Fast spatially-resolved electrical modelling and quantitative characterisation of photovoltaic devices

An efficient and flexible modelling and simulation toolset for solving spatially-resolved models of photovoltaic (PV) devices is developed, and its application towards a quantitative description of localised electrical behaviour is given. A method for the extraction of local electrical device parameters is developed as a complementary approach to the conventional characterisation techniques based on lumped models to meet the emerging demands of quantitative spatially-resolved characterisation in the PV community. It allows better understanding of the effects of inhomogeneities on performance of PV devices. The simulation tool is named PV-Oriented Nodal Analysis (PVONA). This is achieved by integrating a specifically designed sparse data structure and a graphics processing unit (GPU)-based parallel conjugate gradient algorithm into a PV-oriented numerical solver. It allows more efficient high-resolution spatially-resolved modelling and simulations of PV devices than conventional approaches based on SPICE (Simulation Program with Integrated Circuit Emphasis) tools in terms of computation time and memory usage. In tests, mega-sub-cell level test cases failed in the latest LTSpice version (v4.22) and a PSpice version (v16.6) on desktop PCs with mainstream hardware due to a memory shortage. PVONA efficiently managed to solve the models. Moreover, it required up to only 5% of the time comparing the two SPICE counterparts. This allows the investigation of inhomogeneities and fault mechanisms in PV devices with high resolution on common computing platforms. The PVONA-based spatially-resolved modelling and simulation is used in various purposes. As an example, it is utilised to evaluate the impacts of nonuniform illumination profiles in a concentrator PV unit. A joint optical and electrical modelling framework is presented. Simulation results suggest that uncertainties introduced during the manufacturing and assembly of the optical components can significantly affect the performance of the system in terms of local voltage and current distribution and global current-voltage characteristics. Significant series resistance and shunt resistance effects are found to be caused by non-uniformity irradiance profiles and design parameters of PV cells. The potential of utilising PVONA as a quality assessment tool for system design is discussed. To achieve quantitative characterisation, the PVONA toolset is then used for developing a 2-D iterative method for the extraction of local electrical parameters of spatially-resolved models of thin-film devices. The method employs PVONA to implement 2-D fitting to reproduce the lateral variations in electroluminescence (EL) images, and to match the dark current-voltage characteristic simultaneously to compensate the calibration factor in EL characterisations. It managed to separate the lateral resistance from the overall series resistance effects. The method is verified by simulations. Experimental results show that pixellation of EL images can be achieved. Effects of local shunts are accurately reproduced by a fitting algorithm. The outcomes of this thesis provide valuable tools that can be used as a complementary means of performance evaluation of PV devices. After proper optimisation, these tools can be used to assist various analysis tasks during the whole lifecycle of PV products.