# Dynamic meshing - UDF to define bending of tube

• 458 Views
• Last Post 01 March 2018
wbmckinney posted this 27 February 2018

I have a circular tube that is approximately 150 mm long and the last 15 mm of the tube (the tip) will deflect 30 degrees to -30 degrees. I have a velocity inlet and a pressure outlet. I want to simulate fluid flowing through the bending tip. The tip motion is to be defined from an external device and is not due to fluid forces.

I have compiled a UDF in Fluent to assign SHM (DEFINE_CG_MOTION) to the tip boundaries as a rigid body. However, this uses a cosine motion of oscillation that produces an 'elbow' instead of a 'bend' of the tip. Is there a UDF that may be used to help define this bending motion of the tip? Any suggestions or advice would be greatly appreciated.

I have attached a picture detailing how the tip will displace during one oscillation. The red dotted lines indicate where the bending will take place.

Thanks,

Will McKinney

Attached Files

peteroznewman posted this 27 February 2018

I don't know how to write a UDF, but if you want to bend the 15 mm long end, why is all the bending confined to a short elbow instead of the entire length of the 15 mm end?

You want a UDF that creates an equation of a circle as a function of the angle.

wbmckinney posted this 28 February 2018

The bending is confined to a short portion of the tip because that is one of my constraints as it is represented as an embedded flexible joint. Relating radius of curvature of this portion to tip angle may be a possibility to create this motion. However, I am unsure how that would be defined in a UDF.

peteroznewman posted this 28 February 2018

The foreshortening of the tube as it bends should be added to this equation. If I figure that out, I will add it below.

A UDF will be much more efficient, but is it possible to have a Static Structural compute the shape of the 15 mm tip then have Fluent compute the flow?

wbmckinney posted this 28 February 2018

I am not familiar with Static Structural so I am not sure about it computing the shape of the tip. Would I be able compute the flow through the moving tip using this method? I agree a UDF will be more efficient if it is possible to define this type of bending motion.

peteroznewman posted this 28 February 2018

I am very familiar with Static Structural, which was used to mesh that 15 mm long tube and rotate the tip 15 degrees. I believe ANSYS supports 2-way coupling where at each time step, the input angle can be given to Static Structural, which computes a deflected shape, and this shape is given to Fluent to compute the flow. I just don't know how to put them together. The solution to that simple structural model only takes seconds on each time step, but the communication and remeshing in Fluent will take much longer. I could also create a Transient Structural model that is time based and have the tube deform +/- 15 degrees following a sinusoidal function of angle with time as the input.

It would be best if the equation I show above that describes a circular arc that has a variable radius that is a function of the tip angle could be programmed into the UDF. I just don't know how to program that.

wbmckinney posted this 01 March 2018

Static Structural may be the way I have to go if a UDF will not work. Can a Transient Structural model be coupled with Fluent also?

The UDF I have used (DEFINE_CG_MOTION) was used to specify the motion of the tip by providing Fluent with linear and angular velocities at every time step. A cosine function was used to help define the angular velocity in 1 direction. I have attached an animation of this motion in one of my simulations.

However, I do not think I can define the bending motion of the tip I am trying to obtain with just being able to define linear and angular velocity to update node positions on the dynamic zone based on solid-body motion.

I have looked into other UDF’s that could possibly be used:

DEFINE_GEOM can be used to specify the geometry of a deforming zone. Node motion along a planar or cylindrical surface can be defined with time step and position.

DEFINE_GRID_MOTION allows you to control the motion of each node independently. For example, this UDF can update the position of each node based on the deflection due to fluid-structure interaction.

Let me know if anyone has any thoughts as to what may be a good approach.

Regards,

Will McKinney