Historical Development of ICMECopyright: Martin Braun
Integrative Computational Materials Engineering (ICME) is aimed at describing the structure development on all relevant length scales as well as the structure-property correlation along the entire process chain up to the use of materials. In order to be able to better grasp the concept, a historical outline of the development of this still very young discipline is needed.
As computers and computation represent the key ingredient of ICME, the development of this discipline follows the development of computers and their use in society development of computers and computational power goes along with significant changes in their applications. In the 1960s computers filled entire rooms, still had limited capabilities and essentially were operated by experts only. Computers of the 21st century find place on a table with their performance largely exceeding the one of a 1960s computer. Their “operators” changed into “users” and in general are normal people without special computational skills. Nowadays research and development could not proceed as rapidly without computers.
Increasing computational capabilities increasingly allowed for mimicking of real processes in a virtual world. For this purpose models had to be developed, allowing describing physical processes on a computer at all. The most prominent method is the description of phenomena occurring in the continuum of the real world on a numerical grid using Finite Difference Methods “FDM” and Finite Element Methods “FEM” and their further derivatives. These meanwhile have reached a degree of maturity making them even applicable for the qualification and certification of products.
Respective finite element methods, extended FE-methods “X-FEM”, computational fluid dynamics “CFD” and computational damage mechanics “CDM” have been extremely successful and a number of relevant software packages are frequently used to describe and optimize individual processes in the frame of computational engineering “CE”. Most progress in terms of integrative computational engineering “ICE” by now has been made in the area of structural materials and their mechanical properties. Computer simulations of macroscopic processes meanwhile range from CAD-data to the finished product and in the last decade have been intensively used in many fields of application like e.g. virtual crash tests.
Material data entering such simulations, however, by now in many cases have to be taken from experiments, from literature or from other sources of information. Due to a lack of more detailed information or high costs for their determination, such data are frequently assumed as constant, isotropic, homogeneous and/or based on other simplifications. Values for a dedicated material often are not available and are then approximated by drawing on similar materials. The variation of these values across the component, their dependence on temperature and their anisotropy by now in general are even entirely neglected.
The necessity for such approximations is due to the fact that computational models for materials allowing the prediction of respective materials properties did not keep pace with the rapid developments of the macroscopic FEM models.
The historic development of computational materials aiming at the prediction of materials properties at least has two different roots, one originating from the atomistic scale in a bottom-up approach and one starting from a thermodynamic, statistical perspective. Especially the latter has proven the potential to tackle technical alloy systems. Based on the description of the Gibbs energy of the individual phases and the development of respective models, the CALPHAD method has been established in the 1970s to assess a variety of data and combine them into respective, suitable databases. Respective software and databases such as e.g. Thermo-Calc, JMatPro, FactSage or Pandat have meanwhile become key tools for modern alloy development.
The thermodynamic nature of these programs and databases, however, in general only allows the prediction of equilibrium conditions like the fractions of individual phases being in equilibrium at a given temperature, the on-set temperatures for the formation of specific phases and many other most interesting thermodynamic data. Information can however be neither drawn on the evolution of these phase fractions in time (being mandatory to describe the kinetics of phase evolution) nor about their distribution in space (essentially corresponding to the microstructure, which eventually defines the properties of the material).
In the late 1990s, theoretical developments in the area of microstructure modeling such as the phase-field theory and multiphase-field models have been combined with above thermodynamic models. Such combined models nowadays provide the key to describe and to control the microstructure evolution and eventually to tailor effective properties of technical materials and products. The required data and parameter sets for such models can be obtained from even more fundamental models such as the density functional theory, atomistic modelling and molecular dynamic simulations.
Currently all these approaches have reached a level allowing for valuable contributions to modern engineering tasks within knowledge driven production models. The capabilities of the respective software tools and the present availability of computational power make efforts towards the integration of all these approaches possible, meaningful and timely. Such integration offers a relevant step forward in the direction of knowledge based optimal design of tailored materials, components and products with regard to their specific applications. Future “Integrated computational materials engineering (ICME) as an emerging discipline aiming to integrate computational materials science tools into a holistic system will accelerate materials development, transform the engineering design optimization process, and unify design and manufacturing.”