DOE Supercomputers

Point out news stories, on the net or in mainstream media, related to polywell fusion.

Moderators: tonybarry, MSimon

Post Reply
MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

DOE Supercomputers

Post by MSimon »

*

http://www.sciencedaily.com/releases/20 ... 164126.htm

*

Who gets to use them?
physicists from General Atomics will continue their examination of turbulence in fusion reactors.

The 2008 allocations reflect an aggressive program of upgrades to the Cray XT4 Jaguar supercomputer, which is in the midst of the second of two major upgrades in a year. When the current project is complete, Jaguar's 31,000 processing cores will be capable of 275 trillion calculations a second, or 275 teraflops—a fourfold increase over a year ago.

Jaguar's increased processing power is reflected in the size of the allocations being made available to individual projects. Whereas the largest allocation in 2007 was 10 million processor hours, 2008 will see six separate projects with at least 10 million hours, and the largest allocation is 18 million processor hours. The 2008 INCITE program and the large allocations enabled by Oak Ridge's NCCS Leadership Computing Facility will give researchers an invaluable opportunity to continue pushing the boundaries of knowledge, and their efforts promise to improve both our lives and our understanding of the world we live in.
Engineering is the art of making what you want from what you can get at a profit.

drmike
Posts: 825
Joined: Sat Jul 14, 2007 11:54 pm
Contact:

Post by drmike »

It was mostly for nuclear weapons research. I think the big boys are playing with shaped nuclear charges and I would bet a lot of cycles go to that. "High density matter" is what it is usually called. Fun stuff, in terms of pure physics.

scareduck
Posts: 552
Joined: Wed Oct 17, 2007 5:03 am

Post by scareduck »

One of the other threads had mentioned the Nov. 2000 Physics of Plasmas paper by Chacon, Barnes, Miley, and Knoll that laid the groundwork for Bussard's claim of positive energy from an IEC device (their model used a Penning trap) with the proviso that certain conditions have to be met. Unfortunately those conditions include some of the deal-killers Rider mentioned, namely the brehmsstrahlung case where electrons and ions meet.

The interesting part was the acknowledgements where they mentioned using a DOE Cray Origin 2000. I checked and this was a R10000 500 MHz core with a rip-roarin' memory fabric. It would be interesting to see what sort of calculations they did.

scareduck
Posts: 552
Joined: Wed Oct 17, 2007 5:03 am

Post by scareduck »

I found the list of projects that won bids to get this machine time in a Slashdot thread on the same subject. I was originally going to publish the list of all the recipients, but with 54 pages, that's just not gonna work, so here's a shorter list of plasma and/or fusion research project titles:
  • Verification and validation of petascale simulation of turbulent transport in fusion plasmas (8M processor-hours, ORNL Cray XT4; studies plasma turbulent flow relevant to ITER)
  • Fluctuation Spectra and Anomalous Heating in Magnetized Plasma Turbulence (4M processor-hours, ORNL Cray XT4; study of heating in stellar plasma turbulence from first principles)
  • Modeling heliospheric phenomena with an adaptive, MHD-Boltzmann code (850k processor-hours, ORNL Cray XT4)
  • Three-Dimensional Particle-in-Cell Simulations of Fast Ignition (308k processor-hours @ Lawrence Berkeley NL NERSC HPC; inertial confinement fusion simulation)
  • High Resolution Global Simulation of Plasma Microturbulence (2M processor-hours @ Argonne NL IBM Blue Gene/P; tokamak plasma study)
  • Gyrokinetic steady state transport simulations (1.5M processor-hours @ ORNL Cray XT4; burning plasma simulation)
  • High Power Electromagnetic Wave Heating in the ITER Burning Plasma (1M processor-hours @ ORNL Cray XT4)
  • Petascale Particle-in-Cell Simulations of Plasma Based Accelerators (923k processor-hours @ LBNL NERSC HPC; simulation of a proposed compact laser-based particle accelerator)
  • Computational Atomic and Molecular Physics for Advances in Astrophysics, Chemical Sciences and Fusion Energy Sciences (2M processor-hours @ ORNL Cray X1E; simulating laser interactions with matter, and plasma diagnostics in controlled fusion experiments, among others)
The largest single recipient was UCSB's Robert Sugar, who received 27.6M processor-hours at two facilities to study the interactions between quarks and gluons.

Helius
Posts: 465
Joined: Sun Oct 21, 2007 9:48 pm
Location: Syracuse, New York

Post by Helius »

Scareduck; Thanks for the "short List". What I'm trying to figure, is which ones in the list *don't* involve solving simultaneous equations combining Maxwell with Hydrodynamics. Is it that *ALL* American research is to involve MHD Maths?

It's like everybody wins but Science. Schools and Gubberment labs get ever bigger computing facilities, and reserachers get to calculate out to some ever finer detail. This can go on forever.

TallDave
Posts: 3140
Joined: Wed Jul 25, 2007 7:12 pm
Contact:

Post by TallDave »

scareduck wrote:One of the other threads had mentioned the Nov. 2000 Physics of Plasmas paper by Chacon, Barnes, Miley, and Knoll that laid the groundwork for Bussard's claim of positive energy from an IEC device (their model used a Penning trap) with the proviso that certain conditions have to be met. Unfortunately those conditions include some of the deal-killers Rider mentioned, namely the brehmsstrahlung case where electrons and ions meet.
Yes and no. They did not set zero brem as a condition for large Q values, they just didn't include brem in their calculations.

As Nebel said in the MSNBC thread, the interesting conclusion from the paper is that different shapes of the well gave very different results:
It is also shown that realistic parabolic-like wells result in better energy gains than square wells, particularly at large well depths (>100 kV). Operating regimes with fusion power to ion input power ratios (Q-value) >100 have been identified.
So with slightly different assumptions very different results pop out. That says we need to understand the assumptions better to make realistic predictions. That means more experimental data.

Also, Bussard was claiming IEC could produce net gain long before that paper.

MSimon
Posts: 14334
Joined: Mon Jul 16, 2007 7:37 pm
Location: Rockford, Illinois
Contact:

Post by MSimon »

I calculated an ultimate Q of around 22 at reasonable power densities using the pB11 resonance region at around 50 KV well depth.

I can see Qs around 100 if you operate at a reaction rate of around .02 barns. The resonance region is around .1 barns and ultimate is about 1.2 barns.

Generally, once you are operating in an ultimate Q of 20 or above, raising the Q does not change the gain by enough to make it worth the effort.

A Q of below 5 is considered impractical as a commercial energy generator. A Q of more than 20 is unnecessary (if it comes for free or with other advantages I wouldn't turn it down).
Engineering is the art of making what you want from what you can get at a profit.

Post Reply