Output & Visualization¶

Output is split per-solver under code_base/result/<case_name>/:

result/<case>/
├── thermal_fluid_results/     # thermal-fluid solver
├── species_results/           # species transport solver (own uniform grid)
├── mechanical_results/        # mechanical solver
├── coupling/                  # thermal → CA per-step bbox frames (binary)
└── CA_results/
    ├── initial_grain/         # CA initial Voronoi grain structure
    └── scan_grain/            # CA thermal-fluid coupled remelt + resolidification

Output Files¶

Thermal-fluid (`result/<case>/thermal_fluid_results/`)¶

File	Description	When
`<case>_output.txt`	Iteration-by-iteration text log	Every time step
`<case>_vtkmov{N}.vts` + `<case>_vtkmov.pvd`	XML StructuredGrid + zlib-compressed base64 payload (~7-8× smaller than legacy `.vtk`). Open the `.pvd` for time-aware playback.	Every `outputintervel` steps
`<case>_timing_report.txt`	CPU time breakdown by module	End of simulation
`<case>_memory_report.txt`	Peak memory usage (VmHWM headlined; see below)	End of simulation
`<case>_defect_report.txt`	Defect fraction summary	End of simulation
`<case>_maxtemp.vtk`	Max temperature field (VTK)	End of simulation
`<case>_defect.vtk`	Defect classification field (VTK)	End of simulation
`<case>_thermal_history.txt`	Temperature at 10 monitoring points	Every time step
`<case>_thermal_history.png`	Temperature evolution plot	End of simulation
`<case>_meltpool_history.txt`	Melt pool length, depth, width, volume, Tpeak	Every time step
`<case>_meltpool_history.png`	Melt pool geometry evolution plot	End of simulation

Species (`result/<case>/species_results/`)¶

Written only when species_flag=1. Lives in its own folder because the species solver runs on its own uniform X/Y grid, decoupled from the thermal AMR grid.

File	Description	When
`<case>_species{N}.vts` + `<case>_species.pvd`	XML StructuredGrid + zlib-compressed base64. Fields: `Velocity`, `T`, `concentration`. Open the `.pvd` for time-aware playback.	Every `outputintervel` steps

Mechanical (`result/<case>/mechanical_results/`)¶

File	Description	When
`<case>_mech_NNNNN.vtk`	Mechanical VTK (stress, displacement)	Every `mech_output_interval` solves
`<case>_mech_history.txt`	Stress/displacement at monitoring points	Every mechanical solve
`<case>_mech_history.png`	Stress/displacement evolution plot	End of simulation
`<case>_mech_timing_report.txt`	Mechanical solver timing breakdown	End of simulation
`<case>_mech_memory_report.txt`	Mechanical solver peak memory	End of simulation
`<case>_deformation.gif`	Von Mises stress animation (10x deformation)	End of simulation
`<case>_deformation_final.png`	Final stress state (high-res)	End of simulation

CA (`result/<case>/CA_results/{initial_grain,scan_grain}/`)¶

File	Description	When
`Grains.vti.series` + `grainsTimeStates/Grains_*.vti`	Grain-ID time series (VTK XML ImageData + JSON index). Each frame carries `grain_id` (Int32) and `grain_orientation` (Float32×3), zlib-compressed (`vtkZLibDataCompressor`, 32 KB blocks, level 1) and stored with `format="binary"` inline base64. Typical on-disk size is ~25–30 % of the raw 16 B/cell payload (compression dominated by base64's 33 % overhead — the deflate step itself hits ~20× on `grain_id`). Dumps run in a forked child process so the CA main loop is never blocked on disk I/O; the child also calls `fsync` + `posix_fadvise(DONTNEED)` before exiting so the file is flushed and evicted from the page cache, keeping the dirty-page budget available for the concurrent thermal writer. ParaView opens `.vti` natively; no decompression step needed.	Per CA output frequency
`microstructureInfo.txt`	Final grain state (serial dump)	End of step
`grainSizeStatistics.txt`, `average_grain_size.txt`, `grain_size_histogram.png`	Post-processed grain-size stats	Mid-run (every 20 VTK frames) + end
`timeLog.dat`, `memoryLog.dat`	CA solver timing and memory	Mid-run + end

output.txt Format¶

Each time step produces a 4-line block:

  time  iter  time/iter  tot_iter  res_enth  res_mass  res_u  res_v  res_w  [res_spec]
 2.10E-05   60    0.006        60   1.2E-04   5.0E-02  1.3E-02  ...

  Tmax        umax       vmax         wmax       length       depth     width
  1594.47   0.00E+00    0.00E+00    0.00E+00   ...

  north  south  top  toploss  bottom  west  east  hout  accu  hin  heatvol  ratio
    0.0    0.0    0.0    0.0    0.0    0.0    0.0   0.0   0.0   0.0   0.0   0.00

progress%  beam_posx  beam_posy  beam_posz  power  scanspeed  speedx  speedy
     0.70  5.258E-04  2.000E-03  7.000E-04  300.0    1.230    1.230    0.000

Field	Description
`res_enth`	Enthalpy equation residual
`res_mass`	Mass conservation (pressure correction) residual
`res_u/v/w`	Momentum residuals
`res_spec`	Species residual (only when `species_flag=1`)
`Tmax`	Peak temperature in domain (K)
`umax/vmax/wmax`	Peak velocities (m/s)
`length/depth/width`	Melt pool dimensions (m)
`ratio`	Energy balance ratio (should be ~1.0)
`progress%`	Simulation progress (%)

VTK Files¶

XML StructuredGrid (.vts) files with zlib-compressed base64 payloads, indexed by a .pvd Collection for time-aware playback. Native ParaView open, no plugins.

Thermal `.vts` fields (`vtkmov*.vts`)¶

Field Name	Description	Unit
`Velocity`	Flow velocity (cell-centred, zeroed below solidus)	m/s
`T`	Temperature	K
`vis`	Viscosity	Pa·s
`diff`	Thermal diffusivity	m²/s
`den`	Density	kg/m³
`fracl`	Liquid fraction (0=solid, 1=liquid)	-

Species fields (concentration) are NOT written here — they live in their own species_results/<case>_species{N}.vts series so the two solvers' output streams stay decoupled. See Species Transport › Output.

Opening in ParaView¶

File → Open → select <case>_vtkmov.pvd (the time-series index)
Click "Apply"
Select scalar / vector field from dropdown
Use animation controls to step through time

The same workflow applies to species_results/<case>_species.pvd and CA's Grains.vti.series.

Useful ParaView Filters

Threshold: Show only melt pool (T > 1563)
Slice: Cut through domain to see cross-sections
Glyph: Visualize velocity vectors
Calculator: Compute derived quantities (e.g., T - 1563 for superheat)

Timing Report¶

============================================
  PHOENIX Module Timing Report
============================================
  Total iterations (itertot): 17055
  Total CPU time:     2513.425 s
  Total wall time:      765.089 s
--------------------------------------------
  Module              |   Time(s)   |   Ratio(%)
--------------------------------------------
             mod_prop |     440.355  |   17.52%
             mod_sour |     406.017  |   16.15%
             ...
           mod_species|       4.355  |    0.17%

Memory Report¶

All three solvers' memory reports (*_memory_report.txt, *_mech_memory_report.txt, CA_results/*/memoryLog.dat) lead with VmHWM — the only number that should drive machine-sizing decisions.

============================================
  PHOENIX Memory Report
============================================

  *** VmHWM is THE number to look at ***
  Use it to size the machine; the others are diagnostic only.

  VmHWM:   ...   peak PHYSICAL RAM (RSS) ever used during the run.

  --- diagnostic / less important ---

  VmPeak:  ...   peak VIRTUAL memory size (much larger than VmHWM)
  VmRSS:   ...   physical RAM at the MOMENT this report was written
  VmData:  ...   heap + stack, excludes shared libraries

Metric	What it tells you	Use for sizing?
VmHWM	Peak physical RAM ever used	Yes
VmPeak	Peak virtual memory ever requested	No (VM ≠ RAM)
VmRSS	Physical RAM at exit-time snapshot	No (could be < peak after frees)
VmData	Allocated heap + stack	Detail only

The mechanical subprocess writes its own *_mech_memory_report.txt with the same VmHWM-headlined block plus a static breakdown of FEM-array allocations (sig_gp + eps_gp, displacement, stress output) — useful for understanding which mech array dominates the peak.

Defect Report¶

  === Defect Analysis (maxtemp_determ) ===
  Defect fraction:           X.XXXXXX %
  Lack-of-fusion fraction:   X.XXXXXX %
  Keyhole porosity fraction: X.XXXXXX %

Spatial distribution available in thermal_fluid_results/<case>_defect.vtk and thermal_fluid_results/<case>_maxtemp.vtk.

Melt Pool History¶

Time-series log of melt pool geometry, recorded every timestep.

Column	Description	Unit
time	Simulation time	s
length	Melt pool length (along scan)	m
depth	Melt pool depth (from surface)	m
width	Melt pool width (transverse)	m
volume	Melt pool volume (from liquid fraction)	m³
Tpeak	Peak temperature in domain	K
laser_on	Laser state (1=on, 0=off)	-

Auto-generated plot: <case>_meltpool_history.png (4-panel: length, depth/width, volume, Tpeak).

Mechanical Output¶

When mechanical_flag=1, separate VTK files are written for mechanical results.

Mechanical VTK Fields (`<case>_mech_NNNNN.vtk`)¶

Field	Type	Unit	Description
`Temperature`	scalar	K	Temperature at FEM nodes
`ux`, `uy`, `uz`	scalar	m	Displacement components
`phase`	integer	-	0=powder, 1=liquid, 2=solid
`sxx`, `syy`, `szz`	scalar	Pa	Normal stress components
`von_mises`	scalar	Pa	Von Mises equivalent stress
`fplus`	scalar	Pa	Yield function (>0 = plastic)

Note

Mechanical VTK files use a coarsened FEM grid (controlled by mech_mesh_ratio), separate from the thermal VTK files which use the full simulation grid.

Mechanical History¶

Time-series log at 10 monitoring points (same as thermal history), recorded every mechanical solve.

Columns: time, T(1..10), ux(1..10), uy(1..10), uz(1..10), sxx(1..10), syy(1..10)

Auto-generated plots: <case>_mech_history.png (3-panel: temperature, displacement, stress).

Deformation Animation¶

<case>_deformation.gif shows von Mises stress with 10x deformation magnification. Uses matplotlib rendering with jet colormap. Final frame also saved as <case>_deformation_final.png at 200 DPI.