High-performance computing for the BGK model of the Boltzmann equation with a meshfree Arbitrary Lagrangian-Eulerian (ALE) method

Panchatchram Mariappan,Klaas Willems,Gangadhara Boregowda,Sudarshan Tiwari,Axel Klar

2024-08-05

Abstract:In this paper, we present high-performance computing for the BGK model of the Boltzmann equations with a meshfree method. We use the Arbitrary-Lagrangian-Eulerian (ALE) method, where the approximation of spatial derivatives and the reconstruction of a function is based on the weighted least squares method. We have used the Graphics Processing Unit (GPU) to accelerate the code and compared with the CPU code. Two and three dimensional driven cavity problems are solved, where we have obtained the speed up up to 307 times and 127 times in two and three dimensional cases, respectively.

Numerical Analysis

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to perform efficient calculations for the BGK model of the Boltzmann equation in rarefied gas dynamics. Specifically, the authors focus on how to improve the solution efficiency through the meshless Arbitrary Lagrangian - Eulerian (ALE) method and Graphics Processing Unit (GPU) - accelerated computing when dealing with rarefied gas flow problems in Micro - Electro - Mechanical Systems (MEMS).

### Problem Background

1. **Requirements for Rarefied Gas Flow Simulation**:
- In the research of Micro - Electro - Mechanical Systems (MEMS), since the characteristic length is equal to or larger than the molecular free path, resulting in a large Knudsen number, it is necessary to solve rarefied gas flow.
- Rarefied gas flow is usually described by the Boltzmann equation, but directly solving the Boltzmann equation is very complex and computationally costly.

2. **Limitations of Existing Methods**:
- Stochastic methods such as Direct Simulation Monte Carlo (DSMC) have large statistical noise in the case of low Mach numbers and are not suitable for such problems.
- When there are moving boundaries, the fluid domain changes with time, resulting in unsteady flow problems. It is difficult to perform long - time averaging, and it is necessary to run independently multiple times and average the results, which is time - consuming.

3. **Selection of Simplified Models**:
- In order to simplify the calculation, many studies use deterministic methods to solve the simplified model of the Boltzmann equation, such as the Bhatnagar - Gross - Krook (BGK) model.

### Main Objectives of the Paper

1. **Propose an Efficient Numerical Method**:
- Use the meshless Arbitrary Lagrangian - Eulerian (ALE) method to solve the BGK model to adapt to moving boundary and unsteady flow problems.

2. **Utilize GPU - Accelerated Computing**:
- Utilize the powerful parallel computing ability of the Graphics Processing Unit (GPU) to significantly improve the computing speed of 2D and 3D driven cavity problems.
- Experimental results show that the highest accelerations of 307 times and 127 times are achieved in 2D and 3D cases respectively.

### Key Technical Points

1. **Meshless ALE Method**:
- Use the weighted least - squares method to approximate spatial derivatives and function reconstruction.
- Handle moving boundary and unsteady flow problems and adapt to complex geometric shapes.

2. **GPU Implementation**:
- Allocate computational tasks to the cores of the GPU to reduce data transmission between the CPU and the GPU.
- Optimize memory management and thread scheduling to improve computational efficiency.

### Summary

This paper aims to provide an efficient numerical solution scheme for complex flow problems in rarefied gas dynamics by combining the meshless ALE method and GPU - acceleration technology. This method not only improves computational efficiency but also can handle unsteady flow problems with moving boundaries and is suitable for research in fields such as Micro - Electro - Mechanical Systems.