A series of novel algorithms for performing aerostructural shape and topology optimization are introduced and applied to the design of aircraft wings. An isoparametric level set method is developed for performing topology optimization of wings and other non-rectangular structures that must be modeled using a non-uniform, body-fitted mesh. The shape sensitivities are mapped to computational space using the transformation defined by the Jacobian of the isoparametric finite elements. The mapped sensitivities are then passed to the Hamilton-Jacobi equation, which is solved on a uniform Cartesian grid. The method is derived for several objective functions including mass, compliance, and global von Mises stress. The results are compared with SIMP results for several two-dimensional benchmark problems. The method is also demonstrated on a three-dimensional wingbox structure subject to fixed loading. It is shown that the isoparametric level set method is competitive with the SIMP method in terms of the final objective value as well as computation time. In a separate problem, the SIMP formulation is used to optimize the structural topology of a wingbox as part of a larger MDO framework. Here, topology optimization is combined with aerodynamic shape optimization, using a monolithic MDO architecture that includes aerostructural coupling. The aerodynamic loads are modeled using a threedimensional panel method, and the structural analysis makes use of linear, isoparametric, hexahedral elements. The aerodynamic shape is parameterized via a set of twist variables representing the jig twist angle at equally spaced locations along the span of the wing. The sensitivities are determined analytically using a coupled adjoint method. The wing is optimized for minimum drag subject to a compliance constraint taken from a 2g maneuver condition. The results from the MDO algorithm are compared with those of a sequential optimization procedure in order to quantify the benefits of the MDO approach. While the sequentially optimized wing exhibits a nearly-elliptical lift distribution, the MDO design seeks to push a greater portion of the load toward the root, thus reducing the structural deflection, and allowing for a lighter structure. By exploiting this trade-off, the MDO design achieves a 42% lower drag than the sequential result.

Thesis (M.A. Sc.)--University of Toronto, 2007. / Source: Masters Abstracts International, Volume: 45-06, page: 3197.

Thesis (Ph.D.)--Aerospace Engineering, Georgia Institute of Technology, 2009. / Committee Chair: Hodges, Dewey H.; Committee Member: Bauchau, Olivier A.; Committee Member: Johnson, Ellis; Committee Member: Makeev, Andrew; Committee Member: Volovoi, Vitali V.

Thesis (M.S.)--Brigham Young University. Dept. of Mechanical Engineering, 2007. / Includes bibliographical references (p. 65-67).

The main objective of this research is to develop a computational design tool for multilevel optimization of product-material systems under uncertainty. To accomplish this goal, an exponential penalty function (EPF) formulation based on method of multipliers is developed for solving multilevel optimization problems within the framework of Analytical Target Cascading (ATC). The original all-at-once constrained optimization problem is decomposed into a hierarchical system with consistency constraints enforcing the target-response coupling in the connected elements. The objective function is combined with the consistency constraints in each element to formulate an augmented Lagrangian with EPF. The EPF formulation is implemented using double-loop (EPF I) and single-loop (EPF II) coordination strategies and two penalty-parameter-updating schemes. The computational characteristics of the proposed approaches are investigated using different nonlinear convex and non-convex optimization problems. An efficient reliability-based design optimization method, Single Loop Single Vector (SLSV), is integrated with Augmented Lagrangian (AL) formulation of ATC for solution of hierarchical multilevel optimization problems under uncertainty. In the proposed SLSV+AL approach, the uncertainties are propagated by matching the required moments of connecting responses/targets and linking variables present in the decomposed system. The accuracy and computational efficiency of SLSV+AL are demonstrated through the solution of different benchmark problems and comparison of results with those from other optimization methods. Finally, the developed computational design optimization tool is used for design optimization of hybrid multiscale composite sandwich plates with/without uncertainty. Both carbon nanofiber (CNF) waviness and CNF-matrix interphase properties are included in the model. By decomposing the sandwich plate, structural and material designs are combined and treated as a multilevel optimization problem. The application problem considers the minimum-weight design of an in-plane loaded sandwich plate with a honeycomb core and laminated composite face sheets that are reinforced by both conventional continuous fibers and CNF-enhanced polymer matrix. Besides global buckling, shear crimping, intracell buckling, and face sheet wrinkling are also treated as design constraints.

A computationally efficient analysis capability for the geometrically non-linear response of compressively loaded prismatic plate structures was developed. Both a "full" finite strip solution procedure and a "reduced" solution procedure were implemented in a FORTRAN 90 computer code, and comparisons were made with results available in the technical literature. Both the full and reduced solution procedures were demonstrated to provide accurate results for displacement and strain quantities through moderately large post-buckling loads. The full method is a non-linear finite strip analysis of the semi-analytical, multi-term type. Individual finite strips are modeled as balanced and symmetric laminated composite materials which are assumed to behave orthotropically in bending, and the structure is loaded in uniaxial or biaxial compression. The loaded ends of the structure are assumed to be simply supported, and geometric shape imperfections may be modeled. The reduced solution method makes use of a reduced basis technique in conjunction with the full finite strip analysis. Here, the potentially large set of non-linear algebraic equations produced by the finite strip method are replaced by a small set of system equations. In the present implementation, the basis vectors consist of successive derivatives of the non-linear solution vector with respect to a loading parameter. Depending on the nature of the problem, the reduced solution procedure is capable of computational savings of up to 60%+ compared to the full finite strip method. The reduced method is most effective in reducing the computational cost of the full method when the most significant portion of the cost of the full method is factorization of the assembled system matrices. The robustness and efficiency of the reduced solution procedure was found to be sensitive to the user specified error norm which is used during the reduced solution procedure to determine when to generate new sets of basis vectors. In parallel with this effort, a new method for performing global/local design optimization of large complex structures (such as aircraft wings or fuselages) was developed. A simple and flexible interface between the global and local design levels was constructed using response surface methodology. The interface is constructed so as to minimize the changes required in either the global design code or the local design codes(s). Proper coupling is maintained between the global and local design levels via a "weight constraint" and the transfer of global stiffness information to the local level. The method was verified using a simple isotropic global wing model and the local panel design code PASCO. / Ph. D.

Sensitivity analysis is one of the most prominent gradient based optimization techniques for mechanical systems. Model sensitivities are the derivatives of the generalized coordinates defining the motion of the system in time with respect to the system design parameters. These sensitivities can be calculated using finite differences, but the accuracy and computational inefficiency of this method limits its use. Hence, the methodologies of direct and adjoint sensitivity analysis have gained prominence. Recent research has presented computationally efficient methodologies for both direct and adjoint sensitivity analysis of complex multibody dynamic systems. Multibody formulations with joint friction were developed in the recent years and these systems have to be modeled by highly non-linear differential algebraic equations (DAEs) that are difficult to solve using numerical methods. The sensitivity analysis of such systems and the subsequent design optimization is a novel area of research that has been explored in this work. The contribution of this work is in the development of the analytical methods for computation of sensitivities for the most commonly used multibody formulations incorporated with joint friction. Two different friction models have been studied, capable of emulating behaviors of stiction (static friction), sliding friction and viscous drag. A case study has been conducted on a spatial slider-crank mechanism to illustrate the application of this methodology to real-world systems. The Brown and McPhee friction model has been implemented using an index-1 formulation for computation of the dynamics and sensitivities in this case study. The effect of friction on the dynamics and model sensitivities has been analyzed by comparing the sensitivities of slider velocity with respect to the design parameters of crank length, rod length, and the parameters defining the friction model. Due to the highly non-linear nature of friction, it can be concluded that the model dynamics are more sensitive during the transition phases, where the friction coefficient changes from static to dynamic and vice versa. / Master of Science / Mechanisms have been in existence since the earliest days of technology and are more relevant than ever in this age of robotics, artificial intelligence and space exploration. Innovations like myoelectric and neural prosthetics, legged robotics, robotic surgeries, advanced manufacturing, extra-terrestrial vehicles and so on are the modern day manifestations of the traditional mechanisms that formed the backbone of the industrial revolution. All of these innovations implement precision controlled multibody dynamic systems as part of their function. This thesis explores the modelling of such dynamic systems using different mathematical formulations. The contribution of this work is the incorporation of friction in the formulation of such systems. The performance of any dynamical system depends on certain parameters, which can be optimized to meet a certain objective criteria. This is achieved by performing a sensitivity analysis with respect to those parameters on the mathematical formulation of the mechanism. The derivation of this approach has been explored in this thesis. For the benefit of the reader, the application of this method has been discussed using a case study of a simple 3-dimensional slider crank mechanism.

The goal of this thesis is to use biomechanical data describing shoulder motion to determine optimal parameters to assist in the design of a 5 DOF active shoulder exoskeleton. This thesis will provide a proof of concept on optimization techniques using motion data using a simplified 3 DOF model to facilitate future work implementing a full 5 DOF model. Optimization will be performed to determine the link lengths and, consequently, the locations of the joints of the exoskeleton by considering the human's workspace to maximize range of motion and promote user safety by minimizing collisions of the exoskeleton with the user and with the exoskeleton itself. The thesis will detail the development of computational models of the human and proposed exoskeleton, the processing of experimental data used to estimate the human's capabilities, optimization, and future work. This work will contribute to a large-scale NSF-funded project of building an upper body exoskeleton emulator. The emulator will promote the widespread adoption of exoskeletons in industry by providing a test-bed to streamline the rapid design of various assistance profiles for various users and tasks. / Master of Science / An exoskeleton is a robotic assistive device used in industrial and rehabilitative settings. This thesis will use data describing how the human shoulder moves during certain tasks to help design an exoskeleton to assist with theses tasks. A model of the human shoulder and a model of the exoskeleton will be developed and used in an optimization to figure out the best dimensions of the exoskeleton links to support the human's movements.

This thesis describes the design optimization of two different types of vessels. They are LHA(R), a replacement for the US Navy amphibious assault ship and DDG51, a destroyer class vessel. The overall measure of effectiveness (OMOE) and the lead ship acquisition cost (LCA) are considered to be the objective functions. The evaluation of feasibility of the designs and various ship parameter calculations are performed using the US Navy ship design evaluation software ASSET. ASSET is integrated with the design optimization software DARWIN to obtain results representing the best designs over a range of LCA. Model Center software is used to integrate the processes ASSET and Darwin. The results generated will provide the owner with the best designs possible (designs with high OMOE) over a range of LCA. This thesis is mainly of academic interest. The results generated could help the owners to look at various design options available for the amount of money they are willing to spend. / Master of Science

Structural analysis and design optimization is important to a wide variety of disciplines. The current methods for these tasks require significant time and computing resources. Reconfigurable computers have shown the ability to speed up many applications, but are unable to handle efficiently the precision requirements for traditional analysis and optimization techniques. Cellular automata theory provides a method to model these problems in a format conducive to representation on a reconfigurable computer. The calculations do not need to be executed with high precision and can be performed in parallel. By implementing cellular automata simulations on a reconfigurable computer, structural analysis and design optimization can be performed significantly faster than conventional methods. / Master of Science