CAD Idealization for CFD
With geometry representing the first input of simulation, it is important that CAD Idealization is approached strategically to optimize simulation performance (ability to mesh, reduce solver time, & produce valid results).
While a fully detailed building model is necessary for construction, most of the geometric complexity may not be required when leveraging simulation for obtaining reliable performance insight quickly and efficiently. Achieving this is typically accomplished by starting simple and building in complexity as you successfully complete analyses. The degree of acceptable model fidelity is a function of unique simulation goals.
Example: Outdoor courtyard wind analysis
Understanding air velocity and pressure fields of an outdoor courtyard may drive design decisions such as door type, location, need for an air curtain, or even landscape elements (planting or fences). Modeling the building’s interior details is obviously unnecessary in this situation. Even a portion of the exterior details are unnecessary given the simulation goals; especially for a first pass understanding of the situation.
The CAD model signifies the initial assumptions in simulation. Is a fully detailed door model with molding, windows, hinges, and a knob necessary in a room ventilation study? How about a window with framing? The answer to both of these questions is almost always NO. In most instances these components would be completely eliminated as air flow around a door knob would not have a major impact on the overall air distribution or temperatures.
If a homogeneous wall characterization is inadequate—in a situation where heat transfer through the windows and/or doors should be considered—simple volumes or surfaces can be used to represent their existence geometrically. Material or Boundary condition definitions in Simulation CFD would then be used to define their respective performance. Details for including U factors for typical objects such as doors or windows are described here:
Again, when idealizing CAD for simulation it is important to determine what geometry is necessary to obtain your desired outputs. Starting simple and adding complexity after successfully completing simulations is a great method to make progress and understand the impacts of characterizing geometry.
The setup tasks of simulation (materials, boundary conditions, & meshing) are all dependent on, and applied to, the input geometry (CAD). This implies an importance of characterizing geometry for simulation.
Each volume or surface created in CAD is selectable in Simulation CFD. Selectable items are used to define inputs and extract results; making every geometric entity worth evaluating prior to running a simulation, especially considering that the number of surfaces and edges are a factor in simulation run time.
Example: Number of geometric entities in a staircase
When launching the detailed staircase model into Simulation CFD, the small overhangs are translated into small surfaces and edges which negatively influence mesh generation (ability to generate a mesh without errors and the mesh density or total number of elements).
Reducing the number of surfaces present is paramount to optimizing geometry for simulation. Representing the staircase without the overhangs will improve mesh control and reduce simulation run times without impacting the overall flow fields solved.
Replacing the staircase with a ramp would also be an acceptable means to further reduce geometric complexity.
Example: CAD geometry facilitating a simulation task
The air distribution of a room with several supply diffusers is to be simulated. The supply diffusers throw air in a particular pattern, influencing the room’s air distribution.
While a fully detailed model of the diffuser can be simulated, a simpler geometric representation would reduce analysis complexity and run time.
Even though the louvers create the throw pattern in the real world, they are not necessarily needed in simulation. The diffuser throw pattern in simulation can be approximated with velocity boundary conditions applied to selectable surfaces in the model. Details of this characterization are presented here.
TIP: Reducing model complexity (number of CAD surfaces for example) can improve simulation performance and run times without negatively impacting results validity.
Common CAD Considerations
There are a wide variety of CAD model challenges that present themselves as obstacles to making progress in simulation. The more common issues discussed here will serve as the foundation for approaching idealization of any CAD model, regardless of complexity.
Coincident volumes share a surface imprint in Simulation CFD that does not exist in CAD. Hide a smaller volume that touches a larger one to reveal the shared surface.
TIP: Hiding volumes to look for surface imprints is a great way to determine if volumes actually touch each other in Simulation CFD. Tiny gaps between volumes is a common cause for errors in meshing and solving.
|Two volumes in CAD that touch (left) share a selectable surface which is highlighted (right)|
In the previous diffuser example, several selectable surfaces were required to define boundary conditions. The surfaces could have been created explicitly in CAD or created as several extrusions whose imprints would be used to define boundary conditions.
|Explicitly created surfaces (left) vs. extruded volumes (right)|
Interferences may exist in a CAD file intentionally or accidentally. While Simulation CFD can handle interferences that exist in some CAD platforms (not all), interferences should be resolved. Resolving interferences simply means moving faces or bodies such that they are in perfect alignment (zero distance) or to boolean (cut out) the interference.
Interferences are typically found in CAD by using the CAD platform’s respective geometry tools, visually zooming in on components in close proximity, and/or measuring neighboring components. Simulation CFD will attempt to create new volumes from the intersection of interfering parts.
TIP: New volumes in Simulation CFD created from an interference will have a name structure referencing the interfering volumes with a “_U_” between them, such as part1_U_part2. Reviewing the Simulation CFD design study bar volume names can help find interferences.
Marginal geometry refers to any set of geometry which displays problematic behavior. Marginal geometry manifests itself in the form of an inability to perform CAD functions, meshing errors, solver divergence or software crashes. Marginal geometry (created by accident or intentionally) typically consists of interferences, small faces or edges, tiny gaps, minor component alignment offsets, and various other project specific conditions that are problematic to interpret.
Geometry issues are one of the leading causes of simulation failure. Avoiding marginal geometry conditions through the careful creation of geometry is easier than troubleshooting pre-existing problematic geometry. This is why the theme of “starting simple and adding in complexity over time” is a best practice.
Marginal geometry is unique to each file; however, there is a single condition that is common to most marginal geometry set.
Small edges are responsible for the majority of marginal geometry sets. Small is not an absolute number such as 1mm or 1in. Small is relative to the rest of the geometry. Somewhere in every model the smallest edge exists.
The location of small edges can be achieved with the following techniques:
- Geometry check tools
- Measurement / Analysis
- Visual inspection
- Edge Diagnostics
- Mesh Seeds
Idealizing CAD and optimizing it for simulation can be a daunting task. It can often be the most time consuming and critical aspect of simulation.
Every situation cannot be covered but by sticking to these basic concepts, even the most complex CAD models can be idealized.
The key concepts of CAD idealization and geometry characterization presented here can be summarized are as follows:
- Start simple and then build up complexity
- Determine the minimum level of detail necessary
- Consider what assumptions are acceptable
- Evaluate every geometric entity prior to running a simulation
- Leverage geometry to facilitate simulation tasks