Proposal for hydrodynamic added mass #76
Conversation
Signed-off-by: Louise Poubel <louise@openrobotics.org>
|
For people unfamiliar with hydrodynamics, the term "added mass" is not very clear. Can you use the phrase "hydrodynamic added mass" in titles and when introducing the concept to help with clarifying this? |
|
Thanks Steve, I will work through this and add some clarifying details as you suggest. Trick will be for me to keep it short, suggestions on how general an audience to target are welcome. |
adding the word "hydrodynamic" is probably enough for me at first glance, but I'll add additional questions as they occur to me |
…red things some and used the usual greek letter mu in place of n, changed in the description and the xml example. This is fairly standard nomenclature so I think <mu11>1.0<\mu11> in the xml should be clear. This commit does not address changes in the C++ API, I found the comments there a bit confusing as they refer to the center of mass coordinate system of the body as "an inertia frame", which it usually won't be in the usual meaning of that term. Perhaps I am misunderstanding something, but am intending to find out if this is a comment adapted from previous code and I'm mis-understanding something, or if it's a new comment.
Signed-off-by: Louise Poubel <louise@openrobotics.org>
Signed-off-by: Louise Poubel <louise@openrobotics.org>
Signed-off-by: Louise Poubel <louise@openrobotics.org>
Signed-off-by: Louise Poubel <louise@openrobotics.org>
Signed-off-by: Louise Poubel <louise@openrobotics.org>
|
I pushed some more paragraphs in 5c36c87 explaining the motivation for storing the added mass terms separately from the remaining inertial terms, even though they will be combined when fed to the physics engine. I also updated the target version to Garden, since any changes to |
| While many physics engines may not have direct support for added mass, they often | ||
| support directly setting each element of the resulting \\(\bf{M} + \bf{\mu}\\) | ||
| matrix that's multiplied by acceleration. That's the case for | ||
| [DART](https://dartsim.github.io/dart/main/d6/d91/classdart_1_1dynamics_1_1Inertia.html#aa4661f9950d5958efa4c8609d42aedf7), |
There was a problem hiding this comment.
I was just looking more closely at DART's inertia code, and it looks like it will print warnings if a spatial inertia matrix has nonzero values in off-diagonal parts of the translational 3x3 block-diagonal matrix (xy, xz, yz) or the diagonal parts of the block-off-diagonal 3x3 matrix (xp, yq, zr). For rigid bodies, these terms are all 0, so I think DART will complain a bit about using them for fluid added mass, though it might still work despite the complaints, depending on what assumptions they make when inverting the spatial inertia matrix
There was a problem hiding this comment.
I meant to include a link to the dart code: https://github.com/dartsim/dart/blob/5058255853d5b733476fc031b18fb95bdf7d7f5d/dart/dynamics/Inertia.cpp#L292-L333
There was a problem hiding this comment.
Yeah that's a concern that @hamilton8415 and I have had from the beginning. Hopefully it's just a matter of suppressing those warnings. It would be nice to have a prototype implementation out soon so we can have an idea of what breaks and what doesn't.
There was a problem hiding this comment.
From my initial testing, all we need to do is suppress those warnings, see
I'll check with @hamilton8415 in the coming days if the resulting behavior is what's expected though.
Signed-off-by: Louise Poubel <louise@openrobotics.org>
Signed-off-by: Louise Poubel <louise@openrobotics.org>
|
While working on the implementation I realized that since the added mass coefficients are fluid-dependent, it would be convenient to expose the fluid density as a separate parameter. This way, it's trivial to adapt an SDF file for different fluids. I made the changes for that in 60866d4, and will start working on the implementation. I'd appreciate some reviews, since I haven't seen the added mass matrix broken apart like this anywhere in the literature. I may be doing something that may not hold for every use case, or that will make it difficult to export values from other software directly into an SDF file. CC @andermi |
Signed-off-by: Louise Poubel <louise@openrobotics.org>
|
Yes, some bodies will have empirical added mass matrices from CFD or pool tests, and some easier shapes will have closed-form dependent on fluid density. Perhaps, we could support an explicit matrix and a formulaic one? |
Signed-off-by: Louise Poubel <louise@openrobotics.org>
After talking to @andermi today we figured that wasn't a very good idea, so I removed the separate parameter. |
Signed-off-by: Louise Poubel <louise@openrobotics.org>
|
After landing the implementation PR, we are going to give this one more review to confirm that the implementation and documentation match. |
|
|
||
| Newton's second law describes how forces affect a body's motion in an inertial frame. It can be written as: | ||
|
|
||
| $$ (\bf{M} + \bf{\mu}) \ddot{\bf x} = \sum{F({\bf x}, t)} $$ |
There was a problem hiding this comment.
Note that this equation is only valid for translation; for the rotation of a rigid body, this becomes Euler's equations; and in the presence of cross-terms in the spatial inertia matrix, this becomes equation (2.90) from [1].
I'm looking for early feedback on the proposal. I'm planning to circulate this with the ROS Maritime Working Group for feedback after the initial review.
Rendered preview:
http://sdformat.org/tutorials?tut=added_mass_proposal&branch=chapulina/added_mass