A Whirlwind Tour of Beamline OptimisationIntroduction Acceptance Diagrams J. Carpenter, D. Mildner,...

A Whirlwind Tour of Beamline Optimisation

Phil Bentley

European Spallation Source & Uppsala University

18th February, 2016

Phil Bentley (ESS/Uppsala) A Whirlwind Tour of Beamline Optimisation 18th February, 2016 1 / 36

Introduction

Acceptance Diagrams

J. Carpenter, D. Mildner,J. Copley, L. CussenIf you want tounderstand neutronoptics, you have to readthis work


Introduction

Acceptance Diagram Shading

Monte-Carlo tends to beslow for high-resolutionworkRay-tracing could bereplaced by some rastermethod?This idea published2009, but borrowed frommuch older work incomputing

POVRAY

HALF LIFE 2Phil Bentley (ESS/Uppsala) A Whirlwind Tour of Beamline Optimisation 18th February, 2016 3 / 36

Introduction

Acceptance Diagram Shading — Source

Guide entranceparallelogram is definedby bounding values


Introduction

Acceptance Diagram Shading — Propagation &Guides

Propagating beam is ashear operation oncorner points — veryfast!Reflected componentsare reflected twice —require subdivisionfunctions in codeNow we need toattenuate the reflectedpart by the reflectivity


Introduction

Acceptance Diagram Shading — Attenuation

Idealistic reflectivitycurveWeighted phase spacejust like Monte-Carlocodes


Introduction

Acceptance Diagram Shading — SANS Benchmark

SANS agreement is verygoodMore information inacceptance diagrams


Introduction

Acceptance Diagram Shading — Spin Echo & TASBenchmark

Again, agreement withMonte-Carlo is verygood


Introduction

Acceptance Diagram Shading

It can be very fastIt is monochromaticIf you like writing XML, launch software by typing, and go by thename of “Phil”, it’s highly intuitiveUsed on:

ILL — on H5 conceptual design beam extraction (WASP, THALES,D22, IN15 etc)ANSTO — On NBI2 project for cold guide upgrades, BILBYinstrumentESS — Rapid concept exploration, first design on NMX guides,parameter space boundariesBentley & Andersen, NIMA 602 (2009) 564


Introduction

Curved Guide Transmission

D. Mildner — NIMA 290(1990) p.189


Introduction


Analytical study ofcurved guide phasespace


Introduction


Analytical study ofcurved guide phasespaceCan calculate veryquickly the transmissionof the systemNo Monte-Carlo needed!Now have a verypowerful way to exploreparameter space


Introduction


In fact you can write afunction of: w , λ, L, R, mGet a spreadsheet ofcosts from vendors whotrust you(!)Now you can write a(secret) function thatinstantly gives you theprice and performanceof guides(Plus some tweaking formulti-channel benders)


Introduction

Shielding Costs

Get a shopping list of shielding costs:Concrete — standard rate from construction workersLead — London Metal ExchangeSteel — engineer’s priceT0 chopper — talk to chopper teamHeavy shutter — Design & engineering deptLight shutter — One “standard price” (=20,000 british-euro-dollars)


Introduction

Shielding Costs

Now we canout-bean-count thebean-counters!


Introduction

But First This... A Standard Guide

Divergence defined forλmin, λopt w/ optimisedfocussing (fully analytic)Phase space matchingfor zig-zag reflections(fully analytic)Const. sectioncurved/bender in middle(fully analytic)Compression expansionphase space matched(fully analytic)Paper in draft...

Source Detector

Sample

Beam extraction: Focus-defocus system

Straight part: Filtering of unwanted radiation Low m coating

Focusing part: Adapted to the sample size and its environment

Source Detector

Sample




Source Detector

Sample




Source Detector

Sample





Introduction

Reference Suite Instruments

3 groups of instrumentsλ ≥ 4Åλ ≈ 2 Åλ ∼ 1 Å


Introduction

Long Reference Suite Instruments

2 groups of instrumentsλ ≈ 2 Åλ ∼ 1 Å


Introduction

Curved Guide for 150 m 1Å

Line of sight lost within75 metresm = 1.5width = 4 cmR = 17580 mAnother iteration may beneeded for one or twoinstruments


Introduction

Bender for 150 m 1Å

Line of sight lost within25 metresm = 3.0width = 4 cmchannel = 0.5cmnchannels = 8R = 1250 mCost = 1.5 Me


Introduction

Bender for 150 m 2Å

Line of sight lost within25 metresm = 2.5width = 4 cmchannel = 2.0cmnchannels = 2R = 1250 mCost = 350 ke


Introduction

Other Instruments

Can do the same thingfor other lengths,wavelengths etc.


Cost Analysis

Standardised Guide Systems of Different Lengths

Total costs for curvedguides are lower thanstraight guides.Using benders to loseline of sight in thebunker does not look tobe cost-effective at thisstage.Gains are roughlyproportional in eachcase, so a suite costcomparison is not tooinaccurate...


Cost Analysis

Potential Savings

Baseline = optics asproposed by scientists“Standard” = straightguides allowed, reducedspecs on optics“Curved” = no straightguides allowed“Bender” = Get out ofline of sight within 20metres (bunker).Total savings up to 24Me.


Cost Analysis

Master of Puppets

Genetic algorithm-basedsolutionMinimises backgroundeffects in previous slideAvoids mechanicalinterferencesUses standard methodsto find solution

Tournament selectionlogarithmic scoringStandard GAparameters work verywell

Used in Phase 2 toallocate beamports


Cost Analysis

Beam Port / Layout

Each 12-port sector has(n − 1)! ≈ 40,000,000possible instrumentarrangementsVariations in backgroundsensitivity & cross-talk,moderator spectralbrightness, A2Tbackground, spallationbackground ∝ θAdd in chopper overlaps,shutters, walls, vacuumvessels


Cost Analysis

Master of Puppets

Genetic algorithm-basedsolutionMinimises backgroundeffects in previous slideAvoids mechanicalinterferencesUses standard methodsto find solution

Tournament selectionlogarithmic scoringStandard GAparameters work verywell

Used in Phase 2 toallocate beamports

radius5

radius10

radius15

radius20

radius25

radius30

radius35

radius40

radius45

radius50

radius55

ESTIA

EMPTY2

EMPTY5

EMPTY

ODIN

SKADI

FUND

FREIA

EMPTY4

LOKI

EMPTY3

VOR


Cost Analysis

Master of Puppets

North Sector


Cost Analysis

Master of Puppets

East Sector


Cost Analysis

Master of Puppets

West Sector


Cost Analysis

Wolter Optics

Mildner & Gubarev, NIMA 634(2011) S7Can correct opticalabberations and create analmost ideal beam


Cost Analysis

Hybrid Guides

Elliptic Guides are not opticallyperfect systemsCan improve the divergenceby combining elliptic andparabolic sectionsBentley et al, NIMA 693(2012), 268


Cost Analysis

Selene Optics

Developed at PSI by JochenStahn et alTo me, this is the logicalconclusion of an almostperfect beamPoint-to-point imaging withhigh quality optical devicesPrototype built (and works!) atPSIOne of the first ESSinstrumentsStahn et al, arXiv:1102.2747


Cost Analysis

Acknowledgements

Just scratched thesurface4 years of research intoneutron backgrounds,signal-to-noise etc.Many thanks to DavidMilner, Shane Kennedy,Erik Iverson, MichaelFitzsimmonds, TakuSato, Thomas KristAlso Uwe Filges,Emmanouela Rantsiou,Jochen Stahn, KenHerwig, Chris Frost &Matt Fletcher


Cost Analysis

Neutron Optics and Shielding Group

Phil BentleyDoug DiJulioDamian MartinRodriguezCarolin ZendlerNataliia CherkashynaCarsten Cooper JensenValentina SantoroStuart AnsellStudents:

Ellen BrydevallJohn StenanderHanna Björgvinsdottir


Cost Analysis

Thank You

Thank You For Your Attention


A Whirlwind Tour of Beamline OptimisationIntroduction Acceptance Diagrams J. Carpenter, D. Mildner,...

Documents

Transcript of A Whirlwind Tour of Beamline OptimisationIntroduction Acceptance Diagrams J. Carpenter, D. Mildner,...