Reza Azadi - Fluid-solid interaction (FSI)

Fluid-solid interaction (FSI)

What was the project?

Numerical simulation of a moving object in a fluid flow has been challenging for researchers in the last decades. In this project, the aim was to numerically simulate the motion of a high-speed hull with two degrees of freedom (rolling and sinkage), which was interacting with its surrounding turbulent water and air flows. Therefore, I was dealing with three main flow phenomena:

Fluid-solid interaction,
Turbulence and boundary layer, and
Transient two-phase flow.

I used a 6-degrees of freedom (6-DOF) solver to calculate the rigid body dynamics, i.e., translational and angular displacements of the solid object resulting from the forces and moments applied by the surrounding fluid. As a result, the flow's mesh domain and position of nodes were altered and should be updated at each time step.

What was the challenge, and how did I solve it?

The challenge was that even an infinitesimal displacement of the solid object and the following update of the surrounding flow's mesh cells would collapse the initial boundary layer mesh. Therefore, stabilizing the fluid-solid solver was almost impossible, and the approach would fail. I came up with this trick to solve this problem: I divided the fluid domain into one surrounding the solid object, which moved with the solid object and set the rest of the fluid field stationary. The sensitive boundary layer mesh around the solid object was maintained intact by applying this procedure and made the fluid-solid solver was perfectly stable.

Which approach did I employ?

Applying:

The volume of fluid (VOF) model to track the air-water interface,
SST k-ω model for turbulence,
6-DOF solver for the solid object,
Spring-based smoothing approach to update the mesh domain, and
Volume-based stabilization method to stabilize the fluid-solid two-way coupling,

In Ansys Fluent commercial software, I could successfully simulate the problem in 3D and get results that were in excellent agreement with experimental results. For instance, a simple hull body's estimated resistance was at a maximum of 7% difference from the experimental results.

Animations of the bottom view of one-step (top) and two-step (middle) hulls, and representation of the utilized mesh domain (bottom). Color codes in animations show the velocity magnitude of the water, with a maximum of 6.89 m/s.

Novelty?

The final aim of the project was to use a reliable numerical approach (such as mine!) to enhance the performance of the hulls. The first suggestion to achieve this goal was to reduce the hull's resistance by adding steps to the bottom of the hull. These steps can suck air underneath the hull, reduce the drag and increase the efficacy of the body. However, an appropriate design was vital to avoid the stall of the hull. Up to the point that I was involved in the project, I simulated one-step, two-step, and three-step geometries and showed that we could get up to 38% drag reduction in a two-step hull. However, the steps reduced the hull's maneuverability by 15%, which was unsatisfactory!