ignite your...supercomputing 24 jul12_v2
DESCRIPTION
Presenter - Prof Simon J. Cox from the Computational Engineering Design Research Group (CED) with describe The use of Super Computers for Design optimisation. The CED is a centre of excellence for multi-disciplinary engineering simulation and design which combines together a range of analytical, computational, and experimental techniques.TRANSCRIPT
Supercomputing amp Design
Professor Simon Cox Associate Dean Enterprise Faculty of Engineering and the Environment 24th July 2012 sjcsotonacuk
Faculty of Engineering and the Environment
Faculty of Engineering and the Environment
Aeronautics Astronautics and Computational Engineering
Prof Keane - Head of AACE Unit Prof Simon Cox Chair Computational Engineering and Design Group Associate Dean Enterprise
Faculty of Engineering and the Environment
AACE Overview
bull Three research groups (aeronautics amp flight mechanics astronautics and computational engineering amp design) Head of Unit Prof Keane
bull 30 members of academic staff holding pound13m of research grants
bull Three major research centres (Airbus Noise Technology Centre R-R UTC for Computational Engineering Microsoft Institute for HPC)
bull Nearly 80 PhD and EngD students under supervision (major partner in DTC for Complexity)
bull Linked to highly successful undergraduate and MSc teaching programmes (100 student satisfaction in 2008 National Student Survey for Aeronautics-Astronautics course Royal Academy of EngineeringExxonmobil Excellence in Engineering Teaching Award)
3
Faculty of Engineering and the Environment
Design Exemplars Computational Engineering and Design Research Group
Faculty of Engineering and the Environment
5
Rolls-Royce UTC in Computational Engineering
PIs Keane amp Scanlan 5 Academic staff 9 Research fellows 11 EngDPhDs pound12m per annum external funding (EUUK GovtEPSRC) Close integration with Rolls-Royce RampT objectives
Faculty of Engineering and the Environment
Typical Design Improvement Flowchart
6
Background
Effort
CAD
Meshing
CSM
Post
Processing
Results
Extraction
Optimiser
Geometry
Rule
Bases
Final
Design
Boundary
Conditions
Parametric
Base
Geometry
CFD
Initial
Design
Search
Uncertainty
Models
Costs PI Keane
Faculty of Engineering and the Environment Multi-fidelity Optimization
bull 12 geometry variables
bull 10 full car RANS simulations 15h each
bull 120 rear wing only RANS simulations 15h each
7 PI Keane
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Faculty of Engineering and the Environment
Aeronautics Astronautics and Computational Engineering
Prof Keane - Head of AACE Unit Prof Simon Cox Chair Computational Engineering and Design Group Associate Dean Enterprise
Faculty of Engineering and the Environment
AACE Overview
bull Three research groups (aeronautics amp flight mechanics astronautics and computational engineering amp design) Head of Unit Prof Keane
bull 30 members of academic staff holding pound13m of research grants
bull Three major research centres (Airbus Noise Technology Centre R-R UTC for Computational Engineering Microsoft Institute for HPC)
bull Nearly 80 PhD and EngD students under supervision (major partner in DTC for Complexity)
bull Linked to highly successful undergraduate and MSc teaching programmes (100 student satisfaction in 2008 National Student Survey for Aeronautics-Astronautics course Royal Academy of EngineeringExxonmobil Excellence in Engineering Teaching Award)
3
Faculty of Engineering and the Environment
Design Exemplars Computational Engineering and Design Research Group
Faculty of Engineering and the Environment
5
Rolls-Royce UTC in Computational Engineering
PIs Keane amp Scanlan 5 Academic staff 9 Research fellows 11 EngDPhDs pound12m per annum external funding (EUUK GovtEPSRC) Close integration with Rolls-Royce RampT objectives
Faculty of Engineering and the Environment
Typical Design Improvement Flowchart
6
Background
Effort
CAD
Meshing
CSM
Post
Processing
Results
Extraction
Optimiser
Geometry
Rule
Bases
Final
Design
Boundary
Conditions
Parametric
Base
Geometry
CFD
Initial
Design
Search
Uncertainty
Models
Costs PI Keane
Faculty of Engineering and the Environment Multi-fidelity Optimization
bull 12 geometry variables
bull 10 full car RANS simulations 15h each
bull 120 rear wing only RANS simulations 15h each
7 PI Keane
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
AACE Overview
bull Three research groups (aeronautics amp flight mechanics astronautics and computational engineering amp design) Head of Unit Prof Keane
bull 30 members of academic staff holding pound13m of research grants
bull Three major research centres (Airbus Noise Technology Centre R-R UTC for Computational Engineering Microsoft Institute for HPC)
bull Nearly 80 PhD and EngD students under supervision (major partner in DTC for Complexity)
bull Linked to highly successful undergraduate and MSc teaching programmes (100 student satisfaction in 2008 National Student Survey for Aeronautics-Astronautics course Royal Academy of EngineeringExxonmobil Excellence in Engineering Teaching Award)
3
Faculty of Engineering and the Environment
Design Exemplars Computational Engineering and Design Research Group
Faculty of Engineering and the Environment
5
Rolls-Royce UTC in Computational Engineering
PIs Keane amp Scanlan 5 Academic staff 9 Research fellows 11 EngDPhDs pound12m per annum external funding (EUUK GovtEPSRC) Close integration with Rolls-Royce RampT objectives
Faculty of Engineering and the Environment
Typical Design Improvement Flowchart
6
Background
Effort
CAD
Meshing
CSM
Post
Processing
Results
Extraction
Optimiser
Geometry
Rule
Bases
Final
Design
Boundary
Conditions
Parametric
Base
Geometry
CFD
Initial
Design
Search
Uncertainty
Models
Costs PI Keane
Faculty of Engineering and the Environment Multi-fidelity Optimization
bull 12 geometry variables
bull 10 full car RANS simulations 15h each
bull 120 rear wing only RANS simulations 15h each
7 PI Keane
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Design Exemplars Computational Engineering and Design Research Group
Faculty of Engineering and the Environment
5
Rolls-Royce UTC in Computational Engineering
PIs Keane amp Scanlan 5 Academic staff 9 Research fellows 11 EngDPhDs pound12m per annum external funding (EUUK GovtEPSRC) Close integration with Rolls-Royce RampT objectives
Faculty of Engineering and the Environment
Typical Design Improvement Flowchart
6
Background
Effort
CAD
Meshing
CSM
Post
Processing
Results
Extraction
Optimiser
Geometry
Rule
Bases
Final
Design
Boundary
Conditions
Parametric
Base
Geometry
CFD
Initial
Design
Search
Uncertainty
Models
Costs PI Keane
Faculty of Engineering and the Environment Multi-fidelity Optimization
bull 12 geometry variables
bull 10 full car RANS simulations 15h each
bull 120 rear wing only RANS simulations 15h each
7 PI Keane
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
5
Rolls-Royce UTC in Computational Engineering
PIs Keane amp Scanlan 5 Academic staff 9 Research fellows 11 EngDPhDs pound12m per annum external funding (EUUK GovtEPSRC) Close integration with Rolls-Royce RampT objectives
Faculty of Engineering and the Environment
Typical Design Improvement Flowchart
6
Background
Effort
CAD
Meshing
CSM
Post
Processing
Results
Extraction
Optimiser
Geometry
Rule
Bases
Final
Design
Boundary
Conditions
Parametric
Base
Geometry
CFD
Initial
Design
Search
Uncertainty
Models
Costs PI Keane
Faculty of Engineering and the Environment Multi-fidelity Optimization
bull 12 geometry variables
bull 10 full car RANS simulations 15h each
bull 120 rear wing only RANS simulations 15h each
7 PI Keane
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Typical Design Improvement Flowchart
6
Background
Effort
CAD
Meshing
CSM
Post
Processing
Results
Extraction
Optimiser
Geometry
Rule
Bases
Final
Design
Boundary
Conditions
Parametric
Base
Geometry
CFD
Initial
Design
Search
Uncertainty
Models
Costs PI Keane
Faculty of Engineering and the Environment Multi-fidelity Optimization
bull 12 geometry variables
bull 10 full car RANS simulations 15h each
bull 120 rear wing only RANS simulations 15h each
7 PI Keane
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment Multi-fidelity Optimization
bull 12 geometry variables
bull 10 full car RANS simulations 15h each
bull 120 rear wing only RANS simulations 15h each
7 PI Keane
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Multi-scale modelling for nanotechnology
bull Hans Fangohr
bull Nanotechnology length scales are close to the atomic lattice spacings
bull Need to consider atomistic nature of matter to be accurate
bull Need to combine atomistic simulations with continuum approaches to make simulations feasible
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment Multi-physics modelling
for nanotechnology bull Relevant properties
bull magnetic
bull thermodynamic
bull conducting
bull mechanic
bull Optical
bull Crucial for Nanodevices to
consider these together
PI Fangohr
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Design Optimisation of Nanodevices
bull Magnetic storage sensors spintronics
bull Prototype fabrication expensive vs ldquocheaprdquo Simulation
bull Requires
ndash multi-scale simulations
ndash multi-physics simulations
bull Computational Design Optimisation highly beneficial PI Fangohr
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Multi-objective multi-disciplinary design (coronary stents)
Balloon expansion structural integrity Haemodynamics re-endothelialisation
Optimal design patient specific
Systematic design search and optimisation
Recover and maintain healthy blood flow Drug elution
control of restenosis
Flexibility conformability
Stress minimise artery wall damage
PI Bressloff
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
(1) httpwwweverydaychampionsorghospitalcardiac_cath_labphp (2) Stent design tool (NWBressloff) (3) httprohitsaxenawriteswordpresscomauthorrohitklar (4) httpwwwraydiancecomour-technology
(1) Cath Lab
(3) The Cloud
(2) Stent Design Tool (a virtual cath lab)
(4) Manufacture
Future cath-lab
PI Bressloff
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Airbus Noise Technology Centre (AFM Group)
PI Zhang
bull First Airbus-University Technology Centre in the World
bull Opened November 2008
bull Focussing on future aircraft technologies for noise reduction
bull Fifteen academic staff and research students
bull Computation and experiment
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
LES of noise generated by interaction of aircraft engine with
wing (ZhangHu)
Aircraft high-lift device noise generation and control
(ZhangHu)
High-speed train aeroacoustics (Hu amp Thompson (ISVR) )
Sample ANTC projects (AFM Group)
PI Zhang
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Under the Hood hellip tools technologies and platforms
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment Microsoft Institute for HPC
bull PI Cox
bull Workflow + HPC + Data
bull Centre for Fluid Mechanics Simulation (CFMS)
bull Flight simulation HPC
bull μVis ndash X-Ray CT Centre workflow and data management
bull Cloud amp Mobile computing ndash space situational awareness
bull Institutional and Large Scale Data Management
ndash Materials
ndash Chemistry
ndash Medicine
ndash Archaeology
ndash Engineering HPC Data 16
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Engineering Design - from the Desktop to the Enterprise
17
bull Engineering design demands integration of complex systems to set-up execute monitor post-process review and store computational processes data and knowledge
bull Research to tackle new challenges
bull Applied and customised commodity off-the-shelf Microsoft tools technologies and platforms
bull Collaboration with Rolls-Royce Airbus BAE Systems MBDA Microsoft others 20 Million Euro UK TSB (DTI) programme
Microsoft Institute for HPC
Southampton
ldquoWe demonstrate why where and how current and
future Microsoft tools and technologies can make the
engineering design process faster cheaper and betterrdquo
Launched by Bill Gates Nov 2005
Initial Research 1998
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Corporate review
Small group review
Individual review
Semantic structure search and reasoning
Wiki
Concept Computation
Robust reliable and scalable data
intensive collaboration
Airbus
Software + Services for connecting
engineers and experts to users and data
BAE Systems
Orchestration of gas turbine design
calculations
Rolls-Royce
Data
Organizing thoughts and reference
material experience
TAGtivity
Windows Server 2008 Hyper-V RC0 Windows HPC Server 2008 Beta2 SQL Server 2008 CTP6 Office Communication Server
Microsoft Office 2007 SQL Server 2005 2008 Windows Presentation
Foundation Matlab
MediaWiki SQL Server 2008 D2R Server ARQSPARQL
Tagtivity database Filesystem
Wiki database Workflow database Knowledge database Corporate database
Task database Conversation database Workflow tracking Simulation database
Workflow templates Workflow tracking Filesystem
Windows Workflow Foundation Windows CCS 2003 Linux (Interop) Visual Studio 2005 SQL Server 2005
Sharepoint Server Active Directory HP-UX (Interop) Windows Communication Foundation
Sharepoint database Active Directory
Technology
Data sources
Data Driven Product Lifecycle
Management
MBDA
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Exploiting cloud computing for algorithm development
N OrsquoBrien E Hart S Johnston K Djidjeli S J Cox sjcsotonac uk and NeilOBriensotonacuk
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
20
Case study photonic crystals
bull Nanostructures that affect propagation of light
Wing cross-section
Peacockrsquos magnificent
tail
Opal
Man-made example
Man-made example
Morpho butterfly
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Strengths of cloud computing
bull Algorithm development
ndash Time- data- compute-intensive
bull Data dissemination
ndash Sharing large input or output data sets
bull Burst capability
ndash Testingparameter sweeping new revisions
bull Super-scalability
ndash Apply algorithm to big problems
21
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Case study photonic crystal modelling
bull Photonic crystals are periodic dielectric structures we assume periodicity in 119909 and 119910 directions and infinite extent in 119911
22
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Maxwellrsquos Equations
bull Maxwellrsquos equations give the allowed modes of propagation for TM TE modes after applying standard simplifications
minus1
120576∆120595 = 120582120595
minus120571 sdot1
120576120571120595 = 120582120595
bull When the dielectric 120576 varies no analytical solution exists
23
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Algorithms for Photonic Crystal Modelling
bull Plane Wave Methods
ndash Scale poorly + Gibbs Phenomenon
bull Finite Difference Method
ndash Simple to code
ndash Requires fine mesh for accuracy
raquo Large and Sparse Matrices
bull Finite Element Method
ndash Large amounts of code
ndash Requires fine (complicated) mesh for accuracy
raquo Large and Sparse Matrices
bull Meshless Method
ndash Simple to code amp data parallel
ndash Improved geometry handling
raquo Small and Dense Matrices 24
FD Stencil
Radial Basis
Function
FE Mesh
Gibbs Phenomenon
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Meshless Method Formulation
bull Maxwellrsquos equations for 2D problem
bull Simplifying the 2D elliptic Helmholtz equation takes the form
bull Introducing the following notation
bull Then use the Standard RBF approximation of a function to write the Helmholtz equation as a generalized eigenvalue problem
25
uuiiTM
uuiiTE
kk11
k1
k1
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Cloud application
bull Verification by checking a high-resolution run against existing software
ndash Developer workstation free to run other tasks
ndash Institutional cluster queue time avoided
bull Parameter sweeps through high dimensional spaces
ndash Wall-time bounded below by (provisioning time + longest simulation time)
bull Application of the new algorithm to novel shapes bandgap engineering and optimisation
26
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Results
bull Comparison of band diagrams
27
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Concluding remarks
bull Cloud paradigm demonstrated
ndash Accelerated development cycle
ndash Parts of a traditionally-serial workflow ran on Windows Azure in parallel
ndash Enhanced collaboration opportunities
ndash Modularity of design
bull New meshless algorithm demonstrated
ndash Solves Maxwell equations successfully
ndash Applicable for bio-inspired photonic materials
28
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Azure+Windows Phone 7 Atmospheric Science
Professor Simon Cox and Dr Steven Johnston with ASTRA Dr Andraacutes Soacutebester Prof Jim Scanlan and Neil OrsquoBrien Microsoft Institute for High Performance Computing Southampton sjcsotonacuk amp sjj698zeplerorg
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
bull Monitoring the atmosphere
ndash Weather pollution
ndash Volcanic ashhellip
bull Low cost metrological balloons
ndash High altitude
bull Instrument retrieval using WP7 + GPS
bull Predicting the landing location with Azure
Atmospheric Science Through Robotic Aircraft
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
ASTRA architecture
Windows
Azure
University of
Southampton
GSM
Followers
Internet
Windows
Phone 7 Windows
Phone 7
notifications
Flight
prediction
algorithms
and trajectory
updates
Tracker using
Bing maps
on WP7
Device
registration and
GPS logging
notifications
Blob
storage
Logging and
data storage
Meteorological
weather balloon
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
bull ASTRA 7
ndash 18237 meters (59832 feet)
ndash Top speed 145 kmh (90mph)
ndash 1hr 16rsquo
bull ASTRA 8
ndash 21600 meters (71000 feet)
ndash Top speed 82 kmh (51mph)
ndash 2hr 30rsquo
ASTRA flights using WP7 and Windows Azure
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
33
Take Offhellip
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
34
Windows Phone 7 on the edge of Space
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
35
Windows Phone trajectory
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
36
Windows Phone at -58 degrees C
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Thanks
We acknowledge and thank Microsoft and Microsoft Research for support Prof Simon Cox sjcsotonacuk
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
bull Atmospheric Science Through Robotic Aircraft (ASTRA)
httpwwwsotonacuk~astra (Main page)
httpsegozcoukStratosphericflightaspx (WP7 application)
httpwwwsotonacuk~astradiaryhtml
httpwwwmicrosoftcomshowcaseenusdetailsda17789f-7914-4131-b7cb-91937be3d20a (Video)
Further information
Faculty of Engineering and the Environment
Summary
Faculty of Engineering and the Environment
Summary