Center for Radiative Shock Hydrodynamics

Center for Radiative Shock Hydrodynamics

Introductory overviewR. Paul Drake

Page 2

We will take you from overviews to specifics

• This first presentation– Motivation and introduction to the physical system– Overview of the experiments and of the project

• Overviews this morning – Powell on the simulation – Holloway on assessment of predictive capability

• Code and verification this afternoon – Toth on architecture and practices – Myra on tests

• Assessing predictive capability in the morning – Bingham on our first integrated study – Fryxell on 3D sensitivity runs

• Posters today – See the details and meet the people

• You will see how our priorities have been driven by becoming able to assess the capability of our code to predict our year 5 data.

Page 3

We find our motivation in astrophysical connections

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

• Exist throughout astrophysics– Supernovae, accretion, stars,

supernova remnants, collisions

• Our experiments – have behavior and

dimensionless parameters relevant to shocks emerging from supernovae

– We should see any important unanticipated physics

– Good code test in any event

Ensman & Burrows ApJ92

Reighard PoP07

Page 4

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 5

Shock waves become radiative when …

• radiative energy flux would exceed incoming material energy flux

where post-shock temperature is proportional to us

2.

• 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

ous3/2

Initial ion temperature 2 keV

Typ. radiation temp. 50 eV

Page 6

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 7

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, 2007F. Doss, et al. HEDP, submitted 2009

Schematic of radiograph

Grid

Page 8

We have identified key parameters for the code/experiment comparison

• Key measurements at data time– Basic (1D)

• Shock position• Layer thickness

– Multi-D • Distance of kink in shock from tube wall

• Angle of xenon edge just downstream of shock

Page 9

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 10

Last April we reported a CRASH 1.0 3D simulation of the year 5 experiment

Density Temperature

Initialized by calibrated laser code, to be discussed by Ken

Side view

Top view

2008 IRT: The project will live or die based on whether a “reasonably good code” can be built

This font color highlights responses to 2008 IRT recommendations

Page 11

The CRASH research process incorporates many UQ elements. Culture change is here.

Page 12

UQ considerations have driven the project to date

• Had to become able to Assess Predictive Capability – Develop a “reasonably good” code – Code tests and physics tests – “Drill-down” documentation system

– Variability of base experiment (10/08) – Early data to calibrate inputs (12/09) – Learn enough to define the CRASH UQ methodology– Tests of UQ methods

• We are using the tools we have assembled– Going forward, UQ analysis must tell us what code

to write, what experiments to do, and what UQ to do

• Working independently on additional diagnostics – X-ray and imaging Thomson Scattering

Page 13

We are organized to bring leadership and effort to all areas where they are needed• Most of our team members participate in more than one

area (details in supplementary material)

• We use our resource plan, our work plan, and UQ considerations to inform priority decisions

Predictive Capability beyond traditional V&V Scientific Computing PrimaryJames Holloway, Co-PI, lead Primary Quentin Stout, Co-PI, lead Role FTEBruce Fryxell, chief scientist Role FTE Professors 3

Professors 6 Other staff 1Research Scientists 2 1.4 Modeling and Theory Primary

Other staff 4 Jointly led Role FTEGraduate students 6 Professors 10

Research Scientists 1 0.33Other staff 0

Graduate studens10Code Development and Traditional V&VKen Powell, Co-PI, lead Primary Experiments PrimaryGabor Toth, Software Archit.Role FTE Paul Drake, Director, leadRole FTE

Professors 1 Professors 1Research Scientists 4 2.15 Research Scientists 1 0.8

Other staff 2 Other staff 1Graduate studens 6

Page 14

We have expanded our theoretical work

• The IRT recommended more effort in analytic theory

• Paul Drake has invested more time

– Enabled by contributions by Ken Powell and James Holloway

• Ryan McClarren has a paper on the structure of this type of radiative shock in draft form.

– Rob Lowrie (LANL) has collaborated on this.

• Graduate student Forrest Doss continues to work on the

analytic theory of the shocked layer instability

• Igor Sokolov has contributed considerably

• Emilio Minguez (U.P. Madrid) has visited for opacity collaborations

• We are working to engage Dmitri Ryutov (LLNL) and Sasha Velikovich (NRL), who are interested

• We have a CRASH primer which will evolve

Page 15

CRASH will have other applications and users

• Solar group is already doing rad-MHD evolving from CRASH

• Our experiment team (students funded variously) needs CRASH

• Already one other university is eager to use our code (FSU)

• The labs will be users of our trained people more than codes

CRASH simulation of NIF Radiative HydrodynamicInstability experiment at 7.0 ns: 2D, 600 x 80

• Our NIF team may become a key user– Experimental program is

making first university use of NIF

– Excellent opportunity to apply CRASH and see what breaks

Page 16

We intend to accomplish an important result

• Our unique intended contribution – Be the first academic team

to use statistical Assessment of Predictive Capability

to guide improvements in simulations and field experiments

that lead to predictions of extrapolated field experiments

known to have improved accuracy, and to demonstrate this by field measurements.

• This is a sensible goal because– Our codes are almost entirely first-principles

calculations – Our approach will be to add physics not tuning

Page 17

Supplemental material follows

• More details

Page 18

People p. 1

CRASH Faculty Where & what UQ/APCScient. Comp.

Modeling & Theory

Exper-iments

UQ/ Predictive Capability & V&VJames Holloway, Co-PI, lead UM Prof. Nuclear X XBruce Fryxell, chief scientist UM AOSS Res. Sci. X X XNatasha Andronova UM AOSS Res. Sci. XKrzysztof Fidkowski UM Prof. Aero X XBani Mallick TAMU Prof. Stats XVijayan Nair UM Prof. Stats & IOE XDerek Bingham SFU Prof. Stats XJi Zhu UM Prof. Stats & IOE X

Scientific ComputingQuentin Stout, Co-PI, lead UM Prof. CSE X XNancy Amato TAMU Prof. CompSci XLawrence Rauchwerger TAMU Prof. CompSci X

Code Development, Testing, and UQ supportKen Powell, Co-PI, lead UM Prof. Aero X XGabor Toth, Software Archit. UM AOSS Res. Sci. X XIgor Sokolov UM AOSS Res. Sci. XBart van der Holst UM AOSS Res. Sci. XEric Myra UM AOSS Res. Sci. X X X

Modeling and TheoryMarv Adams, Co-PI TAMU Prof. Nuclear X X XBen Torralva UM MSE Res. Sci. XEd Larsen UM Prof. Nuclear X XBill Martin UM Prof. Nuclear X X XRyan McClarren TAMU Vis. Prof. Nuclear X X XJim Morel TAMU Prof. Nuclear XBram van Leer UM Prof. Aero XPhil Roe UM Prof. Aero XSmadar Karni UM Prof. Math XKatsuyo Thornton UM Prof. MSE XTamas Gombosi UM Prof. Space Sci. X

ExperimentsPaul Drake, Director, lead UM Prof. AOSS X X X XCarolyn Kuranz UM AOSS Res. Sci. X

Page 19

People p. 2

Grad Students Advisor UQ/APCScient. Comp.

Modeling & Theory

Exper-iments

Khieu, Loc Van Leer XMiranda, Colin Fidkowsky and Powell XSouza, Marcos PowellZaide, Daniel Powell and Roe XChou, Jason Fryxell and Drake XDoss, Forrest Drake X XPatterson, Nick Thornton and Drake XHuntington, Channing Drake XKrauland, Christine Drake XVisco, Tony Drake XCheatham, Jesse Holloway and Martin XMoran, Tiberus Holloway XDavidson, Greg Larsen XBaker, Eric Holloway and Martin XZhang, Zhanyang Nair/Zhu XDi Stefano, Carlos Drake XGamboa, Eliseo Drake XYoung, Rachel Drake X X XStarinshak, Dave Karni and Fryxell XMukherjee, Ashin Nair/Zhu XBarbu, Anthony Adams&Morel XEdward, Jarrod Adams& Morel XPrabhakar, Avinash Mallick X

StaffMike Grosskopf UM Sr. Res. Eng. X X XDonna Marion UM Technician/Target Fab XErica Rutter UM Technicican/ Codes X XMauro Bianco TAMU Post doc X XDuchwan Ryu TAMU Post doc XW. Daryl Hawkins TAMU Softwr Architect X XSergey Manolov TAMU Staff programmer X XMichael Adams TAMU Staff programmer X X

AdministrativeKathy Norris UM CRASH AdminJan Beltran UM Sr. Admin. Asst.

Page 20

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 21

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 22

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 23

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 24

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 25

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

QuickTime™ and aTIFF (LZW) decompressor

are needed to see this picture.

Emission

Xray Thomson scattering

Interferometry

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

Page 26

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 27

Preliminary analysis of XRTS obtained reasonable temperatures but a better model of Z is needed

•No drive beams, Null shot

•Zfree = .2

•15ns delay•Scattered from radiative pre-heated region•Fit gives Te = 10 eV & Zfree = 12

•19ns delay•Scattered from dense cooling region•Fit gives Te = 55 eV & Zfree = 14

Null Shot Precursor Cooling Layer

Data and fits by Tony Visco

Page 28

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

See E. Vishinac, ApJ 1983

Page 29

U

Vs

Perturbed system Unperturbed 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 30

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

Page 31

The CRASH project has evolved over its first 18 months

Spring 2008Funds arriveCode planningRecruiting

Summer 2008CRASH 1.0 tasksFirst sensitivity

Winter 2009CRASH 1.0 frozen 3D Yr5 simulationUQ methods exploration

Fall 2008CRASH 1.0 betaExpand UQ teamVariability ExptHire 2 FTE +

Spr/Sum 2009CRASH 2.0 Tasks UQ on UQ First end to end UQ3rd hire (expts)

Fall 2009 expectation CRASH 2.0 beta 3D CRASH sensitivity Define 2D UQ studyCalibration experiment

Project status at day zero: To do UQ, needed “reasonably good code” and experimental data on variability and for calibration