Sticking to the issues ...
TWI Bulletin, July - August 2009
bonding composites to steel in double lap shear
Philippa Moore is a senior project leader in Fracture Integrity Management within the Structural Integrity Technology Group at TWI. Philippa runs projects involving fitness-for-service assessment, fracture mechanics testing, and welding engineering, particularly for the oil and gas industry. She is TWI's Institution Representative within the MARSTRUCT Network of Excellence in Marine Structures.
Safa Hashim is a Senior Lecturer in Materials Engineering in the Department of Mechanical Engineering at the University of Glasgow. He currently teaches materials and design. His research interests are in adhesive bonding and moulding of thermoset composites, including multi-scale analysis. He is the Task Leader of composite to steel bonding within the MARSTRUCT Network of Excellence in Marine Structures.
Structural adhesives are finding an increasing number of uses within marine industries as they offer the advantage of achieving lightweight fabrications with corrosion resistance, particularly when bonding composite materials. As Philippa Moore and Safa Hashim explain, in addition to the fabrication of ship superstructures, adhesive bonding of composites is also a potential repair method for cracks and corroded areas on pipelines and offshore platforms. However, there is currently little universal guidance on the design strength of adhesive joints, let alone the added complexity of assessing adhesive joints in composites. One of the most common adhesive joint configurations is the double lap shear (DLS) joint with composite straps.
This article describes the fabrication, testing and numerical analysis undertaken on DLS joints for benchmarking purposes as a joint effort between various partners within the MARSTRUCT project. The aim was to provide guidance on the design and fabrication of DLS joints between 10mm thick steel and composite straps.
Fabricating the double lap shear joints
The fabrication of the DLS test specimens was done by MARSTRUCT partners from the University of Glasgow, and the Technical University of Denmark (DTU). The specimens had 10mm thick mild steel inner adherend bars and outer adherend straps made from laminates of 0°/90° UD CFRP (unidirectional carbon fibre reinforced polymer). Some tests were also carried out using steel or glass fibre straps, but they are not included in this article. The thickness of the CFRP straps was three or six mm, and strap overlap lengths of 25 to 200mm were used (as measured from the centre of the specimen). The DLS test specimen arrangement is shown in Figure 1.
Fig.1. The Double Lap Shear (DLS) joint arrangement prior to adhesive bonding
The whole fabrication process included surface roughening, degreasing, marking, adhesive application, positioning and clamping, and then curing. The bonding surfaces of the steel components were prepared by grit blasting or heavy abrasion by silicon carbide paper. Teflon sheeting was applied to one end of each steel bar to prevent the ends joining, so that all the loading was through shear along the straps rather than a tensile load between the steel bars. The bonding surfaces of the CFRP straps were prepared by light abrasion using 120 grit silicon carbide paper. All bonding surfaces were cleaned and degreased. Markings were put on the steel bars and straps to ensure correct fit-up of the joint when being bonded.
The adhesive was Araldite® 2015 (from Huntsman), a two-part toughened epoxy adhesive. The adhesive was mixed and applied by spatula in two stages; the first stage was to prime the surfaces with a thin layer of adhesive; and in the second stage more adhesive was applied at the centre of the joint, which then spread across the bondline as the joint was pressed closed. The DLS specimens were clamped evenly to give a uniform 0.5mm thickness of adhesive layer, and were then cured for one hour in an oven at 85°C. Afterwards, any excess adhesive was mechanically removed so that there were no effective adhesive fillets within the lap joint (this was done to reduce variability and make subsequent modelling of the specimen configurations easier).
Mechanical testing
The mechanical testing of the double lap shear joint specimens was performed by TWI, in addition to other MARSTRUCT partners (including the University of Glasgow, DTU, University of Galati in Romania, and the National Technical University of Athens). The TWI tests were performed under quasi-static tensile loading, at ambient temperature with a crosshead displacement of 0.5mm/min. Each specimen had three strain gauges attached (Figure 2); one on the steel bar located 5mm away from the composite (identified as strain gauge SG3), and one on the composite strap on either side of the joint, positioned with one edge of the 10mm long strain gauge at the centre line (these were identified as strain gauges SG1 and SG2).
Fig.2. The arrangement of strain gauges on the DLS test specimens
The overall objectives of the mechanical testing part of the work were:
- to assess the quality of the fabrication method
- to determine the joint strength and effect of overlap length for specimens with a range of joint configurations
- to understand aspects of failure and design of DLS joints under quasi-static loading.
The test results from all the institutions were collated and the failure loads plotted against the strap overlap length (Figure 3). The failure load was proportional to the overlap length up to 100mm overlap, after which a plateau was reached. (This trend was also seen in the tests with straps made from steel and glass fibre, albeit at different plateau strengths). Thinner CFRP straps in the longer joints exhibited slightly lower strength than equivalent thicker ones.
Fig.3. Failure loads of steel/CFRP joints with different overlap lengths and strap thickness, showing increasing strength with strap length until a plateau is reached
After testing, the failed specimens were examined and photographed in order to identify the failure mode(s) and initiation locations. FEA models of selected DLS specimens were also generated in order to understand the strength of these joints.
Numerical modelling
Finite element analysis (FEA) models were produced for steel/CFRP specimens with 6mm thick CFRP straps and overlaps of 25, 50, 100 and 200mm. The modelling was done by several universities involved in this MARSTRUCT work. 2-D non-linear models were constructed in ABAQUS (as well as other software) using eight-noded solid quadrilateral plane strain elements (as well as others). The 0.5mm thick adhesive bondline was divided into five layers through thickness with a standardised fine mesh towards the joint ends. This allowed stress and strain data to be taken along paths created at the upper interface of the adhesive with the outer straps, and at the lower interface of the adhesive with the inner steel adherend.
The plies within the CFRP were separated by 0.1mm of matrix resin, so the resin in the CFRP adjacent to the adhesive bondline was modelled into two layers to account for stress details within the resin. Only the first two plies in the 0 and 90° sequence of the CFRP were modelled into layers, while the rest were modelled as a bulk isotropic material. The steel and adhesive were modelled as elasto-plastic, while the composite was modelled as an elastic isotropic layer.
The predictions of the numerical models showed good agreement with the experimental strain gauge results at corresponding locations. The FEA strategy of modelling the first two plies of the CFRP and their resin layers seemed to yield satisfactory results while accounting for maximum stress, and stress values at critical spots, including the matrix resin. For the shorter strap lengths the adhesive seemed to reach full plasticity in shear before any peel occurred. The 50mm overlap seemed to produce a largely plastic behaviour in the adhesive joint, and both the 100 and 200mm lengths developed a similar size plastic zone of about 40mm. A possible failure criterion here is that following the plasticity in shear, a brittle fracture can develop leading to steady crack propagation.
The models for the longer overlaps showed that the maximum principal stresses are highest near the ends of the straps, and also in the resin at the interface with the adhesive at the centre of the joint. High principal stresses can initiate peel failures from these locations. The stress contour showed a very high level of principal stress within the brittle epoxy resin (greater than the resin strength), which could be the source of failure initiation and propagation from the middle of the specimen.
Failure modes in DLS adhesive joints
Assessment of the failed test specimens, the FEA models, and some high speed imaging of the mechanical tests at DTU, allowed the failure modes of these adhesive joints to be evaluated.
- For DLS specimens with short overlap lengths (less than 50mm), the specimens carried loads up to the level where the entire bondline yielded in shear, causing joint fracture (Figure 4a).
- At longer overlap lengths (greater than 100mm), the failure was by peel from the middle of the specimen (Figure 4b). This is because the inner adherend is stiffer than the two CFRP straps.
- Up to 50mm of the joint length for longer specimens was loaded into the inelastic range, carrying much of the applied loading before fracture.
- For specimens with long overlap lengths, assessment of the fracture surfaces suggests that the steel/CFRP joint is failing nearer the interface between the CFRP surface ply (0-direction) and resin and the adhesive bondline. This was perhaps due to resin or adhesive failure starting at the middle of the joint. It is also possible that tensile failure of the laminate had occurred.
Fig.4. The failure surfaces of DLS specimens fabricated from 3mm thick composite straps with overlap lengths of
a) 100mm
b) 25mm
Discussion and recommendations
The fabrication method used in this study can be adopted for practical application. The test results from the fabricated specimens suggest the bonding process is robust, however, some scatter was seen in the joints with longer overlaps (and different strap materials). The test results for the longest overlaps suggest a limited static strength advantage over 100mm overlap, but results from an intermediate overlap length of 150mm would be useful to confirm this.
The FEA results that are based on critical stress or strain points at prescribed distances from critical locations are useful tools to predict joint failure, especially for short overlaps. The modelling also showed that a considerable length (up to 50mm) of the bondline was loaded into the inelastic range before fracture, carrying much of the applied loading, thus showing that attempts at predicting failure of such joints would have to account for non-linear inelastic adhesive behaviour.
Acknowledgements
This work has been performed within the context of the Network of Excellence on Marine Structures (MARSTRUCT) partially funded by the European Union through the Growth Programme under contract TNE3-CT-2003-506141. The work presented in this article is derived from two multi-author publications by the MARSTRUCT task members. Further information can be obtained from these sources.
This work has been performed within the context of the Network of Excellence on Marine Structures (MARSTRUCT) partially funded by the European Union through the Growth Programme under contract TNE3-CT-2003-506141. The work presented in this article is derived from two multi-author publications by the MARSTRUCT task members. The Marstruct tasks and sub-tasks are listed on
www.mar.ist.utl.pt/marstruct/workplan.aspx
