Hardip Bhesaniya

The mission

Proton–boron-11 fusion is one of the most attractive long-term routes to clean energy: its fuel is abundant, and its energy is released as charged alpha particles rather than neutrons, which could allow that energy to be converted directly into electricity. It is also one of the hardest fusion reactions to achieve.

HBF Network was founded to work on this problem step by step — starting with rigorous computational foundations, and moving towards an experimental programme with university and industry partners. HBF Scientific Lab is the first result of that work: an interactive, inspectable toolset that makes the physics of the problem measurable and easy to explore.

What I build

Scientific computing

The calculation engine behind all seven laboratories: reaction-rate integration, relativistic reaction kinematics, magnetic mirror and toroidal field models, a plasma equilibrium solver, relativistic particle-orbit integration, detector image analysis and electrostatic energy-recovery models.

Scientific data

Selecting and integrating the reference datasets — the Sikora–Weller cross-section evaluation, AME2020 atomic masses, Bosch–Hale reactivity data and IAEA nuclear records — so every result can be traced back to its source.

Verification

Benchmarking each calculation against exact solutions, published reference values and separately written reference code, with accuracy targets set before each comparison.

Interface and visualisation

The complete user experience: the seven laboratory workspaces, interactive 3D scenes with a fallback renderer for computers without graphics acceleration, and a fast, private website that needs no server-side processing.

Areas of expertise

• Reaction-rate calculationThermal averaging of measured cross sections with careful interpolation and exact conversion between laboratory and centre-of-mass frames, checked against an analytic reference to a relative accuracy of about 10⁻¹².
• Relativistic kinematicsEnergy release, reaction thresholds and energy sharing between reaction products, with a three-body event generator that conserves energy and momentum exactly in every event.
• Charged-particle orbitsRelativistic orbit integration in magnetic fields, conserving particle energy to about 10⁻¹⁴ over 100 orbits and converging at the expected rate as the time step is refined.
• Plasma equilibriumA second-order solver for the Grad–Shafranov equation, confirmed to converge at its designed order and to agree point by point with a separate Python reference implementation.
• Image analysisAutomatic thresholding, particle-track detection and measurement, with exact statistical confidence intervals for counts.
• Reproducible softwareAccuracy targets fixed in advance, comparisons against separate reference codes and exact solutions, deterministic random seeds and results that reproduce identically from run to run.

How I work

Set the standard first. Every benchmark has its accuracy target defined before the comparison is run.

Check against separate references. Important results are compared with something written separately — an exact solution, a published value or a reference program in another language.

State the scope clearly. Every model documents what it includes and what it leaves out, so results can be interpreted correctly and built on with confidence.