CADORS: Scientific Concepts
Prof. Franck Tancret*
The need for alloy design
Metallic alloys are used for a variety of industrial applications, the latter requiring the constant development of new materials with tailored sets of properties. For years, designing new alloys has been made by trial-and-error which remains long and expensive. However, alloy performance results from the combination of interactions between alloying elements, processing, microstructure and service conditions. Alternative routes for alloy design have thus been developed, and still need to progress; this is the aim of the present project “CADORS”.
State-of-the-art: computational alloy design
New methods have been proposed in the past twenty years, relying on the prediction of alloy properties using flexible regression tools like Neural Networks, NN, or Gaussian Processes, GP. The regressions are made on large databases containing the composition, processing parameters and properties of many existing materials, coming from available data. Such computing tools, once trained on data and assessed, can be trusted to perform reliable predictions in the case of new compositions, and hence to serve as guides for the design of new alloys. In parallel, the properties and processability of alloys being linked to microstructure, in particular to the nature and amounts of the phases they are made of, computational tools have been developed to calculate phase diagrams in multicomponent systems, within the “CALPHAD” framework (CALculation of PHAse Diagrams), or “computational thermodynamics”. Such tools can also be used as guides to design new alloys.
Nevertheless, these tools, i.e. computational thermodynamics or regression tools, if used alone, cannot match all the design requirements. Indeed, the latter usually comprise both mechanical properties targets, which can be addressed by NN or GP, as well as microstructural and/or thermodynamic issues related to processing and/or to phase stability, which can be addressed by computational thermodynamics. In this respect, some recent alloy design approaches made use of both techniques, on the one hand to predict mechanical properties with NN or GP, and on the other hand to check microstructural stability and/or processability with computational thermodynamics. However, although the above method uses computing techniques, the design procedure itself remains rather “manual”, because the search for the set of required properties is also made by trial and error, through individual predictions. Of course, a programming interface may render possible an automatic exploration of a given alloy system. Nevertheless, the computing cost could quickly become not acceptable. Indeed, if for example ten elemental concentrations are varied within a given interval with fifty possible levels, this would make 5010 possible alloys, i.e. 5010 phase diagram calculations. On modern computers, the computation time could reach billions of years! Besides, works have been done on the use of genetic algorithms, or “evolutionary computation”, to deal with the automatic optimisation of complex problems in materials science, for instance in combination with neural networks, or in combination with computational thermodynamics. More recently, we have designed new Ni-base superalloys using a combination of Gaussian processes, computational thermodynamics and genetic algorithms.
Objectives of the CADORS project
The present project aims at taking significant steps forward to increase the performance of the existing approach, by developing new physically-based predictive tools, for an integration in the above-mentioned alloy design procedures. Indeed, to improve the alloy design method, there is a need to add new alloy characteristics to the set of existing ones. As explained earlier, the presently used criteria include the prediction of thermomechanical properties by GP, as well as thermodynamical simulation to predict the phases present at equilibrium and long-term microstructural stability. The present project aims at adding two other important predictive criteria in view of alloy design, on the one hand on dynamic recrystallisation, DRX (to insure a better processability through the production of homogeneous and controlled microstructures) and on the other hand on resistance to hydrogen embrittlement, RHE.
Research axes of the CADORS project
• Overall methodology of the CADORS project
As stated above, the main scientific objectives of the project are to add and to integrate, in a multi-objective alloy design strategy (which uses other already existing models), two new predictive tools relying on physical models, one on dynamic recrystallisation (DRX) and another on the resistance to hydrogen embrittlement (RHE). The developed method will be applied to the design of new Ni-base alloys for specific applications (energy production, aeroengines, chemical engineering…).
Therefore, the research methodology is the following:
• Development of a dynamic recrystallisation model
It is here proposed to extend and adapt existing DRX models to the case of complex and/or multi-phase Ni alloys, to produce at least semi-quantitative predictions or criteria for alloy design. Indeed, most industrial superalloys are multi-phase, containing, for instance, carbides and/or intermetallics. And, it is obvious from the literature that precipitates, either intergranular or intragranular, and depending on their nature (carbides or intermetallics), on their size and on their volume fraction, can have an influence on DRX. For instance, large precipitates concentrate strains and dislocations around them, and can favour DRX by acting as nucleation sites for the formation of new grains, through the mechanism called “particle-stimulated nucleation” (PSN). Inversely, small precipitates can act as pinning centres for grain boundaries, limiting their mobility and hence retarding or inhibiting DRX. Such phenomena have been evidenced in nickel alloys, with only scarce modelling attempts. Nevertheless, computational thermodynamics allow the prediction of the nature and fractions of secondary phases, as well as of the driving force for nucleation, which can be used to evaluate precipitate size and density . The present project aims at using such predictions as quantitative inputs in DRX models, linking precipitate size and density both to the density of nucleation sites for new grains and to the pinning force on grain boundaries, to account for both PSN and the reduction of grain boundary mobility.
• Development of a model for the resistance to hydrogen embrittlement
The second subject of the project in on RHE, for inclusion as an additional alloy design criterion. Most metals suffer from H embrittlement, depending on conditions of temperature and H activity, and the aim of the present project is to predict the RHE of Ni-base superalloys. In the latter, microstructure plays a role through intermetallics like γ’, γ’’, δ or η, or through carbides. The RHE in Ni alloys depends, among others, on the distribution of H between matrix, grain boundaries, precipitate-matrix interfaces and precipitates themselves, which has been described in rare models. Effects can also depend on precipitate size. Nevertheless, as already mentioned, computational thermodynamics allow the prediction of the nature and fractions of secondary phases, and of the driving force for nucleation, which can be used to evaluate the precipitate size and density. Our project aims at using such predictions as guides for the design of alloys with a good RHE, by combining these predictions and the influence of precipitates on RHE reported in the literature.
• Innovative aspects of the CADORS project
The three innovative aspects are (i) the expression of criteria, both for DRX and for RHE, as a function of alloy composition, so that (ii) it becomes possible to integrate them (along with already existing predictive tools) in a multi-objective alloy optimisation method, within (iii) a strategy to design new alloys for actual applications, with potential gains in industrial competitiveness.
To go back to the main page, click here!
* Contact:
Franck Tancret
Université de Nantes
Polytech Nantes
Institut des Matériaux de Nantes – Jean Rouxel (IMN)
BP 50609
44306 Nantes Cedex 3
France
+33 (0)2 40 68 31 97
franck.tancret@univ-nantes.fr
www.univ-nantes.fr