Machine Design

Solid ideas improve fluid simulations

Fluid-flow problems usually involve structures and more frequently, heat transfer.

The brake assembly is modeled and meshed in NX 4. The tree in the menu to the right shows that thermal coupling is set in three places.

The automatic mesher adds more elements to areas that will need them on this close-up of a flow surface.

A menu from NX Thermal shows easy-to-read and select features.

But solving fluidstructure and fluidthermal problems with older FEA codes requires solving the fluid part and somehow mapping its results onto the other. Analysts might also have to tweak the mesh to line up nodes. And two different disciplines could require two different preprocessors, which means a steeper learning curve.

The recent NX 4 Series of CAD software from UGS, Plano, Tex., includes NX Flow for CFD simulations, NX Thermal for FEbased finite-volume heat transfer, and NX Nastran for structural simulations. These modules let engineers more easily simulate flow, thermal, and structural loads.

For example, the software lets users apply the same mesh or a dissimilar mesh for thermal solid surfaces and fluid-flow volumes in CFD problems. The thermofluid iterative solver creates thermal and fluid couplings between the unconnected and dissimilar fluid-to-solid interface meshes at run time. This has several advantages. "First, it lets users examine old flow models with new boundary conditions," says James Osborne, CAE director at UGS. "This is often necessary to upgrade existing designs or to use them as a starting point for new ideas." Although NX Nastran still requires separate setup tools, the thermal and flow spackages improve simulations with a common design database and the same tools for the runready model setup.

Additional features include coupling thermal and structural models to simplify studies that look for stresses caused by thermal loads, such as in an engine-exhaust muffler. A temperature-mapping algorithm treats steady-state or transient temperatures as boundary conditions and accurately places them on the finite-element models that most likely feature different meshes.

Thermal coupling for joining disconnected solid or surface meshes works well on assemblies of different materials and different shapes at their interfaces. "The joining lets engineers create areaproportional conductances between dissimilar element meshes that historically needed complicated mesh-mating connections or CAD modifications," says Senior Engineer Remi Duquette. "Parts that touch in CAD assemblies are thermally disconnected, so the software makes the correct thermal connections at runtime. Coupling characteristics may vary with, for example, temperature and heat loads determined during the solution. The software examines the thermal coupling parameters set by the user and simply computes the proper conductances at runtime between interacting faces which may also feature different materials on either side."

Creating surfaces that convect heat to a fluid simulates convection to or from the fluid on thermal-model surfaces. Solid surfaces and even porous blockages obstructing the fluid trigger the calculation of surfacedrag forces and near-wall effects. "The software solves for conduction and fluid-flow separately," says Osborne. "Flow surfaces can be PC boards, baffles, ducts, chambers, walls, or external complex geometry with or without convection. During simulation, the software establishes heat paths or conductances from the solid surfaces to the adjacent 3D-flow elements. The fluid-mesh nodes need not match the solid-surface mesh nodes, which is a big advantage for thermofluid simulations," he adds.

Connecting disjointed fluid meshes is useful when examining smaller design variations inside larger, unchanging systems. "This technique could be useful when testing different heat sinks, or optimizing the shape of a specific part within an assembly and without remeshing the entire model," says Osborne. The method can quickly swap NX part models to test different designs. A toggle called "Connect disjoint meshes" joins the meshes at runtime. The feature reduces meshing work and model sizes.

Modeling fluid buoyancy and natural convection simulates the effects of gravity on a fluid. "Fluid density affects the total gravitational force on a volume of fluid, so spatial variations in fluid density give rise to socalled buoyancy forces which must be modeled for accurate results," says Osborne. The feature would be useful examining the chimney effect in tall enclosures to more accurately predict temperature differences from top to bottom.

Autoconvect from all surfaces to a fluid avoids identifying the many flow surfaces that have identical properties such as roughness. "These surfaces can convect to fluid elements they adjoin. Similarly, all volumes meshed with nonfluid 3D meshes and not defined as porous-flow blockages convect from their surfaces to fluid elements they adjoin," says Duquette.

Automatic runtime meshing means the fluid domain can be picked as a general volume surrounding all other parts just before kicking off a simulation. Users need only mesh internal components using either 2D mesh on solid surfaces or 3D mesh on solid components within the fluid volume. In addition, many programs require defeaturing a model — removing details that do not influence result accuracy, such as small chamfers and holes. Routines in the mesher can now handle the defeaturing. For example, an engine block that once took a week to defeature and mesh is now ready for analysis in under 3 hr without manual intervention. This feature lets users quickly run multiple "what-if" scenarios involving complex part arrangements within a bounding volume and with multiple part changes. The fluid mesh adapts to the new location of internal part surfaces and volumes and to part-feature changes at run time.

The mesher also creates a boundary layer or skin-layer mesh at user request. "This mesh starts with small elements on a wall and builds the user specified layers of elements away from the surface," says Duquette. "It ensures more accurate calculations of heattransfer coefficients and nearwall effects, such as drag."

UGS, (513) 576-2142,

Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.