No abstract

Until now, there has been no probabilistic modelling or comprehensive risk analysis of the reliability of compressed natural gas (CNG) fuelled vehicles and support systems. This is due to sparse failure and accident data, which, in turn, is largely due to the small number of such vehicles in operation and the relatively new technology compared with diesel and gasoline engines. Direct estimation of the failure frequencies of system components requires a large quantity of data. However, estimation of reliability using probability physical models (i.e. the physics-of-failure approach) is another option that requires less data. This approach is used in this research and discussed in this paper.CNG fuel system components are subject to degradation caused by stress corrosion cracking and corrosion fatigue. A quick risk analysis shows that the storage cylinder is the most risksignificant component in CNG vehicles. The cylinder is a vulnerable component in the system, due to the presence of corrosive constituents in the stored CNG fuel and mechanical fatigue due to frequent fillings. Physics-of-failure modelling is used to estimate the frequency of leakages and ruptures of the CNG cylinders.The analytical model proposed in this paper is based on the probabilistic fracture mechanics of the associated corrosion-enhanced fatigue-failure mechanisms. The proposed model estimates the probability distribution function of the frequency of cylinder failure leading to particular CNG gas-release scenarios, while incorporating the impact of the manufacturing process, material properties, and inspection methodology. The estimated frequency of cylinder failure based on the physics of failure is used to update the overall risk associated with CNG bus systems, which has been the subject of research by the authors in the past.

A model structure for the corrosion-fatigue degradation phenomenon has been proposed for oil pipelines. At first, a well acknowledged model for corrosion-fatigue, such as Wei’s “Super Position Model”, was pursued as a reference. By reiterating the reference model using generic data from literature and applying Monte Carlo simulation, the simplest possible structure for the model was identified. The correlation of the proposed model with the environmental effects, such as loading stress and frequency, surrounding the pipelines were estimated. The sources of random variability emerging from many sources have been reasonably embodied into only two random variables. Yet again, the scarce of field and experimental data for this particular critical degradation phenomenon has compelled the research study to rely profoundly on generic data from open literature. Hence, the best distribution estimates for the two random variables were computed. Subsequent to number of iterations, the proposed model was modified to rather simpler form. All the proposed model forms have been cross checked against the original reference model which resulted in a satisfying agreement.