COST 526 Automatic Process Optimization in Materials Technology (APOMAT)

COST 526

Automatic Process Optimization

in Materials Technology (APOMAT)

Summary

The initiative for this COST Action has grown out of the view that empirical knowledge and computational approximation form the basis of process design and control in process engineering today. This is time-consuming, cost-intensive and also often inaccurate. This is why numerical process simulation is finding increasing application, underpinned by the normally large quantity of free parameters which are involved in very many of these processes and which decide both the quality of the product and the quality of the process. The innovative core of this project is to link the simulator to numerical optimization processes for automatic computation of those parametric configurations which lead to optimization in terms of complex targets.

COST Action 526 is aimed at initiating a new materials technology methodology: numerical optimization for automatic computation of optimized process parameters in the manufacture and processing of materials and components. This involves a quantitative linkage of product quality, process windows, including technical and economic limitations, to process parameters, and the formulation of these factors as a so-called objective function. A numerical process simulation model delivers computed quantities for evaluation of the objective function, in whose value ranges optimization algorithms are applied for computing local extreme values as the optimized process parameters being searched for.

The move towards shorter product cycles and cost-oriented design over the last 20 years has promoted the development of numerical simulation models. These virtual process models are now firmly established among the tools of material and component manufacture. This involves integration of the laws governing materials into the conservation equations of temperature, stress and flow, and calculation of complex, three-dimensional geometries in terms of numerical approximate solutions. The length scales currently range from microsimulation at microscopic space-time resolution to complete plant components or process stages dimensioned in meters. Simulation models of this kind enable visualization of three-dimensional data reflecting the current status of processes normally difficult to capture experimentally. Furthermore, these simulations provide a virtual experimental environment, enabling investigation of the effects of various parameter sets on objective quantities (Fig. 1).

The essentially new aspect provided by COST 526 is to link simulation models to numerical optimization techniques (Fig. 2). This methodology enables automatic computation of the relevant process parameters for given material properties (Fig. 3).

COST 526 is initiating large-scale inter-disciplinary cooperation between material scientists, simulation engineers and optimization experts at the European level aimed at:

defining material properties via constitutive material models and designing process-specific objective functions
making available accurate and high-performance simulation software to defined quality standards
solving multi-dimensional, non-linear optimization tasks.

A major goal of COST 526 at the organizational level is to develop and demonstrate this new methodology for a series of technical material processes of clearly identifiable optimization potential.

Fig. 1: Numerical simulation as a virtual experimental environment is state-of-the-art technology. Investigation can be made into the effects of a range of parameter sets on the target quantities.

Fig. 2: The theme of COST 526 is automatic, numerical process optimization, i.e. linking simulation models to numerical optimization techniques using so-called target or quality functions.

Fig. 3: Numerical process optimization enables automatic computation of the relevant process parameters for given material properties.

back