Cases

Opened in 1996, Cornell University Hospital for Animals is one of the finest veterinary teaching hospitals in North America. One of their clinical specialties in sports medicine is stress tests performed on horses exercising on a high-speed treadmill at their Equine Performance Testing Clinic. Here they identify respiratory obstruction in performance horses, using finite element analysis tools like airway pressure determination and airflow measurement. It usually takes several weeks to prepare a scan of a horse’s airway system for modelling these types of analysis. Intrinsic to human and animal geometry are sharp or thin-walled areas. These require a very fine surface mesh to enable the generation of a volume mesh. Without a good volume mesh, CFD software will fail to return (accurate) analysis results. The surface mesh generated by medical image processing software, however, does not contain the required high level of triangulation. In the past, the Cornell University team would improve surface mesh quality by moving and reshaping triangle after triangle at critical areas. This is a very time-consuming, labour-intensive process.

 

A high-quality surface mesh in a few days Intrigued by the possibility of using Materialise’ Remesh technology to auto mate this mesh optimization process, Cornell University sent us a benchmark. They made CT scans of a race horse’s nasal cavity, nasopharynx and larynx, and converted these into an STL model. The STL model required extensive mesh optimisation before it could be used for CFD. It took our Remesh team only a few days to return the file with a mesh directly transformable in a high-quality volume mesh for CFD.

 

A - Rough triangulation (because the CT scaning wasn't fine enough)
B - Thin wall, with current triangulation it's impossible to fit good tetrahedrons
C - Highly skewed triangles, not meeting the quality requirements

Easy preparation of thin-walled and sharp geometry for CFD

 

The nasal cavity model output produced by the image processing software contained many sharp edges and thin walls. In those areas the triangulation was not fine enough to enable CFD software to fit tetrahedrons inside. Overall, the triangulation didn’t meet the general requisites on triangle shape imposed by the CFD software. Using Remesh software’s wide range of highly advanced, automated tools, we could easily optimise the surface mesh both globally and locally.

 

 

After a global remeshing operation on the entire file, the majority of the triangles conform to the required triangle shape quality. Next, the software allows for automatic selection and inspection of all sharp geometry. The grouping functionality enables the user to view and remesh specific critical areas further. Grouping is a very powerful tool: it makes only the triangles around a certain problem area (defined f.e. by triangle quality or geometry sharpness) visible. The user can iterate over each problem area individually or treat them all simultaneously. You don’t need to spend days rotating and zooming in and out, looking for possible problem areas, some of which may be hidden inside the geometry. In this case, a few of the hardest problem areas (spikes) were deleted and refilled. Then, local remeshing (refining) and smoothing operations were performed on the remaining groups until the mesh was suited for CFD.

Thanks to the high level of automation available in Remesh software, this file was optimised in a day and a half. Preprocessing software would have taken weeks to obtain the same results.