Mesh Independence Studies in Casting Simulation
A rigorous, systematic approach to ensuring simulation accuracy, computational efficiency, and predictive reliability in modern casting CAE workflows.
The Fundamental Challenge:
Discretization
Every casting simulation begins with a deceptively simple question: how should a continuous three-dimensional casting geometry be represented inside a computer? The answer is discretization. By transforming molds, runners, risers, and cavities into a finite collection of computational elements, simulation software converts continuous physical processes into solvable numerical problems. The quality of this discretization ultimately determines the reliability of every thermal, fluid-flow, and solidification prediction that follows.
What Is Discretization?
The continuous casting geometry is decomposed into thousands or millions of computational cells where governing equations for heat transfer, fluid flow, mass conservation, and solidification are solved iteratively.
The Resolution Problem
The fundamental difficulty of meshing lies in deciding how much resolution is enough. Regions most responsible for defect formation often contain the steepest thermal gradients, highest velocity changes, and most complex geometric features. These critical phenomena may disappear entirely if the mesh is too coarse.
Regions That Demand Resolution
The Mesh Resolution Dilemma
Fast But Risky
- Low memory usage
- Short solve times
- Numerical diffusion
- Missed local defects
- Poor gradient resolution
Accurate But Expensive
- High fidelity results
- Sharp interface capture
- Better defect prediction
- Large memory footprint
- Long simulation runtime
The Core Conflict
Every refinement improves numerical accuracy, but computational cost grows rapidly. Increasing mesh density adds more unknowns, larger matrices, additional memory requirements, and substantially longer solve times. Simply refining everywhere is rarely a practical solution for industrial-scale casting simulations.
Mesh independence is demonstrated through a disciplined convergence study—showing that results reflect physics, not the analyst’s choice of element size.
Mesh independence is declared when monitored variables change by less than the accepted threshold between successive refinement levels. Further refinement then yields diminishing returns.
The convergence study should be documented thoroughly, including mesh densities, monitored variables, and the observed changes between levels. This record supports traceability and justifies the selected production mesh for subsequent design studies.
The Mechanics of
Mesh IndependenceRun the same case on increasingly finer meshes.
When refinement stops changing the answer.
Non-converged results are not design data.
It is the boundary between numerical artifact and physical prediction.
Mesh density increases computational cost non-linearly. Doubling elements in 3D multiplies node count by eight. Smaller elements reduce allowable time step size, scaling runtime by an order of magnitude. A study that runs in 4 hours on a coarse mesh may require 40+ hours on a fully refined mesh.
Many casting geometries have symmetry planes. Applying symmetry boundary conditions reduces node count by 50–75%. This is cost-effective but requires validation that flow patterns respect symmetry assumptions.
Advanced strategies refine mesh locally in critical regions — thin walls, sharp transitions, gate junctions, porosity-prone zones. Surrounding mold regions are meshed coarsely, concentrating resources where accuracy matters most.
Auxiliary features like runners, vents, and overflow wells often have minimal influence on cavity flow and thermal patterns. Simplifying or removing them reduces element count substantially while preserving boundary accuracy, enabling faster turnaround.
Balancing Speed and Precision
The Direct Cost of Refinement
Symmetry Planes: Halving the Problem
Variable Mesh Blocks
Geometry Simplification
As casting simulation becomes a core component of industrial product development, mesh independence studies have evolved from academic exercises into standardized engineering procedures. Leading organizations now combine automation, quality assurance, validation protocols, and traceable documentation to ensure simulation results remain reliable throughout the entire development lifecycle.
Modern CAE teams eliminate repetitive setup work through automated convergence pipelines. Batch-processing frameworks generate meshes, configure simulations, launch solver runs, and collect results automatically across multiple refinement levels without analyst intervention.
Automated convergence studies remove repetitive setup effort, reduce the risk of human error, and dramatically shorten the calendar time required for mesh independence verification.
A finer mesh does not guarantee a better simulation. Poor-quality elements can destabilize solvers, increase numerical errors, and produce misleading results. Automated quality screening is therefore mandatory before any production simulation begins.
Best-practice organizations treat the mesh itself as a design variable. Whenever geometry changes introduce new thin walls, gating revisions, sharp radii, or altered thermal pathways, convergence assumptions must be revalidated rather than inherited from previous studies.
Best Practices in
Modern Casting CAEFour Pillars of Modern Casting CAE
Workflow Automation
Automated Convergence Workflow
Instead of Multi-Day Manual Work
Mesh Quality Control Protocols
Geometry Checks
Topology Checks
Common Sources of Numerical Failure
Iterative Validation Strategy
Mesh independence is the non-negotiable foundation for simulation-driven engineering. Without it, results are qualitative at best and misleading at worst.
Rigorous convergence testing ensures outputs reflect physical reality rather than numerical artifacts. Skipping this step introduces systematic error that no amount of post-processing can correct.
Investing in a proper mesh independence study upfront reduces physical trials, tooling modifications, and scrap—producing a compelling return on the computational investment.
The goal is the optimal mesh: coarse enough to be tractable, fine enough to converge, and strategically refined where physical gradients demand resolution.
A grid-independent, optimized mesh transforms casting simulation from an exploratory visualization tool into a genuine predictive engineering instrument—one that earns the trust of the entire product development organization.
Conclusion: The Path to
Predictive ReliabilityNon-converged results carry hidden risk.
Trust eliminates redundant trials.
Optimal, not finest.
From visualization to prediction.
It is the boundary between artifact and prediction.
What's Your Reaction?
