It is essential to make performance measurements of photovoltaics modules in order to quantify the power they will produce under operational conditions. Performance measurements are fundamental throughout the photovoltaic industry, from product development to quality control in manufacturing and installation in the field. Rapid and economic evaluation of photovoltaic performance requires measurements using pulsed illumination solar simulators. However some devices have characteristics which can cause difficulties making these measurements. The aim of this thesis is to overcome these measurement problems focusing particularly on two of the most prevalent and pressing of these problematic characteristics: high capacitance and metastability. A new method for measuring high capacitance modules in a pulsed simulator, based on tailor made voltage ramps, was developed. The voltage ramp is tailor made such that the measurement time is minimised while maintaining high accuracy (0.5 %), allowing the measurement of high capacitance modules in a single 10ms illumination pulse. The necessary inputs for this method are the capacitance and dark current as a function of voltage for each module. In order to make these measurements, at the high forward bias voltages required, a new system was developed. The tailored voltage ramp can be created individually for each module, since the process is rapid an automatic. This makes the method applicable to a production line or to test house measurements. In addition to their use as inputs for the voltage ramp design, the capacitance and dark current also contain other valuable information, including effective minority carrier lifetime. In several thin film technologies, such as CIGS, the efficiency is not a fixed value, rather the module is metastable and the efficiency changes depending on the previous exposure /preconditioning of the device. Preconditioning is normally applied to these devices before measurement in order to put them in a specific state that is repeatable and representative of outdoor operation. Improved preconditioning practices are vital for performance measurements in CIGS modules. Therefore the preconditioning behaviour of a variety of CIGS modules from different manufacturers was investigated. The effect of preconditioning varied for different modules, commonly the fill factor improved substantially, but often changes in open circuit voltage were also seen and in some cases also substantial changes in short circuit current. The rates of preconditioning and relaxation were found to follow stretched exponential behaviour, such that the changes occur linearly on a logarithmic timescale over several orders of magnitude in time. The total time for performance stabilisation was found to vary significantly between different types of module. Because of this stretched exponential behaviour, even though the module took days to fully relax to the dark state, there was significant relaxation within the tens of minutes that it would normally take a module to cool down after light soaking before it could be measured. The major implication of observed kinetics is that in order to achieve repeatable measurement the timing in each element of a preconditioning routine should be controlled such that the fractional error in the duration of each step is small. During the investigation an unexpectedly short timescale preconditioning effect was observed, which occurs on a millisecond timescale and relaxes in seconds. It was shown that the measurement artefacts introduced using this method can be eliminated by using electrical forward bias until immediately before the measurement. Another measurement system was developed to track the dark current and C-V characteristic of the modules during electrical bias preconditioning and subsequent relaxation. These measurements demonstrate that more than one process involved during preconditioning in CIGS. Changes occur both in the doping in the bulk of the absorber and also in charge accumulation occurring near to the absorber / buffer interface. The theoretical models for preconditioning in CIGS were reviewed and compared to the experimental results. A rate model was developed based on the theory of the metastable VSe-VCu defect. This model was shown to correspond well to the rates of preconditioning and relaxation in CIGS. The non-exponential behaviour was shown to be compatible with a distribution of activation energies for the transition between different defect states. The difference in the time taken for modules to stabilise is explained by differences in doping density and the density of VSe-VCu defects. The work presented facilitates more accurate, economical performance measurements for high capacitance devices and CIGS devices, thereby contributing to the large scale implementation of photovoltaics as power source.

This thesis presents an energy prediction tool for photovoltaic (PV) modules, based on the measure-correlate-predict principle. The tool allows quantification of the impact of the different environmental factors influencing PV device efficiency for different sites as they deviate from standardised test conditions and combines their effects for energy yield prediction of different module technologies operating in different climates. Amongst these environmental influences, the impact of angle of incidence has been particularly under-researched. In this work, a systematic investigation of the influence of angle of incidence on PV module performance is realised. This is achieved using both short-term module characterisation and long-term energy yield measurement campaigns. A customised purpose built dual axis tracker for mounting paired sets of modules on a fixed south-facing, 45-degree tilted rack is used to investigate the differences in module performance. The quality and quantity of the composition of the incident irradiance is described for various sky conditions at high latitude locations. Furthermore, an understanding of the entangled effects on photocurrent of both the angle of incidence and spectral variation is presented. This is achieved by analysing data from a system developed especially in this work which integrates an instantaneous all-sky mapping of irradiance from a monochromatic CCD camera with precision measurements of small-aperture normal irradiance from a collimated pyranometer in the short-term measurement campaign. The proposed energy prediction tool is validated using long-term datasets from several locations and is compared to other current methods. This was conducted under the European-funded PV-Catapult and IP Performance projects. The tool's prediction uncertainty falls within the ±5% for crystalline and ±10% for thin films, which is the same accuracy as other methods and within the measurement uncertainty of outdoor measurements.