Vulnerability Assessment of Steel Liquid Storage Tanks Industrial Plants

Abstract

Industrial plants, especially chemical, petrochemical, and oil processing industries, are complex systems of numerous integrated components and processes, which can make them particularly vulnerable to natural hazard events, in particular earthquakes. The interaction of the earthquakes with industrial equipment may trigger relevant accidents resulting in the release of hazardous materials and thus the increase of overall damage to nearby areas. Therefore, each component in the industrial site requires mandatory risk assessment and development of preventive and protective actions. Steel liquid storage tanks are more vulnerable to earthquakes than other equipment in industrial plants because they often handle a large amount of toxic and flammable materials that can ignite and burn easily. Catastrophic failure of tanks observed during past earthquakes caused serious economic and environmental consequences. Many of them were designed with outdated analysis methods and with underestimated seismic loads. Therefore, the evaluation of the seismic vulnerability of existing steel liquid storage tanks located in seismic prone areas is extremely important. The seismic vulnerability of tanks is expressed using fragility curves. These curves are conditional probability statements of potential levels of damage over a range of earthquake intensities. There are a variety of approaches to derive the fragility curves for tanks, e.g., empirical, expert-based, and analytical approaches. Among them, the analytical methods have been widely accepted in recent decades. The development of analytical fragility curves for tanks, in particular existing tanks, faces the challenge of many sources of uncertainty. Therefore, a primary objective of this work is to develop an appropriate methodology to analytically derive fragility curves for existing steel liquid storage tanks with the treatment of uncertainties.

At first, an overview of earthquake damage to steel liquid storage tanks in industrial plants is introduced, together with the definition of critical damage states observed during past earthquakes. Possible numerical models for both above ground and elevated tanks subjected to earthquakes, such as spring-mass models and more refined models, are then presented. Consequently, an efficient procedure for the model calibration of unanchored tanks is proposed. The procedure is mainly based on a static pushover analysis, which is performed using a nonlinear finite element modelling of the steel tank. The model is then validated through a shaking table campaign and a full nonlinear finite element model. An overview of seismic fragility methodologies for tanks is next presented. The attention is paid to analytical methods, e.g., cloud and incremental dynamic analysis methods, which are conducted by using probabilistic seismic demand models and nonlinear time history analyses. The sources of uncertainty, classified into ground motion and modelling parameter uncertainties, are incorporated into the probabilistic seismic demand models. A sensitivity study, based on a screening experiment and an analysis of variance, is performed. This study reveals how different levels of modelling parameters, in turn, affect the seismic response of the tanks. Knowledge of the significance of each modelling parameter will provide insight as to whether its variation should be treated explicitly or may perhaps be neglected. Therefore, results of the sensitivity analysis could be used to reduce the number of parameters considered in the fragility analysis. Subsequently, optimal intensity measures for the probabilistic seismic demand analysis of above ground tanks are presented. Best performance intensity measures are selected based on their efficiency and sufficiency. The applications of the proposed procedure to two case studies of existing above ground and elevated tanks, which are located in seismic prone areas of Italy and Turkey, are finally presented, resulting in fragility curves for different limit states of the tanks. A vulnerability-based design approach of a concave sliding bearing system for the elevated tank is also introduced at the end