CUDA PCISPH

Predictive-Corrective Incompressible Smooth Particle Hydrodynamics developed with NVIDIA CUDA for GPU multithreading. It is a tool for solving Navier-Stokes' equations in a particle-based fluid simulation called Smooth Particle Hydrodynamics in much faster than applications run only with CPUs. This method predicts and corrects future positions for each particle forcing the fluid's incompressibility.

Getting Started

NOTE: This project can only be run on Windows OS.

These instructions will get you a copy of the project up and running on your local machine for development and testing purposes. See deployment for notes on how to deploy the project on a live system.

Prerequisites

Your machine needs to be CUDA capable, i.e., it needs to have a CUDA-enabled Graphics Card Unit (GPU). For more info about CUDA-enabled GPUs, please refer to NVIDIA official developer website or the CUDA page on Wikipedia.

Before proceeding with the CUDA installation, you should first install the latest version of Visual Studio with Desktop Development with C++. For detailed info, refer to this tutorial.

After the successful installation of Visual Studio, if you have a CUDA-capable GPU, you must install the latest version of CUDA Toolkit. Select your OS and version accordingly, it does not matter if you choose exe(local) or exe(network). For detailed info and how to test the installation, please refer to this tutorial.

After the installation of CUDA Toolkit, open Windows Command Prompt and type nvcc --version. You should get something similar to:

Copyright (c) 2005-2019 NVIDIA Corporation
Built on Wed_Oct_23_19:32:27_Pacific_Daylight_Time_2019
Cuda compilation tools, release 10.2, V10.2.89

Physical properties and initial conditions

The physical properties and initial conditions should be set in the /props directory. Each file is self-explanatory. This is where you can change the simulation for your desired conditions.

Deployment

To deploy this project, you should build it with Visual Studio. Open CUDA_PCISPH.sln in Visual Studio, select Release x64 in the solution configuration just left of the green play button on the top of the screen and go to the Solution Explorer (keyboard shortcut Ctrl+Alt+L). Right click on CUDA_PCISPH and select the Build option. After it's done, go to the same folder of CUDA_PCISPH.sln and look for the \bin directory. Open it and go to the \x64 folder and look for the .exe file. This is the application. You can move this file to another directory. Just remember to move the \props directory as well. Double click on it to run.

Post processing

All the simulation results will be on the /results directory, created in the same directory the .exe is located. The post processing can be done using Paraview. It can read the files that end with .pvd and .vtu generated by this project. The .pvd file consists in a group of files with time markings and a .vtu is a single instance in time of the simulation. Open PCISPH.pvd and boundary.vtu to start post processing the simulation.

Contributing

Please read contributing.md for details on our code of conduct, and the process for submitting pull requests to us.

Authors

This project was developed by José Antonio Rico Coque and Bartosz Powalka in West Pomeranian University of Technology (ZUT) on the Faculty of Mechanical Engineering and Mechatronics.

Next Steps and Ideas

1. Create an universal fluid and boundary particles positions generator, to create complex geometries with intersections and sharp edges;

2. Add thermal equations to the analysis;

3. Create a function to calculate the normal of boundary particles globally (which does not happen in this program);

References

1 Solenthaler, B. & Pajarola, R. Predictive-Corrective Incompressible SPH. (2009)

2 Ihmsen, M., Akinci, N., Gissler, M. & Teschner, M. Boundary handling and adaptive time-stepping for PCISPH.

3 Akinci, N., Ihmsen, M., Akinci, G., Solenthaler, B. & Teschner, M. Versatile Rigid-Fluid Coupling for Incompressible SPH.

4 Akinci, N., Akinci, G. & Teschner, M. Versatile surface tension and adhesion for SPH fluids. ACM Trans. Graph. 32, 1–8 (2013).

5 Palyanov, A. & Khayrulin, S. Sibernetic: A software complex based on the PCI SPH algorithm aimed at simulation problems in biomechanics. Russian Journal of Genetics: Applied Research 5, 635–641 (2015).

Notes

1. Some scripts of the code have descriptions. A good order to start reading my code is:

main.cu >> main_functions.cu >> particle_positions.cuh >> particle_parameters.cuh

2. All reference numbers in the code descriptions (noted by brackets []) are related to the reference section in this page.