237 lines (190 loc) · 6.22 KB

V0.1

Planned

add python frontend
add gpu acceleration
reduce todo to 0
[LPT] eliptic particles
[LPT] fluid collision model
[LPT] breakup model
[LPT] dynamic drag model
[LPT] plastic collision

V0.0.4

Planned

[LPT] calculate timestep by characteristic time
[LPT] add simple intra-collision model
[LPT] add breakup models
[LPT] add collision code for LPT
add multilevel method
ray tracing of particles
initial material point method
mpi communication methods
port testing tool over to python
compare timings between runs of the testing tool
[LIB] move more stuff to common lib project
[LIB] refactor kdtree and add more complete test
add intel compiler option
add github website
reduce todo below 50

V0.0.3

Planned

run basic 3D NS Case
run basic sph case
restart
store meta data in output files (version, node etc.)
output performance data
capillary simulation
add support for multiple species
add support for multiple phases
add injection models
add images to readme
allow more than one solver to run at the same time
write out lines as svg
set write out precision
write out memory statistics
refactor comparison with analytical solution to postprocessing
[LBM] calculate total energy/internal energy for output purposes (also output pressure and temperature)
[LBM][POISSON][BUG] the precision of D2Q5 is not correct for Poisson
[LBM] change convergence criterium to some average not maximum
[LBM] fix NEBB bnd cnd
[LBM] add zou-he bnd cnd
[LBM][BUG] Periodic bnd doesn't work always (see step_ns_bug1.json)
Allow in IO functions to set output variable types
[test script] in test script check if bin is current
[test script] run multiple bins
[test script] backup binary for each run
[test script] run test on remote host

moved

load grid from disk
Couette in 3D
Poiseuille flow in 3D
contact angle
[LPT] add tutorial
[LPT] compare solution at multiple time points
determine analytical solutions for 2D poisson cases
[BUG] All wetnode boundary condition case for poiseuille don't converge against a precise enough solution? (overprediction in the center??) (related to the uncertainty about calculating the relaxation??)
move comparison with analytical solution to postprocessing
order boundary conditions by type (is this needed?)
[BUG] D2Q5 still uses diagonal neighbors also this is not necessary(is this needed?)
[LBM] add support for boundary ghost cells in periodic boundary (needed??)
[IO] allow switching of output filetypes by handling a configuration object
[IO] simplify writing output
[IO] warn when overwriting files
[IO] give option to not overwrite files
[IO] set output format for the solvers
[IO] set output variables for the solvers
[IO] check in output functions if fileformat ending is already included
add way to just iterate over leaf cells
add more openmp
allow comments in configuration files
add ns compatible neumann bndcnd
refactor forcing setup
test all bnds for both poisson and NS
- working:

Working on:

add nernst-planck equation case

Done:

Features

Buildsystem

Testing

Usability

IO

Performance

Bugs

Documentation

Refactoring

V0.0.2

Features

Lagrange Particle Solver
- Forces:
  - Gravity
  - Buoyancy
  - Stokes Drag
  - Non-linear Drag
- Integration methods:
  - Forward Euler
  - Implicit Euler
- analytical tests:
  - free fall
  - stokes drag
  - terminal velocity
- simple injection model
Lattice Boltzmann Solver
- add D3Q19 and D3Q27
- Boundary conditions
  - Wet node equilibrium method wall
  - Wet node non-equilibrium extrapolation method wall
  - Wet node non-equilibrium bounce-back method wall
  - BB Dirichlet bnd
  - AntiBB pressure bnd
  - Neumann NEEM bnd
- Analytical test:
  - Poisson:
    - 1D Debye-Hueckel
    - 1D Diffusion Slab
    - 2D Helmholtz Equation
  - Navier-Stokes:
    - Couette
    - Poiseuille
- Examples:
  - Navier-Stokes:
    - single step
    - double step
    - flow around cylinder
  - Poisson
    - single step
    - double step
    - 2D diffusion
    - 2D Debye-Hueckel
Math expression support for boundary conditions

Testing

check for divergence when debug is active
write test report in CSV
for analytical testcases allow alignment with the boundary

Usability

add pressure forcing term in periodic boundary condition
allow to set a bnd per geometry obj
allow setting "conv_interval" for the interval for the convergence check
write output every "60" seconds by default as a keep alive msg
- time interval is settable by "keep_alive_time"

IO

set output dir and solution file name
allow multiple cell output filters
correctly handle switch between int and double in writePoints()

Bugs

segmentation fault after divergence
write out full double precision
[LBM][POISSON] fix D2Q9 for Poisson
error in Base64 encoding

Documentation

improve documentation of testcases

Refactoring

reduce the number of debug levels
refactor output cell filter code
remove shift from base64 which didn't work

V0.0.1

Features

load grid generated from the generator
setup diagonal neighbor connection for D2Q9 and similar methods
periodic boundaries
general bounce back "no-slip" wall boundary condition
general bounce back "no-slip" wall boundary condition with tangential velocity
Poisson Non-equilibrium extrapolation method based dirichlet boundary condition
load bnd conditions from configuration file
added forcing
Postprocessing
- Lines
basic openMP
supports D1Q3, D2Q5 and D2Q9
Support the solution of Navier-Stokes and Poisson-Boltzmann equation

Testing

Added Couette flow testcase
Added Poiseuille flow testcase
option to exclude surfaces from error calculation

Usability

exit on nan during convergence check
Add postprocessing structures
define boundary condition in the configuration file

IO

differentiate between float and integers correctly in the VTK output
write lines output to csv