Till Vallée

The ageing response of pretensioned hybrid joints was investigated in comparison with bonded and bolted reference joints for two structural adhesives, namely a two-component polyurethane (PUR) and a two-component epoxy (EPX), using two substrate surface conditions, namely organically coated steel and hot-dip galvanised steel. Climatic cycling and salt-spray exposure were applied for up to 6 weeks. The experimental basis comprised 400 joint tests, supplemented by 70 bulk tensile tests and 28 DMA measurements. Across both adhesive systems and both surface conditions, the hybrid configuration exhibited the most stable ultimate response. Although ageing affected damage-initiation and energy-dissipation measures, the ultimate lap-shear strength generally remained close to the unaged reference level. By contrast, bonded joints showed the strongest surface- and adhesive-dependent degradation, whereas bolted joints remained mechanically stable. The clearest differencewas observed for epoxy joints on galvanised steel under climatic exposure, where the bondedconfiguration decreased from 14.86 to 0.40 MPa in ultimate lap-shear strength and from 38.9 to 1.1 J in fracture energy, whereas the corresponding hybrid joints remained essentially unchanged in peak strength, from 22.42 to 22.55 MPa. For both adhesives, the fracture-surface observations were consistent with the interpretation that hybrid joints retained mixedfailure and avoided the pronounced adhesive- or near-interfacial failure observed in the most severely degraded bonded subsets. Overall, the results suggested that pretensioned hybrid joints provided the most robust ageing tolerance among the investigated configurations. This behaviour was interpreted as evidence that the pretensioned fasteners provided a parallel, mechanically stabilised load path, such that degradation in the adhesive layer or at the adhesive–substrate interphase did not translate directly into a comparable loss of ultimate joint capacity. However, this interpretation was based on joint-level mechanical response and qualitative fracture observations rather than on direct measurements of local load sharing, crack arrest, bond-line thickness, or preload evolution.

Preloaded bolted–bonded hybrid joints have commonly been characterised using lap-joint specimens loaded predominantly in axial shear. However, many structural connections show complex shear, in which eccentricity and rotation induce additional torsional shear stresses. In this study, a preloaded hybrid joint in complex shear on hot-dip galvanised steel was investigated to extend experimental evidence beyond lap-joint typologies. A slip-resistant bolted reference without adhesive and hybrid configurations using two different epoxy adhesives were examined. Bolt edge distance (𝑒 = 15, 25, 35 mm) and outer splice-plate thickness (𝑡𝑜 = 10, 15, 20 mm at 𝑒 = 15 mm) were varied, and the response was characterised using both machine displacement and DIC-based relative displacement. Across all configurations, a consistent staged response was observed, comprising an initially steep rise, a stiffness transition, attainment of a peak load 𝐹max, a first pronounced load drop, and a sustained post-drop branch. Hybridisation increased 𝐹max substantially relative to the friction-only reference, ranging from about +50% at 𝑒 = 15 mm to about +110% at 𝑒 = 35 mm for Sikadur 370 (with Scotch-Weld 7240 showing smaller but consistent gains). Edge distance reduced peak resistance for all concepts, but the reduction was substantially stronger for the friction-only reference than for the hybrids, indicating a mitigation of rotation-driven localisation by hybridisation. Overall, adhesive selection was shown to influence not only load capacity but also the kinematic pathway into the post-drop regime, and the dataset was intended to serve as a benchmark for subsequent modelling.

Pulsed laser cleaning had offered an attractive option for local repair of steel bridges and offshore structures in marine and de-icing environments, because corrosion and aged coatings could be removed with little secondary waste and high automation potential. However, existing studies had focused mainly on feasibility and microstructural effects on relatively simple corrosion states, and quantitative data on laser-based local preparation of heavily corroded, coated structural steel—particularly with simultaneous control of rust removal, roughness and soluble salts relative to specification limits and conventional methods—remained limited. In this study, pulsed Nd:YAG laser surface preparation of artificially corroded, heavy-duty coated S355 steel was investigated and was benchmarked against grinding, bristle blasting and vacuum suction blasting. Rust removal behaviour, surface roughness and morphology were characterised, and residual soluble-salt contamination was quantified using the Bresle method. A parameter window was identified in which pulsed laser treatment achieved complete removal of thick corrosion layers at practical area rates, generated peak-to-valley roughness comparable to or higher than vacuum suction blasting and reduced residual salt levels to at or near the detection limit, whereas mechanical methods often produced lower roughness and higher salt loadings. The results established a practical process window for laser-based local preparation that satisfied combined cleanliness, roughness and salt criteria for severe exposure and provided a quantitative basis for subsequent performance studies on repair coatings and bonded or hybrid joints.

Adhesive injection into thin bond-lines was studied to establish a computationally efficient prediction tool for flow-front evolution, internal velocities and injection pressures. A reduced-order gap-flow approach based on the Partially Filled Gap Model (PFGM) was extended from squeeze-flow to injection by introducing localised inlets, explicit outlets and gravitational body-force contributions for inclined gaps, and was verified against reference problems for source-flow kinematics, conservation of injected volume and gravity-driven redistribution. Experimental validation was performed in a quasi-two-dimensional glass-plate flow cell using a Newtonian silicone-oil model fluid. Several inlet--outlet arrangements and inclinations were investigated, and flow fronts, in-plane velocity fields (Particle Image Velocimetry) and inlet-pressure histories were measured. The simulations reproduced the observed front shapes and filling sequence, captured the main features of the velocity profiles away from inlet and outlet regions, and recovered the characteristic multi-stage pressure evolution. Systematic offsets in absolute pressure level were attributed to additional hydraulic losses in the upstream supply system not represented in the cavity-scale model. The validated model was then applied to a sensitivity analysis and to a frame-like cavity representative of panel-to-frame bonding, demonstrating strong dependences of pressure and injection-induced forces on viscosity, gap height and in-plane dimensions, and illustrating how gravity and boundary layout influence asymmetric filling. Extensions to adhesive-specific rheology, curing, wetting and discrete gas bubbles remain topics for future work.