Structured or Unstructured Meshes: What Works Best for Turbomachinery CFD

In computational fluid dynamics (CFD), meshing is a critical step for achieving reliable simulations, especially when combined with a robust solver strategy. As turbomachinery blade geometries become more intricate and design cycles shorten, traditional meshing approaches are often not enough. To keep pace, we must adopt advanced methodologies, and more importantly, quantify their impact on results.

GAMM Francis Turbine

This study compares a full turbine hill chart computed using two different mesh types for the bladed components (runner and distributor): a traditional structured mesh and an unstructured surface-to-volume (S2V) mesh in Fidelity CFD. All simulations were performed with Fidelity Flow, leveraging both CPU and GPU architectures to assess performance.

The Challenge with Traditional Meshing

Structured meshing has long been the standard for turbomachinery analysis. However, it often becomes a bottleneck when dealing with complex blade geometry features, making the process difficult to automate.

In contrast, the S2V unstructured meshing workflow in Cadence Fidelity offers a robust alternative that significantly extends the application range of CFD. This raises an important question: how does the solution accuracy of an S2V mesh compare to a traditional structured one?

Setting Up the Comparison

To find out, two different mesh setups were created.

Fidelity Autogrid (Structured)

This approach uses a block-structured mesh with automatic topology and grid point distribution. It features high-quality blade-to-blade (B2B) smoothing, representing the conventional standard for turbomachinery simulations.

Fidelity Hexpress (S2V Unstructured)

This unstructured mesh is hex-dominant and features automatic periodic domain creation.

Key characteristics include:

• Anisotropic refinements at the leading and trailing edges
• Localized surface and volumetric refinements for precision
• Matching periodic boundaries
• An extruded boundary layer mesh

The result is a high-quality unstructured mesh that showcases the efficiency of the S2V approach.

Simulation and Numerics

With the meshes prepared, 60 simulations were run to generate a full hill chart, covering 10 guide vane openings and 6 rotation speeds.

Simulation Setup:

Boundary Conditions: Total pressure and velocity direction at the inlet; static pressure at the outlet.

Walls: Both rotating and non-rotating walls were defined.

Rotor-Stator Interface: A mixing plane was used.

Numerics:

The simulations were run using the Fidelity Flow Solver under steady-state conditions with the k-omega SST turbulence model. This solver is an unstructured, cell-centered, second-order finite-volume solver. It is pressure-based and features full coupling of pressure and velocity, making it ideal for this type of analysis. For computing, high-performance computing with MPI-based parallelization and GPU acceleration was utilized.

The Results: A Close Match in Accuracy

After analyzing global values, the efficiency hill charts produced by both methods were remarkably similar. The peak efficiency difference between the Fidelity Autogrid structured meshes and the Fidelity Hexpress S2V meshes was a mere 0.2%.

This study demonstrates that the automated S2V meshing workflow produces results highly comparable to those of structured meshing, offering a compelling alternative for turbomachinery design cycles without sacrificing accuracy.

*Information in the blog article sourced from Numlberica

If you’re looking to strengthen your understanding of CFD workflows for hydraulic turbines, we’ve put together a short, practical tutorial series using Cadence Fidelity CFD.

Episode 1: GUI and Geometry – navigating the interface and setting up turbine geometry

Episode 2: Meshing – generating meshes for accurate CFD results

Episode 3: Simulation and Post-Processing – running cases and analyzing performance