eSPH is a simple, lightweight 2D SPH code written in MATLAB. The method is based on a high-order, low-dissipation Riemann solver SPH architecture.
It is part of Enson's final year project for his MEng degree at Imperial College London. Please refer to my thesis for more information and references.
The code uses the functions in this repository and MATLAB built-in functions only. After downloading, remember to put all member functions in the same directory as the eSPH.m.
To enable parallel computing, please ensure that the Parallel Computing Toolbox is installed in your local MATLAB.
The current version of the code is tested with MATLAB 2019a. Please report any conflicts with newer MATLAB versions.
The code is run by calling the function eSPH("$FNAME.mat").
The input .mat file contains the followings (must be in exact names):
- A Nx11 double array
fluid:
| Entry | Parameter |
|---|---|
fluid(:,1) |
x-coordinate |
fluid(:,2) |
y-coordinate |
fluid(:,3) |
density |
fluid(:,4) |
mass (constant throughout simulation) |
fluid(:,5) |
pressure |
fluid(:,6) |
x-velocity |
fluid(:,7) |
y- velocity |
fluid(:,8) |
(Reserved) |
fluid(:,9) |
reference density of fluid, rho_0 |
fluid(:,10) |
artificial speed of sound, c_0 |
fluid(:,11) |
kinematic viscosity, nu |
- A Nx4 double array
wall:
| Entry | Parameter |
|---|---|
wall(:,1) |
x-coordinate |
wall(:,2) |
y-coordinate |
wall(:,3) |
x-velocity |
wall(:,4) |
y-velocity |
-
A function handle
ffor the body forcef = @(x,y,t) [f_x; f_y]Note: x, y are Nx1 arrays, so the output of
fhas to be Nx2. You can do this naively byf = @(x,y,t) [f_x; f_y] + 0*x'. -
A 1x8 double array
settings:
| Entry | Parameter | Options |
|---|---|---|
settings(1) |
Kernel function | 3 - Cubic b-spline 4 - Quartic b-spline 5 - 5th order Wendland |
settings(2) |
Kernel support radius, kh | / |
settings(3) |
Gamma in equation of state (usually 7 for water, 1.4 for air) | / |
settings(4) |
Reconstruction scheme | 1 - MUSCL piecewise linear 2 - MUSCL piecewise parabolic (not TVD) 3 - 3rd order WENO-Z otherwise - piecewise constant |
settings(5) |
Riemann solver | 0 - Classical SPH w/o dissipation 1 - Roe solver |
settings(6) |
Switch for second derivative | 1 - on 0 - off |
settings(7) |
Simulation end time (non-dimensionalised by t_ref) | / |
settings(8) |
CFL number (must be <=1) | / |
- A double
dt_save, indicating the time between which each output file is written - A string
dir_name, specifying the name of the directory where the output files are stored in
The code outputs
- a series of
SPHout_$NSTEP.matfiles according to the inputs - a
detail.txtfile containing the settings and runtime
All kinds of discussions and developments are more than welcomed. However, since the project is finished, please do not expect a significant update to the code.
This source code is licensed under the MIT license, which can be found in the LICENSE file.