Real-time calculation of aspect ratio, skewness and orthogonality. Compare with the guide values of OpenFOAM, ANSYS Fluent and Star-CCM+ and evaluate mesh quality instantly.
Parameters
Element type
Presets
Aspect ratio AR
Guide: <5 for interior flow; 100 or more is accepted in boundary layers
Skewness S
0 = ideal, <0.85 as a guide (Fluent recommendation)
Non-orthogonality θ [°]
°
0° = fully orthogonal, <70° as a guide (limit of OpenFOAM's non-orthogonal correction)
Smoothness ratio
Size ratio of adjacent cells; <1.2 recommended
Expansion ratio ER
Boundary-layer growth rate; 1.05–1.2 recommended
Results
1.00
Aspect ratio AR
0.00
Skewness S
1.000
Orthogonal quality cos θ
1.00
Smoothness ratio
1.00
Expansion ratio ER
100
Overall score / 100
Overall rating: Excellent(80 or more = excellent, 50 or more = acceptable, below 50 = needs improvement)
Element shape preview (colour = overall score)
Quality radar chart (per-metric scores 0–100)
Guide values of major solvers
Metric
OpenFOAM (checkMesh)
ANSYS Fluent
Star-CCM+
Equiangle skewness S
Different definition (default threshold 4)
<0.85 recommended (limit 0.95)
Skewness angle <85°
Aspect ratio AR
<1000 (warning threshold)
<5 as a guide for interior flow
—
Non-orthogonality θ
<70° (with correction)
Orthogonal quality cos θ >0.1 recommended (minimum 0.01)
—
Smoothness ratio
Generally <1.2 (size ratio of adjacent cells)
Expansion ratio ER
Boundary-layer growth rate 1.05–1.2 recommended
Definitions and thresholds differ between solver versions and settings. Make the final judgement with each solver's mesh diagnostics (checkMesh / Mesh Quality Report / Mesh Diagnostics).
Mesh quality improvement by Laplacian smoothing (real time)
While paused the animation stands still; press Play to resume it.
$\theta_{equi}$: ideal equiangle (60° for triangles, 90° for quads). $S\in[0,1]$ (0 = ideal)
Aspect ratio: $AR = L_{max}/ L_{min}$
Orthogonal quality: $Q_{orth} = \cos\theta$ (θ = angle between the face normal and the vector joining the cell centroids). The animation approximates it from the cell angles as $1-S$
Overall score: Each metric is normalised to 0–100 and combined with weights AR 25%, skewness 30%, orthogonality 25%, smoothness 10%, expansion ratio 10% (80 or more = excellent, 50 or more = acceptable)
Laplacian smoothing: $\mathbf{x}_i \leftarrow \mathbf{x}_i + \alpha\,(\bar{\mathbf{x}}_{nb} - \mathbf{x}_i)$ (move towards the mean of the neighbouring nodes; the animation uses α=0.14 for 70 iterations)
What Are CFD Mesh Quality Metrics?
🙋
Why is mesh quality such a big deal? Isn't it just the "preparation" before the actual computation?
🎓
Roughly speaking, how badly shaped the mesh is directly sets how trustworthy the results are. In a car aerodynamics analysis, for instance, if the mesh on the bonnet is severely distorted, the pressure distribution there becomes ragged and the real flow is no longer reproduced at all. When several of the five metrics on this simulator's radar chart turn red at once, that is a warning that computing on this mesh may be meaningless.
🙋
Really! So what kind of mesh do I get when I move the "skewness" slider above?
🎓
A mesh with large skewness (close to 0.9, say) has quads squashed into rhombi and triangles stretched into slivers. In practice these distorted elements appear when a complex shape is forced to be filled with mesh. With such elements the gradient calculation in the discretised Navier–Stokes equations goes badly wrong and causes oscillations in pressure and velocity. When you change the value in the tool, the corresponding axis of the radar chart collapses inward and its point turns red, so you can see at a glance that something is wrong.
🙋
"Aspect ratio" and "expansion ratio" sound similar – what is the difference? Both are about differences in size, aren't they?
🎓
Good question! The aspect ratio is the ratio of the longest to the shortest edge within one cell. Thin hexahedral cells that resolve a boundary layer deliberately reach AR=1000 or so. The expansion ratio, on the other hand, describes an abrupt change in volume between neighbouring cells. When it is large, flow quantities such as temperature and pressure become discontinuous across cell faces and the computation tends to diverge. Move the two separately in the simulator and compare how the radar chart responds; the difference becomes clear.
Frequently Asked Questions
The aspect ratio is the length of the longest edge of a mesh element divided by the length of its shortest edge. The closer to 1, the closer the element is to an ideal square or cube; larger values mean a more elongated element. In ANSYS Fluent a value below 5 is the guide for interior flow (100 or more is accepted in boundary layers), and OpenFOAM's checkMesh warns above 1000.
With high skewness the angles inside the element deviate strongly from the ideal shape, so the discretisation error of the numerical scheme grows. The accuracy of gradient calculations and diffusion terms in particular drops, causing divergence or unphysical oscillations of the solution. As a guide, values of 0.8 or more need correction.
The tool shows the guide values of OpenFOAM, ANSYS Fluent and Star-CCM+ side by side in a reference table so you can compare them with your inputs. From the overall score (out of 100) it gives a three-level rating: excellent, acceptable or needs improvement. Because some metrics are defined differently per solver (OpenFOAM's skewness, for example), make the final decision with each solver's own diagnostics.
If orthogonality is poor, review the snapping and layer settings of the mesh. Specifically, reduce the growth rate of the boundary-layer mesh, increase the number of smoothing iterations, or, for hexahedral meshes, adjust the block topology so that cell faces are orthogonal to the flow direction.
Real-World Applications
External aerodynamics of cars and aircraft: Boundary-layer meshes on car bodies and wing surfaces use slender, high-aspect-ratio elements. At body corners, however, skewness tends to grow, and unless it is checked the accuracy of the predicted drag coefficient (Cd) drops sharply.
Thermal-fluid analysis of electronics: The gaps between heat-sink fins and the surroundings of IC chips are geometrically complex regions where mesh quality easily deteriorates. In meshes where the expansion ratio changes abruptly, the heat-transfer coefficient calculation becomes unstable and risks being over- or underestimated.
Combustion analysis with chemical reactions: Near the flame front where reactions occur, concentration and temperature gradients are extremely steep. A mesh with poor orthogonality (high skewness) there distorts the discretisation of the various transport equations and corrupts the computed reaction rates and flame propagation speed.
Integration with CAE software: OpenFOAM's checkMesh, ANSYS Fluent's Mesh Quality Report and Star-CCM+'s Mesh Diagnostics compute and flag the same kind of metrics as this tool. A high-quality mesh is a prerequisite for better solver convergence and for obtaining more reliable results in less time.
Common Misconceptions and Pitfalls
First, there is the misconception that "the mesh is perfect if the whole radar chart is green". In fact, depending on the type of analysis and the region, a metric is sometimes made worse on purpose. In a boundary-layer analysis around an aircraft wing, for example, extremely slender cells normal to the wall (aspect ratio 1000 or more) are used – a "strategic deterioration" that captures the physics correctly. The tolerances in the tool are only general guides; in the end, judge by what you want to compute.
Second, over-confidence that "it passed the tool's check, so it's fine". The tool looks at the geometric quality of the mesh, but that alone is sometimes not enough. In a supersonic flow with a shock wave, for example, the shock position will be wrong unless the mesh resolution along the flow direction is fine enough – something the tool's metrics cannot measure. The balance between geometric quality and physical resolution is what matters.
Finally, overlooking locally bad elements. Even if the overall averages look good, a single element with skewness 0.9 in a critical part of the flow (the separation point behind a car's door mirror, say) can amplify errors from there and ruin the whole flow field. After computing the metrics with the tool, make it a habit to visualise where the badly scored elements actually are.
Enter the aspect ratio (AR) from your mesh statistics. The slider covers 1–100; check larger boundary-layer values separately
Set the skewness (0–1). The closer to 0 the better; Fluent's guide is below 0.85 (limit 0.95). OpenFOAM's checkMesh uses a differently defined skewness (default threshold 4)
Specify the non-orthogonality angle θ (0–89°). cos θ is shown as an output; the smaller the angle, the better the orthogonality
Enter the smoothness ratio and the expansion ratio ER (growth rate) and read the overall score out of 100
Worked Example
A good mesh around an aerofoil: aspect ratio AR=20, skewness=0.35, non-orthogonality θ=25°, smoothness ratio=1.2, expansion ratio ER=1.2. With the weighting on this page the sub-scores are AR 70, skewness 88, orthogonality 85, smoothness 100 and ER 100, giving an overall score of 85 and the rating "Excellent".
Practical Notes
In ANSYS Fluent an equiangle skewness of 0.85 or less is recommended, and cells above 0.95 cause convergence problems. In this tool the sub-score drops to 20 at S=0.85 as well
Set the boundary-layer expansion ratio to 1.05–1.20. Above 1.25 the near-wall resolution becomes coarse and the control of y⁺ and the accuracy of the turbulence model suffer (in this tool ER 1.5 scores 60)
Cells whose orthogonal quality cos θ falls below 0.1 (θ≈84° in this tool) have large gradient errors (numerical diffusion), degrading diffusion and reaction terms. Fluent recommends 0.1 or more, with 0.01 as the minimum
When optimising the mesh of a complex geometry, set an overall score of 75 or more as the initial target and improve step by step; perfecting every metric at once wastes time