estimation of actual free product thickness on the groundwater table
Transkrypt
estimation of actual free product thickness on the groundwater table
Proceedings of ECOpole Vol. 2, No. 2 2008 Iwona DESKA1 and Grzegorz MALINA2 ESTIMATION OF ACTUAL FREE PRODUCT THICKNESS ON THE GROUNDWATER TABLE BASED ON SOIL AND LNAPL PROPERTIES USTALANIE RZECZYWISTEJ MIĄśSZOŚCI WOLNEGO PRODUKTU ROPOPOCHODNEGO NA ZWIERCIADLE WODY PODZIEMNEJ NA PODSTAWIE WŁAŚCIWOŚCI GRUNTU I LNAPL Summary: In order to design the efficient recovery of lighter-than-water non-aqueous phase liquid (LNAPL) from the groundwater table, the data on actual LNAPL thickness should be provided. Unfortunately, the actual thickness of LNAPL (in the porous medium) is always different from the apparent thickness (measured in the monitoring well). This difference depends on parameters of soil and LNAPL. There are several methods developed for estimating the actual LNAPL thickness on the base of the apparent thickness, but the results obtained with different formulas are inconsistent and imprecise. The main limitation of the existing methods is a disregard of many important parameters. Additionally, values of some parameters in these models are difficult to estimate both: in laboratory and field. The results of our study indicate that the appropriate model for estimating the actual thickness should include the properties of soil and LNAPL. Two empirical models developed by us and presented here include key parameters of soils (hydraulic conductivity) and LNAPLs (density and dynamical viscosity), and additionally soil permeability for LNAPL. The values calculated from the developed models corresponded well in many cases with the values obtained during laboratory experiments. Keywords: LNAPL, actual thickness, apparent thickness, laboratory experiments, empirical models The surface spills and leakages from underground storage tanks, pipelines and cisterns are principal sources of soil and groundwater contamination with lighter-than-water nonaqueous phase liquids (LNAPL). In the porous medium, LNAPL floats on the groundwater table, because its density is lower than the density of water. If the layer of LNAPL occurs on the groundwater table the initial remediation step should be its recovery [1-3]. A proper design of this operation requires an assessment of the contamination plume volume. To assess the volume of a plume, observation wells are commonly installed in specified points of the contaminated area to measure the thicknesses of LNAPL layer [1]. Unfortunately, the thickness of LNAPL on the groundwater table (the actual thickness) is different from the thickness observed in the well (the apparent thickness) [4-6]. The difference between apparent and actual thickness depends on the properties of soil, and the properties and amounts of LNAPL released to the soil [7-9]. A number of methods exist for estimating the actual thickness from the apparent thickness data, but the results obtained with different formulas are inconsistent, and in many cases inaccurate [4, 5, 10]. The obtained results of laboratory investigations indicate that the appropriate model for estimating the actual LNAPL thickness should include the properties of soil and LNAPL [9, 11]. The key parameter of soil is hydraulic conductivity, whereas in the case of LNAPL density and 1 Institute of Environmental Engineering, Czestochowa University of Technology, ul. Brzeźnicka 60A, 42-200 Częstochowa, tel. 034 325 09 17, email: [email protected] 2 Department of Hydrogeology and Engineering Geology, AGH - University of Science and Technology, al. A. Mickiewicza 30, 30-059 Kraków, tel. 012 617 24 04, email: [email protected] 304 Iwona Deska and Grzegorz Malina dynamical viscosity play an important role. An additional important parameter that depends on properties of both: soil and LNAPL is the soil permeability of LNAPL [9]. The goal of this paper is to verify the empirical models developed by us for estimating the actual LNAPL thickness on the groundwater table. An additional focus is to compare results obtained based on the proposed models and the existing methods. Materials and methods Studies were carried out in Plexiglas columns with filter-tubes as monitoring wells [7-9, 11]. The experiments were performed with use of six model soils and six types of LNAPLs. The properties of soils and LNAPLs are given in Tables 1-3. Table 1 Properties of soils used in the experiments Soil 1 2 3 4 5 6 Soil grain size [mm] 0.1÷0.25 0.25÷0.315 0.315÷0.5 0.5÷0.63 0.63÷0.8 0.8÷1.0 Medium soil grain size [mm] 0.175 0.2825 0.4075 0.565 0.715 0.9 Hydraulic conductivity at 10ºC k10 [m/d] 11.33 ± 0.07 23.255 ± 0.426 71.334 ± 1.640 107.285 154.627 231.884 Coefficient of permeability κ [m2] 1.75·10–11 ± 1.078·10–13 3.59·10–11 ± 6.674·10–13 1.10·10–10 ± 2.54·10–12 1.66·10–10 2.38·10–10 3.57·10–10 Table 2 Properties of LNAPLs used in the experiments (at temperature of 20ºC) LNAPL Type of LNAPL LNAPL 1 LNAPL 2 LNAPL 3 LNAPL 4 LNAPL 5 LNAPL 6 Petroleum Rape oil Mineral oil “Lotos” Semi-synthetic oil “Orlen” Synthetic oil “Lotos” Synthetic oil “Orlen” Density ρo [kg/m3] 820.5 918 880 872 855 871 Dynamic viscosity ηo [kg/m·s] 1.804⋅10–3 ± 1.42·10–6 0.0718 ± 5.94·10–5 0.3 ± 2.54·10–4 0.219 ± 2.52·10–4 0.181 Specific gravity γo [N/m3] 8049.105 9005.58 8632.8 8554.32 8387.55 8544.51 0.194 ± 2.51·10–4 Table 3 LNAPL permeability (at temperature of 20ºC) LNAPL permeability at 20ºC ko [m/d] Soil 1 2 3 4 5 6 LNAPL 1 6.73 13.82 42.53 63.80 91.81 137.66 LNAPL 2 0.19 0.39 1.19 1.79 2.57 3.86 LNAPL 3 0.04 0.09 0.27 0.41 0.59 0.89 LNAPL 4 0.06 0.12 0.37 0.56 0.80 1.21 LNAPL 5 0.07 0.14 0.44 0.66 0.95 1.43 LNAPL 6 0.07 0.14 0.42 0.63 0.91 1.36 Estimation of actual free product thickness on the groundwater table based on soil and LNAPL properties 305 The columns were hydraulically connected with the equalizing columns. The columns were packed with the soil samples (soils 1-6) and filled with tap water until the water table reached the assumed elevation. After 3-4 days 50 cm3 of diverse LNAPLs, coloured with the pigment - Sudan III, were injected directly above the capillary fringe zone. After subsequent 3-4 days, the apparent and actual LNAPL thicknesses were measured in the well and in soil. The actual thickness was in this case the distance between LNAPL-water interface in soil and air-LNAPL interface in the well (without the capillary fringe of LNAPL in the soil). This procedure was repeated at least 10 times. The water table in the equalizing column was kept constant during experiments. The top of columns was protected against LNAPL evaporation by the styropore covers. Results and discussion On the base of laboratory experiments two empirical models were developed that describe the relationship between apparent and actual LNAPL thicknesses. Model 1 is based on the conductivity of soil for LNAPL and dynamic viscosity of LNAPL. The key parameters of model 2 are the hydraulic conductivity of soil and the specific density and dynamic viscosity of LNAPL [9, 11]. Model 1 The actual LNAPL thickness can be estimated on the base of the equation: H f = H 0 + (ωη o + ϕ )ln k o + χ ln (η o ) − ξ (1) where: Hf - actual LNAPL thickness [cm], H0 - apparent LNAPL thickness [cm], ko - LNAPL conductivity [m/d], ηo - dynamic viscosity of LNAPL [cP], ω, φ, χ and ξ factors dependent on actual LNAPL thickness [-]. Factors ω, φ, χ and ξ can be calculated from the equations: ω = −0.002 H fe + 0.052 (2) ϕ = −0.0087 H 2fe + 0.8185H f + 18.731 (3) χ = 0.1115H fe + 20.992 (4) ξ = 2.138H fe + 110.12 (5) where Hfe - initial estimated actual LNAPL thickness [cm]. Model 2 The actual LNAPL thickness can be calculated from equation: ' Hf = H 0 ⋅ ln k10 ⋅ (ζη o2 + ψη o + σ ) ⋅ e (δ −ερ ro )⋅k10 η o ⋅ ρ ro (6) Factors ζ, ψ, σ, δ and ε can be calculated from the equations: ζ = −0.0059 H fe + 0.0134 (7) ψ = 0.0033H f + 0.0028 (8) 306 Iwona Deska and Grzegorz Malina σ = −3 ⋅10 −6 H fe + 0.0001 (9) δ = −144.6 H fe + 3513.1 (10) ε = 136.92 H fe − 3190.5 (11) where: Hf, H0, Hfe - see above, k10 - hydraulic conductivity [m/d], k10’ - hydraulic conductivity [m/s], ηo - dynamic viscosity of LNAPL [cP], ρro - specific density of LNAPL [-], ζ, ψ, σ, δ and ε - factors dependent on actual LNAPL thickness [-]. Use of these models requires initial estimation of actual thickness (Hfe) to determine the proper values of factors dependent on actual LNAPL thickness. The calculation should be repeated until the calculated actual thickness is equal to the initial estimated value. To verify the developed models the relationships between apparent and actual thicknesses obtained from the models were compared with these calculated with use of the existing methods and with the experimental data. Figures 1 and 2 show these comparisons for selected soils and LNAPLs. Figure 1 presents the best fits. Soil 2 - LNAPL 3 Soil 1 - LNAPL 2 18 Actual LNAPL thickness [cm] Actual LNAPL thickness [cm] 18 16 14 12 10 8 6 4 2 16 14 12 10 8 6 4 2 0 0 0 50 100 150 200 Apparent LNAPL thickness [cm] 250 0 300 20 40 60 80 100 Apparent LNAPL thickness [cm] 140 Soil 5 - LNAPL 5 22 20 18 16 14 12 10 8 6 4 2 0 18 Actual LNAPL thickness [cm] Actual LNAPL thickness [cm] Soil 4 - LNAPL 2 120 16 14 12 10 8 6 4 2 0 0 25 50 75 100 Apparent LNAPL thickness [cm] 125 Laboratory experiments Met. of de Pastrovich Met. of Schiegg min. Met. of Schiegg max. 0 150 20 Met. of Hall 20 40 60 80 Apparent LNAPL thickness [cm] Met. of Blake and Hall 100 Met. of Ballestero 0 0 20 40 60 80 100 Model 1 120 140 160 180 200 Model 2 Fig. 1. The verification of developed models on the base of laboratory investigations and calculations with use of existing methods - the best cases The results of the verification indicate that the values calculated from the proposed models corresponded in many cases to the experimental data much better than results obtained on the base of existing methods. Only for a few compositions of soils and Estimation of actual free product thickness on the groundwater table based on soil and LNAPL properties 307 LNAPLs the results derived from the developed models fit poorly the experimental data. Figure 2 presents these most unfavorable cases. Soil 3 - LNAPL 1 20 16 18 16 14 12 Actual LNAPL thickness [cm] Actual LNAPL thickness [cm] Soil 1 - LNAPL 1 18 14 12 10 8 6 4 2 0 0 20 40 60 80 100 120 140 Apparent LNAPL thickness [cm] 160 10 8 6 4 2 0 180 0 20 Soil 5 - LNAPL 1 100 Soil 6 - LNAPL 1 25 Actual LNAPL thickness [cm] 18 Actual LNAPL thickness [cm] 40 60 80 Apparent LNAPL thickness [cm] 16 14 12 10 8 6 4 2 0 20 15 10 5 0 0 10 20 30 40 50 60 Apparent LNAPL thickness [cm] 70 Laboratory experiments Met. of de Pastrovich Met. of Schiegg min. Met. of Schiegg max. 80 -5 0 5 Met. of Hall 20 10 15 20 25 30 35 40 45 Apparent LNAPL thickness [cm] Met. of Blake and Hall 50 55 60 Met. of Ballestero 0 0 20 40 60 80 100 Model 1 120 140 160 180 200 Model 2 Fig. 2. The verification of the developed models on the base of laboratory investigations and calculations with use of existing methods - the most unfavorable cases The most unfavorable results in the case of model 1 were obtained for the scheme: LNAPL 1 (liquid with the lowest viscosity and density) and soils 5 and 6 (coarse sands). In the case of model 2 the worst results were obtained for compositions: LNAPL 1 and soils 1 and 5. It may suggest that model 1 cannot be applied for coarse sands contaminated with LNAPLs with low viscosities, whereas model 2 cannot be used in the case of LNAPLs with low viscosities. Conclusions 1. 2. 3. 4. The appropriate model for estimating the actual thickness of LNAPL on the groundwater table should include the properties of the specific soil and liquid. The values of actual LNAPL thicknesses calculated from the developed empirical models are in many cases consistent with the measured values. The proposed model 1 in a present form cannot be successfully applied in the case of low viscosity LNAPLs and coarse sands, and model 2 for LNAPLs of low viscosity. Further studies aimed at the improvement of the proposed models should include the investigations with the use of heterogeneous soils and LNAPLs with more diverse properties. 308 Iwona Deska and Grzegorz Malina Acknowledgements This work was financially supported by KBN grant no. 4 T09D 041 25 and partially by BS-401-301/99. References [1] Cooper G.S., Peralta R.C. and Kaluarachchi J.J.: Optimizing separate phase light hydrocarbon recovery from contaminated unconfined aquifers. Adv. Water Resour., 1998, 21, 339-350. [2] Cooper G.S., Peralta R.C. and Kaluarachchi J.J.: Stepwise pumping approach to improve free phase light hydrocarbon recovery from unconfined aquifer. J. Contam. Hydrol., 1995, 18, 141-159. [3] Adamek M., Koślacz R. and Zieliński W.: Wskazówki metodyczne wykonywania rekultywacji gruntów i wód podziemnych zanieczyszczonych produktami naftowymi. MOŚZNiL, Dep. Geologii, Multimedia s.c., Wrocław 1995. [4] Lenhard R.J. and Parker J.C.: Estimation of free hydrocarbon volume from fluid levels in monitoring wells. Ground Water, 1990, 28(1), 57-67. [5] EPA: How to effectively recover free product at leaking underground storage tank sites: A guide for state regulators. (EPA 510-R-96-001), 1996. [6] De Pastrovich T.L., Baradat Y., Barthel Y., Chiarelli A. and Fussell D.R.: Protection of groundwater from oil pollution. CONCAWE, Report 3/79, Den Haag, Netherlands, 1979, 62 pp. [7] Deska I. and Malina G.: Influence of fluid properties on difference between actual and apparent LNAPL thicknesses on groundwater table. Proc. 4th Int. Conf. “Oils and Environment. AUZO 2005”. Gdańsk 2005, 120-125. [8] Deska I., Malina G. and Radło A.: Wpływ uziarnienia gruntu oraz cech paliwa na róŜnicę między miąŜszością pozorną i rzeczywistą LNAPL na zwierciadle wody podziemnej. InŜ. Ochr. Środow., 2003, 6(2), 203-220. [9] Deska I.: Ustalanie rzeczywistej miąŜszości lekkich cieczy organicznych na zwierciadle wody podziemnej. Praca doktorska. WIiOŚ Polit. Częstochowskiej, Częstochowa 2008 (niepublikowana). [10] Deska I. and Malina G.: Laboratory verification of models for estimation of LNAPL actual thickness on the groundwater table. Proc. 14th Central European Conference ECOpole’05. Duszniki Zdrój - Hradec Kralove 2005, 55-59. [11] Deska I. and Malina G.: Model empiryczny wyznaczania rzeczywistej miąŜszości lekkich cieczy organicznych (LNAPL) na zwierciadle wody podziemnej. XIII Konferencja Naukowo-Szkoleniowa Rekultywacja i rewitalizacja terenów zdegradowanych. Puck 25-28 kwietnia 2007, 189-200. USTALANIE RZECZYWISTEJ MIĄśSZOŚCI WOLNEGO PRODUKTU ROPOPOCHODNEGO NA ZWIERCIADLE WODY PODZIEMNEJ NA PODSTAWIE WŁAŚCIWOŚCI GRUNTU I LNAPL Streszczenie: W celu odpowiedniego zaprojektowania operacji sczerpywania lekkiej cieczy organicznej (LNAPL) ze zwierciadła wody podziemnej niezbędna jest znajomość jej objętości. Objętość tę moŜna określić na podstawie pomiarów miąŜszości warstwy LNAPL w studniach obserwacyjnych. Jednak rzeczywista miąŜszość LNAPL (na zwierciadle wody podziemnej) zawsze róŜni się od tzw. miąŜszości pozornej (w studni), a róŜnica między nimi zaleŜy od parametrów gruntu oraz właściwości i ilości LNAPL. Istnieje kilka metod ustalania rzeczywistej miąŜszości LNAPL na podstawie miąŜszości pozornej, ale wyniki uzyskiwane przy ich zastosowaniu są bardzo rozbieŜne. Przeprowadzone badania potwierdziły, Ŝe wartości miąŜszości rzeczywistych ustalone na podstawie tych metod mogą być nieprecyzyjne. Podstawową wadą ww. metod jest to, Ŝe w kaŜdej z nich pominięto bardzo waŜne parametry. Na podstawie uzyskanych wyników badań, korzystając z analizy kluczowych parametrów wpływających na zaleŜność między miąŜszością pozorną i rzeczywistą, opracowano modele empiryczne, będące alternatywą dla obecnie stosowanych metod. Uwzględniają one zarówno parametry gruntu (np. współczynnik filtracji), jak i właściwości LNAPL (gęstość, współczynnik lepkości dynamicznej). Weryfikacja opracowanych modeli potwierdziła, Ŝe w przewaŜającej większości przypadków ich zastosowanie pozwoliło na uzyskanie wyników zbliŜonych do ustalonych w warunkach laboratoryjnych. Słowa kluczowe: LNAPL, miąŜszość rzeczywista, miąŜszość pozorna, badania laboratoryjne, model empiryczny