Peter Friis-Hansen

GRACAT: software for grounding and collision risk analysis

From 1998 to 2000 an integrated software package for grounding and collision analysis was developed at the Technical University of Denmark within the project: Information technology for increased safety and efficiency in ship design and operation. The cost of the software development was 6 man-years. The software provides a toolbox for a multitude of analyses related to collision and grounding accidents. The software consists of three basic analysis modules and one risk mitigation module: (1) frequency, (2) damage, and (3) consequence. These modules can be used individually or in series and the analyses can be performed in deterministic or probabilistic mode. Finally, in the mitigation module risk profiles for the calculated consequences can be calculated and compared to alternative solutions by assignment of a cost function to the consequences. Thus, the possible analyses range from a deterministic crash analysis to a comparative risk analysis of two vessels operating on a specified route where the result is the probability density functions for the cost of oil outflow in a given area per year for the two vessels. In this paper, we describe the basic modelling principles and the capabilities of the software package. The software package can be downloaded for research purposes from www.ish.dtu.dk/GRACAT.

Nature preservation acceptance model applied to tanker oil spill simulations

This paper emphazises the adverse event categorization principle in risk acceptance analysis, and suggests the use of a standard type risk profile of lognormal type for each category of adverse events. The risk profile for a specified category of adverse events and corresponding to a given operation time is defined as the complementary probability distribution function for the accumulated loss during the operation time. The suggestion of the lognormal standard risk profile is based on the following modeling: The sequence of rare adverse events in time is mathematically modeled as a homogeneous Poisson process. It is shown that there is a specific mathematical form of the risk profiles which is robust with respect to variation of distributional assumptions for the losses associated to the single adverse events. The effect of capitalization of the future losses to present time is that the risk profile asymptotically approaches a limit risk profile as the operation time increases. This asymptotic profile is well approximated by the lognormal profile as is the profile for shorter operation time if the single losses are assumed to have lognormal distribution. The risk profile modeling is exemplified by a study of oil spills due to simulated tanker collisions in the Danish straits. It is found that the distribution of the oil spill volume per spill is well represented by an exponential distribution both in Oresund and in Great Belt. When applied in the Poisson model, a risk profile reasonably close to the standard lognormal profile is obtained. Moreover, based on data pairs (volume, cost) for world wide oil spills it is inferred that the condi-tional distribution of the costs given the spill volume is well modeled by a lognormal distribution. By unconditioning by the exponential distribution of the single oil spill, a risk profile for the costs is obtained that is indistinguishable from the standard lognormal risk profile. Finally the question of formulating a public risk acceptance criterion is addressed following Ditlevsen, and it is argued that a Nature Preservation Willingness Index can be defined in a similar way as the so-called Life Quality Index defined by Nathwani et al. [Nathwani JS, Lind NC, Pandey MD. Affordable safety by choice: the life quality method. Institute for Risk Research, University of Waterloo; Waterloo (Ontario, Canada): 1997], and can be used to quantify the risk acceptance criterion for the pollution of the environment. This NPWI acceptance criterion is applied to the oil spill example.

Life Quality Index ? an empirical or a normative concept?

The LQI by Nathwani et al. (1997) is a social indicator that kept constant allocates a part of a countrys GDP to life saving initiatives. Is the LQI built on empirical evidence of a social behaviour that implies a desired balance between free time and work time? It is shown herein that a time balancing principle leads to an explicit mathematical formula connecting work time and value productivity. Comparisons of the formula with available OECD-data show that it fits well to the data for several countries. However, it is argued that the necessary inclusion of the expected life in the LQI definition is beyond empirical verification. Thus the LQI invariance principle is claimed to be a chosen, though reasonable, normative rule of ethical resource allocation. Discrepancies between the logics of the derivations of existing variants of the LQI are mostly epistemological in nature and of less importance for the applications.