Lahar project - Fluid sim

Hi all,
I’m working on modelling the flow of a lahar (volcanic mud flow) down the sides of a model I imported from a DEM file. Basically its an accurate volcano model and I am trying to visualise where these mudflows should go (but mostly don’t!:smiley: )

The fluid is a bit patchy in places and has a tendancy to form blobs and stick rather than flow down the slope. Ideas?

Resolution. It would require at least 300 for the fluidsim to look decent.

Says mpan3, the king of fluid sims!

Hmm, bit of a problem so, the iMac I am using (in college) is maxing out at a resolution of 230. It stalls at 235 or above.
The prospect of getting a better machine or upgrade is highly unlikely.
I was looking into distributed rendering, but does that work for the baking process?
(Im also scared as to what the admin is going to say about putting a distributed system on the network! :eek: )

Oops, Just to be sure, I went back and tested the 300 resolution again. Sure enough Blender stalled, application not responding, BUT after 5 minutes it did actually work. Hooray!
I will post the new pics if and when it finishes baking…

Might not look like much of a change, but the behaviour of the fluid is much better over the terrain in general. Thanks guys!

The above picture was baked at 300? Baking is a highly memory intensive process and you’ll need as much as 1GB for 250 resolution. Otherwise harddrive thrashing will occur, slowing down the baking process significantly.

I have a feeling your domain size is not optimized. Try to reduce the bounding box(domain) to as smaller as possible.

ok, well, I need to ask this as long as we’re on this subject. how long does it take for a high rez bake with fluid sim? If I use just a rez > 120, it can take days and still doesn’t end up right. what size is the smallest for an accurate fluid sim? is there anything else to do to get a higher speed?

ok, since there seems to be some confusion surrounding the fluid issue, i’ll try to address them here:

The key factor that affects the quality of the output and speed is resolution. The higher the resolution, the better the quality but slower. Personally, I usually lower ‘preview’ quality to 1 and just set the ‘Disp Qual’ to final/final. THis way, i can see exactly what I will get when I render. This also reduce the harddrive usage since the preview geometry files will be almost zero-btye.

The easiest way to get a more detailed output without increasing the baking time is to play arund with the bounding box (aka domain) size, there is no reason why you want the domain to be bigger than absolutely necessary. I usually resize the box so that it almost-hugs the fluid on all side. This has no impact on baking time but give you higher quality results.

fluid is more dependent on memroy size than CPU. Although the estimates could be way off sometimes, in general 512mb of system ram will allow you to bake up to 250. 1GB to 350, and 2GB to 450… And since the fluid engine is not multi-threaded, it is not possible to utilize dual-core processors or some sort of ‘render farm’.

Make your real world domain size 10 meters, the maximum that blender allows.

finally I get it, but LOTRJ, why such a large domain? wouldn’t a smaller one be better?
@mpan3 - thank you so much, you saved me 5 days of computer useage!

to help answer LotRJ’s question, a 10 meter domain setting means setting the ‘realworld size’ of the domain box to be 1000 cubic meters (10x10x10 meters). This will make the fluid behave like waters in a small swimming pool, which has roughly the same size.

the default size of 0.0001 meters^3 is only good for a glass of water, at the setting, no matter how much you crank up the baking resolution, the fluid won’t behave like waters in an ocean…

Are we good now?

pretty much, I know what the real world size does, I asked about using a smaller domain for the purpose of baking times. so, a smaller size domain is quicker…right?:confused:
thanks again for helping the fluid-sim-challenged:D

The fluid is a bit patchy in places and has a tendancy to form blobs and stick rather than flow down the slope. Ideas?

Try raising the slip value of the domain

I just suggested a large real world domain size because the area he is trying to simulate is supposed to be covering a mountain side. The closest scale blender can get to this is a 10 x 10 x 10 meter cube.

I really shoud have done this sooner :o

Wow, you do need to give you fluid a lot more velocity to compensate for the greater size!

Thanks for all your great support as always!

And as a side note for Lord Of The Rings Junkie, this is Mt. Ruapehu (aka Emyn Muil) :cool:

Hi again all.

Struck a bit of a problem with my fluid.

The fluid should be moving fast but any time I have the speed up high I start to get effects like this:

“flying blobs”
Any ideas, suggestions or work arounds for this problem?

hmm, do you mind sharing the blender file? You can host it here for free:

Sweeeeeeeet…:smiley: You have just GOT to love the New Zealand landscape.

[Edit] Since you’re trying to simulate a pyroclastic flow, you might want to try to match the velocity that they move at, which is about 100-150 km/hour.

Right so!

Blender File

10.5 MB