The aim of our research is to create better modeling approaches that allow creating and understanding innovative, multifunctional structures under various conditions and loading. This is performed under the current research program titled “Intelligent and Green Marine Vessels”, which is supported by Seaspan Shipyards, NSERC, Vard Marine, Robert Allan and Serco. Several other companies and institutions are involved in some of our research areas.
Main focus areas are described below.
Structural Optimization and Generative Design
We are developing algorithms for topology optimization of complex structures. These algorithms allow computers to create new structural forms (previously unknown), and to improve efficiency of existing designs. We are advancing a few aspects of optimization frameworks in order to handle large number of failure criteria and production requirements, primarily for ships that need to comply with industry regulations.
Main research directions are:
- Neural-reparameterized topology optimization approaches considering structural integrity and vibroacoustics
- Generative AI models
- Constraint-handling approaches to improve swarm and evolutionary algorithms
For more information, please click here
Deep Learning Models
We are working on several types of data-driven and physics-guided models as reduced-order models for analysis and optimization. The aim is to create improved models for early design of complex products where typically finite element analysis is computationally prohibitive. The applications span structural analysis, shape optimization and wave propagation.
For more information, please click here
Mechanics of Lightweight, Multifunctional Structures
We work on modeling nonlinear response (quasi-static, dynamic and shock/impact) of advanced materials and structures for complex vessels. This involves understanding fundamental properties of functionally graded structures (FGMs), sandwich panels and improved stiffened panels.
Lightweight structures can experience high shear deformation which has to be modeled properly. This is done through nonlinear homogenization which allows effective finite element analysis. For this purpose we are developing high-order, non-linear, non-local beam and plate models.
For more information on FGMs, please click here
For more information on nonlinear homogenization, please click here
Welding, Fatigue and Structural Integrity
In order to improve structural integrity and reduce imperfections in metallic and hybrid components, we investigate high-quality, low heat-input welding techniques using experimental and computational techniques. The latter consist of FEA and data-driven models.
For more information, please click here
