Improved quality control procedures and models for solar radiation using a world-wide database.

This thesis deals with- various aspects of broadband horizontal solarirradiance. Quality control of measured datasets are identified and analysed. Itwas found that solar irradiance datasets may contain significant errors. Thesesources of errors were divided in two categories, the inherent instrumenterrors and operation related errors. Methods of assessing the quality of thedatasets were evaluated and found to be unsatisfactory. A new method washence developed to quality control the solar irradiance data.The quality control procedure consists of two tiers of tests. The first tests arephysical tests that identify and remove data points that are physicalimpossibilities. The second tier tests consist of the creation of a mathematicalenvelope of acceptance in a sky clarity index domain. This envelope ofacceptance is based on multiples of standard deviations of the weightedmean of clearness index to diffuse ratio. The available datasets in this studywere thus quality controlled to remove any obvious outliers.Modelling the solar resource is an important tool for engineers and scientists.Such models have been developed since the second half of the 20th century.Some models rely on one or two meteorological parameters to estimate thesolar irradiance, while other models are more complex and require a fargreater number of points. Two of these models have been analysed andevaluated. The two models are all-sky, broadband solar irradiance models.The first model analysed is the Meteorological Radiation Model, or MRM. Thismodel is in fact a sunshine based model, with atmospheric turbidity taken intoaccount as well. The beam irradiance component was found to be acceptablegiven the number of inputs required by the model. Any extra parameterswould increase the complexity of the model, without noticeable improvements.The regressions were modified to take into account sunshine fraction banding.However the diffuse irradiance was identified as one which had the potentialIIIfor improvement. Thus, in the present work an attempt has been made todevelop improved models. The new model was found to be far superior to theolder, original model, thus the name Improved Meteorological RadiationModel, IMRM.The second type of model investigated is the cloud based radiation model.This type of model is simple to use and rely on regressions betweenirradiation, solar altitude angle and the cloud cover. Careful analysis of thecloud distribution reveals certain flaws in the current regressions. Newregressions were formulated and the result was a model superior to all itspredecessors.Clear-sky modelling is important for maximum load calculations; however,there is no method of extracting with accuracy clear-sky broadband data.Clear-sky identification techniques were evaluated and a new method wasdevised. These new datasets were used on four clear-sky models, MRM,Page's Radiation Model, PRM, Yang's radiation model and Gueymard'sREST2 model. It was found that using this new method of extracting extremeclear-sky data, the models performed better than when using quasi-clear-skydata.Solar radiation modelling is not an end by-itself, it must serve a purpose forengineers in their applications. Napier University has installed a 160m2photovoltaic facility in 2003. A 27 -year solar radiation dataset was availablefor Edinburgh, to do feasibility calculations for the project; however thisdataset contained gaps in the data. The cloud radiation model developed inthis study was utilised to this end. In addition a complete life cycle analysiswas performed on the project, and it was found that with an average efficiencyat around 12%, the facility will payback its embodied energy in eight years,and based on a relatively conservative forecast of energy prices, the financialpayback is set at under 100 years