Menu

Bretherton drops and bubbles

Axisymmetric Basilisk solver for a gas bubble or immiscible drop translating through a liquid-filled circular tube. Runtime parameter files control the case; the current validation programme concerns Newtonian bubbles.

The scientific report is in docs/Newtonian-Validation/: main.tex, references.bib, figures, compact comparison data and the compiled main.pdf.

Requirements

  • A project-local Basilisk C installation; the validation campaign uses v2026-08-30.
  • Bash, a C compiler and qcc. The runners load .project_config when present.
  • OpenMP for --threads N; Python with NumPy for flat-film post-processing.
  • Open MPI (mpicc and mpirun) for --ranks N.

Run a case or sweep

# Compile and run the coarse smoke test.
bash runTests.sh --smoke

# Run a single case with the bubble defaults.
bash runSimulation.sh default.params --threads 4

# MPI: the rankfile must map ranks to the allocated physical cores.
bash runSimulation.sh default.params --ranks 4 --rankfile /path/to/rankfile

# Inspect the eight-case grid study, then run it.
bash runParameterSweep.sh grid-sensitivity.params --dry-run
bash runParameterSweep.sh grid-sensitivity.params --threads 4 --parallel 2

default.params defines the physical inputs and numerical controls. Ca is the imposed mean inlet capillary number; the bubble speed is measured from the output. grid-sensitivity.params repeats the four inlet values at maximum levels 9 and 11. The separate grid-sensitivity-fine.params specifies the level-12 follow-up.

Case output defaults to simulationCases/<CaseNo>/. Set OUTPUT_ROOT to a separate run directory for production calculations. Each case contains its parameters, source, executable, diagnostic log, restart dump and intermediate/snapshot-* files. An existing restart resumes that case. Success requires stable deposited central-film measurements and, by default, agreement of the front, rear and centroid speeds. Reaching tmax or the outlet buffer returns an explicit incomplete status with a nonzero exit code. For a refined restart, use a separate case directory containing a copy of the seed dump, set freshFront to its front position, and specify regridBurnR. Changing Ldomain while restoring a dump is rejected; start a fresh case to change the domain length.

For a single-node MPI allocation, --pe-list 0,1,2,3 can replace the rankfile when those are the allocated Open MPI core indices. Set MPIEXEC to select another launcher executable. --build-only compiles without running; --no-build reuses an executable after checking its source, local headers, Basilisk lock and binary checksums. --openmp --threads 1 retains an OpenMP build for one-thread tests.

Analysis and numerical checks

postProcess/bretherton_flat_film.py extracts the flat film and bubble speed from saved interfaces; it uses the compiled getFacets.c helper. postProcess/bretherton_film.py instead reads the minimum film reported in the case log. The validation report explains the difference between these measurements.

# Exact-solution checks followed by the smoke test.
bash runTests.sh

The verification cases state their comparators and tolerances. The smoke test checks compilation and short-time execution.

Source layout

Compile the scientific report

Requires latexmk, a LaTeX distribution and BibTeX:

make -C docs/Newtonian-Validation

This produces docs/Newtonian-Validation/main.pdf. The folder’s README.md lists its sources and the optional figure-regeneration command.

Build and preview the code documentation

bash .github/scripts/build.sh
bash .github/scripts/deploy.sh

The build generates the site in .github/docs/; deploy.sh starts a local preview. See the documentation workflow for source discovery and GitHub Pages publication.

Generated Documentation

Root Directory

postProcess

simulationCases

src-local

testCases

verificationCases