No abstract

Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instruction, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden, to Washington ABSTRACT (maximum 200 words)A Galerkin-based family of numerical formulations is presented for solving nonlinear optimal control problems. This dissertation introduces a family of direct methods that calculate optimal trajectories by discretizing the system dynamics using Galerkin numerical techniques and approximate the cost function with Gaussian quadrature. In this numerical approach, the analysis is based on L 2 -norms. An important result in the theoretical foundation is that the feasibility and consistency theorems are proved for problems with continuous and/or piecewise continuous controls. Galerkin methods may be formulated in a number of ways that allow for efficiency and/or improved accuracy while solving a wide range of optimal control problems with a variety of state and control constraints. Numerical formulations using Lagrangian and Legendre test functions are derived. One formulation allows for a weak enforcement of boundary conditions, which imposes end conditions only up to the accuracy of the numerical approximation itself. Additionally, the multi-scale formulation can reduce the dimension of multi-scale optimal control problems, those in which the states and controls evolve on different timescales. Finally, numerical examples are shown to demonstrate the versatile nature of Galerkin optimal control. SUBJECT TERMS

Solving potential problems, such as those that occur in the analysis of steady-state heat transfer, electrostatics, ideal fluid flow, and groundwater flow, is important in several fields of engineering, science, and applied mathematics. Numerical solution of the relevant governing equations typically involves using techniques such as domain methods (including finite element, finite difference, or finite volume), or boundary element methods (using either real or complex variables). In this paper, the Complex Variable Boundary Element method ("CVBEM") is examined with respect to the use of different types of basis functions in the CVBEM approximation function. Four basis function families are assessed in their solution success in modeling an important benchmark problem in ideal fluid flow; namely, flow around a 90 degree bend. Identical problem domains are used in the examination, and identical degrees of freedom are used in the CVBEM approximation functions. Further, a new computational modeling error is defined and used to compare the results herein; specifically, M = E / N where M is the proposed computational error measure, E is the maximum difference (in absolute value) between approximation and boundary condition value, and N is the number of degrees of freedom used in the approximation.

No abstract

No abstract

The Complex Variable Boundary Element Method, or CVBEM, was first published in Journal of Numerical Methods in Engineering in year 1984 by authors Hromadka and Guymon [1]. Since that time, several papers and books have been published that present various aspects of the numerical technique as well as advances in the computational method such as extension to three or higher dimensions for arbitrary geometries, nonhomogeneous domains, extension to use of a Hilbert Space setting as well as collocation methods, inclusion of the time derivative via coupling to generalized Fourier series techniques, examination of various families of basis functions including complex monomials, the product of complex polynomials with complex logarithm functions (i.e., the usual CVBEM basis functions), Laurent series expansions, reciprocal of complex monomials, other complex variable analytic functions including exponential and others, as well as linear combinations of these families. Other topics studied and developed include rotation of complex logarithm branch-cuts for extension of the problem computational domain to the exterior of the problem geometry, depiction of computational error in achieving problem boundary conditions by means of the approximate boundary technique, mixed boundary value problems, flow net development and visualization, display of flow field trajectory vectors in two and three dimensions for use in depicting streamlines and flow paths, among other topics. The CVBEM approach has also been extended to solving partial differential equations such as Laplace's equation, Poisson's equation, unsteady flow equation, and the wave equation, among other formulations that include sources, sinks and combinations of these equations with mixed boundary conditions. In the current paper, a detailed examination is made of the performance between four families of basis functions in order to assess computational efficiency in problem solving of two dimensional potential problems in a high aspect ratio geometric problem domain. Two selected problems are presented as case studies to demonstrate the different levels of success by each of the four families of examined basis functions. All four families involve basis functions that solve the governing partial differential equation, leaving only the goodness of fit in matching boundary conditions of the boundary value problem as the computational optimization goal. The modeling technique is implemented in computer programs Mathematica and MATLAB. Recommendations are made for future research directions and lessons learned from the current study effort.

In this work, a conceptual numerical solution of the two-dimensional wave partial differential equation (PDE) is developed by coupling the Complex Variable Boundary Element Method (CVBEM) and a generalized Fourier series. The technique described in this work is suitable for modeling initial-boundary value problems governed by the wave equation on a rectangular domain with Dirichlet boundary conditions and an initial condition that is equal on the boundary to the boundary conditions. The new numerical scheme is based on the standard approach of decomposing the global initial-boundary value problem into a steady-state component and a time-dependent component. The steady-state component is governed by the Laplace PDE and is modeled with the CVBEM. The time-dependent component is governed by the wave PDE and is modeled using a generalized Fourier series. The approximate global solution is the sum of the CVBEM and generalized Fourier series approximations. The boundary conditions of the steady-state component are specified as the boundary conditions from the global BVP. The boundary conditions of the time-dependent component are specified to be identically zero. The initial condition of the time-dependent component is calculated as the difference between the global initial condition and the CVBEM approximation of the steady-state solution. Additionally, the generalized Fourier series approximation of the time-dependent component is fitted so as to approximately satisfy the derivative of the initial condition. It is shown that the strong formulation of the wave PDE is satisfied by the superposed approximate solutions of the time-dependent and steady-state components.

Abstract-A new numerical technique is presented for solving optimal control problems. This paper introduces a direct method that calculates optimal trajectories by discretizing the system dynamics using Galerkin numerical techniques and approximates the cost function with quadrature. We show that the Galerkin optimal control method with weak enforcement of boundary conditions leads to improved solution accuracies. Furthermore, we show that a feasible solution exists to the approximation of the constrained nonlinear optimal control problem. Using an example, the Galerkin method described in this paper is shown to be more accurate than some existing methods.