https://doi.org/10.14311/APP.2022.33.0175 Acta Polytechnica CTU Proceedings 33:175–180, 2022 © 2022 The Author(s). Licensed under a CC-BY 4.0 licence Published by the Czech Technical University in Prague BENDING BEHAVIOR OF ADHESIVELY-BONDED ENGINEERED WOOD-CONCRETE COMPOSITE DECKS Qiuni Fua, Libo Yanb, ∗, Bohumil Kasalb a Technische Universität Braunschweig, Institute of Building Materials, Concrete Construction and Fire Safety, Department of Organic and Wood-Based Construction Materials, Hopfengarten 20, 38102 Braunschweig, Germany b Fraunhofer Insitute for Wood Research Wilhelm-Klauditz-Institut WKI, Bienroder Weg 54E, Braunschweig 38108, Germany ∗ corresponding author: l.yan@tu-braunschweig.de, libo.yan@wki.fraunhofer.de Abstract. Four-point bending tests were conducted on five medium-sized (i.e., 2300 mm in length and 215 mm in width) engineered timber (laminated veneer lumber (LVL) and cross-laminated timber (CLT)) - concrete (wood chip concrete and plain concrete) composite decks. The concrete was glued to the wood substrate with epoxy and polyurethane adhesives. The observed failure modes of the composite decks were concrete crushing or wood failure in tension or shear. No failure of the adhesive interface was observed and the decks behaved linearly until failure. In the subsequent analysis, the authors quantified the shear flexibility of transverse layers (stressed perpendicular to the fiber direction) in CLT and LVL boards and its effect on the bending stiffness on the composite decks using γ-method described in EN 1995-1-1 (EC5). The analytical predictions of the effective bending stiffness were verified via experiments, showing consistently good agreement. Keywords: γ-method, adhesive bond, bending behaviour, timber-concrete composite, wood chip concrete. 1. Introduction Timber-concrete composite (TCC) structural ele- ments have advantages of higher stiffness, fire resis- tance and better acoustic performance compared with massive timber. Connection between concrete and wood is usually achieved by conventional mechanical dowel-type fasteners (shear studs) or notches. Me- chanical fasteners are not rigid and there is always a slip between the wood and concrete. Thus, full composite action of TCC connected by mechanical fasteners is hardly achievable. Adhesive bonding is often considered as rigid. The application of glued wood-concrete connections is not fully explored and the knowledge in this area is limited [1–7], let alone the relevant specifications in standards. Studies [1–7] demonstrated that glued TCC sys- tems had higher stiffness (e.g. slip modulus of con- nections and bending stiffness of TCC beams) com- pared to mechanically connected ones, and sufficient adhesion strength for structural applications under short-term loading and ambient-temperature. The wood products involved in these studies [1–7] were solid timber, glulam or vertically used laminated ve- neer lumber (LVL) Assuming the rigid characteristics of the used adhesives, the behavior of these kinds of TCC beams was characterized by assuming continu- ous strain at the wood-concrete interface and using the known transformed cross-section procedure [5, 8– 10].The low shear modulus of transverse layers in the CLT and LVL boards subjected to bending can cause significant shear deformation (or slip) and hence in- crease the deflection. In the present work, the au- thors studied the shear behavior of glued CLT- and LVL-concrete connections [11]. They further experi- mentally investigated the bending behavior of CLT- and LVL-concrete composite decks. Specifically, CLT and LVL boards were bonded wood chip concrete (WCC) and with plain concrete (PC) using epoxy and polyurethane (PUR) and adhesives. The addition of WCC [12] resulted in reduced concrete density and increased thermal insulation properties of plain con- crete. The polyurethane (PUR) used in the present study was stiff, its modulus of elasticity was as high as that of the used epoxy (around 3.5 GPa), showing nearly no deformability in the adhesive layer. The PUR in this study was different from the PUR in Ref. [13] (modulus of elasticity: 0.008 GPa) that resulted in visible slip at bond line (over 1 mm). To describe the slip in transverse layers and its ef- fect on the bending stiffness of composite decks, the authors used the shear analogy method [14] for quan- tifying shear deformation of the transverse layers in combination with γ-method of EC 5 [15]. The predic- tions of the effective bending stiffness were compared with the experimental results, showing a good match. 175 https://doi.org/10.14311/APP.2022.33.0175 https://creativecommons.org/licenses/by/4.0/ https://www.cvut.cz/en Q. Fu, L. Yan, B. Kasal Acta Polytechnica CTU Proceedings CLT Flatwise LVL Thickness (mm) 80 (20 + 40 + 20) 69 (Thickness of each veneer: 3 mm) Shear modulus (MPa) 50 (Rolling shear, Gr ) 22 (Perpendicular to grain, G90) Elastic modulus in grain direction (E0) (GPa) 12 11.4 Table 1. Mechanical properties of the engineered wood products [16]. 2. Experimental studies 2.1. Overview of test plan Engineered timber, CLT and LVL, and concrete, WCC and PC used in this study were identical to those in the previous study [11]. Epoxy and PUR were selected due to their bonding perfor- mance [11]. The authors constructed five medium- sized specimens, WCC-Epoxy-LVL, WCC-PUR-LVL, WCC-Epoxy-CLT, WCC-PUR-CLT and PC-Epoxy- LVL, and tested them under bending. Figure 1 (e) shows the configurations of the specimens. 2.2. Materials The mechanical data and configurations of CLT and LVL are shown in Table1. Moisture content of the used wood was about 10%. The beech wood chips used in WCC were in a relatively flat shape, having around 20 mm side length and 2.4 mm thickness. The density of the wood chip was 0.68g/cm3. The mass ratio of the constituents of WCC was CEM II/B-S 42.5 R cement: water: 0-2 mm sand: 2-8 mm gravel: 8-16 mm gravel: wood chip = 1: 0.60: 2.00: 1.27: 2.32: 0.11. The mass ratio of the constituents of PC was CEM II/B-S 42.5 R cement: water: 0-2mm sand: 2-8 mm gravel: 8-16 mm gravel = 1: 0.60: 2.00: 1.27: 2.72. The compressive strength of the WCC and PC cylinders were 25.8 MPa (SD = 3.79 MPa) and 27.9 MPa (SD = 1.73 MPa) based on five samples, re- spectively. The respective elastic modulus of PC and WCC were 29.9 GPa and 12.3 GPa. According to the product of PRIMETM 20LV from [17] datasheet, the elastic modulus of epoxy is 3.5 GPa. The pull-off tests in authors previous study [11] showed that the load-displacement curves of the used epoxy and PUR almost overlapped at the initial loading stage, so the elastic modulus of the used PUR is similar to that of the epoxy. 2.3. Specimen preparation When the concrete was cured for 28 days, the con- crete surfaces were cleaned and glued to shaped wood decks with wood decks on the top for the first 3 days. The specimens were turned upside down since the fourth day and tested after the seventh day. The de- tailed processes can be found in Ref. [16]. 2.4. Experimental setup A load (F ) was quasi-statically applied on the TCC decks with a rate of 0.35 mm/s through a two-point spreader, as shown in Figure 2. It also shows the loca- tions of the support and loading points. A deflection at the mid span was measured by two LVDTs (linear variable differential transducers). 3. Results and discussion 3.1. Load-carrying capacities The load versus mid-span deflection curves of all the TCC decks showed linear behaviour until the ulti- mate failure. Figure 3 shows the response of WCC- PUR-CLT as an example, and the ultimate loads, the ultimate mid-span deflections and bending stiff- ness (EI)test of all the specimens are summarized in Table 3. The calculation of (EI)test was based on the ascending slope of a load versus mid-span deflection curve as follows [16]: (EI)ef = F w a ! 3l2 − 4a2 " 48 (1) where w is the average reading value of the two LVTDs; a is the shear span; and l indicates the span of the deck. From Table 3, it follows that the bending stiffness of a beam was not affected by adhesive type. The rea- son was that a rigid connection was achieved by using either epoxy or PUR, which was also demonstrated by the absence of slippage between concrete and wood decks (see Section 3.2 and Table 3). The specimens comprising CLT boards were stiffer than the counter- parts made of LVL (Table 3). The inherent reasons are discussed in Section 4. On the contrary, the ulti- mate load-bearing capacities of the LVL-specimens were higher than that of CLT-specimens., due to different failure modes (Table 3). As the PC had higher elastic modulus and strength compared with WCC, PC-Epoxy-LVL had the largest flexural stiff- ness (4.12 × 1011N.mm2) and ultimate load-carrying capacity (63.2 kN) [16]. The average bending stiffness and load-bearing ca- pacity of WCC-Epoxy-LVL and WCC-PUR-LVL was around 78.9% and 70.0% of those of PC-Epoxy-LVL, respectively. Meanwhile, the thermal conductivity of WCC could be only 68.9% of that of PC. [18]. Be- sides, wide application of WCC can reuse wood waste, and reduce the demand for natural aggregates. The load-bearing capacity of the weakest specimen, WCC- Epoxy-CLT, was 39.1 kN, equivalent to a uniformly- distributed load of 85.4 kN/m2. Considering service- ability limit design, EC 5 [15] allows the maximum 176 vol. 33/2022 Bending of WoodConcrete Composite deflection of a slab to be l/300 (i.e. 7.1 mm). This deflection corresponds to an equivalent slab load of 34.7 kN/m2 in the case of WCC-Epoxy-CLT. 3.2. Failure modes The failure modes of all the specimens are sum- marized in Table 3. Neither of the specimens was failed by adhesive bonding interface, indicating that the used adhesives can provide satisfactory bonding performance. The different mechanical properties of CLT and LVL caused different failure modes of the corresponding TCC decks. -see [16] for details. One can also have a deeper appreciation of the structural behaviour of LVL and CLT in TCC decks in Section 4. 4. Prediction of bending stiffness In building slab design, deflection is one of the important serviceability considerations. Structural engineers usually calculate deflections with Euler- Bernoulli beam theory that does not take shear de- formation into account. However, when engineered wood (e.g. CLT) is under bending, transversely stressed layers have significant shear deformation, be- cause of low rolling shear modulus (Gr ) and low shear modulus in the perpendicular direction (G90) (Ta- ble 1). To consider the shear flexibility of transverse layers, the authors regarded it as sources of partial composite action and calculated the effective bend- ing stiffness (EI)ef for all the specimens based on γ- method in EC 5 [15]. The γ-method [15] allows a composite cross-section to be comprised of three parts with two connecting interfaces featured with the respective degree of com- posite action (γi) relative to the middle layer (referred as "Part 2" in Figure 4 (a)). A term γi ranges between 0 (no interaction) and 1 (rigid connection). Then (EI)ef of a composite deck is calculated with Equa- tion (2) [15]. (EI)ef = 3# i=1 ! Ei Ii + γi Ei Ai a2i " (2) where Ei is the modulus of elasticity of Part i in the longitudinal direction; the term Ii means moment of inertia of Part i; the term Ai denotes the cross-section area of Part i; the term ai indicates the distance be- tween the neutral axis (i.e. N.A. in Figure 4 (a)) and the centroid of Part i. According to EC5 [15]: γi = $ 1 + π2 Ei Ai si % (Ki l2) & −1 (i = 1 and 3) (3) where Ki is the slope of a shear force - slip curve (i.e. slip modulus) of a connection, and si denotes the equidistance between shear fasteners connecting Part 2 and Part i. For a continuous adhesive connection, Ki can be defined based on a given length, e.g. 1 mm. Then, si is equal to 1 mm. As explained in Ref. [16], the CLT board (Part 2) and concrete (Part 1) were rigidly bonded based on the experimental obser- vations (i.e. K1 = ∞ γ1 = 1). Part 2 is the reference layer, so γ2 is equal to 1. In this paper, the slip mod- ulus K3 relates to the shear connection between the top (Part 2) and bottom (Part 3) longitudinal layers of CLT. Therefore [16], K3 = Gr · b · (1000 mm) / htr (4) where Gr denotes the rolling shear modulus of the CLT ( refer to Table 1), htr denotes thickness of the transversal layer, and b is the width of the decks. According to EC [15]: a2 = γ1 E1 A1(h1 + h2) − γ3 E3 A3(h2 + h3) 2 3' i=1 γi Ei Ai (5) The terms a1 and a3 can be obtained based on the geometric relationship in Figure 4 (a). To model the shear flexibility of dispersed transver- sal veneers in LVL, the longitudinal and transversal veneers were gathered into the respective two layers for simplification. The full concept is illustrated in Figure 4 (b). For details regarding the modelling and analysis, please refer to the authors’ previous work in Ref. [16]. In simple terms, Equations (2) through (5) are applicable for TCC decks using LVL, with h1 = 0 and G90 replacing Gr . All the detailed deriva- tions can be found in Ref. [16]. For comparison, γi and (EI)ef of TCC decks comprising the respective CLT and LVL boards are presented in Table 2. The degree of composite action in the TCC decks com- prising LVL (0.52) was lower compared with that of the TCC decks comprising CLT (0.70). The reason was that the transversely stressed veneers have a very low shear modulus (22 MPa). The (EI)ef of CLT- involved TCC decks was slightly higher than that of LVL-involved TCC decks, which agrees with the test results (Table 3). The higher (EI)ef of CLT-involved TCC decks stemmed from a larger γ3 and a larger a3 (see Fig- ure 4). The specimen made of PC had the largest stiffness due to the larger elastic modulus of the plain concrete. The larger elastic modulus of PC led to the largest stiffness of PC-Epoxy-LVL. As shown in Table 2, the (EI)ef calculated in this study conservatively predicted the corresponding test results with an error of around 10% [16]. Ignoring the shear deformation (i.e. γ3 = 1), EI was overestimated by around 34% on average. With the latter kind of calculation, another undesired situation is that the errors of predictions of EI varied significantly (CoV = 14.1%), depending on different types of engineered wood products having different degrees of shear flexi- bility. This problem was solved by this study through quantifying the shear flexibility of transversal wood layers [16]. The errors of predicted (EI)ef of CLT- and LVL-involved TCC decks had small variations 177 Q. Fu, L. Yan, B. Kasal Acta Polytechnica CTU Proceedings Figure 1. Dimensions of fabricated TCC decks. Concrete LVDTȬ1ȱ(2) 710ȱmm 710ȱmm 710ȱmm Timber F/2 F/2 F355ȱmm 355ȱmm Roller Figure 2. Setup of 4-point bending test. Specimen ID γ1 γ3 EI a (N.mm2) EI/(EI)test (EI)ef (N.mm2) (EI)ef /(EI)test WC-Epoxy-CLT 1 0.70 4.13×1011 1.14 3.21×1011 0.89 WC-PUR-CLT 1 0.70 4.24×1011 1.13 3.33×1011 0.89 WC-Epoxy-LVL \ 0.52 4.62×1011 1.42 2.90×1011 0.89 WC-PUR-LVL \ 0.52 4.90×1011 1.50 3.05×1011 0.93 PC-Epoxy-LVL \ 0.52 6.15×1011 1.49 3.82×1011 0.93 Average 1.34 0.91 Coefficient of variance 14.0 % 2.6 % a The term EI denotes the flexural stiffness of TCC decks, ignoring the shear flexibility of transvers layers. Table 2. Predictions of flexural stiffness. Figure 3. Load vs. displacement responses of WCC- PUR-CLT at mid-span. (CoV = 2.6%) (Table 2). Thus, this study provides an effective solution for the calculation of (EI)ef (or deflection) of TCC decks using engineered wood. 5. Conclusions From this study, some conclusions can be drawn, as follows: 1. The PUR- and epoxy-bonded wood and concrete could form a strong and rigid interfaces to transfer the shear in TCC decks subjected to quasi-static bending, without failure and visible slip. 2. The failures of TCC decks in this study were brit- tle, which were governed by wood fracture and con- crete crushing. 3. Wood chip concrete is feasible to be used with the engineered wood as composite structural slabs. En- gineered wood could be used together with WCC to realize a sustainable solution, with satisfactory structural performance. 4. The method in this study provides an effective so- lution for the calculation of (EI)ef (or deflection) of TCC decks under bending. Acknowledgements The research was financially supported by Fachagentur Nachwachsende Rohstoffe e. V. (FNR, Agency for Re- newable Resources) founded by Bundesministerium für Ernährung und Landwirtschaft (BMEL), under the Grant Award No.: 22011617, and by Bundesministerium für Bil- dung und Forschung (BMBF) (Grant No.: 031B0914A). 178 vol. 33/2022 Bending of WoodConcrete Composite (a) Concrete-CLT composite decks Concreteȱ(EC) Adhesive* 40 20 1000 ȱLongitudinalȱlayer ȱLongitudinalȱlayer Transversȱlayer Gr 20 E0 E0 3ȱ (5 ) 50 h t r h 2 h 3 h 1 Model N.A. a 1 a3 h2 /2 h1 /2 h3 /2 a 2 *ȱTheȱmomentȱofȱinertiaȱofȱtheȱadhesiveȱlayerȱisȱignored,ȱbutȱitsȱthicknessȱisȱconsideredȱwhenȱdeterminingȱgeometricȱpositionsȱforȱotherȱmaterialȱlayers. Construction Concreteȱ(E1) Adhesive 40 20 1000 ȱLongitudinalȱlayer ȱLongitudinalȱlayer K3=Gr*b*(1000mm)/htr Partȱ1 Partȱ2 Partȱ3 20 K1=infinite E3 E2 3ȱ (5 ) 50 (b) Concrete-LVL composite decks Partȱ3 K3=G90*b*(1000mm)/htr h 3 =5 4 h 2 3ȱ (5 ) ȱRigidȱadhesiveȱlayerAdhesive:ȱrigidȱbond htr=15ȱmm Model Construction Concreteȱ(EC) ȱLVLȱ(E0,ȱG90) ȱLongitudinalȱlayerTransversȱlayer 3 3 ȱLegends 69 1000 3ȱ (5 ) 50 Concreteȱ(E2) ȱLVLȱ(E3) 34 ,5 1000 50 34 ,5 27 27 Centroidȱline ofȱwood Partȱ2 Figure 4. Analytical model of composite decks (unit: mm) - quoted from Ref. [16]. References [1] M. Brunner, M. Romer, M. Schnüriger. Timber- concrete-composite with an adhesive connector (wet on wet process). Materials and Structures 40(1):119-26, 2006. https://doi.org/10.1617/s11527-006-9154-4. [2] J. Kanócz, V. Bajzecerová, M. Mojdis. Experimental and Numerical Analysis of Timber-Lightweight Concrete Composite with Adhesive Connection. Advanced Materials Research 969:155-60, 2014. https://doi.org/10.4028/www.scientific.net/AMR .969.155. [3] P. L. Clouston, C. P. Quaglia. Experimental Evaluation of Epoxy Based Wood Plank-concrete Composite Floor Systems for Mill Building Renovations. The International Journal of the Constructed Environment 3(3):63-74, 2013. https: //doi.org/10.18848/2154-8587/CGP/v03i03/37394. [4] J. H. J. d. O. Negrão, C. A. Leitão de Oliveira, F. M. Maia de Oliveira, et al. Glued Composite Timber-Concrete Beams. I: Interlayer Connection Specimen Tests. Journal of Structural Engineering 136(10):1236-45, 2010. https: //doi.org/10.1061/(asce)st.1943-541x.0000228. [5] J. H. J. d. O. Negrão, F. M. Maia de Oliveira, C. A. Leitão de Oliveira, et al. Glued Composite Timber-Concrete Beams.II: Analysis and Tests of Beam Specimens. Journal of Structural Engineering 136(10):1246-54, 2010. https: //doi.org/10.1061/(asce)st.1943-541x.0000251. [6] M. Smilovic, D. Cubela, J. Radnic, et al. Experimental testing of wood-concrete and steel-concrete composite elements in comparison with numerical testing. Materialwissenschaft und Werkstofftechnik 44(6):562- 70, 2013. https://doi.org/10.1002/mawe.201300026. [7] M. Schäfers, W. Seim. Geklebte Verbundbauteile aus Holz und hoch- bzw. ultrahochfesten Betonen. Bautechnik 88(3):165-76, 2011. https://doi.org/10.1002/bate.201110017. [8] Dias A M P G, Mechanical Behaviour of Timber-Concrete Joints. 2005, Technische Universiteit Delft: Netherland. [9] L. Boccadoro, R. Steiger, S. Zweidler, et al. Analysis of shear transfer and gap opening in timber-concrete composite members with notched connections. Materials and Structures 50(5), 2017. https://doi.org/10.1617/s11527-017-1098-3. [10] V. Bajzecerová, J. Kanócz. Timber-Concrete Composite Elements with Various Composite Connections. Part 2: Grooved Connection, Wood research, 59(3):27-638, 2014. 179 https://doi.org/10.1617/s11527-006-9154-4 https://doi.org/10.4028/www.scientific.net/AMR.969.155 https://doi.org/10.18848/2154-8587/CGP/v03i03/37394 https://doi.org/10.1061/(asce)st.1943-541x.0000228 https://doi.org/10.1061/(asce)st.1943-541x.0000251 https://doi.org/10.1002/mawe.201300026 https://doi.org/10.1002/bate.201110017 https://doi.org/10.1617/s11527-017-1098-3 Q. Fu, L. Yan, B. Kasal Acta Polytechnica CTU Proceedings [11] Q. Fu, L. Yan, T. Ning, et al. Interfacial bond behavior between wood chip concrete and engineered timber glued by various adhesives. Construction and Building Materials 238, 2020. https: //doi.org/10.1016/j.conbuildmat.2019.117743. [12] T. Ning. Interfacial Bond and Structural Performance of Adhesively-Bonded Light-Weight Timber-Wooden Chip Concrete Composite Structures, in Institut für Baustoffe, Massivbau und Brandschutz (iBMB), Technische Universität Braunschweig: Braunschweig, Germany, 2018. [13] J. R. Cruz, S. Serȩga, J. Sena-Cruz, et al. Flexural behaviour of NSM CFRP laminate strip systems in concrete using stiff and flexible adhesives. Composites Part B: Engineering 195, 2020. https: //doi.org/10.1016/j.compositesb.2020.108042. [14] Q. Fu, L. Yan, T. Ning, et al. Behavior of adhesively bonded engineered wood - Wood chip concrete composite decks: Experimental and analytical studies. Construction and Building Materials 247, 2020. https: //doi.org/10.1016/j.conbuildmat.2020.118578. [15] P. Mestek, H. Kreuzinger, S. Winter. Design of Cross Laminated Timber (Clt), in 10th World Conference on Timber Engineering, Miyazaki, Japan, 2008. [16] Gurit. PrimeTM 20lv-Epoxy Infusion System, 2018. https://www.gurit.com/sitecore/content/Old-Pro duct-Pages/Other-Products/Laminating-Infusio n-Systems/PRIME-20LV. [17] EN 1995-1-1, Eurocode 5 Design of Timber Structures Part 1-1: General - Common Rules and Rules for Buildings. European Committee for Standardisation, Brussels, 2004. [18] M. Bederina, L. Marmoret, K. Mezreb, et al. Effect of the addition of wood shavings on thermal conductivity of sand concretes: Experimental study and modelling. Construction and Building Materials 21(3):662-8, 2007. https: //doi.org/10.1016/j.conbuildmat.2005.12.008. 180 https://doi.org/10.1016/j.conbuildmat.2019.117743 https://doi.org/10.1016/j.compositesb.2020.108042 https://doi.org/10.1016/j.conbuildmat.2020.118578 https://www.gurit.com/sitecore/content/Old-Product-Pages/Other-Products/Laminating-Infusion-Systems/PRIME-20LV https://doi.org/10.1016/j.conbuildmat.2005.12.008