From the little bit I read about this method of predicting the ferrite number in Austenitic Stainless steels, I believe this a more precise method of modelling compared to the WRC-92 method mainly because of the fact that the Bayesian method uses neural networks as opposed to the constitutional diagrams previously found in Schaeffler-1949, DeLong-1973, and the WRC-1992. The Function Fit model (mathematical formula) which uses the free energy between austenite and ferrite, and relates it to the ferrite number... It is as accurate as the WRC-1992 CD but, is limited by composition range. Constitution Diagrams lack generality because elements like Silicon and Titanium are neglected... In the data set, most of the variation in the ferrite number is attributed to Cr, Ni, Mo, and N followed by Ti, V, Cu and Co... This makes metallurgical sense!!!
The Bayesian model is much more precise as it is shown in the chart found in the link I provided below that compares previous methods of predicting the ferrite number in Austenitic SS according to the percentage value of RMS error for each method... The Bayesian method was found to have the lowest amount of RMS error - 1.9 compared to the WRC-1992 which was 5.8, and the Function Fit model @ 5.6 respectively.http://www.msm.cam.ac.uk/phase-trans/2001/ferrite.number.html/index.html
Not being a formal metallurgist or material science major, I would like to read some input, nad opinions from others more familiar with Constitutional Diagrams, and varying Cr & Ni equivalents used in the ones that are mentioned in the presentation linked above to see if this is indeed supported by others in the field of material science/metallurgy...
Foe myself, I understand most of the presentation and yet ,there are certain aspects from the charts that I need to read over in order ot have all of the data make sense in my head :) :) :)
Enjoy the presentation from the folks at the Materials Joining Center of the Indira Gandhi Center for Atomic Research, Kalpakkam, India & The Department of Metallurgy and Material Science @ *Cambridge University, UK... M. Vasudevan, M. Murugananth* and A.K. Bhaduri.