TAI Meshing Support

This section is intended to provide help with meshing. If your team requires rapid support for meshing, please contact our engineering services team. The data below is largely archival. It should be considered relevant to general meshing ideas alone, as both meshing software and meshing methods have evolved since this page was written.

Review of Codes to Create Polygon Meshes
NOTE: This page was written in 1998 and is kept for archive purposes only. For updated information about meshing, contact ThermoAnalytics directly: sales@thermoanalytics.com

Creating Efficient Rhino Surfaces
Meshing With Rhino
Rhino's Detailed Mesh Controls
Quick Meshing Example With Rhino: Zipped Powerpoint slideshow illustrating a simple meshing example.

Mesh Generation at ThermoAnalytics for Thermal Analysis and IR Signature Prediction

Generating a mesh for thermal analysis is a complicated task requiring modeling expertise and advanced tools. ThermoAnalytics provides thermal meshing and modeling services for its customers. The following describes the thermal mesh generating process and the tools and techniques that TAI employs when building thermal models.

The process of generating meshes is not a linear process. Starting with an existing geometry and producing a quality thermal model depends on the geometry type (mesh or CAD source), the modeler’s preferences, the condition of the source geometry, special thermal modeling techniques, and the desired final model resolution. It is often much more difficult to generate a low-resolution thermal model than a high-resolution one because every element in a low-resolution mesh is critically important. This requires careful planning. The modeler must choose which features are thermally benign and disregard them. Features such as fillets, rounds, and antennas may be eliminated from the model. Also there are special techniques used to represent specific components. For example, on a road wheel it is preferred to use a wagon-wheel shaped mesh. This will provide more thermal mass toward the center of the wheel in the thermal solution. Another example is an aircraft wing. The wing physically has two sides, but the MuSES modeler might use a single plane of elements to represent the wing for a low-resolution model. Also, any compartments, windows, or doors are typically sealed with rubber or disconnected. In these cases, the mesh must be discontinuous to break the conduction path in the thermal solver.

Source Geometry Considerations

In creating a thermal mesh, the first factor to consider is the source of the geometry. The two main types of geometry are CAD models or meshes. For models created in Pro-Engineer, TAI has developed a plug-in to interface Pro/ENGINEER directly to MuSES. However, for most CAD geometry, including complex Pro/ENGINEER geometry, TAI creates the mesh using Rhinoceros (Rhino3D), Fluent’s Gambit, or ICEM CFD. Depending on the condition of the geometry and required detail level, the modeler will choose the appropriate package. When the CAD information is loaded into the mesher, due to tolerance issues, it usually must be healed. The process of healing fixes gaps at the vertices and edges. If the mesh is not healed, it is impossible for the mesher to automatically appropriate coincident vertex points along the edges. In Rhino3D, the healing is done by hand on the generated mesh or by rebuilding the geometry. Once healed geometry is generated, ThermoAnalytics employs one of several meshing programs depending on the model and application requirements.

Mesher Operation

Different meshers operate in different ways. The most commonly used ICEM mesher uses a loop process to fill regions of geometry with mesh elements. The loop process requires the modeler to manually create a patch on the model surface by specifying the geometry edges that define the patch. The looping process generates a plane of elements and then pulls the vertices to the surface, thus filling the looped surface with quad dominant elements. In areas where there is a large amount of curvature, the mesher uses triangular elements to avoid creation of out-of-plane (warped) quadrilateral elements. The loop method ensures that edges of features are preserved and creates a highly structured mesh using a growth technique from the boundaries.

If the geometry starts as a mesh, there are two distinct paths. The first is to use Eclectic, a tool developed by U.S. Army TACOM. Eclectic meshes over the top of an existing mesh by body-fitting a plane of quadrilateral elements over the existing mesh. The user must establish where hard and soft features exist in a semi-automated fashion before creating the Eclectic mesh. Hard features, like large-angle edges, will always be present in the final mesh; soft features need not be. The second path for starting with a mesh is to use Rhino3D, ICEM, or Gambit. This technique requires re-creation of surfaces and edges from the original mesh. In ICEM and Gambit this is done automatically, but they both require that the geometry goes through a healing process. Rhino3D depends on the modeler to create the geometry by hand. Once the geometry has been built, any of the various meshing algorithms can be used to create the final mesh.

The common factors in the meshing process are:

1. Import the mesh or CAD data into the mesher.
2. Clean up the geometry so that edges and vertices line up (heal, repair, or create geometry).
3. Identify the hard (large angle feature) edges.
4. Mesh using an algorithm that preserves the geometry as best as possible, while minimizing the element count and maintaining element quality.

Some Meshing Tools used by Engineers

Rhino3D:
Inputs: Mesh or CAD geometry
Meshing algorithms: Moderate control over mesh output
Mesh editing: Good mesh editing and hand meshing capability
CAD features: Excellent CAD package for geometry creation

ICEM CFD:
Inputs: Mesh or CAD geometry
Meshing algorithms: Several different meshing techniques. Allows very high control over final mesh and quality
Mesh editing: Excellent mesh editing capabilities
CAD features: Primitive CAD abilities

Eclectic:
Inputs: Mesh geometry
Meshing algorithms: Moderate control over mesh output
Mesh editing: Moderate mesh editing ability CAD features: None

Creating a Mesh for WinTherm, RadTherm, or MuSES

The creation of a mesh for thermal analysis consists of seven major parts:

1. The arrangement of thermal elements and facets must capture the important heat transfer features.
   • Geometry must be correct to allow for accurate calculation of the facets’ exposure to solar load, convection, and radiation exchange both external and (for convection and radiation) internal to the vehicle.
   • Heat transfer linkages must be accurately modeled. The conduction paths between internal heat sources and sinks to the outer surface facets must be included in the model at the appropriate areas. The conduction across the vehicle surface must also be modeled correctly. Both factors place constraints on the size, shape, and location of facets.
2. The facets and parts must define the boundaries of changing material type, surface optical properties (paints), thickness, and thermal boundary conditions.
3. The mesh must be of sufficient resolution to model the roundness of surfaces.
4. Adjacent facets must share common vertices (e.g. equivalence mesh). This constraint creates the need for staggered arrays of triangular elements between high- and low-resolution regions, such as between the small facets needed to define the roundness of a gun and the large facets that model the turret to which the gun is attached.
5. The mesh must incorporate modeling assumptions. Examples of modeling assumptions include simplified conduction links between engine and hull surfaces, imposition of adiabatic (insulated) boundary conditions, and modeling many-layered parts.
6. Internal heat sources, such as drivetrain components and electronics, must be included in the thermal model. Depending on the model and the application requirements, the heat sources can be modeled as heat-generating geometry or more simply as an imposed heat load on an internal air node (e.g., engine compartment air).
7. When computation speed must be minimized, such as for ultra-low-resolution thermal models, the mesh should contain as large as facets as possible, given all of the above factors and constraints. These large facets should encompass high conductance, high thermal mass, and uniform thermal property surfaces that are insulated from heat transfer drivers and hence tend to have low temperature gradients across their surfaces.

The resulting mesh must satisfy all the general properties of a good WinTherm / RadTherm / MuSES mesh:

• All adjacent polygons share common vertices (equivalence mesh).
• All polygons are 3 or 4-sided (triangles or quads).
• All polygons are convex.
• All polygons have an aspect ratio near unity (e.g., no long and skinny polygons).
• Polygons are spread uniformly across the surface (e.g., avoid fans of polygons).
• No overlapping or repeated facets.
• Surface mesh only (e.g., thin plates represented by their exterior surface only).
• Mesh elements are grouped into meaningful parts.

Future Research

ThermoAnalytics actively researches new meshing algorithms and techniques. At the forefront of this investigation is the drive to automate the process of creating thermal meshes and models. Since thermal meshes are derived not only from geometry but also from consideration of heat transfer features and boundary conditions, it is difficult to fully automate the process. The current thermal mesh building process is a man-in-the-loop procedure in which modelers use a variety of automatic meshing and gridding tools to construct the thermal mesh. There are primarily two possible avenues for more fully automating the meshing process. The first technique involves the generation of high- or ultra-high-resolution meshes only. In this technique a body-fitted, quad-dominated mesh is created through subdivisions of automatically generated tetrahedrons. CEI's prototype HARPOON code utilizes this technique. The second technique starts with an ultra-low or low-resolution thermal model. The low-resolution mesh must be of high quality, in that it must accurately model important heat transfer features (linkages to internal heat sources, boundaries of paints, material types, and surface optical properties, etc.). Through iteration, the thermal model is analyzed and the low-resolution mesh is refined in regions of high thermal gradients, thus producing medium- and high-resolution thermal models. Note that neither avenue will be completely automated since the creation of thermal models requires a man-in-the-loop to make modeling decisions and assumptions. In the first approach, the man-in-the-loop operations occur at the end of the process to turn the high-resolution mesh into a high-resolution model. In the second approach, the creation of the low-resolution model at the start of the process- requires manual intervention.

Learn More About Meshing

1998 Archive: Review of Codes to Create Polygon Meshes

There are many codes that will create a polygon mesh. Most of these will not create a mesh suitable for thermal analysis. The following is a review of some of these tools. We do not endorse any specific code; rather, we supply this information as a guide.

• Read more