Highlights • A new atmospheric dispersion model is developed based on combination of Cellular Automata and Artificial Neural Networks (CA-ANN). • Comparisons are made with CFD RANS standard k-ε model on 2D free field dispersion of methane.

Atmospheric powder dispersion is modeled in a complex urban area.• 290 daily mean concentration measurements are recorded.• Artificial Neural Network (ANN) model is trained and evaluated.• The ANN model satisfies air quality model evaluation criteria.• The ANN model computing time is nearly instantaneous (less than one second).

No abstract

Résumé Entretien avec Hélène Guillemot : GIEC vs climatosceptiques

This paper presents the results of a small scale experimental study of BLEVE overpressure effects. Testing consisted of a sealed aluminum tube (0.6 L) filled with either water or propane, being heated by a flame until the internal pressure led to catastrophic failure and explosion. Three parameters were controlled during the experiments: the failing pressure, the weakened length on the tube and the fill level. BLEVEs were obtained with tests involving water and propane. Blast gages and optical techniques were used to characterize the shock wave escaping from the failing tube. The results obtained suggest that the lead shock was primarily generated by the vapor space. Overpressure results obtained were compared with the predictions of existing models and found to be in reasonable agreement except for overpressures measured vertically above the cylinder where the overpressures were highest. A prediction model based on only vapor space characteristics was developed. Images show that the shock was fully formed at some distance away from the vessel opening and this was due to the non-ideal opening of the vessel. The model developed was based on the characteristics of the shock when fully formed away from the tube. These characteristics were defined using a combination of imaging, pressure measurements, and predictions from shock tube theory.