Shear loaded friction-welded crosswise arranged timber boards

https://doi.org/10.1016/j.ijadhadh.2016.10.016Get rights and content

Abstract

Friction welding of wood is a bonding technology that can be used for joining timber elements, and in which the adhesive is formed from thermally modified cell wall material within the substrate during the welding process. In this paper, this principle is used for manufacturing prototypes of friction welded crosswise arranged timber boards, structural element bearing similitude to cross-laminated timber (CLT) tested under in-plane shear loads. In addition to the experimental investigations, three different approaches of strength prediction were performed. Beside a simply analytical method, two different probabilistic principles based on the non-local Weibull theory of brittle failure were conducted.

Introduction

Friction welding of wood is a bonding technology in which the adhesive is formed from thermally modified cell wall material of the substrate during the welding process. The idea to achieve wood bonds by means of friction welding technology originates from a patent issued to Sutthoff [1], subsequently pursued by iBOIS [2], [3], [4], [5], [6], and by other research institutes [7], [8].

In order to achieve bond strength, the interface is quickly heated up by a combination of mechanically induced friction (thus the name), and pressure. As a result, wood cell material is thermally decomposed into a viscous mixture of ligno-cellulosic monomers [2]. Yet only investigated on a research level [3], [4], [5], [6], [7], [8], [9], friction welding of wood is seen as a promising complement to currently used chemically based adhesives. When interrupting the friction movement, usually after a few seconds, the viscous material released at the interface hardens under a defined cooling pressure. The final result is a stiff joint that achieves significant adhesive and cohesive strength, which results in joint strengths sufficient for structural timber applications [6], as for example shear walls.

In theory, almost any wood species has the potential to be friction-welded, with specific welding parameters, namely frequency f, amplitude a, welding pressure Pw and welding time tw, to be experimentally determined. Two frictional movements can be distinguished: Linear Vibration Welding (LVW), and Circular Vibration Welding (CVW). First investigations were carried out on small specimens, starting at a few square centimetres in size, but recently developed welding devices made it possible to weld larger surfaces up to 500 cm2.

With friction welding technology increasingly moving towards full scale samples, research for potential structural applications emerged, making it paramount to allow for estimations of corresponding joint strengths. Research on predictive methods for the strength of friction welded single lap joints [9], and double lap joints [4], showed that load bearing capacity in function of overlap length is bound to a threshold value, above which strength cannot be increased, an observation similar to what has been described for adhesively bonded joints [10], [11] to which welded joints are conceptually very close.

Previous investigations at IBOIS revealed that structures conceptually similar to Cross Laminated Timber (CLT) can be manufactured by means of friction welding [6]. Cross Laminated Timber [12] constitutes a plate-like engineered timber product which is optimized for bearing loads in and out of plane. Cross Laminated Timber is composed of an uneven number of layers each consisting of side-by-side placed boards which are crosswise arranged to each other normally under an angle of 90° and quasi-rigidly connected by adhesive bonding, usually Melamine-Urea-Formaldehyde (MUF) and one-component polyurethane adhesives (1K-PUR). In timber construction CLT panels are used for both slab and wall elements which fulfil a dual function for vertical load transfer as well as for horizontal bracing. For the latter load transfer occurs by transmission of shear stresses induced by in-plane torsion moments within the interface between the intersections of the orthogonal timber boards. Blaß and Görlacher [13] developed an analytical model for the design of horizontally loaded glued CLT panels, illustrated in Fig. 1, further used and discussed in Section 3.1 of this manuscript.

All of previous experimental work on welded timber connections [4], [6], [8], [9] has repeatedly shown that once the maximum load of the very stiff friction welded bond is attained, failure occurs in a sudden and brittle manner without noticeable plastic deformation prior to rupture. Such brittle failure mode is usually associated with a strong dependency of nominal material strength with the size of the sample upon which it has been determined. A statistical theory that considers size effects for brittle failure has been introduced by Weibull and furthermore developed by, among others, Freudenthal [14], and more recently by Bažant [15], [16]; associated models are usually labelled probabilistic. Probabilistic strength prediction approaches have successfully been applied on linearly loaded Fibre Reinforced Polymers (FRP) [17], glued timber lap joints [18] as well as friction welded interfaces [4], [9]; subsequently this approach has been extended towards friction welded moment resisting connections [6].

The probabilistic strength prediction approach requires an accurate estimation of the stress state at the interface. For the case of moment resisting connections and torsion, analytical approaches of St.-Venant and his followers are mostly applicable for limited ranges of geometrical shape, idealized loadings and isotropic material properties. Although analytical closed-form solutions are available, Finite Element Analysis (FEA) has to be preferred in order to take into account the complexity caused by factors such as anisotropic material properties and a three-dimensional stress state [19], [20]—and has already proven to be an accurate tool for simulation of wood welded joints [21]. Due to the incapacity of stress redistribution within rigid interfaces, failure is usually caused by stress concentrations of interacting transverse tensile and in-plane shear stresses, particularly towards the end of the overlaps [4], [5], [6], [18]. As is most strength prediction methodologies, stresses obtained from FEA have to be verified against a failure criterion that takes into account the multiaxial nature of the acting stresses.

Mechanical problems involving stress singularities can be solved using fracture mechanics, which require a pre-existing crack. However, all methods derived from linear elastic fracture mechanics (LEFM) usually require extensive and often tricky mechanical characterisation of materials [22], [23], [24], resulting in additional material, and system, parameters (as energy release rates). Accordingly, the implementation of LEFM-based methods into engineer adapted design routines is, by far, not straightforward.

The investigations presented herein aim to transfer knowledge gathered previously on single moment resisting connections [6] to more complex systems in form of crosswise arranged friction welded timber boards. First, the load-bearing behaviour with regard to in-plane shear, occurring when used for bracing, is experimentally investigated. Secondly, three numerical methods of strength prediction have been applied and evaluated with regard to their reliability. A comparison between experimental and numerical load capacity concludes the paper.

Section snippets

Material

Almost flawless, spruce boards (Picea abies) have been used for the production of the panels. The material has been stored under dry climate conditions of 40 °C and 27% relative air humidity (r.H.), resulting in a low moisture content of 4%. According to previous investigations by Hahn et al. [5] a reduction of the moisture content during manufacture leads to a reduction of upscaling effects on the quality of the bond, and a significant decrease of the scattering of joint capacities. Although

Analytical approach

The simplest possibility to model the crosswise arranged friction welded timber board is to consider the shear load F, following Schickhofer et al. [26] which used a similar approach to model CLT, being equally distributed through torsional moments acting at each welded surface within the panel: this can be modelled as a rigid grid, illustrated in Fig. 1, Fig. 5. Accordingly, at failure under the maximum load, Fult, shear stresses, τlim, are determined following Eq. (3)τlim=MT(n1)·Ip·b2=2·Fult·

Experimental results

Mechanical testing of the shear loaded friction-welded crosswise arranged timber boards showed, after some initial slip, relatively linear load-displacement behaviour. Stiffness of the panels under load is comparable within the series, which indicates a good reproducibility of the manufacturing process. Failure always occurred in a brittle manner, without any visible signs of ductility, and almost always followed the rupture of one of the nine individual welded contact zones. The scattering of

Conclusions

Friction-welded crosswise arranged timber boards have been produced and tested under in-plane shear in order to evaluate their potential, as for example for horizontal bracing elements in timber construction. The experimental investigations on six samples showed linear-elastic load-displacement prior to sudden and brittle failure of the welded interface at the maximal load level. Three different approaches of numerical strength prediction have been compared to the experimental values.

  • First,

Acknowledgements

The present research work was funded by the Swiss National Foundation and is part of the project "SNF-Sinergia Project no. CRSI22—127467/1".

References (37)

  • D. Taylor et al.

    Some new methods for predicting fatigue in welded joints

    Int J Fatigue

    (2002)
  • Sutthoff B. Verfahren zum Reibschweißartigen Fügen und Verbinden von Holz;...
  • B. Stamm et al.

    Joining wood by friction welding

    Holz als Roh- und Werkst

    (2005)
  • B. Stamm et al.

    Joining of wood layers by friction welding

    J Adhes Sci Technol

    (2005)
  • B. Hahn et al.

    Linear friction welding of spruce boards: experimental investigations on scale effects due to humidity evaporation

    Wood Sci Technol

    (2014)
  • B. Hahn et al.

    Moment resisting connections composed of friction-welded spruce boards: experimental investigations and numerical strength prediction

    Eur J Wood Wood Prod

    (2014)
  • B. Gfeller et al.

    Wood bonding by vibrational welding

    J Adhes Sci Technol

    (2003)
  • B. Gfeller et al.

    Wood bonding by mechanically-induced in situ welding of polymeric structural wood constituents

    J Appl Polym Sci

    (2004)
  • Cited by (7)

    • Friction rivet joining towards high-performance wood-metal hybrid structures

      2020, Composite Structures
      Citation Excerpt :

      The applications for the permanent joining of woods and their hybrid structures have attracted attention, which combined mechanical interlocking with chemical bonding [11–13]. Vallée et al. [14] applied friction welding to joining spruce (Picea abies) boards, and stated that the self-generated adhesive was formed from thermally-modified wood cell to obtain chemical bonding at the welding interface. Omrani et al. [15] studied the utilization of linear friction welding on the joining of beech (Fagus longipetiolata).

    • Numerical investigations and capacity prediction of G-FRP rods glued into timber

      2018, Composite Structures
      Citation Excerpt :

      The FEA indicated that for all considered series, magnitude of transverse tensile stresses range between slightly below 4 MPa to around 6 MPa, which would be corresponding to between slightly twice and thrice the nominal transverse tensile strength. However, these maxima were only attained at relatively sharp stress peaks, at the end of the embedment length, confirming previous findings [8,17,57,62–65] that not only the magnitude of the stresses matters, but also the size of the volumes on which they act. Shear stresses, although having their maxima at the end of the embedment, developed less marked stress peaks, that tended to flatten with increasing rod diameter.

    View all citing articles on Scopus
    View full text