|dc.description.abstract||Gas explosions occurring in chemical process industries, onshore and offshore modules of gas
and oil industries, inside buildings, inside process equipment, can have catastrophic consequences such as loss of lives, property damage, environmental contamination, and so on. Consequently, achieving an acceptable level of safety regarding gas explosion related accidents is a major concern of all chemical process, and oil and gas industries. Gas explosion hazard assessment is therefore very important to prevent or reduce gas explosion accidents. Since reliable predictive computational tools are needed to provide consistent and accurate estimates of gas explosion hazard assessment using numerical simulations, this master thesis focused on the study and use of the computational fluid dynamics (CFD) tool KFX-EXSIM for modeling gas explosions in small to realistic geometries. In addition, the XiFoam solver of the OpenFOAM
toolbox was also studied and used to simulate a small scale gas explosion with and without the presence of an obstacle. Three experiments of the gas explosion of stoichiometric hydrogen- and methane-air, and near stoichiometric natural gas-air mixtures in small, large vented, and realistic geometries were simulated for validation of both codes to experimental data.
Several simulations were performed for the three gas explosion experimental scenarios, and the main investigation of the experiments, such as peak overpressure, occurrence time of the peak overpressure, and flame speed were compared to results from both KFX-EXSIM and XiFoam.
For a small scale stoichiometric hydrogen-air explosion scenario, a reasonably good agreement between experiment and both KFX-EXSIM and XiFoam simulations was found in terms of the peak overpressure with an average error of 51 and 26%, respectively, for all configurations. Results of the simulations obtained from XiFoam show good agreement with the experimental results for flame speeds. The XiFoam simulations also show a good agreement with the experimental data with respect to the occurrence time of the peak overpressure. In KFX-EXSIM, the occurrence times of the predicated overpressures were delayed by more than 1.5 ms. As a result, it was concluded that the quasi-laminar combustion model used in the KFX-EXSIM explosion model gave a slow initial acceleration of flame.
For a large scale stoichiometric methane-air vented explosion experimental scenario, the numerical results obtained from the KFX-EXSIM simulation were in good agreement with the experimental data in terms of the first larger overpressure peak associated with both the Helmholtz oscillation and external pressure. However, the acoustically derived second larger overpressure peaks were completely damped by the simulations. Regarding the flame speed, the simulations show good agreement with the experimental data.
For the realistic scale near stoichiometric natural gas-air gas explosion scenarios, the simulations were in relatively good agreement with the experimental data in terms of peak overpressures. Most of the overpressure predictions fall within the band factor of 2.
The results from all simulations show that the KFX-EXSIM explosion model is capable of predicting the influence of the ignition point location, vent size, and different geometries. However, to increase the accuracy of the code, modifications on some of the values of the constant in the Porosity/Distributed Resistance (PDR) concept, and improvement in the quasi-laminar combustion model are recommended.