Implementation of multijunction photovoltaic cells with III-V-based elements for terrestrial applications

Abstract

This thesis presents a modular and integrated approach to optimising the terrestrial deployment of multijunction photovoltaic cells (MJPVCs) based on III-V elements, addressing key challenges in power conversion, control stability, and system-level data acquisition. While III-V-based cells such as Gallium Arsenide (GaAs) offer nominal efficiencies exceeding 30% in space applications, their terrestrial implementation suffers from spectral mismatch, dynamic load variability, and control inefficiencies. To counter these limitations, three subsystems were designed, prototyped, and validated over a 50-month research period, each contributing to the overall system performance and traceability.

A Portable Internet of Things Station (PIoTS) was developed as a configurable edge device, integrating a Programmable Logic Controller (PLC) and multi-protocol gateway. The station consolidates up to 500 registers across three nodes via Modbus Transmission Control Protocol (TCP) and broadcasts data to a Message Queuing Telemetry Transport (MQTT) server, achieving a maximum data rate of 667 bits/s with a latency of 12 seconds. This architecture enables synchronised data acquisition across all subsystems, with unified timestamping for temporal alignment and traceability.
The power conversion stage was implemented using a Multiphase Interleaved Synchronous Buck Converter (MISBC), designed for compatibility with both Silicon (Si) and Gallium Nitride (GaN) Field-Effect Transistors (FETs). The converter supports switching frequencies from 25 kHz to 500 kHz and reliably delivers up to 250 watts of output power. Experimental validation demonstrated output ripple voltages below 5%, overshoot transients under 20%, and conversion efficiencies up to 97% under dynamic load conditions.

Control optimisation was achieved through a Modified Perturb and Observe (MP&O) algorithm, featuring a proportional variable step function compiled for deployment on a dsPIC33 microcontroller. Compared to conventional P&O methods, the MP&O algorithm stabilised the Maximum Power Point (MPP) up to five times faster, reduced ripple, and contributed to a circuit- level efficiency gain of 2% when integrated with the MISBC in closed-loop operation.

Future research will focus on fault-tolerant control, predictive analytics for grid interaction, and the extension of this modular platform to support emerging PV technologies and decentralised energy ecosystems.

Date of Thesis submission	2025
Original language	English
Awarding Institution	Western Sydney University
Sponsors	Australian Government Research Training Program Scholarship
Supervisor	Ali Hellany (Supervisor), Mahmood Nagrial (Supervisor) & Jamal Rizk (Supervisor)