Cranfield University MSc Computational & Software Techniques in Engineering Dr. Peter Sherar Applied Mathematics & Computing Group School of Engineering Peter Sherar AMAC – March 2007

Download Report

Transcript Cranfield University MSc Computational & Software Techniques in Engineering Dr. Peter Sherar Applied Mathematics & Computing Group School of Engineering Peter Sherar AMAC – March 2007

Cranfield University
MSc
Computational &
Software Techniques
in Engineering
Dr. Peter Sherar
Applied Mathematics & Computing Group
School of Engineering
Peter Sherar
AMAC – March 2007
1
Cranfield University
Peter Sherar
AMAC – March 2007
2
Cambridge
Cranfield
Oxford
London
Peter Sherar
AMAC – March 2007
3
Bedford
Cranfield University Campus
Cranfield
Milton Keynes
Peter Sherar
AMAC – March 2007
4
How do we do it?
100
90
 83% of Cranfield’s
income is from sources
other than government.
80
– Highest for any UK university
%
70
– also strong links to
government
60
50
Peter Sherar
 Strong links with
industry
Cranfield Oxbridge Imperial
AMAC – March 2007
5
Teaching @ Cranfield
 UK’s only postgraduate only university
– teaching diploma, MSc and PhD
 Over 20 years experience of European double degrees
– in partnership with ESTIA for ~16 years
 One of Europe’s top MBA programmes
– accredited in Europe, UK & US
 Queen’s Award for Education
 Employment : 97% (graduates in employment within 6 months)
– one of top 3 UK universities for graduate employment
~ 1 in 10 UK postgraduate engineers is Cranfield trained
Peter Sherar
AMAC – March 2007
6
MSc Computational &
Software Techniques in Engineering
Specialist options in
– Computer Aided Engineering (CAE)
“using software to produce and evaluate engineering solutions”
– Digital Signal and Image Processing (DSIP)
“processing signals / images to extract information”
– Grid Computing & e-Engineering (GCEE)
“Internet 2!, the computational grid - sharing large scale computing
resources across the internet”
Peter Sherar
AMAC – March 2007
7
Course Structure
 Core modules:
– Programming (C, C++, Java) & Software Engineering
– Visualisation & GUI’s
– Computational Methods
– Management for Technology
CAE modules:
Geometric Modelling
Finite Element Analysis
Finite Difference Analysis
CAD/CAM Applications
CAD/CAM Advanced Apps
DSIP modules:
GCEE modules:
Signal Analysis
Digital Signal Processing
Digital Image Processing
DSIP Applications
Computer Vision
Grid Fundamentals & e-Science
Grid Middleware
High Performance Computing
Grid Computing Infrastructures
Grid Development – e-Engineering Apps
 Emphasis : developing practical engineering skills for CAE/DSIP/GCE
Peter Sherar
AMAC – March 2007
8
Research : Signal Processing
 Vibro-acoustic Analysis
– Fault detection & Condition Monitoring
e.g. Compressor Blades (TV3-117 Engine)
Without Crack
With Crack
Rolls Royce
Peter Brotherhood
Caterpillar
Peter Sherar
AMAC – March 2007
9
Research : Signal Processing
 Vibro-acoustic Fault
detection & Condition
Monitoring
Peter Sherar
AMAC – March 2007
10
Automotive Vision
Gaze Camera
Forward looking camera
Dashboard view camera
Touch screen recording
Radar
Foot Camera
 Task : enhance driver
sensing capabilities
– improve situational
awareness
– maintain concentration
– hazard notification
 Application
– safety
– improved driver ability
hazardous conditions
Peter Sherar
AMAC – March 2007
11
Real-time IR Enhancement
Raw IR Image
Global Equalisation
Near Foreground
Enhancement
 Improving Visual Clarity
– greater road awareness
– increased hazard warning
Depth Relative Enhancement
Peter Sherar
– “valued added IR”
AMAC – March 2007
12
MSc Thesis Project
 MSc = 50% taught modules + 50% thesis project
– thesis project is industry sponsored
 Past company sponsors include
Peter Sherar
In UK:
In France:
Sony
Jaguar Racing
Delcam (CAD/CAM company)
GKN Technology
Matra Datavision
AspenTech
BAe Systems
INBIS
(Assystem)
Texas Instruments
Turbomeca
Dassault Systemes
Alstom
Renault Trucks
EADS Astrium
Cap Gemini
Snecma Moteurs
Airbus
Eurocopter
AMAC – March 2007
13
Graduate Employment
 Employment prospects are very good
– Cranfield : 97% of students within 6 months
– demand for CAE / DSIP/GCEE engineers strong
 MSc provides many transferable skills
– software development, project management (thesis), presenting skills,
engineering design, technical writing, …..
 Past students currently employed with:
–
–
–
–
–
–
Peter Sherar
BAe Systems
EDS Unigraphics
Sony
Cap Gemini
Flow3D
Earth Observaion Sciences
Nortel Networks
Oracle Consulting
Lucent Technologies
Matra Datavision
Delcam
GKN Technology
AMAC – March 2007
Logica
….
14
MSc: Computational & Software Techniques in
Engineering
options in:
Computer Aided Engineering
Digital Signal & Image Processing
Grid Computing & e-Engineering
Student Thesis Project Examples
64-bit ACM Porting Project
Project Objectives
Outcomes and Benefits
 To undertake a practical investigation into the
technical issues involved in porting a large mixed
language engineering modelling software application
(ACM) to the 64-bit Intel/AMD architectures
 Increase the size, complexity and accuracy of
simulations that can be undertaken:
 Help increase sales and consulting
 Provide a strategy for moving mixed language
engineering applications from 32-bit to 64-bit
 To convert a key part of ACM to 64-bit with
performance comparisons for internal evaluation.
Dr LAWRENCE DANIELS
Principle Technologist, AspenTech
“Moving to 64-bit is of strategic importance. Utilising
Cranfield staff, students and facilities to analyse the
problems involved and provide some real technical
solutions in this area has proved valuable for the
company.”
DULCENEIA BECKER
AMAC student
“Being exposed to some real software issues facing an
engineering IT company has allowed me to apply
technical knowledge to an area that is receiving much
attention in the commercial world. Interactions with
engineers working onsite and using the latest software
tools provided valuable insight into the constraints of
the problem, overall a very positive experience!”
Dr Peter Sherar
Cranfield University
Academic Supervisor
“Working closely with our industrial partner and applying
expertise and knowledge built-up in the department to
help find solutions to improve performance of the
company’s process simulation software has been a
rewarding experience for all involved.“
Author: Dulceneia Becker, 2006
Supervisors: Dr Peter Sherar,
Dr Lawrence Daniels
Thesis projects by: Laurent Muller
Supervisor: Peter Sherar
MSc Computational & Software Techniques in
Engineering,
option Digital Signal and Image Processing, 2006
High Scan Rate Telemetry Thermocouple Calibration Software
Real engine telemetry system sending
temperature data from rotating blades to
the casing through ultra high frequencies
Functional Engine Simulator
Simulator to test transmission
or
160 temperature values per sample
33 samples per second
14 bits resolution
Generic software based engine simulator capable of replicating
the electric signals sent by many different engines. Data
simulated is related to shaft speeds, temperatures, vibrations,
etc…
Receiver
The purpose of such a system is to have an easy to use
environment to develop and the validate different types of engine
test sets in a simulated environment prior to the last validation
steps on a real engine.
Morphing of Complex Composite
Geometry
Objective: Using a morphing process developed by Delcam to Tools:
create an application solving two problems encountered by  PowerSHAPE
shoe designer:
 ACIS kernel
1. Changing the toe spring of a shoe model
 C++/Visual Studio
2. Making a shoe model fit a new shoe last
Toe spring model with cubic S-decay
curve (factor 0.3)
PowerSHAPE user interface
GUI developed around the ACIS kernel for
input and display of results
Morphing the shoe model with a
pair of curves
Main features of a shoe last
Author – Olatokunbo Ogundana
Supervisor – Dr Peter SHERAR
MSc: CSTE, CAE option, 2004
Elastomeric Fatigue Crack Growth
– Integration into Abaqus
Project objectives:
 To establish a general methodology
for the automation of model remeshing
for elastomeric fatigue crack growth
and to compute the solution for a
extension test piece with a crack in it.
Crack area
Extension
Loading
Extension
Loading
2D Simplification
 Material : Rubber  Hyperelastic
behaviour (no analytical solution)
Zone1
Zone2
Zone3
Transition blocks
Mesh generation
Final Batch process
Pre-Processor
Mesh Generation
Strain energy density
around the crack tip
Abaqus processing
Abaqus job
Pre-processing
Data extraction
Post-processing
Post-Processor
 Calculates the new crack length and
configuration
Data storage
 Abaqus produces a binary file
containing the output data
 Postprocessor reads from data extractor
 We are interested in strain energy
density around the crack tip :
this data must be extracted
 Writes all the corresponding information into
a file, as well as all other information required
by Pre-Processor for next iteration
Author - Astride AREGUI
Supervisor – Dr A Zafrani/Dr S Leefe
 A simple FORTRAN subroutine
executes this task and writes the
result to a file
MSc in Computational & Software Techniques in Engineering,
Computer Aided Engineering option, 2003
Modelling of the Dispersion of
Gases Using Computational Fluid
Dynamics
MSc Computational and Software Techniques in Engineering
OBJECTIVE
The object of this work is to investigate a real scenario involving the accidental release of
methane from a pipeline conveying natural gas flow. The commercial code ANSYS CFX 10.0
was used.
INTRODUCTION
A three-dimensional Computational Fluid Dynamics (CFD) approach was applied to simulate
the mean flow field and tracer dispersion of different gases at the temperature of 25 C in the
vicinity of an idealized cubical building. Air and methane alone, as well as a mixture of these,
were investigated in three scenarios. Time-dependent external flow was used to model these
simulations using a buoyancy parameter.
 Two different geometries
 Two different gases: Air & Methane
 One model: Shear Stress Transport (SST) and
 Three different grids
Representation of the dispersion of air & rising methane from a
source using vectors
NAVIER-STOKES (N-S) EQUATIONS FOR VISCOUS FLOW
These equations are the mathematical formulas that computers solve
and the the entire science of viscous flow theory has been developed
based on these.
Mass Equation:

 div( u )  0
t
X, Y, Z Momentum Equations:
 (  ui )
p
 div (  u i u )  
 div ( grad u i )  S Mj 
t
j
Internal Energy Equation:
CONCLUSION
The mixture of air and
methane produced the highest
maximum pressures. Air alone
created approximately 50%
more maximum suction
compared to methane alone.
The case of two-face inlet
showed that the main mixing
effect starts taking place past
the cubical box, continuing
towards the outlet.
Furthermore, the boundary
layer thickness increased at
about 15 % of the box
surfaces, and after that
distance it was subjected to
transition leading to boundary
layer separation and
recirculation.
 ( i )
 div( iu)   pdivu  div( kgrad T )    S i
t
Representation of the main mixing effect using velocity contour
Thesis By: Nikolaos Spyridon Annitsakis
Supervisor: Professor Chris Thompson
Date: September 2006
New Transforms for Non-stationary Signals with
Non-linear Variation of Instantaneous Frequency
Project objectives:
 To create a new generic transform that can match any nonlinear
variation of instantaneous frequency.
 Such a transform is important for transient signals from radar, sonar
and mechanical systems. This problem has not been investigated in
literature.
The signal
Instantaneous
frequency
The Wigner
distribution
The new sine-Wigner
transform
 To develop a new adaptive time-frequency transform for sinusoidal
or exponential variation of instantaneous phase of a complex signal.
 To investigate the chirp-Wigner transform in the case of arbitrary
polynomial variation of the instantaneous frequency.
The new generic transform:


W (t , f ; μt )   x (t  ) x(t  ) K  ; μt  e
d
2
2

 j 2 f 

g
g

The new sine-Wigner transform:
Ws (t , f ; μt ) 




 
 
 x (t  2 ) x(t  2 )exp  j 2   cos2 t sin        e

 j 2 f 
d

The new exponential-Wigner transform:
Ws (t , f ; μt ) 



 x (t  2 ) x(t  2 )exp  j 2    exp  t 2 sinh 


2       e  j 2 f  d
Jeremy D Gould
MSc: Computational & Software
Techniques in Engineering, DSIP
option, 2006
Supervisor: Professor L Gelman
The new generic adaptive time-frequency transform and new
high order spectrum (HOS) technique
FOR NON-STATIONARY SIGNALS WITH NON-LINEAR VARIATION
OF INSTANTANEOUS FREQUENCY
High Order Spectra analysis based on the new transform
applied to fatigue crack detection
UNCRACKED BEAM. ALPHA = 0.90
Fisher value
Exponent
bicoherence
Author: Esteban Lapeña
Supervisor: Professor L.Gelman
Classical
bicoherence
Gain
MSc:9.24
Computational
& Software
Techniques
E+04
6.85
E+04
1.35 in Engineering,
Digital Signal and Image Processing option, 2006
QinetiQ
Site Security Using High Resolution
Radar
Cranfield University - School of Engineering
Aim: The project set out to develop an existing
Networking: The original radar was a stand-
radar system to improve its performance as a
site security radar system
alone system. The software was extended to
allow sending of target information from
multiple radars to a single display.
Clutter Handling: The original system did not handle clutter effectively. A normalisation process was added which
caused static objects to be normalised out, leaving only changes in the scene as bright regions.
The figure on the
right shows results
from the previous
system. Much of
the image is
masked out, and
the circled person
is difficult to
identify.
The figure on
the right shows
performance
after applying
normalisation.
No masking is
required, and
the person is
clearly visible.
Author: Christopher Anstey, Supervisor: Peter Sherar
MSc: Computational & Software Techniques in Engineering, 2006
Sony AVIO Digital TV Project
Project background, objectives






Access to the channel list, service description via a
remote control.
Provide an audible rendition of the service.
Creation in JAVA of a User Interface.
Feasibility: Human factors
For visually impaired people
Flexibility for
- the developer: improve the service
- the user: configuration
29/08/2002
24
AMAC, Amiens Teaching Hospital, UTC
Objectives
To optimise the structure of an edge detection
algorithm.
Techniques employed
Preprocessing thresholding & morphological
opening
B-spline fitting from radial sampling
Cohen’s snake balloon model iteration
Optimisation of an Edge Detection
Algorithm for Echocardiagraphic Images
Software interface showing application of gradient
vector flow snake algorithm
Optimised algorithm flow chart
Examples of good detections
Author – Alexandre SALVADOR
Supervisors – Dr Shan FU, Jean-Francois LERALLUT
MSc - Computational & Software Techniques in Engineering,
option Digital Signal and Image Processing, 2003
Department of Retinal Screening
Cheltenham General Hospital
Detection of Diabetic Retinopathy in
Digital Fundus Images
Project Background
 In the Gloucestershire Diabetic Eye Screening Service,
experts in the Department of Retinal Screening at
Cheltenham General Hospital are currently responsible for
the checking and grading of images of diabetic subjects in the
area.
 There is a need for an automated grading system.
 Image Processing is one area of study that could form the
basis for computer assisted automated detection of Diabetic
Retinopathy.
Objectives
 To investigate methods based on based on filter matching
and mathematical morphology for extracting and analysing
pixel based and region based features.
 To develop a software application with a user interface for
the purpose of displaying and manipulating digital colour
fundus images.
Graphical user interface
Author: Jennifer Rabone
MSc: Computational & Software Techniques in
Engineering (DSIP option)
Histogram Equalisation/Gradient Supervisor: Dr Shan Fu
Mask Example
International Coatings Ltd., Newcastle.
Surface Deformation Analysis in the
Investigation and Detection of Coating Cracks
Context
Example of uncoated T-girder courtesy
of International Coatings Ltd.,
Newcastle.
2D- Digital Image Correlation
 T-girder components
used in construction of
ballast hulls for freightcontainer vessels at sea.

 In-situ stresses on Tgirders lead to cracks
forming on surface of
protective coating applied
during production by
International Coatings Ltd,
Newcastle.

 Digital image
processing techniques used
to investigate cause of
cracks on T-girder before
and after protective coating
is applied to establish link
between strain distribution
and subsequent coating
cracks.
Author: Stephen J Learmonth
Supervisor: Dr Shan Fu
Used to measure surface
displacement between digital images
of object surface before and after
deformation.
Limited to integer-valued pixel
displacements.
Measuring Surface Strain
Involves

Sub-pixel interpolation using bi-cubic
spline function.
 Image pattern intensity matching
using
normalised cross-correlation
function.


Iterative recursive search
algorithm using Newton-Raphson
method to estimate strain parameters.
Mapping measured displacement/
strain distribution to object surface
Next
MSc Computational and Software Techniques in .Engineering, 2004.
International Coatings Ltd., Newcastle.
Surface Deformation Analysis in the
Investigation and Detection of Coating
Cracks
Experimental Setup of Optical Imaging System
CCD
camera
52mm lens
attachment
micrometer-induced
deflection
Author: Stephen J
Learmonth
Supervisor: Dr Shan Fu
upright
plate
welded region
base plate
region where
surface
displacement
has occured
due to out-ofplane
movement
welded-region
viewing
perspective
T-girder sample

T-girder fixed securely to optical table with
micrometer induced stress applied to upright
plate.

CCD Camera with 1394 FireWire interface to PC
and 52mm lens capturing the welded-joint region
of the T-girder.


Welded-region of T-girder captured
by CCD camera for analysis.
Cross-correlation reference grid
superimposed on region of interest.
MSc Computational and Software Techniques in Engineering, 2004.
back
A Grid Service Based Architecture for Persistent Online Applications
The idea: Persistent programs are very complex, the work can be divided in
autonomous layers (E.G.: Advertisement, Weather model, AI…). Each layer
being a kind of parallel generic GRID services run by specialized companies
Two main qualities:

Enable to outsource the development work

Enable to outsource runtime work => more scalability
Implementation: Development of a demonstrator game using two Grid
services: Advertisement & Artificial intelligence
Thesis by:
Mikaël Wozniak, 2006
Supervisor:
Prof. F Wang
MSc: Computational &
Software Techniques in
Engineering, option Grid
Computing & e-Engineering
The three main entities involved and their roles:
The players’ program (C++/OpenGL/windows):
Do the graphical work and listen to Users’
input event (mouse)
Connect to the game producer’s program
for knowing other players’ coordinates and
for using Grid service features
The game producer’s program (Java/UNIX):
Answer to players’ requests
Outsource job to Grid services
Aggregate the work done by Grid services
Here is how the demonstrator game works with the 2 Grid services:
…
…
The Grid services (Java/UNIX)
Handle a narrow category of tasks
Answer to the game producer’s program’s
requests
Background
Storage Web Service Development
using Globus Toolkit 4
 Grid Computing and e-Science is an innovative extension
of distributed computing technology.
 It allows participants of a virtual collection of organisations
to share computing resources.
Objective
 Using Globus Toolkit for configuring the grid environment,
design and implement a new web storage service.
Adopted design for web storage service
Main client dialog window
Transfer options dialog
Design advantages:
 Service providers get direct control over the services
FTP Integration
 Local service deployment
 Every client can create a new instance of the service
There can be unlimited number of instances
Bill preview dialog
Author – Lukasz GRYZBON
Takes a lot of work away from administrators Supervisors – Prof Frank Wang
MSc - Computational & Software Techniques in
 Simple log file for service usage
Engineering, option Grid Computing & e-Engineering, 2006