Risk assessment and risk management for natural gas distribution

Natural gas distribution is a network industry the operations of which result at times in catastrophic events of locally concentrated impact. Its traditional practice for assessing and managing technological risk has been to focus on aggregate risk measures and not on the distribution of risk burden throughout the region serviced. Methods both for assessing the regional risk distribution and for making use of it in operational and strategic decisions are proposed. A risk analysis framework is followed leading to the generation of risk maps representing risk distribution in the service region. By integrating knowledge about the pipeline network geography and technical characteristics, point risk information is estimated in closed form. The estimation of risk during natural disasters is more complicated and requires the application of Monte Carlo simulation methods. We have developed an efficient variance reduction technique to estimate regional risk when risk sources are correlated. The important decision of prioritizing pipe line segments for replacement in the presence of spatial interactions in the cost structure is addressed. An efficient optimization model is proposed based on Edmonds' optimum branching method. This method introduces a way to avoid overly simplistic assumptions of linearity in the maintenance cost structure. Finally, the use of risk maps in strategic decisions of Natural Gas Distribution companies is discussed. Guidelines are offered for two important problems, evaluating the productivity of a given safety budget, and legitimating or mitigating the distribution of risk in the service region. All these methods are easily expandable or directly applicable to the more general problem of hazardous material distribution. Together, they form a set of decision tools for technological risk analysis and management that avoids broad brush decision indicators and stresses the use of detailed decision inputs for better technological and organizational improvements.