Help with overriding drag coefficients

[Project-Evoris]

Hi,

Firstly, thank you for developing such a great tool.
Currently I am working on simulating a AUV for researching control algorithms.
I am wanting to override the automatically calculated drag coefficients.
But it is unclear from the documentation and the paper how these are calculated in stonefish.

Specifically when calculating the drag coefficients, is the reference area used the wetted area or the projected area.

Thank you in advance, any insight you can provide would be greatly appreciated.

[patrykcieslak]

Hello!

Stonefish uses a hybrid approach. On one side it estimates the drag coefficients based on an ellipsoid approximation of the geometry and on the other it uses the actual geometry (phyiscal meshes) to compute the projected area.
However, to allow for more realism in the behaviour of the robot, the projection part is taking into account local water velocity at each triangle. If the robot is moving in a straight line and there is no currents defined, this should simplify to the typical drag formula. If the robot is rotating or some currents are defined, the water velocity acting on different triangles is not the same, so the computation is not following the typical model, but is providing more of an emprirical solution. The value of the drag force and torque in this case may be far from reality but the behavour of the robots is much more reallistic compared to a simple point mass model used by other simulators. For example in my simulator a robot entering a current will not just move sideways but will also rotate because the current will first hit the front of the robot and gradually affect the whole body.

The automatically estimated drag coefficients can be overriden defining:
<hydrodynamics viscous_drag="0.05 0.08 0.08" quadratic_drag="0.2 0.6 0.6"/>
inside dynamic body or link definitions.

To be honest simulating hydrodynamics reallistically in realtime is beyond any current technology and papers on the topic show that mathematical models lead to 50% errors when comparing with experimental results so for me it was never a goal to get right the values of the forces but to get a behaviour of the robots that shows similar effects as in reality.

I hope this answers your doubts :)