Claudia Fecarotti

System design and maintenance modelling for safety in extended life operation

It is frequently the most cost effective option to operate systems and infrastructure over an extended life period rather than enter a new build programme. The condition and performance of existing systems operated beyond their originally intended design life are controlled through maintenance. For new systems there is the option to simultaneously develop the design and the maintenance processes for best effect when a longer life expectancy is planned. This paper reports a combined Petri net and Bayesian network approach to investigate the effects of design and maintenance features on the system performance. The method has a number of features which overcome limitations in traditionally used system performance modelling techniques, such as fault tree analysis, and also enhances the modelling capabilities. Significantly, for the assessment of aging systems, the new method avoids the need to assume a constant failure rate over the lifetime duration. In addition the assumption of independence between component failures events is no longer required. In comparison with the commonly applied system modelling techniques, this new methodology also has the capability to represent the maintenance process in far greater detail and as such options for: inspection and testing, servicing, reactive repair and component replacement based on condition, age or use can all be included. In considering system design options, levels of redundancy and diversity along with the component types selected can be investigated. All of the options for the design and maintenance can be incorporated into a single integrated Petri net and Bayesian network model and turned on and off as required to predict the effects of any combination of options selected. In addition this model has the ability to evaluate different system failure modes. The integrated Petri-net and Bayesian network approach is demonstrated through application to a remote un-manned wellhead platform from the oil and gas industry.

An extension of the fractional model for construction of asphalt binder master curve

Knowledge and prediction of viscoelastic behaviour of asphalt binder is of great interest in order to design asphalt mixtures for civil construction of road and airports with good performances. The capability of a fractional model – requiring a very limited number of parameters – to describe and predict the linear viscoelastic behaviour of asphalt binder subjected to sinusoidal oscillations is investigated. Experimental data of complex modulus, |G*|, and phase angle, δ, are used to validate the proposed constitutive model. Based on the proposed extension of a fractional model, complex modulus isotherms for a range of frequencies can be created simply starting from isochronals at frequency value of 1 Hz. Furthermore, a mathematical procedure derived from the same model is proposed in order to compute the shift factors and automatically create the master curve, thus avoiding further errors due to manual adjustments. This shifting procedure requires only a few isotherms to create a smooth master curve able to describe asphalt binder behaviour under a wide range of temperature and loading conditions. When the fractional model is adopted and the mathematical shifting procedure is applied, only a temperature sweep test at 1 Hz is needed in order to create complex modulus and phase angle master curves.

A Petri net model for railway bridge maintenance

This article describes the application of the Petri net modelling approach to managing the maintenance process of railway bridges. The Petri net model accounts for the degradation, inspection and repair processes of individual bridge elements in investigating the effectiveness of alternative maintenance strategies. The times governing the degradation and repair processes considered are stochastic and defined by the appropriate Weibull distribution. The model offers a capability for modelling the bridge asset which overcomes the limitations in the currently used modelling techniques reported in the literature. The bridge model also provides a means of predicting the future asset condition as a result of adopting different maintenance strategies. The solution of the Petri net model is performed using a Monte Carlo simulation routine. The application of the model to a typical metal railway bridge is also presented in the article.