bout++ workshop 2015 - bout++ project€¦ · bout++ overview what bout++ is: a toolbox for solving...

BOUT++ workshop 2015

Ben Dudson

York Plasma Institute, Department of Physics,University of York, Heslington, York YO10 5DD, UK

BOUT++ Workshop

14th September 2015

Ben Dudson, YPI BOUT++ Workshop 2015 welcome (1 of 21)

Fire exits: first floor

Fire exit from side of meeting room

Escape Stairs

FireExit

GN/107 GN/106

GN/103

GN/111

GN/109

PhDOffice

FusionLearningStudio

GN/116

GN/112

Store

MailArea

GN/113Cl.

Shower

GN/108

DrNWoolsey DrJPasley ProfGTallents

GN/110

Male WC

Remote TokamakControlRoom

GN/104

IndustryOfficer

UnisexWCUnisexWC

FIRSTFLOOR

Meet outside at the front of the building


Fire exits: ground floor

Two exits: the front, and the back right corner

Escape Stairs

FireExit

DrRVannDrKGibson DrBDudson

DrKImadaDrTO’Gorman

GN/002

DrDO’Connell

GN/006ProfGPert

DrEWagenaars

GN/007

MainEntrance

Lobby

CommonRoom

Alfv nMeeting Room[U+e9]

wc

ProfTGans

PhotocopyingArea

KitchenStore

UnisexWC UnisexWC

Receptioncounter

Bulkhead

GN/005

GN/018 GN/017 GN/016 GN/015 GN/012 GN/011GN/010

MrRArmitage

GN/009

Switchroom

VisitorsDirector[U+2018]s OfficeProfHWilson

Reception/AdminMs RStephenson

GN/013

GN/019 GN/020

GN/003

Cleaners

GROUND FLOOR

GN/004

Meet outside at the front of the building


Workshop events

Coffee: Morning and afternoon coffee breaks

Lunches: Buffet lunches with sandwiches, hot and cold drinks

Dinner: The workshopdinner will be tonight in theAle House on Walmgate,starting at 7:00pm. Pleasebring cash (around £20).


Miscellaneous

Taxis The phone downstairs in the common area can calllocal numbers. Some taxi companies are

Ebor taxis: 641 441Station taxis: 623 332

WiFi internet access. If you need a guest account please seeme. Connect to eduroam or UoY Setup.

Printing and Fax. If you need anything printed, or faxed,please see Ben or Sharon

Lecture room not locked. Please see Ben about storingvaluables.

Welcome to York, and have a good meeting


Introduction to BOUT++


Background: Transport in the tokamak edge

Plasma edge region criticalto reactor design

Core performanceHeat fluxes to divertorParticle fluxes to first wall

Understanding of transportprocesses incomplete

Leads to uncertainty inpredictions for future devices

*Figure: www.euro-fusion.org/glossary/divertor/


Background: Tokamak divertor exhaust problem

ITER designed for 500MW fusionpower for 50MW input

Of this 500MW, 100MW goes intothe plasma→ 150MW must beexhausted from the plasma

Area available to handle power

A ' 2πRλq/ sinα

R ' 6m α ' 2.5o λ ∼ 2 − 20mm

→ A ' 1.7m2

Power fluxes could be∼ 100MW/m2∼ 10× technological limits!



1T.Eich et.alNucl. Fusion 53 (2013) 093031Ben Dudson, YPI BOUT++ Workshop 2015 welcome (9 of 21)


A solution is thought to exist for ITER, even with λq ∼ 1mmBeyond ITER, a demonstration power plant (DEMO) will needto handle even higher powers in order to be economical(Power increased ×5)Machine cannot be much larger than ITER

Heat flux becomes 500MW/m2 rather than 100MW/m2

“A reliable solution to the problem of heat exhaust isprobably the main challenge towards the realisation ofmagnetic confinement fusion”

– EU-EFDA Report: Fusion Electricity A Roadmap to therealisation of fusion energy (2012)


Tokamak edge simulations

Transition region betweenhot core and materialsurfaces

Complex nonlinear physics:plasma turbulence, neutrals,impurities, material surfaces

Complex geometry:X-points, shaped divertorsurfaces, snowflakeconfigurations,...


BOUT++ development aims to tackle these problems

BOUT++ has been designed(and re-designed) to handle thiscomplicated physics

Curvilinear coordinates andbranch-cuts allow quitegeneral geometries (*)

Extendable code allowsexperimentation withnumerical methods andphysics models


BOUT++ overview

What BOUT++ is:

A toolbox for solving PDEs on parallel computers. Aims toreduce duplication of effort, and allow quick development andtesting of new models

A collection of examples and test cases

Focussed on flute-reduced plasma models in field-alignedcoordinate systems, though has more general capabilities

And is not:

A single plasma model or simulation

A general library of numerical methods. Other libraries likePETSc are available for that

Magic. Appropriate numerical methods depend on theproblem, and must be chosen intelligently by the user


BOUT++: A toolbox for plasma simulations

Collection of useful data types and associated routines.Occupies a middle ground between problem-specific codesand general libraries (e.g. PETSc, Trilinos, Overture,Chombo,...)

Has its origins in the BOUT code123. Re-written andre-designed (at least once) in C++45

Researchers can make use of a (mostly) well tested library ofsimulation code and input / output tools

Greatly reduces the time needed to develop a new simulation

1X.Q. Xu and R.H. Cohen, Contrib. Plasma Phys. 38, 158 (1998)2Xu, Umansky, Dudson, Snyder, CiCP 4, 949-979 (2008)3Umansky, Xu, Dudson et al. Comp. Phys. Comm. 180, 887-903 (2008)4Dudson, Umansky, Xu et al. Comp. Phys. Comm 180, 1467 (2009)5Dudson et al. J. Plasma Phys. (2014).

Preprint: http://arxiv.org/abs/1405.7905Ben Dudson, YPI BOUT++ Workshop 2015 welcome (14 of 21)

http://arxiv.org/abs/1405.7905

Growing BOUT++ community

Open source project, available on github

http://boutproject.github.io

Users in labs and universities in UK, US, China, India, Japan,S.Korea, Denmark,...


http://boutproject.github.io

BOUT++ workshop

Aims of this workshop are to:

Get new users started with using BOUT++

Share experience and expertise in plasma edge modellingand tools

Discuss problems and future work


BOUT++ workshop

Aims of this workshop are to:

Get new users started with using BOUT++

Share experience and expertise in plasma edge modellingand tools

Discuss problems and future work

Outline

Getting started (This afternoon, Tuesday afternoon)

Recent work and developments (This morning, uesdaymorning)

More advanced uses (Tuesday afternoon)

Divertor modelling activities (Wednesday morning)


Getting started (This afternoon)

Getting started, runningsome “simple” test cases

Heat conduction,turbulence, plasma blobsReading data into Python.Also possible in IDL,Mathematica, Matlab,Octave,...Some new tools forrunning simulations andvisualising outputs

Introduction to where theimportant bits of the codeare

Data handling, PDEsolvers, integrationAnatomy of a physicsmodel

0 10 20 30 40 50 60 70Radius [ρs ]

0

10

20

30

40

50

60

70

Bino

rmal

dire

ctio

n [ρs]

0.990

1.035

1.080

1.125

1.170

1.215

1.260

0 50 100 150 200 250X index

0

50

100

150

200

250

Z in

dex

Normalised plasma density fluctuation n

0.96

0.72

0.48

0.24

0.00

0.24

0.48

0.72

0.96

1.20


Applications in tokamak physics

Filaments / blobsTransport of heat and particles in SOL2D examples run on a laptop

Edge turbulenceFormation of blobsNear SOL heat transport (λq)

Divertor simulationsSpreading of particle and power fluxesto surfacesInteraction with neutral gas anddetachment

Pedestal physics and ELMsGyro-fluid models to capture FLR,Landau damping, and drift resonanceeffectsL-H transition physicsStability and nonlinear dynamics

0 50 100 150 200 250

0

50

100

150

200

250

Density [×1018m−3 ]

0.00

0.06

0.12

0.18

0.24

0.30

0.36

0.42

0.48

0.54

0.60


Recent developments

Many people contributing fixes, features and improvements:George Breyiannis, David Dickinson, Luke Easy, Erik Grinaker, JoeHenderson, Peter Hill, Jarrod Leddy, Michael Loiten, Peter Naylor,Jens Madsen, John Omotani, Kevin Savage, David Schwoerer,Luke Townley, Nick Walkden (apologies for omissions)

In the last year(git diff --stat)

534 files changed

22674 insertions(+)

47251 deletions(-)


Recent developments

Some of the new features at this workshop are:

Implementation of the Flux Coordinate Independent (FCI)scheme (P.Hill, B.Shanahan)

Implicit-Explicit (IMEX) schemes for time integration(N.Walkden, B.Dudson, D.Dickinson)

Non-orthogonal meshes for realistic tokamak geometries(J.Leddy, N.Walkden)

Interaction with neutral gas (S.Mekkaoui, J.Leddy, B.Dudson)

Finding eigenvector/eigenvalues using SLEPc (D.Dickinson,B.Dudson)


Issues and future work

Increasing time-step size: Often limited by fast electronparallel dynamics to δt < 1/ΩciCalculating the potential φ from vorticity efficiently androbustly

Inclusion of impurities

Development of neutral gas models and divertor turbulencestudies

...

Lots of things to discuss!


bout++ workshop 2015 - bout++ project€¦ · bout++ overview what bout++ is: a toolbox for solving...

Documents