Ansys HFSS Electromagnetics Simulation Out of Memory

  • Last Post 21 November 2018
kjohnson10 posted this 13 November 2018

Good Afternoon,

I am running an antenna simulation much smaller than my typical simulation size (so I think), but I keep getting an error saying that I am running out of memory. 


I have attached images for all the information I deemed pertinent for troubleshooting this issue.

If anyone could assist me in solving this issue, I would certainly appreciate it.



Kerry Johnson 



Order By: Standard | Newest | Votes
Peter Serano posted this 13 November 2018

Hi Kerry,


What would be more helpful is the 'Profile' and 'Mesh Statistics' information which can be found by right clicking on 'Results' in the project tree and clicking 'Solution Data'. Can you please post this information?



Peter Serano - Senior Application Engineer - High Frequency Products

kjohnson10 posted this 15 November 2018

I got my simulation to run by upping the number of cores allowable for the simulation.


Aside from that, the only change I made was that I made my open region at the central frequency of my sweep. However, this drastically changed my results for the worse. is there a rule of thumb for implementing open regions for radiation boundaries?

Peter Serano posted this 21 November 2018

Hi Kerry,

The radiation boundary should be at least lambda/2 away from any outer surface of your antenna (with lambda being based on the lowest frequency point in your sweep).

Instead of using an open region, I would recommend creating a box defined as vacuum with equal dimensions 'rad_box_size' equal to [lambda (at lowest freq point) + max antenna dimension (from the origin)]. Then set the X, Y, and Z lengths of the box to 'rad_box_size' and center it at coordinate '-rad_box_size/2,-rad_box_size/2,-rad_box_size/2' so that the box is centered around the isocenter (where your antenna should be placed).

After you make the box, press 'f' to go into face select mode and select all 6 faces of the box you just created and define the faces as a Radiation Boundary.

You can also now setup a parametric sweep of the variable 'rad_box_size' and see how it effects your results. You should see the sweeps start to converge to the same result as you increase the size of the box towards lambda.


Peter Serano - Senior Application Engineer - High Frequency Products