Skip to Main Content
Article navigation
Purpose

With the development of high-performance computers, the requirements of numerical simulations on the large scale of girds are increasing. However, the quality of ultra-large-scale grids is difficult to meet the demand.

Design/methodology/approach

This study proposes an improved high-quality parallel tetrahedral generation algorithm. Key methodological steps include: (1) Employing an enhanced dual graph weight-based domain decomposition method to optimize interface elements, ensure rapid partitioning, and prevent bad cell exposure at interfaces. (2) Developing the extendable node numbering method (ExNN) to achieve fully decoupled parallelization of curve, surface, and solid meshes, enabling boundary updates/partition merging with near-linear time complexity and zero communication overhead, and flexible load balancing in curve/surface parallel processes. (3) Advancing surface projection via parametric coordinates in recursion to ensure efficiency and avoid cell flipping.

Findings

The algorithm enhances the mesh quality of the partitioned interface and the generation efficiency of massively parallel tetrahedral meshes. With the improved decoupling of the parallel algorithm, an engineering verification of meshing complex geometric models with more than 1 billion elements in 2 minutes on a 1000-core high-performance computer is passed, demonstrating that the method possesses good parallel efficiency and scalability.

Originality/value

This work contributes critical innovations to large-scale parallel mesh generation: (1) The ExNN method achieves fully decoupled parallelization of curve, surface, and solid meshes, resolving high communication overhead and complexity. (2) The dual graph weight-based decomposition balances interface optimization and partitioning speed, while parametric coordinate projection ensures efficiency and cell stability. (3) ExNN-based fast partition merging (without template restrictions) preserves freedom for parallel refinement.

Licensed re-use rights only
You do not currently have access to this content.
Don't already have an account? Register

Purchased this content as a guest? Enter your email address to restore access.

Please enter valid email address.
Email address must be 94 characters or fewer.
Pay-Per-View Access
$41.00
Rental

or Create an Account

Close Modal
Close Modal