ENH: Workspace integration

[bocklund]

This branch only has WIP changes for directly using Workspace in simulate_scheil_solidification

Outstanding TODO items:

• Fix broken tests
• Port ordering handler to Workspace
• Port simulate_equilibrium_solidification to Workspace
• Support getting properties in the API via output keyword argument (analogous to PyCalphad equilibrium output)? See Add options to save equilibrium results during scheil simulation #38. Also would want to export the properties to the SolidificationResult object API.

[bocklund]

@Chuangye-Wang @sorkins I think I'm happy with Workspace support here in general.

The latest commit adds notional support to demo the custom property use case. The idea is that you can pass anything that works for PyCalphad Workspace to output

    sol_res = simulate_scheil_solidification(dbf, comps, phases, initial_composition, start_temperature, step_temperature=20.0, verbose=True, output=["HM", "CPM", "NP(LIQUID)"])

These properties give the properties of the state of the properties are for 1 mole of solidifying liquid values at any instant in the simulation (that is, we don't account for the properties of the previously solidified solid at all or the remaining amount of liquid). This is conceptually the same thing that happens in the implementation of #38, i.e. leaving it all up to the user.

I'm a little hesitant to not account for the simulation's sense of how much liquid remains and the properties of the frozen-in solid, since that's what the equilibrium picture would look like if we took the same kind of approach there. I don't like the idea that equilibrium and Scheil solidification simulations give different physical pictures of the properties. But I'm open to the idea of deferring on supporting the equilibrium solidification support for now until we reconcile handling the Scheil case better.

Does this approach serve your use case? I'm curious on how you're addressing handling the properties of the frozen-in solid if that's relevant for your use case. Happy to accept feedback in general.

[sorkins]

Thanks so much for the work here @bocklund! We'll test things out with this branch ASAP and give some feedback.

We'll think on the frozen-in solid question, it's nothing too complex so I'm sure it wouldn't be a problem to share some details.

[Chuangye-Wang]

@bocklund Thanks for this nice implementation! I tried to use this branch to get solidified fraction during solidification, but the results turned out to be different from the branch I made. I got a lot of duplicated values for the fraction_solid in the scheil's results. I tried to debug a bit, and found multiplicity_aware_stable_phases only include liquid phase from the fourth iteration, which leads to the duplication. Could you help to identify the root cause for this issue? I used alloy Fe-0.3Cr-0.02C in mole fraction. Sorry I cannot provide our internal database for you to debug. If you need I provide a test case to debug, I will try to make one for you.

Branch I made.

Current branch.

Debugged at:

[Chuangye-Wang]

@bocklund I created an example for you to test on.
You can see that there are lots of duplicated values in the fraction_solid as temperature decreases.
test_scheil_solidification_ternary_system.zip

[sorkins]

@bocklund reminder

[bocklund]

@bocklund I created an example for you to test on. You can see that there are lots of duplicated values in the fraction_solid as temperature decreases. test_scheil_solidification_ternary_system.zip

Apologies, I've been swamped recently. Thank you for the test, it helps a lot debugging this stuff. I just want to verify that the expected behavior of the test is that it passes (it does for me), but you see that there are duplicated values in the fraction of solid? I am able to reproduce that behavior (test passes, but has duplicated fraction liquid/solid values)

[Chuangye-Wang]

@bocklund Correct. It passes but the results contain a lot of duplicated values.

[sorkins]

@bocklund reminder on this when you have the time. If there's anything you can point us at to help figure this out, please let us know!

[sorkins]

@bocklund reminder, no rush.

[bocklund]

Do you guys want to try this again on this branch with the newly released version of PyCalphad (0.11.1)? Seems to be the fix for me locally:

from pathlib import Path
import matplotlib.pyplot as plt
import numpy as np
from pycalphad import Database, variables as v
from scheil import simulate_scheil_solidification

here = Path(".").resolve()

dbf = Database(here / "CCrFe-2014Khvan.tdb")
comps = ["FE", "CR", "C", "VA"]
phases = sorted(dbf.phases.keys())

initial_composition = {v.X("CR"): 0.3, v.X("C"): 0.02}
start_temperature = 1750

sol_res = simulate_scheil_solidification(dbf, comps, phases, initial_composition, start_temperature, step_temperature=1, verbose=False)

fig, ax = plt.subplots()

for phase_name, amounts in sol_res.cum_phase_amounts.items():
    if amounts[-1] > 0:
        ax.scatter(sol_res.fraction_solid, amounts, label=phase_name, s=2)
ax.scatter(sol_res.fraction_solid, sol_res.fraction_liquid, label='LIQUID', s=2)
ax.set_ylabel('Phase Fraction')
ax.set_xlabel('Fraction solid')
# ax.title('Al-30Zn Scheil simulation, phase fractions')
ax.legend(loc='center left', bbox_to_anchor=(1.02, 0.5))
fig.savefig("Fe-Cr-C_scheil_phase_fractions.png", bbox_inches='tight')

[Chuangye-Wang]

@bocklund Thanks a lot! I will test and let you know next week.

[Chuangye-Wang]

@bocklund @sorkins Confirmed that it fixes the duplicated values with pycalphad 0.11.1.
@bocklund Thanks a lot for your help!

[bocklund]

Great, I'll try to make any last polishing changes and try to get it merged and a new release out 😄 Thanks for your patience through my busy summer.