Purdue University Graduate School
msg_design_dissertation_221207.pdf (13.86 MB)


Download (13.86 MB)
posted on 2022-12-09, 15:10 authored by Su TianSu Tian

Composite materials have been invented and used to make all kinds of industrial products, such as automobiles, aircraft, sports equipment etc., for many years. Excellent properties such as high specific stiffness and strength have been recognized and studied for decades, motivating the use of composite materials. However, the design of composite structures still remains a challenge. Existing design tools are not adequate to exploit the full benefits of composites. Many tools are still based on the traditional material selection paradigm created for isotropic homogeneous materials, separated from the shape design. This will lose the coupling effects between composite materials and the geometry and lead to less optimum design of the structure. Hence, due to heterogeneity and anisotropy inherent in composites, it is necessary to model composite parts with appropriate microstructures  instead of simplistically replacing composites as black aluminum and consider materials and geometry at the same time.

This work mainly focuses on the design problems of complex material-structural systems through computational analyses. Complex material-structural systems are structures made of materials that have microstructures smaller than the overall structural dimension but still obeying the continuum assumption, such as fiber reinforced laminates, sandwich structures, and meta-materials, to name a few. This work aims to propose a new design-by-analysis framework based on the mechanics of structure genome (MSG), because of its capability in accurate and efficient predictions of effective properties  for different solid/structural models and three-dimensional local fields (stresses, strains, failure status, etc). The main task is to implement the proposed framework by developing new tools and integrating these tools into a complete design toolkit. The main contribution of this work is a new efficient high-fidelity design-by-analysis framework for complex material-structural systems.

The proposed design framework contains the following components. 1) MSG and its companion code SwiftComp is the theoretical foundation for structural analysis in this design framework. This is used to model the complex details of the composite structures. This approach provides engineers the flexibility to use different multiscale modeling strategies. 2) Structure Gene (SG) builder creates finite element-based model inputs for SwiftComp using design parameters defining the structure. This helps designers deal with realistic and meaningful engineering parameters directly without expert knowledge of finite element analysis. 3) Interface is developed using Python for easy access to needed data such as structural properties and failure status. This is used as the integrator linking all components and/or other tools outside this framework. 4) Design optimization methods and iteration controller are used for conducting the actual design studies such as parametric study, optimization, surrogate modeling, and uncertainty quantification. This is achieved by integrating Dakota into this framework. 5) Structural analysis tool is used for  computing global structural responses. This is used if an integrated MSG-based global analysis process is needed.

Several realistic design problems of composite structures are used to demonstrate the capabilities of the proposed framework. Parameter study of a simple fiber reinforce laminated structure is carried out for investigating the following: comparing with traditional design-by-analysis approaches, whether the new approach can bring new understandings on parameter-response relations and because of new parameterization methods and more accurate analysis results. A realistic helicopter rotor blade is used to demonstrate the optimization capability of this framework. The geometry and material of composite rotor blades are optimized to reach desired structural performance. The rotor blade is also used to show the capability of strength-based design using surrogate models of sectional failure criteria. A thin-walled composite shell structure is used to demonstrate the capability of designing variable stiffness structures by steering in-plane orientations of fibers of the laminate. Finally, the tool is used to study and design auxetic laminated composite materials which have negative Poisson's ratios.


Degree Type

  • Doctor of Philosophy


  • Aeronautics and Astronautics

Campus location

  • West Lafayette

Advisor/Supervisor/Committee Chair

Wenbin Yu

Additional Committee Member 2

William A. Crossley

Additional Committee Member 3

Dianyun Zhang

Additional Committee Member 4

Gary J. Cheng