Hello,
I am working on the same project as mehnoush.
We were able to import an appropriate inital structure.Thank you for your help!
But now, after starting the simulation process, we noticed another problem.
We are developing a model for a steel with a (average) carbon content of 0.4 wt-%. It is clear, that the local carbon concentraion differs from the global. Due to the reason, we imported a section/extract of a real structure we noticed an average concentration of 1,2 wt-% within our domain. Now we have two options:
1.) find an appropriate initial strucutre which results in an average concentration of carbon with 0.4% by trial and error
or 2.)to change the boundary conditions. But which of the options is expedient? I think only to change the concentration field from insulation ("i") to periodic wont be sufficient, because the carbon will stay within my domain and is not able to "leave".
Is there a option like "open boundary" to choose for such a case?
What do you think will be the best way to handle this problem?
Thank you and Best Regards,
Raphael
importing the initial microstructure
Re: importing the initial microstructure
Dear Rafael,
What you describe is a general problem of consistency which is connected to the procedure of reading in initial microstructures and which I know very well! To get the correct alloy composition not only depends on correctly measured and representative values of the phase compositions, but also on the phase fractions!
Of course, you could use boundary conditions as you propose, a fixed boundary condition in this case, in order to manipulate the composition in the way you want. However, you would have to do it as an intermediate "correction" simulation step, because you don't want to fix the composition afterwards (when e.g. you have precipitation and the compositions must change). Furthermore, the outcome would be quite arbitrary and artificial (comparable to option 4. below)...
Apart from this, in general, you have several options:
1.) Try to use a sufficiently large and representative region. This is often impossible because you would need a too large area, because your metallographic images are not representative, or because there are systematic measurement errors.
2.) Read the phase distribution only and manually assign (constant) phase compositions. This helps to some extent, but requires at least that the phase fractions are close to representative. Otherwise, phase compositions have to be chosen far from realistic in order to achieve the correct alloy composition. And in case of many different phases, it gets difficult anyway
3.) Phase fractions can be corrected to some extent either during image processing (e.g. choice of suitable threshold values) or in MICRESS by using a suitable number of steps for initialisation of the interface (which always shrinks small particles, see here). With representative phase fractions, method 2) works perfectly.
4.) You can always manually manipulate the data which you read in from file, e.g. by applying correction factors, etc. However, you are fully responsible for yourself what you get and whether it makes sense.
5.) Create a fully synthetic consistent initial microstructure by other means (MICRESS input, Dream3D, ...)
6.) Simulate the initial microstructure. This is a method I often use with complex multiphase systems if the initial microstructure e.g. is a as cast structure which can be simulated with MICRESS. However, this is a lot of extra effort, and cannot be done e.g. for deformed materials.
Bernd
What you describe is a general problem of consistency which is connected to the procedure of reading in initial microstructures and which I know very well! To get the correct alloy composition not only depends on correctly measured and representative values of the phase compositions, but also on the phase fractions!
Of course, you could use boundary conditions as you propose, a fixed boundary condition in this case, in order to manipulate the composition in the way you want. However, you would have to do it as an intermediate "correction" simulation step, because you don't want to fix the composition afterwards (when e.g. you have precipitation and the compositions must change). Furthermore, the outcome would be quite arbitrary and artificial (comparable to option 4. below)...
Apart from this, in general, you have several options:
1.) Try to use a sufficiently large and representative region. This is often impossible because you would need a too large area, because your metallographic images are not representative, or because there are systematic measurement errors.
2.) Read the phase distribution only and manually assign (constant) phase compositions. This helps to some extent, but requires at least that the phase fractions are close to representative. Otherwise, phase compositions have to be chosen far from realistic in order to achieve the correct alloy composition. And in case of many different phases, it gets difficult anyway
3.) Phase fractions can be corrected to some extent either during image processing (e.g. choice of suitable threshold values) or in MICRESS by using a suitable number of steps for initialisation of the interface (which always shrinks small particles, see here). With representative phase fractions, method 2) works perfectly.
4.) You can always manually manipulate the data which you read in from file, e.g. by applying correction factors, etc. However, you are fully responsible for yourself what you get and whether it makes sense.
5.) Create a fully synthetic consistent initial microstructure by other means (MICRESS input, Dream3D, ...)
6.) Simulate the initial microstructure. This is a method I often use with complex multiphase systems if the initial microstructure e.g. is a as cast structure which can be simulated with MICRESS. However, this is a lot of extra effort, and cannot be done e.g. for deformed materials.
Bernd