Farfield Size and Shape

tcarrigan · August 7, 2014, 09:17

I had an someone ask me about my best practices for defining a farfield. What's a good size? What's a good shape? I thought I'd share my response with the community.

Depending on whether you are generating a structured or unstructured grid, both the shape and the size of the farfield domain can have an effect on the solution.

First, the size. I've found that for an airfoil or a wing you'll want to have a farfield that has a radius of at least 100 reference chord lengths. This will help ensure that the flow field instability caused by the wing does not propagate beyond the farfield boundary. If you take a simple airfoil example and compute the coefficient of lift while changing the farfield size, you'll notice it changing to a certain point at which any changes are negligible.

Second, the topology. The topology can often be dictated by the choice in mesh type, structured or unstructured. If you're creating a structured domain, it's relatively easy to generate an o-type or c-type topology. I've found the o-type to be very easy to generate and good at computing the coefficient of lift. However, it's not that great for computing the coefficient of drag due to the lack of wake resolution (see this video, http://youtu.be/SwLkbrZMYMo). A c-type grid is much better at computing the coefficient of drag due to the topology's inherent ability to capture the wake.

Lastly, the shape. Similar to my comments above, the shape is also often dictated by the choice in the mesh type. If a structured mesh is desired, you'll find creating a hemispherical farfield or a c-type farfield more natural. For unstructured, you can create any farfield shape you desire. I've traditionally used hemispherical farfields because they are easy to construct, have a constant radius, and it's much easier to specify the boundary conditions for the solver. Most solvers have a farfield type boundary condition that you can assign to the entire farfield domain.

Vamsi · April 15, 2016, 19:29

Hi.
Could you tell what is the meaning of farfield and farfield resolution?

August 7, 2014, 09:17	Farfield Size and Shape	#1
tcarrigan Senior Member Travis Carrigan Join Date: Jul 2010 Location: Arlington, TX Posts: 161 Rep Power: 16	I had an someone ask me about my best practices for defining a farfield. What's a good size? What's a good shape? I thought I'd share my response with the community. Depending on whether you are generating a structured or unstructured grid, both the shape and the size of the farfield domain can have an effect on the solution. First, the size. I've found that for an airfoil or a wing you'll want to have a farfield that has a radius of at least 100 reference chord lengths. This will help ensure that the flow field instability caused by the wing does not propagate beyond the farfield boundary. If you take a simple airfoil example and compute the coefficient of lift while changing the farfield size, you'll notice it changing to a certain point at which any changes are negligible. Second, the topology. The topology can often be dictated by the choice in mesh type, structured or unstructured. If you're creating a structured domain, it's relatively easy to generate an o-type or c-type topology. I've found the o-type to be very easy to generate and good at computing the coefficient of lift. However, it's not that great for computing the coefficient of drag due to the lack of wake resolution (see this video, http://youtu.be/SwLkbrZMYMo). A c-type grid is much better at computing the coefficient of drag due to the topology's inherent ability to capture the wake. Lastly, the shape. Similar to my comments above, the shape is also often dictated by the choice in the mesh type. If a structured mesh is desired, you'll find creating a hemispherical farfield or a c-type farfield more natural. For unstructured, you can create any farfield shape you desire. I've traditionally used hemispherical farfields because they are easy to construct, have a constant radius, and it's much easier to specify the boundary conditions for the solver. Most solvers have a farfield type boundary condition that you can assign to the entire farfield domain. __________________ Travis Carrigan Manager, Business Development Pointwise, Inc.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Optimum Size of Rotating Domain in MRF	vismech	STAR-CCM+	5	April 24, 2020 00:15
[ANSYS Meshing] Help with element size	sandri_92	ANSYS Meshing & Geometry	14	November 14, 2018 07:54
domain size for 3d external aerodynamics	edi	FLUENT	3	January 1, 2018 03:33
Size functions and boundary layer with Gambit 2.2	Ramine	FLUENT	4	August 1, 2006 08:54
*VARIABLE CELL SIZE IN LES	p.cooper	Main CFD Forum	1	July 28, 2002 02:27

April 15, 2016, 19:29		#2
Vamsi New Member VAMSI YERRAGURAVAGARI Join Date: Mar 2016 Location: Houston-TX Posts: 16 Rep Power: 10	Hi. Could you tell what is the meaning of farfield and farfield resolution?