Domino Accidents in the Process Industry-A Review of Current Analysis Methods and Upcoming Trends

D'ALESSANDRO A., TONELLI S.M.,

Cartagena de Indias

Congreso; 5th CCPS Latin American Conference on Process Safety; 2013

Center for Chemical Process Safety(CCPS-AICHE)

With the development of the chemical industry, many large-sized complex facilities were located in huge chemical sites, resulting in the increase of storage of flammable, toxic and explosive substances. On other hand, distances between equipments and between plants were getting smaller and smaller, particularly in the industrialized countries, due to economic and land use reasons. When an accident occurs, the surrounding equipments can be damaged. If these equipments fail, an additional accident scenario can arise. Thus, a minor accident can initiate a sequence of events and lead to more severe consequences. This is usually called a domino effect (Darbra, 2010). In spite that some of the most destructive accidents in process industries were domino accidents, there is no single definition of domino effect. On a 2010 paper, Reniers lists thirteen definitions of such effect, putting into evidence the lack of thorough understanding of the mechanisms involved in the spread of an accident. However, many authors agree in considering the following pre-conditions (Cozzani et al., 2006): 1. A primary event that occurs in a certain unit. 2. The propagation of the accident to one or more units or plants, in which secondary‖ accidents are triggered as a result of the primary event. 3. An escalation‖ or intensification effect that leads to an increase in consequences, resulting in a global accident more severe than the primary one. Taking into account these guidelines, a domino event will be defined as an accident in which a primary event propagates to nearby equipment, triggering one or more secondary or higher level events resulting in overall consequences more severe than those of the primary event. Accidents including domino effects were responsible for the most serious accidents recorded in chemical plants and process facilities, as shown by the incidents listed by Abdolhamidzadeh et al. (2010). Massive series of explosions at an LPG complex in Mexico City (1984) caused more than 500 deaths and 6000 burns victims. Fire and explosion in HPCL Refinery, India (1997) led to over 60 deaths. Recently in 2009, the spread of fire after the derailment of two wagons with LPG in Viareggio, Italy resulted in the death of 32 people and forced the evacuation of 1000 people. Since the 50´s, there have been 318 domino-effect accidents with a balance of 2400 deaths and a substantial economic loss (Chen, 2012). These examples indicate that the analysis of the dangers in handling hazardous substances should not be limited to the occurrence of isolated incidents. However, the evaluation of the mechanisms that lead to a domino effect is an undeveloped area of study. Historically, the possibility of occurrence of domino events have been controlled and reduced through the use of preventive measures (safety distances, thermal insulation, emergency devices, etc.) recommended by numerous technical standards. Pioneering works in this field propose a qualitative assessment of this phenomenon (Gledhill and Lines, 1998; Delvoselle, 1998) and are essentially based on the estimate of the frequency of domino effects from historical data of accidents (Bagster and Pitblado, 1991; Kourniotis et al., 2000). Since then, there have been numerous publications focused on specific aspects of the problem, for example the techniques for assessing the likelihood of damage to a equipment following a primary event (Khan and Abbasi, 1998; Cozzani et al. 2001; Cozzani and Salzano, 2004a and b ; Gubinelli et al. 2004). Khan and Abbasi (2001) were the first to develop a systematic methodology to evaluate a string of accidents. They present a systematic sequence to assess the likelihood and type of damage to equipment within the impact radius of a primary accident. Reniers et al. publications (2005) propose a method which combines techniques such as HAZOP and What-If with matrices. In regard to quantitative methodologies as QRA analysis, the first attempts to incorporate the domino effect adopt a very simple approach: consequences or frequency of the primary stage are increased . In all cases it is necessary to perform simplifications, due to the methodological constraints and computational tools available. The first systematic approach to quantitatively assess the contribution of domino effect to the total risk of an industrial installation is done by Cozzani et al. (2005). It assesses the consequences and frequencies of all possible secondary events. One limitation of this methodology is that the propagation beyond secondary events is not considered. Moreover, its implementation in a program that uses GIS technology (Cozzani et al, 2006) requires significant simplifications in the lay-out of plants. A recent work introduces new frequency estimation techniques such as Monte Carlo simulation (Abdolhamidzadeh et al., 2010). In addition, Reniers (2010) use the game theory to decide whether to invest in prevention measures of internal and external domino effects. In this work we present a survey of the methods and models used to analyze the domino effect. In a comprehensive discussion of the available literature, the papers are classified according to their objectives: Papers analyzing specific case study to identify detailed mechanism and consequences. Papers exploiting information from historical case databases to evaluate, in a global way, mechanisms and expected frequencies of domino effects. Papers focused on the revision of available models or the formulation of new ones for the quantitative estimation of damage caused by physical effects. Papers dealing with QRA techniques and new methodologies to estimate spread probability in an accident string, in order to compare QRA with and without domino effect. We also discuss the new trends and future works in the most recent publications.