International Journal of Structural Integrity

1 downloads 2 Views 2MB Size Report
and PP honeycomb cores for civil engineering structural applications: Effects of introducing ... Purpose – The purpose of this paper is to present experimental ...

International Journal of Structural Integrity Emerald Article: GFRP sandwich panels with PU foam and PP honeycomb cores for civil engineering structural applications: Effects of introducing strengthening ribs J.R. Correia, M. Garrido, J.A. Gonilha, F.A. Branco, L.G. Reis

Article information: To cite this document: J.R. Correia, M. Garrido, J.A. Gonilha, F.A. Branco, L.G. Reis, (2012),"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, Vol. 3 Iss: 2 pp. 127 - 147 Permanent link to this document: http://dx.doi.org/10.1108/17579861211235165 Downloaded on: 25-05-2012 References: This document contains references to 28 other documents To copy this document: [email protected]

Access to this document was granted through an Emerald subscription provided by ECOLE POLYTECHNIQUE FEDERALE DE LAUSANNE For Authors: If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service. Information about how to choose which publication to write for and submission guidelines are available for all. Additional help for authors is available for Emerald subscribers. Please visit www.emeraldinsight.com/authors for more information. About Emerald www.emeraldinsight.com With over forty years' experience, Emerald Group Publishing is a leading independent publisher of global research with impact in business, society, public policy and education. In total, Emerald publishes over 275 journals and more than 130 book series, as well as an extensive range of online products and services. Emerald is both COUNTER 3 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation. *Related content and download information correct at time of download.

The current issue and full text archive of this journal is available at www.emeraldinsight.com/1757-9864.htm

GFRP sandwich panels with PU foam and PP honeycomb cores for civil engineering structural applications

GFRP sandwich panels

127

Effects of introducing strengthening ribs J.R. Correia, M. Garrido, J.A. Gonilha and F.A. Branco Department of Civil Engineering and Architecture, Instituto Superior Te´cnico/ICIST, Technical University of Lisbon, Lisbon, Portugal, and

L.G. Reis Department of Mechanical Engineering, Instituto Superior Te´cnico/ICIST, Technical University of Lisbon, Lisbon, Portugal Abstract Purpose – The purpose of this paper is to present experimental investigations on the structural behaviour of composite sandwich panels for civil engineering applications. The performance of two different core materials – rigid plastic polyurethane (PU) foam and polypropylene (PP) honeycomb – combined with glass fibre reinforced polymer (GFRP) skins, and the effect of using GFRP ribs along the longitudinal edges of the panels were investigated. Design/methodology/approach – The experimental campaign first included flatwise tensile tests on the GFRP skins; edgewise and flatwise compressive tests; flatwise tensile tests on small-scale sandwich specimens; and shear tests on the core materials. Subsequently, flexural static and dynamic tests were carried out in full-scale sandwich panels (2.50 £ 0.50 £ 0.10 m3) in order to evaluate their service and failure behaviour. Linear elastic analytical and numerical models of the tested sandwich panels were developed in order to confirm the effects of varying the core material and of introducing GFRP ribs. Findings – Tests confirmed the considerable influence of the core, namely of its stiffness and strength, on the performance of the unstrengthened panels; in addition, tests showed that the introduction of lateral reinforcements significantly increases the stiffness and strength of the panels, with the shear behaviour of strengthened panels being governed by the ribs. The unstrengthened panels collapsed due to core shear failure, while the strengthened panels failed due to face skin delamination followed by crushing of the skins. The models, validated with the experimental results, allowed simulating the serviceability behaviour of the sandwich panels with a good accuracy. Originality/value – The present study confirmed that composite sandwich panels made of GFRP skins and PU rigid foam or PP honeycomb cores have significant potential for a wide range of structural applications, presenting significant stiffness and strength, particularly when strengthened with lateral GFRP ribs. Keywords Composite materials, Mechanical properties of materials, Strength of materials, GFRP sandwich panels, Polyurethane foam, Polypropylene honeycomb, GFRP skins and ribs, Experimental tests Paper type Research paper

International Journal of Structural Integrity Vol. 3 No. 2, 2012 pp. 127-147 q Emerald Group Publishing Limited 1757-9864 DOI 10.1108/17579861211235165

IJSI 3,2

128

1. Introduction The structure of sandwich panels follows a typical basic pattern that comprises relatively thin, stiff and strong skins enclosing a relatively thick and light core (Davies, 2001). Many alternative forms of sandwich construction are possible by combining different skin and core materials, with varying geometries, therefore enabling optimum designs to be produced for particular applications. Among the main advantages of composite sandwich panels are the high-specific strength and stiffness, the lightness, the high thermal insulation, the possibility of producing complex geometries and the improved durability provided by the fibre reinforced polymer (FRP) skins. As drawbacks, composite sandwich panels generally present low acoustic insulation, poor performance for relatively high temperatures and lack of design codes or guidelines. Sandwich construction was initially restricted to a few applications, namely in aerospace and naval industries. Subsequently, their field of application was extended to the automotive industry and, more recently, to the offshore oil and wind industries. The present need for civil engineering structures with low self-weight, high strength, and durability has been fostering the development of FRP solutions (Correia, 2008), among which composite sandwich panels are playing an increasingly important role (Almeida, 2009). Sandwich panels are being used as structural elements in both vehicular and pedestrian bridges (Davalos et al., 2001), in new constructions and in the replacement of degraded decks (de Freitas et al., 2010, 2011). For secondary structures, sandwich panels have been developed for a wide range of applications that include cladding (Shawkat et al., 2008), facades (Sharaf and Fam, 2008; Sharaf et al., 2010) roofing (Keller et al., 2008a, b) and also partition wall elements, possibly with translucent properties (Keller et al., 2004). Standard composite sandwich panels are often made of glass fibre reinforced polymer (GFRP) skins and rigid polyurethane (PU) foam core. Despite their very competitive costs, the stiffness and strength of these conventional sandwich panels are hardly compatible with their structural use in building floors or bridge decks, at least for standard spans and loads. The main weaknesses of these panels stem from the low stiffness and strength of the core, and the top skin susceptibility to delamination and buckling, owing to the local mismatch stiffness and the inexistence of reinforcements bridging the core and the skins. In order to improve the mechanical performance of standard sandwich panels, different strengthening techniques have been proposed and studied, namely: . The use of stitches connecting the top and bottom skins. . The use of reinforcing ribs. Potluri et al. (2003) studied the effects of introducing stitches and concluded that the mechanical properties of the panels, including their strength, stiffness, and fatigue behaviour can be improved depending on the type of composite material and sewing technique. Lascoup et al. (2006) also studied the effect of introducing stitches crossing the two skins and the core and concluded that stitching increases significantly the stiffness and the ultimate strength, in terms of bending, shear and flatwise compression strengths. Reis and Rizkalla (2008) and Dawood et al. (2010) studied the effect of introducing fibre reinforced stitches with a 3D architecture joining GFRP skins and foam cores and concluded that such strengthening technique is able to increase significantly the stiffness of the sandwich panel; in addition, with such technique,

the uncracked foam used as a filling material confines the fibres and contributes to increase significantly the core shear modulus. Fam and Sharaf (2010) studied different configurations of ribs and their influence on the mechanical properties of panels made of GFRP skins and PU foam core. They concluded that the ribs significantly increase the strength and stiffness of the panel. This paper presents results of further experimental investigations on the mechanical behaviour of composite sandwich panels, taking into account their possible structural use in civil engineering applications, such as building floors or footbridge decks. The panels studied are constituted by GFRP skins and the influence of the two following parameters was evaluated: (1) Characteristics of the core material – a PU rigid foam and polypropylene (PP) honeycomb core were compared. (2) Strengthening of the panels with GFRP ribs along the longitudinal edges of the panels. The experimental investigations first included material characterisation tests on small-scale specimens. Subsequently, flexural static tests were carried out in order to evaluate the service and failure behaviour of the sandwich panels. Finally, flexural dynamic tests were performed in order to evaluate the dynamic properties of the panels. Alongside the experimental programme, 3D finite element (FE) models were developed for all panels tested, allowing to simulate their serviceability behaviour (i.e. within the linear elastic domain) and further evaluate the effects of varying the core material and introducing strengthening elements. Results of experiments are compared with predictions obtained from analytical and numerical models. 2. Experimental programme 2.1 Objectives and materials In order to evaluate the performance of two alternative core materials – PU rigid foam and PP honeycomb – and the effects of introducing strengthening ribs on the longitudinal edges of the panels, the following four types of composite sandwich panels made of GFRP skins were produced and studied: . Two standard sandwich panels, without lateral reinforcements, constituted by a core of either PU or PP (panels PU-U and PP-U, respectively). . Two sandwich panels comprising GFRP ribs, each one with the aforementioned core materials (panels PU-R and PP-R, respectively). The PU foam used has a nominal density of r ¼ 70 kg/m3. The supplier did not report any mechanical properties. However, an average shear modulus of G ¼ 6.5 MPa was estimated from the literature for PU foams with similar density (Davies, 2001; Zenkert, 1997). This value is consistent with the shear modulus of G ¼ 6.97 MPa reported by Huang and Gibson (1991) for PU foams with a density of r ¼ 64 kg/m3. The aforementioned author additionally reports a shear strength of tu ¼ 0.3 MPa. The PP honeycomb, supplied by NidaCore (n.d.) (reference H8HP), has the following properties: density of r ¼ 110 kg/m3; shear modulus of G ¼ 9 MPa; shear strength of tu ¼ 0.6 MPa; tensile strength of stu ¼ 0.6 MPa; compressive strength of scu ¼ 2.4 MPa; and compressive modulus of Ec ¼ 50 MPa.

GFRP sandwich panels

129

IJSI 3,2

130

2.2 Specimen manufacturing The sandwich panels were produced using the hand lay-up technique by company ALTO – Perfis Pultrudidos, Lda. The GFRP skins were manufactured using three different types of mats, embedded in an isopthalic polyester resin matrix: (1) Surface veil mats (SVM, 40 g/m2). (2) Chopped strand mats (CSM, 300 g/m2). (3) 08/908 symmetric woven fabric mats (WFM 800 g/m2). A total of nine plies were used in the skins with the following stacking sequence: 3 2 SVM 6 CSM 7 7 6 7 6 6 WFM 7 7 6 7 6 6 CSM 7 5 4 WFM S

The lateral reinforcements (ribs) of panels PU-R and PP-R consisted of GFRP pultruded laminates which were bonded to the lateral faces of the panel cores using polyester resin; In addition, to further fix the GFRP laminates to the panels, the two top plies of the upper skin were folded along the border of the bottom skin in a width of approximately 15 cm. Figure 1 shows the basic geometry of the unstrengthened and the strengthened panels. Figure 2 shows different stages of the manufacturing of panel PU-U. The nominal dimensions of the panels were as follows: core thickness tc ¼ 90 mm; skins thickness ts ¼ 7 mm; and ribs thickness tr ¼ 6 mm. The manufacturing process used resulted in some variability of the GFRP skins geometry: although an average thickness of 7 mm was guaranteed, locally thicknesses ranging from 5 to 9 mm were measured (systematic measurements indicated a coefficient of variation of 18.8 percent),

(a)

Figure 1. Basic geometry

(b)

Note: (a) unstrengthened panels and (b) strengthened panels

GFRP sandwich panels

131

Notes: (i) Moulding table; (ii) application of the first layer of polyester matrix; (iii)-(v) application of alternating layers of glass reinforcement and polyester matrix; (vi) and (vii) positioning of the PU foam core; (viii) steel weights placed on top ofthe PU foam

Figure 2. Different stages of the manufacturing process of panel PU-U

and such variation may have influenced their mechanical performance. The panels were 2.50 m long and had a 0.50 m width (b). 2.3 Test programme The experimental programme first included material characterization tests on small-scale sandwich panels and their constituent materials in order to evaluate the basic material properties. In particular, the following tests were performed: . Flatwise tensile tests on the GFRP skins. . Flatwise compressive. . Edgewise compressive. . Flatwise tensile tests on sandwich specimens. . Shear tests on the cores (Figure 3). Subsequently, static and dynamic flexural tests were carried out on full-scale sandwich panels, in order to evaluate their structural behaviour and some of their most relevant

Notes: (i) Tensile tests on skins; (ii) flatwise and (iii) edgewise compression and (iv) flatwise tensile tests on sandwich specimens; (v) shear tests on cores

Figure 3. Material characterisation tests

IJSI 3,2

132

mechanical properties (stiffness, strength, failure modes), as well as to investigate their dynamic behaviour. 3. Material characterisation tests 3.1 Tensile tests Tensile tests were carried out according to ISO 527-1,4 standard (ISO 527, 1997) on GFRP laminates extracted from the skins of the panels. Specimens with a length of 300 mm, a width of 25 mm, a thickness varying from 6 to 8 mm, and a grip distance of 150 mm were loaded in the longitudinal direction using an Instron universal hydraulic testing machine (with a load capacity of 250 kN). The tests were performed under displacement control at a speed of 0.18 mm/s and a grip pressure of 40 bar. During the tests, the load and displacement of the machine were registered on a PC using a HBM data acquisition system with eight channels, Spider8 model. In some specimens, the axial strains in the longitudinal direction were also measured with TML electrical strain gauges. Figure 4 shows the axial stress vs strain curves of specimens with thicknesses of 6 mm (T6) and 8 mm (T8). As expected, the GFRP laminates presented linear-elastic behaviour almost up to failure, which occurred in a brittle manner. These tests allowed determining the tensile modulus (Et), the tensile strength (stu) and the strain at failure (1tu) of the GFRP skins in the longitudinal direction (Table I) – these properties are also valid in the transverse direction owing to the symmetric architecture of the reinforcing mats. Results obtained show that there is a considerable dependency of the mechanical properties of the GFRP laminates on the skins thickness, with thinner specimens exhibiting higher mechanical performance. This result is most likely due to the higher compaction and lower volume of voids in thinner specimens. 3.2 Flatwise compression tests The flatwise compressive behaviour, which is particularly relevant in the support zones or under concentrated loads, was evaluated according to test standard

Figure 4. Tensile tests on GFRP skins: axial stress-strain curves for 6 and 8 mm thick specimens

Table I. Tensile properties of the GFRP skins (average ^ standard deviation)

Thickness (mm) 6 8

Et (GPa)

stu (MPa)

1tu (mstrain)

20.5 ^ 0.9 12.2 ^ 0.9

228 ^ 1.59 132 ^ 7.1

11,278 ^ 843 13,892 ^ 601

(ASTM C365, 2003). Compressive tests were performed on 100 £ 100 mm2 sandwich specimens (six for the PP core and five for the PU core). The load was applied with the same machine used in tensile tests and the values of the load and displacement were registered with the same data acquisition system (also used in the remaining tests). The load-vertical displacement curves (Figure 5) show that both core materials initially exhibited an approximately linear behaviour up to the maximum load, which had an average value of 24.1 kN for the PP honeycomb specimens and 3.01 kN for the PU foam specimens (about 12 percent of PP strength). Then, after a load reduction, that was less pronounced in the PU specimens, the load-deflection curves of both materials showed a plasticity plateau with an increase of displacement for approximately constant loads of about 10-12 and 2.75 kN for the PP and PU cores, respectively. At this stage no signs of failure were distinguished in the PU foam; however, for the PP core, it was possible to observe progressive crushing/buckling of the honeycomb cells (Figure 5). Residual deflections after complete unloading were relatively high for both cores, due to considerable plastic deformations. The load-displacement curves allowed determining the following mechanical properties in flatwise compression (Table II): stiffness (Kc,f), strength (scu,f), and apparent elasticity modulus Eap c,f , the latter based on the assumption that the GFRP skins are rigid. Although the densities of the core materials are different, results show that PP honeycomb is significantly stiffer (10.4 times) and stronger (eight times) than rigid PU foam for flatwise compressive loading. Regarding the PP honeycomb core, while the compressive strength measured in the tests matches the value provided by 30

133

PP

PP PU

25

GFRP sandwich panels

Load [kN]

20 15 PU

10 5

Figure 5. Flatwise compression tests: load-deflection curves and failure modes

0 0

5

10 15 Deflection [mm]

20

25

Test

Core

K (kN/mm)

su (MPa)

Eap (MPa)

Flatwise compression

PP PU PP PU PP PU

10.4 ^ 1.5 1.0 ^ 0.1 65.5 ^ 10.8 68.0 ^ 9.6 13.31 ^ 1.72 1.62 ^ 0.11

2.4 ^ 0.16 0.3 ^ 0.01 8.19 ^ 0.77 4.70 ^ 1.11 0.76 ^ 0.02 0.35 ^ 0.04

93.6 ^ 13.5 9.0 ^ 0.9 – – – –

Edgewise compression Flatwise tension

Table II. Mechanical properties of sandwich panels for flatwise and edgewise compression and flatwise tension

IJSI 3,2

134

the manufacturer, for the compressive modulus the value obtained in the test is almost the double of that indicated in the technical sheet (Nidacore, n.d.). 3.3 Edgewise compression tests Edgewise compression tests were carried out according to ASTM C364 (1999) standard, in order to evaluate the in-plane compressive behaviour of the sandwich panels made of both core materials. Tests were performed on specimens (six for each type of core) with nominal dimensions of 250 £ 250 £ 104 mm3. The loaded surfaces (250 £ 104 mm2) were levelled in order to avoid any non-uniform or transverse loading. Load was applied with the same hydraulic machine used in the previous tests. In some specimens, two displacement transducers were placed at the centre of each skin to measure their out-of-plane horizontal displacements therefore identifying the beginning of their instability. The load-axial deflection curves of both types of specimens (Figure 6) show a non-linear initial behaviour most likely due to the loading system, namely the adjustment of the metal plates to the specimen. Then, load increased up to a value of about 110 kN and, subsequently, the curves exhibit a small segment with deflection increase for an approximately constant load. After this small branch, which corresponded to a considerable increase of horizontal deflection (as attested by displacement transducers measurements), load increased again almost linearly up to failure. Failure of PP honeycomb specimens occurred due to either: (1) buckling of the skins (Figure 6), sometimes followed by delamination, with core-skin debonding in the specimen intrados; or (2) by crushing of the skins near the metal plates. In most PU foam specimens, failure was triggered by buckling of the skins, followed by delamination and, in some cases, shear failure of the core (Figure 6) and/or crushing of the skins next to the metal plates of the testing machine were also observed. Based on the load-deflection curves, the axial compressive stiffness (Kc,e) and the edgewise compressive strength (scu,e) were determined (Table II). The average failure PU

300 PP PU

250

C+.PU.4

Load [kN]

200 150 PP 100 50

Figure 6. Edgewise compression tests: load-deflection curves and failure modes

0 0

1

2

4 3 Deflection [mm]

5

6

7

load of PP honeycomb specimens (212.9 kN) was approximately twice than that of PU core specimens (122.3 kN) – such difference most likely stems from the higher flexibility of the PU foam in the out-of-plane direction (causing much less restriction to the skins deformation) and also the higher flatwise tensile strength provided by the PP core (Section 3.4). The axial stiffness of the two types of panels was similar, as it depends mainly on the skins. Simple elasticity theory calculations show that the axial compressive stiffness of the tested panels is considerably lower than that of current concrete or timber floors (Almeida, 2009); such result points out the need to adopt thicker skins if sandwich panels are to be used in building floors. For the panels/cores used in the present study, a similar axial compressive stiffness to that of current timber floors may be obtained by doubling the skins thickness and/or by adopting longitudinal GFRP ribs.

GFRP sandwich panels

135

3.4 Flatwise tensile tests Flatwise tensile tests were carried out according to ASTM C297 (2004) standard on 100 £ 100 mm2 sandwich specimens (three for each type of core), whose GFRP skins were bonded to two thick steel plates with an epoxy adhesive supplied by SIKA (Sikadur 330). Test specimens were prepared guaranteeing a very accurate alignment between all bonded parts. Tensile loads were applied by pulling the steel bars that had been previously welded to the centre of the steel plates. Figure 7 shows the load-deflection curves that allowed determining the initial stiffness (Kt,f) and strength (stu,f) of both types of specimens for flatwise tension (Table II). It can be seen that PU foam exhibited linear-elastic behaviour up to failure, which was triggered within the core where cracking was first observed (Figure 7), while PP honeycomb cores first presented linear response (with considerably higher stiffness compared with the PU core), and then progressively started exhibiting non-linear behaviour up to failure (which occurred also for a much higher load). In the specimens with the PP core, failure occurred at the interface between the core and one of the GFRP skins (Figure 7). The tensile strength measured for this type of specimens was slightly higher than the value provided by the manufacturer for the tensile strength of the PP core, which is 0.6 MPa (Nidacore, n.d.). 10

PP PU

PP core

8 GFRP sKin Load [kN]

6 4 GFRP sKin 2

PP core

0 0.0

0.5

1.0

1.5 2.0 2.5 Deflection [mm]

3.0

3.5

4.0

GFRP sKin

Figure 7. Flatwise tensile tests: load-deflection curves and failure modes

IJSI 3,2

136

3.5 Shear tests Shear tests were carried out based on ASTM C273 (2000a) standard on 800 mm long £ 50 mm wide specimens (five for each type of core), which were bonded to two steel T-shaped profiles with the same epoxy adhesive used in the flatwise tensile tests. Tensile loads were applied by pulling a fixing device that was bolted to the steel profiles. This device allowed for specimen rotation in the shear plane. During the tests some slipping in the grips was observed. Figure 8 shows the load-deflection curves obtained in the shear tests. Regarding the PU core, the load-deflection curves were linear-elastic up to failure (with the exception of one specimen), which was always caused by premature debonding between the PU core and the steel profile (thereby preventing the determination of the core shear strength). PP specimens also exhibited initial linear-elastic behaviour and premature debonding was also observed in two of the specimens. In the remaining three specimens, load-deflection curves progressively started to exhibit a non-linear response which corresponded to considerable distortion of the PP core. In those specimens failure was due to the progressive separation between the PP cells (Figure 8), enabling the determination of the shear strength of the core. Table III presents the main parameters obtained from this test, namely the stiffness (Ks), the shear modulus (Gc) and the shear strength (tu) of the core materials. Owing to the slipping of the fixing device and the premature debonding of PU specimens, values presented for Ks and Gc are an underestimate of the actual material properties. Although affected by the slipping of the fixing system, the stiffness values point out the higher shear stiffness of the PP honeycomb when compared to the PU core. Additionally, it is worth mentioning that the average shear strength of the PP core is 25

PP PU

Load [kN]

20

PP

PU

15 10 5

Figure 8. Shear tests: load-deflection curves and failure modes Table III. Mechanical properties of the core materials in shear (for the PU core, obtained from five specimens; for the PP honeycomb, obtained from five specimens (K and G) or three specimens (tu))

0 0

5

10 15 20 25 Deflection [mm]

30

35

Core

K (kN/mm)

G (MPa)

tu (MPa)

PP PU

. 2.97 ^ 0.45 . 1.69 ^ 0.18

. 6.92 ^ 1.09 . 3.96 ^ 0.37

0.52 ^ 0.03 . 0.20 ^ 0.03

slightly lower than the value indicated in the manufacturer technical sheet, which is 0.6 MPa (Nidacore, n.d.). 4 Full-scale flexural tests on sandwich panels 4.1 Static tests 4.1.1 Test setup. Full-scale flexural tests were carried out on each type of panel (one specimen for each type) according to ASTM C393 (2000) standard, in a four-point bending configuration (Figure 9). The sandwich panels, which were 2.50 m long £ 0.50 m wide £ 0.104 m thick, were tested in a 2.30 m span and the loaded sections were distanced of 0.76 m. The supports were materialized by steel rollers with top steel plates (0.06 £ 0.50 m2) – both end supports allowed for free rotation and one of them also allowed for longitudinal sliding. Load was applied with a 300 kN Enerpac hydraulic jack and was measured with a 200 kN Novatech load cell. Steel spreading plates (0.10 m wide £ 0.50 m long) were used between the hydraulic jack and the top skin to guarantee uniform load application throughout the width of the sandwich panel. Sandwich panels were monotonically loaded up to failure. During the tests vertical deflections at midspan and under loaded sections were measured with 100 mm stroke TML displacement transducers. Axial strains at both top and bottom skins were measured at midspan with four TML strain gauges, bonded at a distance of 7.5 cm from the centre of the panel to avoid any interference with the displacement transducer. 4.1.2 Results and discussion. Figure 10(a) shows the load-deflection at midspan curve of all sandwich panels subjected to four-point bending tests. All panels exhibited an approximately linear behaviour up to failure, with a slight stiffness reduction prior to collapse. Such linearity was also attested by strain gauge measurements (e.g. Figure 10(b) for panel PU-U) that also reflect the symmetry of the panels – neutral axis is located approximately at mid-height. Figure 11 shows the failure modes of the different panels. Both unstrengthened panels (PP-U and PU-U) collapsed due to shear failure of the core, followed by skin-core delamination. In panel PP-U a vertical fracture surface developed within the honeycomb cells, at a distance of about 50 cm from one of the supports, while in panel PU-U the failure surface developed at an angle of about 458 at an average distance of

GFRP sandwich panels

137

Figure 9. Static flexural test setup

IJSI 3,2

Section height [mm]

PU-U PP-U PU-R PP-R

80

Load [KN]

138

100

60 40 20 0 0

20

40

60

80

Midspan deflection [mm] (a)

Figure 10. Static flexural tests

100

100 90 80 70 60 50 40 30 20 10 0 –3,000 –2,000 –1,000

F = 5 KN F = 10 KN F = 15 KN F = 20 KN F = 25 KN F = 30 KN

0

1,000 2,000 3,000

Axial strain [mstrain] (b)

Notes: (a) Load vs midspan deflection curves for all panels; (b) axial strains of panel PU-U as a function of section depth for different load levels

Figure 11. Static flexural tests: failure modes

about 10 cm from the support. The strengthened panels (PP-R and PU-R) presented a different failure mode: . The compression stresses in the top skin first led to its debonding from the core at midspan section. . Then, the unbonded part of the top skin soon started to wrinkle. . Final failure was triggered by the crushing/delamination of the fibres, starting from the tip of the wrinkle, with the crack then developing along the width of the top skin and propagating towards the lateral reinforcements. None of the strengthened panels showed any signs of core failure. Table IV lists, for all panels tested, the load (Fu) and deflection (du) at failure, the initial (elastic) stiffness (K), the maximum longitudinal stress in the skins (sx,max,

based on strain measurements) and, in addition, the maximum shear stresses in the core of the unstrengthened panels (txz,max). The last two parameters were determined based on the elasticity theory for sandwich structures (Allen, 1969). In terms of deformability, panel PP-U was slightly stiffer (about 25 percent) than panel PU-U, which is in accordance with the higher stiffness of the PP core (attested in both shear and compressive tests). The stiffness of the strengthened panels was very similar for the two different core materials and, as expected, the lateral GFRP ribs led to considerable stiffness increases of 71 and 129 percent for the PP and PU cores, respectively. In the strengthened panels, the core contribution for overall stiffness (and, in particular, for shear stiffness) is considerably reduced – this role is played essentially by the lateral ribs. Therefore, the slightly higher stiffness of panel PU-R compared to panel PP-R (contrasting with the lower stiffness of panel PU-U when compared to panel PP-U) is most likely due to small differences in the thickness of the skins/lateral ribs. The failure load of both unstrengthened panels (PU-U and PP-U) was very similar, indicating that the shear strength of their cores (which caused collapse in both cases) was also very similar (in the shear tests, the shear strength had only been determined for the PP core). More importantly, the maximum shear stress attained in the test of panel PP-U (0.30 MPa) was only about 58 percent of the average shear strength (0.52 MPa) measured in the material characterization tests. Such difference, which is an important result of the present study, suggests the possible existence of a significant size effect in the shear strength of the PP cores. The failure load of the strengthened panels, which presented similar failure modes for the two types of cores was also much higher than that of the unstrengthened panels – the lateral GFRP ribs provided strength increases of 158 and 171 percent for the PP and PU cores, respectively. The fact that core shear failure did not occur in the strengthened panels is naturally due to the presence of the lateral GFRP ribs. As expected, these elements took a very significant portion of shear load, allowing for a major reduction of shear stresses in the core materials – this effect was later confirmed in the numerical calculations (Section 5).

GFRP sandwich panels

139

4.2. Dynamic tests 4.2.1 Test setup and methodology. The dynamic tests were performed in the same panels that were later subjected to the static tests described in the previous section. Panels were tested also in a simply supported span of 2.30 m. The dynamic excitation of the panels was caused by manual strikes, applied approximately at midspan. Vertical vibrations were measured with two accelerometers from Bruel & Kjaer, symmetrically positioned at midspan, at a distance of 0.05 m from the lateral edge of the panels. The accelerometers, with a precision of 0.01 mm, were connected to Bruel & Kjaer amplifiers (model 2635). The signal readings were performed at a rate of 400 Hz and recorded in a PC using a data acquisition unit with eight channels from HBM, Spider8 model. The output of Property

PP-U

PU-U

PP-R

PU-R

Fu (kN) du (mm) K (kN/mm) sx,max (MPa) txz,max (MPa)

28.26 51.57 0.685 43.58 0.30

31.74 72.54 0.544 48.94 0.33

72.83 72.30 1.173 78.09 –

86.13 89.16 1.247 85.81 –

Table IV. Static flexural tests: summary of results

IJSI 3,2

140

Figure 12. Dynamic flexural tests of panel PU-U

Table V. Dynamic flexural tests: summary of results

the readings was then treated using the Fast Fourier Transform (FFT) algorithm. The spectral values of the half-sum and half-difference of the accelerometers readings were used to estimate the natural flexural and torsional frequencies, respectively. These parameters were then used for further calibration and validation of the numerical models (Section 5). 4.2.2 Results and discussion. Results obtained in the dynamic tests of panel PU-U are shown in Figure 12 as an illustrative example. In particular, Figure 12(a) plots the vertical vibrations (accelerations) measured in both accelerometers (A1 and A2), while Figure 12(b) shows the FFT analysis of the half-sum and half-difference of vibration measurements; as already mentioned the peaks presented by those two parameters allow identifying the natural flexural and torsional frequencies of the panel. Table V summarizes the results obtained in the dynamic flexural tests, listing the average values (from five tests) of the first flexural and torsional frequencies. The first flexural frequencies of the unstrengthened panels (PP-U and PU-U) are very similar, which is due to the fact that their flexural stiffnesses are also similar and also because the lower shear modulus of the PU core is counterbalanced by its lower self-weight (about 64 percent of the PP core). As expected, the introduction of GFRP lateral ribs led to a considerable increase (21-25 percent) of the flexural frequencies, which were very similar for both types of cores. Regarding the torsional frequencies, among the unstrengthened panels, panel PP-U presented higher frequency than panel PU-U most likely due to the higher stiffness of its core. As for the flexural frequencies, the introduction of strengthening ribs led to a considerable increase of torsional frequencies (36-57 percent). Unlike their unstrengthened counterparts, panels PP-R and PU-R exhibited very similar torsional

(a)

(b)

Notes: (a) Vertical accelerations measured in both accelerometers (A1 and A2); (b) FFT of half-sum and half-difference of measured vibrations

Panel

Flexural frequency (Hz)

Torsional frequency (Hz)

PP-U PU-U PP-R PU-R

25.6 25.7 31.0 32.1

70.8 61.7 96.1 96.6

frequencies, as their torsional stiffness is mainly governed by the ribs, being almost independent of their core material stiffness. 5. Numerical study 5.1 Description of the numerical models 3D FE models of tested panels were developed using FE commercial package ANSYS , release 11.0 (Ansysw, 2007). All panels were modelled with the same geometry of the tested panels. For the GFRP material, an average thickness of 7 mm was considered, together with its corresponding tensile properties, namely an elasticity modulus Et ¼ 16.4 GPa (Table I). All components of the sandwich panels (skins, ribs and cores) were divided in four layers and modelled using 3D quadratic isoparametric 20-node elements with three degrees of freedom (DOF) per node (translations in directions x, y and z), which allow modelling orthotropic materials and non-linear behaviour (Figure 13(a)). The applied loads and the supports were defined as being distributed through an area corresponding to the metal plates and rollers used in the tests, thereby minimizing the development of excessive stress concentrations. Applied loads were defined as surface pressures at the top surface of the beams. For the supports, in order to accurately reproduce the displacements and rotations restraints of the tests, one-dimensional rigid bar FEs with two nodes and two DOF per node were used – these bars connected extra nodes positioned along the axis of the support rollers, and all nodes from the bottom surface of the beam located above the metal plates (Figure 13(b)). All materials were modelled assuming linear-elastic behaviour, isotropic for the PU and PP cores and orthotropic for the GFRP skins/ribs. Properties determined in the material characterization tests were used whenever available (Tables I-III), together with the following values taken from the literature: shear modulus of the PP and PU cores of GPP ¼ 9.0 MPa (Nidacore, n.d.) and GPU ¼ 6.5 MPa (Davies, 2001) (as already mentioned, in the shear tests, the slipping of the fixing system prevented the determination of the shear moduli); shear modulus and Poisson ratio of the GFRP material of Gxz ¼ 3.6 GPa and y xz ¼ 0.28, respectively. All panels were subjected to both linear-elastic static and dynamic modal analyses.

GFRP sandwich panels

141

5.2 Results and discussion 5.2.1 Static behaviour. Figure 14 shows the comparison between load-deflection curves of panels PU-U and PU-R obtained with the numerical models and the corresponding experimental data. Analytical calculations based on Timoshenko beam theory (i.e. taking into account the shear deformability of the core and ribs for panels PU-U

(a)

Notes: (a) Overall perspective; (b) detail ofthe supports

(b)

Figure 13. FE model of panel PU-U

IJSI 3,2

142 Figure 14. Load-midspan deflection curves of panels PU-U and PU-R: comparison of experimental, numerical, and analytical results

and PU-R, respectively) is also shown. It must be stressed that analytical and numerical predictions are only valid in the linear elastic domain of the panels’ response. Figure 14 shows that the numerical deflections at midspan for both PU-U and PU-R panels corresponding to the serviceability limit state (i.e. within the linear branch of the experimental curve) were very similar to corresponding experimental data. Numerical and analytical load-deflection curves obtained for panels PP-U and PP-R, namely the comparison with their experimental counterparts were also very similar. Table VI compares the measured and calculated elastic stiffness of the panels tested – relative errors for numerical and analytical predictions are relatively low, varying between 0.5-8.0 and 2.0-13.4 percent, respectively. Figure 15 shows the comparison between experimental, analytical and numerical results for axial tensile and compressive strains of panel PU-U. As for the load-deflection at midspan curves, for load levels corresponding to a serviceability response (i.e. within the linear elastic range of the panel response), a fairly good agreement was obtained for both analytical and numerical predictions (maximum relative errors of 12 and 10 percent, respectively). For higher load levels, namely in the brink of collapse, analytical and numerical load-deflection at midspan curves deviate considerably from their experimental counterparts and relative differences are now significantly higher, varying between 13 and 22 percent (numerical) and between 8 and 20 percent (analytical). Such higher differences are due to the fact that both types of models did not simulate the non-linear response of the panels and their constituent materials, namely the non-linear response of the core in the unstrengthened panels and the progressive top skin-core delamination

Table VI. Experimental (Kexp), numerical (Knum) and analytical (Kanl) elastic stiffnesses and corresponding relative errors (Dnum and Danl)

Panel

Experimental Kexp (kN/mm)

PP-U PU-U PP-R PU-R

0.685 0.544 1.173 1.247

Numerical Knum (kN/mm) dnum (%) 0.630 0.554 1.190 1.253

28.0 1.8 1.4 0.5

Analytical Kanl (kN/mm) danl (%) 0.593 0.513 1.149 1.217

213.4 29.8 2.0 22.4

GFRP sandwich panels

143 Figure 15. Axial strain curves of panel PU-U: comparison of experimental, numerical, and analytical results

and skin wrinkling in the strengthened panels. These behavioural aspects will be addressed in future investigations. The numerical models were also used to calculate the stress distributions in the panels tested for a load level corresponding to service conditions. As an example, Figure 16 shows a longitudinal section across the centre of panels PU-U and PU-R plotting the shear stresses within their core for a load level of 10 kN (for which panels tested presented linear elastic behaviour). Numerical results confirmed the effects of introducing the ribs along the lateral edges of the panels. For the unstrengthened panels, the shear stresses in the core vary considerably along the thickness, but are approximately uniform across the width. In opposition, shear stresses in the core of the strengthened panels are approximately uniform along the thickness, but vary considerably across the width, presenting lower values near the panels’ edges. In the strengthened panels, this variation is due to the fact that the GFRP lateral ribs absorb a significant proportion of shear force. In fact, as shown in Figure 16, for a 10 kN load the maximum shear stress in the core of panel PU-U is about twice than that corresponding to panel PU-R. 5.2.2 Dynamic behaviour. The numerical models were also used for modal characterization of the panels. As an example, Figure 17 plots the first flexural and torsional vibration modes of panel PU-U. Table VII lists for all panels the values of the first flexural and torsional frequencies obtained with the numerical models and the comparison of such values with experimental data.

(a)

–60

–108 –84

–12 –36

36 12

(b)

84 60

108

Figure 16. Shear stresses (in kN/m2, according to the colour scale) in the core of panels

IJSI 3,2

144

Figure 17. Numerical vibration modes of panel PU-U

Table VII. Experimental (fexp) and numerical (fnum) flexural and torsional frequencies and corresponding relative errors (D)

Results presented in Table VII show a generally good agreement between the numerical and experimental values of flexural and torsional frequencies (the average relative error considering all panels and modes was 10.6 percent). 6. Conclusions The present study confirmed that composite sandwich panels made of GFRP skins and PU rigid foam or PP honeycomb cores have significant potential for a wide range of structural applications, presenting significant stiffness and strength, particularly when strengthened with lateral GFRP ribs. Based on the experimental, analytical, and numerical investigations reported herein, the following main conclusions are outlined: (1) All sandwich panels tested in bending presented linear-elastic behaviour for considerable values of the applied load, exhibiting only a slight stiffness reduction prior to failure. (2) Unstrengthened sandwich panels made of PP honeycomb core were stiffer than those made of PU foam core, essentially due to the higher shear modulus of the PP core, as attested by the results of the shear tests. (3) The load capacity of the unstrengthened panels was very similar and was governed by the shear strength of their core. In panel PP-U, the maximum shear stress in the PP core attained in the flexural tests was only about 58 percent of the shear strength measured in the material characterization tests. Such result points out the possible existence of considerable size effects in the shear strength of PP honeycomb cores. This topic should be addressed in further depth in future studies. (4) The introduction of GFRP lateral ribs had a remarkable effect in the panels performance, as they absorbed a considerable proportion of shear load:

(a)

(b)

Note: (a) Flexural and (b) torsional modes

Panel

fexp,f (Hz)

Flexural fnum,f (Hz)

Df (%)

fexp,t (Hz)

Torsional fnum,t (Hz)

Dt (%)

PP-U PU-U PP-R PU-R

25.6 25.7 31.0 32.1

25.9 25.4 35.0 37.1

1.2 2 1.2 12.9 15.6

70.8 61.7 96.1 96.6

59.0 55.1 107.9 110.0

2 16.7 2 10.7 12.3 13.9

stiffness increased significantly (between 71 and 129 percent when compared to the unstrengthened panels); . strength also increased considerably (between 158 and 171 percent); and . vibration amplitudes and natural frequencies were also affected, with the latter having increased very significantly. (5) The ultimate strength of the strengthened sandwich panels was governed by the top skin-core adhesion and, subsequently, by the compressive strength of the unbonded top skins. (6) For service conditions, midspan deflections of sandwich panels in bending can be estimated using simple beam theory, provided that the shear contribution to overall deformability is taken into account. (7) Similarly, simple numerical models developed with commercial FE package were able to simulate with reasonably good accuracy the static and dynamic behaviour of the sandwich panels, particularly for service conditions. In future developments, a more complete consideration of the material properties of the cores, namely of their non-linear response in shear, and the incorporation of cohesive elements in the modelling of adhesive interfaces will allow improving the accuracy of the models, especially for failure conditions. .

Acknowledgements The authors wish to acknowledge the support of FCT (Grant No. PTDC/ECM/113041/2009) and ICIST for funding the research. The third author also wishes to thank FCT for the financial support through Grant No. SFRH/BD/70041/2010. The support of company ALTO, Perfis Pultrudidos, Lda. for manufacturing and supplying the GFRP sandwich panels and company SIKA for supplying the epoxy adhesive used in the material characterization tests is also acknowledged. The authors also wish to thank MSc students Ineˆs Almeida and Diogo Ferreira for their help in the experiments. References Allen, H.G. (1969), Analysis and Design of Structural Sandwich Panels, Pergamon Press, Oxford, p. 283. Almeida, I.A. (2009), “Structural behaviour of composite sandwich panels for applications in the construction industry”, MSc dissertation in Civil Engineering, Instituto Superior Te´cnico, Technical University of Lisbon, Lisbon. Ansysw (2007), Ansysw Academic Research, Release 11.0, Ansys Inc., Canonsburg, PA. ASTM C393 (2000), Standard Test Method for Flexural Properties of Sandwich Constructions, American Society for Testing and Materials, West Conshohocken, PA. ASTM C365 (2003), Standard Test Method for Flatwise Compressive Properties of Sandwich Cores, American Society for Testing and Materials, West Conshohocken, PA. ASTM C364 (1999), Standard Test Method for Edgewise Compressive Properties of Sandwich Constructions, American Society for Testing and Materials, West Conshohocken, PA. ASTM C273 (2000a), Standard Test Method for Shear Properties of Sandwich Core Materials, American Society for Testing and Materials, West Conshohocken, PA.

GFRP sandwich panels

145

IJSI 3,2

146

ASTM C297 (2004), Standard Test Method for Flatwise Tensile Strength of Sandwich Constructions, American Society for Testing and Materials, West Conshohocken, PA. Correia, J.R. (2008), “GFRP pultruded profiles in civil engineering: hybrid solutions, bonded connections and fire behaviour”, PhD thesis in Civil Engineering, Instituto Superior Te´cnico, Technical University of Lisbon, Lisbon. Davalos, J.F., Qiao, P., Xu, X.F., Robinson, J. and Barth, K.E. (2001), “Modeling and characterization of fibre-reinforced plastic honeycomb sandwich panels for highway bridge applications”, Composite Structures, Vol. 52 Nos 3-4, pp. 441-52. Davies, J.M. (2001), Lightweight Sandwich Construction, Blackwell, Oxford, p. 370. Dawood, M., Taylor, E. and Rizkalla, S. (2010), “Two-way bending behavior of 3-D GFRP sandwich panels with through-thickness fiber insertions”, Composite Structures, Vol. 92 No. 4, pp. 950-63. de Freitas, S.T., Kolstein, H. and Bijlaard, F. (2010), “Composite bonded systems for renovations of orthotropic steel bridge decks”, Composite Structures, Vol. 92 No. 4, pp. 853-62. de Freitas, S.T., Kolstein, H. and Bijlaard, F. (2011), “Sandwich system for renovation of orthotropic steel bridge decks”, Journal of Sandwich Structures and Materials, Vol. 13 No. 3, pp. 279-301. Fam, A. and Sharaf, T. (2010), “Flexural performance of sandwich panels comprising polyurethane core and GFRP skins and ribs of various configurations”, Composite Structures, Vol. 92 No. 12, pp. 2927-35. Huang, J.S. and Gibson, L.J. (1991), “Creep of polymer foams”, Journal of Materials Science, Vol. 26 No. 3, pp. 637-47. ISO 527 (1997), Determination of Tensile Properties – Part 1: General Principles. Part 4: Test Conditions for Isotropic and Orthotropic Fibre-reinforced Plastic Composites, International Organization for Standardization, Geneva. Keller, T., de Castro, J. and Schollmayer, M. (2004), “Adhesively bonded and translucent glass fiber reinforced polymer sandwich girders”, Journal of Composites for Construction, Vol. 8 No. 5, pp. 461-70. Keller, T., Haas, C. and Valle´e, T. (2008a), “Structural concept, design, and experimental verification of a glass fiber-reinforced polymer sandwich roof structure”, Journal of Composites for Construction, Vol. 12 No. 4, pp. 454-9. Keller, T., Valle´e, T. and Murcia, J. (2008b), “Function-integrated GFRP sandwich roof structure – Experimental validation of design”, Proceedings of the 4th International Conference on FRP Composites in Civil Engineering (CICE2008), Zurich, CD-ROM. Lascoup, B., Aboura, Z., Khellil, K. and Benzeggagh, M. (2006), “On the mechanical effect of stitch addition in sandwich panel”, Composites Science and Technology, Vol. 66 No. 10, pp. 1385-98. Nidacore (n.d.), “Structural honeycomb materials”, Product data sheet: H8PP and H8HP, available at: www.nida-core.com (accessed 30 June 2011). Potluri, P., Kusak, E. and Reddy, T.Y. (2003), “Novel stitch-bonded sandwich composite structures”, Composite Structures, Vol. 59 No. 2, pp. 251-9. Reis, E.M. and Rizkalla, S.H. (2008), “Material characteristics of 3-D FRP sandwich panels”, Construction and Building Materials, Vol. 22 No. 6, pp. 1009-18. Sharaf, T. and Fam, A. (2008), “Flexural load tests on sandwich wall panels with different rib configurations”, Proceedings of the 4th International Conference on FRP Composites in Civil Engineering (CICE2008), Zurich, CD-ROM. Sharaf, T., Shawkat, W. and Fam, A. (2010), “Structural performance of sandwich wall panels with different foam core densities in one-way bending”, Journal of Composite Materials, Vol. 44 No. 19, pp. 2249-63.

Shawkat, W., Honickman, H. and Fam, A. (2008), “Investigation of a novel composite cladding wall panel in flexure”, Journal of Composite Materials, Vol. 42 No. 3, pp. 315-30. Zenkert, D. (1997), The Handbook of Sandwich Construction, Engineering Materials Advisory Services, Clifton-upon-Teme, p. 442. About the authors J.R. Correia graduated in Civil Engineering and received his MSc and PhD degrees at Instituto Superior Te´cnico (IST) – Technical University of Lisbon (TUL), Portugal, where he is an Assistant Professor. His areas of interest are fibre reinforced polymers, new materials, construction technology, building pathology and rehabilitation. J.R. Correia is the corresponding author and can be contacted at: [email protected] M. Garrido, received his MSc degree in Civil Engineering at IST – TUL, Portugal, where he is a Research Assistant and PhD student. His areas of interest are fibre reinforced polymers, in particular composite sandwich panels.

J.A. Gonilha, received his MSc degree in Civil Engineering at IST – TUL, Portugal, where he is a Research Assistant and PhD student. His areas of interest are fibre reinforced polymers, in particular GFRP pultruded profiles and hybrid structures.

F.A. Branco, graduated in Civil Engineering at IST – TUL and received his MSc and PhD degrees at the University of Waterloo and IST – TUL, respectively. Presently, he is a Full Professor and Head of the Construction Section at IST – TUL and his areas of interest are construction technology, building pathology and rehabilitation and fibre reinforced materials.

L.G. Reis graduated in Mechanical Engineering and received his MSc and PhD degrees at Instituto Superior Te´cnico (IST) – Technical University of Lisbon (TUL), Portugal, where he is an Associate Professor. His areas of interest include mechanical design, mechanical characterization of materials, structural integrity of mechanical components, fatigue and fracture.

To purchase reprints of this article please e-mail: [email protected] Or visit our web site for further details: www.emeraldinsight.com/reprints

GFRP sandwich panels

147