An adaptive global–local generalized FEM for multiscale advection–diffusion problems

Research output: Contribution to journalArticlepeer-review


This paper develops an adaptive algorithm for the Generalized Finite Element Method with global–local enrichment (GFEMgl) for transient multiscale PDEs. The adaptive algorithm detects a subset of global nodes with trivial enrichments, which are exactly or close to linearly dependent from the underlying coarse FEM basis, at each time step, and then removes them from the global system. It is based on the calculation of the ratio between the largest and smallest singular values of small sub-matrices extracted from the global system of equations which introduces little overhead over the non-adaptive GFEMgl for transient PDEs. Compared to existing adaptive multiscale approaches, where either an a-posterior error estimate, a change in physical quantities, or a local problem residual is calculated, the proposed approach provides an innovative framework based on singular values. The proposed approach is shown to be robust for solving advection–diffusion problems that require detecting initial conditions with spikes and capturing moving/morphing/merging mass plumes in heterogeneous media and non-uniform flow with sharp fronts. Specific examples are provided for applications in groundwater contaminant and heat dissipation. The accuracy of the proposed adaptive GFEMgl closely matches reference fine-scale FEM solutions in the L norm.

Original languageEnglish (US)
Article number116548
JournalComputer Methods in Applied Mechanics and Engineering
StatePublished - Jan 5 2024


  • Adaptivity/adaptive
  • Advection–diffusion
  • Generalized finite element method
  • Localized features
  • Multiscale
  • Singular value decomposition

ASJC Scopus subject areas

  • Computational Mechanics
  • Mechanics of Materials
  • Mechanical Engineering
  • General Physics and Astronomy
  • Computer Science Applications


Dive into the research topics of 'An adaptive global–local generalized FEM for multiscale advection–diffusion problems'. Together they form a unique fingerprint.

Cite this