PhD Oral Exam - Mehdi Nikoo, Civil Engineering

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

Cross-laminated timber (CLT) is increasingly used in structural applications, but the characterization of rolling shear strength and prediction of panel flexural capacity are affected by test configuration, specimen geometry, material variability, and modeling assumptions. This research investigates the rolling shear strength and flexural capacity of CLT using experimental testing, regression, and AI models. Rolling shear strength was examined in two experimental stages using the Modified Planar Shear (MPS) test. First, 18 three- and five-layer SPF CLT specimens were tested using three inclination-angle selection approaches. Based on the measured response, repeatability, and failure localization, the fixed 14° configuration produced the most consistent rolling shear response for both layups within the tested conditions. A second experimental program included 23 CLT specimens with different layer configurations and geometries to examine the effects of layup, specimen width, and loading configuration on rolling shear strength, stiffness, and failure mode. These experimental data were combined with published results to establish a database of 128 SPF CLT specimens. Moth-Flame Optimization (MFO) combined with multiple linear and nonlinear regression as well as artificial neural networks (ANNs) to predict rolling shear strength. Among these, the MFO-ANN-13 model matched the measured data most closely, with an R² of about 0.96 and an RMSE under 0.10 MPa. A separate database of 110 four-point bending tests from published studies was used for the flexural analysis. Density, moisture content, number of layers, width, span, thickness, local bending stiffness, and global bending stiffness were used as numerical inputs, while wood species was represented by one-hot encoding. The MFO-optimized ANN was evaluated using 42 two-hidden-layer topologies and compared with MFO-based regression models and the Gamma and Shear Analogy methods. The MFO-ANN-6-4 topology provided the best prediction of maximum bending moment capacity, with a test R² of 0.91, MAE of 2.77, RMSE of 3.56, VAF of 92%, and a correlation coefficient of 0.96. The results indicate that MPS test configuration affects the measured rolling shear strength and that the developed AI models can predict rolling shear strength and flexural capacity within the material, geometric, and parameter ranges represented by the respective databases.