Publications

Fault trees and event trees have for decades been the most commonly applied modelling tools in both risk analysis in general and the risk analysis of hydrogen applications including infrastructure in particular. It is sometimes found challenging to make traditional Quantitative Risk Analyses sufficiently transparent and it is frequently challenging for outsiders to verify the probabilistic modelling. Bayesian Networks (BN) are a graphical representation of uncertain quantities and decisions that explicitly reveal the probabilistic dependence between the variables and the related information flow. It has been suggested that BN represent a modelling tool that is superior to both fault trees and event trees with respect to the structuring and modelling of large complex systems. This paper gives an introduction to BN and utilises a case study as a basis for discussing and demonstrating the suitability of BN for modelling the risks associated with the introduction of hydrogen as an energy carrier. In this study we explore the benefits of modelling a hydrogen refuelling station using BN. The study takes its point of departure in input from a traditional detailed Quantitative Risk Analysis conducted by DNV during the HyApproval project. We compare and discuss the two analyses with respect to their advantages and disadvantages. We especially focus on a comparison of transparency and the results that may be extracted from the two alternative procedures.

For hydrogen fuelled vehicles to attain significant market penetration it is essential that any potential risks be controlled within acceptable levels. To achieve this goal on-board vehicle hydrogen storage systems should undergo risk analyses during early concept development and design phases. By so doing the process of eliminating safety-critical failure modes will help guide storage system development and be more efficient to implement than if undertaken after the design-freeze stage. The focus of this paper is the development of quantitative risk analyses of storage systems which use onboard reversible materials such as conventional AB5 metal hydrides the complex hydride NaAlH4 or other material candidates currently being researched. Collision of a vehicle having such a hydrogen storage system was selected as a dominant accident initiator and a probabilistic event tree model has been developed for this initiator. The event tree model contains a set of comprehensive mutually exclusive accident sequences. The event tree represents chronological ordering of key events that are postulated to occur sequentially in time during the accident progression. Each event may represent occurrence of a phenomenon (e.g. hydride chemical reaction and dust cloud explosion) or a hardware failure (e.g. hydride storage vessel rupture). Event tree branch probabilities can be quantified using fault tree models or basic events with probability distributions. A fault tree model for hydride dust cloud explosion is provided as an example. Failure probabilities assigned to the basic events in the fault tree can be estimated from test results published data or expert opinion elicitation. To account for variabilities in the probabilities assigned to fault tree basic events and hence to propagate uncertainties in event tree sequences Monte Carlo sampling and Latin Hypercube sampling were employed and the statistics of the results from both techniques were compared.

Previous experiments demonstrated that the accidental release of high pressure hydrogen into air can lead to the possibility of spontaneous ignition. It is believed that this ignition is due to the heating of the mixing layer between hydrogen and air that is caused by the shock wave driven by the pressurized hydrogen during the release. Currently this problem is poorly understood and not amenable to direct numerical simulation. This is due to the presence of a wide range of scales between the sizes of the blast wave driven and the very thin mixing layer. The present study addresses this fundamental ignition problem and develops a solution framework in order to predict the ignition event for given hydrogen storage pressures and dimension of the release hole. In this problem only the mixing layer between the hydrogen and air is considered. This permits us to use much higher resolution than previous studies. This mixing layer at the jet head is advected as a Lagrangian fluid particle. The key physical processes in the problem are identified to be the mixing of the two gases at the mixing layer the initial heating by the shock wave and a cooling effect due to the expansion of the mixing layer. The results of the simulations indicate that for every storage pressure there exists a critical hole size below which ignition is prevented during the release process. Close inspection of the results indicate that this limit is due to the competition between the heating provided by the shock wave and the cooling due to expansion. Furthermore the results also indicate that the details of the mixing process do not play a significant role to leading order. The limiting ignition criteria were found to be well approximated by the Homogeneous Ignition Model of Cuenot and Poinsot supplemented by a heat loss term due to expansion. Therefore turbulent mixing occurring in reality is not likely to affect the ignition limits derived in the present study. Comparison with existing experiments showed very good agreement.

The hydrogen detection system is a key component of the hydrogen safety systems (HSS). Any HSS forms a second layer of protection for the assets under accidental conditions when a first layer of protection - passive protection systems (separation at “safe” distance natural ventilation) are inoperable or failed. In this report a performance-based risk-informed methodology for establishing of the explicit quantitative requirements for hydrogen detectors allocation and actuation is proposed. The main steps of the proposed methodology are described. It is suggested (as a first approximation) to use in a process of quantification of a hydrogen detection system performance (from safety viewpoint) a five-tiered hierarchy namely 1) safety goals 2) risk-informed safety objectives 3) performance goal and metrics 4) rational safety criteria 5) safety factors. Unresolved issues of the proposed methodology of Safety Performance Analysis for development of the risk-informed and performance based standards on the hydrogen detection systems are synopsized.<br/><br/>

Transport by pipe is one the most usual way to carry liquid or gaseous energies from their extraction point until their final field sites. To limit explosion risk or escape to avoid pollution problems and human risks it is necessary to assess nocivity of defect promoting fracture. This need to know the mechanical properties of the pipes steels. Hydrogen is considered to day as a new energy vector and its transport in one of the key problems to extension of its use. Within the European project NATURALHY it has been proposed to transport a mixture of natural gas and hydrogen. 39 European partners have combined their efforts to assess the effects of hydrogen presence on the existing gas network. Key issues are durability of pipeline material integrity management safety aspects life cycle and socio-economic assessment and end-use. The work described in this paper was performed within the NATURALHY work package on ’Durability of pipeline material’. This study makes it possible to emphasize the hydrogen effect on mechanical properties of several pipe steels as X52 X70 or X100 in fatigue and fracture and in two different environments: air and hydrogen electrolytic.

Computational fluid dynamics (CFD) is used to numerically solve the sudden release of hydrogen from a high pressure tank (up to 70MPa) into air. High pressure tanks increase the risk of failure of the joints and pipes connected to the tank which results in release of Hydrogen. The supersonic flow caused by high pressure ratio of reservoir to ambient generates a strong Mach disk. A three dimensional in-house code is developed to simulate the flow. High pressure Hydrogen requires a real gas law because it deviates from ideal gas law. Firstly Beattie-Bridgeman and Abel-Noble real gas equation of states are applied to simulate the release of hydrogen in hydrogen. Then Abel-Noble is implied to simulate the release of hydrogen in air. Beattie-Bridgeman has stability problems in the case of hydrogen in air. A transport equation is used to solve the concentration of Hydrogen-air mixture. The code is second order accurate in space and first order in time and uses a modified Van Leer limiter. The fast release of Hydrogen from a small rupture needs a very small mesh therefore parallel computation is applied to overcome memory problems and to decrease the solution time. The high pressure ratio of the reservoir to ambient causes a very fast release which is accurately modelled by the code and all the shocks and Mach disk happening are observed in the results. The results show that the difference between real gas and ideal gas models cannot be ignored.

The scenario of ignition of fuels by the passage of shock waves is relevant from the perspective of safety primarily because shock ignition potentially plays an important role in deflagration to detonation transition. Even in one dimension simulation of ignition between a contact surface or a flame and a shock moving into combustible mixture is difficult because of the singular nature of the initial conditions. Indeed initially as the shock starts moving away from the contact surface the region filled with shocked reactive mixture does not exist. In the current work the formulation is transformed using time and length over time as the independent variables. This transformation yields a finite domain from t = 0. In this paper the complete spatial and temporal ignition evolution of hydrogen combustible mixtures of different concentrations is studied numerically. Integration of the governing equations is performed using an Essentially Non-Oscillatory (ENO) algorithm in space and Runge-Kutta in time while the chemistry is modeled by a three-step chain-branching mechanism which appropriately mimics hydrogen combustion.

Experiments with a hydrogen jet were performed at two different pressures 96 psig (6.6 bars) and 237 psig (16.3 bars). The hydrogen leak was generated at two different hole sizes 1/16 inch (1.6 mm) and 1/32 inch (0.79 mm). The flammable shape of the plume was characterised by numerous measurements of the hydrogen concentration inside of the jet. The effect of the nearby horizontal surface on the shape of the plume was measured and compared with results of CFD numerical simulations. The paper will present results and an interpretation on the nature of the plume shape.

The United States has 11 distinct natural gas pipeline corridors: five originate in the Southwest four deliver natural gas from Canada and two extend from the Rocky Mountain region. This study assesses the potential to deliver hydrogen through the existing natural gas pipeline network as a hydrogen and natural gas mixture to defray the cost of building dedicated hydrogen pipelines.

Plasmonic Ni nanoparticles were incorporated into LaFeO₃ photocathode (LFO-Ni) to excite the surface plasmon resonances (SPR) for enhanced light harvesting for enhancing the photoelectrochemical (PEC) hydrogen evolution reaction. The nanostructured LFO photocathode was prepared by spray pyrolysis method and Ni nanoparticles were incorporated on to the photocathode by spin coating technique. The LFO-Ni photocathode demonstrated strong optical absorption and higher current density where the untreated LFO film exhibited a maximum photocurrent of 0.036 mA/cm² at 0.6 V vs RHE and when incorporating 2.84 mmol Ni nanoparticles the photocurrent density reached a maximum of 0.066 mA/cm² at 0.6 V vs RHE due to the SPR effect. This subsequently led to enhanced hydrogen production where more than double (2.64 times) the amount of hydrogen was generated compared to the untreated LFO photocathode. Ni nanoparticles were modelled using Finite Difference Time Domain (FDTD) analysis and the results showed optimal particle size in the range of 70–100 nm for Surface Plasmon Resonance (SPR) enhancement.