2.4 kN/m2, 55 mm and 3000 mm, respectively, an expected midspan deflection of 18.72 mm was obtained on the beams. Since deflection values experimentally measured on the transversal and longitudinal beams was 13.8 mm and 4.6 mm, respectively, the value of LDF  was calculated as 0.75 and 0.25 in transversal and longitudinal beams, respectively. 7. Conclusion This paper has presented a novel modular prototype floor to be used as a part of an emergency house. The prototype consists of a skeleton of GFRP tubular pultruded profiles, and a floor formed by three sandwich panels with GFRP skins and a PU foam core. A fitting connection system is utilized to appropriately assemble the different components. The developed prototype is capable of being prefabricated and easily transported to the site, and rapidly installed. This functionality illustrates the high potentiality of this system to be used in prefabricated emergency houses. An experimental program has been conducted, studying the behavior of a single floor panel, two and three connected panels and the whole modular floor prototype, subjected to residential service loads. Additionally, an analytical assessment has been developed to conduct a deeper study of the failure mechanisms, the influence of placing U-shape GFRP profiles inside the panels, the efficiency of the connections between panels and the deformational behavior of the prototype. The following concluding remarks were drawn from this work: 1. The prototype demonstrates a high potentiality to be integrated in the production lines for temporary residential building, and more specifically to be used in places with high demanding for habitation due to natural disasters. The assembly/disassembly process is simple and fast due to the lightness of the components and simplicity of the connections. 2. For the considered load levels, typical of a building structure, the sandwich panels have presented a linear-elastic behaviour. Their maximum deflection under service loads, taking into account the creep effects, fulfils the requirement established by standard. 3. The ultimate load carrying capacity of the sandwich panels was substantially greater than the design demand levels. The failure has occurred due to a local outward buckling, known as wrinkling. The failure has started when the outward tensile stress between the GFRP skin and PU foam core has attained the tensile strength of the PU. The debonding has propagated towards the centre of the panel, leading to a loss of integrity between GFRP skin and PU foam core. 4. The behaviour of connected two and three sandwich panels has exhibited adequate flexural performances and has fulfilled the requirements in both SLS and ULS, in terms of deflection and strain. Moreover, the proposed connection system has demonstrated to be effective in transferring loads between the panels, guaranteeing deformation compatibility. 5. The floor prototype has presented a flexural behaviour more predominant in one direction. However, beam-panel and panel to panel connectors contributed for the development of a certain level of transversal curvature, therefore the floor panels have behaved as a two-way spanning slab with load distribution factor of around 75% and 25% for the longitudinal and transversal supporting beams, respectively. The excellent performance showed by the proposed prototype, along with the fulfilment of long-term behaviour requirements, highlighted the potential capacity of the proposed system for being used as a temporary floor building. 6. A theoretical prediction employing Winkler hypothesis and utilizing mechanical properties of the constituent materials were employed to evaluate the failure mode of single sandwich panel. Accordingly, two kinds of stresses, namely interfacial out-of-plane stress and critical wrinkling stress were evaluated. It was shown that the interfacial out-of-plane stress between the PU foam core and the GFRP skins occur has exceeded the tensile strength of the PU foam, justifying the debonding of the top GFRP skin observed experimentally. The calculated critical wrinkling stress based on the proposed equations was in good agreement with the experimental measured values. 7. The first-order shear deformation theory was employed to predict the deformational behavior of single sandwich panel, as well as connected sandwich panels. A very good agreement between the experimental and analytical values was observed The importance of using U-shape GFRP profiles inside the sandwich panels as ribs for increasing the flexural stiffness of the panel was clearly observed. The analytical analysis has showed that when no U-shape GFRP profile is used the contribution of flexural and shear deformation for the total deflection of the panels is 60% and 40%, respectively. However, the contribution of the shear deformation for the total deflection of the panels has decreased significantly with the increase of the number of the U-shape GFRP profiles. For the four profiles per panel (the number used in the developed panels) this contribution was 3%. Additionally, it was observed that using a connector in two and three connected panels resulted in decreasing middle span deflection when compared to a unique panel. Acknowledgements This work is part of the research project ClickHouse - Development of a prefabricated emergency house prototype made of composites materials, involving the company ALTO – Perfis Pultrudidos, Lda., CERis/Instituto Superior Técnico and ISISE/University of Minho, supported by FEDER funds through the Operational Program for Competitiveness Factors – COMPETE and the Portuguese National Agency of Innovation (ADI) - project no. 38967. Special thanks are given to company ALTO who manufactured all the elements (GFRP profiles and sandwich panels) involved in this work. The authors also want to express their gratitude to engineer Luís Oliveira, for his invaluable collaboration, help and support in this research. References [1] Johnson C. Impacts of prefabricated temporary housing after disasters: 1999 earthquakes in Turkey. Habitat International. 2007;31:36-52. [2] Arslan H, Cosgun N. Reuse and recycle potentials of the temporary houses after occupancy: Example of Duzce, Turkey. Building and Environment. 2008;43:702-9. [3] Barakat S. Housing reconstruction after conflict and disaster.: London: Overseas Development Institute., 2003. [4] Dodoo A, Gustavsson L. Life cycle primary energy use and carbon footprint of wood-frame conventional and passive houses with biomass-based energy supply. Applied Energy. 2013;112:834-42. [5] Datin PL, Prevatt DO. Using instrumented small-scale models to study structural load paths in wood-framed buildings. Engineering Structures. 2013;54:47-56. [6] Imperadori M, Salvalai G, Pusceddu C. Air Shelter House Technology and its Application to Shelter Units: the Case of Scaffold House and Cardboard Shelter Installations. Procedia Economics and Finance. 2014;18:5529. [7] Ljunggren F, Ågren A. Potential solutions to improved sound performance of volume based lightweight multi-storey timber buildings. Applied Acoustics. 2011;72:231-40. [8] Winandy JE, Hunt JF, Turk C, Anderson JR. Emergency housing systems from three-dimensional engineered fiberboard : temporary building systems for lightweight, portable, easy-to-assemble, reusable, recyclable, and biodegradable structures. General Technical Report FPL-GTR-166. Madison, WI: U.S. Department of Agriculture, Forest Service, Forest Products Laboratory2006. [9] Zenkert D. An intoduction to sandwich construction. Stockholm: Cradley Heath, Warley:Engineering Materials Advisory Services., 1995. [10] Vinson JR. The behavior of sandwich structures of isotropic and composite materials. United States of America: Technomic publishing company,Inc, 1999. [11] Kootsookos A, Burchill PJ. The effect of the degree of cure on the corrosion resistance of vinyl ester/glass fibre composites. Composites Part A: Applied Science and Manufacturing. 2004;35:501-8. [12] Nguyen CH, Chandrashekhara K, Birman V. Multifunctional thermal barrier coating in aerospace sandwich panels. Mechanics Research Communications. 2012;39:35-43. [13] Allard JF, Atalla N. Propagation of Sound in Porous Media: Modelling Sound Absorbing Materials2009. [14] Sousa JM, Correia JR, Cabral-Fonseca S, Diogo AC. Effects of thermal cycles on the mechanical response of pultruded GFRP profiles used in civil engineering applications. Composite Structures. 2014;116:720-31. [15] Correia JR, Cabral-Fonseca S, Branco FA, Ferreira JG, Eusébio MI, Rodrigues MP. Durability of pultruded glass-fiber-reinforced polyester profiles for structural applications. Mechanics of Composite Materials. 2006;42:325-38. [16] Rizkalla S, Lucier G, Dawood M. Innovative use of FRP for the precast concrete industry. Advances in Structural Engineering. 2012;15:565-74. [17] Oppe MW, Knippers J. Application of bolted connections in fibre-reinforced polymers. Proceedings of the Institution of Civil Engineers: Structures and Buildings. 2011;164:321-32. [18] Shawkat W, Honickman H, Fam A. Investigation of a novel composite cladding wall panel in flexure. Journal of Composite Materials. 2008;42:315-30. [19] Sharaf T, Shawkat W, Fam A. Structural performance of sandwich wall panels with different foam core densities in one-way bending. Journal of Composite Materials. 2010;44:2249-63. [20] Correia JR, Garrido M, Gonilha JA, Branco FA, Reis LG. GFRP sandwich panels with PU foam and PP honeycomb cores for civil engineering structural applications: Effects of introducing strengthening ribs. International Journal of Structural Integrity. 2012;3:127-47. [21] Keller T, Haas C, Valĺe T. Structural concept, design, and experimental verification of a glass fiberreinforced polymer sandwich roof structure. Journal of Composites for Construction. 2008;12:454-68. [22] Mousa MA, Uddin N. Global buckling of composite structural insulated wall panels. Materials & Design. 2011b;32:766-72. [23] Garrido M, Correia JR, Keller T, Branco FA. Adhesively bonded connections between composite sandwich floor panels for building rehabilitation. Composite Structures. 2015;134:255-68. [24] ASTM. Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials. ASTM D3039/D3039M-14. United States: West Conshohocken, PA 19428-2959; 2000a. [25] ASTM. Standard Test Method for Flatwise Compressive Properties of Sandwich Cores. ASTM C365. United States: West Conshohocken, PA 19428-2959; 2005a. [26] ASTM. Standard Test Method for Flatwise Tensile Strength of Sandwich Constructions. ASTM C297-04. United States: West Conshohocken, PA 19428-2959; 2010. [27] Garrido M. Composite sandwich panel floors for building rehabilitation. UNIVERSIDADE DE LISBOA: INSTITUTO SUPERIOR TÉCNICO; 2015. [28] ASTM. Standard Test Method for Tensile Strength of Concrete Surfaces and the Bond Strength or Tensile Strength of Concrete Repair and Overlay Materials by Direct Tension (Pull-off Method) ASTM C1583. United States: West Conshohocken, PA 19428-2959; 2005b. [29] ASTM. Standard Test Method for Tensile Properties of Plastics. ASTM D638. United States: West Conshohocken, Pennsylvania 19428-2959; 2004. [30] ASTM. Standard Test Method for Flexural Properties of Sandwich Constructions. ASTM C393. United States,West Conshohocken, PA 19428-2959,2000. [31] Zenkert D, Shipsha A, Persson K. Static indentation and unloading response of sandwich beams. Composites Part B: Engineering. 2004;35:511-22. [32] Petras A, Sutcliffe MPF. Indentation resistance of sandwich beams. Composite Structures. 1999;46:413-24. [33] Rizov V, Shipsha A, Zenkert D. Indentation study of foam core sandwich composite panels. Composite Structures. 2005;69:95-102. [34] Petras A, Sutcliffe MPF. Indentation failure analysis of sandwich beams. Composite Structures. 2000;50:311-8. [35] Borsellino C, Calabrese L, Valenza A. Experimental and numerical evaluation of sandwich composite structures. Composites Science and Technology. 2004;64:1709-15. [36] Fam A, Sharaf T. Flexural performance of sandwich panels comprising polyurethane core and GFRP skins and ribs of various configurations. Composite Structures. 2010;92:2927-35. [37] Council INR. CNR, Guide for the Design and Construction of Structures made of FRP Pultruded Elements. ROME2008. [38] Abdolpour H, Garzón-Roca J, Escusa G, Sena-Cruz J, Barros J, Valente I. Development of a composite prototype with GFRP profiles and sandwich panels used as a floor module of an emergency house. Journal of Composite Structures: Elsevier; 2016. [39] Mousa MA, Uddin N. Flexural behavior of full-scale composite structural insulated floor panels. Advanced Composite Materials. 2011a;20:547-67. [40] Allen GH. Analysis and Design of Structural Sandwich Panels. . London: Pergmon Press, 1969. [41] Zenkert D. The Handbook of Sandwich Construction. London, United Kingdom: Chameleon Press Ltd., 1997. [42] Carlsson LA, Kardomateas GA. Structural and Failure Mechanics of Sandwich Composites. New York Springer, 2011. [43] Mousa MA, Uddin N. Structural behavior and modeling of full-scale composite structural insulated wall panels. Engineering Structures. 2012;41:320-34. [44] The Buckling of Sandwich-Type Panels. Journal of the Aeronautical Sciences. 1945;12:285-97. [45] Mousa MA, Uddin N. Debonding of composites structural insulated sandwich panels. Journal of Reinforced Plastics and Composites. 2010. [46] Eskandari H. The effect of shear deformation on material selection for bending components. Materials & Design. 2003;24:143-9. [47] Timoshinko S. Theory of plates and shells. New York: McGRAW-HILL Book company, 1989. [48] Ginjeira do Nascimento J. Transitional architecture in emergency scenarios. Clickhouse case study: modular shelter made of advanced composite materials. Portugal: Técnico Lisboa, 2015. List of tables Table 1. Materials used in sandwich wall panels. Table 2. Registered deflection and strain in the floor prototype. Table 3. Results obtained in the flexural test performed on single composite sandwich floor panels. Table 4. Flexural response of connected panels and unique sandwich slab. 120 1000 1000 1000 120 3400 120 3000 120 3400 120 1000 1000 1000 120 3400 120 3000 120 FP 1 FP 2 FP 3 1000 1000 3400 Sec. a-a Detail 1 eb=55 eb=55 Le=2890 (a) (b) 55 300 290 300 1000 300 290 300 1000 300 290 300 55 1000 Detail 2 Detail 3 Detail 4 110 110 120 120 120120 (c) 13 4213567 7 56 3 57 6 8 Detail 1 Detail 2 Detail 3 Detail 4 Legend: (1) GFRP beam; (2) GFRP column; (3) GFRP square profile; (4) steel profile; (5) GFRP skin; (6) GFRP U profile; (7) PU foam core; (8) steel bolt Fig. 9. Schematic presentation of the floor prototype: (a) frame structure; (b) frame structure and sandwich floor panels; (c) Cross section a-a (all units in mm). Fig. 10. Stages of the assembling process:(a) placing columns; (b) connecting beams to the columns; (c) mounting panels along beam-panel connection; (d) placing panel-panel connector; (e) installing the last beam; (f) final prototype. Fig. 4. Material characterization test setups: (a) GFRP tensile test; (b) PU compressive test; (c) PU tensile test; (d) PU shear test; (e) pull off test; (f) epoxy tensile test. 500 500 S1/S4 S2/S5 S3/S6 D3 D2 D1 D5D4 S: strain gauge and D: LVDT S1-S3 on bottom skin and S4-S6 on top skin 150 850 500 500 850 150 (a) (b) Fig. 5. Single panel four-point bending test: (a) test setup; (b) instrumentation. 500 500 S2 S1/S5 D3 D2 D1 D5D4 500 500 S4/S6 S3 D8 D7 D6 D10D9 500 500 150 S2 S1/S8 850 500 500 850 150 D3 D2 D1 D5D4 500 500 S5 S4/S9 S3 D8 D7 D6 D10D9 500 500 S7/S10 S6 D13 D12 D11 D15D14 150 850 500 500 850 150 Two adjusted panels Note: S1-S4 on bottom skinS5-S6 on top skin Three adjusted panels Note: S1-S7 on bottom skinS8-S10 on top skin (d) (e) Fig. 6. Connected panel flexural test: (a) two connected panels; (b) three connected panels; (c) connection study; (d) instrumentation for two connected panels; (e) instrumentation for three connected panels. (a) S2 S1 D2 D1 S5 S4 S3 D5 D4 D3 S7 S6 D7 D6 D11 D9 D10 D8 S8 S9 S10 S11 S12 S15 S13 S14 500 500 500 500 500 500 120 120 C2 C1 C4 C3 Beam 1 Beam 2 Beam 3 Beam 4 3240 120 3000 120 3240 FP 3 FP 2 FP 1 (b) Fig. 7. Prototype test setup: (a) loading procedure; (b) monitoring system. 0100 200 300 -16 -12 -8 -4 0 SLS_U ULS_U ULS_L SLS_L FP1 FP2 FP3 FP4 Midspan deflection (mm) Time (s) 0100 200 300 0 200 400 600 FP1 FP2 FP3 FP4 Strain (Microstrain) Time (s) SLS_L ULS_L ULS_U SLS_U (a) (b) Fig. 8. Results of single sandwich floor panel tested under uniformly distributed load: (a) time versus midspan deflection; (b) time versus strain. 030 60 90 120 150 0 10 20 30 D1 D2 D3 Load (kN) Deflection (mm) -4000 -2000 0 2000 4000 0 10 20 30 S1 S2 S3 S4 S5 S6 Load (kN) Strain (Microstrain) (a) (b) Fig. 9. Single panel tested up to its failure: (a) load versus deflection; (b) load versus strain. (a) (b) (c) Fig. 10. Single panel tested up to its failure: (a) load versus deflection; (b) load versus strain. 0 4 8 12 16 20 0 4 8 12 16 20 Load (kN) Midspan deflection (mm) Two connected panels Three connected panels 0100 200 300 400 500 0 5 10 15 20 Two connected panels Three connected panels Load (kN) Strain (Microstrain) (a) (b) Fig. 11. Flexural behaviour of connected panels: (a) load versus midspan deflection; (b) load versus strain.