Hi;

I'm an engineer working for an architecture firm. I'm trying to compare Flovent and Airpak, and one obvious difference between the two is the meshing -- Flovent uses a structured cartesian (or rectilinear, to add another technical adjective) mesh, whereas Airpak uses an unstructured mesh.

The buildings we design typically aren't too geometrically complex, but we will frequently have walls that are angled in one plane or broad curves. If I'm not mistaken, these features would require some grid stepping for Flovent to mesh them. I'm curious if anyone had any opinions on the extent, if any, to which this would effect the accuracy/speed of analysis.

The primary reason why I'm concerned is that one of the things we'd like to use CFD to look at is the downdrafts caused by the cold inside surfaces of windows during winter, where I would guess that the quality of the mesh at the cold surface would be important. In one project that we're working on right now, this glazing surface slopes to the outside at 10° from floor to ceiling. If Flovent would have to step its grid to mesh this, I'm worried it could affect the accuracy of solution. Are my worries valid or completely misplaced?

Many thanks,

Josh

Hi Josh:

I am not too sure how FLOVENT handle the non axis-aligned edges. Anyway, if FLOVENT uses the stair-case approach in approximating the complex boundary, accuracy will be undoubtedly affected. This problem will be more severe if one desires to study the local flow (in the vicinity of slanted glazing surface), in detail.

From the post-processor of FLOVENT, apparently, the boundary is represented by steps. However, the numerical treatment at the boundary is all dependent on the FLOVENT solver. In the case of FLOVENT implements some sort of Immersed Boundary techniques in this context, I would say that the accuracy can be improved, as compared to that of the stair-case approach.

-khai ching-