[GCAero]

Hi everyone!
I'm trying to simulate the stage separation of a two-stage rocket using the transient module in Fluent. The objective was to simulate the whole separation process, but I wasn't able to create an initial mesh that would work in the dynamic meshing module while having both stages still joined together. To bypass this, the simulation starts with a 1mm gap between stages, so there's already some mesh for the remeshing tools to start with (I'd like to solve this issue later on). The rocket (with approximate dimensions of D=0.098 m and L=3.3 m) is put inside a cylindrical enclosure of r=4 m and h=14 m.
To create the mesh, I've used tetrahedrons and several sizing methods, as well as a sphere of influence sizing to refine the mesh around the 1mm gap (it allows for about 2 elements between the stages).
The setup is a transient density-based simulation with energy and viscous (Transition SST) models. The boundary conditions are set as: booster (wall, no slip), upper-stage (wall, no slip), cylindrical enclosure (pressure far field).
All 3 methods for dynamic meshing are enabled (Smoothing, Layering and Remeshing) and configured as follows:
- Smoothing is set to diffusion method.
- Layering is height-based, with a split factor of 0.4 and collapse factor of 0.2
- Remeshing uses local cell and local face methods with no sizing function and the default parameters taken from the mesh scale info.
From the dynamic mesh options I've activated 6DOF and Implicit Update. In the 6DOF settings, the mass and inertia tensor of the booster are introduced and saved as a property. As for the Implicit Update settings, I've been trying different values for the parameters to see how it affected convergence, but right now they're set to: Update Interval=3, Motion Relaxation=0.9, Residual Criteria=1e-5.
I've created a single dynamic mesh zone for the booster, defined it as a rigid body and assigned to it the property created previously. Solution Stabilization is enabled and set to a coefficient-based scale factor of 0.5; just as with Implicit Update, I've tried different values but this is the current one.
I haven't been able to preview the mesh motion (it shows the booster floating, but no mesh and no movement).
Hybrid Initialization is achieved within 10 iterations (constant pressure, iter.10 → 6.181946e-08).
Since the goal is to simulate 2 s of stage separation, the Run Calculation setup used is:
- Fixed and User-Specified time advancement.
- Number of time steps=20000
- Time step size=0.0001
- Max. iterations/time step=40 (also used 80)

As mentioned before, I have run several cases varying implicit update and solution stabilization parameters, and recorded the resulting error for each. I've also tried reducing the time step size and increasing the max number of iterations/time step. This is what I have so far:
results.png
For the first case I did get results up to about 0.016 s (after more than a day running on 20 processes) before it started diverging, so I used it as reference.

Since it's the first time I run a transient simulation of this caliber, it's quite possible I'm misusing some tools or estimating values for certain parameters that are off so, if this is the case, please let me know.
At this point, I don't really know what else to try so if anyone has some insight on how I could proceed, it will be much appreciated. Thanks in advance!

Best regards,
G



PS: This is my first thread, please tell me if there's anything I haven't explained well enough, if there are issues with the format,...