SFGRAD (mapdl.sfgrad) vs SF (mapdl.sf)

[VHB4910]

I wanted to discuss about these two options to apply Neumann Boundary Conditions on surface areas. This is my understanding for a steady state heat conduction problem:

1. SFGRAD: I could not quite understand from the manual what this does but here is my guess:
   This is the heatflux vector $q = (q_1,q_2,q_3)^T = \kappa \nabla(T)$. So the input is going to be this:

mapdl.sfgrad("HFLUX", 0, "X", q_1)
mapdl.sfgrad("HFLUX", 0, "Y", q_2)
mapdl.sfgrad("HFLUX", 0, "Z", q_3)

2. SF: This is the magnitude of the heatflux vector $q_{mag} = \sqrt(q_1^2 + q_2^2+ q_3^2)$ and so the input is going to be:

mapdl.sf("Nlist","HFLUX", q_mag)

Am I thinking it right?

By trial, I see that this works if I need to apply the HFLUX as a vector :

mapdl.sf("Nlist","HFLUX", q_1,q_2,q_3)

Although I am not sure if q_2 and q)3 are even being taken as inputs my PyMAPDL?

[mikerife]

Hi @VHB4910
SF and SFGRAD work together. Let's set aside SFGRAD for now.

SF defines only one magnitude of heat flux. The direction is found by the set of nodes ("nlist"). A positive value of heat flux is into the material, and a negative value is out of the material. Here is a small model of a 2x2 square meshed with 4 quad elements. The gplot is showing the lines of the square with the mesh divisions and the nodes with the node numbers turned on:

Now if we issue the commands:

mapdl.nsel('s', 'node', '', 2)
mapdl.nsel('a', 'node', '', 4, 7)
mapdl.cm('node_1', 'node')
mapdl.allsel()
mapdl.sf('node_1', 'hflux', 10)

And issue the commands to turn on surface load symbols as arrows we get the image above. As you can see the two element edges on the 'top' have heat flux in the -Y direction but the two edges on the right side have heat flux in the -X direction.

Now we can add a gradient to the heat flux. Say we wanted a gradient in the global X direction of 10 per unit length. And isssue the commands:

mapdl.nsel('s', 'node', '', 2)
mapdl.nsel('a', 'node', '', 4, 7)
mapdl.cm('node_1', 'node')
mapdl.allsel()

mapdl.sfgrad('hflux', 0, 'x', 0, 10)
mapdl.sf('node_1', 'hflux', 10)

We would get the following applied heat flux:

We can see that nodes 4, 6, 7 are along X so their heat flux is a gradient. But nodes 2, 4, 5 have a constant X so those two edges have a constant heat flux.

Since we issue both SF and SFGRAD commands they each contribute to the total value of heat flux. If you want no 'offset' from the SF command then the SF command needs to be applied with a value of 0 for heat flux.

Mike

[VHB4910]

Thanks @mikerife for the explanation.

I am trying to understand SF first:

From your explanation, I understand that SF takes the dot product of the heat flux vector with the surface normal ($\mathbf{q} \cdot \mathbf{n}$) and hence it works for the case of a regular geometry as in your example above. Assume, a surface with a non-constant surface normal as below:

To the irregular surface boundary on top, I want to apply a Heatflux (Neumann) Boundary condition. In this case, ideally I'd want to apply the heat flux vector $\mathbf{q}$ and let Ansys figure out the surface normal $\mathbf{n}(x,y,z)$ and eventually apply $\mathbf{q}\cdot \mathbf{n}$ internally. But I guess, since SF requires the dot product of $\mathbf{q}\cdot \mathbf{n}(x,y,z)$ as input.
a) In this case, what the best way to calculate the surface normal $\mathbf{n}(x,y,z)$?
b) SFGRAD seems also not to be useful in this case, and the gradients of $\mathbf{q} \cdot \mathbf{n}(x,y,z)$ is also not constants in x and y directions.

[mikerife]

@VHB4910
a) we don't need to calculate the surface normal. SF only take a magnitude. The solver will find the surface normal from the 'nlist' nodes and apply the heat flux appropriately.

b) well yeah...I mean it's in the name of the command. Surface Force GRADient.

[mikerife]

@VHB4910 @germa89
I think you are looking for SFCONTROL which is not yet implemented in PyMAPDL. Please look it up in the MAPDL Command Help.
Mike

[VHB4910]

Thanks, Mike. Will take a look