Informations et ressources scientifiques
sur le développement des zones arides et semi-arides

Accueil du site → Master → Etats Unis → 2011 → Modeling passive solar distillation production in Las Vegas, Nevada

University of Nevada, Las Vegas (2011)

Modeling passive solar distillation production in Las Vegas, Nevada

Santos Noe I.

Titre : Modeling passive solar distillation production in Las Vegas, Nevada

Auteur : Santos Noe I.

Université de soutenance : University of Nevada, Las Vegas

Grade : Master of Science in Engineering (MSE) 2011

Résumé
A study has been performed to examine the effects of daily weather on the performance of commercial solar distillation basins (solar stills). The objectives of this study were to evaluate the long term performance of solar stills, to instrument two solar stills and record sub-hourly thermal properties, to evaluate existing heat transfer modeling methods for hourly production, and to create new models to predict daily production using experimental distillate production and local weather data by utilizing artificial neural networks, genetic algorithms, and multivariate regression. A system dynamics model was also created to determine the required basin area and storage volume to produce enough water to meet year round potable water demand. Solar still production was measured between January 2011 and September 2011. The average daily yield of solar still #1-A (SS1-A) and solar still #1-B (SS1-B) ranged from 2.11 ± 0.35 L/m2 and 2.00 ± 0.46 L/m2 (winter season) to 5.53 ± 1.01 L/m 2 and 5.64 ± 1.06 L/m2 (summer season), respectively. The artificial neural network model performed with a mean absolute error as low as 9.4% with up to 92.4% of production predictions within 0-20% of the actual daily production. The genetic algorithm model performed with a mean absolute error as low as 11% with up to 91% of production predictions within 0-20% of the actual daily production. The multivariate regression model performed with a mean absolute error as low as 9.7% with up to 94.1% of production predictions within 0-20% of the actual daily production. Analysis of the sub-hourly performance data indicated that large distilland volumes resulted in a greater proportion of production occurring during the night compared to smaller distilland volumes. Hourly temperature data was used to calculate heat transfer coefficients which could predict hourly distillate production with a mean absolute error between 26% and 53%.

Présentation

Version intégrale

Page publiée le 22 octobre 2012, mise à jour le 30 octobre 2018