NSF Gives Supercomputer Time For 3-D Model of Spill

CWmike writes "Scientists have embarked on a crash effort to use one the world's largest supercomputers to create 3-D models to simulate how BP's massive Gulf of Mexico oil spill will affect coastal areas. Acting within 24 hours of receiving a request from researchers, the National Science Foundation late last week made an emergency allocation of 1 million compute hours on a supercomputer at the Texas Advanced Computing Center to study how BP's gusher will affect coastlines. The computer model they are working on 'has the potential to advise and undergird many emergency management decisions that may be made along the way, particularly if a hurricane comes through the area,' said Rick Luettich, a professor of marine sciences and head of the Institute of Marine Sciences at the University of North Carolina in Chapel Hill, who is one of the researchers on this project. Meanwhile, geographic information systems vendor ESRI has added a social spin to GIS mapping of the BP oil spill."

Comments

In Time?

[MozeeToby]

Is it really possible to develop, test, and run a complex simulation of the gulf's currents and weather in time for it to be useful for the recovery? It seems to me like the kind of job that could take a couple years to get right, even assuming we know everything we need to make such a simulation. Maybe they want the simulation developed today so that it can be used to evaluate potential future disasters? To help quantify the risk involved in this kind of drilling?

Re:In Time?

[alfredos]

I think the answer is "no". Which takes us back to the summary: The NSF has allocated compute hours. Nothing else. Therefore it's just a piece of the whole thing, and the other pieces (the simulation itself, getting reliable data to feed it -see GIGO-, etc) are orders of magnitude more complex.

Re:In Time?

[Monkeedude1212]

Well, You're actually not too far from the truth. Surprisingly enough, video game engines provide great simulators for these kinds of things.

I know of a few cases where people have licensed the Unreal Engine for creating their own simulators, specifically I think it was to deal with some high collision testing. Since the Unreal team has started with some fluids physics, its not hard to see someone possibly using that engine to simulate the effects.

Then, all thats needed is the writing of Entities, which is usually done in C++, not VB. But the idea is the same, you only need about a weeks worth (40 hours) to write the basic entities you'll need, and maybe a bit longer for the complicated ones. Depends on what you are simulating, but if say there are only 4 different types of matter they have to consider, water, oil, air, dirt, than there really isn't that much to add.

So, really, the GUI would be already made, it'd probably be the Unreal Map Editor. The Code they would be writing would be C++, which most /.ers can at least recognize, and you won't need to make any models, since you are going to be using brushes for the landscape and liquids for the rest.

Not that they had to us Unreal, they also could use the Source Engine, though I don't think source offers much in the lines of liquids. But there are actually many ways to go about this. In fact, some modelling and animation software lets you simulate water physics, one in particular I know but I can't recall the name right now. (A plugin for Maya or 3dsMax or Softimage? Bah).

So really, the only LENGTHY part of development is duplicating the ocean currents, which if you have recent records on, is about as trivial as stitching their co-ordinates to the map you create, and making your entities flexible enough to handle the varied input.

I'm not trying to belittle the task of simulating something like this, but with a team of 10 or so people I could see it being done rather quickly. You have to realize that simulations are never quite 100% like the real thing. They have to cut off at some approximation point to keep the simulation running, otherwise the computer would hang trying to figure out the googelth digit. So, really, because the gulf is so large, its not like they have to calculate everything down to the particle.

Re:In Time?

[Stephen Samuel]

Considering that BP doesn't even bother measuring how much oil is spilling

Well, I'm pretty sure that they have a rough idea, and it's way more than what they're saying in public. They just don't want to formalize their estimates, because then they'll have to report the numbers. I did some napkin (units(1)) calculations based on the volume of dispersant that they say they've been using and, if they're using at the suggested dilution, then at a minimum they're dealing with 60K~500K barrels per day.
((don't have the actual calculations on hand, right now, and I'm on a different computer, so I can't even just look at my command line history))

And that's a minimum.... the volume that they're using may be limited by the supply chain.

I'm guessing that, in internal conversations, they're duck-speaking their way around solid numbers... For example, they can talk about how much dispersant they'll need ( a number based on oil flow), but there's probably an unwritten rule about never mentioning the oil flow estimate that underlies that calculation, because it'll be seekable in disclosure requests, and they'll still be able to 'truthfully' claim that they've never talked about the actual oil flow.

Photos of Louisiana Shores

[Anonymous Coward]

I caught this link a day or two ago. Take a look.

Re:Photos of Louisiana Shores

[Anonymous Coward]

I found the aerial pictures of the oil booms really interesting, thanks AC!

You mean, the way they are fucking done completely fucking wrong.?

Re:Where's Bruce Willis when you need him?

[camperdave]

Well, maybe not a nuke. But how about a large rubberized canvas funnel connected to a hose to the surface?

Oh, and why don't oil rigs have a large containment boom around them in the first place?