Elsevier

Journal of Hazardous Materials

Volume 262, 15 November 2013, Pages 517-526
Journal of Hazardous Materials

Probabilistic spill occurrence simulation for chemical spills management

https://doi.org/10.1016/j.jhazmat.2013.09.027Get rights and content

Highlights

  • A model for simulating probabilistic spill occurrence time series is proposed.

  • The model characterizes spill's temporal and spatial randomness.

  • The model quantifies spill's aleatory and epistemic uncertainties.

  • Probabilistic benzene spill occurrences are simulated in the St. Clair River AOC.

  • Spill occurrence management based on simulation results is discussed.

Abstract

Inland chemical spills pose a great threat to water quality in worldwide area. A sophisticated probabilistic spill-event model that characterizes temporal and spatial randomness and quantifies statistical uncertainty due to limited spill data is a major component in spill management and associated decision making. This paper presents a MATLAB-based Monte Carlo simulation (MMCS) model for simulating the probabilistic quantifiable occurrences of inland chemical spills by time, magnitude, and location based on North America Industry Classification System codes. The model's aleatory and epistemic uncertainties were quantified through integrated bootstrap resampling technique. Benzene spills in the St. Clair River area of concern were used as a case to demonstrate the model by simulating spill occurrences, occurrence time, and mass expected for a 10-year period. Uncertainty analysis indicates that simulated spill characteristics can be described by lognormal distributions with positive skewness. The simulated spill time series will enable a quantitative risk analysis for water quality impairments due to the spills. The MMCS model can also help governments to evaluate their priority list of spilled chemicals.

Introduction

Inland spills have been identified as one of the major sources of pollution of the Great Lakes and pose great threats to water quality there [1]. Unlike tanker spills in oceans, inland spills originate from industrial and municipal lands [2], including production sites, local product stores, and transportation corridors. Every year, hundreds of chemical spills occur in Southern Ontario, resulting in surface water pollution and other negative environmental impacts [3]. These spills involve in thousand types of chemicals and 89% of them only occurred less than 10 times, which brings difficulty for investigating spill probability distribution.

There have been some initiatives targeting spill prevention, preparedness, and management in Canada. For instances, the Environmental Emergencies Branch of Environment Canada has developed a Priority List for chemical spills to focus on research and development efforts for the most frequently spilled and harmful chemicals [4]. The top priority chemicals have been focused on through the development of analytical techniques and the preparation of chemical-specific response manuals. Ten-year period is suggested to be an appropriate time period to re-evaluate the Priority List because spill statistics may change with time due to the changes of chemical use and transportation patterns. In Ontario, the Spill Action Centre (SAC) was established to record spill events and other urgent environmental events on a daily basis, initiate or coordinate a response as required, and provide support to municipalities. The City of Toronto has been obliged to mitigate water contaminants from spills in order to preserve the water quality of Lake Ontario under the Great Lakes Water Quality Agreement between the Canadian and American governments.

Most current spill-related research is focused on oil spills. A Tactical Decision Problem (TDP) associated with oil spill cleanup operations was formulated as a general integer programme to optimize total response time to the spill over a planning horizon [5]. An optimization procedure for this TDP model was developed based on an aggregation scheme and strong cutting plane methods [6]. A multiperiod mixed-integer linear programming model was developed under economic and responsive criteria and coupled with oil transport and weathering model to simultaneously predict the optimal time trajectories of oil slick's volume and area, transportation profile, response resource utilization levels, cleanup schedule, and coastal protection plan [7], [8]. However, not enough research on the effect of inland chemical spills on fresh water has been justified and there is a lack of models for forecasting the probabilistic quantifiable occurrences of spills that can be used to aid decision making. Ref. [3] presented a comprehensive chemical spill management framework, which consists of a spill pollution prevention plan, a spill control plan, and an emergency response plan. Statistical and spatial analysis of spill data and stochastic spill occurrence prediction model are significantly important components of this framework because the former can identify the extent of spill problems and potential locations where management measures should be implemented and resources effectively allocated.

To investigate probabilistic events of a stochastic process (e.g. spill occurrences), it is important to analyze available historical data and determine the probability distribution (PD) of observations. Many researchers have discussed the applications of linear least square, maximum likelihood, moments, and order statistics for estimating the PD parameters of a stochastic process [9], [10], [11], [12], [13]. A traditional approach to identify the sensitive parameters or uncertainties of a mathematic model is to conduct sensitivity analysis by changing a parameter values (e.g. ±10%) from each input. However, it is fundamentally meaningless to run this analysis to determine uncertainties in final point estimates [14].

The sources of uncertainties include the uncertainties of nature, model structure, model parameter, data, computation, and operation [15]. They are categorized as either aleatory or epistemic. The former is caused by randomness in nature, while the later arises from the lack of systems knowledge or the paucity of data. It is impossible to reduce aleatory uncertainties but epistemic uncertainties could be reduced through increasing the knowledge and a longer history of quality data [15], [16]. In terms of the occurrences of inland spills, they exhibit both aleatory uncertainties in time and space due to their feature of randomness, and epistemic uncertainties due to the inadequate representation of historical spill data.

With high performance computers, studies on stochastic processes and uncertainty analysis can be achieved through Monte Carlo simulation (MCS), which is a technique that generates random values of stochastic input parameters according to their respective probabilistic characteristics [17]. This paper presents a MATLAB(2011b)-based MCS (MMCS) model to simulate the probabilistic quantifiable occurrences of inland chemical spills by time, magnitude, and location as determined by its 2012 North America Industry Classification System (NAICS) code. This model also integrates bootstrap re-sampling method [18] to quantify the model's aleatory and epistemic uncertainties. The simulation results will be used for further risk analyses of potential water quality impairments due to the spills. Benzene spills in the St. Clair River area of concern (hereinafter referred to as the River AOC) are used as a case study to demonstrate the MMCS model.

Section snippets

Methodology

The MMCS model that simulates the probabilistic quantifiable occurrences of inland chemical spills by time, magnitude, and NAICS code location is depicted in Fig. 1. The correlations among variables characterizing spills are examined to acquire variables correlativity. Through statistical analysis of historical spill data, the PD NAICS code parameters, spill inter-event time, and spilled mass can be determined by the maximum likelihood method. The resulting PDs are applied to generate random

Case study: Benzene spills in the St. Clair River AOC, Ontario

Benzene is a colourless, sweet smelling, flammable organic chemical liquid with the molecular formula C6H6 and also a confirmed carcinogen for humans [21]. It is toxic to blood, bone marrow, central nervous system, and haematopoietic system (at low concentration); it may also damage liver and urinary system and cause a continuum of haematological changes [21], [22]. Benzene can be used in a number of products such as paint, rubber, detergents, tires, shoes, drugs, plastics, synthetic rubber,

Spill occurrence management

The prediction of spill occurrences is a critical component in the comprehensive spill management planning framework [3]. The simulation models, MMCS could serve as a base for carrying out more research into a spill pollution prevention plan, which is “the use of processes, practices, materials, products, substances or energy that avoid or minimize the creation of pollutants and waste, and reduce overall risk to human health or the environment” [36]. The historical spill analysis and

Conclusions

This paper responds to the present challenge of spill occurrence prediction and management by developing a MATLAB-based Monte Carlo simulation model for simulating the probabilistic quantifiable occurrences of inland chemical spills by time, magnitude, and location based on NAICS code. The model can be extended to simulate probabilistic occurrences of any type of chemical or oil spills inland or in water from any source. The novelties of the model include characterizing spills’ temporal and

Acknowledgements

Financial supports from Natural Science and Engineering Research Council and Ryerson University are highly appreciated. The authors also would like to express their gratitude to Ontario Spill Action Centre and Sarnia Lambton Environmental Association for providing the spill databases.

References (36)

  • M. Fingas et al.

    Development of a new chemical spill Priority List

  • W.E. Wilhelm et al.

    Prescribing tactical response for oil spill clean up operations

    Management Science

    (1997)
  • F. You et al.

    Mixed-integer dynamic optimization for oil-spill response planning with integration of a dynamic oil weathering model

    AICHE Journal

    (2011)
  • P. Bhattacharya et al.

    A study on Weibull distribution for estimating the parameters

    Journal of Applied Quantitative Methods

    (2010)
  • R. Izsák

    Maximum likelihood fitting of the Poisson lognormal distribution

    Environmental and Ecological Statistics

    (2008)
  • S.J. Wu

    Estimations of the parameters of the Weibull distribution with progressively censored data

    Journal of the Japan Statistical Society

    (2002)
  • K.M. Thompson et al.

    Monte Carlo techniques for quantitative uncertainty analysis in public health risk assessments

    Risk Analysis

    (1992)
  • V.P. Singh et al.

    Risk and Reliability Analysis: A Handbook for Civil and Environmental Engineers

    (2007)
  • Cited by (3)

    • Intelligent computational techniques in marine oil spill management: A critical review

      2021, Journal of Hazardous Materials
      Citation Excerpt :

      It is essential to consider significant interactions of influential attributes in OSRM selection and process optimization. Additionally, sources of oil spill risk, such as human errors, and influential uncertainties in MOSM have also been evaluated using mathematical and AI techniques (Cao et al., 2013; Ye et al., 2020). Most decision-making models have been trained based on previously happened oil spill incidents and hypothetical cases; the trained models were used to predict future incidents and/or select suitable OSRM by analyzing the influential attributes of the new incidents (Cao and Fan, 2013; Hu et al., 2020; Huang et al., 2020; Ivshina et al., 2015).

    View full text