Parameterization Models for Scalable, Predictive Simulation of Production Machines (PARADICMA)

Projekttitel

PARADICMA - Parameterization Models for Scalable, Predictive Simulation of Production Machines

Thematische Stichwörter

Parameter, Komponentenmodelle, Simulation, Werkzeugmaschine

Kurzbeschreibung/Abstract

The success of a new production machine design depends on the predictability of the machine properties during the design stage. Reliable simulation-based predictions about the properties of a new machine require a detailed model description and lead to complex models, usually Finite Element Models (FE-Models). But a good agreement between simulation and reality can only be achieved with precise knowledge about the structure and particularly the parameters of nonlinear components. One problem related to the modelling task is that the required parameters are dependent on a very large number of influencing conditions. Even for a given set of influencing conditions essential parameters are often unknown and cannot be acquired (e.g. through measurements) with justifiable time and effort. As a result parameters for abstract component models are often estimated based on knowledge about the parameters of a seemingly similar system. Thus, incorrect predictions can result from intuitive assumptions that are made during the modelling and parameterization process.

The aim of this project therefore is to develop practically applicable methods for handling model parameterization and transferability problems for virtual prototypes of production machines. We propose a layered approach based on parameter models (bottom layer) embedded into linear and nonlinear component models including the join patch between components (intermediate layer), which, in turn, make up the machine model (top layer). Parameters, strategies for parameter determination and the associated general conditions will be analysed and generalised. We will characterise the determinant influences on parameters and analyse the model robustness towards parameter variations. In addition to that, we will study the transferability of model knowledge between

·         different model formulations used in different stages of the design process,

·         different scales of similar structures, e.g. small and large linear guides,

·         different levels of abstraction, e.g. material damping vs. modal damping, and

·         different system configurations with different general and boundary conditions.

Based on the won awareness new parameter models will be developed to replace the existing constant parameter models. This shall considerably simplify the system identification and transferability of parameter models.

As the result of the research proposed here, methods will be available for efficient and reliable predictive determination and characterisation of model parameters. These can be applied to build efficient, accurate and reliable virtual prototypes on the different levels of abstraction and formulations required throughout the production machine design process. Re-using parameter models and validated scalable component models will reduce the modelling effort and particularly the modelling error rate, thus enhancing the accuracy and reliability of simulative predictions.

Weiterführende Informationen

• SimTech

Ansprechpartner

• Dipl.-Ing. Alexandru Dadalau