Output files reference

All files land in runs/{run_id}/.

File	Written by	Columns / content
params.json	Python scripts	Input parameters (copied in)
logstats.dat	BioReactor	i t dt #Cells wall_time cpu_time — one row per 0.1 non-dim time
normf.dat	BioReactor	i t Omega_liq_avg Omega_liq_rms ... ux ... uy ... — vorticity and velocity norms in liquid phase
vol_frac_interf.dat	BioReactor	i t f_liq_sum f_liq_interf posY_max posY_min — liquid volume, interface length, interface y-extent
tr_oxy.dat	BioReactor	i t oxy_liq_sum oxy_liq_sum2 [c_liq...] c2_liq_sum c2_liq_sum2 ... — dissolved O₂ and tracer integrals; written from t_mix onward
results.json	postprocess.py	19 KPIs — see the table below
checkpoint.dump	BioReactor	Binary Basilisk dump at the end of each run (for chain restart)
frames/	BioReactor-video	Raw binary frame dumps (grid + VOF field per timestep) — temporary; render_videos.py consumes and deletes this directory
volume_fraction.mp4	render_videos.py	VOF interface animation, body frame (rotates with the bag)
volume_fraction_lab.mp4	render_videos.py	Same field, lab frame (fixed camera)

All .dat files have a one-line header beginning with i. Time t is non-dimensional; t_physical = t × T_bio.

kLa is NaN if tr_oxy.dat contains no data — this happens when t_end < t_mix (run finished before oxygen injection started).

results.json KPIs

All 19 keys postprocess.py writes, verified against a real run (see Your first simulation):

Key	Description	Unit
kLa_10/25/50	O₂ transfer rate at 10/25/50 % saturation (5-point log-linear fit)	h⁻¹
kLa_inst_10/25/50	Same, estimated instantaneously via dC*/dt	h⁻¹
dtmix_0.50/0.75/0.95	Time for tracer to reach 50/75/95 % homogeneity	seconds
vor_mean	Period-averaged mean absolute vorticity (steady streaming strength)	1/s
vel_rms_qss	RMS velocity over the quasi-steady-state window	non-dim
kla_fit_rmse_25	Fit quality (RMSE) of the kLa_25 log-linear regression	non-dim
tau_95/98/100_qss	Median shear-stress percentile over the QSS window	Pa
tau_95/98/100_max	Max shear-stress percentile over the QSS window (NOT the whole run — see Checkpoint restart for why that boundary matters)	Pa
tau_mean_max	Max over time of the spatially-averaged shear stress over the QSS window	Pa

Any key can be NaN if its underlying data file is absent or too short — e.g. every tau_* key is NaN if oxygen transport was disabled for the run (and vice versa: kLa_* needs tr_oxy.dat, unrelated to whether shear stress was tracked). See Non-dimensionalization for exactly how the kLa and tau physical-unit conversions work.