New and Conventional Measures for Quantifying Risk in Rail Transport by Prof. Dr.-Ing. Eckehard Schnieder; Ing. Roman Slovák; and Dipl.-Ing. Stefan Wegele Braunschweig, Germany   Page:  1 | 2 | 3 Different European countries currently use different organizational structures, strategies and rules of operation to ensure the safety of rail transport. All national operators and manufacturers have developed their own requirements and operating procedures for a safe transport system, each according to his own safety philosophy and complying with national safety regulations. As a result of the common European market and interoperability requirements, efforts are now underway to harmonize these different approaches (EU Directives RL96/48, RL2000/16). In this context, CENELEC has published the first European standards for the safety, reliability, availability and maintainability of the railway system [Ref. 1]. In contrast to the conventional, absolute concept of safety, standard EN 50126 defines safety in rail transport in terms of the absence of an unacceptable risk of damage. Risk of damage is defined as the product of the probability that an undesirable event (hazard) will occur, and the related damage. The standard views safety only in terms of hazards to human life. A risk-based definition of safety requires railway operators to determine safety requirements based on risk analysis when introducing a new system. Under standard EN 50126, risk analysis forms part of safety planning for the entire lifecycle of a system. The aim is to ensure that a new system will be able to guarantee an adequate safety level for rail operations. Following an overview of conventional measures of risk, this article introduces a new approach to quantifying risk, followed by an illustration using a concrete example. Conventional Measures of Risk We can analyze existing risk from two different perspectives: Collective risk, which corresponds to the number of deaths expected within a specific period. Here we take the point of view of rail operators or of society as a whole. Individual risk, which considers the personal risk of each individual exposed to the rail transport system. The collective risk of a system can be expressed, for example, in terms of the number of deaths per year caused by this system. To achieve better comparability between systems, performance-related reference values are frequently used (e.g., deaths per passenger-kilometer, deaths per train-kilometer), or factors relating to specific system characteristics (e.g., distance-kilometer, number of level crossings). Individual risk is normally expressed in terms of the number of deaths per system user and year. To be able to include casualties in a risk assessment, standard EN 50126 recommends using a conversion factor to express casualties as deaths, with one death being equivalent to 10 serious or 100 slight casualties. The ratios returned by this approach differ from those that result from a calculation of risk in financial terms [Ref. 2]. Following the analysis of existing risk, it is essential to determine the level at which a risk becomes unacceptable for society or for individual system users. To determine risk acceptance, standard EN 50126 suggests three possible approaches, which are based on an application of the risk acceptance criteria below: MEM (minimum endogenous mortality) – requires that the total risk from all technical systems affecting an individual must not exceed minimum human mortality (2E-4 deaths per person per year). Assuming that an individual may be exposed to up to 20 technical systems simultaneously, this would result in an acceptable risk of 1E-5 deaths per person per year for the whole railway system (both vehicles and track). ALARP (as low as reasonably practicable) – differentiates three different target groups of individuals who are exposed to risk as represented by the railway system (employees, commuters, residents). For each target group, there is a defined upper limit of acceptable risk (e.g., for commuters, risk must not exceed 1E-4 deaths per person per year), and a defined lower limit (a risk of less than 1E-6 deaths per person per year [commuters] is always acceptable). If the resulting risk is found to be within these limits, risk reduction methods should only be applied if they make economic sense. This criterion thus requires a cost-benefit analysis. GAMAB (globalement au moins aussi bon) or GAME (globalement au moins équivalent), both meaning “globally at least as good” – can be applied when looking at either individual or collective risk. This criterion is based on the requirement that the total risk inherent in any new rail-borne transport system must not exceed the total risk inherent in comparable existing systems. It is assumed that the risk level of existing systems can be assessed (e.g., using existing statistics). The respective risk levels of an existing system and a new system can only be compared if both systems have comparable performance characteristics and operating conditions. "Time and again, safety requirements have thus resulted in oversized solutions, mostly resulting in high costs, and frequently restricting availability of the overall system." In addition to the approaches referred to in the standard, the railway industry also uses other methods to assess risk and risk acceptance (e.g., MGS – Mindestens Gleiche Sicherheit – “at least equivalent safety,” and NMAU – Nicht Mehr Als Unvermeidbar – “no more than is unavoidable”) [Ref. 3]. A cost-benefit analysis could use the life quality index as a basis [Ref. 4]. This establishes a relationship between the share of the gross domestic product available for risk reduction, average life expectancy, and the number of working hours required to earn an average income. For highly developed Western European countries, this criterion sets the value of a human life at 4 million euros. New Measures of Risk Motivation The above criteria all have one flaw and are thus difficult to apply widely: their reference value for risk bears little relevance to real life and is therefore hard for the general public to grasp. As a result, use of a risk-based method to derive safety targets for innovations in railway technology has been limited. Time and again, safety requirements have thus resulted in oversized solutions, mostly resulting in high costs, and frequently restricting availability of the overall system. The above description shows that an acceptance of individual risk is based on minimum human mortality. The approach thus assumes that all users of the railway system are exposed to a natural (endogenous) risk. The additional risk that stems from using the railway system definitely increases this endogenous risk. Exposure by an individual to other technical systems throughout his or her lifetime, or to activities in which he or she engages, also increases the endogenous risk. The information currently available does not make people aware of the risks that exist in their lives, since it is based on accident statistics. Such statistics are usually highly specific to one transport system and use different reference values. Given this lack of a universal measure of risk, people have so far been unable to develop an objective sense of risk, and thus an idea of what is an acceptable level of natural risk. A person’s risk that results from using a system depends to a large extent on the cumulative time of exposure and is thus proportionate to human life expectancy. In Germany in 2003, this was 78.09 years on average (men: 75.11 years, women: 81.07 years), which corresponds to an average mortality rate (the reciprocal value of the average life expectancy) of 1/78.09 = 0.0128 per year. Assuming minimum human mortality (2E-4 deaths per year), this results in a maximum life expectancy of 5,000 years. Risk assessment can thus no longer be intuitively grasped. Shortening of the Human Lifetime As has already been discussed, risk is defined in terms of the frequency and severity of damage. Severity of damage can be influenced particularly with passive safety measures (e.g., vehicle or route safety) or rescue management. Risk frequency, however, also depends on technical system parameters (hazard rate in [hrs-1]) of the control system, route and vehicles) and on operational traffic characteristics (vehicles or trains per hour). The time of exposure is a crucial factor in the assessment of individual risk. It depends on the traveling time required and on its potential extension as a result of operational unavailability (e.g., in hours per annum). In this respect, a close correlation between availability and safety is evident. Since the factors mentioned must be quantified in time-related units, it also appears appropriate to use time as a reference value for risk. As a measure to assess individual risk, the number of deaths per person per year is often recommended; risk acceptance is based on a comparison of this figure with the minimum endogenous mortality rate. An alternative approach is to consider risk as a potential shortening of a human lifetime. This measure of risk is often used in medicine, particularly to illustrate the negative impact of habits that are damaging to health (such as smoking) or physical activities (such as certain sports or occupations) [Ref. 5].    NEXT PAGE »