INTEGRATED ENVIRONMNETAL RISK ASSESSMENT THROUGH A GIS-BASED DECISION-SUPPORT SYSTEM
Z. Chen, G.H. Huang, and A. Chakma
Environmental Systems Engineering program, Faculty of Engineering, University of Regina, Regina, Saskatchewan Canada S4S 0A2
PREPARED FOR THE Esri 1998 USER CONFERENCE
April 1998
© Chen et al. 1998
ABSTRACT
With an urgent need for effective management of petroleum contaminated sites, a GIS-aided simulation and risk analysis (SIMRA) system is presented in this study. The SIMRA contains two components: environmental modeling and geographical information system (GIS), which are integrated within a general framework. The modeling component includes (i) simulation for the contaminants in subsurface unsaturated and saturated zones, and (ii) risk assessment of pollution impacts on receptors of concern. The GIS component provides (a) a comprehensive database of contaminated site conditions, (b) majority of inputs for the modeling studies, (c) tools for spatial and statistical analysis, and (d) visual presentation of the modeling results. Especially, combination of the risk assessment results with spatial information will be meaningful for identifying and assessing pollution impacts on specific receptors.
The operation of the SIMRA is explained through a case study which is related to a petroleum-contaminated site in the western Canada where the soil and groundwater are contaminated by petroleum-derived product and its constituents due to leakage of pipeline and under ground storage tanks. The results indicate that SIMRA is useful for not only industries themselves but also governments when they are making decisions regarding waste management, pollution control, site remediation, impact/risk assessment.
Key words: petroleum waste, simulation, risk assessment, geographical information systems, decision-support
INTRODUCTION
Problems of leakage/spill from pipeline and storage tank in petroleum industry have been paid significant attention in the last decades (Dowd 1984; Treadway 1990; Newton 1991). The number of under ground storage tanks (USTs) for petroleum products in the North America was estimated to be between 1.5 and 2 millions (Predpall et al. 1984). Tejada (1984) reported that as much as 23% of all the tanks leak. Soil and groundwater in thousands of sites have been found being contaminated by petroleum derived products.
Previously, a number of studies for environmental risk assessment for petroleum-contaminated sites have been undertaken. For example, Lo (1991) proposed an oil spill risk simulation model based on a probabilistic approach. Hallenbeck and Flowers (1992) undertook a study of risk assessment for worker exposure to benzene. Most of the previous risk analysts argued that risk should be measured through probability (relative likelihood) of possible contamination and magnitude (seriousness) of consequences from the contamination. Thus, risk could be expressed as a probability distribution over a number of adverse consequences. In fact, when attempting to model behaviors of environmental processes, analysts often suffer from a lack of data or imperfect knowledge about the processes. This may frustrate rigorous probabilistic studies (Lein 1992).
Another major approach for risk assessment is through fuzzy set theory, which is suitable for situations when probabilistic information is not available (uncertainties present as fuzzy membership functions) (Bardossy et al. 1991). There has been a lack of previous studies using fuzzy risk assessment for petroleum waste management. In comparison, many applications to other areas have been reported. For example, Ganoulis et al. (1995) proposed a fuzzy arithmetic for ecological risk management under uncertainty. Dahab et al. (1994) developed a rule-based fuzzy set approach for risk analysis of nitrate-contaminated groundwater. Generally, the fuzzy risk analysis methods have advantages in their effectiveness in reflecting uncertainties and their applicability to practical problems. It is typically more difficult for planners/engineers to specify probability distributions than to define membership functions. Thus, extension of the methods to petroleum waste management area would be desirable for generating effective evaluations and decisions.
The fate of leaked petroleum product (nonaqueous phase liquids, NAPLs) in the subsurface is determined by several processes described in the following simplified scenario (Figure 1) (Mercer and Cohen 1990). Upon release to the environment, NAPL will migrate downward under the force of gravity. If a small amount of NAPL is released to the subsurface, it will move through the unsaturated zone where a fraction of the hydrocarbon will be captured by capillary forces as residual globules in the soil pores, finally the movement of NAPL in subsurface stops. If sufficient NAPL is released, it will transport downward until it encounters a physical barrier or is affected by buoyancy forces near the water table. Once the capillary is reached, the NAPL may move laterally as a continuous, free phase layer along the upper boundary of the water-saturated zone due to gravity and capillary forces. The chemicals inside NAPL will dissolve into the water due to water infiltrating through NAPL plume and groundwater flow. Also, the convection, dispersion, diffusion, adsorption, and biodegradation govern the transport of a variety of petroleum-derived chemicals in groundwater system. In addition, volatilization will result in further spreading of contamination. The up-to-date modeling technique is the subsurface multi-phase multi-component systems analysis such as MOVER and BIOF&T simulators in Draper Aden Environmental Modeling, Inc. and UTCHEM simulator in University of Texas at Austin.
Important progress has been made during the last decade with regard to the availability of comprehensive spatial data for supporting hydrological modeling. In a spatial hydrology model, the emphasis is firstly on digital description of the environment, and then on the formulation of process models which can fit the available data and environment description. Especially, GIS can be used to represent the landscape by means of locationally referenced data describing the character and shape of geographic features. Namely, GIS has been widely used in environmental modeling and risk assessment (Albertson et al. 1992, Hiscock et al. 1995, Lull et al. 1995, Lovertt et al. 1997). However, most of them were limited within scope of interactive and visual representation of modeling results level (Bober, 1996, Stein, 1996). More recently, exploration of in-depth integration of GIS with environmental models was conducted. For example, Schhenk et al. (1993) studied the integration of a three-dimensional groundwater modeling techniques with a multi-dimensional GIS system. Batelann et al. (1993) integrated a groundwater model within a GIS GRASS system. However, very few studies could reach the third level as pointed out by Ehlers et al. (1989) which is a full integration to be achieved through a single software package.
The objectives of this research includes:
SIMRA SYSTEM
Multicomponent Transport Model
A typical subsurface transport media has five regions, Van Genuchten and Wierenga (1976): 1) voids filled with air, 2) mobile water located inside the larger inter-aggregate pores or fractures, 3) immobile water located mainly in the intra-aggregate pores or in the porous media surrounding fractures, 4) a dynamic soil region, in equilibrium with the mobile phase, and 5) a stagnant soil region where mass transfer is diffusion limited.
The general transport equation was formulated as:
� (q mCwm)/� t + � (q imCwim)/� t + � (f r Pwm) +� [(1-f )r Pwim]
= � (q m Dij � Cwm/� xj)/� xi - � (qiCwm)/� xi - qsCws ������...[1]
where q m and q im are the fraction of soil filled with mobile and immobile water, respectively; Cwm and Cwim are the concentration [ML-3] of species in the mobile and immobile water respectively; qi is the Darcy velocity [LT-1]; Pwm and Pwim areadsorbed phase concentration of species w in the mobile and immobile phase [MM-1] respectively; f is the fraction of sorption site which is in direct contact with the mobile liquid; r is soil bulk density [ML-3]; qs is the volumetric flow rate of fluid injection (or withdrawal) per unit volume of the porous medium; Cws is the concentration of species w in the injected fluid; and Dij is the hydrodynamic dispersion tensor defined as:
q mDij = dL |q| dij + (dL � dT) qiqj/ |q| + qmt Dcd ij ��������.[2]
where dL and dT are longitudinal and transverse dispersivities, respectively; d ij is the Kronecker delta; t is tortuosity; Dc is the coefficient of molecular diffusion; and |q| is the absolute value of the Darcy velocity.
Using the continuity equation for water flow:
- � (qi)/� xi = � (q m)/� t - qs ����������������[3]
and assuming linear adsorption (P = KdC), equation [1] can be written as:
� Cwm/� t [q m + f r kd] + � Cwim/� t [ qim + (1 � f ) r kd]
= � (q mDij� Cwm/� xj)/� t - qi� Cwm/� xi � q(Cws � Cwm) �����[4]
The concentrations of mobile and immobile phases have the following relation:
� Cwim/� t [qim + (1 � f )r kd] = X (Cwm � Cwim)��������.[5]
where X is a mass transfer coefficient [T-1] for diffusive mass exchange between the mobile and immobile phases.
Incorporating decay losses l wm and contaminant loading from a hydrocarbon source to the mobile phase Hw in equations [4] and [5], we have:
� Cwm/� t [qm + f r kd] + � Cwim/� t [ qim + (1 � f ) r kd]
= � (q mDij� Cwm/� xj)/� xi- qi� Cwm/� xi � q(Cws � Cwm) - lwm + Hw�[6]
� Cwim/� t [qim + (1 � f )r kd] = X (Cwm � Cwim) - lwm ������.[7]
The more detailed formulation and solution process of the multi-phase and multi-component transport model in porous media are described in detailed by Kaluarachchi and Parker (1990), Katyal and Parker (1992), and Katyal (1997).
Fuzzy Relation Analysis (FRA)
Based on the BIOF&T 3D simulation results, a fuzzy relation analysis (FRA) model is developed (Chakma, et al. 1997) for comprehensively evaluating health and/or environmental risk due to multiple pollutant sources under uncertainty. The concept of fuzzy relation was firstly applied to medical diagnosis by Zadeh (1969). It illustrates the link between malfunctioning of a system and the possible symptoms. In a very general setting, the process of fuzzy relation analysis can be conveniently described by pointing out relationships between a collection of pattern features and their class membership vectors. It is useful for multifactorial evaluation and risk assessment under imprecision and uncertainty (Asse 1987; Pedrycz 1990). The axiomatic framework of fuzzy set operation provides a natural setting for constructing multiattribute value functions in order to sort a set of potential actions and make an effective assessment. Previously, this theory has been applied to a number of practical problems, such as assessment of environmental quality, evaluation of industrial products, and classification of acute toxicity from poisons (Jennings 1988; Ivanov and Ryvkin 1991).
Decision-Support System
The proposed SIMRA is capable of addressing uncertain and interactive features of a petroleum waste management system (Figure 2). The main tasks for the remedial investigation or diagnostic study include scoping process, sampling plan, site characterization, and treatability analysis. For the feasibility study phase or prognostic study, it contains project scoping, identify alternatives, screen alternatives, and technical evaluation. As a software package, SIMRA was developed based on the following technical considerations::
The structure of the proposed decision support system (DSS) is shown in Figure 3.
CASE STUDY
Applicability of SIMRA is demonstrated through a real case study in western Canada. The study involves the following six steps:
Step 1: Digitization and GIS data base construction. As shown in Figure 4, the study site is located in western Canada. This step provides database for further modeling study. The information sources include existing maps, digital data sources, and site investigation reports (Figure 5).
Step 2: Site specification. Background information of the study site is investigated, which includes contamination source, artesian soil and aquifers, pollutant discharge, monitoring wells, and surrounding environmental features (such as residential zones, river, lakes, etc.) (Figure 6).
Step 3: Grid system. The computational grid is often needed by numerical models. This decision support system (DSS) can generate any two-dimensional grid using the internal GIS function conveniently. If a three-dimensional computational mesh is needed, an external mesh editor may be launched to create and edit finite element meshes. The external mesh editor allows designing irregular quadrilateral meshes in two and three dimensions. The computational grids and the soil/groundwater properties in the study domain being considered can be managed through the mesh editor(Figure 7).
Step 4: Running simulator. The BIOF&T simulator and risk assessment model (FRA) can be launched through the proposed event-driven interface. After the modeling inputs prepared from the GIS database, the multi-phase and multi-component simulator was launched to simulate the transport of benzene, toluene, and xylene in the groundwater flow. The outputs were filtered through an internal data file convertor produced by C ++ to abstract data into a readable file for the system. On the other hand, as soon as the outputs from pollutant transport model were obtained by the system, the input file for the FRA risk analysis model was ready for further computation. Avenue was used for generating the inputs for the FRA model based on the BIOF&T simulation results.
Step 5: Results representation. Results o the SIMRA can be presented in several ways including GIS spatial risk distribution map, statistical chart, and contaminant-concentration -distribution surface (or contours). For instance, Figures 8, Figure 9, and Figure 10 present the simulation results of contaminant concentration distribution surface over the study domain; Figure 11 gives the predicted risk level for a specific receptor under consideration; and Figure 12 gives the final integrated risk assessment results based on the modeling studies for three major pollutants (benzene, toluene, and xylene).
CONCLUSIONS
A GIS-based decision-support system was proposed to petroleum waste management. The system not only provides a powerful data management tool for spatial information associated with petroleum-contaminated sites but also links 3D simulation model, risk assessment model, and graphical input/output presentation within a user-friendly interface. In addition, the system is adjustable and can be used for generating a variety of remediation alternatives.
The developed SIMRA itself is: (1) an object-oriented simulation environment, an open software architecture, and a convenient tool for solving large and complex real-world problems.
Results from a case study indicate that: the SIMRA is useful for realizing real-time management of site-remediation activities. It can also help the industries to identify desirable techniques for site remediation.
Development of the SIRMA has reached a stage when several modules are to be integrated within a common user interface and simultaneous display. The long term goal of full integration which would be achieved with a single software will be explored in the further improvement.
ACKNOELEDGEMENT
This research has been supported by the Natural Sciences and Engineering Research Council of Canada.
REFERENCES
