Bond line performance at elevated temperature in engineered wood products

Abstract

This thesis investigates bond line performance at elevated temperatures of engineered wood products, particularly cross-laminated timber (CLT). The main interest is heat induced delamination, a phenomenon where parts of timber lamella detach at the bond-line interphase at temperatures less than those required for significant timber pyrolysis (< 300 °C). When not accounted for by structural designers, loss of composite action and subsequent heat induced delamination induce reductions in load bearing capacity, compromising both fire compartmentation and structural integrity during and after fire.

In this thesis, a bottom-up, multi-scale methodology was adopted to study the coupled thermal, hygrothermal, and mechanical factors influencing bond line performance. Two timber species, two 1-component polyurethane (1-c-PUR) adhesives (one standard (HB S) and one ‘heat-resistant’ (HB X)) and one melamine–urea–formaldehyde (MUF) adhesive were used across scales and compared to elucidate material and species effects on bond line performance.

At the microscale, thermogravimetric analysis (TGA), differential scanning calorimetry (DSC), and dynamic thermo-mechanical analysis (DTMA) were used to characterise thermal degradation and thermo-mechanical degradation of adhesives, timber, and their joints (shear laps) at different temperatures.

Small scale neutron imaging experiments were performed to elucidate the effects of bond line presence and timber species choice on heat induced moisture movement. This work shows increased neutron attenuation in bond line surrounding, where the neutron beam intensity decreases as it passes through hydrogen-rich zone, suggesting presence of moisture lag and/or accumulation This might cause plasticization in adhesive, lower glass transition temperature, or contribute to the added pressure gradients and stresses on bond line. It might also be beneficial by slowing down the heat transfer due to the increased moisture content in specific region. Timber species also influenced the response, when compared to Norway Spruce, Radiata Pine showed more rapid deterioration of elastic modulus and faster tangential moisture transport, both of which may influence bond line response.

Intermediate-scale experiments on shear-lap joints and cantilever beams were used to quantify the onset of heat induced delamination under controlled heat fluxes and structural loads. In these experiments, both 1-c-PURs, HB S and HB X, were susceptible to heat induced delamination (HID) when the heating conditions were increased up to and including flaming combustion, and when structural loading applied was equal above 12% of shear strength utilisation. No unique critical bond line temperature at which the failure occurred was observed, with the range of temperatures depending on heating regime and structural loading. Samples bonded with HB S failed either at their glass transition or adhesive softening temperature. For HB X adhesive, initiation of HID occurred at glass transition when structural loads were below 6% of shear strength utilisation, and at moisture plateau, from 100-120°C, when structural loading was increased to/above 12%.

Despite HID in both HB S and HB X bonded products, cantilever intermediate scale results show that HB X bonded samples have higher stiffness, slower crack propagation, smaller crack width, and are less prone to char fall-off. This could be due to their different chemical composition and subsequent thermal degradation, where HB S shows softening and HB X charring behaviour.

The findings in this thesis have three key practical implications. First, they offer conservative critical bond‐line temperature thresholds and temperature dependent deterioration of elastic modulus, which can be integrated into heat‐transfer and structural models to improve predictions of mass‐timber behaviour in fire scenarios. Second, they suggest that microscale methods can be useful for understanding material responses at larger scales and for enabling pre-screening of adhesive systems before conducting large-scale testing, which often involves significant financial and time costs. Finally, they highlight the relevance of incorporating representative structural loading in (standardised) experimental assessments aiming to ensure adequate adhesive performance in fire for engineered wood products, thus enhancing the reliability of fire design recommendations for mass‐timber buildings.