Dear Bernd,
Thanks for the reply. I have increased the mobility for 1000 times and it is still on the run. BTW kindly find my last dri file as attached. Thanks for your attention and efforts.
Regards,
PF simulation of Eutectic solidification
Re: PF simulation of Eutectic solidification
- Attachments
-
- MoSIb_Adjusted_dri.txt
- (39.87 KiB) Downloaded 364 times
Re: PF simulation of Eutectic solidification
Dear omid,
I checked your driving file and found the following "obvious" errors (I mean obvious for someone experienced with MICRESS
):
1.) In your linearized phase diagram description, all slopes are negative. To my understanding, this is not consistent with a cooperative ternary eutectic growth.
I would expect the following signs for the slopes of component 1 and 2: (0/1):+/-, (0/2): -/+, (0/3): -/-.
Inconsistent definition of the phase diagram typically leads to completely unexpected behaviour
2.) In your phase diagram description, phase 3 has composition 0 for both elements. This is only consistent with a completely stoichiometric phase. Probably you mixed up phase 2 and 3 in the definition of stoichiometric components...
At this stage, I would recomment to define all three solid phases as completely (for all elements) stoichiometric. The solidus slopes then can be defined arbitrarily (they are not used). I would suggest to put "-999.99" as a placeholder, because this string is also used as placeholder in the output of linearisation data (.log .TabLin).
3.) Your initial temperature is 1000K, which is more than 1000K below the eutectic temperature
This will lead to extremely high driving forces which destroy all your interfaces if the mobility is not very small or you cut them down by use of dGMax.
But for a reasonable behaviour which respects curvature (and thus microstructure), driving forces must be smaller than the curvature undercooling, i.e. typically 0.1-1 K. Thus, you should start with an initial bottom temperature which is equal to the eutectic temperature, minus the temperature drop between the expected growth front and the bottom boundary according to the temperature gradient .
4.) You use a fixed concentration boundary condition at the top of the 1D-extension which is correct. But the concentration value there should be the eutectic composition. Otherwise, you will never get steady-state ternary eutectic growth...
5.) The interface thickness must be at least 2.5 cells (better 3 cells)
Two more remarks:
- if you use automatic time stepping without limits, then it can happen that the automatic time step is very small and the simulation freezes. To check whether this is the case, reduce the tab_log output frequency to a very small interval and check the .TabT output.
- You have chosen a maximum number of nuclei of 25 in the nucleation input. This is OK for testing. The risk is, however, that you forget it afterwards when you increase the domain size. The problem is that it may be difficult to diagnose later that a strange nucleation behaviour is due to exceeding this maximum number. My recommendation: set a big value like 1E6 already now to avoid that risk.
Bernd
I checked your driving file and found the following "obvious" errors (I mean obvious for someone experienced with MICRESS
):1.) In your linearized phase diagram description, all slopes are negative. To my understanding, this is not consistent with a cooperative ternary eutectic growth.
I would expect the following signs for the slopes of component 1 and 2: (0/1):+/-, (0/2): -/+, (0/3): -/-.
Inconsistent definition of the phase diagram typically leads to completely unexpected behaviour
2.) In your phase diagram description, phase 3 has composition 0 for both elements. This is only consistent with a completely stoichiometric phase. Probably you mixed up phase 2 and 3 in the definition of stoichiometric components...
At this stage, I would recomment to define all three solid phases as completely (for all elements) stoichiometric. The solidus slopes then can be defined arbitrarily (they are not used). I would suggest to put "-999.99" as a placeholder, because this string is also used as placeholder in the output of linearisation data (.log .TabLin).
3.) Your initial temperature is 1000K, which is more than 1000K below the eutectic temperature
This will lead to extremely high driving forces which destroy all your interfaces if the mobility is not very small or you cut them down by use of dGMax. But for a reasonable behaviour which respects curvature (and thus microstructure), driving forces must be smaller than the curvature undercooling, i.e. typically 0.1-1 K. Thus, you should start with an initial bottom temperature which is equal to the eutectic temperature, minus the temperature drop between the expected growth front and the bottom boundary according to the temperature gradient .
4.) You use a fixed concentration boundary condition at the top of the 1D-extension which is correct. But the concentration value there should be the eutectic composition. Otherwise, you will never get steady-state ternary eutectic growth...
5.) The interface thickness must be at least 2.5 cells (better 3 cells)
Two more remarks:
- if you use automatic time stepping without limits, then it can happen that the automatic time step is very small and the simulation freezes. To check whether this is the case, reduce the tab_log output frequency to a very small interval and check the .TabT output.
- You have chosen a maximum number of nuclei of 25 in the nucleation input. This is OK for testing. The risk is, however, that you forget it afterwards when you increase the domain size. The problem is that it may be difficult to diagnose later that a strange nucleation behaviour is due to exceeding this maximum number. My recommendation: set a big value like 1E6 already now to avoid that risk.
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
Thanks very very much for your constructive guides. I have implemented your advises and now it looks much better. Just for a double check kindly have a look on the way in which I defined the phase diagram:
# Input of the phase diagram of phase 0 and phase 1:
# --------------------------------------------------
# Which phase diagram is to be used?
# Options: linear linearTQ
linear
# Temperature of reference point? [K]
2260
# Entropy of fusion between phase LIQUID and 1 ? [J/(cm**3 K)]
0.9
# Input of the concentrations at reference points
# Reference point 1: Concentration of component 1 in phase 0 ? [at%]
17
# Reference point 2: Concentration of component 1 in phase 1 ? [at%]
25
# Reference point 1: Concentration of component 2 in phase 0 ? [at%]
7.7
# Reference point 2: Concentration of component 2 in phase 1 ? [at%]
0
# Input of the slopes at reference points
# Slope m = dT/dC at reference point 1, component 1 ? [K/at%]
80
# Slope m = dT/dC at reference point 2, component 1 ? [K/at%]
120
# Slope m = dT/dC at reference point 1, component 2 ? [K/at%]
-80
# Slope m = dT/dC at reference point 2, component 2 ? [K/at%]
-999.9
# Input of the phase diagram of phase 0 and phase 2:
# --------------------------------------------------
# Which phase diagram is to be used?
# Options: linear linearTQ
linear
# Temperature of reference point? [K]
2260
# Entropy of fusion between phase LIQUID and 2 ? [J/(cm**3 K)]
0.9
# Input of the concentrations at reference points
# Reference point 1: Concentration of component 1 in phase 0 ? [at%]
17
# Reference point 2: Concentration of component 1 in phase 2 ? [at%]
12.5
# Reference point 1: Concentration of component 2 in phase 0 ? [at%]
7.7
# Reference point 2: Concentration of component 2 in phase 2 ? [at%]
25
# Input of the slopes at reference points
# Slope m = dT/dC at reference point 1, component 1 ? [K/at%]
-80
# Slope m = dT/dC at reference point 2, component 1 ? [K/at%]
-999.9
# Slope m = dT/dC at reference point 1, component 2 ? [K/at%]
+180
# Slope m = dT/dC at reference point 2, component 2 ? [K/at%]
-999.9
# Input of the phase diagram of phase 0 and phase 3:
# --------------------------------------------------
# Which phase diagram is to be used?
# Options: linear linearTQ
linear
# Temperature of reference point? [K]
2260
# Entropy of fusion between phase LIQUID and 3 ? [J/(cm**3 K)]
0.9
# Input of the concentrations at reference points
# Reference point 1: Concentration of component 1 in phase 0 ? [at%]
17
# Reference point 2: Concentration of component 1 in phase 3 ? [at%]
17
# Reference point 1: Concentration of component 2 in phase 0 ? [at%]
7.7
# Reference point 2: Concentration of component 2 in phase 3 ? [at%]
7.7
# Input of the slopes at reference points
# Slope m = dT/dC at reference point 1, component 1 ? [K/at%]
-80
# Slope m = dT/dC at reference point 2, component 1 ? [K/at%]
-999.9
# Slope m = dT/dC at reference point 1, component 2 ? [K/at%]
-120
# Slope m = dT/dC at reference point 2, component 2 ? [K/at%]
-999.9
Did I understand you completely? (I'm pretty sure that I didn't )
Moreover, with pass of the time I see one of the nuclei which belongs to solid phase 1 dissolves! I guess this belongs to a probable tiny mistake in the phase diagram definition
Thanks very very much for your constructive guides. I have implemented your advises and now it looks much better. Just for a double check kindly have a look on the way in which I defined the phase diagram:
# Input of the phase diagram of phase 0 and phase 1:
# --------------------------------------------------
# Which phase diagram is to be used?
# Options: linear linearTQ
linear
# Temperature of reference point? [K]
2260
# Entropy of fusion between phase LIQUID and 1 ? [J/(cm**3 K)]
0.9
# Input of the concentrations at reference points
# Reference point 1: Concentration of component 1 in phase 0 ? [at%]
17
# Reference point 2: Concentration of component 1 in phase 1 ? [at%]
25
# Reference point 1: Concentration of component 2 in phase 0 ? [at%]
7.7
# Reference point 2: Concentration of component 2 in phase 1 ? [at%]
0
# Input of the slopes at reference points
# Slope m = dT/dC at reference point 1, component 1 ? [K/at%]
80
# Slope m = dT/dC at reference point 2, component 1 ? [K/at%]
120
# Slope m = dT/dC at reference point 1, component 2 ? [K/at%]
-80
# Slope m = dT/dC at reference point 2, component 2 ? [K/at%]
-999.9
# Input of the phase diagram of phase 0 and phase 2:
# --------------------------------------------------
# Which phase diagram is to be used?
# Options: linear linearTQ
linear
# Temperature of reference point? [K]
2260
# Entropy of fusion between phase LIQUID and 2 ? [J/(cm**3 K)]
0.9
# Input of the concentrations at reference points
# Reference point 1: Concentration of component 1 in phase 0 ? [at%]
17
# Reference point 2: Concentration of component 1 in phase 2 ? [at%]
12.5
# Reference point 1: Concentration of component 2 in phase 0 ? [at%]
7.7
# Reference point 2: Concentration of component 2 in phase 2 ? [at%]
25
# Input of the slopes at reference points
# Slope m = dT/dC at reference point 1, component 1 ? [K/at%]
-80
# Slope m = dT/dC at reference point 2, component 1 ? [K/at%]
-999.9
# Slope m = dT/dC at reference point 1, component 2 ? [K/at%]
+180
# Slope m = dT/dC at reference point 2, component 2 ? [K/at%]
-999.9
# Input of the phase diagram of phase 0 and phase 3:
# --------------------------------------------------
# Which phase diagram is to be used?
# Options: linear linearTQ
linear
# Temperature of reference point? [K]
2260
# Entropy of fusion between phase LIQUID and 3 ? [J/(cm**3 K)]
0.9
# Input of the concentrations at reference points
# Reference point 1: Concentration of component 1 in phase 0 ? [at%]
17
# Reference point 2: Concentration of component 1 in phase 3 ? [at%]
17
# Reference point 1: Concentration of component 2 in phase 0 ? [at%]
7.7
# Reference point 2: Concentration of component 2 in phase 3 ? [at%]
7.7
# Input of the slopes at reference points
# Slope m = dT/dC at reference point 1, component 1 ? [K/at%]
-80
# Slope m = dT/dC at reference point 2, component 1 ? [K/at%]
-999.9
# Slope m = dT/dC at reference point 1, component 2 ? [K/at%]
-120
# Slope m = dT/dC at reference point 2, component 2 ? [K/at%]
-999.9
Did I understand you completely? (I'm pretty sure that I didn't
)Moreover, with pass of the time I see one of the nuclei which belongs to solid phase 1 dissolves! I guess this belongs to a probable tiny mistake in the phase diagram definition
Re: PF simulation of Eutectic solidification
Looks much better! The only doubt I have is whether there really exists a ternary eutectic system which melts at 2260 K - but as long as you use linearized thermodynamic data, temperatur is just a number, and you even can use it in units of Fahrenheit if you correct the slope and dS values correspondingly...
With respect to the exact temperature which is necessary for growth, it is not only the position of the nucleus (distance from bottom), but also curvature which matters. That means you need a certain undercooling to overcome curvature, especially for nucleation. This undercooling scales with the interfacial energy values you have chosen. A value of 1.E-4 may be realistic for intermetallic phases with facetted morphology, but is perhaps too big for a metallic phase like Mo (ss).
Bernd
- but as long as you use linearized thermodynamic data, temperatur is just a number, and you even can use it in units of Fahrenheit if you correct the slope and dS values correspondingly...With respect to the exact temperature which is necessary for growth, it is not only the position of the nucleus (distance from bottom), but also curvature which matters. That means you need a certain undercooling to overcome curvature, especially for nucleation. This undercooling scales with the interfacial energy values you have chosen. A value of 1.E-4 may be realistic for intermetallic phases with facetted morphology, but is perhaps too big for a metallic phase like Mo (ss).
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
Firstly I have to thank you for your constant and comprehensive supports!
Actually the behavior of my system is totally strange! And a friend of mine and his colleges are working on this system experimentally, so it looks to be the same temperature! But as you said this is just a number and is of minor of importance
btw, I do not see the chess pattern lamellae boundary and it looks to be at least rational set up. I just have some minor problems; How can I adjust the minimum temperature of the system? when I run my model I see that the nuclei grow till some levels and then stop. I have checked the Tabl and -tabf file and found that the temperature reaches to -200 K!
And I guess that's why my systems crashes and does not go ahead! It may be possible to define a connecting point in the temperature gradient, Yes?
---Another problem is that I still do not have the new nucleation
It's getting a vital psychologically bothering problem!
---For continuing my research, I guess the next step for me is to play with the numerical parameters until I can generate a photo which is fully compatible with the experimental result, so then I can claim that I have simulated the procedure correctly. Am I right?
Firstly I have to thank you for your constant and comprehensive supports!
Actually the behavior of my system is totally strange! And a friend of mine and his colleges are working on this system experimentally, so it looks to be the same temperature! But as you said this is just a number and is of minor of importance
btw, I do not see the chess pattern lamellae boundary and it looks to be at least rational set up. I just have some minor problems; How can I adjust the minimum temperature of the system? when I run my model I see that the nuclei grow till some levels and then stop. I have checked the Tabl and -tabf file and found that the temperature reaches to -200 K!
And I guess that's why my systems crashes and does not go ahead! It may be possible to define a connecting point in the temperature gradient, Yes?---Another problem is that I still do not have the new nucleation
It's getting a vital psychologically bothering problem!---For continuing my research, I guess the next step for me is to play with the numerical parameters until I can generate a photo which is fully compatible with the experimental result, so then I can claim that I have simulated the procedure correctly. Am I right?
Re: PF simulation of Eutectic solidification
Dear omid,
Good to hear that things are going to the right direction. I see it same way as you, that for continuing your research you need to improve numerical and physical parameters until you reach a result with is as close as possible to the experiments. But there are too many parameters and combinations of parameters to do it by trial and error.
Therefore I propose to advance step by step. I guess that the next step is to adjust the correct length and time scales. The length scale of the problem, i.e. the lamellar distance, is defined by the material parameters (diffusion coefficient of the melt, interface energy, liquidus slopes and tie lines) and the process parameters (cooling rate). The numerical parameters (mainly grid resolution and interface mobility) must be chosen such that they fit this length scale. If not, unstable behaviour is the consequence, whichever other numerical parameters are chosen.
Let's take the diffusion coefficients and interface energy as given (they have reasonable values). Then, from the side of the material parameters, only the liquidus slopes remain to be a source of uncertainty. They must be given in K/at% and seem to be a bit high. E.g. the liquidus slope of component 2, phases 0/3: If you increase the composition of component 2 in the liquid from 7.7 to 8.7 at%, the melting temperature drops by 120K. At the same time, this element is segregating strongly because it has no solubility in phase 3. If I look at the liquidus projection diagram of Mo-Si-B with the isotherms, I have the impression there is only one steep side (the Moss, and even there the value of the slope is not much higher than 10 K/at%...
With respect to the process parameters, you define a cooling rate of 100K/s and a gradient of 10000K/cm. This means, the front should grow at a velocity of 100 µm/s. I guess these parameters stem from an experiment, so they cannot be changed. The question is which lamellar distance one would expect and whether resolution is chosen high enough.
For numerical parameters, I cannot tell whether the resolution is suitable or not. A reasonable procedure is to take it from experiments, and to adjust interface energies if necessary (as they are typically unknown) to reach correct lamellar spacing in simulation. The interface mobility can be estimated from the equation
v = µ ΔG = µ ΔS ΔT
ΔT should be small compared to curvature undercooling (~1K), therfore µ should be at least 0.01 cm4/Js so that the phases can grow with the expected velocity at an undercooling of about 1K. In your case (µ=1.E-7) an undercooling of 10000K is necessary (if you would not cut at dGMax=100 J/cm3 which corresponds to about 110 K). If you check the .driv output, I am sure you see a value of -100 which means that you are in the cutting regime).
I hope that this discussion helps you to reasonably adjust the length and time scale. I assumed that the parameters you use are still those of your last driving file you posted (apart from my suggested changes). It would be nice if you could show again your input file after the latest adjustments.
Bernd
Good to hear that things are going to the right direction. I see it same way as you, that for continuing your research you need to improve numerical and physical parameters until you reach a result with is as close as possible to the experiments. But there are too many parameters and combinations of parameters to do it by trial and error.
Therefore I propose to advance step by step. I guess that the next step is to adjust the correct length and time scales. The length scale of the problem, i.e. the lamellar distance, is defined by the material parameters (diffusion coefficient of the melt, interface energy, liquidus slopes and tie lines) and the process parameters (cooling rate). The numerical parameters (mainly grid resolution and interface mobility) must be chosen such that they fit this length scale. If not, unstable behaviour is the consequence, whichever other numerical parameters are chosen.
Let's take the diffusion coefficients and interface energy as given (they have reasonable values). Then, from the side of the material parameters, only the liquidus slopes remain to be a source of uncertainty. They must be given in K/at% and seem to be a bit high. E.g. the liquidus slope of component 2, phases 0/3: If you increase the composition of component 2 in the liquid from 7.7 to 8.7 at%, the melting temperature drops by 120K. At the same time, this element is segregating strongly because it has no solubility in phase 3. If I look at the liquidus projection diagram of Mo-Si-B with the isotherms, I have the impression there is only one steep side (the Moss, and even there the value of the slope is not much higher than 10 K/at%...
With respect to the process parameters, you define a cooling rate of 100K/s and a gradient of 10000K/cm. This means, the front should grow at a velocity of 100 µm/s. I guess these parameters stem from an experiment, so they cannot be changed. The question is which lamellar distance one would expect and whether resolution is chosen high enough.
For numerical parameters, I cannot tell whether the resolution is suitable or not. A reasonable procedure is to take it from experiments, and to adjust interface energies if necessary (as they are typically unknown) to reach correct lamellar spacing in simulation. The interface mobility can be estimated from the equation
v = µ ΔG = µ ΔS ΔT
ΔT should be small compared to curvature undercooling (~1K), therfore µ should be at least 0.01 cm4/Js so that the phases can grow with the expected velocity at an undercooling of about 1K. In your case (µ=1.E-7) an undercooling of 10000K is necessary (if you would not cut at dGMax=100 J/cm3
which corresponds to about 110 K). If you check the .driv output, I am sure you see a value of -100 which means that you are in the cutting regime).I hope that this discussion helps you to reasonably adjust the length and time scale. I assumed that the parameters you use are still those of your last driving file you posted (apart from my suggested changes). It would be nice if you could show again your input file after the latest adjustments.
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
Vielen dank für Ihre Antwort
It seems that I already have a long way to generate the completely reasonable results. But with your advantageous guides I hope to overcome it finally
Regarding to lamellar distance, is there any output which shows the numerical value of the distance? Moreover, I guess I didn't enable a suitable output to see the .driv output. I already just see _driv.mcr which does not return any numerical values!
Regarding to the liquidus slope, I should say that you are totally right! I have talked to my colleague who is working on this system, and he said that he saw this phenomena so many times! The problem is that this system is a little bit unknown and the data are not verified completely! I have already asked him to generate a valid database or at least a phase diagram, but I have to wait for the response since they are too busy nowadays! For now I prefer to stick to the ternary eutectic point in order to get rid of liquidus slope's barrier! Do you think that until I am dealing with the ternary eutectic point completely, it will cause such a problem for me? Does is matter anymore to have not access to the right tie lines and slopes?
--The solubility of the components inside the phase 3 was defined in a silly way by me I have already corrected that!
With respect to the process parameters, both the cooling rate and gradient have imaginary values! My colleague has never measured them, I just wrote a number there without any sense!
Btw, I have attached my code through this post, Kindly help me to:
Get the driv output which shows me those numerical values
Extract the lamellar spacing, Which is one of my research's scope. I am hopefully going to investigate i.e the effect of cooling rate on the lamellar spacing
Make an educated guess about the cooling rate and temperature gradient
make your comments about my concern in regard to the liquidus slope
Regards
Vielen dank für Ihre Antwort
It seems that I already have a long way to generate the completely reasonable results. But with your advantageous guides I hope to overcome it finally
Regarding to lamellar distance, is there any output which shows the numerical value of the distance? Moreover, I guess I didn't enable a suitable output to see the .driv output. I already just see _driv.mcr which does not return any numerical values!
Regarding to the liquidus slope, I should say that you are totally right! I have talked to my colleague who is working on this system, and he said that he saw this phenomena so many times! The problem is that this system is a little bit unknown and the data are not verified completely! I have already asked him to generate a valid database or at least a phase diagram, but I have to wait for the response since they are too busy nowadays! For now I prefer to stick to the ternary eutectic point in order to get rid of liquidus slope's barrier! Do you think that until I am dealing with the ternary eutectic point completely, it will cause such a problem for me? Does is matter anymore to have not access to the right tie lines and slopes?
--The solubility of the components inside the phase 3 was defined in a silly way by me
I have already corrected that!With respect to the process parameters, both the cooling rate and gradient have imaginary values! My colleague has never measured them, I just wrote a number there without any sense!
Btw, I have attached my code through this post, Kindly help me to:
Get the driv output which shows me those numerical values
Extract the lamellar spacing, Which is one of my research's scope. I am hopefully going to investigate i.e the effect of cooling rate on the lamellar spacing
Make an educated guess about the cooling rate and temperature gradient
make your comments about my concern in regard to the liquidus slope
Regards
- Attachments
-
- MoSIb_Adjusted_dri.txt
- (39.96 KiB) Downloaded 334 times
Re: PF simulation of Eutectic solidification
Dear Omid,
You say that you don't know anything about the real liquidus slopes. Then just stick to your assumptions and keep in mind that rescaling them would rescale the thermodynamic driving force (solutal undercooling) for a given concentration deviation and thus kinetics of growth.
You already have written the .driv (=_driv.mcr under Windows) output. It is a graphical output, so you should open it with DP_MICRESS
The cooling rate and temperature gradient resemble a fast quenching process. As I do not know anything about the process, I cannot tell. But what I explained is that if you stick to these parameters, you have to increase the interface mobility values by 5 orders of magnitude! If you prefer you can also decrease the cooling rate by the same factor, but I would keep the thermal parameters until you have some realistic ones, and increase the interface mobilities.
The lamellar spacing can be extracted once the simulation is running correctly, and steady state growth of lamellae can be observed. Then you just count the lamellae for calculating the average distance. Please note that the correct ternary eutectic growth morphology cannot be simulated in 2D. At some later point you will have to switch to 3D simulation.
Bernd
You say that you don't know anything about the real liquidus slopes. Then just stick to your assumptions and keep in mind that rescaling them would rescale the thermodynamic driving force (solutal undercooling) for a given concentration deviation and thus kinetics of growth.
You already have written the .driv (=_driv.mcr under Windows) output. It is a graphical output, so you should open it with DP_MICRESS
The cooling rate and temperature gradient resemble a fast quenching process. As I do not know anything about the process, I cannot tell. But what I explained is that if you stick to these parameters, you have to increase the interface mobility values by 5 orders of magnitude! If you prefer you can also decrease the cooling rate by the same factor, but I would keep the thermal parameters until you have some realistic ones, and increase the interface mobilities.
The lamellar spacing can be extracted once the simulation is running correctly, and steady state growth of lamellae can be observed. Then you just count the lamellae for calculating the average distance. Please note that the correct ternary eutectic growth morphology cannot be simulated in 2D. At some later point you will have to switch to 3D simulation.
Bernd
Re: PF simulation of Eutectic solidification
Dear Bernd,
Thanks for the reply.
Regarding the driv file, I thought that there may be an output with this format which I do not know it
I have already implemented your advises. One of the remaining problems is how can I determine the minimum temperature? For example, due to the experimental procedure, which is laser melting I have to define a really fast cooling rate, which results in a high amount of driving force, and obviously since I have limited the driving force to max 100, I will not follow the experimental procedure, for example, in the experiments the domain in 2.5 second gets totally solid, but in my simulation after 4 second I have only % 25 solid and the rest are liquid! However the domain's temperature reaches to -200 K during process running! How can I adjust the temperature to not come below i.e 50 K regardless the other parameters?
Regards,
Thanks for the reply.
Regarding the driv file, I thought that there may be an output with this format which I do not know it
I have already implemented your advises. One of the remaining problems is how can I determine the minimum temperature? For example, due to the experimental procedure, which is laser melting I have to define a really fast cooling rate, which results in a high amount of driving force, and obviously since I have limited the driving force to max 100, I will not follow the experimental procedure, for example, in the experiments the domain in 2.5 second gets totally solid, but in my simulation after 4 second I have only % 25 solid and the rest are liquid! However the domain's temperature reaches to -200 K during process running! How can I adjust the temperature to not come below i.e 50 K regardless the other parameters?Regards,
Re: PF simulation of Eutectic solidification
Dear Omid,
you are using a "moving_frame" temperature boundary condition, i.e. the simulation domain will follow the growing eutectic front along the temperature gradient. As a consequence, even if you define a cooling rate, the temperature should remain at a stationary value which should be close to the eutectic temperature. But the front can only grow fast enough if you define a sufficiently high interface mobility, and if all 3 phases grow simultaneously.
I just noticed, that in the phase diagram data for phases 0/3 you defined an identical value for the solidus and liquidus composition: This means that there is no segregation at all, and that this phase will grow much faster than all other phases. Please correct that and put the same value as initial concentration!
I also noticed that you use a far-field composition as fixed boundary condition which is far from the eutectic composition. This will not work, because with this simulation you want to focus on the ternary eutectic growth and not on growth of a primary phase. Please put the ternary eutectic composition as fixed boundary condition.
And in all cases where you defined the solidus slope as -999.99, you must define the corresponding element of the phase as stoichiometric:
1 2
2 1 2
3 1 2
no_more_stoichio
Otherwise you get error messages...
Bernd
you are using a "moving_frame" temperature boundary condition, i.e. the simulation domain will follow the growing eutectic front along the temperature gradient. As a consequence, even if you define a cooling rate, the temperature should remain at a stationary value which should be close to the eutectic temperature. But the front can only grow fast enough if you define a sufficiently high interface mobility, and if all 3 phases grow simultaneously.
I just noticed, that in the phase diagram data for phases 0/3 you defined an identical value for the solidus and liquidus composition: This means that there is no segregation at all, and that this phase will grow much faster than all other phases. Please correct that and put the same value as initial concentration!
I also noticed that you use a far-field composition as fixed boundary condition which is far from the eutectic composition. This will not work, because with this simulation you want to focus on the ternary eutectic growth and not on growth of a primary phase. Please put the ternary eutectic composition as fixed boundary condition.
And in all cases where you defined the solidus slope as -999.99, you must define the corresponding element of the phase as stoichiometric:
1 2
2 1 2
3 1 2
no_more_stoichio
Otherwise you get error messages...
Bernd