Sizing of stand-alone photovoltaic / wind turbine hybrid renewable energy systems using Markov modeling and data simulated with empirical copulas

A novel methodology was developed for optimally sizing components of stand-alone photovoltaic/wind turbine hybrid renewable energy systems (HRES's). The ultimate goal was to promote reliable, clean, low-cost HRES's for basic electricity and water pumping needs in remote locations, particularly small farmsteads in the developing world. A thorough literature review was undertaken on HRES sizing methodologies. Many methodologies utilize weather data sequences of relatively short duration. Others simulate data using univariate based methods. The author here used samples of solar irradiation, wind speed, and modeled load data and uniquely applied empirical copula data simulation methodology to generate large sets of simulated data. Additionally he developed a unique sizing methodology involving Markov modeling of HRES performance, which, along with the copulas, explicitly incorporates autocorrelation and cross-correlation of solar, wind, and load variables. This dissertation also contains unique options of water storage in artificial ponds and crop irrigation using `as generated' energy utilization.;Case studies used here focus on small systems with modeled and water pumping loads ranging from hundreds of watts to tens of kilowatts. Locations selected for testing the methodology were: San Juan, Puerto Rico; Golden, Colorado; and Paloma, Arizona. Additionally, an artificial data set with specific values of loss of load probability with respect to energy (LOLP energy) for given levels of storage, was generated as a means of comparing sizing methodologies.;Results of the Markov based methodology were compared to a Monte Carlo based methodology and available commercial software for validation. In many cases, estimates of LOLPenergy and surplus generation probability (SGP) found using the Markov model based methodology, were slightly lower than those found with Monte Carlo modeling. Values of LOLPenergy and SGP differed, slightly, from results obtained using the software package, however component sizes of optimal systems compared favorably. In the future, the method should be useful in predicting the reliability of grid-connected hybrid systems where storage is becoming more needed as more renewable energy systems come on line. Water storage ponds and as-generated crop irrigation systems were found to be less costly than the alternatives and should be considered for use in stand-alone HRES's.