Initial Concentration

technical aspects of .dri file generation (e.g. debug mode ) etc...
Bernd
Posts: 1504
Joined: Mon Jun 23, 2008 9:29 pm

Re: Initial Concentration

Post by Bernd » Thu Apr 27, 2017 5:14 pm

Dear Rafael,

I suspect that the reason for the concentration problem is the different treatment of concentration units for weight percent (wt%) in the case of linearized and TQ-coupled simulations. In the TQ case, internal treatment and storage of compositions is always in at% as the physically relevant unit. If the user specifies "wt%" as composition unit, all compositions are converted to at% after input and converted back to wt% for output.
In the linearized case, this transformation cannot be done because of missing molar volumes (in fact, the elements even have no name and could be everything...). Moreover, linearized diagrams which are given in w% in literature would lead to different results if converted and back-converted.
Thus, there is a fundamental problem which we want to solve in one of the next versions by giving the user the possibility to input molar volumes in linearized calculations and/or to decide whether the phase-diagram should be used with wt% internally or not.

This leads to the situation, that presently if you do sim1 using database, concentrations in the .rest file (which is an extraction of the internal composition units) are in at%. When you read them in sim2 with linearized phase diagram and wt% specification, they are re-interpreted as wt%. The difference between at% and wt% is big for carbon and nearly 0 for Mn...

The solution for you is to do sim1 and sim2 both either linearized or TQ-coupled, or to perform the linearized simulations in at% (you need a phase diagram description in at%).

If you anyway want to use a linearized phase diagram, you use TQ-coupling only for getting proper diffusion data. Then it may be easier to use a manual input which include the T-dependency in form of an Arrhenius description ("diagonal d"), or to tabulate the diffusion coefficient as function of T and read it from file ("diagonal f"). Both is also possible including off-diagonal terms ("multi dd" or "multi ff").
Your trick to set diffusion of carbon to 0 works in principle if you start from a homogeneous initial structure and do no heat treatment afterwards. However, the physical reason for no partitioning is the fast (virtually instantaneous) phase transformation in reality which does not leave time for diffusion. Thus, if you want to combine diffusion processes of carbon with your martensitic transformation, you need to find another solution...

Bernd

Post Reply