A variational framework for the strain-smoothed element method

TL;DR

This paper establishes a rigorous mathematical foundation for the convergence of the strain-smoothed element (SSE) method, explaining its improved performance through a novel variational principle and supporting the analysis with numerical experiments.

Contribution

It introduces a new variational principle that unifies the SSE method with other strain smoothing methods, providing the first theoretical convergence analysis.

Findings

The SSE method converges under the proposed variational framework.

Theoretical analysis explains the improved convergence of SSE.

Numerical experiments validate the theoretical convergence results.

Abstract

This paper is devoted to a rigorous mathematical foundation for the convergence properties of the strain-smoothed element (SSE) method. The SSE method has demonstrated improved convergence behaviors compared to other strain smoothing methods through various numerical examples; however, there has been no theoretical evidence for the convergence behavior. A unique feature of the SSE method is the construction of smoothed strain fields within elements by fully unifying the strains of adjacent elements. Owing to this feature, convergence analysis is required, which is different from other existing strain smoothing methods. In this paper, we first propose a novel mixed variational principle wherein the SSE method can be interpreted as a Galerkin approximation of that. The proposed variational principle is a generalization of the well-known Hu--Washizu variational principle; thus, various existing strain smoothing methods can be expressed in terms of the proposed variational principle. With a unified view of the SSE method and other existing methods through the proposed variational principle, we analyze the convergence behavior of the SSE method and explain the reason for the improved performance compared to other methods. We also present numerical experiments that support our theoretical results.