Multiphase modelling (VOF) of solidification and melting

  • Last Post 03 April 2020
cfdstar posted this 28 March 2019

The problem that I am dealing with is related to simulating multiphase melting and re-solidification of a moving body using ANSYS Fluent. 


The physical problem is a solid slab which is moving at a constant speed and is heated locally above its melting point with a stationary heat source. The heat source is applied by defining an energy source term at the material-air interface. I used the VOF model to free surface deformations of the molten region. Figure 1 shows a similar problem simulated in other software by others.

Figure 1


To numerically model this problem in ANSYS Fluent, I made a case and employed the “velocity inlet” boundary condition at the inlet to determine the slab and air velocities and an “outflow” boundary condition at the outlet. However, after running the simulation I found that the solidified region close to the boundaries moves and deforms, which is physically unrealistic (see figure 2).

Figure 2


Any heads up, or hints on how to model this problem in ANSYS Fluent using a moving reference frame (i.e. without modelling the source-term movement explicitly) are highly appreciated.

Order By: Standard | Newest | Votes
rwoolhou posted this 29 March 2019

Don't use an outflow bc: they're not suitable for VOF (or anything much now). If you're setting a velocity boundary Fluent must push that much material into the domain: check the pressure in your model as it could be fairly high. 

cfdstar posted this 29 March 2019

Thanks for your reply.

What would be a good setup for the boundary conditions in this problem?

What is known in this problem is the velocity of the workpiece at the inlet boundary and we do not know what is the surface morphology when it leaves the domain to prescribe a velocity. The value of the pressure and its distribution is also unknown since the solidification model for all relative velocities to be zero in the solid region at the outlet. 

rwoolhou posted this 01 April 2019

Not sure without going into a lot of detail which I can't on a public forum. 

One option would be to fix the solid and move the heat source: that neatly avoids the problem? 

cfdstar posted this 01 April 2019

The intention is to avoid using the moving heat source since it will eventually increase the computational costs. 

cfdstar posted this 01 April 2019

Is there a private forum for ANSYS customer? how can I ask this question in the private forum?


rwoolhou posted this 02 April 2019

We have the Customer Portal, via for commercial customers. This is not open to students, but your supervisor (or department expert) may be able to log a query. 

luigilusini posted this 03 April 2020

Dear cfdstar! 

I have the same task, but unfortunately I could not make the model. Can you help me?

Thanks for advance...


abenhadj posted this 03 April 2020

I would go with having a moving heat source per UDF.

Best regards, Amine