Domain for Pipe Flow

Swarup

Dear All,

How can I represent a computational domain for a non-axisymmetric circular c/s cylindrical pipe flow without any swirl? That is, I have an inherently 2D flow & variations in radial & axial directions are only important. Mind you, the third dimension (theta) is finite but of no significance as far as flow pattern is concerned.

The first geometry that comes into mind is a rectangle of height equal to diameter & length equal to the cylinder length. How appropriate is this approximation for the abovementioned pipe flow? Is it adequately representing the physical domain? Same domain can also be built for flow through two parallel plates of infinite z-direction (into & out of the screen).

Any advice will be highly appreciated.

Regards,

Swarup.