2

Note the water wheel (selected) failing to spin

Apologies if this doesn't light correctly, I omitted the HDRI I was using to save space for upload and download of the .blend. Plug in your HDRI of choice and the scene should look fine (or turn off HDRI and light as you desire!).

I'm trying to turn a physics object through contact with the Mantaray liquid simulation and it's not working.

The simulation particle system is set to render as a simple object with a sphere rigidbody with sphere collision.

The domain also has a passive rigidbody and collision system - added this out of desperation, and it doesn't seem to be doing much.

The water wheel is a Fluid effector set to collision.

The water wheel also has an active rigidbody and collision system set to mesh - which doesn't seem to be doing much.

The wheel attached to the pole behind it with an empty set to Generic physics constraint (having some problems getting that to rotate reliably on the Y axis, but that's not the main problem).

Can anybody tell me how to get this waterwheel spinning with Mantaflow?

Reverend Speed
  • 1,109
  • 12
  • 25
  • I'm trying to turn a physics object through contact with Mantaray liquid. Do you have any resources, which say that this would work? – Leander May 06 '20 at 12:49
  • Uh, I encountered a couple of videos that seemed to demonstrate this happening. I'll try to track them down for you. Have I been barking up the wrong tree with this? I'd admit, after working with the Effectors set to Collision for a bit, I figured this action would be available. If you can definitively correct me on this, I'd really appreciate it! – Reverend Speed May 06 '20 at 12:57
  • It would be knew to me, that doesn't mean no one has figured out a way to do it. – Leander May 06 '20 at 12:59
  • Belatedly, I note that on of my references is faked, as mentioned in the comments: https://www.youtube.com/watch?v=RVlXncgtQus Leander, I don't suppose I could ask you a related question? https://blenderartists.org/t/mantaflow-liquids-effector-collision-friction-and-viscosity-why-is-it-soup/1225713 – Reverend Speed May 06 '20 at 13:00
  • I don't have a blenderartists account right now, sorry. – Leander May 06 '20 at 13:06
  • More faking in that answe.r – Leander May 06 '20 at 13:07
  • Here's an example of driving the animation purely through physics, but the result isn't amazing (and it's somewhat lacking in instruction). https://blender.stackexchange.com/questions/104582/can-blender-2-79-fluid-physics-act-on-rigid-body-objects This is what I was trying to emulate by turning the particles into rigidbodies. – Reverend Speed May 06 '20 at 13:08
  • In his youtube description, he writes, that he is using the cubesurfer addon. – Leander May 06 '20 at 13:17
  • Gah. Okay, I'm going to leave this open for a bit in case anybody else has something useful to contribute, but after a day, feel free to answer this as "it's not possible" and I'll mark your answer correct. =( – Reverend Speed May 06 '20 at 13:19
  • Don't worry, question may stay open for as many months as necessary. Hopefully someone has a cool solution. – Leander May 06 '20 at 13:29
  • Thanks @Leander. I'll delete this comment in a few minutes, but I don't suppose you'd have any thoughts on this -> https://blender.stackexchange.com/questions/177275/how-can-i-reduce-the-friction-of-mantaflow-colliding-with-effectors – Reverend Speed May 06 '20 at 13:31
  • In general, true fluid-structure-interaction is quite difficult to simulate. So much that it is even a challenge in dedicated CFD-software. So I guess your only chances are either faking it or using something like openFOAM... – haarigertroll May 06 '20 at 15:31
  • Aha, great info @haarigertroll ! Could I ask you to give me some conclusive reference on this so I can close the question? – Reverend Speed May 06 '20 at 16:50
  • I am afraid that this is rather "anecdotal evidence" from my work experience as an engineer. For example, some years ago I tried to simulate a certain type of passive flow-regulating hydraulic valve. I managed to get the mechanical behavior and the fluid dynamic behavior modeled with standard tools pretty well, but not the coupling. So in the end, I solved it iteratively by exchanging the results between the two simulations and re-running them several times. – haarigertroll May 07 '20 at 05:37
  • Out of interest, I searched a bit myself... Maybe this is of interest? https://www.research.manchester.ac.uk/portal/files/70404430/FULL_TEXT.PDF – haarigertroll May 07 '20 at 05:47
  • Oh goodness. Well, that definitely suggests that a solution for this is currently above my paygrade. Thank you @haarigertroll - would you care to submit your comments as an answer? – Reverend Speed May 07 '20 at 08:08
  • While getting the water to turn the water wheel may be challenge (I would simulate it turning starting when the water hits it and as it can be a collision object it can affect the water!), here are some other pointers that would move you closer.
    • Effector: adj Surface Thickness to >= 0.5 (bug workaround) & Sampling Substeps >0, Use Effector
    • WWheel: Sampling Substeps >0, Use Effector Surface Thickness to >= 0.5
     – james_t Jan 29 '21 at 22:12

3 Answers3

1

Bi-directional fluid-structure interaction is quite an advanced topic. (Bi-directional meaning that the fluid reacts to forces applied by the structural domain and vice versa)

There is of course specialized software which can perform these simulations. However, most current simulation packages focus on simulating either the fluid or the rigid (or deformable) bodies accurately and model the other as simply as possible.

That's why in almost all those nice 3D visualisations of fluid-structure-interaction (at least outside the scientific or engineering world) the motion of the rigid (or soft) bodies will most certainly be faked with non-simulated animation.

Further reading e.g: research.manchester.ac.uk/portal/files/70404430/FULL_TEXT.PDF

haarigertroll
  • 1,157
  • 7
  • 12
1

Here is another example of an "active" water wheel that is affected by the water particles. As noted by others, much tuning would be required to make the weight and velocity of the water affect the wheel, include the friction and weight of the wheel itself. But this is proof of concept that this can be done in blender.

enter image description here

james_t
  • 5,446
  • 8
  • 29
0

Here is a modification to your blend file where I:

  • most important: modify the effector to overcome a problem in fluid collision requiring > 0.5 Surface Thickness (reported)
  • enable "Use Effector" for effector, WWheel
  • I moved the effector and Flow closer to the WWheel to shorten my simulation time
  • Per some comments above, I cheated to simulation the rotation of the WWheel, yet used it as a collision obj on the Flow
  • changed the cache location to be relative so as to work for me.
  • disable Mesh for my convienence

enter image description here

james_t
  • 5,446
  • 8
  • 29