Learning Forum

[LorenzH]

I have a problem when trying to create the topology with the ATM Method in Turbogrid. It works perfectly fine, if I exclude the Splitterblades, but fails to find the right topology when included them.

I get always the same error message:

“An error was thrown generating the span geometry for MID. This probably means the selected topology is not appropriate. Details are given below: U value outside of its bounds. Value = -0.00343816”

The value changes slightly, when modifying the radii for the round off of the blades, or when increasing the number of points along the blade or the LE and TE but never vanishes.

I had constructed my radial inflow turbine with the Ansys DesignModeler. I tried even with the BladeGenerator a similar blade and export it to the Turbogrid but with little more success.

The blades are defined as followed:

*Number of Blades: 10 (+10 Splitter Blades)

*Trailing and Leading edge of both Main and Splitterblades are rounded off

[rfblumen]

Hi Lorenz, The problem is that the splitter topology is set up for a centrifugal pump and not a radial inflow turbine. If you LMB on Topology Set and click on the Topology Viewer tab, TurboGrid shows you the geometry it's expecting for this topology (see the image below):nn It's assuming that:n-The full blade is at the low theta positionn-The full and splitter blades have their trailing edge at the same radius. The location of the splitter leading edge is downstream of the full blade.nTo fix this case in DM:n-Under FlowPath2, swap the locations for the Inlet Contour and Outlet Contour. n-Under Blade3 and Spliiter2, swap the locations of the leading edge and trailing edgennRobn

[LorenzH]

Hello Rob, nthank you for the quick reply. I found the constrains in the UserGuide just moments after writing my post. I will try your proposed idea. nRegarding the meshing process in Turbogrid, is it possible to mesh a stator that consists of arbitrary blades such as a NACA profile? Or does the cad file must consists of a blade with thickness and theta distribution? If not possible, the best solution for a similar mesh would be the use of ICEM Blocking, or do you have a different suggestion regarding this. nGreetingsnLorenzn