Firstly, an existing thermodynamic model for the predicting of martensite start temperature of commercial steels has been improved to include more elements such as N, Si, V, Mo, Nb, W, Ti, Al, Cu, Co, B, P and S and their corresponding composition ranges for Martensitic transformation. The predicting ability of the existing model is improved considerably by critical assessment of different binary and ternary systems i.e. CALPHAD approach which is by wise selection of experimental data for optimization of the interaction parameters. Understanding the degree of variation in multi-component commercial alloys, various ternary systems such as Fe-Ni-X and Fe-Cr-X are optimized using both binary and ternary interaction parameters. The large variations between calculated and the experimental values are determined and reported for improvements in thermodynamics descriptions.Secondly, model for the prediction of Epsilon martensite start temperature of some commercial steels and shape memory alloys is newly introduced by optimizing Fe-Mn, Fe-Mn-Si and other Fe-Mn-X systems considering the commercial aspects in the recent development of light weight steels alloyed with Al and Si.Thirdly, the effect of prior Austenite grain size (pAGS) on martensite start temperature is introduced into the model in the form of non-chemical contribution which will greatly influence the Gibbs energy barrier for transformation. A serious attempt has been made to describe the dependency of transition between lenticular and thin-plate martensite morphologies on the refinement of prior Austenite grain size.Finally, the model is validated using a data-set of 1500 commercial and novel alloys. Including the newly modified thermodynamic descriptions for the Fe-based TCFE9 database by Thermo-Calc software AB, the model has the efficiency to predict the martensite start temperature of Multi-component alloys with an accuracy of (±) 35 K. The model predictability can be further improved by critical assessment of thermodynamic factors such as stacking faults and magnetism in Fe-Mn-Si-Ni-Cr systems.

Les aciers TWIP se déforment par maclage et par glissement de dislocations, avec pour conséquence de forts taux d’écrouissage. Les mécanismes de déformation sont contrôlés par l’énergie de faute d’empilement (EFE). Un modèle de prévision de l’EFE et une régression de TNéel (transition antiferro/paramagnétique) de l’austénite sont proposés pour les systèmes FeMnXC (X = Cu, Cr, Al, Si et Ti). Les nuances FeMnCuC étudiées ont une EFE plus faible que la nuance de référence Fe22Mn0,6C. La formation de martensite [epsilon]?se substitue au maclage, sans dégradation des caractéristiques mécaniques en traction. La contrainte d'écoulement diminue avec la teneur en carbone et la formation de martensite [alpha]' aux plus basses EFE réduit l'allongement à rupture. La substitution d'une partie du manganèse par du cuivre permet un gain de 20% sur le module d'Young à température ambiante, en abaissant TNéel en dessous de 0ºC. La précipitation intragranulaire de carbures de vanadium augmente la limite d’élasticité mais n’influence pas le taux d’écrouissage. Aucune interaction entre précipités et macles n'a été observée en microscopie. Les calculs de cohérence et les mesures au MET montrent que les carbures ont une relation d'orientation avec l'austénite et sont semi-cohérents avec une faible cohérence résiduelle. Les contraintes induites ne semblent pas suffisantes pour piéger de grandes quantités d'hydrogène. Les alliages FeMnC + TiC présentent un fort durcissement par effet composite en début de déformation, tandis que l'écrouissage par effet TWIP n'est pas modifié par la présence des particules TiC. Cependant, le clivage des précipités primaires de grande taille réduit l'allongement à rupture / TWIP steels deformation occurs by twinning and by dislocations gliding which leads to high a strain hardening. The deformation mechanisms are controlled by the stacking fault energy (SFE). A model for the prediction of the SFE and a law for TNéel (antiferro to paramagnetic transition) for austenite are proposed in FeMnXC systems (X = Cu, Cr, Al, Si et Ti). The studied FeMnCuC grades have a lower SFE than the Fe22Mn0,6C reference. The formation of [epsilon]-martensite replaces twinning without any deterioration of the mechanical properties. The flow stress decreases with the carbon content and the formation of [alpha]'-martensite at the lowest SFEs reduces the elongation to fracture. Substituting a part of the manganese content by copper leads to a 20% increase of the Young's Modulus at room temperature by decreasing TNéel below 0ºC. The precipitation of intragranular vanadium carbide increases the yield stress but does not influence the strain hardening rate. No interaction between precipitates and twins has been observed by microscopy. The coherency calculations and the TEM observations show that the carbides have an orientation relation with the austenite and are semi-coherent with a low residual coherency. The resulting stresses do not seem to be high enough to trap large quantities of hydrogen. The FeMnC + TiC alloys exhibit a strong hardening by composite effect at the beginning of deformation, while the strain hardening due to TWIP effect is not modified by the presence of the TiC particles. Meanwhile, cleavage occurs in the largest primary precipitates, which reduces the elongation to fracture