Article Preview
TopIntroduction
Petroleum spills happen more often than we may think in larger cities (Liang, Ryvak, Sayeed, & Zhao, 2012; Mendelssohn et al., 2010). There were over 1000 petroleum spills observed in Albany NY over a ten year period which averaged 100 per year. The researchers hypothesized there were patterns when petroleum leaks were likely to occur in certain industries or on particular days of the week. The day of week is important for humans and therefore it is a criterion for investigating man-made disasters.
The researchers tried to quantify these uncertainties into a probability distribution which could predict future spills. Knowing the methodologies to predict petroleum disasters could help emergency management agencies, insurance companies, and other stakeholders, to plan their operational strategies, human resources and schedules (Liang et al., 2012).
Awareness of risk analysis methods for disaster planning will help emergency management stakeholders avoid the loss of lives as well as to reduce the costs of mitigation (Taleb, 2007; Vanasselt & Renn, 2011). Unfortunately though one of the major limitations with existing disaster planning models and tools is that they are too complex for most managers to apply (Ellram, 1996; Mangan, Lalwani, & Gardner, 2004; Sproull & Keisler, 1991; Worthington, 2009). Several researchers recommended that scholars ought to develop new models or migrate existing tools into easy-to-use low cost spreadsheet software (Krajewski, Ritzman, & Malhotra, 2010; Nersesian, 2012; Strang, 2012).
Our current world is uncertain. Disasters are not necessarily normal in the statistical sense. Black swan events are seen everywhere such as in weather patterns when two supposedly 100-year hurricanes - Sandy and Isaac - occurred back to back (Tang, 2006). The sea wall protecting the Fukushima nuclear power plants in Japan against tsunamis was built to withstand any that had occurred in recorded history, but unfortunately the one that struck in 2012 was twice that height (Nersesian, 2012).
Man-made disasters may illustrate different probabilities as compared to natural catastrophes. Petroleum spills are generally caused by humans although some may be triggered after a natural disaster (Adams & Berry, 2012; Wagner et al., 2009). Therefore, natural and man-made disasters cannot be expected to approximate normal probabilities. Nassim Taleb provided salient advice: “only a fool would base his expectations on a normal probability distribution” (Taleb, 2007, p. 12).
This may be especially pertinent when developing disaster-related models where humans are involved with highly corrosive, toxic petroleum spills. Therefore, disaster planning models for petroleum spills should be developed using empirical data and not just simulations. “Mathematical models cannot totally replace human logic, and mistakes can be made, so it is essential to apply quality assurance when using software tools to assist with disaster planning” (Strang, 2013, p. 2). “Logistical models should not [totally] rely on simulated data – they should be tested with empirical quantitative and qualitative data (such as using the case study or experiment methodology) […] so that business managers can readily understand, interpret and apply them to their problems” (Strang, 2011, p. 13).
This study discussed the theoretical literature related to analyzing and applying probability distributions for risk analysis. A case study was used to develop and apply a discrete distribution model for risk analysis using spreadsheet software. Petroleum spill data was collected from the New York State Department of Environmental Conversation. A theoretically-selected ten-year empirical sample was evaluated for the Albany NY area to demonstrate the probability distribution model (N=1005).