NONLINEAR PROGRAMMING SOLVERS FOR HYBRID FINITE ELEMENT
The focus of this work is on analyzing and developing nonlinear solvers for performing nonlinear
structural analysis for large displacements in both elastic and inelastic cases. The response of a structure to a load application is shown by its equilibrium path which may include snap back and snap through behavior. Material and geometric nonlinearities are taken into consideration while developing the response path of a multi-element structure with sections discretized using fiber elements. Traditionally, Newton’s method is employed for solving the system of nonlinear equations but it comes with certain challenges. The response determination becomes difficult when stiffness matrix becomes singular at turning point. It also requires the calculation of the inverse of a Hessian matrix, which is costly. Newton’s method gives quadratic convergence but as the scale of the structure increases, resorting to Newton’s method becomes difficult.
These limitations motivate us to explore new solvers. Hence, in this study we analyze and develop various nonlinearly constrained optimization solvers for a recently suggested hybrid finite element. In particular, we compare the performance of conjugate gradient method with or without preconditioning, Sequential Quadratic Programming method and augmented Lagrangian method. For the case of structural response with snap back and snap through behavior, a new method called the implicit path continuation method is developed to ensure path continuation and solution convergence. The various solvers are then validated by obtaining responses of three benchmark structural problems with large displacements and rotations, and comparing the results with the conventional Newton’s method and a variant of Newton’s method with submatrices.
TABLE OF CONTENTS
In structural analysis, there is an increasing demand for performing analysis of large scale structural systems with nonlinear behavior. The load response of a structure can be seen physically in the form of deflections, rotations or vibrations. For a preliminary analysis, the material behavior is considered linearly elastic, i.e. stress (axial or bending) varies linearly with the corresponding strain and upon unloading the body recovers back to its original configuration. This is an idealized model which is very straightforward for obtaining structural element behavior. The stiffness matrix is constant in this case. The structural deformations can be also superimposed for this linear case.
However, for real life problems with large loads many complexities arise due to nonlinearities. Nonlinear phenomena arise due to physical variables related by nonlinear quantities. In nonlinear structural analysis, the relationship between the stress and strain becomes nonlinear and the stiffness matrix does not remain constant. When a structure is tested for large loads up to collapse, we observe bifurcations and instabilities in structural response. Hence, we define two types of nonlinearities: material and geometric. Material nonlinearly is commonly seen in all elements. As the name suggests, this type of nonlinearity is attributed to material behavior, as the force increases the stress-strain behavior becomes nonlinear. In many structures, geometric nonlinearity is also observed due to large displacements and large rotations. Geometric nonlinearity arises when the change in geometry becomes significant and is taken into account in equilibrium and compatibility equations. For small deformations, geometric nonlinearity does not come into picture as it introduces very small error. But as the displacements and rotations increase in magnitude,
Figure 1 Material Nonlinearity
Figure 2 Geometric Nonlinearity
the frame of reference is significantly different from the original one and the stress-strain curve of body is nonlinear.
To completely define the behavior of continuum problems, it is required to have deformation information for all the sections of the body. This is a very cumbersome task since initially we know only about the applied loading and the general body configuration. To obtain structural deformations and internal forces for a body, we need a solution strategy so that we scan define, analyze and obtain structural responses. Hence finite element methods are used which are very popular in several engineering fields and provide approximation to boundary value problems.
In displacement based finite element methods, compatibility is strictly enforced due to their formulation, but equilibrium of forces is not exact which results in certain errors. In force-based finite element methods, the nodal forces are unknowns and exact values of forces are obtained. Since compatibility is implemented by integrating, the deformations have now some errors in this case. In order to eliminate the limitations of these methods, mixed and hybrid finite element methods have been proposed in several works. From the available studies in the literature there is still need for development of methods that can overcome several limitations of current techniques.
This work is based on developing computational solutions which enhance the efficiency of hybrid finite element methods. In this regard, we explore various optimization techniques which are being successfully implemented in other fields of engineering. The aim is to explore optimization algorithms for solving the finite element problems by improving upon the classical methods. We want to improve the accuracy of solution and computational cost. In this regard, a hybrid finite element model was developed by  for large displacements and rotations and Newton’s method has been used to iteratively obtain the structural response. Hence the overall goal of this work is to search for alternative methods than second order Newton’s methods in solving problems with multiple degrees of freedom based on this recently suggested hybrid element.
- The new hybrid finite element formulation will be modeled and analyzed using Newton’s constrained optimization technique.
- The Newton’s equation will be solved using conjugate gradient method and preconditioned conjugate gradient method.
- The hybrid finite element problem will also be analyzed using sequential quadratic programming method.
- The hybrid finite element problem will also be analyzed using augmented Lagrangian method.
- For snap back and snap through case, Crisfield’s arc length method will be implemented and a new implicit path continuation method will be developed to capture structure behavior in these cases.
- The methods mentioned above will be used to check cantilever beam with vertical load, cantilever beam with moment at free end, toggle frame and Lee’s frame for elastic case and for inelastic case.
- A comparison of the performance of the above mentioned methods will be provided and future
directions will be discussed.
NONLINEAR PROGRAMMING SOLVERS FOR HYBRID FINITE ELEMENT