J. Hydrol. Hydromech., 55, 2007, 2, 111 121 CONTAMINANT TRANSPORT MODELLING IN THE VICINITY OF BZENEC WATER-WITHDRAWAL AREA LIBOR GVOŽDÍK, JANA VALENTOVÁ ČVUT v Praze, Fakulta stavební, Thákurova 7, 166 29 Praha 6; mailto: libor.gvozdik@fsv.cvut.cz This article deals with numerical modelling of contaminant transport in a locality near Bzenec. From the 1970s to the 1990s, this locality was subjected to groundwater contamination by chlorohydrocarbons (PCE, TCE, DCE). The locality is known for its drinking water supplies, which serve for over 100 000 people. Since 1992 remediation of the locality has been in progress, with several breaks due to funding problems. Numerical modelling was used as a method for assessing the efficiency of remediation and for predicting the contaminant transport until the end of 2006. In order to model contaminant transport, a 3D groundwater flow model was first created, calibrated and verified in steady state. Then the transport model was built to simulate contaminant transport. The modelling of contaminant transport was solved by using several scenarios where the input values for the dispersion, sorption and decay parameters were verified using measured values of contaminant concentration in the region of interest. KEY WORDS: Numerical Modelling, Groundwater Flow, Contaminant Transport, DNAPL. Libor Gvoždík, Jana Valentová: MODELOVÁNÍ TRANSPORTU ZNEČIŠTĚNÍ V BLÍZKOSTI JÍMACÍHO ÚZEMÍ BZENEC. Vodohosp. Čas., 55, 2007, 2; 27 lit., 5 obr., 2 tab. Článek se zabývá numerickým modelováním šíření znečištění v blízkosti Bzence. V průběhu 70. až 90. let minulého století došlo v této lokalitě ke kontaminaci podzemní vody chlorovanými uhlovodíky (PCE, TCE, DCE). Tato lokalita je významným zdrojem pitné vody pro více než 100 000 obyvatel. Od roku 1992 probíhájí v lokalitě sanační práce, které byly z finančních důvodů několikrát přerušeny. Pro ověření účinnosti sanačních prací a pro predikci šíření znečištění do konce roku 2006 byla využita metoda numerického modelování. Aby bylo možné simulovat proces šíření znečištění, byl nejprve sestaven, zkalibrován a verifikován třírozměrný model proudění podzemní vody pro ustálený stav. Potom byl vytvořen transportní model. Transport kontaminantu byl modelován v několika scénářích, lišících se hodnotami parametrů disperze, sorpce a rozpadové konstanty. Hodnoty těchto parametrů byly verifikovány pomocí měřených koncentrací znečišťujících látek v oblasti. KLÍČOVÁ SLOVA: numerické modelování, proudění podzemní vody, transport znečištění, DNAPL. Introduction Groundwater pollution presents a serious threat to human health, in particular with respect to the potential contamination of groundwater sources and supplies. In a locality near Bzenec (South Moravia) the sources of drinking water for nearly 100 000 people have already been contaminated (Fig. 1). As soon as the groundwater contamination was recognized, the discharge schedule in the withdrawal area was changed and remediation was started. Due to these measures the source of groundwater was contaminated only to a small extent (a maximum concentration of 30 µg l -1, the limit for drinking water being 10 µg l -1 ). In the region of major contamination the concentrations amounted to 1500 µg l -1, however, this is located approximately 400 m from the nearest intake structure. The groundwater in the locality of interest is contaminated by chlorohydrocarbons (ClHs) PCE (tetrachlorethylene) and TCE (trichlorethylene), which were used in the KOVO and DESTA production plants (Fig. 1) for degreasing metal parts. Another significant contaminant present here is cis-1,2-dce (dichlorethylene), which is a product of TCE biodegradation (Wiedermeier et al., 1999). The migration of organic contaminants in the subsurface is a complex process. Chlorohydrocarbons belong to a group of DNAPL substances (Dense Non Aqueous Phase Liquids), which may be con- 111
L. Gvoždík, J. Valentová Fig. 1. Plan of modelled territory Bzenec withdrawal area. Obr. 1. Situace modelovaného území jímací území Bzenec. sidered immiscible with water and which are heavier than water. DNAPLs migrate downward through the unsaturated zone and through the saturated zone to the impervious bottom of the aquifer (Šráček et al., 2002). On the bottom, a pool can form which moves in the direction of the bottom gradient. In the course of transport through the unsaturated and saturated zone, the residual free phase of the contaminant remains in the medium and it becomes a long-term source of contamination. Although chlorohydrocarbons may be considered immiscible with water, they are slightly soluble and their water solubility exceeds the water quality standards. They dissolve, form a plume of contaminated water and migrate as a part of the contaminant system (Abriola, Pinder, 1985). ClHs were used at the KOVO and DESTA plants from the early 1970s, and the annual consumption was 12 15 t of ClHs (Střemcha et al., 2003). The use of ClHs was terminated in 1992. Termination of contaminant effluent into the aquifer, however, can be assumed as late as 1997, when remediation of the unsaturated zone came to an end. Contamination with chlorohydrocarbons was first proved in 1989. The contaminant plume is relatively clearly defined, being elongated in the direction of groundwater flow (Fig. 1). Remediation pumping began in 1997 within the whole area of the contaminant plume. The total amount pumped out was 20 l s -1 in approximately 18 wells. Maximum concentrations of ClHs within the industrial plants were measured in 1995, amounting to nearly 40000 µg l -1 of ClHs. In 2005, the maximum concentrations measured were 800 µg l -1. The front of the contaminant plume, located 2000 m from the contaminant source, recorded maximum concentrations of around 1500 µg l -1 of ClHs. In 2005, the mean concentrations were measured at around 500 µg l -1 of ClHs (but when the remediation works were interrupted, they increased up to a value of 1881 µg l -1 of ClHs). Thus the contaminant plume still represents a major source of pollution and a serious hazard for the withdrawal area of Bzenec I. From the beginning, the remediation works performed in the locality were supported by the results of numerical modelling, in connection with the design of an efficient remediation system of groundwater pumping (Patzelt, 1995). In 2005, a new, more comprehensive model of the groundwater flow and contaminant transport was created (Gvoždík, 2005). This model was used for verifying the efficiency of the ongoing remediation and for simulating the contaminant transport until the 112
Contaminant transport modelling in the vicinity of Bzenec water-withdrawal area remediation is completed. Some results of this numerical modelling are presented in this paper. In order to predict the contaminant migration, various mathematical models have been developed. These models can be grouped into two categories: those that deal with the immiscible flow of a water and a nonaqueous contaminant phase, and those that describe the migration of the water soluble components as a contaminant plume (Abriola and Pinder, 1985). An overview of models dealing with multiphase / multicomponent transport in porous media is given in Kueper and Frind (1992). Dorgarten and Tsang (1992) published a numerical model which takes into consideration the flow of three moving phases, namely the water phase, the organic phase and the gas phase, as well as the water and gas phase transport. The interphase mass transfer processes are taken into account (dissolution, volatilization and air-water partitioning), and transformation of the pollutants due to adsorption and biodegradation is included in the model. Some commercial models are available, for example TOUGH (Císlerová and Vogel, 1998), TMVOC (Pruess and Battistelli, 2002). Models of this type are not used very often for the field applications, because it is often impossible to identify a large number of parameters and processes from field data. For field studies, models capable of describing the transport of the water-soluble fraction of a pollutant are widely used (Gray and Hoffman, 1983). In this study, a numerical model of this type, the commercial model MT3DMS (Zheng, 1990) was used. In 2006, a new approach based on stochastic modelling was published (Chan and Govindaraju, 2006). Geological and hydrogeological conditions The western part of the region (the contamination source zone) lies in an area of blown sands with porous permeability allowing easy infiltration of atmospheric precipitation. The disadvantage is that blown sands lack an impermeable rock cover, and this allows the rock medium and groundwater to be contaminated. The permeability of blown sands is relatively low, the filtration coefficient calculated from the grain-size curve ranging between 1.47x10-4 and 4.04x10-6 m s -1. These aeolian sediments gradually merge into fluvial sediments of the Morava River flood plain (withdrawal area of Bzenec I). The Morava River flood plain is composed of two strata formations of stream-laid sediments flood loams and sandy gravels. The genetic and lithofacial difference of the two strata formations may cause local changes in the natural groundwater flow pattern due to the effect of fast alternation of sediments with different permeability. The thickness of the flood loams, which form the impermeable hanging roof of the base-level aquifer, is fairly variable (1 5 m). The sandy gravels with sufficiently large pores for groundwater circulation provide the most favourable conditions for accumulating larger reserves of groundwater. The filtration coefficient of the medium- to coarse-grained sandy gravels of the flood plain reaches a value of 1 x 10-3 m s -1. The average thickness of this group of strata is 6 10 m, reaching up to 50 m in over-deepened parts of the flood plain. The water bearing of these sediments is mainly due to the underlying Neogene clays with variable contents of sandy admixtures. These rock types are practically impermeable, forming the Quarternary base-level aquifer. The groundwater sources are mainly infiltration territories lying on the edges of the flood plain, with also some river water infiltration from the Nová Morava River. At lower discharges, the Nová Morava serves for drainage purposes, discharging groundwater along its river bed. The natural groundwater flow direction in the region is roughly perpendicular to the Nová Morava canal (Fig. 1). Model concept In order to model the contaminant transport, several linked-up steps were necessary to perform. First of all, a 3D model of the region was created. The groundwater flow model was calibrated and verified. The resulting model was further used for simulating several states of groundwater flow under different hydraulic conditions. The calculated velocity fields served as input data for contaminant transport modelling. In order to model contaminant transport over a period of 30 years, several simplifications of the real state were adopted. Changes in hydraulic conditions (changes in groundwater intake during remediation pumping, water level fluctuation in the Nová Morava) were approximated by means of a sequence of several steady states. The sources of contamination were also simplified to non-point sources with constant contaminant effluent for the whole time of action. Contaminant transport was modelled for summary contents of PCE, TCE and DCE. The values of the dispersion, sorption and decay parameters for the respective domain on all 113
L. Gvoždík, J. Valentová model layers (regardless of materials) were kept constant. Due to the ongoing, long-term remediation, relatively large quantities of data, i.e. measured values of concentrations in remediation and monitoring wells, were available (Střemcha et al., 2003, 2005). This data was compared with modelled values of the concentrations at identically situated observation points. In this way, the accordance of the model with the real state could be verified. Numerical model The groundwater flow modelling and the subsequent contaminant transport modelling was carried out with the use of Visual MODFLOW 4.0 software (Visual MODFLOW, 2004). This software provides a user interface for a group of programmes solving groundwater flow, contamination spread, computation of flow paths, etc. The groundwater flow was solved with the help of MODFLOW 2000 software, which simulates non-stationary 3D groundwater flow in a non-homogeneous anisotropic saturated domain. The governing equation that is solved by the programme takes the form (Valentová, 2001): h h Kxx + Kyy + x x y y h h + Kzz W = SS, (1) z z t where K xx, K yy, K zz are diagonal components of the hydraulic conductivity tensor [m s -1 ], x, y, z the axes of the coordinate system parallel to the principal axes of the hydraulic conductivity tensor, h the hydraulic head [m], W characterizes sources and sinks (volume flow per volume unit [s -1 ]), S s the specific domain storativity [m -1 ], and t is time [s]. In order to obtain a numerical solution of the governing flow equation, MODFLOW applies the control volume method. This also implies that the domain is meshed into an orthogonal mesh of cells. Due to the high density of remediation and monitoring wells and the size of the model, the cell size selected was 20 x 20 m. The model is vertically subdivided into 8 non-parallel layers of variable thickness, depending on the individual material strata. The groundwater inflow across the boundary of the solved region (Vacek, 1983) was simulated by means of the Neumann boundary condition. The Nová Morava flood release canal was simulated by an inside boundary condition of a river with a clogged riverbed. The model uses mean values of infiltrated and pumped out groundwater quantities in individual wells of the withdrawal area. All remediation and monitoring wells of the remediation pumping system were also introduced into the model. The numerical modelling of the contaminant transport applied MT3D software (Zheng, 1990), version MT3DMS. This software simulates 2D or 3D transport of contaminants dissolved in groundwater through advection, dispersion and chemical reactions (sorption, decay). The transport is described by means of an advection-dispersionreaction equation in the form: C C R = Dij ( vc i ) + t x i x j xi qs ρ C b + s λ C + C sorb, Θ Θ (2) where C is the concentration of the contaminant dissolved in groundwater [µg l -1 ], t time [s], R the retardation factor [ ], x i the distance on the axes of the coordinate system [m], D ij the hydrodynamic dispersion coefficient [m 2 s -1 ], v i porous velocity [m s -1 ], q s the volume flow representing sources and sinks [s -1 ], C s the source or sink concentration [µg l -1 ], θ porosity [ ], ρ b the rock medium volume density [kg m -3 ], C sorb the contaminant concentration sorbed on to the porous material surface [kg kg -1 ], λ the first order rate (decay) constant [s -1 ]. The groundwater flow porous velocity v i is calculated from the relation (Darcy's law): vi Kii h xi = Θ, (3) where i represents a direction parallel to axis x, y or z. The hydraulic head h is calculated from the governing Eq. of flow (1), which explicitly links the transport model to the groundwater flow model. The solution of the transport equation in the MT3D model applies the mixed Euler-Lagrange method. The Lagrange method is used for solving the advection member of the transport equation (by means of moving particles). The dispersion, sink (source) and reaction members are solved by the finitedifference method within a fixed mesh. To create the transport model, it is necessary to define the input values of the parameters present in 114
Contaminant transport modelling in the vicinity of Bzenec water-withdrawal area the transport Eq. (2) with regard to the processes to be modelled hydrodynamic dispersion, sorption, biodegradation. The effect of dispersion is given by the longitudinal, transversal horizontal, vertical dispersivity values and the value for molecular diffusion, which may be neglected due to the prevailing advection member. The effect of sorption in the model is assumed to be in the form of a linear isotherm, with the equilibrium distribution coefficient K d. Contaminant biodegradation (i.e. contaminant degradation with the help of micro-organisms) is a significant process which reduces the concentrations of the contaminant in the saturated zone. The degradation is considered to fit a decay model of the first order kinetic with the decay constant λ. While modelling contaminant transport, boundary conditions (source of contamination) and initial conditions are also defined. There was, however, not enough background data to make a precise determination of the source of the contamination, and so it had to be estimated. Since the groundwater was most likely contaminated throughout the time when the chlorohydrocarbons were used, the source of contamination applied was the infiltration of contaminated water into the aquifer (recharge concentration). The concentration of the contaminants in the infiltrated water was calculated from the total assumed contaminant amounts penetrating into the aquifer during the time under consideration. It is assumed that a total of 5000 kg of chlorohydrocarbons leaked into the saturated zone (Střemcha et al., 2005). Initial conditions at the start of individual time periods were chosen in accordance with the modelled scenario. In some cases, the initial concentration was based on the modelling results of the previous time period, while in other cases it was based on measured data. Results of numerical modelling of contaminant transport The accuracy of the results obtained from numerical modelling depends mainly on the reliability of the model input parameter values dispersion, sorption, decay, sources of contamination, etc. (Quiot et al., 2005). Due to the lack of necessary background data, numerical modelling of contaminant transport was at first used for calibrating the parameter values. By comparing modelled and measured values of ClH concentrations in the monitoring wells, the values of dispersion, sorption and decay were calibrated in the scenarios that modelled the historical development of the contamination plume. The modelling confirmed a minimum effect of dispersion on the contaminant transport. The observed spread of the contaminant plume fitted best to longitudinal dispersivity α L = 5 m and transversal dispersivity α T = 0.5 m. Vertical dispersivity was assumed to be 0.05 of the α L. This dispersivity is about two orders lower than value calculated by the generally used method (Wiedermeier et al., 1999), where α L is taken as 0.1 of the plume length (α L = 200 m for plume length 2000 m). The dispersivity value of 5 m, however, falls within the range (2.13 to 7.5 m for longitudinal dispersivity in sandy gravels) described by Adams and Gelhar (1992) and Naymik and Barcelona (1981). For ClHs and other nonionic organic chemicals, it has been observed that the fraction of organic content f oc of the subsurface material is the dominant soil characteristic affecting sorption (Knox et al., 1993). Thus, the distribution coefficient K d is calculated using the following equation: K d = f oc K oc, (4) where K oc is the partition coefficient [ml water kg oc -1 ]. The fraction of organic carbon content of the contaminated aquifer is discussed in Střemcha et al. (2003). The mean value in the aeolian and sandy gravel sediments is 0.05% (f oc = 0.0005), and the calculated values of the distribution coefficient were 183 ml kg -1 for PCE, 69 ml kg -1 for TCE and 10 ml kg -1 for DCE, respectively (Střemcha et al., 2003). However, this method produces a reasonable estimate of sorption only for subsurface materials with an organic content greater than 0.1% (La Grega et al., 1994). At lower levels, interactions with the mineral surfaces become significant. For these low organic carbon sediments, K d values of 100 ml kg -1 for PCE and 50 ml kg -1 for TCE respectively were determined (Hoffman, 1995). Due to dominant contamination of TCE and DCE, the model input value of the distribution coefficient was 50 ml kg -1, and scenarios with values of 10 ml kg -1, and 0 ml kg -1 were also simulated. The distribution coefficients that resulted in the best match were 50 ml kg -1 in sandy gravels and 0 ml kg -1 (no sorption) in blown sands. Modelling also showed the significant effect of biodegradation on the reduction of dissolved contaminant concentrations in groundwater. Biodegradation processes occurred in the central part of the contaminant plume, where there are suitable anaerobic conditions and where cis-1,2-dce is the dominant contaminant in the plume. Much research 115
L. Gvoždík, J. Valentová has been undertaken to determine biodegradation rates. The main difficulty in determining the rates is due to the complex transport processes such as advection, dispersion, sorption, degradation, and difficulties in separating them. In recent years, a number of methods and relationships have been derived that enables us to calculate the first-order decay rate constant λ (Bedient et al., 1999, MPCA, 2006). However, for an approximation of the specific model value a simplified approach was used. The value of λ was estimated from the time related concentrations in the monitoring wells situated in the biodegradation area. The value was calculated using the first-order decay equation (Císlerová and Vogel, 1998): c 0 = c 1. e -λt, (5) where c 0, c 1 are the measured concentration values in the monitoring wells at time t 0 and t 1 and t = t 1 t 0. The calculated λ was 0.001 day -1, and the value is similar to the rates discussed in the literature. In Suarez and Rifai (1999), the mean values of the biodegradations rates are 0.0046 day -1 for TCE and 0.004 day -1 for DCE. Aronson and Howard (1997) indicate a biodegradation rate ranging from 0.00014 day -1 (lowest measured value) to 0.0025 day -1 (mean value). Approximately a half of the calculated value was derived due to solving the transport of the summary content of ClHs. Chlorinated solvents PCE TCE DCE are degraded by the subsequent removal of the chlorine atom, and the contaminant is transformed from one to another. Thus, the biodegradation rate value includes both the DCE degradation (the decrease in the summary content of ClHs) and the TCE degradation (transformation of TCE to DCE only and no decrease in the summary content of ClHs). Due to the transformation, the decrease in the summary content of ClHs will not occur until cis- 1,2-DCE is the dominant contaminant in the aquifer. This was observed in the historical scenario, where the biodegradation process with a decay rate of 0.001 day -1 (assumed for the whole time period) was faster than the other processes and stopped the spread of the contaminant plume. In the other scenarios, DCE is the main contaminant in the biodegradation area, and a value of 0.001 day -1 gave relatively good results for the contaminant transport. Assessment of remediation efficiency In order to assess the efficiency of the remediation of 2000 2004, two scenarios were designed. In one of them, the contaminant transport was simulated without any remediation, while in the second it was simulated with on-going remediation pumping. Apart from the comparison between the two scenarios, the simulated concentrations were also compared with the concentrations that were measured. The scenario without remediation modelled a hypothetical situation, under the assumption that the remediation works in the locality had not been restarted at the beginning of 2000. Thus, both the spread and the reduction of the contamination from the rock medium occur only due to natural (attenuation) processes. Fig. 2 displays the contaminant transport at the beginning of the modelled time period in 2000 (the initial condition being identical for both scenarios). Fig. 3 shows the contaminant transport at the end of the modelled period in 2004 for the scenario without remediation. Fig. 4 illustrates the contaminant spreading at the end of the same period for the scenario with remediation. The figures also display the groundwater level isolines, which document different hydraulic conditions in the locality for the two scenarios. The figures show that the version without remediation would imply a significant shift of contamination in the direction to the Nová Morava. The front of the contaminant plume with a concentration of 500 µg l -1 would shift by approximately 400 m in the groundwater flow direction. Remediation well SA-21 thus has a significant function in the medium part of the contaminant plume, where by pumping it serves as a hydraulic barrier to further contaminant spread. The figures also clearly show that there was a considerable reduction of the contamination, in the area of the industrial premises and in the centre of the contaminant plume, even without the remediation. This is due to contaminant biodegradation in this area. The time-related pattern of concentrations in well SA-21 (Fig. 5) shows that due to the remediation the concentrations in this well were reduced. The time course reflects a sharp decrease in concentrations at the start of the modelled period, due to remediation pumping. The decrease in concentrations in this well is probably also due to the dilution process. Despite the current state, the lower concentrations modelled at the end of the period are probably caused by overestimating the degradation effect in the area adjacent to this well. Tab. 1 presents the total contaminant balance in the aquifer for the modelled period of 2000 2004. Due to the overestimated degradation effect, the concentrations modelled in the domain of the con- 116
Contaminant transport modelling in the vicinity of Bzenec water-withdrawal area Fig. 2. Contaminant spreading at the beginning of the modelled period in 2000 (initial condition for both scenarios). Obr. 2. Rozšíření znečištění na počátku modelovaného úseku v roce 2000 (počáteční podmínka pro oba scénáře). Fig. 3. Contaminant spreading at the end of the modelled period in 2004, for the scenario without remediation. Obr. 3. Rozšíření znečištění na konci modelovaného úseku v roce 2004 ve scénáři bez sanačního zásahu. Fig. 4. Contaminant spreading at the end of the modelled period in 2004, for the scenario with remediation. Obr. 4. Rozšíření znečištění na konci modelovaného úseku v roce 2004 ve scénáři se sanačním zásahem. 117
L. Gvoždík, J. Valentová ClS concentration [µg.l -1 ] 2 000 1 500 1 000 500 SA-21 (pump-and-treat - measured ClS concentration) SA-21 (pump-and-treat - simulated ClS concentration) SA-21 (w ithout pump-and-treat - simulated ClS concentration) 0 1.1.2000 31.12.2000 31.12.2001 31.12.2002 31.12.2003 30.12.2004 Fig. 5. Concentration time-pattern in well SA-21 during remediation (modelled versus actual state) and for a version without remediation. Obr. 5. Průběh koncentrací ve vrtu SA-21 během sanace (modelovaný a skutečný stav) a ve variantě bez sanace. T a b l e 1. Contaminant balance in the aquifer for the solved period (1. 1. 2000 31. 12. 2004). T a b u l k a 1. Bilance kontaminantu ve zvodni za řešené období 1. 1. 2000 31. 12. 2004. Total contaminant amounts in the aquifer at the start of the solved period [kg] Total contaminant amounts in the aquifer at the end of the solved period [kg] Contaminant amounts eliminated by remediation pumping for the entire period [kg] Contaminant amounts eliminated by degradation for the entire period [kg] Contaminant amounts eliminated for the entire period, in total [kg] Modelled version with remediation Modelled version without remediation Difference Total balance estimate based on actual remediated amounts 2754 2754 0 5285 1563 2058 +495 3000 769 1450 (Střemcha et al., 2005) 443 696 +253 835 1191 696-495 2285 taminant plume at the start of the given period are lower than the current state. It may, therefore, be assumed that the total modelled amount of the contaminant in the aquifer at the beginning of the given period will also be lower than in reality. The respective balance, as it is, cannot serve for deriving unambiguous conclusions valid for the current state. The balance rather represents a rough estimate of the real situation. The total balance in the version with remediation shows that in the course of the solved period 1191 kg of ClHs were removed from the groundwater, which is roughly 45% of the total amount of the contaminant in the aquifer. Of this, remediation pumping accounted for 63% and degradation for 37% of the contaminant reduction. In the version without remediation, 696 kg of ClHs were removed, which is 25% of the total contaminant content. This is 20% less than in the version with remediation. It is interesting to compare the contribution of biodegradation to contaminant reduction for the individual versions. The balance shows that in the version without remediation, by 253 kg of ClHs more were eliminated through degradation than in the version with remediation, i.e. nearly 60% more. This follows from the form of Eq. (5). Based on the actual remediated contaminant amount, which was 1450 kg for the entire solved period, the modelled contaminant balance served for estimating the actual state balance. This means that at the beginning of the modelled period in 2000 there were 5285 kg of ClHs present in the groundwater. Therefore, roughly 6000 7000 kg of ClHs could leak from the source of contamination into the aquifer (taking into account ongoing degradation even before 2000). The balance further shows that 835 kg of ClHs were eliminated by degradation during the solved period. Nevertheless, at the end of 2004, there were still 3000 kg of ClHs present in the aquifer. 118
Contaminant transport modelling in the vicinity of Bzenec water-withdrawal area Prediction of contaminant spreading The prediction of the spread of the contaminant refers to the modelling of transport with remediation described in the previous paragraphs. The data on the amounts pumped at the remediation structures were updated in the model. This prediction was modelled until the end of 2006, when the remediation in the given locality is to finish. The modelling revealed that in the domain of the contaminant plume only a very slow decrease in the contaminant concentration would occur. Based on the model results (Tab. 2), the concentrations neither within the industrial areas nor outside the areas will decrease below the remediation limit by the end of 2006, and so remediation will have to continue. T a b l e 2. Summary of remediation limits and modelled ClH concentrations at the end of 2006. T a b u l k a 2. Přehled sanačních limitů a modelovaných koncentrací ClU na konci roku 2006. Remediation limit Modelled concentration Industrial areas 250 µg l -1 425 µg l -1 (SA-25) Outside areas 30 µg l -1 122 µg l -1 (SA-21) Conclusion This paper presents numerical modelling of organic contaminant transport in groundwater. Groundwater flow and transport of dissolved chlorinated solvents in the vicinity of a waterwithdrawal area was simulated using MODFLOW and MT3DMS software. The transport model includes processes of advection, dispersion, sorption and biodegradation. Thanks to a detailed technique for monitoring the progress of contamination for the last 10 years, the model calibration process enables us to verify the values of the dispersion, sorption and degradation parameters, the direction of contamination spread and the size of the contamination plume. The model derived values of dispersivity, distribution coefficient and first order rate constant were compared with values published in the scientific literature. Based on the values discussed here, several scenarios were modelled: the efficiency of the on-going remediation was verified and a prediction of contaminant spreading at the end of the remediation was modelled. The model results show that we can apply this model, which is not based on a multiphase flow approach to simulate DNAPL transport. This model can provide important information concerning the design of remediation actions and prediction of contaminant transport. However, it is necessary to calibrate and verify the model using measured values of contaminant concentrations. We must at the same time bear in mind that the model approach used here does not involve migration of the free phase product, which becomes a long-term source of contamination. List of symbols C concentration dissolved in groundwater [µg l -1 ], c 0, c 1 measured concentration values at time t 0 and t 1 [µg l -1 ], C s source or sink concentration [µg l -1 ], C sorb concentration sorbed on to the porous material surface [kg kg -1 ], D ij hydrodynamic dispersion coefficient [m 2 s -1 ], f oc fraction of organic content [ ], h hydraulic head [m], K d equilibrium distribution coefficient [ml kg -1 ], K oc partition coefficient [ml water kg -1 oc ], K xx diagonal components of the hydraulic conductivity tensor [m s -1 ], q s volume flow (sources and sinks) [s -1 ], R retardation factor [ ], S s specific domain storativity [m -1 ], t time [s], v i porous velocity [m s -1 ], W volume flow (sources and sinks) [s -1 ], x i distance on the axes of the coordinate system parallel to the principal axes of the hydraulic conductivity tensor [m], longitudinal dispersivity [m], α L α T transversal dispersivity [m], θ porosity [ ], λ first order rate (decay) constant [s -1 ], ρ b volume density [kg m -3 ]. REFERENCES ABRIOLA L., M., PINDER G. F., 1985: A Multiphase Approach to the Modeling of Porous Media Contamination by Organic Compounds, 1. Equation Development. Water Resources Research, 21, 1, 11 18. ADAMS E.E., GELHAR L.W., 1992: Field study of dispersion in a heterogeneous aquifer. 2. Spatial moments analysis. Water Resources Research, 28, 12, 3293 3307. ARONSON D., HOWARD P.H., 1997: Anaerobic Biodegradation of Organic Chemicals in Groundwater: A Summary of Field and Laboratory Studies. Prepared for the American Petroleum Institute, SRC TR-97-0223F, USA. 137 138. BEDIENT P.B., RIFAI H.S., NEWELL Ch.J., 1999: Ground Water Contamination: Transport and Remediation. (2nd Edition.) Prentice-Hall, Inc., New Jersey. CÍSLEROVÁ M., VOGEL T., 1998: Transportní procesy. Skriptum, ČVUT Praha. CHAN T. P., GOVINDARAJU RAO S., 2006: A stochasticadvective transport model for NAPL dissolution and degra- 119
L. Gvoždík, J. Valentová dation in non-uniform flows in porous media. J. Contaminant Hydrology, 87, 253 276. DORGARTEN H. W., TSANG C. F., 1992: Three-phase simulation of organic contamination in aquifer systems. In: Subsurface Contamination by Immiscible Fluids, p. 149 158, A.A. Balkema, Rotterdam. GVOŽDÍK L., 2005: Numerická simulace šíření znečištění v oblasti jímacího území Bzenec. [Diplomová práce.] Praha, 98 p. GRAY W. G., HOFFMAN J. L., 1983: A numerical model study of groundwater contamination from Prince s landfill, New Jersey. II. Sensitivity analysis and contaminant plume simulation. Ground Water, 21, 1, 15 21. HOFFMAN F., 1995: Retardation of Volatile Organic Compounds in Ground Water in Low Organic Carbon Sediments. Lawrence Livermore National Laboratory, USA. 17 p. KNOX R.C., SABATINI D.A., CANTER L.W., 1993: Subsurface Transport and Fate Processes. Lewis Publisher, Boca Raton, Florida. KUEPER B.H., FRIND E.O., 1992: Numerical modeling of multiphase/multicomponent flow and transport in porous media: An overview. In: Subsurface Contamination by Immiscible Fluids, p. 3 18, A.A.Balkema, Rotterdam. LaGREGA M.D., BUCKINGHAM P.L., EVANS J.C., 1994: Hazardous Waste Managment. McGraw-Hill, Inc., New York. MPCA, 2006: Guidelines: Natural Attenuation of Chlorinated Solvents in Ground Water. Minnesota Pollution Control Agency, Site Remediation Section, St. Paul, Minnesota. 49 p. NAYMIK T.G., BARCELONA M.J., 1981: Characterization of a contaminant plume in ground water, Meredosia, Illinois. Ground Water, 19, 5, 517 526. PATZELT Z., 1995: Bzenec - Moravský Písek. Matematické modelování proudění podzemních vod a transportu kontaminantů. ProGeo, Brtníky. 17 str. PRUESS K., BATTISTELLI A., 2002: TMVOC, A Numerical Simulator for Three-Phase Non-isothermal Flows of Multicomponent Hydrocarbon Mixtures in Saturated-Unsaturated Heterogeneous Media. Lawrence Berkeley National Laboratory, Berkeley, USA, LBNL-49375. QUIOT F., ROLLIN C., BOUR O., JORDANA S., RUIZ E., GUIMERA J., SCHWARTZ J., POIROT N., DAN A., GOBLET P., 2005: Modelling of Chlorinated Solvents Transport and Natural Attenuation in Groundwater. Transpol Research Program, INERIS, France. 8 p. STŘEMCHA J., HOSNÉDL P., PATZELT Z., PATZELTOVÁ B., PRINZ J., ŠEPS M., ZAJÍC J., 2003: Sanace těkavých chlorovaných uhlovodíků v podzemní vodě v předpolí prameniště Bzenec. Aktualizace analýzy rizika 2003. Sakol, Praha, 68 str. STŘEMCHA J., HOSNÉDL P., PRINZ J., ZAJÍC J., 2005: Sanace těkavých chlorovaných uhlovodíků v podzemní vodě v předpolí prameniště Bzenec. [Roční zpráva.] Sakol, Praha, 76 str. SUAREZ M.P., RIFAI H.S., 1999: Biodegradation Rates for Fuel Hydrocarbons and Chlorinated Solvents in Groundwater. Bioremediation Journal, 4, 337 362. ŠRÁČEK O., DATEL J., MLS J., 2002: Kontaminační hydrogeologie. Karolinum, Praha. VACEK Z., 1983: Bzenec - komplex. Čerpací zkouška. [Závěrečná zpráva.] Vodní zdroje Praha, Holešov, 412 str. VISUAL MODFLOW v. 4.0, User s Manual, 2004, Waterloo Hydrogeologic Inc., Ontario, Canada. VALENTOVÁ J., 2001: Hydraulika podzemní vody. [Skriptum.] Vydavatelství ČVUT, Praha. WIEDERMEIER T.H., NEWELL C. J., RIFAI H. S., WIL- SON J. T., 1999: Natural Attenuation of Fuels and Chlorinated Solvents in the Subsurface. Wiley & Sons, Inc., USA. ZHENG C., 1990: MT3D, Reference Manual. Waterloo Hydrogeologic Inc., Waterloo, Canada. Received 11. September 2006 Review accepted 20. March 2007 MODELOVÁNÍ TRANSPORTU ZNEČIŠTĚNÍ V BLÍZKOSTI JÍMACÍHO ÚZEMÍ BZENEC Libor Gvoždík, Jana Valentová Článek je věnován problematice numerického modelování transportu polutantů v blízkosti města Bzenec. Podzemní voda v této lokalitě byla v průběhu 70. až 90. let kontaminována chlorovanými uhlovodíky (PCE, TCE, DCE), které byly používány jako odmašťovadla v blízkých průmyslových podnicích. Bylo spotřebováno asi 300 tun těchto polutantů a z toho přibližně 6 tun způsobilo znečištění podzemní vody. Zmiňovaná lokalita je známa jako významný zdroj pitné vody pro více než 100 000 obyvatel. V roce 2005 se střed kontaminačního mraku s koncentrací 1500 µg l -1 chlorovaných uhlovodíků nacházel asi 400 metrů od nejbližší jímací studny. Vzhledem k ohrožení zdroje podzemní vody probíhají v oblasti od roku 1992 sanační práce. Aby bylo možné analyzovat šíření kontaminantu v podzemní vodě, byl pomocí software MODFLOW a MT3DMS sestaven model proudění podzemní vody a model transportu polutantů. Modely pokrývají celou kontaminovanou oblast, ve které se nacházejí sanační a monitorovací vrty, oblast jímání podzemních vod, která je ohrožena šířícím se kontaminačním mrakem a koryto Nové Moravy. Model proudění podzemní vody byl kalibrován a verifikován za ustáleného stavu pro dva různé časové úseky, vybrané z celkového období 30 let od počátku kontaminace, ve kterých je známá poloha hladiny podzemní vody. Modelem stanovená pole rychlostí podzemní vody byla použita jako vstupní hodnoty pro model transportu znečištění. V transportním modelu, který zahrnuje procesy advekce, disperze, sorpce a biodegradace, byl uvažován sumární obsah chlorovaných uhlovodíků. Při modelování transportu látek bylo specifikováno několik scénářů, ve kterých byly zkoumány parametry disperze, sorpce a biodegradace. Vzhledem k tomu, že na lokalitě probíhala v minulých letech podrobná měření koncentrací znečištění v monitorovacích vrtech, bylo možné parametry transportu verifikovat na základě porovnání simulovaných a měřených hodnot koncentrací. Získané hodnoty parametrů byly porovnány s údaji publikovanými v odborné literatuře. Pomocí takto stano- 120
Contaminant transport modelling in the vicinity of Bzenec water-withdrawal area vených parametrů byl poté simulován průběh šíření znečištění v minulosti, byla hodnocena účinnost sanačních opatření a byla uskutečněna predikce transportu znečištění. Namodelované a naměřené hodnoty byly porovnány. Výsledky modelování ukázaly, že pro simulaci transportu chlorovaných uhlovodíků je možné použít transportní model, který není založen na principu multifázového proudění. Model může sloužit pro predikci šíření znečištění a může poskytnout důležité informace pro návrh sanačních opatření. Je ovšem nezbytné kalibrovat a verifikovat ho pomocí měřených hodnot koncentrace kontaminantů. Současně je zapotřebí si uvědomit, že použitý zjednodušený přístup nezahrnuje pohyb volné fáze, která může být zdrojem dlouhodobé kontaminace. Seznam symbolů C koncentrace rozpuštěného znečištění [µg l -1 ], c 0, c 1 koncentrace v časech t 0 a t 1 [µg l -1 ], C s koncentrace zdroje nebo propadu [µg l -1 ], C sorb koncentrace sorbovaná na povrch pevné fáze [kg kg -1 ], D ij koeficient hydrodynamické disperze [m 2 s -1 ], f oc váhová frakce organického uhlíku [ ], h hydraulická výška [m], K d rovnovážný distribuční koeficient [ml kg -1 ], K oc distribuční koeficient pro organickou hmotu [ml water kg -1 oc ], K xx diagonální složka tenzoru hydraulické vodivosti [m s -1 ], q s objemový tok zdroje nebo propadu na jednotku objemu [s -1 ], R retardační koeficient [ ], S s specifická storativita [m -1 ], t čas [s], v i pórová rychlost [m s -1 ], W objemový tok zdroje nebo propadu na jednotku objemu [s -1 ], x i osy souřadného systému rovnoběžné s hlavními osami tenzoru hydraulické vodivosti [m], podélná disperzivita [m], α L α T příčná disperzivita [m], θ pórovitost [ ], λ rozpadová konstanta (rozpad 1. řádu) [s -1 ], ρ b objemová hmotnost [kg m -3 ]. 121