Technical investigations and engineering arguments: what the physics says, how the models behave, where the assumptions fail, and what the design implications are.
What it took to make Concorde carry passengers at Mach 2, from shock waves, delta-wing aerodynamics and supersonic intakes to aerodynamic heating, fuel-based trim and the sonic boom — and why building a successful supersonic airliner remains such a difficult systems-engineering problem.
An engineering investigation into how tokamak fusion reactors confine, heat, stabilise, and control hundred-million-degree plasmas—and what it takes to turn fusion into useful power.
Fusion requires plasma temperatures exceeding one hundred million kelvin—far beyond the limits of any material container. This article explores how tokamaks use charged-particle dynamics, helical magnetic fields, plasma current, MHD equilibrium, heating, stability, and turbulence control to construct a magnetic bottle capable of reaching fusion conditions.
Reaching fusion conditions is only the beginning. This article examines how a burning tokamak couples alpha self-heating, neutron transport, tritium breeding, superconducting magnets, plasma exhaust, state estimation, and multirate feedback control into a complete fusion power system.
High-fidelity simulation can analyze almost anything, but exhaustively exploring a modern design space is computationally impossible. This four-part series investigates how multi-fidelity surrogate models, active learning, and reinforcement learning solve this bottleneck—and why the future engineer must act as a guardian of physics rather than just an operator of software.
Modern engineering systems are governed by interactions between multiple physical domains. Using a pulse jet case study, this article explores why changing a single design variable can simultaneously affect fluid dynamics, combustion, thermal loading, and structural behaviour.
Traditional optimisation methods become impractical when every design evaluation requires an expensive physics simulation. This article examines why gradient methods, evolutionary algorithms, and conventional search strategies struggle within complex engineering design spaces.
A surrogate model that knows what it does not know is only useful if the optimisation system can act on that uncertainty. This article explores active learning, Bayesian optimisation, and multi-fidelity methods for intelligently allocating computational resources across complex engineering design spaces.
Autonomous optimisation systems can explore enormous design spaces, but they cannot determine whether their predictions represent physical reality. This final article examines model exploitation, uncertainty, validation, and the evolving role of engineers in trustworthy computational design.
Why QCW Dual Resonant Solid State Tesla Coils are considered the pinnacle of amateur resonant power electronics, and how a simple change in operating philosophy transforms the engineering challenge into a tightly coupled nonlinear control problem.