Center for Radiative Shock Hydrodynamics Fall 2010 Review Introductory overview R. Paul Drake.

Center for Radiative Shock Hydrodynamics

Fall 2010 Review

Introductory overviewR. Paul Drake

Page 2

What lies ahead

• This first presentation– Motivation and introduction to the physical system– Overview of the past year: progress, challenges, decisions

• Following presentations today– Drake on the integrated project– Adams on transport physics and UQ – Powell on the simulations – Holloway on assessment of predictive capability

• Code and verification tomorrow morning– Toth on architecture and practices

• Other highlighted contributions tomorrow morning (little time! ) – Kuranz, Sokolov, Morel

• Posters today – See the details and meet the people

• You will see how our priorities have been driven by becoming able to conduct a sequence of integrated UQ studies.

Items in this color are directly responsive to 2009 review

Page 3

We are showing a visualization of CRASH 2.1+ output on the other screen

• Simulation details – 9600 by 960 effective resolution in 2D – Multigroup diffusion (30 groups, 0.1 eV to 20 keV) – 5 materials, 3 AMR levels, CRASH EOS & Opacity

• Also see scale models in the room

7.6 ns

Page 4

We find our motivation in astrophysical connections

• Radiative shocks have strong radiative energy transport that determines the shock structure

• Exist throughout astrophysics

Ensman & Burrows ApJ92

Reighard PoP07

SN 1987A

Cataclysmic binary star(see Krauland poster)

Page 5

A brief primer on shock wave structure

• Behind the shock, the faster sound waves connect the entire plasma

Denser,Hotter Initial plasmaShock velocity, us

Mach number M > 1

unshockedshocked

Mach number M = us / csound

Page 6

Shock waves become radiative when …

• radiative energy flux would exceed incoming material energy flux

where post-shock temperature is proportional to us2.

• Setting these fluxes equal gives a threshold velocity of 60 km/s for our system:

Material xenon gas

Density 6.5 mg/cc

Initial shock velocity 200 km/s

shockedunshockedpreheated

Ts4 ous

3/2

Initial ion temperature 2 keV

Typ. radiation temp. 50 eV

Page 7

The CRASH project began with several elements

• An experimental system that is challenging to model and relevant to NNSA, motivated by astrophysics

• A 3D adaptive, well scaled, magnetohydrodynamic (MHD) code with a 15 year legacy and many users

• A 3D deterministic radiative transfer code developed for parallel platforms

• A strong V&V tradition with both codes

• Some ideas about how to approach “UQ” in general and specifically the Assessment of Predictive Capability

Space weather simulation

Page 8

CRASH builds on a basic experiment

• Basic Experiment: Radiography is the primary diagnostic. Additional data from other diagnostics.

A. Reighard et al. Phys. Plas. 2006, 2007

F. Doss, et al. Phys. Plas. 2009, HEDP 2010

Schematic of radiographGrid

(see Doss poster)

Page 9

What we predict

What the radiograph fundamentally shows us is where dense Xe exists

Grid

• We predict scalar quantities– By predictive modeling we mean– computing an estimate of the

probability distribution function (pdf) of the outputs generated by the pdf of the inputs for a prospective field experiment, informed by both simulation and prior field experiments

• We predict the area where dense xenon exists on a radiograph and selected moments of the distribution of such locations– Holloway will show much more

about this– Grosskopf has a poster on the

integrated metrics

Page 10

CRASH 2.1+ has substantial capability

• Dynamic adaptive AMR• Level set interfaces• Self-consistent EOS and

opacities for 5 materials• Multigroup-diffusion

radiation transport• Electron physics and flux-

limited electron heat conduction

• Ongoing – Laser package – Multigroup preconditioner– I/O performance upgrade– Use of other EOS 3D Nozzle to Ellipse @ 13 ns

Material & AMR

Log Density

Log Electron Temperature

Log Ion Temperature

Page 11

CRASH has proven useful

• Design simulations of radiative reverse shock experiments

• Simulations of ongoing NIF experiment

• Simulations of x-ray driven radiative-shocks

• We used CRASH to help select some details of the radiative reverse shock design (Krauland poster)

x-ray driven radiative-shock (Myra poster)

Page 12

We have accomplished a lot during the past year

• UQ and predictive studies – Predictive study involving calibration – Two papers – Radiograph interpreter for integrated

metrics– Deeper analysis of experimental and of

all sources of uncertainty – Extensive studies of output sensitivity

to code details– New tests– H2D 104 run set – Predictive study with calibration from

H2D run set– Analysis of H2D limitations– 3D Hydro experiment design

• CRASH 2.0 released and used– X-ray-driven modeling – Pure hydro nozzle study– Application to other experiments – Detailed examination of axial structures– Hydro instability studies

• Code improvements – Flux limited electron heat xport– EOS source adaptivity – Laser package – Progress on multigroup preconditioner– Hydro adjoint implementation– Reduced alchemy– Improved parallel I/O? – Vastly improved PDT scaling

– Physics – Radtran & radhydro theory papers – X-ray driven walls theory – Further work on wall shock – Obtaining STA opacities – Work on non-LTE effects – SN/FLD comparison

– Experiments – Shock breakout measurements – Initial attempts at other early time

measurements – Late time (26 ns) measurements for

predictive study – Radiograph analysis (compression,

background)– UQ-driven planning for year-3

experiments– Metrology comparison

Items in this color are directly responsive to 2009 review

Page 13

We have also encountered some challenges

Page 14

Initializing CRASH with Hyades proved problematic

• H2D has a laser package and (now) rezoner– Did run set for Dec 09 expt– Superseded by 104-run set

done in early 2010– This has produced results

• But using Hyades has proven impractical – Rezoner had fidelity issues – Code revisions were slow – UQ was problematic– Results differ vs CRASH– H2D is manpower intensive

• The rezoner works fine for typical design studies but not for predictive science

• Comparison using 6 vs 3 zones in auto-rezoner:

Decision:do a laser package in CRASH

Page 15

The simulated morphological features were not useful for UQ

• The CRASH code has yet to reliably produce the observed morphology in runs using Hyades initialization for laser drive

Decisions: 1. Focused effort for several months, then moved on; later

improvements made a difference: see talk by Ken Powell

2. Adopted integrated metrics that are independent of morphological detail: see poster by Mike Grosskopf

3. Did predictive study with calibration using 1D simulations: see talk by James Holloway

Spring 2010 Fall 2010

Page 16

Politics precluded integration of CRASH and PDT

• One of the TST members indicated that at the labs the combined code would be considered UCNI – We sought a ruling, and what came out of DOE HQ was: “The final

authority believes that the guidance is wrong and should be changed, but under current rules such a code would be UCNI”

– We are told this will be addressed, “slowly” • This is despite the fact that several US universities and numerous

foreign researchers are writing and even publishing codes with analogous capabilities.

• Doing an UCNI code is for us a practical non-starter

Decision: until this situation changes, we will pursue correlated studies to understand the impact of limited fidelity

It might prove useful for the Review Team to make a very strong recommendation to DOE about addressing this

Page 17

Predictive simulation roadmap

Page 18

We are now ready for multi-D integrated studies

• Our code is “good enough” and is getting better

• We have carried out the UQ elements needed

• The primary limitation going forward is computational– Details and implications to be discussed at length later – Includes core-hours limitations but also much more – Affects approach to UQ (following talks)

• We intend to be the first academic team– to use statistical Assessment of Predictive Capability– to guide improvements in simulations and field experiments – that lead to predictions, known to have improved accuracy, of

field experiments having extrapolated parameters (not physics) – and to demonstrate this by field measurements.

Page 19

Supplemental material follows

Page 20

People p. 1

Page 21

People p. 2

Page 22

Our experimental sequence will improve and test our assessment of predictive capability

• A conceptually simple experiment– Launch a Be plasma down a

shock tube at ~ 200 km/s

• Year 5 experiment – Predict outcome and accuracy

before doing year 5 experiment

• Goals– Improve predictive accuracy

during project– Demonstrate a predictive

uncertainty comparable to the observed experimental variability

– A big challenge and achievement

Page 23

Conservation of energy forces the shock wave to develop complex structure

Shocked xenon layer Compressed 40x Traps thermal photons

Preheated regionThermal photons escape upstream

Other fun complications: Instabilities Wall shocks

Page 24

Our experiments are at the Omega laser

Omega 60 beams30 kJ in 1 ns0.35 µm wavelength

One of our shots at the Omega laser Related experiments LULI & PALS & RAL, LIL (soon?) NIF & LMJ maybe someday

Page 25

How to produce radiative shocks

Gas filled tubes

Laser beams launch Be piston into Xe or Ar gas at > 100 km/s

Piston drives shock

Diagnostics measure plasma properties

Gold grids provide spatial reference

Parameters1015 W/cm2 0.35 µm light1 ns pulse 600 µm tube dia.

Targets: Korbie Killebrew, Mike Grosskopf, Trisha Donajkowski, Donna Marion

Experiments: Amy Reighard, Tony Visco, Forrest Doss

Page 26

The laser first creates structure at the target surface

• The laser is absorbed at less than 1% of solid density

Ablation pressure from momentum balance:

Typical pressures of tens of Mbars

From Drake, High-Energy-Density Physics, Springer (2006)

p ~ 8.6 I142/3 / µm

2/3 Mbars

Radiative shocks need thinner targets than the one shown here

Page 27

For radiative shocks, target acceleration produces the high required velocities

• Profiles at 1.3 ns shown

Laser produced pressure accelerates Be plasma

Expanding Be drives shock into Xe gas

Acceleration pushes velocity into radiative shock regime

Page 28

Researchers are studying these shocks with a range of diagnostics and simulations

• Radiographs Emission

Xray Thomson scattering

Interferometry

Data credits: L. Boireau S. Bouquet, F. Doss M. Koenig, C. Michaut, A. Reighard, T. Visco , T. Vinci

Page 29

Radiography is our workhorse; we also use other diagnostic methods

Radiographs (1 or 2 views)

Data by grad students Amy Reighard (Cooper), Tony Visco, Forrest Doss, Channing Huntington Christine Krauland

Transverse Streaked Optical Pyrometer (SOP)

Transverse VISAR

UV Thomson Scattering

X-ray Thomson Scattering

Page 30

Lateral structure within the shocked layer is expected from a Vishniac-like mechanism.

See E. Vishinac, ApJ 1983

Page 31

U

Vs

Perturbed systemUnperturbed system

BeZ = H

Z = 0

Vorticity features

Shocked Xe

Unshocked Xe

Theoretical analysis shows structure internal to shocked layer for the experimental case

• Wavelength and growth rate of instability in reasonable agreement with observations

• Stereoscopic experiments will seek further evidence

Forrest Doss, et al. in preparation

-Vs

.

Page 32

Simulating these shocks is challenging but not impossible

• Optically thin, large upstream• Electron heating by ions• Optically thin cooling layer • Optically thick downstream

This problem has• A large range of scales• Non-isotropic radiation• Complex hydro 20