Recent interest in sustainable design has resulted in timber products being considered for a variety of construction projects. This has especially been the case for engineered wood products (EWPs), such as glue-laminated timber (glulam) and cross-laminated timber (CLT). Research into the performance of these massive timber products has been ongoing, where the methodology employed has generally favoured experimental approaches on undamaged members, combined with simplified analytical methods. Relatively little attention has been given to more sophisticated numerical methodologies and to the effects of repeated loadings on the same specimen. This study intends to contribute to the literature by investigating the viability of full-scale finite element models to simulate the behaviour of timber elements at high strain rates and proposing a generalized structure for dynamic models that is capable of adequately recreating realistic failure modes.
Three glulam specimens and three CLT specimens were subjected to simulated blast loads under four-point bending with simply supported boundary conditions using the University of Ottawa Shock Tube Test Facility. The behaviour of the glulam specimens during the dynamic testing was consistently linear-elastic until flexural failure was reached. Conversely, the failure behaviour of CLT panels was more complex and included flexural failure, rolling shear failure, or a combined behaviour where both modes developed simultaneously.
Single-degree-of-freedom (SDOF) and finite element analysis (FEA) methodologies were used to predict the behaviour in terms of displacement-time histories and failure modes. The inputs for the analytical methods relied on values sourced from literature or manufacturer data. A finite element (FE) material model was implemented into ABAQUS/Explicit through a dynamic user subroutine (VUMAT). The model used continuum damage mechanics to alter the material stiffness matrix once the elastic strengths were exceeded.
SDOF analysis was shown to effectively predict the maximum mid-span displacement of glulam members subjected to blast loads, within a 20% error margin. However, the model was found to be incapable of consistently predicting the displacement and time of failure, especially for CLT panels, where up to 50% error was observed. This degree of error was attributed to the model’s inability to account for multiple failure modes, namely rolling shear and flexural failure. The resistance curves implemented in the SDOF models generally agreed with experimental results, particularly with regard to initial stiffness, and were deemed sufficiently accurate from the perspective of design.
The finite element models simulated specimen ultimate behaviour reasonably well. Relatively accurate analytical predictions were also obtained for both maximum mid-span displacements and corresponding times. However, computational issues with damage transfer prevented the modeling of repeated tests on CLT panels. The FE model was capable of producing resistance-displacement relationships which correlated well to experimental results, despite the presence of numerical fluctuations. This is a significant outcome for the potential application of FEA to blast behaviour of timber components, since SDOF models require resistance curves as input and are unable to predict the force-displacement response of members.
| Identifer | oai:union.ndltd.org:uottawa.ca/oai:ruor.uottawa.ca:10393/42623 |
| Date | 02 September 2021 |
| Creators | Oliveira, Damian |
| Contributors | Doudak, Ghasan, Viau, Christian |
| Publisher | Université d'Ottawa / University of Ottawa |
| Source Sets | Université d’Ottawa |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf |
Page generated in 0.0022 seconds