Comparative Analysis of Finite Difference and Finite Element Method for Audio Waveform Simulation

TL;DR

This paper compares the accuracy, efficiency, and applicability of Finite Difference and Finite Element Methods in simulating audio waveforms, highlighting their respective advantages and limitations through practical examples.

Contribution

It provides a systematic comparison of FEM and FDM for waveform simulation, including implementation details and analysis of their performance on real-world and analytical cases.

Findings

Both methods achieve similar accuracy levels.

FEM converges faster with complex geometries.

FDM offers quicker computation for simpler configurations.

Abstract

In many industries, including aerospace and defense, waveform analysis is commonly conducted to compute the resonance of physical objects, with the Finite Element Method (FEM) being the standard approach. The Finite Difference Method (FDM) is seldom used, and this preference is often stated without formal justification in the literature. In this work, the accuracy, feasibility, and time of simulation of FEM and FDM are compared by simulating the vibration of a guitar string. Python simulations for both methods are implemented, and their results are compared against analytical solutions and experimental data. Additionally, FDM is applied to analyze the sound of a cycling bell to assess its reliability compared to a real cycling bell. Final results show that both FEM and FDM yield similar error margins and accurately predict the system's behavior. Moreover, the errors from FEM and FDM follow the same periodicity with a phase shift when varying the assumed analytical tension and without a phase shift when changing the time interval. However, FEM converges faster with increasing mesh complexity, whereas FDM demonstrates quicker computational performance and achieves stable solutions even with bigger time intervals. Despite this FDM is limited to simpler configurations and often demands extensive mathematical formulation, which can become cumbersome for intricate shapes. For example, modeling a hemispherical object using FDM results in significant simulation times and big calculations. In conclusion, while FDM may offer faster convergence and computation time in certain cases, FEM remains the preferred method in industrial contexts due to its flexibility, scalability, and ease of implementation for complex geometries.