Research interests include:
- Nuclear reactor thermal hydraulics
- Nuclear Fusion
- Hybrid, coupled numerical methods
- Electromagnetic, acoustic and elastodynamic wave propagation
- Boundary integral equations
- Finite element methods
- Applications of differential forms to numerical methods
- Multipole methods for fast numerical solutions of discrete problems
- Model based event correlation in networks
As is the case with the majority of academics at Imperial College, I have a strong interest in research and am a member of the Nuclear Engineering research focus area (Computational Mechanics) in this department. This group conducts research into nuclear thermal hydraulics and neutronics and computational methods for a wide range of engineering disciplines including nuclear engineering, solid mechanics, heat transfer, acoustics, electromagnetics.
During my time at Imperial I have been involved in all of these areas and have initiated much of this work. As an example, I developed some of the first implicit techniques for the solution of time dependent integral equations in acoustics and electromagnetics. This seminal work has subsequently initiated interest internationally and a number of groups are now heavily involved in the study of these and related methods.
I have been involved in the identification of numerous projects in recent years, some of which are listed below:
Keeping the nuclear option open (knoo)
In collaboration with colleagues in Chemical Engineering, we are undertaking a series of coupled projects studying the reflood phase of large break loss of coolant accidents in PWR's. At the more 'applied' end of the spectrum, we are developing mechanistically coupled models of flow patterns within a reflooding core as the core geometry is altered by the creep of the cladding. We are here working collaboratively with the US NRC, coupling their system thermal hydraulics code TRACE to the British Energy MABEL fuel structural mechanics models. This is underpinned with more fundamental experimental and computational studies.
Computation for advanced reactor engineering (CARE)
Application of Differential Forms to Computational Electromagnetics
This work provides a framework for the construction of successful approximations of arbitrary degree for all numerical schemes. Conventionally, such approximations are arrived at entirely heuristically and result in low-order simplistic representations of the field. This work was of such novelty that some years of mathematical effort were required before tangible results could be reported.
Advanced Methods in Computational Electromagnetics
This project involved the development of advanced high-order finite element methods for anisotropic materials. The sophistication of finite element methods for electromagnetics lags significantly behind that of analogous techniques in the study of, say, stress analysis. This work has sought to address this failing.
Massively parallel computation for Computational Mechanics
Field computations are very expensive in terms of computing resources. One approach to this problem is to employ large clusters of processors and software has been developed to take advantage of the potential performance gains offered. Codes have been installed on a range of massively parallel platforms including those at BAE Systems.
Algorithm developments to reduce cost scaling of Computational Wave Propagation problems
As stated earlier, such field computations are expensive. Indeed, computational costs conventionally increase with the sixth power of the incident excitation frequency and this places great limitations on the size and sophistication of the models that can be studied. I have developed significant advances in the implementation of integral equations which can reduce the cost-scaling considerably, even down to the third power of frequency. This has resulted in practical calculations needing small fractions of the resources of more conventional schemes.
Development of Integral Equation methods for Time Domain Elastodynamics
The development of time domain integral equation methods in acoustics and electromagnetics in which I have been involved are also applicable to elastodynamics. Stable implicit schemes have been developed under my supervision for application to problems arising in areas such as non-destructive testing.
Fast Multipole Methods for Elastostatics
Accelerated methods developed initially in acoustics and electromagnetics can also (in principle) be applied to solid mechanics. I have been involved in the development of a Taylor series based fast multipole method for the solution of elastostatic problems. The aim of this work is to extend current capabilities in terms of model size and complexity.
Computational Modelling of Corrosion Protection Systems
The corrosion of marine vessels is a complex problem involving electrostatics and electro-chemical kinetics. I have worked on a project in collaboration with Frazer-Nash plc for the development of novel methods for the prediction of corrosion on a variety of marine structures. This has led to a massively improved modelling capability, hugely increasing the complexity of problem that can be analysed.
The investigation of the projects listed above is not a lone task. An essential part of each project is the supervision & training of staff. Notably, I have supervised or co-supervised ~15 PhD students and postdoctoral researchers.
The results of this work are not merely of academic interest. This research has led directly to the development of software products purchased by the likes of the MoD & BAE Systems. The provision of industrial strength software is not normally associated with academic institutions and the development of quality assured software requires a high degree of management and coordination of our research staff. Some years ago, a colleague and myself instigated a set of procedures and controls by which all software development was to be carried out. This framework enabled high quality software to be developed which met the stringent demands of our customers. Indeed in some ways our development procedures were more demanding than those required by the various NATO and ISO standards.
Emeritus Professor Geoffrey F. Hewitt, Keeping the Nuclear Option Open
Dr Walker, Computational Mechanics
Professor Sir John Pendry, The Extraordinary Behaviour of Light
Research Student Supervision
CHATZIKIRIAKOU,D, Keeping the Nuclear Option Open
KEDDIE,A, Computational Modelling of Corrosion Protection Systems
MCGOURTY,G, Advanced Computational Electromagentics
MCKINNELL,J, The Extraordinary Behaviour of Light
REFIG,A, Application of FMM to photonics