Tim Senden Department of Applied Mathematics Deputy Director (Technology Development) Research School of Physics and Engineering

CTLab Integrated Workflow for Digital Material Analysis

Research School of Physics & Engineering ANU College of Physical and Mathematical Sciences School Structure

• The largest university-based physics activity in Australia

• ~ 170 full time academic staff

• ~ 150 graduate students

• ~ 110 technical and administrative staff

• Full undergraduate teaching program (3-4 years)

• 9 Research Departments

Fundamental Research • Nanoscience • Quantum & atom optics • Non-linear physics • Mathematical physics • Atomic & molecular physics • Antimatter-matter studies • Nuclear science • Materials & surface science • Space science • Plasmas/fluids • Gravity waves

4 Major National Facilities • Heavy ion accelerator – largest in southern hemisphere • The Australian Plasma Fusion Research Facility • National Antimatter (positron) Facility • Australian National Fabrication Facility ACT node • CTLab – national CT Facility

5 Dept Applied Mathematics

 Founded by Prof Barry Ninham in 1971  Chemists, physicists, mathematicians and engineers  Theoreticians and experimentalists working together  What shapes matter?  How does matter interact?  Complex and hierarchical (multiscale)  Fields: geometry and topology in nature, molecular force measurement, nanomaterials, packing in granular materials and flow in porous materials  Applications : biomineralisation, composite materials, adhesion, oil & gas, printing and paper making, mineral processing, crystallographic theory, textiles, osteoporosis…..  Mesoscale is the next level bigger than ‘nano’.

People in Applied Mathematics • Prof Stephen Hyde – self assembly, hyperbolic structures • Prof Vince Craig – surface chemistry, pH, salt effects • Prof David Williams – stat mech and polymer physics • Prof Barry Ninham – theory of intermolecular forces • Prof Denis Evans – non-equilibrium thermodynamics • A/Prof Adrian Sheppard – multi-phase flow in porous media

• Prof Ross Stephens – nanobiomed, radiochemistry • Dr Vanessa Robins – digital topology/Morse theory • Dr Andrew Kingston – discrete transforms/reconstruction • Dr Shannon Notely – surface chemistry, 2D materials, graphene • Dr Mohammad Saadatfar – granular materials • Dr Glenn Myers – iterative methods, X-ray phase contrast

• 9 technicians – mechanical, software, instrument engineers • 15 students

In 1661 Kepler said about 1/4 of a box of oranges is empty.

In 1998 Hale proved him correct. Dilatacy & cohesion Shake/tap

Going from disorder to order We need to characterize the structural changes in terms of an order parameter Stable but frustrated polytetrahedra

Two Face-connected tetrahedra

3 or more face-adjacent dense tetrahedra = a Polytetrahedron configuration a a

Zoology of polytetrahedra

5-ring icosahedron

b c b c a

b c

Hanifourb et al Phys Rev Lettc (2014) • Increase of bipyramids, rise of HCP

• Branched tetrahedra; resilient from below RCP into crystal

• 5-rings: the system increases the packing fraction by forming these frustrated structures.

• Analogues with glass

• High resolution local volume tracking based on Delaunay triangulation of grain centres. • Local shear and strain buildups result in local dilatancy (increase of porosity) where grain >120,000 tetrahedra breakage occurs at the following stage. • >60% grain failure on local dilation Francois, Saadatfar, Sheppard, Phys Rev Lett (2013) Global contact numbers of frictional sphere packings and ellipsoids can be explained by a combination of local contact function and a conditional probability is dependent on global packing fraction and NOT ellipsoids' aspect ratios, coefficient of friction (material) and structural anisotropy (preparation protocol)

Heliscan

Sheppard et al. Nuclear Instruments and Methods in Physics Research: Section B (2014) Precise reconstruction and no geometric distortions

1 mm

Conventional circular scanning Helical scanning cone <10° cone ~110°

Kingston et al Med. Phys. 2011 Varslot et al SPIE 2012 Under the same • Higher geometric and signal fidelity conditions a conventional • > 3 to 10 times faster than conventional µCT micro-CT might • arbitrary lengths at highest resolution only scan this • resolution is source limited ~200nm to 5µm much. • giga- to tera-voxel datasets

40 Gbyte scan of carbonate core (5mm diameter, 20mm long, 3.5 micron resolution) Mango image analysis software • Scalable, distributed-memory parallel toolkit for image processing tasks: – tomographic reconstruction (openCL and CUDA) – registration – grayscale filtering – segmentation – morphological ops, skeletonisation, grain labelling – pore-network extraction • written in C++, uses message-passing interface (MPI), more recently some openCL and CUDA

Adrian Sheppard is theme leader EPINET – framework atlas epinet.anu.edu.au CAD models 3D printed mould

cast dynamic tomography compute

Finite Element Analysis

N. Francois et al, Philosophical Transactions of the Royal Society A (2014)

Drishti volume rendering software • Cross-platform, open source volume rendering software for visualizing 3D X- ray or electron tomographic data, confocal data, etc. • Drishti is used it for exploration, measurement and presentation. • 100’s of users world wide, growing communities in Europe and US • supports colour 3D printing • FREE

https://github.com/AjayLimaye/drishti Digicore Consortium A joint industry consortium (established 2006) Spun-out Digitalcore/Lithicon - 2009 Primarily in field of oil/gas but inclusive of all materials research. Conventionals, Unconventionals, carbon sequestration

• BG • BP • Conoco • EOG • FEI • JOGMEC • Maersk • OMV • ONGC • Petronas • PetroBras • Shell • Total Past members– Saudi Aramco, ExxonMobil, Attracts ~70 participants, • Whiting Chevron, Schlumberger, Baker Hughes, Abu about 20-25 from industry Dhabi Onshore, BHP Log & sample select image model

Physical Parameters Reservoir Descriptors Electrical Conductivity Oil Saturation Dielectric Permittivity Water Saturation Does image capture Neutron Gas Saturation the mineralogy, Borehole Pressure Porosity Sound Velocity Permeability wettability, NMR Response heterogeneity…. Gamma-ray x-section Capillary Pressure

Tomography provides the framework

Permeability Permeability (D)

Fractional porosity ~ 1 mm3 sandstone showing flow simulation Image registration • Metric-based method, affine or rigid • relatively resource intensive when applied to large datasets • scalable implementation

Latham et al SCA2008-35 2D-3D registration

μCT image close up 2D-3D registration

SEM image close up Registration: Porosity mapping

Heterogeneous tight sand 36x8mm

Imaged at 7200x16002 5μm voxels

Difference between saturated and dry image yields porosity map dry saturated difference

Imaging of CO2 dissolution

before after Barrow Island rock sample before and after treatment with carbonic acid for

329 hours under 1 MPa pCO2 at 15-20°C. Field of view 0.8 x 0.7mm

29 Simple in- situ drainage

Benthiemer sandstone, 5mm sample 2.8μm voxels Simple in- situ drainage

Benthiemer sandstone, 5mm sample 2.8μm voxels Dynamic Tomography

• In 4D tomography we need far fewer projections to reconstruct the changes between successive frames. cf. MPEG compression: the information required to encode the changes from one moment to the next is much less than that required to encode each frame in isolation • The geometrical characteristics of fluid flow are highly constrained and well suited. • Reconstruction is compute intensive for fully constrained sequence, best suited to GPU.

~ 20min water drainage sequence in sand 500 tomograms Myers et al Med Phys 2011 analysis vs. material vs. configuration

A collaborative process

1. Distillation the problem 2. Method design 3. Process validation 4. Transfer of knowhow

eg. Tools (seg/LB)

eg. Methodology (mylar/carbon fibre) Workflow specimen Alter specimen state CT scan

FIB

MIP 3D registration 2D microscopy

Phase segmentation 2D registration

Network Pore / grain analysis labelling

Topological / structural analysis Digital process Physical process Simulation of Petrophysical properties Calibrated step • Diamond coring in situ • Water plasma specimen • Cryo-cell • saturation • Tri-axial cell (multiphase, • Wire cutting Alter specimen compaction, scCO2, state temperature) CT scan • Xenon labelling Hardware • Helical ex situ • nanoCT • Oil-in-place • Long scan FIB • Wettability alteration • E-sensitive CT • Surface/mineral labelling • Phase retrieval MIP 3D registration Recon s/ware • Non-rigid rego 2D microscopy • Auto-focus • Feature libraries • Helical • SIRT • Dynamic Phase • Dual energy 2D registration • Interative • Multivariate approach segmentation • Informed Segmentation • BET Surface area Network Pore / grain analysis labelling • Grain tracking • Mineral Liberation Analysis

Topological / structural analysis • Morse Theory Production Development • upscaling Simulation of Petrophysical properties In Design • Multiphase flow The lasting benefits

• curve-outs preserved: research and service • <25% Govt funded • 11 CT Facilities sold • 10 FTE for >10 years • >160 publications 2012 • 9 patents • 2 endowments

Endowments

Stjepan Marcelja Proof-of-concept fund Visiting Fellowship scheme CTLab

• 200 m2 space with visualisation room & 3 colour 3D printers • Five 180kV CT (1 nano-, 3 micro, 1 meso-CT) • 300kV CT large body scanner (mid-2016) • SEM-based mineral mapping, petrophysical mapping • Sample prep equipment • Dedicated 10Gbit/s pipe to NCI

National Computational Infrastructure CTLab Acknowledgements

• AFOSR and ANFF • Australian Research Council • Australian Microcopy and Microanalysis Research Facility • National Computational Infrastructure (NCI) • NeCTAR • Education Infrastructure Fund • Digicore Constortium of companies • FEI

Invitation to collaborate