Heidari, Ali (2012) Numerical simulation of deflagration to detonation transition in hydrogen explosion. (PhD thesis), Kingston University, .
Abstract
The issue of deflagration to detonation transition (DDT) is one of the key factors influencing safety standards, risk assessment and risk managements in the petrochemical industries. It is also one of the most outstanding problems in combustion theory. Despite the efforts from a number of scientists around the world, numerical predictions of DDT is still an un-resolved problem due to the high level of complexities involved. Although there have been relatively more experimental efforts, a comprehensive database to assist model validation and development is still lacking. The present thesis includes numerical analysis of a wide range of combustion regimes to establish the critical conditions under which transition from deflagration to detonation occurs. In order to facilitate the study, new correlations for hydrogen burning velocity are derived from curve-fitting to experimental data from literature and implemented in the code for simulation of initial stages of flame acceleration and deflagration propagation. DetoFOAM, a code for solving transient and fully compressible Euler equations, has been developed within the framework of the OpenFOAM toolbox for numerical simulations of gaseous detonation. The detonation solver uses the total variation diminishing (TVD) numerical schemes which are suitable for shock capturing. A one step reaction mechanism has been developed following first principle and tuned for both small and large scale simulations. Since the numerical solver for DDT simulations must be capable of handling both deflagration and detonation as well as the transition, a new solver, DDTFOAM, which is based on solving fully compressible and transient Navier-Stokes equations has also been developed. DDTFOAM also uses the TVD numerical schemes for shock capturing and uses the Implicit Large Eddy Simulation (ILES) approach as a compromise for accuracy and computational efficiency [131]. Implementing an adequate chemical reaction mechanism in the DDTFOAM has been challenging to ensure that the right amount of chemical energy release is supplied in the right place and at the right time. Incorrect models for chemical energy release can significantly modify the flow behaviour. The available reactions in the literature are very limited and valid for limited range of conditions, e.g. for laminar flames only. A single step Arrhenius type reaction has been designed, tuned and implemented in DDTFOAM. The reaction mechanism has been carefully designed to reproduce flame properties e.g. laminar flame speed and thickness as well as detonation properties such as detonation thickness, propagation velocity, etc. The main difference between DetoFOAM and DDTFOAM that the former is designed for supersonic combustions (detonations) only; therefore it neglects the diffusive effects and solves reactive Euler Equations, whereas in DDTFOAM full Navier Stokes Equations are solved. The detonation solver is mainly designed for large scale detonation simulations therefore the derived reaction mechanism for this solver is obtained trough slightly different procedure compared to the DDT solver. Obtaining the reaction mechanism for DDTFOAM is more challenging as it has to reproduce properties of deflagrations as well as detonations correctly. The computational power which is required to carry out the simulations is extremely high. Different techniques have been employed to reduce the computational cost without compromising accuracy. These include using the ILES approach in cooperation with adaptive mesh refinement and multiple meshes. Numerical predictions have been conducted for different combustion regimes including laminar flames, turbulent flames and detonations as well as the actually DDT processes. The predictions of deflagrations waves are found to be In reasonably good agreement with some published experiment data. In case of detonations, detailed studies have been conducted on the detonation front structure, cellular structures as well as large industrial scenarios. This work involved contributions to Buncefield explosion investigations [109-110]. Finally, numerical simulations of some standard DDT tests have been carried out. The predictions have again achieved reasonable agreement with published experimental data and previous simulations. Successful simulations of large scale detonation in the present work represent the capability of the present study to address the increasing demands from the industries to study real scale accidental scenarios. Furthermore the obtained results for DDT simulations compare well with the medium scale experimental works and provide a step forward towards large scale and unconfined DDT studies.
Actions (Repository Editors)
 |
Item Control Page |