Why do solar panels hate the sun?

The problem, made visible.

I modelled the panel layer by layer (ETFE front sheet, EVA encapsulant, monocrystalline cells, PET, adhesive, carbon-fibre backing) and simulated it under solar load in ANSYS Fluent. Uncooled, the stack settles above 53 °C: an efficiency loss of 11–14% exactly when irradiance peaks.

Same geometry, same sun. The only difference is the fluid loop.

Designed part by part.

I built the panel and its integrated heat exchanger as native SOLIDWORKS parts and assembled them into a full PVT module, checking geometry, material arrangement, and manufacturability before any simulation ran. The mesh was verified for quality (skewness 0.617) before any results were trusted.

Then tested against a real postcode.

I then dropped the module into a whole-building Polysun model for Wandsworth, London (real irradiance, real hot-water demand) and compared it across three system configurations. The simulated system generates roughly 806 kWh of electricity and 835 kWh of useful thermal energy a year at a performance ratio of 0.79, with the harvested heat preheating domestic hot water and covering 21.4% of the annual heating demand.

Honest about its limits.

Everything here lives in simulation: idealised weather, no manufacturing tolerances, no leaks. The dissertation says so, and maps the next step: a physical prototype, dynamic weather models, and advanced coolants (nanofluids, phase-change materials) to hold efficiency through winter. That prototype is the work I want to do next.