NPLE
NPLE
Hi Bernd,
I found MICRESS has a model for NPLE. But there is no document detailing it. The only literature is "J Rudnizki et al. phase-field modeling of austenite formation from a ferrite plus pearlite microstructure during annealing of cold-rolled dual-phase steel. Metal Mater Trans. 2011".
Could you please give some details.
I have the general idea of NPLE. But for a limited mobility, how is the driving force calculated and how does partitioning of elements occurs between phases, e.g. in austenite-to-ferrite transformation in Fe-C-Mn alloy?
Thank you.
Ben
I found MICRESS has a model for NPLE. But there is no document detailing it. The only literature is "J Rudnizki et al. phase-field modeling of austenite formation from a ferrite plus pearlite microstructure during annealing of cold-rolled dual-phase steel. Metal Mater Trans. 2011".
Could you please give some details.
I have the general idea of NPLE. But for a limited mobility, how is the driving force calculated and how does partitioning of elements occurs between phases, e.g. in austenite-to-ferrite transformation in Fe-C-Mn alloy?
Thank you.
Ben
Re: NPLE
Hi Ben,
the specific characteristics of NPLE (no-partitioning local equilibrium) in the context of MICRESS are:
- there are slow-diffusing elements and fast-diffusing elements
- the driving force is so high that the slow-diffusing elements are not limiting the phase-transformation (their pile-up due to local equilibrium is "overrun"). With lower driving force, the transformation would effectively stop ( i.e. drop by several orders of magnitude, and also global partitioning of the slow elements would be observed)
- the fast-diffusing elements are still limiting the transformation speed (otherwise, the transformation would be "massive")
- the pile-up of the slow elements is compensating part of the driving force and thus reducing the transformation speed (otherwise it would be para-equilibrium)
For MICRESS, the problem consists in the different length-scales involved with the slow and fast-diffusing elements. As only the fast elements are limiting the transformation speed, those mainly contribute to microstructure formation. Therefore, typically, microstructure simulation will be done on the scale of their diffusion length.
This means that in a typical MICRESS simulation the pile-up of the slow elements cannot be resolved. Without the special NPLE model, although local (quasi-) equilibrium is fulfilled, the driving force is not evaluated correctly - only the innermost grid cell of the interface with a very small fraction of the vanishing phase would feel the correct compostion and thus driving-force.
The strategy which is followed in MICRESS is:
- Redistribution inside each grid cell is modified such that all over the interface the fraction of the vanishing phase is set to phMin. To do so, the moving direction is needed.
- The information on the moving direction of the interface is taken from the time-step before
- If the front did not move in the time-step before (because the driving force was too low), the (potential) moving direction is changed
This is all! You can observe the different mode of redistribution by checking the corresponding phase composition outputs (*.cXXphaY).
The redistribution behaviour (nple, para, normal) can be specified for each element independently and should be set to "nple" for all slow-diffusing elements and to "normal" for the fast elements.
Best wishes
Bernd
the specific characteristics of NPLE (no-partitioning local equilibrium) in the context of MICRESS are:
- there are slow-diffusing elements and fast-diffusing elements
- the driving force is so high that the slow-diffusing elements are not limiting the phase-transformation (their pile-up due to local equilibrium is "overrun"). With lower driving force, the transformation would effectively stop ( i.e. drop by several orders of magnitude, and also global partitioning of the slow elements would be observed)
- the fast-diffusing elements are still limiting the transformation speed (otherwise, the transformation would be "massive")
- the pile-up of the slow elements is compensating part of the driving force and thus reducing the transformation speed (otherwise it would be para-equilibrium)
For MICRESS, the problem consists in the different length-scales involved with the slow and fast-diffusing elements. As only the fast elements are limiting the transformation speed, those mainly contribute to microstructure formation. Therefore, typically, microstructure simulation will be done on the scale of their diffusion length.
This means that in a typical MICRESS simulation the pile-up of the slow elements cannot be resolved. Without the special NPLE model, although local (quasi-) equilibrium is fulfilled, the driving force is not evaluated correctly - only the innermost grid cell of the interface with a very small fraction of the vanishing phase would feel the correct compostion and thus driving-force.
The strategy which is followed in MICRESS is:
- Redistribution inside each grid cell is modified such that all over the interface the fraction of the vanishing phase is set to phMin. To do so, the moving direction is needed.
- The information on the moving direction of the interface is taken from the time-step before
- If the front did not move in the time-step before (because the driving force was too low), the (potential) moving direction is changed
This is all! You can observe the different mode of redistribution by checking the corresponding phase composition outputs (*.cXXphaY).
The redistribution behaviour (nple, para, normal) can be specified for each element independently and should be set to "nple" for all slow-diffusing elements and to "normal" for the fast elements.
Best wishes
Bernd
Re: NPLE
Hi, Bernd.
Thank you for quick reply.
Here is my understanding:
1)to simulate nple conditions e.g. in Fe-C-Mn alloy, chose C as "diff" and "normal" while Mn as "no_diff" and "nple"; the thermodynamic data e.g. del_S and slopes, are the same as ortho equilibrium condition.
2) the Mn concentration in new phase (e.g. ferrite) is the same as the average concentration of the material, while the Mn concentration in vanishing phase (e.g. austenite) is the product of the partitioning coefficient and Mn in new phase.
Also, this Mn concentration in vanishing phase is the same within the interface thickness, i.e. the spike peak is extended to the whole interface.
Thank you for quick reply.
Here is my understanding:
1)to simulate nple conditions e.g. in Fe-C-Mn alloy, chose C as "diff" and "normal" while Mn as "no_diff" and "nple"; the thermodynamic data e.g. del_S and slopes, are the same as ortho equilibrium condition.
2) the Mn concentration in new phase (e.g. ferrite) is the same as the average concentration of the material, while the Mn concentration in vanishing phase (e.g. austenite) is the product of the partitioning coefficient and Mn in new phase.
Also, this Mn concentration in vanishing phase is the same within the interface thickness, i.e. the spike peak is extended to the whole interface.
Re: NPLE
That is right!
It it not absolutely required to set the diffusion of Mn to "no_diff"! Nevertheless, as the diffusion coefficient of Mn is typically much smaller than that of C, including diffusion of Mn will not have much of an effect anyway!
Bernd
It it not absolutely required to set the diffusion of Mn to "no_diff"! Nevertheless, as the diffusion coefficient of Mn is typically much smaller than that of C, including diffusion of Mn will not have much of an effect anyway!
Bernd
Re: NPLE
Hi Bernd,
I have one question that with an extremely high mobility, the solute concentrations will approach to the equilibrium which are C1 and C2 in the attached figure. But the carbon concentration near the interface edge should be C3 which is a little bit different from C2 in terms of the Carbon concentration.
By the way, I cannot find the example file ‘Gamma_Alpha_NPLE_in.txt" on the website.
I have one question that with an extremely high mobility, the solute concentrations will approach to the equilibrium which are C1 and C2 in the attached figure. But the carbon concentration near the interface edge should be C3 which is a little bit different from C2 in terms of the Carbon concentration.
By the way, I cannot find the example file ‘Gamma_Alpha_NPLE_in.txt" on the website.
- Attachments
-
- Untitled.png (18.68 KiB) Viewed 9599 times
Re: NPLE
Hi Ben,
I do not understand what is the meaning of C3 and how you construct it - why should it have another Carbon composition than C2?
The examples on the webpage have not been uddated yet. But the actual examples are on the installation CD - I can send it to you if you wish!
Bernd
I do not understand what is the meaning of C3 and how you construct it - why should it have another Carbon composition than C2?
The examples on the webpage have not been uddated yet. But the actual examples are on the installation CD - I can send it to you if you wish!
Bernd
Re: NPLE
Hi Bernd,
It would be quite appreciated if you could send me the example (zhubenqiang@gmail.com).
The line between C2 and C3 is isoactivity line for carbon.
You can see the schematic of LENP below.
It would be quite appreciated if you could send me the example (zhubenqiang@gmail.com).
The line between C2 and C3 is isoactivity line for carbon.
You can see the schematic of LENP below.
- Attachments
-
- 1-s2.0-S1359645408000402-gr1.jpg (33.62 KiB) Viewed 9595 times
Re: NPLE
Hi Ben,
I do not see why the composition at the interface should be different from C2! NPLE implies local equilibrium which is defined by the tie-line!
Madeleine Durand-Charre shows this in her book "Microstructure of Steel and Cast Irons" on page 188:
Or do I understand something wrong?
Bernd
I do not see why the composition at the interface should be different from C2! NPLE implies local equilibrium which is defined by the tie-line!
Madeleine Durand-Charre shows this in her book "Microstructure of Steel and Cast Irons" on page 188:
Or do I understand something wrong?
Bernd