Ehsan Arzaghi

Dynamic risk-based maintenance for offshore processing facility

Processing facilities in a marine environment may not remain safe and available if they are not well maintained. Dynamic risk‐based maintenance (RBM) methodology is a tool for maintenance planning and decision making, used to enhance the safety and availability of the equipment. It also assists in identifying and prioritizing the maintenance of equipment based on the level of risk. This article discusses an advanced methodology for the design of an optimum maintenance program integrating a dynamic risk‐based approach with a maintenance optimization technique. In this study, Bayesian Network (BN) is employed to develop a new dynamic RBM methodology that is capable of using accident precursor information in order to revise the risk profile. The use of this methodology is based on its failure prediction capability which optimizes the cost of maintenance. The developed methodology is applied to a case study involving a failure of a separator system in the offshore oil and gas production platform considering marine environments. The result shows it is essential that the valve system in the separator needs to be planned for maintenance once every 25 days; however, the cooler system can be planned for repairs once only biennially. A sensitivity analysis is also conducted to study the criticality of the operating system.

Multi-level optimization of maintenance plan for natural gas system exposed to deterioration process

In this paper, a risk-based optimization methodology for a maintenance schedule considering Process Variables (PVs), is developed within the framework of asset integrity assessment. To this end, an integration of Dynamic Bayesian Network, Damage Modelling and sensitivity analysis are implemented to clarify the behaviour of failure probability, considering the exogenous undisciplinable perturbations. Discrete time case is considered through measuring or observing the PVs. Decision configurations and utility nodes are defined inside the network to represent maintenance activities and their associated costs. The regression analysis is considered to model the impact of perturbations on PVs for future applications. The developed methodology is applied to a case study of Chemical Plant (Natural Gas Regulating and Metering Stations). To demonstrate the applicability of the methodology, three cases of seasonal observations of specific PV (pressure) are considered. The proposed methodology could either analyse the failure based on precursor data of PVs or obtain the optimum maintenance schedule based on actual condition of the systems.

An ecological risk assessment model for Arctic oil spills from a subsea pipeline

There is significant risk associated with increased oil and gas exploration activities in the Arctic Ocean. This paper presents a probabilistic methodology for Ecological Risk Assessment (ERA) of accidental oil spills in this region. A fugacity approach is adopted to model the fate and transport of released oil, taking into account the uncertainty of input variables. This assists in predicting the 95th percentile Predicted Exposure Concentration (PEC95%) of pollutants in different media. The 5th percentile Predicted No Effect Concentration (PNEC5%) is obtained from toxicity data for 19 species. A model based on Dynamic Bayesian Network (DBN) is developed to assess the ecological risk posed to the aquatic community. The model enables accounting for the occurrence likelihood of input parameters, as well as analyzing the time-variable risk profile caused by seasonal changes. It is observed through the results that previous probabilistic methods developed for ERA can be overestimating the risk level.