LS-OPT 3.3 released
Problem Description
Optimization
LS-OPT papers dealing with optimization
An Overview of New Features in LS-OPT Version 3.3
N. Stander, T. Goel (Livermore Software Technology Corporation - LSTC); David Björkevik (Engineering Research Nordic) This paper summarizes the development status of LS-OPT® Version 3.3 and focuses mainly on the following new features: (i) Radial Basis Function Networks, (ii) Multi-objective Optimization (MOO), (iii) Metamodel-based MOO, (iv) User-defined metamodeling, (v) User-defined queuing
Multi-Objective Optimization Using LS-OPT
Goel T., Stander N.; Conference Proceedings, 6th DYNAmore User Forum, Frankenthal, Germany, 2007 Most engineering optimization problems have more than one objective. Often these objectives conflict such that no single solution can be considered optimum with respect to all objectives. Then, the optimum to this problem is a set of solutions known as Pareto optimal set....
On the Robustness of a Simple Domain Reduction Scheme for Simulation-Based Optimization
Stander N., Craig K.J.; Livermore Software Technology Corporation, Livermore, USA, 2002 This paper evaluates a Successive Response Surface Method (SRSM) specifically developed for simulation-based design optimization, e.g. that of explicit nonlinear dynamics in crashworthiness design. Linear response surfaces are constructed in a subregion of the design space using a design of experiments approach with a D-optimal experimental design...
ANSA as a Pre-Processor for LS-OPT - Optimization Applications
Korbetis G. BETA CAE, DYNAmore Infoday, June 2008
Analyzing 'Noisy' Structural Problems with LS-OPT: Probabilistic and Deterministic Fundamentals
Willem Roux and Nielen Stander Livermore Software Technology Corporation, Livermore CA System identification of 'noisy' structural design optimization problems: the sources of uncertainty, the competing roles of bias and variance, and the interaction of uncertainty and deterministic effects. Two test problems are used to clarify the effect of different approaches.
Crashworthiness Optimization in LS-OPT: Case Studies in Metamodeling and Random Search Techniques
Nielen Stander, Willem Roux, Mathias Giger, Marcus Redhe, Nely Fedorova, Johan Haarhoff; 4th European User Conference 2003; This crashworthiness optimization study compares the use of three metamodeling techniques while using a sequential random search method as a control procedure. The three methods applied are (i) the original Successive Linear Response Surface Method, (ii) the Neural Network method and (iii) the Kriging method. It is shown that, although NN and Kriging seem to require a larger number of initial points, the three metamodeling methods have comparable efficiency. The random search method is surprisingly efficient in some instances, but by nature much less predictable.
Shape Optimization of a Crashbox using HyperMorph and LS-OPT
H. Wang (Benteler), H. Müllerschön (DYNAmore), T. Mehrens (HAW Hamburg) The aim of this paper is to demonstrate how to improve the performance of a crashbox by the application of shape optimization using HyperMorph and LS-OPT. Two load cases for the crashbox are considered: low speed front crash with initial load in x-direction and low speed crash with initial load 10° rotated with respect to the x-axis. Possible beads and geometry variations are parameterized with HyperMorph. The parameters that define the shape variations in HyperMorph are controlled by LS-OPT and be exchanged via a dedicated interface. The main quality criterion for the crashbox is the smoothness of the force values which occur during the energy absorption. This means, a force-intrusion curve with a horizontal line, after a specific force level is reached, would be the ideal case. The formulation of the optimization problem takes this into account by minimizing the difference between the maximum and the minimum force values of the force-intrusion curve during the folding process. Simultaneously several restrictions regarding the producibility and the folding mechanism have to be considered.
Reliability Based Design Optimization with LS-OPT for a Metal Forming Application
H. Müllerschön, D. Lorenz (DYNAmore), Prof. K. Roll (Daimler) The purpose of this paper is to account for uncertainties in the manufacturing processes of metal forming in order to evaluate the random variations with the aid of FE-simulations. Various parameters of the Finite-Element model describing the investigated structural model are affected by randomness. This, of course, leads to a variation of the considered simulation responses such as stresses, displacements, and thickness reductions. On this, for the simulation engineer basic questions arise regarding: (1) the dimension of the range of variation of the simulation responses (2) the significance/contribution of the (input) parameters with respect to specific responses and (3) the reliability of the process design with respect to constraints (failure, damage, requirements, ...). In order to find solutions to these questions several methodologies may be applied that are available in the commercial optimization software LS-OPT (Stander et al. [5]). Some of the methodologies, such as Monte Carlo simulation, meta-model based Monte Carlo simulation, stochastic fields, are discussed in this paper and are demonstrated by means of a metal forming problem. For this, a non-robust design with respect to the specified constraints has been detected. By utilizing reliability based design optimization (RBDO) through LS-OPT, the failure probability (violation of constraints) could be reduced significantly.
FE-Simulation Based Optimization of an Adaptive Restraint System Considering Multiple Front-Crash Load Cases using LS-OPT
M. van den Hove, B. Mlekusch (AUDI), H. Müllerschön (DYNAmore) The purpose of this paper is to explore some interesting aspects of optimization for crashworthiness occupant safety applications and to propose optimization strategies for highly nonlinear problems. With the today’s state of technology it is possible to identify specific load cases and different types of occupants in the car. System parameters of the restraint system, such as trigger time for seat-belt, airbag and steering column can be adapted to particular load cases. This is refered to an adaptive restraint system. In the first part of the paper different optimization strategies are discussed and pros and cons are compared. In addition, a methodology to get a reliable surrogate model using neural networks is introduced. The surrogate model (Meta-Model or Response Surface Model) approximates the relationship between design parameters and a physical response and can be used to visualize and explore the design space. In the second part the application of the Successive Response Surface Scheme (SRSM) for the optimization of an adaptive restraint system is conducted. For this, several front crash load cases are considered. This is performed using LS-OPT (Stander et al. [11]) as optimization software and PAM-Crash as solver for the finite element occupant safety simulations. The procedure of generating an advanced meta-model to get an approximation of the global design space using neural networks is demonstrated for this example. Furthermore, the visualization of multi-dimensional meta-models in two- and three-dimensional design space is illustrated.
Two-Stage Stochastic and Deterministic Optimization
T. Rzesnitzek, H. Müllerschön, F. Günther, M. Wozniak (2002)
