CORDIS - EU research results



The project aims to tackle the extremely complex multi-disciplinary problem of FA, the conditions for DDT and detonation. The key objectives, which are tailored in order for the Fellow to transfer knowledge to the host and/or bring knowledge to Europe, are as follows:

• To deliver a robust sub-model for predicting flame acceleration around obstacles on the basis of the coherent flame model which is a variant of the flame surface density approach;
• To validate the above model with laboratory scale test data and DNS predictions;
• To apply the above model to examine FA in vapour cloud explosions, and examine in particular the effect of repeated obstacles, e.g. cars, storage tanks, buildings and trees, on FA and possible DDT.
• To implement a reaction model and combine it with fine tuned chemistry to predict vapour cloud detonation.
• To test and validate the model with large scale detonation test data.
• To apply the above model to analyse some recent proprietary experimental data from industrial collaborators, previously published test results as well as historical accident scenarios in which there was strong possibility that large vapour cloud detonated. In particular the effect of cloud height and thickness on the development of overpressure and its decay will be examined.
• To draw conclusions and guidelines towards design safety to mitigate explosion hazards

The project has successfully delivered all its planned objectives. Major achievements include:
• In order to simulate large-scale hydrogen combustion and explosion, a modelling approach and technique has been developed to use single-step chemistry and transport coefficient model for fuel-air mixture combustion. The model has been developed and validated initially for hydrogen while work is continuing to extend this to propane and methane as well as blended fuels like hythane.

• Large-scale outward-propagating detonation interaction with obstacles in an open space has been numerically simulated with the modified OpenFOAM code. Some interesting phenomena such as detonation deceleration, decoupling and re-coupling, transition from normal reflection to Mach reflection and transverse overdriving detonation have been captured by the predictions and found to bear considerable similarity with small-scale cases. This is crucial evidence to support the rational of developing large-scale chemistry model and modelling strategy from small scale validations.

• An evaluation were performed to judge whether a combustion or detonation occurs in a long turbine tube, filled with H2-O2-N2-CO2 mixture (Mole fraction: hydrogen 0.122 nitrogen 0.656 oxygen 0.183 and carbon dioxide 0.039) with the total mass flow 16.1kg/s initial temperature 500℃ and ambient pressure. It was fund that the critical self-ignition temperature is less than 500 ℃ for current mixtures and self-ignition was predicted. The maximum over pressure is less than 3atm and the maximum temperature is nearly 1800K.

• For propane-air mixture, a newly developed single-step chemistry model is presented as

The predictions of this single step model for key detonation parameters within the whole explosive limit are in close agreement with that of the detailed GRI-Mechanism. The new single-step reaction mechanism has been developed for propane-air mixture, covering the entire spectrum covering flame acceleration, transition to detonation and detonation.

• Numerical study of a shock or detonation wave propagating in a channel with a U-bend has been carried out using a reactiver Navier Stoke solver with the newly developed single step chemistry model. The effects of the bend on shock or detonation wave can be divided into four modes: (1) firstly decelerating and then accelerating the detonation wave; (2) firstly accelerating and then degenerating the detonation wave; (3) degenerating the detonation wave; and (4) degenerating a shock wave followed separately by a flame.
When passing through the U-bend, the detonation wave generally decays and often re4sulting in the decoupling of the leading shock and the reaction zone. Upon emerging out through the U-bend, the distorted leading shock is accelerated due to Mach reflection or normal reflection on the side wall.

• The above model has also been validated with a vented duct case which involves flame propagation along a vent duct containing grids, the predicted flame front is in good agreement with the measurements. The results suggest that the physical processes which can lead to vapour cloud explosions in petro-chemical plant are dominated by the structure of the underlying fluid mechanics.

• Six cases of under-expanded hydrogen and hydrogen/methane jet fires are simulated using open source CFD code FireFOAM in the LES frame. Combustion and radiative heat transfer are computed using the eddy dissipation concept for multi-component fuels recently proposed by the authors and the Finite Volume Discrete Ordinates Model. The predictions are found to be in very good quantitative agreement for flame length and radiant fraction with the measurements of Schefer et al. (2007), Studer et al. (2009) and Ekoto et al. (2012), demonstrating the good potential of the FireFOAM code as a predictive tool for hazard analysis of hydrogen and hydrogen/methane jet fires.

• Numerical study has been conducted flame acceleration and detonation in obstructed semi-confined flat layer. First, to satisfy the needs of current simulation and even the future large-scale hydrogen explosion simulation, the hydrogen-air single-step chemistry model and the corresponding transport coefficient model was newly developed based on relative experimental parameters such as adiabatic flame temperature, detonation velocity, laminar flame speed, half reaction length etc. Then these models were validated against experiments in obstructed straight channel (Teodorczyk,2008) and numerical comparison of one-dimensional ZND detonation wave and free-propagating laminar flame with Oran et al.(2007)’s single-step chemistry model. Especially, using this new singe-step chemistry model and its transport coefficient model, the flame acceleration and detonation in obstructed semi-confined flat layer were simulated. The flame shape, wave evolution and flame acceleration/transition mechanism etc are analysed and further comparison with experiments was performed in pressure histories at specific points, flame acceleration curve and deflagration-to-detonation distance etc.

• Numerical studies have been conducted for a range of liquefied natural gas (LNG)
pool fires on land or water with pool diameters ranging from 14 to 400 m in the presence of a cross
wind. The code uses the extended Eddy Dissipation Concept and a newly developed soot model based on the smoke point concept in the large eddy simulation framework. The trend of fire behaviour change with diameter is analysed. Comparison is made about the change of flame length, tilt angle and surface emissive power with diameters. For the first four cases where full scale test data is available, comparison is also made with the measurements.
• A modified version of the open source CFD code, OpenFOAM for flame acceleration, DDT and detonation studies for fundamental academic research as well as industrial applications related to consequence analysis of accidental fuel release.
• A new single-step reaction mechanism for propane-air mixture, covering the entire spectrum covering flame acceleration, transition to detonation and detonation.
• A new single-step reaction mechanism for hydrogen-air mixture, covering the entire spectrum covering flame acceleration, transition to detonation and detonation.
• A modified combustion model based on the eddy dissipation for both single and multi-component fuels.
• A soot model based on laminar flamelet concept linked with the above.

The above modelling techniques, developed codes and reaction mechanisms, combustion and soot models can be used by both academic researchers and industrialists for similar applications involving turbulent combustion, deflagration, deflagration to detonation transition and detonation modelling.