Y. ‘hgimigiiisi‘sinr. an m 9 MM?! mum WHY! “WWW mmmmmmm {£1315 T‘LIBRARIES 111111111 u. |CH|GAN GANUSTATE 1111 11 11 111111 1111‘ 111 1‘11 1 This is to certify that the dissertation entitled AQUEOUS PHASE TRANSPORT IN SOILS CONTAMINATED WITH A MULTI-COMPONENT LIQUID HYDROCARBON AND SUBJECTED TO VAPOR FLOW presented by Michael D. Annable has been accepted towards fulfillment of the requirements for Ph.D. degree in CiVil & Environmental Engineering I anjor professor Date /0 /7 i/ /// MSU is an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY Michigan State 1 Unhealty PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE ____11 i__11____ . 1 , :11—1 11 MSU Is An Affirmative Action/Equal Opportunity Institution cadmium-9.1 AQUEOUS PHASE TRANSPORT IN SOILS CONTAMINATED 'ITH R.HULTI-COMPONENT LIQUID RYDROCRRBON AND SUBJECTBD TO VAPOR FLOW by Michael D. Annable A DISSERTATION Submitted to Michigan state University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Civil and Environmental Engineering 1991 677- 7607 ABSTRACT AQUEOUS PHASE TRANSPORT IN SOILS CONTAMINATED UITH A NULTI-CONPONBNT LIQUID HYDROCARBON AND SUBJECTBD TO VAPOR PLOW by Michael D. Annable Leachate from soils containing an immobile, multi- component, non-aqueous phase liquid was investigated. Soil column experiments, conducted with soil containing three fluid phases, air, water and gasoline, were designed to simulate conditions in the unsaturated zone. The apparatus was capable of sustaining flow of water with a continuous air phase and residual gasoline present in the soil. During flow leachate was collected and analyzed for volatile constituents. Measured concentrations of benzene, toluene, ethylbenzene, m&p-xy1ene, Oexylene and naphthalene (BTEX & N) in leachate were equal to batch equilibrium measurements. Long term measurement of concentrations in leachate showed that there ‘was selective ‘transport of constituents. ‘This leaching process was simulated using a numerical model assuming equilibrium partitioning between phasesc The model suggested that transport out of the column, within the vapor phase, was ii likely occurring-during both the setup and leaching phases of the experiments. Aqueous phase transport was investigated in gasoline contaminated soil columns that had been subjected to soil venting. Each column was leached with water following a different soil venting duration. Aqueous phase transport experiments indicated that chemical equilibrium between phases was reached. Results showed that as venting time increased, gasoline constituents were removed in the order indicated by their vapor pressure; high vapor pressure compounds were removed first. Concentrations in leachate of BTEX & N were reduced over four orders of magnitude, eventually reducing concentrations below 1 ug/l. An experimental technique was developed in which soil columns were subjected to simultaneous flow of air and water. During this process, effluent water samples were collected and analyzed allowing the soil venting process to be continually monitored, as well as, providing a direct measure of the level of remediation achieved. Modelling the selective transport observed in experiments indicated that the majority of BTEX & N removal could be simulated acceptably using a chemical equilibrium based approach. As constituents were reduced to very low concentrations, some deviation between the model and data were observed; rate limited transport between gasoline and air is suggested as a likely explanation. Copyright by MICHAEL DAVID ANNABLE 1991 ACKNOWLEDGBNBNTS I would first like to thank my advisor Dr. Roger B. 'Wallace for’ his encouragement and assistance through. my research. I would also like to thank the members of my committee, Dr. David C. Wiggert and Dr. Thomas C. Voice of the Department of Civil and Environmental Engineering, and Dr. Raymond Kunze of the Department of Crop and Soil Sciences. I am grateful to several groups which have provided support through various programs. Shell Oil Corporation, the Michigan Oil an Gas Association, the Michigan Department of Natural Resources, along with the Institute of Water Research and the Division of Engineering Research and the Civil and Environmental Engineering Department all at Michigan State University. I also thank office mates and friends who have made my studies much more enjoyable including; George Zalidis, with whom I worked very closely in the early phase of my research, Yakup Darama, a good friend through my entire time at MSU, and.Nancy Hayden for the many discussions and suggestions that certainly improved the research. I thank my parents for their continued encouragement and understanding. My greatest thanks and love to my wife Anne. iii TABLE OP CONTENTS 9532133 EASE LIST OEEIGURES ......................................... Vii LIST 0! TABLES ......................................... ix NONENCLATURE ............................................ x CHAPTER I. INTRODUCTION, PROBLEN SCOPE, mucxanom..........................OOO. 1 1.1 Introduction .................... ........... ... 1 1.2 Objective and Scope of the Study ...... ........ 5 1.3 Background .................................... 8 1.3.1 Introduction ........................... 8 1.3.2 Unsaturated Flow in Laboratory Columns ..................... 8 Leaching in Gasoline Contaminated.soils ..................... 12 Multi-Component.NAPL Partitioning ...... 21 Soil Venting ........................... 23 U U Hie H 01¢ .3. .3. 1.4 smary...............................0....... 27 CHAPTER II. A LABORATORY METHOD FOR STUDYING THE AQUEOUS PHASE TRANSPORT OE DISSOLVED CONSTITUENTS FROM RESIDUALLY HELD IN UNSATURATEDSOILCOLUNNS.................. 29 2.1 IntrOduction....................... ........ ... 29 2.2 Experimental Design ................. ....... ... 31 2.3 Apparatus ..........................O0......00. 32 2.4 Materials ..................................... 35 2.5 Procedure.......................... .......... . 35 2.6 Results ............................ ..... ...... 41 2.7 summary ....................................... 46 iv V CHAPTER III. LONG TERM LEACHING OF A MULTI-COMPONENT NAPL, EXPERIMENTAL CONSIDERATIONS . . . . . . . . Introduction .................................. Experimental Procedure ........................ Experimental Conditions ....................... Results ....................................... Numerical Modelling of Results ......... ....... CODCIUSj-ons ..................................O CHAPTER IV. A LABORATORY INVESTIGATION OF GROUND 4.5 WATER LEACHATE CHARACTERISTICS rOLLO'INGBOILVMING................... IntrOduction .................................. Background and.Theory ......................... Experimental Procedure ........................ Experimental Conditions ..... ........ .......... Results ....................................... 4.4.1 Experimental leachate Data ............. 4.4.2 Influence of Soil Venting Duration ..... 4.4.3. Model.Simulation ....................... summaryandconCIuSions......O................ CHAPTER V. A NUMERICAL MODEL AND EXPERIMENTAL INVESTIGATION OF GROUND WATER LEACHATE CHARACTERISTICS DURING SOIL VENTING . . . . . . IntrOdUCtion................ ..... . ............ Numerical Simulation .......................... 5.2.1 MOdeIDevelopment...................... 5.2.2 Scaled Model Simulations ............... Laboratory Experiments .................. ..... . 5.3.1 Apparatus .............................. 5.3.2 Procedure .........................O.... 5.3.3 Experimental Conditions .. ............. . Experimental Results .......................... Discussion .................................... 48 48 52 53 54 65 72 74 74 77 85 88 89 89 96 100 104 106 106 109 109 112 120 120 123 123 124 132 vi 5.6 summary....................................... CHAPTERVI. SUMMARYANDEECOMMENDATIONS.............. 6.1 Summary 6.2 Recommendations............................... APPENDIX A: Detailed Experimental Procedures andErrorEstimates........................ APPENDIX B: Input Parameters for Numerical Model . . . . . . . APPENDIX c: Numerical Code for Venting and or Leachingsimulations LIBTornnrnwcns...................................... 143 145 145 149 151 157 160 163 Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 1.1. 1.2. 2.1. LIST OF FIGURES Gasoline release into the subsurface ..... Matrix suction head vs. elevation in calm................................OO. Soil column for VOC's leaching in an unsaturated state and vacuum control sample collection.system ................. Benzene effluent concentration levels .... O-xylene effluent concentration levels ... Effluent concentrations of benzene ....... Effluent concentrations of toluene ....... Effluent concentrations of o-xylene ...... Effluent concentrations column J ......... Effluent concentrations column.K ......... Effluent concentration benzene for all columns on a scaled time axis ........ Model simulation of columns J & K, leaching process ......................... Model simulation of columns J & K with air flow of 2.7 ml/hr ............... Effluent concentrations of saturated flowc01umn...-000............OOOOOOOOOO. Soil vapor extraction of contaminated soils ....................... Idealized pore with three fluids and conceptual model of pore ............. Time to reach equilibrium concentrations in water as a function of pore water saturation for the conceptual model ...... vii 12 33 42 42 55 56 57 60 61 64 67 68 71 76 81 84 Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure Figure 4.4. 4.5. 4.8. 5.1a. 5.1b. 5.28. 5.2b. 5.3. 5.4a. 5.4b. 5.5. 5.6. 5.7. A1.1. viii Column and collection system ............. 86 Concentration in leachate, column DH vented6hours........................0.0 92 Concentration in leachate, column EA vented48hours......O..........OOO ...... 94 Aqueous phase concentrations for columns vented different durations ....... 98 Simulation of aqueous phase concentrations in vented soils ........... 103 Scaled model simulation of air venting ... 115 Scaled model simulation of air venting on a semi-log scale 116 Scaled model simulation of waterfluShing................ ........... 117 Scaled model simulation of water flushing on a semi-log scale . ............ 118 Apparatus for dual flow experiments ...... 121 Dual flow experimental results ........... 125 Dual flow experimental results on a semi-log scale . . .................... 126 Mass fractions by data integration ....... 130 Experimental results using Teflon ends ... 133 Comparison of model and experimental data, benzene and toluene .... ...... . ..... 134 Comparison of model and experimental data, ethyl-benzene, o-xylene, andnaphthalene............... ........... 139 Column packing device . . . . . . . . . . . . ........ 152 Table Table Table Table Table Table Table Table Table 2.1. 2.2. 2.3. 4.1. 4.2. 5.1. LIST OP TABLES Soil and fluid characteristics .............. 36 Characteristics of four replicate - 8°11c01umn8.......................O.... ..... 37 Effluent concentrations ......... ........... . 43 Initial fluid saturation for leaching columns . . ........ . . ................. 54 Percentage of initial mass removed inleachatewater................. ........... 65 Properties of glacial sand . . . . . ............. 89 Aqueous phase concentrations from ventedSOils......OOOOOOOOOOOO OOOOOOOOOOOOOOO 97 Initial fluid saturations and air flowapplied.........................0.....O.127 Calculated initial mass fractions from dual flow experiments . . . . . . . . . . . . . ..... . 130 ix NOMENCLATURE area [L2] constant = S.” K, constant = term in brackets Equation 5.7 [1] biological losses [MIT] concentration in water [ll/L3] aqueous phase concentration at chemical equilibrium [H/L’] concentration of k in water [M/DH coefficient of longitudinal dispersion [IFWT] flux of water [IF/T] flux of contaminant [M/Dfi gravitational acceleration [141?] concentration of compound k in air [M/Lf) head in porous media [L] vapor concentration over pure constituent [M/Dfi air-water partitioning coefficient [M/L3 / M/L’] soil-water partitioning coefficient [M/M / M/DH concentration of k in the NAPL [M/DH soil sorption partitioning coefficient [M/M / M/Dfi hydraulic conductivity [L/T] soil distribution partitioning coefficient [M/M / M/L’] Darcy's Constant [UT/M] saturated hydraulic conductivity [L/T] column length [L] total moles of k in NAPL [1] X xi total moles of NAPL [1] mass of k in air [M] mass of k in water [M] mass of k in NAPL [M] mass of k in soil [M] molecular weight of i [M] empirical soil parameter [1] number of compounds in NAPL [1] capillary pressure [M/LT’] vapor pressure of component i [MJIHFJ pure liquid vapor pressure [M/LTH column outlet pressure [M/LT’] column inlet pressure [M/LTH flow rate of air [L3/T] fluid flow rate [L’IT] water flow rate [L3/T] radial coordinate [L] radial distance in a pore to the water-NAPL interface [L] radial distance in a pore to the soil-water interface [L] retardation coefficient [1] universal gas constant [MLz/T’] matrix suction head [L] air entry suction head [L] gasoline saturation in porous media [1] concentration of k sorbed [M/M] water saturation in porous media [1] .n on 0 U H < xii time [T] scaled time Dt/rf [1] time required for clrg)/chm to reach 0.99 [T] temperature [1] average pore velocity [L/T] volume conversion factor [1?] volume of air in system.[l?] volume of NAPL in system.[l?] volume of water in system.[l?] initial mass of gasoline [M] mass of soil [M] mole fraction of i in the immiscible phase [1] vertical coordinate [L] Greek Letters rate transfer coefficient [1/T] first order decay constant [1] molecular weight of k [M] pressure in soil water [M/LTH ratio rzlrl [1] soil bulk density [M/Lfi density of gasoline [M/Lfi density of NAPL [M/L’] density of water [M/LH zero order production term [M/Lfifl scaled venting/flushing time cypNt/W; [1] CHAPTER 1 INTRODUCTION, PROBLEM SCOPE, AND BACKGROUND 1.1 Introduction Hazardous liquids immiscible with water pose a complex set of problems when released into the subsurface environment. Examples of non-aqueous phase liquids (NAPL’s) of concern in the environment are single component organic liquids such as benzene or multi-component mixtures such as gasoline or fuel oil. Products such as these can be released into the subsurface from leaking underground storage tanks or piping systems as well as from surface spills. Released.NAPL's flow through porous media in the vadose zone and can potentially reach the groundwater table or saturated flow region (Figure 1.1). Depending on the relative density of the NAPL to water the fluid may float on the water table or sink through the water saturated media. In either case some quantity of NAPL will become immobilized above the water table, in the vadose zone, and can contribute to groundwater contamination for many years. Individual constituents of an immobile NAPL can sorb to soil particles as well as partition to other fluids, such as water, present in thejporouS'media. Effluent water from soils 1 .womuusmnsm on» 09:“ mmomawu mcwaommo .H.H wusofim Zn:GflF-¢m:h:swr=§zé=¢‘N Zn:GfiF-CWhKS?&=<_p ,zuzuwm_mn222n3wm=+h 3 is referred to as leachate, and if contaminated it can contribute to groundwater contamination in the saturated flow region, potentially degrading a usable water supply. The contaminant concentrations in leachate depend on the characteristics of the NAPL and whether the partitioning between phases is at equilibrium, or some rate dependent interphase mass transport is occurring. To evaluate the length of time that a NAPL can continue to contribute to groundwater contamination it is important to know the initial quantity of NAPL present, and the rate at which constituents are being removed. NAPL that becomes immobilized in 'the vadose zone following a release is often referred to as residual or trapped liquid. Due to the effects of capillary forces and the presence of three distinct fluid phases in the vadose zone, mechanical pumping methods are not effective for removing residual NAPL's. Remediation techniques are being developed to address this group of subsurface contaminants. Any technique designed to remediate soil contaminated with a trapped NAPL in the vadose zone must also be capable of reducing concentrations of NAPL constituents in leachate, affecting a significant reduction in the long term threat to groundwater. One technique aimed at removing residual liquids from the vadose zone is soil vapor extraction or soil venting. This method is gaining acceptance and its use continues to increase. Soil vapor extraction uses the flow of air through 4 partially saturated soil to remove volatile contaminants present. Soil vapor extraction is also called soil venting, in-situ volatilization, and soil vacuum extraction. Soil venting can potentially provide an economical and simple means to remediate NAPL contaminated soils. Many aspects of the soil vapor extraction process are not well understood. Simulating the placement of a NAPL into the soil structure and studying the leachate characteristics of vented soils allow a controlled evaluation of the level of remediation that can be achieved with this technique. Multi- component liquids, such as gasoline, present additional complexities and raise concern, about the overallvcapability of soil vapor extraction, particularly with regard to reducing constituent concentrations in leachate. The ability of soil venting to reduce to low concentrations both light, highly volatile compounds, and heavy semi-volatile compounds existing in a NAPL mixture, has not been demonstrated. An investigation was undertaken to address some of the concerns mentioned above. In evaluating the effectiveness of soil venting, the study focused on concentrations in leachate produced by both vented and non-vented soils to better understand the benefits of remediation efforts. Gasoline, a multi-component NAPL, representing a major source of subsurface contamination problems, was used to study the difficulties a mixture would impose on the soil venting process. Studying soil venting capabilities for cleaning gasoline contaminated soils, with regard to these concerns, 5 will help move this technique closer to becoming a well accepted clean-up method or help identify its limitations so that the method can be selected to achieve realizable goals. 1.2 Objective and Scope of the Study The overall objective of this study was to investigate aqueous phase transport characteristics in soils contaminated with a residual quantity of NAPL under a variety of conditions including saturated and unsaturated flow of water, prior to, during, and following the application of soil venting. Studying constituent concentrations in leachate, produced by gasoline contaminated soils, provided a better understanding of the benefits of soil venting as a remediation technique. Specific objectives included.determining if concentrations in leachate were in chemical equilibrium with residual gasoline present in the soil, or if a non-equilibrium interphase transfer model was necessary to describe the process. Another aim was to determine if soil venting could reduce concentrations in soil leachate to acceptable levels from a regulatory perspective. This study was conducted in four phases and the results of each phase are presented in separate chapters. Each of he chapters is written to essentially stand alone and has been or is in the process of being published in a research journal. In the first.phase:of the study (Chapter 2) aqueous phase transport of benzene, toluene, ethylbenzene m&p-xylene and o- 6 xylene (BTEX) in a gasoline contaminated soil was investigated. Soil columns, capable of maintaining unsaturated flow of water, were developed and used to determine how BTEX partition from.immobile gasoline, held at residual saturations in the soil, to the mobile aqueous phase. Experimental procedures were developed to produce soils contaminated with gasoline at low residual saturations. Unsaturated water flow was established within the soil column and effluent samples were collected and analyzed. A sandy loam soil and unleaded gasoline were used. The second phase of the study (Chapter 3) was a continuation of experiments investigating concentrations of BTEX in leachate from gasoline contaminated soils. In this phase, leaching experiments were conducted for long periods of time to observe selective constituent removal from the columns. This selective transport was simulated using an equilibrium based numerical model. .The effect of different initial fluid saturation histories within the soil was also observed. In the third phase of the study (Chapter 4) columns developed for unsaturated flow were modified to allow air flow through the soil. Soils contaminated with gasoline at low residual saturations were vented by forcing a steady flow of air through the columns. Following a period of venting the columns were subjected to saturated and unsaturated water flow to measure changes in BTEX aqueous phase concentrations. During both saturated and unsaturated flow experiments, flow 7 rates of water were varied to determine if the interphase transport process was taking place under equilibrium conditions. A series of soil columns with similar initial conditions ‘were leached. with. water after each. had been subjected to different durations of air venting. The resulting changes in aqueous phase concentrations due to increased air venting were compared with results from an equilibrium based numerical model. The final phase of the research (Chapter 5) involved an experimental technique in which the soil columns were subjected to simultaneous flow of air and water. With this procedure, soil leachate could be monitored continuously for effluent concentrations during the soil venting process. This allowed the evaluation of soil venting effectiveness with regard to soil leachate in a single column experiment. The continuous data was very useful for modeling purposes as well as for determining which processes were effective or controlling during the transport experiments. This experiment also provided an evaluation of the ultimate capability of soil venting to reduce leachate concentrations in gasoline contaminated soils. While each chapter includes references to relevant literature, such that it can stand alone, some understanding of basic background material from the areas of multi-phase flow, inter-phase mass transfer, miscible transport in water, and soil vapor extraction is necessaryu The remainder of this chapter is devoted to a brief discussion of these areas to 8 provide an overview of important information for the reader who may be unfamiliar with one topic or another. 1.3 Background 1.3.1 Introduction A large body of information is available related to multi-phase flow and miscible transport. The areas directly related to this study - water leaching and air venting of immobile NAPL's in soil — are discussed in the following sections. 1.3.2 Unsaturated Plow in Laboratory Columns Unsaturated flow has been simulated in the laboratory using acrylic columns (van Genuchten and Wierenga 1986). The principle component of the soil column is the capillary barrier or pressure plate at the lower end. The plates used have the property of allowing only the wetting fluid (in many cases water) and not the non-wetting fluid (typically air) to pass through when saturated and properly operated. The pressure required in order to force air through the plate is known as the bubbling point or air entry pressure. The plate allows the water in the soil contacting the plate to have a controlled negative pressure or suction (ie. positive capillary pressure) as required for unsaturated flow. This is 9 achieved by controlling the fluid pressure on the lower side of the plate. With a suction maintained at the bottom of the column and a constant flux of water, maintained less than the saturated hydraulic conductivity of the soil and supplied at the top, unsaturated flow conditions can be maintained. Supplying a steady flux of water to the top of the column for a sufficient time produces steady flow conditions. A mathematical solution for steady flow through an unsaturated soil can be obtained analytically for simplified cases (Gardner 1958). We begin with Darcy's flow equation - .93 o, AKUfl dz (1.1) where, Q, = Water flow rate [L3/T] Hydraulic conductivity [L/T] N ll Vertical coordinate [L] .A = Cross sectional area [1?] and the head H’is defined as H= «LP—+2 . (1.2) 9.9 where z is positive downward, p, is the density of water,g is gravitational acceleration and ‘P is water pressure or pressure in the wetting phase. In addition we define the flux: f=32 [E] (1.3) In this analysis it is assumed that the pressure in air, the non-wetting phase, is atmospheric; therefore the negative of the capillary pressure pc, is equal to ‘1’ . The matrix suction 10 head S, is defined as ‘1’ S:—-- 1.4 pg ( ) where S takes on a positive value with units of length in partially saturated soil. Substituting Equations 1.2-1.4 into Equation 1.1 we have: f = K(S)[-g—-‘:+1 (1.5) Appropriateness of signs can be confirmed by noting the following: 1) The + 1 gives a positive downward flux (i.e. in the direction of gravity) and 2) A positive dS/dz, (i.e. higher suctions at greater depths), implies a positive flux downward. For steady flow with a constant flux we can solve for z: = d3 z ‘r—_" (1.5) 1((5') Using a function for 1((3) from Campbell (1985) in the form: K = :95“ (1.7) where, B = Constant = 5,? K, S, = Air entry suction [L] R; = Saturated hydraulic conductivity [L/T] n = Empirical soil parameter [1] then, I z _ d5 = _§ dS §_n_1 f 50-2 (1.8) 3 1" With n=3, a realistic value for soil, the solution is z = -—§—ln[ (7—S)3]+ 1 tan‘ 25+—Y—]+C (1.9) 61"!2 73-53 72/5 N5 11 where, 3=5 1.10 v f ( ) The constant C can be evaluated using a known boundary condition. For example, Equation 1.9 was solved and a plot of matrix suction head vs. depth presented in Figure 1.2 using the parameters estimated for the Metea soil employed in this study (11 = 3, K, = 0.029 cm/hr, S, = 10.2 cm 11,0, and f = 0.015 cm/hr). Figure 1.2 indicates that if a suction head of 60 cm of H20 is applied to the bottom of a column, then a unit gradient condition, such that flow is driven only by gravitational forces, is established in the upper portion of the column. The transition zone from high suction heads to a uniform suction head with elevation is fairly rapid. Given that unit gradient flow conditions, ie. uniform vertical water saturations, is desired in column experiments, the theory indicates that the optimum vacuum to apply is where unit gradient flow occurs for the applied flow of water, or approximately 13 cm suction in Figure 1.2. Applying a suction head slightly above that for unit gradient flow conditions will produce uniform water saturations for the majority of the column. The solution presented is contingent upon having homogeneous soil conditions throughout the column. This may not always be true since the column packing may produce hydraulic soil property variations and the distribution of NAPL may vary from top to bottom. Solutions for other values of n can be found using a numerical integration scheme to 12 solve Equation 1.6. Corey (1986) presents a good discussion of layered soils and pressure distributions under steady flow conditions. action vs. one." 40 ’5 q .0 _ «Wren of Comm 2: . 3, 3 20 - 13 _ w «I! 5 '] +Preeetr. Plato o I I I I I I 0 20 40 D Suction (a m) Figure 1.2 Matrix suction head vs. elevation in column. 1.3.3 Leaching in Gasoline Contaminated Soils If an immiscible fluid like gasoline is allowed to drain from a soil, some amount will be retained in the soil matrix. This fluid, sometimes referred to as residual saturation, can contain constituents that partition to other fluids in the soil.matrixu If fluids such.as air and water are mobile, then contaminants can be transported from the soil. This can be 13 described as a leaching process. A comprehensive review of leaching processes has been presented by Nielsen et a1. (1986). 1.3.3.1 Residual Saturation in Soils To understand leaching in gasoline contaminated soils, it is important to understand the quantity of gasoline that can be retained in vadose zone soils. Given sufficient drainage time, gasoline trapped in porous media above a water table will achieve a hydrostatic condition where the water is in mechanical equilibrium and there is no net driving force. A better understanding of this drained state can be gained by looking at the similar behavior of water in soils. water saturation vs. height above a water table has been studied in great detail. Several methods are used to determine the pressure-saturation relationship for two fluid systems in porous :media. Two commonly ‘used. techniques are column drainage and pressure cell methods (Prill et al. 1965). The two methods are similar, the advantage of the pressure cells being that one can employ a short soil column which can be made to represent any point on the moisture profile by controlling the applied pressure. Prill and Johnson (1967) found that moisture profiles determined using these two techniques were comparable. Soil moisture at great distances above a water table or at high water tensions can be found in a state of pendular 14 saturation (Smith 1933) .' Pendular saturation, defined by Versluys (1917), is described as the region in which single liquid rings exist around soil contact points. For water in a pendular state to reach hydrostatic equilibrium with a water table, water must ultimately be transported through the vapor phase which is a very slow process. Gasoline, like water, can reside in a pendular state in an unsaturated soil profile. Hoag and Marley ( 1986) performed column experiments investigating drainage of gasoline in both dry soils and soils at field moisture. They found that particle grain size was of primary importance to the residual saturation of gasoline. In initially dry, coarse and fine soils, the degree of gasoline saturation varied from 14-55% (by pore volume). They also found a 20-30% reduction in gasoline saturations for initially wet medium sands, and 60% reductions in initially water wet fine sand. ‘Wilson. et al. (1990) also jperformed experiments to determine residual saturation of organic liquids. Sufficient suction was applied to the wetting fluid in order to achieve irreducible fluid saturation (i.e. the saturation at which further increases in applied suction resulted in only negligible decreases in fluid saturation.) In a sandy medium they found organic liquid residual saturations of about 9%. In these experiments Soltrol-130 a mixture of Clo-C12 isoparaffins was used as the organic phase. The study used visual techniques to determine how fluids were held in both saturated and unsaturated soils. They concluded that residual 15 organic liquids were held in the vadose zone as films, pendular rings, wedges surrounding aqueous pendular rings, and filled pore throats. In a study of residual saturations of gasoline in soils having different initial water saturations, Zalidis et al. (1991) showed experimentally that the sum of water saturation and gasoline saturation in a soil was equal to the saturation of gasoline only in an air-gasoline system at the same equilibrium capillary pressure. In other words, the more water that was present in the soil the less residual gasoline one would correspondingly expect when drained to equilibrium. They also found that if water saturations were high (i.e. the saturation of water' was greater' than the saturation of gasoline that would exist in an air-oil system at residual levels), then the quantity of gasoline remaining after drainage was a constant low value independent of the water saturation. The state of immobile residual saturation is desired for unsaturated transport experiments in this study. It is ideal to produce a uniform distribution of NAPL ‘that can be described mathematically, and one at low saturations so that mobilization will not be significant. Residual saturation is also desired since the majority of an unsaturated zone of significant thickness may be in such a state after a NAPL source has been discontinued and drainage to a near static condition has occurred. 16 1.3.3.2 Local Equilibrium Interphase Transport Two methods of describing how contaminants move between phases in a soil are equilibrium and 'non-equilibrium interphase mass transport (Nielsen et al. 1986). It is important to understand which condition is operating during leaching in NAPL contaminated soils. Non-equilibrium interphase transport is discussed in Section 1.3.3.3. In systems operating under equilibrium transport, or where the local equilibrium assumption (LEA) is true, all phases existing in the soil are in a state of chemical equilibrium. The equation used to describe transport of dissolved constituents in water through soil columns, assuming chemical equilibrium between phases, is the classical one dimensional advection-dispersion equation 6c . .62- 92- Rd-a—t- DLaz“ vaz now (1.11) where Rd, the retardation coefficient is defined as: Rae-1 . 9‘31“ (1.12) In these equations c is the concentration in water, t is time, BL is the longitudinal dispersion coefficient in the flow direction 2, V'is the average pore water velocity, K is a first order decay term, a is a zeroth order production term, pb is the soil bulk density, k; is the soil sorption coefficient, and 0 is the volumetric water content. Equation 1.11 has been solved for a wide range of boundary conditions 17 by van Genuchten and Alves (1982). Corapcioglu and Baehr (1987) present a multi-phase transport model for petroleum products which operates under the local equilibrium assumption (LEA). The LEA partitions the constituents between air, water, soil, and immiscible fluid phases. The model is very general and includes bulk flow of all fluid phases. Chemical constituents are partitioned between the phases using equilibrium relationships. The problem of immobile immiscible contaminant leaching into unsaturated water flow is a sub-category of the scope of problems the model can simulate. Baehr and Corapcioglu (1987) simulated immobile gasoline leaching into unsaturated water flow using realistic parameters in their model. In their simulation benzene was substantially depleted first due to its high solubility and low mass fraction in the gasoline. The analysis shows however that ten years must pass before a column of 10 cm in length begins to see significant reductions in total gasoline hydrocarbon in effluent water. Baehr (1987) used the same model to show selective transport of hydrocarbons in the unsaturated zone. The simulations showed changes in concentration for gasoline constituents in the aqueous phase as a function of time. The situation modelled included a surface boundary condition that allowed maximum hydrocarbon loss to the atmosphere. With this boundary condition, the simulation showed lighter hydrocarbon concentrations such as benzene and toluene decrease with time, 18 while heavier, less volatile compounds increase due to their increasing mass fraction in the NAPL. A number of researchers have conducted laboratory investigations of the transport processes in soils containing NAPL. Schwille (1988) performed solubilization experiments using chlorinated hydrocarbons (CHC's). He found that water will become saturated with perchloroethylene (PER) at a flow rate of 1 m/day after passing through only 1 meter of soil at residual levels of PER. From this it was concluded that the assumption may be made with some certainty that a relatively short contact time is needed to produce high CBC levels even with moderate water flow rates. Van der Waarden et al. (1971) performed laboratory experiments on the transfer of residual hydrocarbons to trickling water. They used lOO-cm-long columns with a bottom boundary consisting of a water table. Under these conditions their columns were partially saturated and partially unsaturated depending on the flux of water supplied. They concluded that for non-adsorbing glass-particle packs, the water-extractable components were leached from an oil zone by trickling water at equilibrium concentrations. In these experiments adsorption was negligible because glass beads were employed. In soils adsorption may be significant due to varying mineral composition, clay content and organic matter. Adsorption may act to remove a fraction of the source material and alter its ability to transfer constituents to the leaching fluid. 19 1.3.3.3 Mon-Equilibrium.Mass Transport Situations where the phases present in a soil are not in chemical equilibrium imply that diffusion of contaminants through a region in one of the phases or at the interface of two phases is limiting transport. Non-equilibrium transport behavior can potentially result in significantly lower aqueous phase concentrations. A rate dependent transfer formulation can be used to simulate transport in an aqueous system where chemical equilibrium is not reached. (A simple solution using a linear rate transfer model proposed by Lapidus and.Amundson (1952) is developed here. We solve for the concentration distribution in water along the length of a column where water flows past a uniformly distributed NAPL source. Here the mass flux from the NAPL source to the moving fluid is driven by the difference between the concentration in the fluid, and the equilibrium concentration for the contaminant in the fluid: :11 = —g-(c - cm) (1.13) where, m = Mass transfer from NAPL to water [M/TLH a = Mass transfer coefficient [1/T] = Equilibrium concentration [M/Lfi cut Incorporating Equation 1.13 into the advection-diffusion equation: ig- vfi' (1.14) 20 For steady state conditions where C(zsfl) = 0, the solution from Carslaw and Jaeger (1959) is: a v-(v’ +41%?) (1 . 15) c = cuc e 20L 2 NIH This expression can be used to solve for a when experimental column data indicate concentrations in leachate at less than cut. This equation shows that as the position z, increases, the concentration in water increases and approaches cut. The effect of increasing velocity v causes the system to require longer distances z to reach equilibrium. The above derivation is for a simplified condition where transport is driven by differences between aqueous phase concentration and the maximum achievable concentration, can. A similar but more general model that considers the porous media to consist of two regions has been proposed by Cameron and Klute (1977), and is based on the Coats and Smith (1964) formulation. The model assumes that exchange between immobile and. mobile fluids takes place ‘through a ‘time dependent mechanism. The mathematical formulation of the model which uses subscripts im for immobile and m for mobile is as follows: ac] Ohnacnn E? m.__g;a¢n 1 16 "5? ET: "—322 9,7; " ’ acim _ a __ —6t -E(Cm cm) (1'17) The model can be applied to systems that have water isolated from the region of flow. These could include anything from 21 water located within soil grains to water isolated by large scale heterogeneities. This model is also called the dual porosity model. Experimental work, aimed at investigating non-equilibrium transport in porous media containing immobile NAPL, was conducted by Miller et al. (1990). In their work short water saturated columns packed with glass beads, with immobile toluene, were subjected to different water flow rates in order to test for the existence of non-equilibrium.transport. They concluded that equilibrium was reached over a wide range of aqueous phase velocities and NAPL saturations. 1.3.4 Multi-Component NAPL Partitioning When working with a multi-component liquid such as gasoline, the equilibrium concentrations that result in fluids contacting the NAPL are a function of the NAPL composition. When a pure or single component NAPL is equilibrated with air, the resulting vapor pressure is the same as that over the pure liquid. When the same compound is a constituent of a NAPL mixture the vapor pressure will no longer be equal to that over the pure liquid. The resulting vapor pressure can be described using Raoult's Law (Brown and LeMay 1977): p1 = X11310 (1.18) which states that £3, the vapor pressure of i is equal to Pf, the pure liquid vapor pressure multiplied by X1, the mole 22 fraction of compound 1 in the mixture. As the mole fraction for a given constituent increases or decreases so does the corresponding vapor pressure. Assume the NAPL compounds behave as ideal gases in the vapor phase employ the ideal gas law, ‘ P,v, = nRT (1.19) where V, is the volume occupied by gas, 11 is the number of moles of compound 1, R is the universal gas constant and T is. temperature. Incorporating the ideal gas law into Raoult's Law, the compound concentration in air as a linear function of the mole fraction of i in the NAPL can be determined, G, = H}, x, (1.20) where G, is the concentration in air, and H}, is the air-NAPL partitioning coefficient. The transfer of contaminants between water and air is described by Henry’s Law, which states that the solubility of a gas C, is directly proportional to the partial pressure P, of the gas above the solution (Brown and LeMay 1977): c, = k9, (1.21) The proportionality constant k is referred to as Henry's constant. Using the ideal gas law, Henry's Law can be written in terms of concentrations with a new proportionality constant, Hf.“ valid for constant temperature systems: G 1 H" = 2;? (1.22) The partitioning to soil of constituents of a multi- component NAPL is also assumed to be unaffected by compound interactions and a simple linear relationship S, = KdC, is 23 often applied. Here 55 is the concentration of i adsorbed on the soil and R; is a partitioning coefficient often referred to as a distribution coefficient. Non-linear relationships more appropriate for some soil systems are discussed by Valocchi (1985) and a review of sorption phenomena is provide by Weber et al. (1991). 1.3.5 Soil Venting A limited number of laboratory studies have been performed on volatilization of organic compounds from soils. Marley’ and. Hoag (1984) conducted. venting' experiments in laboratory columns. In their experiments 65-cm-long columns were brought to .residual gasoline levels by allowing a gasoline saturated soil to drain freely. Air flow between 1 and 10 liters/min was then forced through the column. Air samples were collected in gas-tight syringes and analyzed. The column venting results were then modelled using equilibrium partitioning between phases. To calculate the vapor phase concentration of constituent i, Raoult's Law and the ideal gas law were applied to obtain the following equation: G(t), = P1 X125,“ VMW, (1.23) where, V is a volume conversion factor (cm’) and MW, is the compound molecular weight of 1. Using this equation and 51 compounds identified in their gasoline, mass loss rate of total gasoline‘was calculated over 24 the course of venting. These results agreed quite well with experimental data. This modelling effort did not compare individual components of gasoline and how they change with time. By only looking at total gasoline it is not certain that the individual components, each of which is a small fraction of the gasoline, will behave according to the model. Continuation of this research (Baehr et al. 1989) used similar experiments to look at the adequacy of the LEA for modelling the soil venting process. By looking at soil venting at three different flow rates, and using the lowest to calibrate the flow'model, they found the simulation of higher flow rate experiments was good, indicating the process was operating under local equilibrium conditions. Vapor transport in soils has been investigated by Johnson et al. (1987). In examining the role of’mass exchange between vapor and water phases, one-dimensional diffusion as well as advection experiments were conducted. They concluded that mass exchange between pore water and pore vapor is sufficiently rapid to be described as an equilibrium process for most scenarios. They also concluded that removing residual hydrocarbons from the subsurface may be difficult in a short time frame due to the presence of low-vapor pressure compounds found in gasoline, volatile compounds however, can be removed readily. The Texas Research Institute (1983) investigated forced venting to remove gasoline vapors from.a model aquifer in the laboratory. The model aquifer was 3 by 3 meters with a depth 25 of 1.2 meters. The tanks had a water table with a known gradient and three types of gasoline removal were monitored: gasoline as NAPL removed by the flowing water, dissolved gasoline removed in the water, and gasoline vapor removed in the effluent air. Different venting geometries were studied in the tanks using two different sand permeabilities. Work aimed at quantifying the removal of multi-component liquids from soils was presented by Johnson et al. (1990a). They developed a numerical model assuming equilibrium partitioning between soil, water, air, and liquid hydrocarbon. It was shown that a scaled mass removal curve could be developed for’ a given liquid. hydrocarbon and that this relationship could be used to estimate soil vapor extraction times for different initial contaminant volumes and air flow rates. Analytical solutions to problems of flow and transport of contaminants in air were obtained by Wilson et a1. (1988). Their models assumed local equilibrium conditions and took into account compressibility of the gas phase. In a one dimensional laboratory column the pressure distribution and linear velocity were given by P2_P2 3, P‘Z) =[Pg— IL p212 (1.24) where, £3 = column inlet pressure (atm) P; = column outlet pressure (atm) = column length (cm) 26 and, 1 V . £1122 p2-_‘£r’_:P_3->z"5 (1.25) 2L 1 L where: [(0 = Darcy's constant (cm2/atm.s) v = linear velocity (cm/s) Solutions for flow of compressible gas through one dimensional columns showed that compressiblilty becomes significant only at very high flow rates. The gas flux through a column is given by: 00 = urgva (1-25) where: 65 = Flux of gas through the column (mol/s) rc = column radius (cm) G = Gas concentration (mol/cmfl v = soil void fraction Wilson et al. (1988) developed a similar model for radial flow to a well. This model was used to study venting efficiency for various well geometries in field settings. The models were applied to a field data and found that using Henry’s constants for pure compounds gave removal rates much too large. ‘This could have indicated that the effect of other compounds was significant or that diffusion limited transport was occurring. The soil used in this study was very humic, and this was offered as a possible explanation by the authors. In a second paper by Gannon et al. (1989) effects of impermeable caps, soil permeabilities and evaporative cooling are investigated. Wilson et a1. (1987) develop a mathematical model for analyzing forced venting and applied it to the experimental 27 data of Wootan and Voynick (1984). They concluded that the hydrocarbon mixture (gasoline) should be treated as a multi-component liquid. They also found the presence of residual NAPL presented a complex problem and may require modelling random distributions of spherical liquids. While a limited amount of laboratory research has been completed in the venting area, considerable site clean-up has been attempted using this method. Hutzler et al. (1989) present a review of soil vapor extraction technology. Seven pilot-scale and ten field-scale studies are reviewed, and based on the results achieved, Hutzler et al. that soil vapor extraction can effectively be used to remove a wide range of chemicals from the subsurface. Results also indicate that chemicals can be extracted from clay and silt but at a lower rate with intermittent pumping being the more efficient approach. Johnson et al. (1990b) present an overview of in-situ soil venting methods and discuss aspects of the design, operation, and monitoring of field systems. 1.4 Summary The previous sections have reviewed background information upon which the research in the following chapters is based. The general area of study -- air venting and water leaching of gasoline contaminated soils -- encompasses a large body of related research. The present discussion was intended to assist the reader in understanding material in subsequent 28 chapters or alternatively, assist the researcher interested in finding additional published material related to this study. Literature directly related to each phase of the research is discussed in the corresponding chapter. CHAPTER 2 A LABORATORY METHOD FOR STUDYING THE AQUEOUS PHASE TRANSPORT OP DISSOLVED CONSTITUENTS PROM RESIDUALLY HELD NAPL IN UNSATURATED SOIL COLUMNS 2.1 Introduction As the number of subsurface contamination sites continues to rise so does the need for understanding the long-term implications of contaminated soils. Soils contaminated with nonaqueous-phase liquids (NAPLs) in the vadose zone have the potential to release chemical constituents to recharging groundwater for many years. Understanding how these chemicals are transported to vadose zone water will help in decisions governing cleanup of contaminated sites. NAPLs, by virtue of their immiscibility with water, may move as a separate liquid phase through soil. Nevertheless, a soil has the ability to retain a given amount of NAPL depending on the physical and chemical conditions existing at a spill site. This residually held NAPL can partition to other fluids, such as air or water, in the soil matrix. Residual NAPLs threaten groundwater because they can be a 29 30 long-term source of slightly soluble contaminants. A very common contamination situation involves gasoline, a mmlti- component NAPL, held residually in vadose zone soils. Past experiments on contaminant leaching from an immobile NAPL have been reported (Schwille 1988, van der Waarden et al. 1971, Pfannkuch 1984, Hunt et al. 1988 and Miller et al. 1990). Schwille [1988] performed solubilization experiments using chlorinated hydrocarbons (CHC). van der Waarden et al. [1971] performed experiments on the partitioning of residual hydrocarbons to trickling water. Pfannkuch [1984] looked at the dissolution of hydrocarbon components from an oil lens resting on top of a free water surface. Miller et al. [1990] investigated non equilibrium transport from trapped toluene in glass beads. Both Fried et al. [1979] and Hunt et al. [1988] performed experiments on the dissolution of trapped NAPL. All of these studies agree in their conclusion that water extractable components leached at rates determined by their respective water partitioning coefficients. Some conditions used in these studies, such as saturated flow or glass beads, may not be adequate to simulate NAPL leaching behavior in a natural soil setting. There is an apparent absence of methods and previous experiments described in the literature involving the simultaneous use of a multi-component NAPL, unsaturated water flow and real soils. Despite the complexity, it is important to conduct experiments simulating these conditions because all occur in natural settings. The laboratory experiments 31 conducted were designed to a) use real soil, b) attain a uniform residual NAPL saturation, simulating a multi-component spill after the NAPL saturation has been reduced to residual levels, c) establish unsaturated flow over a range of rates characteristic of those found in the vadose zone, and d) maintain immobile air and NAPL phases simulating an immobile source that has not been subjected to advective vapor movement. 2.2 Experimental design Apparatus, materials and experimental procedures were designed to incorporate the above criteria. A sandy loam soil was used with gasoline being the residual multi-component NAPL, while unsaturated conditions simulated the vadose zone. In addition to providing an opportunity to develop an experimental procedure for leaching studies, the experiments were intended to provide preliminary information on the long- term leaching effect of residually held gasoline in soil columns under unsaturated water flow conditions. Contaminant levels in the aqueous leachate were monitored and compared with levels produced when water was equilibrated with gasoline in a batch experiment. The soil column, flow system and collection device were designed to ensure that measured effluent concentrations were the result of transport mechanisms governing the actual process and not artifacts produced by the experimental system. 32 2.3 Apparatus The columns (30 cm long and 5.44 cm in diameter) were constructed with materials selected to minimize sorption losses. They included borosilicate glass for the column walls, Teflon o-rings to seal the end parts, a Teflon cap on the downstream end and stainless steel fittings (Figure 2.1). The column design allows easy assembly or disassembly and can easily be adjusted to various lengths using different length glass tubes. For some applications column end caps should be constructed of stainless steel. For example our experiments have shown that Teflon end caps having contacted gasoline (NAPL) are difficult to clean. .After the Teflon end caps were rinsed with hexane and acetone and dried, clean water passed through an empty column showed measurable BTEX levels. For experiments where this would affect the results stainless steel is recommended. A saturated ceramic pressure plate, having a nominal gasoline bubbling point of 500 mbars, was placed in the downstream end of the column prior to leaching in order to establish residual levels of gasoline in the soil. A porous stainless steel pressure plate, with a nominal water bubbling point of 250 mbars, was maintained.in the bottom.of the column during leaching to establish unit gradient conditions. A constant low water flux was delivered by a multi-channel 33 Inflow From Pump 1 /Venting PM ugflflmm P °J 1 V4 " Glass Walls Sepmm with _ i | Plus Threaded Rod Tensiometer w o Preunre Plate; ' FTeflon O 41an WSW—...... 1 From Column Stainless Steel Valve \ 118 " Teflon ‘Ihbe Connected To Regulated Vacuum ‘— Sample 1 Liter Vawum Flask 1 Vacuum Transferred Through Syringe Figure 2.1 Soil column for VOC’s leaching in an unsatu- rated state and vacuum control sample collec- tion system. 34 syringe pump (Wierenga et al. 1973) . To maintain the critical seal with the Teflon o-rings at the pressure plates the edges of the plates were sealed. This was accomplished by grinding the edges of stainless steel plates or sealing the edges of ceramic plates with an epoxy paint. A venting port consisting of a 30 gauge needle next to the inflow port was provided. This can be opened or closed to control vapor phase movement during the experiment. Tensiometers with stainless steel plates were used to monitor the distribution of water pressure during unsaturated flow. A hand-held pressure transducer (Tensimeter; Soil Measurement Systems, 7344 N. Oracle Road, Tucson, AZ 85704, USA) was used to periodically read the tensiometers. A sample' collection device (Figure 2.1) was developed that minimized volatilization and adsorption losses. The samples were collected in clean glass syringes within a vacuum flask. The syringes were attached directly to a short Teflon tube which was attached to the outlet of the column with a stainless steel fitting. A constant vacuum was maintained in the flask with a vacuum regulator. The low friction of the glass syringe allowed the vacuum to be transferred to the pressure plate. A stainless steel valve in the Teflon tube could be closed when the syringe was removed from the flask in order to keep the pressure at the plate constant during sample removal. This valve also prevented air entry into the tube when the syringe was disconnected. Leachate samples collected 35 with ‘this apparatus had. minimum. exposure to air during collection. 2.4 Materials 0.001 M CaSO, deaired water was used as leaching fluid to avoid clay dispersion and air entrapment. Metea sandy loam (Table 2.1) soil was selected for these experiments. ‘The soil was wet sieved to remove particles smaller than 0.0075 mm and larger than 800 mm. A large quantity of unleaded gasoline, obtained in April, 1989, was stored in sealed glass bottles with minimal headspace to provide a consistent NAPL source. The five individual compounds.of interest and their'respective mass fractions were benzene 0.017, toluene 0.065, ethylbenzene 0.015, m- and.p-xylene 0.059 and o-xylene 0.029. 'Periodically the mass fractions of these compounds were measured using gas chromatography in order to ensure that the gasoline remained unchanged. The soil and fluid characteristics are described in Table 2.1. 2.5 Procedure The procedure discussed below describes soil packing, introduction and redistribution of NAPL to establish residual levels, initiation and maintenance of unsaturated flow, effluent sample collection and concentration measurement. Reproducible soil packing was ensured by using a device 36 Table 2.1. Soil and Fluid Characteristics Properties of Metea sandy-loam soil Particle Fraction' Medium Sand Fine Sand Fines 28.2% 60.4% 11.1% Saturated hydraulic conductivity: 0.075 cm/min Porosity: 0.41 m3/m3 % organic carbon: 0.44 Particle density: 2.61 Mg/m3 Classification: Arenic Hapludalts, sandy over loamy, mixed mesic Origin: Michigan State University Research Farm in East Lansing, north central portion.of Ingham.County, between 42 and 43 degrees latitude and 84 and 85 degrees longitude Fluid Characteristics (temp = 2011 C) Fluid Density Dynamic Surface Viscosity Tension (Mg/m3) (cp) (dyne/cm) Water(0.001M CaSO4) 1.00:0.002 O.98:l:0.01 71.7510.92 Gasoline 0.7210.02 0.35:0.07 21.99io.44 1 indicates one standard deviation 37 (Zalidis, 1987) that maintains a uniform bulk density within and between different columns. The packing device consists of a reservoir to be filled.with air dry soil, a valve to control the flux of soil, and a 50 cm long, 3 cm diameter tube with a 2 mm screen in the lower part to help maintain a uniform distribution of soil particles. The screen was positioned 5 cm above the soil surface and maintained in this position by moving the packing device upward with a pulley. The tube was rotated to ensure uniform distribution of soil particles. The bulk densities of four columns packed with the Metea air-dried soil were very similar (Table 2.2) , indicating that our attempt to pack the columns reproducibly was successful. Table 2.2. Characteristics of Four Replicate Soil Columns Column Residual Bulk Density Saturations A 0.110 1.54 B 0.124 1.54 C 0.136 1.54 D 0.119 1.52 Average 0.122 1.53 Standard Deviation 0.0099 0.01 Water Flow Rate 1.07 ml/hr One-hundred milliliters of gasoline were injected into each.packed column over'a.period.of 30 hours. ‘This controlled infiltration was slow enough to prevent air entrapment. 38 During gasoline introduction, in addition to controlling the infiltration rate, the venting port was kept open, further minimizing the chance for air entrapment. This volume of gasoline represented a gasoline saturation (0.34) which was well above residual saturation levels (Table 2.2). The gasoline was introduced into the top of the column and then brought to residual levels by applying a vacuum of 300 mbars to a gasoline saturated ceramic plate at the bottom of the column. The vacuum was maintained for one week, which was sufficient to allow gasoline drainage to cease. Residual gasoline saturations were measured in the four columns and were found to be 0.12, with a standard deviation of 0.0099 (Table 2.2). This value agrees well with pressure-saturation experiments on Metea soil at an equilibrium pressure of 300 mbars. Agreement between pressure-saturation data and the columns brought to 300 mbars with respect to the gasoline indicates that the long columns have a uniform distribution of NAPL. After the columns were drained to residual gasoline levels, gasoline saturated ceramic plates were replaced by oven dried (104°C, 24 hours) stainless steel pressure plates. At this point column caps used to bring the soil to residual gasoline were cleaned or replaced before water flow was initiated. Air displaced by the wetting front during the initial stages of the experiment must be released so that excess air pressure does not develop. This was accomplished by using a very small initial flux and maintaining an air- 39 dried bottom plate until the wetting front arrived. A vacuum was applied to the lower end of the column when the infiltrating water reached and saturated the pressure plate. This vacuum was necessary to produce unit-gradient unsaturated-flow conditions throughout the soil column. Initially, the water flow was less than 1 ml/hr with a vacuum of 70 mbars. Consistent readings of -35 mbars in both tensiometers were attained using a vacuum of 50 mbars and water flow of 1 ml/hr indicating steady unsaturated flow. Water suction measured by the tensiometers agreed well with predicted values from a steady state solution for unsaturated flow (Gardner, 1958) using estimated.hydraulic properties for our soil. Effluent sample collection was initiated by attaching a glass syringe to the outlet of the column, reapplying the vacuum and opening the stainless steel valve. After more than five milliliters of sample were collected, the valve was closed and the syringe detached. The syringe was then fitted with a clean, stainless steel needle and five milliliters of leachate were transferred into a clean headspace vial (20 ml glass serum bottle with Teflon-faced septum and aluminum crimp cap) and immediately sealed. Samples were then labelled and stored in the dark, at 4°C until analysis. Samples were analyzed within an appropriate period of time. The quantification of soluble gasoline components in leachate water was performed using headspace gas chromatography with a flame ionization detector and external 40 standards using the method of Voice et al. [1990]. A multipoint calibration curve was determined for each of the five components of interest. A DB-624 megabore capillary column (30 m * 0.53 mm i.d., J & W scientific) was employed. The initial oven temperature‘was 60°C ramping in 10 minutes to a final temperature of 200°C. The injector and detector were maintained at 200°C. The sample equilibration temperature and time were 80°C and one hour respectively. The retention time of m- and p-xylene, using the technique outlined above, were the same; therefore, these two compounds were quantified as a single component. Batch equilibrium experiments were conducted for gasoline following the method outlined by API [1985]. Overall the experimental procedure succeeded in producing and controlling the desired conditions and processes. It was able to a) reproducibly pack a real soil b) create and maintain a residual immobile NAPL source, c) simulate unsaturated conditions without altering sample effluent concentrations as the result of the imposed vacuum, and d) maintain immobile air and NAPL phases using pressure plates, a well-sealed design, and non-adsorbing materials. In addition, the sampling' procedure employed. did. not allow subsequent losses of the volatile components. 41 2.6 Results Effluent concentrations of benzene and o-xylene are shown in Figures 2.2 and 2.3 for a leaching period of one month. Concentrations for the other components of interest are not presented although they exhibit similar trends to the data presented here. The straight lines represent aqueous-phase batch equilibrium concentrations for the corresponding compounds partitioning between water and gasoline used in this study. Results from the four columns presented in Figures 2.2 and 2.3 and in Table 2.3 indicate good reproducibility between these columns. The table shows values of time-averaged concentrations of the five compounds associated with each of the four replicate soil columns; the standard.deviation.of the instantaneous concentration. measurements about the time- averaged values are also provided. The four values of time- averaged concentration for each compound were used to obtain the four-column average concentrations and associated standard deviations for each compound. The good reproducibility between columns is shown by the small four-column standard deviations. Table 2.3 and Figures 2.2 and 2.3 show that effluent benzene concentrations were slightly below batch equilibrium concentrations while this relationship was reversed for o- xylene. As pure compounds, some of the organic chemicals, Benzene Concentration (mg/l) Figure 2.2. O-Xylene Concentration (mg/l) Figure 2.3. 42 1 “ML?" ‘5' 'v ° 01" 0 e e 20d 1 1 Column 0 o-W ‘ a V 1 Y j W f 1' v 1 ' V ' j V V V ‘ nef—oo—f‘r'"? * 5,. O 1 4 204 ' q 3 Column C ‘0 u . . 4 . C O 4 o Q O . . 0 C O O o a O 201 o 1 Column 8 40- 3 I ‘ .e '5" 8 0 . 3° ° e 0"“, ° 0 e . ° 0 e . w . 20.4 1 Column A n 0 5 ‘0 15 20 25 30 TIME (days) Benzene effluent concentration levels. .—WWW ColumnD lO- 0 o o o. 9.. o o (#7 n v Y‘ 'Vj"roYYT’V'Y‘VYT'T‘V‘YVYY‘ 4 ColumnC 1.. 0 . o 0. £. @0099 0 OJ YVVVYVWVVTVVTVIV v VfI VVY V'Y‘VYV‘ ColumnB 10-1 . Gee ”0°. 'fio"! .0 e e e 9 O o V '7' ' VV'Vfi' 'Y7‘ 'VYYv‘fiVVYTYYVT' 1 ColumnA 10~ . o 0‘0 w°e- A“'0000 0. o e x 0 o VVTV'V'YVIY" "fi‘vv'fiv‘Vv'v 'Vrl ll 5 lo 15 20 25 30 TIME (days) O-xylene effluent concentration levels. 43 mad $6 $6 2.6 counten— emflo>< 85309 3803. 33.59 06:. Emu—Sm van—mam $6 «6 $6 $6 $6 3.0 hvd use—.mXAv .meaudHuddOdoo usuaauuu mod. end «.2 3... 3.2 ao.~ 59.2 2.— Sam— _n._ mafi— 25,—. 863x12 . m. N OHQUB :6 3.3. 3.3. 3.: an; 3.— 35 undo mém and aaé n~.m 3.3. «Gm 3.5 3.6 36 a. .v 3.: Na.~ mndm 2..— 36 m_.w 5.5. 36 mean 3.6 San cod 3%: Rd 3.3” omEo>< cogen— om80>< c2339 own—03. Eaveflm 08F Beam 2:2. macaw 25,—. omensomé ode—Boa. econ—sum Eons—E850 Sagan 85m 8855 288. own—o}. 5:28 Son 55.8 44 such as benzene and toluene, exhibit moderate solubility (1800 and 550 mg/l respectively) in the water phase. On the other hand, when they exist as part of a gasoline mixture, their aqueous phase concentrations decrease to approximately 58 and 33 mg/1 respectively (API 1985) . These values vary depending on the composition of gasoline used. Changes in aqueous phase concentrations for all gasoline components will occur whenever the mass fraction of individual components changes. Such changes may occur as a result of losses in one or more of the compounds. For example, reduction of benzene concentration in gasoline, would be expected to produce a reduction in the aqueous phase benzene concentration. Such reductions may result in increased aqueous phase concentrations of other gasoline constituents because their mass fraction in gasoline exhibits a net increase. This is a likely explanation for the results of our column experiments (Tables 2 and 3); measured effluent benzene values were slightly below batch equilibrium levels while o-xylene values were slightly above due to this mass fraction change in the remaining residual NAPL. Volatile losses during gasoline introduction in the batch equilibrium experiments were likely to be significantly lower than losses during gasoline introduction into the columns. Although every precaution was taken, small losses of the more volatile gasoline constituents, including benzene, could have occurred. This initial loss of certain gasoline components may result in small compositional changes within the gasoline during the initial stages of our experiment. 45 Contaminant transport from an immobile NAPL source to a mobile fluid can be described mathematically using either a local equilibrium assumption or a mass transfer formulation. The local equilibrium assumption (LEA), treats all phases as existing in a state of chemical equilibrium. Partition coefficients describe the distribution of constituents between the air, water, Soil and NAPL phases. The LEA should apply under conditions for which mass transfer between phases is fast relative to the rate of the advective mass transport. If mass transfer rates are slow relative to advective mass transport, the LEAuwill not be applicable (Bahr et al., 1989). In our study the concentrations for all five compounds were approximately equal to the batch equilibrium levels over the 30 day period of time. This suggests that the LEA was satisfactory for describing the transfer of soluble hydrocarbons in our experiments. This is in agreement with other recent literature (Miller et al. 1990, Schwille 1988, Fried et al. 1979, and van.der Waarden et al. 1971). The fact that aqueous-phase concentrations for the five compounds remained relatively constant for the entire 30-day period demonstrates that the mass fractions for these compounds in the residually-held source of hydrocarbons is essentially unchanged. Hunt at al. [1988a] developed theoretical predictions of single component NAPL.dissolution into groundwater, and based on that, calculated the lifetimes of trapped NAPL under flooding. It is possible to calculate the lifetimes of the 46 five compounds in our experiments directly from our data based on the assumption that effluent concentrations will stay close to batch equilibrium values. These lifetimes were calculated based on the experimental gasoline residual saturations and fluxes. For these conditions the calculated lifetimes for benzene, toluene, ethylbenzene, m- and.p- xylene and o-xylene were 430, 1540, 4670, 6020, and 4870 days respectively. Caution should be taken before interpreting field conditions on the basis of these values for the following reasons: a) the annual average fluxes in the field could be much lower than the ones employed in this study; b) the vertical extent of the spill in the field would probably be greater than the length of column ‘used in 'this study; and c) estimated lifetimes based on batch equilibrium concentrations will be short, even under the LEA, since the lifetimes are.governed by the mass fraction of each component which will eventually decrease due to varying solubilities of the compounds present in gasoline. The above reasons, especially the first two, can greatly increase the true lifetimes. 2.7 Summary An experimental technique has been developed and tested} which is fundamental for studying the movement of dissolved volatile compounds from an immobile NAPL source under unsaturated water flow conditions. This capability now enables the study of the long and short term behavior of 47 immobile volatile multi-component contaminants in a real soil environment, the associated mass transport mechanisms, and the physical and chemical factors influencing these phenomena. In addition, it is an excellent experimental tool for verifying the ability of various remediation techniques, such as forced air venting, surfactant flushing or bioremediation, to remove immiscible contaminants from the soil. The effectiveness of these techniques can. be evaluated on. the basis of the resulting leachate concentrations. Measured effluent BTEX concentrations were reproducible between soil columns and remained relatively constant, at levels that ‘were very close to batch equilibrium concentrations. This suggests that a local equilibrium assumption is valid for the conditions employed.in this study. A minimum estimate of the lifetime of the five compounds was made based on the assumption that effluent levels remain close to these equilibrium levels. CHAPTER 3 LONG TERN LEACHING OP A HULTI-COHPONENT NAPL: EXPERIMENTAL CONSIDERATIONS 3.1. Introduction When liquid hydrocarbons have been released into the subsurface, the potential exist for hazardous constituents to partition to water in the porous media. This water can then enter the saturated groundwater flow system, contaminating a valuable water source. Understanding the long term effect of any immiscible or non-aqueous phase liquid (NAPL) trapped in the unsaturated zone will help to evaluate the need for and advantages of efforts at remediation. While improving this understanding as a goal, laboratory experiments were undertaken to simulate leaching of gasoline contaminated soils with steady unsaturated flow of water. .A specific aim of this study was to consider experimental procedural influences on leachate results. In support of this objective a simple chemical equilibrium. model was employed to simulate the chemical transport process. In order to simulate leaching processes in the 48 49 unsaturated zone, experiments were setup with conditions representative of a region below a release of a NAPL. The conditions assumed to exist in the unsaturated zone, for the purpose of this study, are those found after a release has been discovered and the source of NAPL has been eliminated. The remaining NAPL in the subsurface has subsequently drained, possibly reaching the water table should a large enough quantity have been released. After a sufficient period of time drainage of NAPL will become very slow and the bulk of the unsaturated soil above the water tabLe will be at low saturations of NAPL (Figure 1.1). This condition will be referred to as residual saturations and can be simulated in the laboratory by controlling fluid.pressures using capillary barriers. This representation of a field setting has assumed homogeneous soil conditions and could be greatly complicated by soil heterogeneities. When percolating water flows past an immobile NAPL, chemical partitioning between phases takes place. If the partitioning is at equilibrium, the concentration produced in the aqueous phase for a given compound is proportional to the concentration of that compound in the NAPL. If the NAPL is composed of a single compound the resulting aqueous phase concentration will be equal to the compound solubility at the system temperature. In a multi-component NAPL such as gasoline, the resulting aqueous phase concentration for a given compound will not be at solubility but a fraction of that depending on the composition of the NAPL. As a multi- 50 component NAPL is leached by water percolating through a soil, the composition of the NAPL can change. Compounds having high ‘solubility in water may be removed from the NAPL at a higher rate than low solubility compounds, this can produce a NAPL that is enriched in low solubility compounds. The resulting concentrations in leachate for these compounds will in turn increase. This process produces selective removal of constituents from NAPL and is fundamental to understanding the long term leaching behavior of soils contaminated.with multi- component NAPL’s. A number of studies have investigated leaching of immobile NAPL's in porous media (Miller et al. 1990, Schwille 1988, Hunt et al. 1988, Pfannkuch 1984, Fried et al. 1979, and van der Waarden et al. 1971). Miller et al. (1990) performed experiments aimed at determining rate limited transport coefficients for saturated flow in porous media containing trapped toluene. They found that for flow rates realistic for groundwater flow systems partitioning from toluene to water was at equilibrium. Additional studies of leaching with trapped NAPL’s in saturated flow include Schwille (1988) and.Hunt.et al. (1988). Hunt et al. included gasoline as well as pure hydrocarbons. Dissolution of a gas—oil was investigated by Fried et al. (1979). Selective impoverishment.of the oil.product with.time due to differences in the solubilities of the hydrocarbons was Observed. A study on the transfer of hydrocarbons from residual oil to trickling water was performed by van der 51 Waarden et al. (1971). The oils used included gasoline, kerosine and gas-oil. All of these studies supported the idea that under normal groundwater flow conditions the contaminants in the aqueous phase were at equilibrium concentrations for the hydrocarbons present. Research investigating compositional changes of multi- component liquid hydrocarbons during leaching is limited. A study by Eastcott et al. 1987 used soil box experiments to look at leaching of diesel fuel and a synthetic hydrocarbon mixture. The experimental design exposed the hydrocarbons to evaporation and bio-degradation. Preliminary simulations of the transport processes were conducted. The compositional changes of the hydrocarbon mixtures were discussed in qualitative terms by categorizing grOups of compounds that elude together in gas chromatographic analysis. Generally they found that the character of the liquid hydrocarbon after exposure to evaporation and leaching could be simulated assuming chemical equilibrium between phases. The present study looks at individual hydrocarbon aqueous phase concentrations over long periods during unsaturated flow in soil columns. The observed data are compared to that predicted using a simple equilibrium based mass removal model. Considerations of experimental methods are discussed for improving studies of leaching of multi-component NAPL's. 52 3.2. Experimental Procedure Leaching experiments were conducted using laboratory columns with conditions designed to simulate those existing in the vadose zone. The details of the columns and methods used are presented in Chapter 2. The fluids present in the soil when leaching was initiated were intended to simulate those at a NAPL release site where drainage to a deep water table has occurred. In some columns water was imbibed into the soil through the capillary barrier and later drained to residual levels. NAPL was introduced to all columns using syringe injection and allowing the NAPL to distribute throughout the columns. The NAPL was then drained to residual saturations using a gasoline saturated capillary barrier producing a contaminant intended to remain immobile for the duration of the leaching experiment. Steady water flow, under partially saturated conditions, was established by applying a constant source of water to the top of the column and. maintaining' a constant capillary pressure at the bottom by applying a negative fluid pressure to ‘the water ‘through a stainless steel water saturated capillary barrier. The flow rate of water and capillary pressure applied were designed to produce uniform fluid saturations vertically or unit gradient flow conditions throughout the columns. Effluent samples were collected and analyzed for benzene, toluene, ethylbenzene, m&p-xylene, and 53 o-xylene (BTEX) using a head space gas chromatographic analysis method described in Chapter 2. With this experimental setup 'the :removal by ‘water of constituents initially within a residual NAPL could be monitored over long periods of time. 3.3 Experimental Conditions Six 30-cm-long columns containing residual saturations of gasoline were leached with water for extended durations. The columns were packed with air dry Metea soil (Table 2.1), a sandy-loam, and brought to initial fluid saturations prior to the initiation of steady unsaturated flow of water. Columns E-H had gasoline as an initial residual fluid on an air dry soil, while J & K had residual water present prior to the introduction of gasoline. Residual gasoline saturations are lower in the columns that had been brought to residual saturations of water (Table 3.1). The columns were leached at very low flux rates to maintain unsaturated flow conditions at low water saturations. The flux rate (2 ml/hr) and suction applied to the lower pressure plate (60 cm of H20) resulted in an average saturation level of 0.40 or approximately twice the residual saturation for this soil. The flux rates used are equivalent to an annual application rate of 755 cm/year, much above typical infiltration to groundwater. 54 Table 3.1. Initial saturations of water and gasoline for leaching columns. Column S, S, E 0.016 0.11 F 0.016 0.12 G 0.016 0.14 H 0.016 0.12 J 0.20 0.051 K 0.21 0.049 3.4 Results The column experiments discussed here were leached over periods of up to 14 months. The first four columns (E-H) had initial conditions of residual gasoline in an air dry soil. Leaching was then initiated and continued for a 9 month period. The aqueous phase concentrations for benzene, toluene and o-xylene are presented in Figures 3.1a, b & c as a function of time. Early in the experiments there is evidence of NAPL gasoline passing through the capillary barriers and into the sample vials. The measured high concentrations, much above the equilibrium values for fresh gasoline determined in batch experiments (41.9 benzene, 44.3 toluene and 5.19 o- xylene), and well above the majority of the measurements, 55 .OCTNCOQ MO mCOHUMHHCTUr—OU HEODHHHm .mH.m wusmfim I 4 O O n— + m D Amxoov 08:. "wCESHOU owN O¢N OON 00 r ONP 0w 0.? .1fiflf¥¢.q _ _ n _ q _ _ _ _ . Ea. "0mg: 03.0 U D 3% H.859 3 HTML .# .Tflww. U + g 1+ o 0— ON on 0+. on Om. Ox: uogioliuaauog auazuae (I/Bw) .OCOSHou uo mcowunuucuocoo anomauum .na.n Guzman I d O O n.— + m D 56 Amxoov oEE. "mgaou O¢N OON Om? ON_. GO O¢ O E _ _ _ a d _ fl _ _ _ _ _ O 1 OF 1 ON + a £454 66 4 1 on +4 0.88.56 5.55% a. a. + 90 as 1 9. bamfiw awe A. Amw¢ Bu D fl. bAv + Av 0 ms nu Dn_ +&¥w L on 38% 5% e. D Did? 1. 1 OO 1 Oh I ow 1 Om a < 4 OOP uogimiuaauog auanloi (I/Ew) 57 owcwflglo MO mCOflHMHHn—OUGOU HGTSHHHW O 0 ... Amxo 3 0E:. ON— .OH.n uudmwm "msESHOU mm 4 O— ON on O¢ on uononuaauog auaMx-o (I/fiw) 58 indicates that NAPL gasoline had entered the sample vials. Visual evidence of an oil sheen could be seen in some early samples also supporting this conclusion. This may explain some of the scatter present in the four column data set as the amount of gasoline in the columns is unknown and may not be reproduced between columns. This observation highlights the need to create an initial condition in which the NAPL would remain immobile. The continuation of BTEX measurements shows that NAPL movement out of the columns had stopped after approximately a one month period. Following this, benzene concentrations in effluent water were reduced continually while toluene and o- xylene over the same time period were increasing slightly. This is an indication of the selective transport process acting to remove constituents of a multi-component liquid hydrocarbon. This process can be clarified by looking at a form of Rauolt's Law, Ck = 5:: XI: (3.1) which states that the concentration of constituent k in water, C}, is equal to the solubility of k in water, 5%, multiplied by the mole fraction of compound k in the NAPL, Xk. This relationship can explain the rising trend in the toluene concentration. As the mass of some soluble fraction of gasoline is removed, the mole fraction of toluene for example remaining in the mixture.could increase resulting in increased concentrations in water. This behavior will be investigated later using a simple equilibrium partitioning model. 59 Experimental results from columns J and K, which had initial water present prior to the introduction of gasoline, are presented in Figures 3.2 & 3.3. The results show no evidence of gasoline passing through the capillary barrier as in columns E-H. This discrepancy in the experiments can be explained by looking at the wetting sequence in a three fluid system. In soils, water is considered the wetting fluid and oil (gasoline) is the intermediate fluid between air and water (Corey 1986) . The relationships between residual fluids in three phase systems is described by Zalidis et al. (1991) . In pressure cell experiments, they found that if the initial saturation of water in a soil was greater than the residual oil saturation determined in an air-oil system, then the resulting residual saturation of oil in this wet soil was at a low saturation that was independent of the initial water saturation. They defined the initial water saturation, above which residual oil saturations are independent of water saturation, as the critical saturation, and the saturation of oil in this region as a minimum oil saturation. This is the case that exists in columns J & K where residual gasoline saturation is 0.050. The initial water saturations were well above residual gasoline saturations found in the air dry soils (ie. 0.20 vs. 0.13 Table 3.1). As the water saturations increased in columns J & K after water flow was initiated, the gasoline should not mobilize as gasoline saturations are at minimum values. In the columns brought to residual gasoline in an air dry soil (columns E-H) the residual gasoline 60 .b sasHoo mcofiumuusmosoo Hammauum 4 comm Amxovv 0Ctk 0 .0H .N.n mfisdmwm Or ON On O¢ On 00 Oh uogionuaouog aioqooaj u! (I/fiw) 61 .M sasdoo ecoHunuucuosoo ucusauum a comm Amxoov oath Av _0H .m.m Guzman + com n. Or ON on Cd 00 OO Oh U011DJ1U83 U00 810143081 U! (|/6L.U) 62 saturations are higher 0.13 and. much above the minimum saturation defined above. In these columns it would be expected that as water enters the soil some gasoline may be displaced by water resulting in a condition where gasoline saturations, intermediate between water and air, are higher than the minimum saturation and enough to induce a measurable flow of gasoline. These results point out the need to have initial water saturations, prior to placement and drainage of gasoline, higher than the fluid saturation dictated by the air-oil residual saturation for the soil in order to avoid mobilization of NAPL. The concentration in leachate for columns J and K (Figures 3.2 & 3.3) is more continuous than columns (E-H) and show very good agreement with each other. The increasing aqueous phase concentrations of toluene is more evident and all compounds other than benzene show concentration increases before eventually decreasing. As discussed earlier this selective process can be predicted making use of Raoult’s Law for multi-component hydrocarbon mixtures. The effluent water concentration data from all six column experiments cannot be compared on actual experimental time scales due to the different quantities of gasoline initially present in each column. By scaling the time axis using the initial mass of gasoline present and the flOW'Of water applied this problem is overcome. A dimensionless time is defined t Qw 9N W. 1. T = (3.2) 63 where, 0,, Flow of water [L3/T] W. 1 Initial mass of NAPL [M] pN Density of NAPL [M/L’] This scaled time is simply a ratio of the volume of water passed through the column to the initial volume of gasoline present. Scaling the data for all six columns using T shows similar trends with a possible difference between columns with and without initial water present (Figure 3.4). Column E, without initial water present fits well with columns that had water present. One possible explanation may be associated with the presence of vapor transport as described later. To further investigate the mass transport process, accumulative mass removed is examined. The total benzene removed from the columns by water can be determined by integrating the areas below the curves in Figure 3.4. After integrating to determine the quantity of mass removed, this can be compared to the initial mass present in the columns at the beginning of the experiments. The percentages of the initial mass that can be accounted for in leachate water are presented in Table 3.2. The results indicate that in [general less than 50% of the initial quantity of benzene present has been removed within the effluent water. This suggests that some mass has either left the system unaccounted for or remains in the system but does not partition to water. Soil samples collected when the columns experiments ended were analyzed for BTEX. The results for benzene in columns J & K Showed very low levels in the soil samples that when summed .mflunfl 05w.“ UOHOUM M £0 mad-HOD HHM .HOH OCONCB COflUMHUn—OUCOU UCQfiHHHm X D 1 X I Q 0 O ... + m D ... 0E2. 8.00m 555500 O 64 .1...” 0.962 or ON on Cd On 00 Oh UOl‘lDJlUSDUOO BUBZUBB (I/Bw) 65 for the entire soil mass would only represent 2% of that introduced. This suggest that the benzene has left the system or been degraded. A likely possibility is that losses to the atmosphere are occurring during the leaching process. This would tend to remove the more volatile fraction of the gasoline and may explain why heavier molecular weight compounds increased in effluent concentrations initially before eventually decreasing. To better understand the mass removal process a simple mathematical model will be employed. Table 3.2. Percentage of initial mass removed in leachate water. Column Benzene (%) Toluene (%) E 45 ~40 F 29 ~29 G 25 -- H 29 -- J 49 51 K 54 51 3.5 Numerical Modelling of Results The processes removing BTEX from the columns was investigated using a simple numerical model . The model assumed chemical equilibrium between NAPL, water and air, for individual compounds found in gasoline. The model assumed 66 that the distribution of phases and constituents within phases were uniform throughout the columns. The model simply kept track of the mass of gasoline components present in the column and removed mass from the system within the flowing fluids, ie. water and or air. The model used was based on a version by Marley and Hbag (1984) and that used by Johnson et al. (1990). The mathematical development of the model is presented in Chapter 5. The model was used to simulate the removal of BTEX from the laboratory columns. To do this a characterization of the gasoline used was performed. The parameters necessary for the model were determined by laboratory analysis of air, water, and gasoline at equilibrium. The details of the procedures are described in Chapter 4. In the first model application the known experimental flow of water with all other fluids static was simulated as shown in Figure 3.5. The time scale used in Figure 3.5 is equivalent to those in Figures 3.2 & 3.3 so a direct comparison can be made. The model results inaccurately' predict ‘the ibehavior of ‘toluene and. xylene compounds in gasoline» The model does not.predict any rise in toluene concentrations and only minimal increases in xylene. Based on this simulation and earlier accumulative mass removal results, there is a strong indication that a mass removal process other than water occurring. The model was therefore run with air flow'present. .Assuming some unknown flow of air, the model was calibrated to best fit the data with 2.7 ml/hr giving the best fit as shown in Figure 3.6. Clearly the loss 67 .mmmooum Dawsonua .x a b msadaoo no sowunasawm ammo: OO¢ was .5: OON .m.m Shaman ‘ Il/ comm axis. / 3x10 com 3% 0* (l/Enuu) aioqaoe') u] uogionuaauog 68 .HA\HE h.m mo 30am was mafia x a b msfidaoo uo soflunasaww ammo: .w.m A963 9.5:. 00¢ CON 0 TI!) _ — Jr} _ .rnlizlxlrnmunn/z/Ir comm )//// .slllrmlxlxxxxx EXIo 1 JK I1 com _o._. wusmwm O— .ON On O¢ on Om Oh (l/fiw) eioqooej u! uogioliuaouoo 69 of volatile compounds from the gasoline is capable of simulating the selective transport of BTEX observed. While actual flow of air may not be the true cause of mass removal from the column, some process removing compounds favoring the gas phase is strongly indicated. A possible explanation is that some small leak was present in the column allowing vapor to diffuse to the atmosphere. To look at this further diffusion of gas through a cross sectional area is examined. Using the steady state diffusion equation, C-C 17: SDEEE==_D_J__3 dx L (3.3) where F'is the flux of contaminant which is proportional to the difference in concentration (c,-c2). Assuming that the flux of contaminant must correspond to 2.7 ml/hr of air and that the distance L is 1 cm, and D=0.1 chs, the cross 2 or a single hole 1.0 mm. sectional area required is 0.0075 cm This very simple calculation shows the fairly significant diffusion area that is required and suggests an inadequate seal is not likely the cause» Diffusion through.material used in the column such as Teflon is a potential source of the vapor transport. In the initial phase of the experiments a port at the top of the column was left open with a 20 gage (0.2 mm diameter) needle. This valve was subsequently closed and no effect was observed on the effluent sample data. This observation supports the idea that diffusion through the column apparatus may be the cause of the vapor loss. The results here indicated that the experiments simulated 70 leaching in the unsaturated zone with some vapor transport occurring. The vapor transport measurably influenced the long term leachate characteristics of the gasoline contaminated soil. This points out the need to better understand and quantify vapor transport in the unsaturated zone. Based on these findings an experiment was conducted attempting to eliminate all vapor losses. The column was run under saturated flow conditions and all seals were very carefully checked for vapor leaks by pressurizing the columns and submerging in water. Measuring concentration in leachate over time (Figure 3.7) and integrating the area under the benzene curve showed that 85% was accounted for in the leachate water. The behavior of the BTEX compounds as leaching continued was much different.with toluene showing no significant rise and fall as in the previous data. Ethylbenzene and xylene showed a very slight rising trend very similar to the model with no vapor transport (Figure 3.5). These results indicate that vapor flux may be controlled however, it is difficult to draw conclusions as to the exact cause of the vapor flux as the later experiments were run under different conditions than the earlier columns. The last column was run using 10 cm columns under saturated flow conditions and had 1/4 the initial mass of gasoline compared to columns J S K. The model was used to simulate the slight rise and curvature evident in toluene concentrations in Figure 3.7, by curve fitting to an unknown air flow. The results indicate that the best fit flow rate is ~0.9 ml/hr which is 71 .CESHOO 3oHu Umumusunm mo mcoflumuucuocoo usOSHumm .>.n madman VA “2 3on x 3qu a $3 0 .8 8m n. Amxoov oEfi ON we or NF or w m N o .xu__m_. __ __ 17. 0.- W'HI‘H . 0. 0 0V 0! Av Av . I a I 4 1.. .. as! m - - . 4 . . . 71 n. r... Z '2 _ O_. ON on O¢ On (I/Btu) alDHDDe'] u] uogionuaouoo 72 1/3 the magnitude of that determined for columns J & K. This result suggests the loss is relatively constant and may be proportional to the size of the columns used. The effect can be minimized by running experiments at a faster rate. 3.6 Conclusions The experimental results presented here show the capabilities of the experiment for monitoring concentration in leachate from unsaturated flow of water through soils. Several columns were leached with water for long periods of time to look at the effects of a multi-component NAPL present in the soil. The results showed a selective removal process as the composition of the gasoline changes as various constituents were removed at different rates. Calculating the mass removed by water indicated that only a fraction of the initial mass present in the columns had been removed by water. A numerical model was used to simulated the data; the model assumed chemical equilibrium between water-NAPL-air. Simulation of the leaching process indicated that some vapor transport was taking place during the experiments. The vapor transport required was on the same order as the flow of water applied and may have been due to poor seals or diffusion into or through the apparatus materials used. An attempt to eliminate the vapor transport in additional experiments showed improvement but not total elimination. In Chapter 5 a technique is used to control the system. by imposing’ a 73 controlled vapor flow through the columns. The results of this work demonstrate the ability of a very simple local equilibrium model to simulate leaching of a multi-component NAPL. The experiments also stress the importance of considering vapor transport in simulating leaching processes in the unsaturated zone. CHAPTER 4 A LABORATORY INVESTIGATION OF GROUND'ATER LEACHATE CHARACTERISTICS POLLOUING SOIL VENTING 4 . 1 Introduction When liquid hydrocarbons such as gasoline have been re leased into the subsurface, the primary goal of remediation efforts is to minimize or eliminate resulting groundwater Gasoline located in the unsaturated zone can individual contamination . contribute to groundwater contamination when components partition to infiltrating water. This mobile water can then enter the saturated groundwater flow system and Potentially a water supply well. Liquid hydrocarbons c301'1'tzributing to groundwater contamination can persist for Years. For this reason, efforts are undertaken to remediate cOntaminated soil in the vadose zone. One technique aimed at removing liquid hydrocarbons from the vadose zone is soil venting, also referred to as soil Vapor extraction. Soil venting is a process in which air flow is established through soils containing volatile contaminants, transporting those that partition to the mobile air (Figure 74 75 4.1) . These volatile contaminants are removed from the subsurface in the air flow and subsequently contained, destroyed, or released to the atmosphere. One of the appealing aspects of this technology is the in-situ application that is possible and the economic cost reductions that can result. Soil contaminated with gasoline presents an interesting problem. for remediation efforts. lNot only is gasoline immiscible with water, so that it flows as a separate phase, but it is also a complex mixture of a large number of individual compounds which complicates the removal by soil venting. As gasoline is vented from a soil, it's chemical character changes. This is because individual compounds are removed at different rates from the remaining gasoline mixture. In mixtures, compounds with high vapor pressures are removed more rapidly because they partition more readily to air than low vapor pressure compounds do. This selective removal process could potentially result in difficulty significantly reducing high vapor pressure compounds within a mixture which also contains compounds of low vapor pressure. Limited effectiveness of soil vapor extraction may result in continued contamination of groundwater if percolating water passes through the vented soils. For this reason, aqueous phase transport characteristics provide a good measure of the success or limitations of the remediation technique. Measurement of leachate from vented soils can show whether both high and low vapor pressure compounds can be removed to 76 Venllng well Ground surface Infiltraflng W019! I fl (1:: => (y it Alr Flow 3 Al=r Flow :9 <1: t <1: 0 — v 3;: Leachate Soluble plqu Figure 4.1. Soil vapor extraction of contaminated soils. 77 a degree that produces an acceptably remediated soil. The work presented here investigates the leachate characteristics of gasoline contaminated soils that have been air vented for varying durations. Transport of gasoline constituents in the aqueous phase is evaluated to determine if the interphase transport process is an equilibrium process, or if it is rate limited. The laboratory" techniques used are capable of maintaining saturated or unsaturated flow while determining leachate concentrations produced by a NAPL immobilized within the soil. The experiments provide a direct means to quantify the level of soil remediation that can be achieved through soil venting with regard to the potential for groundwater contamination. 422 Background and Theory A limited amount of research has been conducted to understand the capabilities and limitations of soil vapor extraction, none however have focused on the capability of this technique to reduce concentrations below regulatory limits in percolating water passed through vented soils. Research that is related to leachate characterization of vented soils is discussed here. Wootan and Voynich (1984) investigated forced venting to remove gasoline vapors from soil in a 3 m x 3 m x 1.2 m model aquifer in the laboratory. The laboratory tank experiments had a flowing water table with a floating gasoline lens and a 78 mobile air phase above. The experiments were monitored for three types of gasoline removal: NAPL gasoline removed by the flowing water, dissolved gasoline removed within the flowing water and vaporous gasoline vapor removed in the effluent air. While the focus of the study was on the effectiveness of different venting geometries, aqueous phase concentration measurements showed compositional changes towards heavy fractions of gasoline during venting. The mass balance showed that aqueous phase transport accounted for only a small percentage of the total mass removed. The experiments were discontinued before total removal of gasoline was achieved. This, along with the complexities of the system used, limit conclusions regarding the efficiency of the technique for reducing potential groundwater contamination. Vapor transport in soils has also been investigated by Johnson et al. (1987) . In examining the role of mass exchange between vapor and water phases, both one dimensional diffusion and one dimensional advection experiments were conducted. They concluded that mass exchange between pore water and pore vapor was sufficiently rapid that it could be correctly described as an equilibrium process for most scenarios. They point out that low-vapor pressure components found in gasoline may make it difficult to remove residual hydrocarbons from the subsurface in a short time frame. Volatile compounds could however be removed rapidly. While the previous two studies involved vapor transport in the presence of water-wet soils their focus was not on the resultant aqueous phase 79 concentrations as is the focus of this study. Work aimed at studying evaporation of hydrocarbon spills floating on water has a close relationship to this investigation. Burris and MacIntyre (1986) monitored aqueous phase concentrations in stirred tank systems. The tanks contained water with a floating hydrocarbon mixture and an overlying vapor phase which was continually purged with nitrogen; there was no porous medium. Constituent concentrations in the water and in the NAPL were monitored over the course of the experiment. The four constituent hydrocarbon.mixture used resulted in selective transport with the most volatile constituent removed first. The experiment was not carried out to observe substantial reductions in the constituents, as the goal was to observe the selective transport due to the varying solubilities and vapor pressures of the constituents present in the hydrocarbon mixture. The effect of the size of the water volume and its contact area with NAPL resulted in non-equilibrium between the two phases. This effect made it difficult to extend the results to other systems. A fairly large body of work has investigated aqueous phase transport from NAPL contaminated soils. Much of the focus of this work has been aimed at identifying whether the interphase transport process is adequately represented as an equilibrium process. Many studies have concluded that for natural flow conditions in porous media equilibrium transport is an appropriate representation (Miller et al. 1990, Zalidis 80 et al. 1991, Schwille 1988, Fried et al. 1979, and van der Waarden et al. 1971). In the present study leachate characteristics of vented soils were investigated under both saturated and unsaturated flow conditions. A simple mathematical model can be formulated in order to develop a better understanding of the potential for non-equilibrium transport under flow conditions used in the column experiments. Flow of water through porous media with a distributed NAPL, which provides the source of constituents that are leached out of the soil, can be idealized as flow of water through an annular region in a tube (Figure 4.2b). As shown in the idealized pore in Figure 4.2a the wetting order is assumed to be water-NAPL-air. The water is viewed as flowing between the soil grains and the NAPL. In the idealization of the porous media using a hollow cylinder, I} is the radius from the center of the pore to the water-NAPL interface, and r5 is the radius to the soil-water interface. This characterization of the partially saturated media, while highly idealized, can provide a means to calculate how quickly chemical equilibrium is established in the water by a constituent diffusing from a NAPL source. The time required to reach equilibrium can be determined for different water saturations and provides some insight into water saturation effects in porous media. The solution for the travel distance required for a fluid particle to reach equilibrium concentration with a NAPL source, can be simplified by neglecting diffusion and 81 Soil Grain a) Idealized Pore Soil Surface b) Conceptual Pore Model Figure 4.2. Idealized pore with three fluids and conceptual model of pore. 82 dispersion in the direction of water flow and moving the reference frame at the average fluid velocity. The problem is a one dimensional in radial coordinates. We begin with the one dimensional diffusion equation, 29.. 22.2 11‘: 4.1 at 61" + 181' ( ) where c is the concentration of a constituent in water, t is time, D is the diffusion coefficient for the compound in water, and r is radial distance. At the NAPL-water interface it is assumed that the concentration in the water is in chemical equilibrium with NAPL. At the soil-water interface it is assumed there is no flux of mass into the soil. The boundary conditions for this simple model are: t = 0, .z',<1:r'n:aining gasoline. Effluent concentrations are monitored u“tier different flow conditions in order to determine if chemical equilibrium is achieved over the length of the laboratory columns. Leachate from vented soils are used to 5S9Ied time T .99 84 50 40— 30 — 20 - 10— 0 1% n L a 1 1 1 1 0.4 0.5 0.8 1 Figure 4.3 Pore Water Saturation Sw Time to reach equilibrium concentrations in water as a function of pore water saturation for the conceptual model. 85 potential groundwater contamination at spill sites. 4.2 Experimental Procedure Laboratory experiments were performed using the apparatus and.procedures.described.by Zalidis et al. (1991). Figure 4.4 shows the column and collection system. The columns were constructed of glass, Teflon, and stainless steel components. Columns dimensions were 10 cm in length 5.4 cm in diameter. The soil columns, originally designed for unsaturated flow of water, were modified to allow air flow through the partially liquid saturated media. The modification involved a stainless steel ring and screen that was used to replace capillary barriers (pressure plates) at the lower end of the columns. The screen allowed air flow while supporting the soil mass. Air was supplied through the building compressor system and passed through activated carbon and drying material. The air flow was controlled using a Matheson variable air flow meter and regulator. Columns were first packed with air dried soil, saturated with water and brought to a low water content. This was established by holding air at atmospheric pressure and reducing the pressure in the water at the capillary barrier to -300 mbar. The column design allowed the water saturated capillary barrier to be removed while a vacuum was continually applied. After removal, the columns were weighed to determine the quantity of water remaining in the soil. Gasoline was 86 Inflow From Pump 1 / Venting P0" H/fiflu O - l/4 ” Glass Walls Threaded Rod O Preewre Plate 0 [4% fl Teflon 0ng 1 From Column Stainless Steel Valve \ l/B " Teflon Tube Connected To Regulated Vacuum 1 Liter Vawum Flask Vacuum Transferred Through Syringe Figure 4.4. Column and collection system. 87 then introduced into the soil and drained to low saturation using a gasoline saturated ceramic capillary barrier which had replaced the water saturated barrier. A -300 mbar pressure was applied to the gasoline. Once again the capillary barrier was removed and the columns weighted to determine the quantity of gasoline remaining. The fluid saturations achieved are intended to simulate conditions that may be found at a site where a release of NAPL has had time to drain vertically to a nearly immobile state. By applying a -300 mbar pressure to the fluids in the columns, they are representative of a point in the field above the capillary fringe where fluids are at or near residual saturations. With residual fluids in place, the column top, bottom, and o-rings were replaced with clean Teflon components and the stainless steel screen was inserted. Air flow was then initiated through the column. Each column was vented.for'a different length of time and air samples were collected prior to stopping the venting process. Columns were then fitted with stainless steel top and bottom components, saturated with water, sealed and allowed to equilibrate for a 24 hour period in preparation for aqueous flow experiments. Aqueous phase flow was initiated using a Harvard Apparatus syringe pump and effluent water samples were collected for analysis. The column experimental configuration was capable of maintaining flow under both saturated and unsaturated conditions using a capillary barrier in the lower end. With this capability, selected columns were leached under saturated flow then rapidly switched to 88 unsaturated to look at the effects of water saturation on the aqueous phase transport process. 4.3 Experimental conditions Experiments presented here were run using a glacial outwash sand, the properties of which are presented in Table 4.1. Initial fluid saturations established for the 25 columns were 0.11 t 0.01 water and 0.046 r 0.005 gasoline. Columns with replicate initial conditions were vented with air at an average flow'rate of 29.5 ml/min, 4.1 cm/min (average air'pore velocity) or 20 pore volumes per hour. Column venting durations ranged from 6 hours to 10 days. Following the venting phase of the experiment, columns were leached with water under saturated flow conditions at flow rates ranging from 1 to 100 ml/hour (0.011 to 1.1 pore volumes per hour). Water saturations achieved by capillary infiltration were 0.85. In columns subjected to unsaturated flow conditions, flow rates of 2 to 10 ml/hr were used, resulting in water saturations of approximately 0.35. Effluent water samples were collected and analyzed for benzene, toluene, ethyl- benzene, m-xylene, o-xylene (BTEX), and naphthalene using a head space gas chromatograph technique (Voice et al. 1990). 89 Table 4.1 Properties of Glacial Sand Mean grain diameter 0.3 mm Average porosity 0.37 Bulk density 1.66 g/cm3 Organic content 0.2% Hydraulic conductivity 0.0036 cm/min 4.4 Results 4.4.1 Experimental Leachate Data The results of the laboratory experiments provide insight and understanding regarding leachate contributions to groundwater contamination from soils containing gasoline that have been subjected to venting. The analysis of results focuses entirely an aqueous phase concentrations resulting from the leaching phase of the column experiments. In order to characterize leachate from vented soils and to be able to compare results between column experiments an evaluation of the aqueous phase transport process was necessary. It was also important to understand what affect the column length and the flow of water applied had on the concentrations of contaminants produced in effluent water. If the columns were homogeneous and leachate was at chemical equilibrium the geometry of the columns and flow rate of water applied would .have no influence. To view the columns as representative of a point in a field setting this aspect of the leaching experiments had to be investigated. It was also of interest 90 to determine if aqueous phase flow conditions, saturated or unsaturated, affect the leachate concentrations produced. If no significant differences were observed further research could dispense with the complications of working in partially saturated flow systems. To investigate these questions selected column experiments were subjected to changes in aqueous flow rates and water saturation levels while effluent concentrations were monitored. A flow interruption method similar to that used by Brusseau et al. (1989) was also used to investigate transport processes. Before investigating the experimental results for the presence of chemical equilibrium, the solution for the simple model, Equation 4.3, is applied to the experimental conditions. The maximum seepage velocity applied in the laboratory eXperiments under saturated flow conditions was 0.003 cm/s which corresponds to a vertical application rate of 4.3 cm/hour. Using the diffusion coefficient for toluene in water, 0.9x10°¢m¥/s (Miller et al. 1990), and assuming a pore size equal to the average grain size of 0.03 cm, at.a gasoline saturation of 0.05 the travel distance required to reach 99% of equilibrium is 0.7 cm. This estimate of travel distance is a conservative estimate becausejpore sizes in the media are likely to be less than the average grain size and the model does not account for tortuosity of flow paths. While the model is highly simplified it does provide some estimate of the transport conditions in porous media. This analysis indicates that if the only resistance to interphase transport 91 is molecular diffusion within the aqueous phase, then chemical equilibrium will be reached relatively quickly. Effluent concentrations of BTEX and naphthalene vs. flow of water scaled by soil pore volumes that are representative of vented soils in this study are presented in Figure 4.5. The seepage velocity applied is indicated for different periods of the experiment. Intervals where flow was stopped (33.2 and 35.0 p.v.) are indicated by breaks in connected data. In cases where flow was stopped the interval was 48 hours or longer in which the column was static and sealed. It was clear from the data that evidence of significant rate limited transport was not present; average concentrations before flow stoppage and following are within 10% for all compounds monitored. If non-equilibrium transport was taking place in the columns, effluent concentrations following a static period should be higher than those prior to flow stoppage and should decrease back to initial concentrations as flow continues. Also when flow rates of water were increased it would be expected that effluent water concentrations would correspondingly decrease as the column residence time for water decreases. In the early phase of leaching experiments concentrations rise as pore volumes increase, this was believed due, in part, to the presence of uncontaminated water in the lower column from the barrier and below. This water has not contacted soil as the columns were saturated from the bottom. As this water was flushed from the system the effluent concentrations approach constant values. 92 0.. T .musos e ouuca> mo sasHoo .muonooaa cw sawunuusaocoo .m.e assume nose—36> Boa 25E 35.33:: 26: 8.22am Ema cozazcoocoo omega msooso< 5.182 G _>x-o fl 3x5 0 seem O .3 D com I 93 Data shown in Figure 4.5 is for a soil that has been vented a relatively short period of time (6 hours), and, therefore, other than benzene, all compounds mbnitored were present in effluent water at relatively high concentrations. Figure 4.6 shows aqueous phase transport results for a column vented for a comparatively long period of time (48 hours). Here all compounds except. naphthalene ‘were at ‘very‘ low concentrations with benzene, toluene, and ethylbenzene at non- detectable levels. The concentration of naphthalene on the other hand has doubled from the levels in soil vented 6 hours (Figure 4.5). These data have been presented on a semi-log scale to accommodate the wide range of concentrations. The data indicate that with a NAPL containing constituents a very low mass fractions (ie. m&p-xylene and o-xylene, Figure 4.6), the transport process in the aqueous phase for these compounds continues to behave under equilibrium conditions. Again concentrations before and after flow stoppage are within 10%. While the data presented on a log scale de-emphasize fluctuations possibly caused by non-equilibrium transport, it should be realized that when characterizing leachate from vented.soils, concentrationS‘will.range.over’several orders of I magnitude during the course of venting. The interest here was to look for evidence of "significant" non-equilibrium conditions, that would influence the conclusions drawn regarding the level of remediation achieved. The data in this regard maintain constant leachate levels over a range of flow conditions. 94 .muson we pausa> mm Canaan .munsoemH ca sawunuusaocou N? or ewe“; *0 meta-O) Pan. 0 v —o— fir— - L I HEEo Qm m w. l Ssh—filings o. z _ _ luv—lull .EEo Qm _ Ilia—mag: 30E 85.5.3 V .232 Q 3x6 ‘ .>x-E Av 6.1. 02:62 €95 5:92.350 o9 omega. 38:3. Door 089 95 The influence of water saturation on leachate characteristics was also investigated. The effects of fluid saturation were estimated using the analytical solution for the idealized model (Equation 4.3). Saturations in columns were determined by weight, from approximately 0.85 to 0.35. Using Figure 4.3 the time to reach 99% equilibrium at the soil water interface (:2) can be determined. The ratio of times for a transition from saturated to unsaturated conditions using the above range of saturations is 1/60. This implies that equilibrium is achieved 60 times faster during unsaturated flow. The model clearly implies that if equilibrium conditions are operating under saturated flow, they also are very likely during unsaturated flow. In laboratory experiments columns initially leached under saturated flow conditions could be quickly brought to partially saturated conditions by imposing a partial vacuum - across the capillary barrier in the bottom of the column. Changes from saturated to unsaturated flow conditions are indicated in Figures 4.5 & 4.6. Leachate concentrations before and after saturation changes show good agreements being within 10% of one another. Minor fluctuations evident during changes from saturated to unsaturated flow may in part be contributed to any non-uniformities in the vertical distribution of contaminants within the columns. Excess fluid present for saturated flow is drained during the transition period in a matter of a few minutes. Concentrations in water exiting the column during this time may not be representative 96 of conditions at the lower end of the columns as the contact time with.this region is very shorts The experiments indicate that an evaluation of potential contributions to groundwater from gasoline contaminated soils can be made under saturated or unsaturated conditions producing comparable results. 4.4.2 Influence of Soil Venting Durations To investigate the full range of leachate concentrations produced from vented gasoline contaminated soils, effluent concentration data were collected from 25 columns, each having been vented for a different length of time. Aqueous phase transport experiments conducted following soil venting were analyzed to determine the concentrations that the vented soil was capable of producing. These values were used to characterize the leachate generated by soil that had been vented by the applied volume of air. These average aqueous phase concentrations are presented in Table 4.2. In addition, three compounds have been plotted as a function of the length of time the column was vented (Figure 4.7). A scaled venting time (T) has been employed to account for variations in air flow rates applied and the initial mass of gasoline, T=_t_o¢_fll (4,7) ”1 where t is time, 0‘I is the flow rate for air, W, is the initial mass of gasoline, and p, is the density of gasoline (0.72 g/cm3 for this study). The scaled time is simply a 97 .mafiom oousa> Scum msowunuucaosoo moose nsoodv< .~.¢ manna nx8. : 88.3 .a 68 as a... 58 8.~ 5.8 u 8992 t V V V V V 885 38 new new on v V V V V V .58 88 o. 5 m2 m0 V V V V V V 82: :8 new 88 o: m V V V V V 288 8.... «.8 8. x: 88 V V V V V 88¢ mom «.8 8. .2: m... V V V V V «88 38 m8 «3 m: 8:. V V V V V e: 6 88 .98 3; mm .8 3 3 V V V e88 New «.8 m3 >0 ma m 3 m V V 88 New e. 8 8. mo 88 V 3 V V V 803. 8.~ Q8 ....8 E 88 m m V V V 83. :.~ 28 as v5 88 m 3 V 3 m 88» 88 m8 8 x: 88 ms 3 V 3 m 88 ca 38 8 >: 88 3 F V P V 8am Ba 08 8 2,: 688 V V V V V 88m 88 m8 m8 :1 88 a... as E m... V 888 «8 QB 38 <1 88 em 3 V 3 «a :58 Ba «.8 8 E 88 t m V V V .83 55.. e8. 8 98 .m:0aumusc unmuwuufiu cmunm> mcazaoo Mom mcodunuucmocoo manna mzooavd .>.v wusmfim .. 2:: 38> 8.8m ooooow oooom oooom 089m 9500 oooom 88v oooom oooom 008.. o _ q 4 fl fig ‘ _ a ‘1‘ I _ _ P o D D I I o o D D a D I or 6. HM mg I 8.. 0 89¢ 5:928:00 O I # manna. 2533 OH}! 82 . z w 0 O o o o: D “ D WD 88? I. 880.. 5:82 9 3x6 B .8 l 99 ratio of the volume of air passed through the soil to the volume of gasoline initially present. The scaling procedure is valid if equilibrium transport is taking place within the air phase. This has been shown to be true for gasoline contaminated soil.with.air flow rates more than 20 times those employed in this study (Baehr et a1. 1989). These aqueous phase concentration measurements capture the complexities of working with a mixture of hydrocarbons such as gasoline. The presence of many compounds having different vapor pressures resulted in selective removal of individual components. Understandably the higher vapor pressure compounds were removed early in the venting process. Not as clear is why lower vapor pressure compounds had "increased" their concentrations in leachate as venting proceeded. The explanation lies in the fact that the liquid hydrocarbon remaining following a period of venting is enriched in lower vapor pressure compounds such as naphthalene. With a higher fraction of the remaining NAPL composed of naphthalene, the resulting aqueous phase naphthalene concentration was elevated. The results in Figure 4.7 have been plotted on a semi-log scale to observe the full range of concentrations measured and in addition the ultimate levels to which venting was capable of reducing leachate concentrations. While the data are not continuous and smooth it is clear that the more volatile compounds such as toluene were removed first as expected. .All six compounds monitored were eventually reduced below 1 ug/l 100 in effluent water samples. For the single column with leachate levels less than ling/l for all six compounds, it was noted that the chromatograms had no detectable peaks for retention times earlier than naphthalene. 4.4.3 Nodal simulation A simple vapor transport model was employed to better understand how processes influencing the removal of gasoline during soil vapor extraction, influence leachate concentrations. The model assumed equilibrium partitioning between NAPL, water, and air for individual compounds. The model also assumed uniform conditions throughout the column and simply kept track of the mass of individual gasoline components present in the column and removed mass from the system by flowing air. The model used was based on a version by Marley and.Hoag (1984) and.was similar to that developed.by Johnson et a1. ( 1990) . A mathematical development of the model is presented in Chapter 5. By assuming chemical equilibrium the model simply partitioned individual constituents to air or water based on the mole fraction remaining in the gasoline. A basic finite difference scheme was used to solve the temporal removal of gasoline constituents from the system- A. mathematical development is not presented here but can be found in Marley and Hoag (1984) or Baehr et al. (1990). To simplify the problem further it was assumed that the mass present in the 101 immiscible phase (NAPL) dominates that present in all other phases. For example compounds present in the aqueous phase or the sorbed phase have no impact on the‘ quantity that partitions to the air. This assumption was most likely true for high concentrations of components in the gasoline but may fail as quantities were reduced to very low levels. The assumption was also made that biological losses were not significant. To simulate gasoline removal by soil vapor extraction it was necessary to determine partitioning coefficients for all compounds present in gasoline as well as the composition of the fresh gasoline. Partitioning coefficients between.water- NAPL as well as air-NAPL were determined from laboratory measurements conducted on water, air, and gasoline which had been in contact for sufficient time to reach chemical equilibrium. Due to difficulties in identifying the great number of compounds present in gasoline, concentrations of all constituents present in gas chromatographic analysis of gasoline were determined by correlations using reference peaks (see Hayden 1991 for details and additional compound identification). Molecular weights of each compound, necessary to determine the mole fractions of gasoline, were estimated based on the nearest known value and comparisons with other information available for gasoline (Marley and I-Ioag 1984, Johnson et a1. 1990). Using estimated mole fractions for 104 compounds in,‘gasoline, as described. above, and measured concentrations in water and air, partitioning 102 coefficients between air and NAPL and coefficients between water and air could be determined. Air-water partitioning coefficients were compared with theoretical values and were found to be high for all BTEX compounds. Because of greater uncertainty in gas phase concentrations measured it was decided to adjust the air-NAPL partitioning coefficients so that air-water coefficients matched theoretical values for BTEX. All other unknown compounds were adjusted based on a reference peak. The values used for the numerical model are presented in Appendix II. The model was run for the gasoline used in this study and results were converted to aqueous phase concentrations that would result from leaching the column at any point in the venting process. These data can be directly compared to laboratory data using the scaled time axis previously employed (Figure 4.8). The results exhibit the selective removal process discussed earlier. Naphthalene was removed over a much longer time frame than the other compounds monitored. The overall order of removal and the time of removal agree between model and data. The slow rise and continuation at high levels (5000-9000 ug/l) for naphthalene was very similar between model and data. The naphthalene data after long venting times becomes somewhat scattered but eventually decreases as in the model. The model results indicated that as compounds were reduced to low levels (i.e. less than 10 ug/l) they continued to be rapidly removed towards zero. This trend was not 103 .mafiom omucm> ca mcowumupcmocoo muonm msomnvm mo covaHSEHm Hobo: .w.¢ musOAm 2:; 8:55 8.8m o88c 082: 88m 88 884 808 o 4L «4»— . i l. or cow 89: 5:938:00 ommcd £5ng coop oooow L F ooooo _. 5:32 _>x-o zoom . ..... _o._. -.- com I 104 readily supported by the aqueous phase experimental data. While reductions down to approximately 10 pg/l were similar to model predictions, some compounds continue to leach from soils that were vented approximately two times longer than the time after which the model predicts their disappearance. It should be pointed out that this was after reductions of over three orders of 'magnitude Zhave occurred. and further ‘that the compounds were ultimately removed. 4.5 Summary and Conclusions The leachate characteristics of vented soils containing gasoline have been investigated. Aqueous phase concentrations were used as a measure of the level of remediation achieved using soil vapor extraction. Individual columns with similar initial conditions were vented for various durations and then leached with water to characterize the contaminated soils potential impact on groundwater. In evaluating aqueous phase transport that occurred in the columns following soil venting it was found that transport processes were approximately at equilibrium for both saturated and unsaturated flow conditions. Constituents ranging from benzene to naphthalene all exhibited equilibrium aqueous phase transport over a very wide rage of concentrations in water. Analysis of 25 columns, each vented a different length of time, from 150-4400 pore volumes of air, indicated that leachate levels in vented soils can be reduced below 1 ug/l 105 for all compounds monitored in this study (BTEX and naphthalene) Comparing the results with a simple equilibrium based model indicated that the overall removal of constituents from the gasoline contaminated soil was similar between model and experiment. There was some evidence to suggest that individual constituents will leach from vented soils at very low levels < 10 ug/l for a measurable period beyond that predicted by the simple model. In general it was found that soil vapor extraction was efficient for reducing the potential impact to groundwater of gasoline contaminated soils. It was found that the very simple model can.predict the bulk of gasoline removal. It was clear from our study that using the model to predict clean-up of a gasoline spill and calculating the required venting time on a low vapor pressure compound such as naphthalene would provide a good conservative estimate. CHAPTER 5 AN EXPERINENTAL AND NUMERICAL NOBEL INVESTIGATION OF GROUND WATER LEACEATE CHARACTERISTICS DURING SOIL VENTING 5.1. Introduction In Chapter 4 a study was undertaken to investigate groundwater leachate characteristics of soils following the application of soil vapor extraction. The study made use of experimental techniques developed in Chapter 2. The aqueous phase concentrations measured following soil venting provided a direct measure of the capability of the technique to remediate gasoline contaminated soil with regard to groundwater contamination potential. The results were compared with a chemical equilibrium based model capable of simulating the soil venting process. In this chapter the adequacy of this simple model is investigated using a more sophisticated experimental method. With the ability to maintain partially saturated flow of water through soil, the air filling the remaining' pore space can. be mobilized, resulting in vapor transport of gasoline constituents during the leaching process. This technique, referred to in this 106 107 chapter as a dual flow experiment, has the advantage that a single column can provide leachate characteristics of a NAPL contaminated soil through the entire soil venting process. This continuous data is more suitable for evaluating the acceptability of assuming chemical equilibrium because the method can detect changes in transport behavior as constituents are reduced to non-detectable levels and information on trends in data can be evaluated. A study similar in nature, but aimed at studying evaporation of hydrocarbon spills on water, is closely related to this investigation. Burris and.MacIntyre (1986) monitored aqueous phase concentrations in stirred tank systems. The tanks contained water with a floating hydrocarbon mixture and an overlying vapor phase continually purged with nitrogen. The goal of the study was to observe the selective transport due to the varying solubilities and vapor pressures of the compounds present in the hydrocarbon mixture. The effect of the size of the water tank and contact area with NAPL:resulted in non-equilibrium conditions. This effect limited the ability to scale the results but the experiments provided a better understanding of the complex interactions of a NAPL- vapor-water system. Vapor extraction of hydrocarbons from soils can be viewed as a very similar process. In porous media the fluid flow is much more confined and has large surface area to fluid volume ratios, causing the system to reach chemical equilibrium much faster. This fact can greatly simplify mathematical simulation of the process. 108 . In the present study a model is presented with the capability of simulating transport in soil columns containing residual gasoline that are subjected to simultaneous flow of air and water. The transport is simulated using chemical equilibrium partitioning between phases. Different transport behavior under conditions of air venting and water flushing are investigated using the numerical model. Laboratory experiments were conducted supporting flow of both air and water through soils containing NAPL at residual saturations. During the flow experiments aqueous phase samples were collected and analyzed for the following six constituents of gasoline: benzene, toluene, ethylbenzene m&p-xylene, o-xylene (BTEX) and naphthalene. These results were compared to numerical model predictions. The goal was to observe the selective transport caused by soil venting and compare the results to those predicted by a simple numerical model. In addition, observations were made on the complete removal of constituents in the effluent water. Previous experimental studies investigating soil venting have focused on total removal of gasoline from soils and have not looked at the ability of the technique to remove individual compounds to extremely low levels, nor at the adequacy of an equilibrium model for simulating this constituent transport. 109 5 . 2 . Numerical simulation 5 . 2 . 1 . Model development To develop a better understanding of the remediation process during air venting or water flushing of gasoline from soil, a simple model is employed. The model keeps track of all constituents present in a NAPL and can simulate removal of mass by either flow of water or air. The mass remaining at any time during the process can be used to predict the concentration of any constituent of gasoline in air, water, or NAPL. The model assumes equilibrium partitioning of gasoline constituents between NAPL, water and air. The model also assumes that the distribution of mass for each phase and for each constituent within a phase are uniform throughout the column. The model simply keeps track of the mass of each gasoline constituent present in the column and removes mass from the system within the flowing fluids, ie. air and/or water. The model used is based on a version by Marley and Hoag (1984) and that developed by Johnson et al. (1990). The model development and symbols used here follow those by Baehr et al. (1990). Making use of Rauolt's Law, the model partitions between air and NAPL according to, Gk=H£vXk (5-1) 110 where the mole fraction X; is XI: = '1; = — (5-2) and, Ch = Concentration of compound k in air [M/Dfi id; = Vapor concentration over pure constituent [M/L’J X} = Mole fraction in the immiscible phase [1] an = Total moles of k in NAPL [1] .m, a Total moles of NAPL [1] I} = Concentration of k in the NAPL [M/Dfi wk = Molecular weight of k [M] .NC = Number of compounds in the NAPL The partitioning of constituents between air and water incorporate Henry's Law in the following relationship, G C. = 1: (5.3) H.’ . where, C} = Concentration of k in water [M/Dfi H}, = Air-Water partition coef. [M/L3 / M/L’] Finally any mass sorbed to the soil organic mater partitions according to S, = H}. c, (5-4) where, 5; = Concentration of k sorbed [M/M] In: = Soil-water partitioning coefficient [M/M / M/L’J Having defined equilibrium relationships between all phases present. within. the soil, the total mass of any 111 constituent k can be written as Mk =Mh +Mh+Mm+Mh . (5.25) here: M,“I = Mass of k in air [M] Mass of k in water [M] it u 5&w = Mass of k in NAPL [M] Mk, = Mass of k on soil [M] then, M, = GkV. + Cka + IkVN + skws (5.6) where V is the volume of the identified phase (a-air, w-water, and N-NAPL) and W3 is the mass of soil. Substituting Equations 5.1-5.4 into Equation 5.6: Ht NC I Ht k Mk=Xk 5v. + ‘fvwer wkEJVN+ ...-me. (5-7) H" k-l 0:: H" The removal of mass from a NAPL contaminated soil can be obtained by solving (W E5 = - 0““er — 04-1er - Bk (5'8) where, Q = Flow of indicated phase [L3/T] Bk = Biological losses [M/T] Writing Equation 5.8 in finite difference form: Mk(t+At) =Mk(t)-At(ouer)-At(0ncuck) - At(Bk) (5.9) Equation 5.9 can be solved explicitly for MR values over time. This quantity along with 14,. can then be used to calculate concentrations in air or water at any time during the soil venting or water flushing process. 112 5.2.2. scaled model simulations The model was run simulating soil air venting and water flushing. The model was simplified by assuming the total mass in the NAPL is much larger than that in all other phases, Mm>Mn +Mn+Mu (5.10) or that the third term in brackets in Equation 5.7 is much larger that the other three. This assumption is easily met when the initial mass of gasoline is introduced into the column. The assumption may fail late in the venting process when the total remaining mass of NAPL is very small. The assumption is also :made that. biological losses are not significant. For all experiments run in this study several factors limit the effectiveness of biological losses. The initial concentrations of BTEX in water equilibrated with fresh gasoline are very high and approach toxicity levels for biological activity. The Columns also have no supply of nutrients such as nitrogen or phosphorous required for biological activity. The model was run using transport parameters for the gasoline employed in the laboratory experiments. The gasoline was analyzed for mass fractions and partition coefficients between air and water (H&,H:§,) for six compounds of interest in this study (BTEX and naphthalene). Partitioning coefficients, mass fractions and. molecular weights were estimated for the other 99 compounds that were evidenced by 113 peaks on the gas chromatograms (see Appendix II). The process used to determine these values has been described in more detail in Chapter 4. As stated in Chapter 4, time can be scaled in soil venting simulations using a combination of time, mass removal rate, and initial mass of contaminant. The simulations here are scaled according to the ratio of the volume of air passed through the column to the volume of initial NAPL present, T = M (5,11) ”1 where Qf is the fluid flow rate, p” is the density of the NAPL, and W1 is the initial mass of NAPL. It must be noted that with two fluids transporting constituents from the system simultaneously that time is most meaningfully scaled when the transport rate of one fluid is much larger than the other. In this case the low mass transport fluid can be neglected and 0, represents the flow of fluid causing transport. When this is true systems with different conditions can be easily compared using '1‘. For most induced vapor flow in laboratory columns or field settings the transport in air dominates that in the aqueous phase. Looking at mass removal by air, 0.1.1sz: and water, outerck, the fluid dominating transport can be determined by calculating the ratio Qaier/Qwaterck and comparing to unity. The air water partitioning coefficient HJL, from Equation 5.3, which is simply a ratio of the concentration [M/L’] in the air to that in water, can be incorporated 114 resulting in, M > H"; (5.12) 0&1: and if true then flow of water is dominant and should be used for scaling. Strictly, for a multi-component mixture, every compound would have to satisfy the above inequality. An approximation where a majority of compounds agree with the inequality may be adequate to justify scaling the data. Flow systems with transport in each phase of similar magnitudes can also be scaled but can only be compared to systems with the same fluid flow ratios. Removal of gasoline was simulated under conditions where air was the only mobile fluid (Figure 5.1 a & b), as well as conditions where water transport was taking place with no air flow (Figure 5.2 a 8 b). The results highlight the differences in miscible transport of a multi-component liquid when the transporting fluid is water vs. air; Under air flow, constituents of gasoline are removed according to the vapor pressures, whereas under water flow constituents are removed according to their solubility in water. Transport under air flow results in increasing concentrations of the heavier compounds as the very volatile compounds are removed first (Figure 5.1a). The most volatile compound shown, benzene, also increases concentration initially because there are approximately 20 compounds more volatile (that can be identified by laboratory analysis). Results look very different under conditions in which.water transport dominates (Figure 5.2 a). The compounds of interest in this study 115 oooom 803 b _ 800.. — _ .maucm> ham «0 downwasawm H0005 cwawom 80$. _ 7 as: 555.; 8.8.0. 5:82 .MH.m Ouamwm oooow oooow oooom oooov oooom oooow oooon 080$ 83 5.9828 39.5 8.0264. 116 .mamom moHuflamm a co ucwuco> “Hm Ho cowumassfim H0605 cmamom .na.m ouaaflm 2:: 38> 8.8m o88. 858. 88w F cocoa / o 1 1 - F ooooo q)— “r- 89; cozaccoocoo 33.... «58:2 r. oooow - 882 £82 - 3x6 .- - comm ..... _o._. !- com I .mcwnmsam 00003 HO cofiuwasawm H0005 umawom .w~.m 0090:: 117 25 382“. 88m 88F 88: 083 88. 88: 08 88 08.. 88 o + 1 1. 1 ._ I m l_ : i m o % 082 m 0 p -1 808 $03 5:858:00 $3.... 058:3. .... cocoa I 0000.: 118 .mamom modifiamm a no vcHnmaau umuns uo nodumazafim kuoa omamom 2:: 9.52“. 38m 88m com: ooomv comm P 083 comm Tl ¢ ,, + 4 ¢ L_ * 5..an i _>x-o conm .. .0... I com I .a~.m muavfim or 00., :33 5:328:00 89.n— mzoozg ooow 800.. 8000 _. 119 (BTEX) are among the most soluble compounds in gasoline which contributes to their concern for groundwater contamination. In the water flushing simulation highly soluble compounds are removed first, therefore BTEX compound concentrations are reduced from the start and are continually reduced to low levels. Reductions of constituent concentrations to very low levels less then 1 ug/l are shown on a semi-log scale in Figure 5.1 b & 5.2 b. Air venting (Figure 5.1b) tends to cause compounds with low vapor pressures such as naphthalene to be removed at a much slower rate than high vapor pressure compounds such as xylene. Simulation of transport under conditions with both flow of air and water produces results between the above solutions. The model simulation was used to determine if the air flow and water flow rates used in later experiments could be scaled by the air alone. With a ratio of water flow to air flow Qatar/Q“, equal to 0.008 the model simulation indicated that the flow of water has very little impact on BTEX removal. The ratio is greater than H}: partitioning coefficients for nearly all compounds measured in this study (Appendix II). Low vapor pressure compounds such as naphthalene, which had an Hf, value that was only 1.5 times greater than the fluid flow ratio, were more affected by water flow. Model simulations with water flow included showed naphthalene was reduced to lug/1 approximately 20% earlier than simulations without water flow. The modeling simulations show that the process of 120 removing a multi-component liquid hydrocarbon from soil using air or water is complex. The concentrations in water are varying with time due to the changing composition of the NAPL. One consistency in the simulations is that individual compounds are removed to very low concentrations (<1ug/l) rapidly after levels fall below 100 ug/l. This aspect of the transport process can be explained by the fact that as compounds reach very 10w fractions of the remaining gasoline the effect of other compounds changing the NAPL mixture is minimal on the low concentration compound and the compound removal is mainly a function of the mass of that compound. This phase of the transport process will be discussed further with the laboratory results. 5.3. Laboratory Experiments 5.3.1. Apparatus Glass columns used in this study have been described in Chapter 2. Minor modifications were required to permit dual flow of air and water. During dual flow operation, the columns used (Figure 5.3) had two fluid entry ports at the top. A central port was used for inflow of water while an second off-set port supplied air flow. The bottom of the columns used in this study contained a capillary barrier (pressure plate) used to maintain negative fluid pressures in the water thus achieving low water saturations with unit 121 olr SUPPIY ‘ water supply Flow V Meter -LL Syrlnoe Pump Soll Column fl olr exit wafer collecflon Regulated Vacuum Figure 5.3 Apparatus for dual flow experiments. 122 gradient flow conditions. A constant flow of water was supplied to the top of the columns using a Harvard Apparatus 2000 syringe pump. Constant flow of air was supplied to the top of the column using compressed air from the building system. The air was passed through activated carbon, drying material and a variable area flow meter so that flow rates could be monitored and controlled. Air was allowed to exit the bottom of the columns through a gap above the top edge of the stainless steel plate where the Teflon o-ring, in place when packing and bringing the soil to residual fluid saturations, had been replaced by glass beads. 5.3.2. Procedure The procedure used to bring the soil columns to initial residual fluid saturations was the same as in all previous experiments. The modification to the procedure was introduced when the columns reached residual saturations of gasoline. At this point the gasoline saturated capillary barriers were removed and a volume of soil approximately 0.5 cm thick was removed from the bottom of the column and weighed. This soil was replaced by uncontaminated soil prior to emplacement of the stainless steel capillary barrier used to maintain unsaturated flow. The purpose of this operation was to remove any gasoline at the center of the column near the plate. Gasoline located here had the potential to be isolated from the flow of air due to the fact that air must exit the columns 123 at the bottom outer edge. At this point in the procedure the column was sealed by all three o-rings and flow of water was initiated. After a short period of time steady unsaturated flow was achieved and effluent water samples were collected. After a few samples had. been collected. tot characterize the initial effluent concentrations the lower pressure plate was removed and the o- ring above the plate was replaced by glass beads having a _diameter equal to the o-ring thickness. With the column reassembled, air flow was initiated through the top of the columns and intensive effluent water sampling was initiated. Water samples were collected in ‘the vacuum. chamber and analyzed using a gas chromatograph head space procedure as described in Chapter 2. 5.3.3. Experimental Conditions All dual flow experiments performed used a glacial outwash sand (Table 4.1) with unleaded gasoline as the NAPL source. Initial fluid saturations were 0.11 water and 0.066 gasoline. The total mass of gasoline was 4.2 grams. The flow of water was initially 20 ml/hour resulting in a water saturation of approximately 0.40. Water flow was later reduced to 10 m1/hour due to permit less frequent real time sampling. 5 ml water samples were collected in 20 ml head space vials with salt added to enhance volatilization. These were analyzed for benzene, toluene, ethylbenzene, m-xylene, o- 124 xylene, and naphthalene using a gas chromatogram with an auto- headspace sampler. Flow of air was maintained at approximately 38 ml/hr for the entire experiment. 5.4. Experimental Results Effluent aqueous phase concentrations from dual flow experiments were collected over periods of up to 12 days representing scaled time up to 110,000. Data representative of these experiments are presented in Figures 5.4a and 5.4b. The column conditions for this experiment and others can be found in Table 5.1. The aqueous phase concentrations collected during soil venting are plotted vs. scaled venting time on linear scales and show' the removal of various constituents from the columns (Figure 5.4a) . The removal order for the six constituents monitored was directly related to their vapor pressures. Benzene concentration in water was reduced to low levels prior to xylene compounds which in turn were reduced much earlier than naphthalene. Like the model simulation (Figure 5.1a) each compound monitored initially increased in aqueous phase concentrations as the composition of the gasoline changed. This was due to removal of highly volatile compounds (higher vapor pressures than benzene) causing the gasoline composition to become enriched in the remaining compounds, their respective mole fractions in turn increased. This increase was reflected by increases in effluent aqueous phase concentrations. 125 COOON .muasmmu Housmawummxm 30am Hana Doom? 00va as: use; 38m 88? come 5:82 4 3x6 Av :25 O .8 D :00. .M¢.m 0H50wm o comm- F o coco? oooom oooom ooocv oooom ooooo oooon oooom A33 5:828:00 ommcd maooao< 126 .n¢.m ounce: €03 5:558:00 002d m:00:0< .mamum modnwaom 0 c0 muasmmu Housmaauwmxm 30am dado 08F o:::o> 00.3w oooow F 8009 oooow 880 880 o T I 1 l w 0 5.132 4 sxé 0 Sam 9 .8 D 8m I 127 Table 5.1 Initial saturations and air flow applied. Column I.D. Water Gasoline Air flow (ml/min) DA 0.11 0.051 ‘ 29 DB 0.11 0.056 30 DM 0.099 0.067 41 DP 0.011 0.068 38 VV 0.011 0.050 38 Comparisons between the model simulation (Figure 5.1a) and the experimental data (Figure 5.4a) show very similar behavior in the concentrations present in effluent water throughout the course of the soil venting process. It must be pointed out that for the experimental data presented, T20 is set at the time at which air flow was initiated (Figure 5.4). Prior to the initiation of air flow only water flow was occurring. The data during this period show concentrations slowly rising and an abrupt rise in concentration is observed when the venting process in initiated. After venting was initiated, a comparison of the different constituent concentrations between the model prediction and the data is quite good. Focusing on toluene one can see the very steep rise and peak at ~72,000 ug/l and rapid decent is also predicted by the model. The general behavior and magnitudes of BTEX in water samples and 128 model predictions are in agreement through the venting process. This is an indication of the adequacy of the equilibrium model. The mass removed in the experiments can be compared to the initial mass in the columns by integrating the area under the curves and calculating the total mass of a constituent removed in the fluid flow: Massflremoved) = [(QnCeer+Qaier) dt (5' 13) O In the experiments the total mass removed by air and water can be estimated using the air-water partitioning coefficient Hi, and the flow of air applied: k Massk(removed) = fck [oyster + air HOV] dc (5'14) 0 The scaled data can be converted to mass fraction removed for an individual constituent with the following relationship, "1 [deT (5.15) Cfdt:= I; k Qairpg 0 then assuming flow of water is insignificant, Initial Mass fractionk = p f Ck dT (5-15) 9 0 Equation 5.16 calculates an initial mass fraction by integrating the mass removed in the air and scaling this by the initial mass of gasoline present in the columns. These calculated mass fractions can be compared between different column experiments as well as to the actual mass fraction of 129 the initial gasoline introduced to the columns. Calculated mass fractions for the six compounds monitored were determined using Equation 5.16 for 5 column experiments and are compared with the initial mass fraction for the gasoline that was measured by gas chromatogram analysis (Table 5.2 & Figure 5.5). The integration technique used to calculate the area under curves was a simple trapezoid formulation. The result show that the average of the 5 experiments has an error of less than 20%. Benzene integrations are less accurate due to the rapid decrease and the limited number of data points collected, 'The actual.mass fraction.and.those calculated from the experiments show reasonably good agreement for the six compounds monitored. These results along with the good match between the model and data in Figures 5.1a and 5.4a is strong indication that equilibrium transport can adequately represent the process for the laboratory column experiments. While the data indicate that constituent removal at relatively high concentrations by soil vapor extraction can be described by using the local equilibrium assumption, the transport process at very low concentration must also be investigated. Figure 5.4b presents the same set of data as Figure 5.4a using a semi-log scale to look at how low the venting process was capable of reducing effluent aqueous phase concentrations. The scaled time axis has been extended to include the removal of naphthalene. Aqueous phase concentrations changed over five orders of magnitude during the course of soil venting. All six constituents monitored 130 Clbllllted hm Mess fractious W Column DM H)? r DA DE W Meme 1 $ error Mm frat. Benzene 0.013 I 0.010 0.” 0.” 0.015 0.011 24.4 0.017 Toluene 0.50 0.051 0N7 0N6 0.074 0.” 12.2 0W 3W 0.015 0.013 0.010 0.017 0.010 0.016 11.9 0.010 l-xyklle 0.051 0.0‘7 0W 0N1 0N6 0.050 12.8 0W o-Xylcse 0025 0023 0.033 0.031 0.026 0.027 13.5 0.025 puma 0005 0004 000: 9.0 0004 Table 5.2. Calculated initial mass fractions from dual flow mom: :°-'F*Omw1'| experiments. Columns I D” 5 DP I DA I as a w Measured 0'08 T Gasoline 0.07 -. 0.06 1 ;__- E 0.05 -» g E 0.04 ._ g g *- s s 0.02 .- ié: f g r; '- ’1' E E E g 0.01 — , g (L g I? I; g E 0'4 llf E; ~ 3% fit 2; 3% E: IINthi Benzene Toluene Ethylben. m-Xylene o-Xylene Naphthal. Figure 5.5. Mass fractions from data integration. 131 were reduced below the lower quantification limit of l ug/l (1 Ppb) - The continuous data resulting from the dual flow experiments ‘was ‘useful for’ evaluating’ the 'ultimate effectiveness of the soil venting process with regard to leachate levels. While past investigates (Marley and Hoag 1984) have shown that soil venting was capable:of removing 99% of gasoline present in soil, the data presented here show that individual constituents in gasoline that represent a threat to groundwater can be reduced to extremely low levels. Comparisons of the model results to the data on semi-log scales can give additional insight into the mechanisms controlling transport.of'mass out.of the columnsw Figure 5.1b shows the model results on.a semi-log plot and can.be compared with the experimental data in Figure 5.4b. Comparisons indicate that both the model and the experimental data show reductions of leachate concentrations to less than 1 ug/l in the same constituent order and at approximate scaled times. Also evident is that the model predicts a smooth and rapid decrease in aqueous phase concentrations as the final mass of a given constituent is removed. The data on the other hand have a break in slope, possibly indicating a change in transport process when small quantities of a constituent remain in the NAPL. This feature of the data is discussed further in Section 5.4. A note must be made regarding the experimental apparatus used, as it affects the results. Early experiment using the 132 dual flow technique used caps made of Teflon on the bottom effluent-end of the columns. The results showed significant tailing of aqueous phase concentrations at low levels ie. < 100 ug/l (Figure 5.6) . Later experiments were run using stainless steel components on the lower end of the column which reduced the tailing. The possibility remained that the single Teflon o-ring, used in the later experiments, could still be affecting the aqueous phase concentrations. To look at this potential problem, new Teflon o-rings were used in experiments and periodically the o-rings were replaced and the data analyzed for possible deviations. For the data in Figure 5.4b, 0-rings were changed at scaled times 19000, 30000, and 105000. If o-rings were in fact acting as a source of mass entering the water, when that source was removed the aqueous phase concentration should drop significantly. This effect was not observed as a result of changing o-rings in the experiments. 5.5. Discussion The following discussion focuses on deviations between the data and the local equilibrium model examining the mathematical relationships used in the model formulation. The data and model result have been superimposed for benzene and toluene in Figure 5.7. The point of deviation, as compounds are reduced to low concentrations, between the model and data becomes clear. The results indicate that aqueous phase 133 .moco coamoa moan: muasmou Houcoawuomxm .o.m endow: s as: use; 8.80 cocoon ooooom ooooom 808v ooooom OOOOON 0800? o s i l l l l i F O O 0 o 0 0 0 _I I. _l -r__- 0.9 06. .9 OJOO lime” LFIMIio 2 O O O 00 l “I DDDDmmm auowAvoo 5000 0am 00% minim DMD D flu A 8. 4 $03 5:228:00 0005030033. 83 800.. Scoop £532 4 .500 0 5am 0 _o» D 5m I 134 .ocosaou 0:0 0:00:00 .0000 Houcoawummxm 0:0 H0003 mo comwu0eaoo 2:: use; 8.80 808 880 808 80.3 808 88. 809 88 o AW 1 1 1 1 1 1 . , O O O O O O ’ ,, 0 w 00/ A/\ , O ,1 O 11 O > 1, O {0 , / 1 / 1, \, /\ // . 4T 0 , / t \, I C / O /, LT O _ my .8. 0 com 0 35822 3 com .252 i .s.m 0.33: or 00.. $03 :0:0::00:00 0005. 030033 coop oooow 135 concentrations measured were remaining high over a longer time frame than the model predicts. The deviation observed between the experimental data and the numerical model can be better discussed by looking'at the basic transport equation (Equation 5.8). Assuming flow of water to be insignificant, conservation of mass can then be written: dM} dt The rate of mass removal from the system once the = -0“er (5. 17) concentration of a constituent k in water or air becomes low can be approximated by recognizing that the mole fraction, Xk=mk/m,., is a very small value in which case 111,, the total moles of NAPL may be considered a constant. Assuming an is in fact constant, all the terms in brackets in Equation 5.7 are constants and the total mass of k, A5 can be written, M,r = 311‘: (5.18) where 81 is a constant. Then making a substitution using Equation 5.1, k 5.191: = JEN—0516* (5.19) dt lg Since Ck=Gk/Hf, we can write, dC 11.5.0 1 k - -—‘_I.Ck (5020) dc B1 so that the relationship is written in terms of the concentration in water, Ct. Equation 5.20 can be solved for ckm, «fies.» (5.21) or, 136 lan = 10111 - fiaingt (5-22) where A1 is a constant to be eval1uated using an initial condition. Equation 5.22 indicates that removal at low concentrations will approach a straight line with a slope of -H.'§, 9.11/31 on a semi-log plot. As the mole fraction approaches zero for a constituent k this approximation should become exact. The model simulation presented on a semi-log p10t (Figure 5.2b) shows this to be true for constituents in the 1-10 ug/l range. The data show clearly a very different slope from the model predictions. From Equation 5.22 the slope of the line is governed by H5; 0.11/Bu Since 0.1, is controlled during the experiment it is a known constant. The only means of causing a less steep slope as observed in the data is by increasing 31 or decreasing H5). The possibility that 81 is larger can be discussed by looking at Equation 5.8. The model simulation has assumed that the mass of constituent k in the air, water and adsorbed phases were much less than that in the NAPL. Clearly this assumption may be violated as the NAPL is greatly reduced. Having included the other terms in Equation 5.8 would have resulted in a lower slope in Equation 5.22. To better understand the discrepancies between the experimental data and the numerical model the data trends must be investigated as the removal process proceeds. In the numerical model the slope of the function, C,,(t:) , becomes steeper (higher negative number) as time increases. This can be explained by looking at Equation 5.22 and realizing that 31 137 is actually decreasing with time because B1 is proportional to m, which can only decrease. To include a constant term, such as the adsorbed soil phase from Equation 5.7, into 81 the result would still show a semi-log slope that becomes steeper with time. The experimental data however show the opposite trend, a slope that is becoming less steep with time or is approaching a zero slope rather than a slope of -w. A model that will capture the behavior of the experimental data requires the slope (I15, 0.12/31) to tend towards zero with time. One likely mechanism that might cause this is that 111;, is decreasing as time proceeds. This would imply that partitioning from NAPL to air is decreasing or the constituent k is favoring the NAPL phase. This same behavior would result if the air phase was no longer in chemical equilibrium with NAPL and some diffusion limiting process was acting to limit transport into the air. This mechanism requires that the air phase removing a constituent such as benzene or toluene was no longer removing the compound under equilibrium conditions, at the same time the experimental data indicate that the water continued leaching at high levels indicating that equilibrium partitioning was maintained between the water and remaining NAPL. To explain this behavior one could argue that the remaining mass of a compound was in some way isolated from the air but not from the water. Several situations may give rise to this situation. One possibility was that transport through the NAPL 138 toward the air has become limited within the NAPL due to some "skinning" of heavy constituents at the air-NAPL interface. This phenomenon has been suggested as a rate limiting transport process in studies on evaporation of hydrocarbons (Stiver and Mackay, 1984) . In laboratory experiments by Yang and Wang (1977) involving evaporation of oil floating on water, a thin film was observed on the oil surface which caused a drastic drop in the volatilization rate. If the wetting sequence water-gasoline-air is occurring in the soil, then the development of a skin, limiting transfer to air, would be consistent with the fact that the aqueous phase concentrations remained high. Transport between gasoline and water would not be affected by phenomena at the gasoline-air interface. The flow rate of water through the column being much lower than air would also tend to allow the water to reach chemical equilibrium. The "skinning" mechanism is also supported by looking at the model comparison with experimental data for naphthalene (Figure 5.8). These data do not exhibit the break in slope and deviation from the model behavior like the BTEX compounds. This. would be in agreement with the "skinning" model because naphthalene would be a constituent compatible with the layer consisting of low vapor pressure compounds. Naphthalene therefore would not be expected to have diffusion limitations through the skinning layer. The following explanation is based on the physical setting of the air flow in the column. As a constituent is removed from the column, the contaminant in the top portion is 7139 .0:0H0:9:Q0: 0:0 0:0.2070 6:00:0012390 .0900 H0u:oa«uonxo 0:0 H000: mo :omflu09000 .0.m 0.56.“: we: use; 8.80 oQXNF 9x50: 09x0 09x0 09x3 oQXN 0 T d 7 + 1 AW :1 _. Q / 4 O O ... g // ‘ ‘ O ... / 44 OO 1 g / fl 00 .... AVG // “ 0.1.. 11 OF 4 /, 44 MW... / .. 0 0 / <4 Q / 4 Q .1 8: Q G Am « W €03 :0:0::0o:00 Q 6 0005. 300:3 0%. 4 n AN .00 1.00. m 82 4 4%.: \ 4§ I <0? GXNN 10421455. Aw WW \1- 88. .r 9250: 5502 Q 3x6 4 :20 O 5.132 .822 ....... 502282 -- :20 .822 I 140 removed first and the final remaining mass is ultimately located in a thin zone at the bottom of the column. This layer as it becomes thin has a shorter and shorter contact time with the flowing air. As this contact time decreases equilibrium conditions between NAPL and air are no longer maintained or equivalently H"; decreases. This would explain the deviation between data and model but the argument must be made that water and the remaining NAPL.do achieve equilibrium conditions. This may not be unreasonable due to the much slower flow rate of water as compared to air. Another possible explanation could be that some quantity of a given constituent would adsorb to the soil surfaces and be later removed under conditions producing non-equilibrium transport between air and sorbed phases. Non-equilibrium transport between soil sorbed phase and water has been documented in a review by Brusseau and Rao 1989. This mechanism would suggest that a concentration gradient must exist within the water between the soil and air; This however would imply that water flow'rates.and.saturations'would.affect effluent concentrations. As was seen in Chapter 4 significant rate limited transfer processes were not observed in the aqueous transport over a wide range of NAPL conditions. A similar result would occur if some quantity of the NAPL became entrapped in water and.hecame isolated to the air flow but was in contact with water. This, too, is a condition that would likely result in rate limited transport at the column scale in the water which as stated above was not observed in leaching 14]. experiments. 0 Two potential experimental device considerations must.be stated. The first is desorption from the Teflon o-rings used. This was tested by replacing the o-rings several times during the flow experiments and observing if a shift in the effluent concentrations resulted. Some of the o-ring changes were during periods where constituents were exhibiting deviation from the model and no shifts in data were observed. A second experimental explanation may be caused by the center of the column near the capillary barrier being isolated from the air flow. Water containing gasoline constituents enters the region isolated from flow of air and some mass is sorbed to the soil. Later, as incoming water concentrations of a constituent are reduced, the sorbed mass desorbes, maintaining increased leachate levels not representative of the entire column. The experimental results from Chapter 4, where leachate samples were collected from columns, each column venting a different length of time, also show evidence of aqueous phase concentrations remaining high long after the model predictions. These columns were operated with air flow uniformly exiting the bottom of the column. This suggest that the observed continued high water concentrations are not due to the unique flow conditions used in the dual flow experiments. Of the several possible explanations given above the "skinning" of NAPL or the reduction in air contact time seem most likely. At this point one might ask what the effect of 142 the observed deviation from equilibrium between model and data has on large scale venting operations. The first observation is that deviations in the data appear after aqueous phase concentrations have been reduced over two orders of magnitude. This indicates that more than 99% of the initial mass of a constituent has been.removed. In.a field setting this level of removal may be adequate to meet regulatory goals in groundwater. If the process producing non-equilibrium transport is distributed through the column, as the ”skinning" concept would imply, then the ability to transfer the information to a larger setting is somewhat difficult and may require a detailed numerical model. The "skinning" mechanism may also introduce a dependence on the initial quantity of gasoline. A thicker layer of gasoline may show the effects of skinning sooner than a thinner layer of gasoline. The case of contact time decreasing due to accumulation of the remaining mass at the exit of the column, would also occur at field venting sites: The tailing caused.under these conditions would not be significantly different from that seen in the columns and when compared to the large mass removed.at a site would likely have an insignificant effect of the required clean-up time. At field sites, difficulties with large scale heterogeneities in media properties and NAPL distribution are likely to overwhelm those discussed here. A conservative estimate of venting time required for larger venting geometries may be to use a low vapor pressure compound such as naphthalene which does not 143 exhibit a clear deviation from model simulations. 5.6 Summary Aqueous phase concentrations produced by gasoline contaminated soils were investigated using laboratory experiments and. a numerical model. A simple numerical transport model was used to simulate the removal of a multi- component liquid hydrocarbon through either air flow or water flow. The effect of different fluids was discussed along with complications of the multi-component transport process. The model result indicated a smooth continuous removal process and rapid removal of constituents after low levels are reached. A laboratory technique using both flow of air and water . was used to investigate the soil vapor extraction process. Water samples were collected allowing continuous evaluation of the character of the gasoline throughout the process. The data were compared to model simulations using parameters from the gasoline and the experimental conditions applied. Constituent removal compared well between data and model, indicating that chemical equilibrium transport was a good assumption for modeling transport in laboratory columns. The leachate levels for BTEX and naphthalene were reduced to less than 1 ug/l over the course of the experiments. At low concentrations in water samples it was observed that the data deviated from the model, indicating some form of non-equilibrium transport. The comparison implied that 144 transfer between gasoline and air was not at equilibrium, but water and gasoline were at equilibrium. Several potential explanations for this behavior were put forth. Of these the development of a skinning layer of low vapor pressure liquid hydrocarbons causing non-equilibrium between air and NAPL, but allowing equilibrium between water and gasoline, was the most likely explanation. CHAPTER 6 DISSERTATION SUMMARY AND RECOMMENDATIONS 6.1 Summary The previous four chapters contain results of studies on leachate characteristics of soils containing an immobile multi-component NAPL; both laboratory experiments and numerical models were employed. Each phase of the research has been presented in a separate chapter that can stand alone and a general summary of the research is presented here. The reader interested in more detailed summaries of individual phases of the work is directed to the summary section at the end of each chapter. In the first phase of the study, laboratory columns were developed to simulate field.conditionS‘with'three fluid phases present, air, NAPL and water. The columns could support flow of water in the presence of a continuous air phase and residual NAPL, both immobile (Chapter 2). The purpose of this scenario was to better understand the contribution to groundwater contamination that such conditions produce. After developing laboratory techniques capable of simulating these conditions, leachate behavior of gasoline 145 146 contaminated soils was investigated under different levels of imposed soil venting. To establish background conditions for gasoline contaminated soils, unsaturated flow columns were leached with water immediately following the introduction of gasoline. Results of these initial experiments showed that aqueous.phase concentrations onBTEX in leachate*were equal to equilibrium concentrations measured by batch experiments. Gasoline contaminated soils were leached for long periods of time to observe selective aqueous phase transport of constituents found in gasoline (Chapter 3). The experiments demonstrated the effect of initial conditions on the mobilization of gasoline in the early phases of leaching. It was determined that by establishing residual levels of water in soil, prior to the introduction of gasoline, mobilization of gasoline during the experiments could be avoided. Selective:removal.of constituents was.observed.over'the course of the leaching experiments. This behavior was simulated by employing a numerical model based on equilibrium partitioning between phases. The modelling results indicated that some vapor transport was likely taking place. The source of vapor transport.was not identified, but tests suggested.the loss was constant with time and.could.be:minimized when.column leaching time was reduced. Aqueous phase transport was also investigated in soils subjected to different levels of soil venting (Chapter 4). A set of 25 columns, each.vented.a different length of time, was subsequently leached with water to evaluate their potential to 147 contaminate groundwater. An evaluation of aqueous phase concentrations in vented soils showed that concentrations were at chemical equilibrium with other phases in the system and were independent of water saturation present in the soil. These results indicated.that.the length of column used, 10 cm, was long enough for water to reach equilibrium with the contaminant source and therefore these experimental results can be representative of others systems at equilibrium. This result also indicates that in laboratory studies of leachate from NAPL contaminated soil, a distinction between saturated and unsaturated flow may not be required. Analyzing the leachate concentrations, produced by soil vented over a range of applied air volumes, showed that concentrations were reduced over 4 orders of magnitude in effluent. water, and. that all constituents ‘measured. were reduced to concentrations below lug/l. Higher vapor pressure compounds were removed much earlier than low vapor pressure compounds. A numerical model, used to simulate the column experiments, showed results in reasonable agreement with the experimental data. The experiments directly demonstrated the effectiveness of soil venting for reducing concentrations of BTEX and naphthalene in leachate from gasoline contaminated soils. In the last phase of the research a technique was developed where flow'of both air and water were simultaneously maintained in laboratory columns (Chapter 5). These flow experiments were continuous throughout the soil venting 148 process and allowed the collection of leachate samples without disturbances to the system. The data were used to evaluate the adequacy of a simple equilibrium transport simulation model. The results indicated that the majority of mass removal from the columns could be simulated by equilibrium relationships. Data at very low concentrations in water showed clear deviations from the simple model employed. Analysis suggested that non-equilibrium transport between air and gasoline was most likely causing the discrepancies between experiment and model. Through this research effort our understanding of how non-aqueous phase liquids in the unsaturated zone contribute to groundwater contamination has been advanced. Experimental equipment and methods were developed to conduct investigations of leachate from an immobile NAPL source in porous media. The process of removing constituents of a multi-component NAPL by water leaching and air venting was studied. 149 6.2 Recommendations The following list includes directions for future research based on the present study: - Investigate different NAPL mixtures for non-equilibrium transport characteristics. - Identify a pure NAPL that will exhibit non-equilibrium transport characteristics at low fluid velocities and investigate transport at different fluid saturations within the soil columns. - Develop or acquire a model that includes the sorption to soil in multi-component hydrocarbon systems and investigate this effect. - Develop or acquire a model that includes non-equilibrium transport between phases and incorporate transfer coefficients appropriate for gasoline. Test this model for simulating the dual flow experimental data. - Investigate the effect of heterogeneities located in the soil columns on soil venting and subsequent leachate. APPENDICES APPENDIX A Detailed Experimental Procedures and Error Estimates Introduction: The following pages describe the procedures used to bring soil columns used in this study to initial conditions and procedures associated with the venting and leaching phases of the experiments. The procedures have evolved over the period of study and my not reflects those used in the earliest phase of the research. This is followed by error estimates for measurement significant to the analysis of data. Column Packing: Prior to packing the columns with soil the individual components were washed and.oven.drieda IPortions of the column assembly were then weighed so that later determinations of soil and fluid masses could be determined. To determine the residual mass of fluid in the column at a later time, the weight of the column apparatus and soil need to be subtracted out. Therefor the weight of materials located above the capillary barrier must be weighed separately. This includes the glass tube, sleeves, threaded rods and nuts (these represent a single unit always left assembled and referred to as the column center), along with.the top and bottom teflon 0- rings, top cap, and nuts. With the total weight of these components measured a column could be weighed after removing the capillary barrier, and the mass of fluid determined. The second critical weight needed is the weight of the column assemble before packing. For this the column center, capillary barrier with teflon o-ring on the upper side and a rubber o-ring on the lower side, and the bottom cap and nuts are assembled and weighed. All weights are made after calibrating the scale to a 500 gram standard. The columns are then ready for packing. . The packing device consists of a reservoir to be filled with air dry soil, a valve to control the flux of soil, and a 50 cm long, 3 cm diameter tube with a 2 mm mesh screen in the lower part to help uniformly distribute the soil particles (Figure Al.1). The screen was positioned approximately 2-3 cm above the capillary barrier and the valve was opened to a position that filled the columns at a rate of about 1 mm/s. As the column fills the position of the screen was maintained at 2-3 cm above the soil surface by moving the packing device upward with a pulley. The tube was rotated to ensure uniform distribution of soil particles. When the columns were filled a cap was placed on the top and the columns were dropped 10 times from a height of 5 cm onto a one inch thick hard rubber 151 152 / / FUNNEL 7" VALVE |L_l SCREEN [___I son. COLUMN l‘___'J—" Figure A1.1. Soil Packing Device. 153 pad. Any additional soil needed was then added and a rubber stopper, of slightly less diameter than the glass tube, was placed on the soil surface. On top of this was placed a 4 Kg weight and the columns were again dropped 10 times from a height of 5 cm. At this point the packing was complete and only small amounts was soil were added or removed.to bring the surface of the soil even with the top of the glass tube. The columns were then weighed and the mass of soil determined. The top o-ring and cap were then assembled. Saturation procedure: The columns were then saturated with water. To do this a constant head burette was attached to the bottom outlet of the column and the zero head position placed near the top of the column. Water imbibed into the soil and was allowed a minimum of 24 hours to distribute throughout the soil mass. At this point a first check for any system leaks could be made. After water redistribution the column outflow port was attached to a vacuum. Generally this was done using a teflon tube put through a #10 rubber stopper that was place in a one liter vacuum flask. The time was recorded when drainage of . water was initiated. Drainage was allowed to continue until the rate became very slow. The 10 cm columns used in this study were drained for 5 days. After the drainage period was complete, the capillary barrier was removed for water mass determination. The capillary barriers are removed with the vacuum continually attached. The columns are first tipped to a shallow angle with the bottom lower than the top. The reason for this is that should the seal between the bottom cap and the capillary barrier be broken the water below the ceramic plate will run out of the column and not into the soil. At this point the wing nuts holding the bottom in place are evenly loosened and the bottom cap and capillary barrier carefully removed from the soil. Any small amounts of soil attached to the ceramic plate was carefully brushed onto paper and put back in the column. The columns were then weighed upside down and the mass of water in the column could be calculated. A gasoline saturated capillary barrier was then placed in the column and the rubber o-ring and bottom cap attached. Gasoline was then injected into the top of the columns using 10ml glass syringes. The ports were sealed between injections. Enough gasoline was injected to over saturate the soil well above the eventual residual saturation achieved (greater than 5 times the final volume). This gasoline was allowed 24 hours to distribute throughout the soil. The gasoline drainage was initiated by attaching the outflow port to a 300 mbar vacuum. The gasoline drained was collected in a vial within the vacuum system. 10 cm columns used in this study were drained for a period of 24 hours. The capillary barrier was then removed as before and the columns weighed to determine the mass of gasoline remaining. This process was done quickly to minimize vapor 154 losses. At this point the columns have residual fluids and are ready for leaching, venting or dual flow experiments. Leaching experimental procedure: After the columns were drained to residual fluid levels, the gasoline saturated capillary barriers were replaced by oven dried (104°C, 24 hours) stainless steel capillary barriers. At this point column caps and o-rings used to bring the soil to residual fluids are replaced with clean o-rings and stainless steel end caps. The replacement of the lower teflon o-ring between the capillary barrier and the glass tube is difficult as the soil is packed next to the ring. The 0- ring is removed and some soil is cleared using a Q-tipR swab. A clean o-ring is carefully placed on the glass tube and the stainless steel barrier put in place. The edge of the stainless steel barrier is wrapped with teflon tape to improve sealing. The columns were then saturated with water using a constant head burette as above. A.period of 48 hours or more was given for the water to distribute and chemical equilibrium to be achieved. Saturated flow of water could be initiated by attaching a teflon inflow line to the column from a Harvard Apparatus syringe pump. The flow rate is set by the pump and samples can be collected from the outflow line. On the exit end of the column a stainless steel 12? gauge needle is used to which glass syringes can be attached for sample collection. The columns are designed so that the water flow condition can be saturated or unsaturated. The transition from saturated to unsaturated simply involves applying a vacuum to the capillary barrier in the bottom of the column. The outflow needle is placed through a #10 rubber stopper so that it can be placed in a one liter vacuum flack and the vacuum transmitted to the soil. With the vacuum applied, the soil will not drain until.a port is opened at the top of the column to allow air to enter the column- Opening the 2nd.port on the top of the column causes the column to drain and rapidly come to a steady state unsaturated flow condition. Samples under these conditions are then collected using glass syringes in the vacuum flask: When a sample is collected the valve on the outflow line is closed while the syringe is attached so the water does not imbibe into the soil leaving an air slug in the line. This minimizes the contact of samples with air. When a syringe is full it is detached and 5 ml of sample is transferred to a 20 ml headspace vial (20 ml glass serum bottle with teflon-faced septum and aluminum crimp cap) and immediately sealed. Samples were then labelled and stored in the dark, at 4°C until analysis. Samples were analyzed within an appropriate period of time. The quantification of soluble gasoline components in leachate water was performed using headspace gas chromatography with a flame ionization detector and external standards using the method of Voice et al. ( 1990) . A 155 multipoint calibration curve was determined for each of the five components of interest. A DB-624 megabore capillary column (30 m * 0.53 mm i.d., J 8 W scientific) was employed. The initial oven temperature was 60°C ramping in 10 minutes to a final temperature of 200°C. The injector and detector were maintained at 200°C. The sample equilibration temperature and time were 80°C and one hour respectively. The retention time of m- and p-xylene, using the technique outlined above, were the same; therefore, these two compounds were quantified as a single component. Batch equilibrium experiments were conducted for gasoline following the method outlined by API (1985). Venting experimental procedure: Following the establishment of residual fluids as described above, gasoline saturated capillary barriers were replaced with a stainless steel screen held in a ring. The ring was of the same size as the capillary barriers and could simply replace them easily. The end caps and o-rings were replaced with clean materials as in the leaching preparations. With the screen in place an air flow line was attached to the top of the column and air flow was regulated with a Matheson variable area flow meter and regulator. Prior to the meter the air is dried using Dryrite‘ and filtered through activated carbon to remove any hydrocarbons in the air stream. Flow of air through the column is verified by measuring the flow out the column exit line using a bubble meter. Dual flow experiment procedure: Following the establishment of residual fluids as described above, the columns were saturated with water using the same procedure for the leaching experiments. After the columns were saturated with water the bottom was removed and a quantity of soil was removed equalling a thickness of approximately 0.5 cm. This soil mass was weighed. The soil removed was replaced with uncontaminated soil to bring the column back to full. The o-ring normally used between the glass cylinder and the capillary barrier was replaced by glass beads having the same diameter as the o-ring thickness. Approximately 15-20 beads were used leaving gaps for air to exit the bottom of the column. The stainless steal plate was then placed on the beads and a stainless steal bottom was used. The bottom was carefully tightened in place because the glass beads can easily fracture the glass tube. At this point tape is wrapped around the gap between the sleeve and bottom to minimize convective transport of air out of the column. Water flow was then initialed under unsaturated conditions as described in the leaching procedure. After steady flow conditions were established and some water samples had been collected air flow was initiated through the 2nd port on the top of the columns and the tap around the columns removed. 156 Flow of air could be verified using a soap solution or drawing air samples near the capillary barrier and analyzing on the GC. Water samples were collected continuous through the course of the venting experiment. Error estimates for experimental measurements: Fluid Saturation Determinations: Made by differential column mass, balance calibrated at 500 g. Balacne error is 10.03 grams. Estimated error for mass of gasoline in 10 cm columns is 2%. Flow of water: Controlled by Harvard Apparatus syringe pump estimated error for this instrument is 12%. Controlled by Soil Measurements Systems cycling syringe pump and estimated based on records of inf luent water measurements over time is 15%. Flow of air: Made by monitoring calibrated Matheson variable area flow meter. Reading of the meter varied approximately 5015 units. The calibration error for the instrument is 5%. Estimated error in air flow for venting experiments is 15%. Density of gasoline: Measured by pycnometer method. Error is estimated at 0.7210.2 g/cm3 or 3% error. Dimensionless Time T: Determined using the above measurements. The accumulated error for this quantity is approximately 120% error. Concentrations of BTEX and Naphthalene' in water: Error estimates based on standard calibration curves and use of check standards during GC runs is approximately 5% for water samples. Mass fractions of BTEX and Naphthalene in gasoline: Error estimates her are also on the order of 5% however higher molecular weight compounds such as Naphthalene errors increase. Errors for unknown compound estimates are uncertain. Peak # ‘OQQGUIQQNH APPENDIX 3 Input parameters for numerical model. Ihn 8670 45300 21700 8400 3950 4920 1670 4150 1990 1740 1660 1250 964 696 350 331 350 0 336 243 194 435 176 174 185 147 121 -123 40.4 70.1 90.5 68.8 50 146 62.6 0 8.2 0 18.3 22.8 MW 44 44 58 58 58 70 70 72 72 72 73 86 86 86 86 86 78 78 86 86 84 78 100 100 100 100 100 100 100 100 114 100 100 92 114 114 114 114 114 114 fraction fraction 0.000001 0.000016 0.001983 0.006136 0.076103 0.002082 0.002334 0.106927 0.042811 0.008753 0.018610 0.027007 0.066702 0.030664 0.035032 0.010811 0.014970 0.000707 0.006081 0.028757 0.020608 0.017390 0.01716 0.015200 0.006324 0.001317 0.002646 0.009798 0.002758 0.003145 0.016577 0.011469 0.001422 0.060280 0.010361' 0.000690 0.002391 0.000837 0.001145 0.003332 157 mole 0.000004 0.000033 0.003078 0.00952 0.118076 0.002676 0.003001 0.133642 0.053507 0.01094 0.022942 0.028259 0.069796 0.032086 0.036657 0.011313 0.017271 0.000815 0.006363 0.030091 0.022077 0.020063 0.015442 0.013678 0.005691 0.001185 0.002382 0.008817 0.002482 0.00283 0.013085 0.010321 0.001279 0.058962 0.008179 0.000545 0.001887 0.00066 0.000904 0.00263 Km/Kn 31400 1480 197 193 1080 56.4 77.1 136 160 16 0.812 16.7 91.9 2.36 33.6 569 22.7 22.4 OOOOOIFPOU‘O 0‘ \O N 6.19 30.6 596 0.211 10.113 Peak f 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 MW 128 128 128 106 106 106 106 140 140 140 140 140 140 140 140 140 140 140 150 150 150 160 160 160 140 140 140 140 140 150 150 150 150 170 170 170 128 150 150 150 160 150 150 150 150 150 150 158 fraction fraction 0.001487 0.008004 0.004196 0.017803 0.061899 0.000331 0.024745 0.000999 0.001010 0.002137 0.004131 0.001785 0.005445 0.026421 0.011020 0.007351 0.032851 0.002099 0.002416 0.000628 0.007415 0.008033 0.003233 0.006102 0.005704 0.002394 0.001886 0.003652 0.005660 0.003266 0.003023 0.006830 0.001975 0.000529 0.003398 0.001434 0.003564 0.000794 0.001180 0.00032 0.001997 0.000474 0.001346 0.000849 0.000728 0.001666 0.00512 mole 0.001045 0.005627 0.00295 0.015114 0.052549 0.000281 0.021007 0.000642 0.000649 0.001373 0.002655 0.001147 0.0035 0.016983 0.007083 0.004725 0.021116 0.001349 0.001449 0.000377 0.004448 0.004518 0.001818 0.003432 0.003667 0.001539 0.001212 0.002347 0.003638 0.001959 0.001813 0.004097 0.001184 0.00028 0.001799 0.000759 0.002505 0.000476 0.000708 0.000191 0.001123 0.000284 0.000807 0.000509 0.000436 0.000999 0.003071 00000000 00000000 Peak # 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 IAN O O \O O UIU'IOUlOubch-FGOQUINOOON OO O OOOG NO. 0 eeee eem MW 150 150 150 150 150 150 150 150 150 150 160 160 160 160 160 160 160 159 fraction fraction 0.003497 0.000231 0.000066 0.000055 0.000187 0.00064 0.001191 0.001975 0.000518 0.000209 0.000463 0.000077 0.000055 0.000043 0.000043 0.000010 0.000019 mole 0.002098 0.000139 0.000039 0.000033 0.000112 0.000383 0.000714 0.001184 0.000311 0.000125 0.00026 0.000043 0.000031 0.000024 0.000024 0.000006 0.000011 Identified compounds for this study: Benzene Toluene Ethylbenzene m&p-Xylene o-Xylene Naphthalene #22 #34 #44 #45 #47 #77 Km/Hm 27. 37. OOOOOOOOOOOOOOOmu HIV 0.034 0.000 OOOOOOOOOOOOOOO APPENDIX C Numerical code for venting and or leaching simulations. Based on a model published by Michael Marley and George Hoag in a paper titled "Induced soil venting for the recovery [restoration of gasoline hydrocarbons in the vadose zone," Presented at the NWWA/API Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground Water, Houston, TX, November 5-7,1984. $LARGE REAL X(200),XMAS(200),XML(200),XMOL(200),ZT(200) REAL XMLR(200),CMW(200),KWATER(200) REAL VP(200),NCI(200) INTEGER PEAK(200) REAL MASRT TEMP=294.25 R=82.04 v=1000.00 INPUT FLOW RATE AIR,WATER. (Qair ml/min, Qwater ml/hr) ratio used as a coefficient of the initial mass 0000 READ(8,*) Q,QWATER,ratiO WRITE(*,*) Q,QWATER,ratio READ NUMBER OF COMPOUNDS AND THOSE OF INTEREST NC number of compounds NI number of compounds to be printed NCI identification number of compound NC,NI,NCI(I) READ (8,*) NC,NI,(NCI(I),I=1,NI) READ IN TOLERANCE MAX ITERATION NSKIP FOR WRITE READ (8,*) TOL,MAXIT,NSKIP READ FOUR TIME STEPS READ (8,*) DELT1,DELT2,DELT3,DELT4 J=1 READ IN PARAMETERS FOR NC COMPOUNDS PEAK identification number VP partitioning coef. H.i CMW compound molecular weight XMAS compound mass fraction 000000 0 0 000 0000000 160 161 X compound mole fraction kwater partitioning coef. H“ 01000 READ (8.*)(PEAK(I).VP(I):CMW(I).XMAS(I).X(I). lkwater(i),I=1,NC) WRITE (*,10) (PEAK(I),VP(I),CMW(I),XMAS(I), 1X(I),I=1,NC) 10 FORMAT (I5,5X,F6.4,3X,F8.1,2X,F8.3,5X,F7.5) TMR=TOTAL MASS REMOVED (MG) MASRT=MASS RATE LOSS (MG/MIN) TOT=TOTAL MASS REMOVED (MG) TOTMOL=TOTAL # OF MOLES REMAINING TIME=0.0 ' TOTMOL=0.0 TOT=0.0 C CALC totmas Do 12 I=1,NC xmas(i)=xmas(i)*ratio xmol(i)=xmas(i)/cmw(i) totmol=totmol+xmol(i) totmas=totmas+xmas(i) 12 CONTINUE 00000 DO 14 I=1,NC X(I)=XMOL(I)/TOTMOL 14 CONTINUE Do 600 J=1,MAXIT,1 TMR=0.0 TOTMOL=0.0 MASRT=0.0 IF (TIME.LT.10.0) DELT=DELT1 15 IF (TIME.GE.10.0) DELT=DELT2 IF (TIME.GE.60.0) DELT=DELT3 IF (TIME.GE.300.0) DELT=DELT4 TIME=TIME+DELT Do 100 I=1,NC,1 CALCULATE ZT=CONCENTRATION'OF COMPONENT IN‘VAPOR.PHASE CALCULATE XMLR=MASSLOSS RATE OF COMPONENT IN TIME DELTA T CALCULATE XML=MASSLOSS OF COMPONENT IN TIME DELTA T 000000 ZT(I)=(VP(I)*X(I))/V XMLR(I)=Q*ZT(I) XML(I)=XMLR(I)*DELT =TMR+XML(I) MASRT=MASRT+XMLR(I) XAQU=QWATER*DELT*X(I)*KWATER(I)/(60000*V) 162 XMAS(I)=XMAS(I)-XML(I)-XAQU IF(XMAS(I).LE.0.0) XMAS(I)=0.00000000 70 XMOL(I)=XMAS(I)/CMW(I) TOTMAS=TOTMAS-XML(I)-XAQU TOTMOL=TOTMOL+XMOL(I) C 100 CONTINUE c CALCULATE NEW MOLE FRACTION OF EACH COMPONENT 0 DO 200 I-l,NC,l IF (TOTMOL.EQ.0.0) X(I)=0.0 c WRITE(*,*) TOTMOL IF (TOTMOL.EQ.0.0) GO TO 900 X(I)=XMOL(I)/TOTM0L 200 CONTINUE c c WRITE(*,*) TMR c IF(TMR.LE.TOL) GO TO 900 TOT=TOT+TMR c c PRINT TO A FILE 0 IF (MOD(J,NSKIP).EQ.0.0) THEN WRITE(*,300) TIME,(X(NCI(I))*kwater(nci(i)),I=1,NI) 1,TOTMAS WRITE(7,300) TIME,(X(NCI(I))*kwater(nci(i)),I=1,NI) 1,TOTMAS ENDIF 300 FORMAT(3X,F8.1,5X,6F12.S) 600 CONTINUE 900 CONTINUE STOP END LIST OF REFERENCES API. 1985. "Laboratory Study of Solubilities of Petroleum Hydrocarbons in Groundwater", API #4395. Baehr, A.L. and M.Y. Corapcioglu. 1987. "A Compositional Multi-phase Model for Groundwater Contamination by Petro- leum Products, 2, Numerical Solution," Water Resources Research, Vol. 23 pp. 201-213. Baehr, A.L. 1987. "Selective transport of hydrocarbons in the unsaturated zone due to aqueous and vapor phase partitioning," Water Resources Research, Vol 23, pp. 1926- 1938. Baehr, A.L., G.E. Hoag and M.C. Marley. 1989. "Removing volatile contaminants from the unsaturated zone by induced advective air-phase transport," J. Contaminant.Hydrology, Vol. 4, pp. 1-26. Brown, T.L., and R.E. LeMay, Jr. ' Science, Prentice-Hall, Inc. New Jersey, pp. 813. Brusseau, M.L., and P.S.C. Rao. 1989. "Sorption nonideality during organic contaminant transport in porous media," Crit. Rev. Environ. Control, Vol. 19, pp. 33-99. Brusseau, M.L., P.S.C. Rao, R.E. Jessup, and J.M. Davidson. 1989. "Flow interruption: .A method for investigating sorption nonequilibrium," .Journal of Contaminant.Hydrolo- gy, Vol. 4, pp. 223-240. Burris, D.R. and W.G. MacIntyre. 1986. "Solution of Hydrocar- bons in a Hydrocarbon-Water System with Changing Phase Composition due to Evaporation," Environ. Sci. Technol., Vol. 20, pp. 296-299. Cameron, D.R. and Klute, A. 1977. "Convective-Dispersive Solute Transport with a Combined Equilibrium and Kinetic Adsorption Model," Water Resources Research, Vol. 13, NO. 1. Campbell. G-S- 1985. "WW WW." Elsevier pp- 150- 163 164 Carslaw, H.S., and Jaeger, J.C. 1959. anfigg;19n_gfi_fig§§_in Solids, Oxford and the Clarendan Press. Coats, K.H., and B.D. Smith. 1964. "Dead-End Pore‘Volume and Dispersion in Porous Media," Soc. Pet. Eng. J., Vol. 4, Pages 73-84. Corapcioglu, M.Y. and H.L. Baehr. 1987. "A Compositional Multi-phase Model for Groundwater Contamination by Petro- leum Products, 1, Theoretical Considerations," Water Resources Research., Vol. 23 pp. 191-200. Corey, A.T. 1986. ueChaniCs_of_lmmis:1ble.Eluids_in_£orous Media1_Eater_BesourCes_Publications. Littleton. ColoradO- pp. 255. Fried, J.J., P. Hunter and L. Zilliox. 1979. "Ground-Water Pollution by Transfer of Oil Hydrocarbons," Groundwater, Vol. 17, No. 6, pp. 586-594. Gannon, K., D.J. Wilson, A.N. Clarke, R.D. Hutch, and J.H. Clarke. 1989. "Soil clean up by in-situ aeration. II. Effects of impermeable caps, soil permeablitity, and evaporative cooling," Separ. Sci. Tech., Vol. 24, pp. 831- 862. Gardner, W.R. 1958. "Some Steady State Solutions of the Un- saturated Moisture Flow Equation with Applications to Evap- oration from a Water Table", Soil Sci., Vol. 85, pp. 228- 232. Hayden, N .J . 1991. "Gasoline retention, enhanced volatiliza- tion, and leaching in unsaturated soil." Ph.D. Disserta- tion, Michigan State University, East Lansing, MI. Hoag, G.E. and Marley, M.C. 1986. "Gasoline Residual Saturation in Unsaturated Uniform Aquifer Materials," J. Environ. Engineering, Vol. 112, No. 3, 586-604. Hunt, J.R., N. Sitar and KIS. Udell. 1988a. "NOnaqueous Phase Liquid Transport and Cleanup I", Analysis of Mecha- nisms, Water Resources Research, Vol. 24, pp. 1247-1258. Hunt, J.R., N. Sitar and K.S. Udell. 1988b. "Nonaqueous Phase Liquid Transport and Cleanup II", Experimental Studies, Water Resources Research, Vol. 24, pp. 1259-1269. Hutzler, N.J., B.E. Murphy, and J.S. Gierke. 1989. "State of technology review soil vapor extraction systems," EPA/60- 0/2-89/024, CR-814319-01-1. Johnson A.I. 1967. "Specific Yield-Comparison of Specific Yields for Various Materials," U.S. Geol. Survey Water-Su- pply Paper 1662-D, 74 pp. 165 Johnson, P.C., M.W. Kemblowski and J.D. Colthart. 1990a. "Quantitative analysis of the cleanup of hydrocarbon-con- taminated soil by in-situ soil venting," Ground Water, Vol. 28, pp. 413-429. Johnson, P.C., C.C. Stanley, M.W. Kemblowski, D.L. Byers, and J .D. Colthart. 1990b. "A practical approach to the design, operation, and monitoring of in situ soil-venting sys- tems,"Ground Water Monitoring Review, Vol. X, No. 2, pp. 159-178. Johnson, R.L., C.D. Palmer and J.F. Keely. 1987. "Mass transfer of organics between soil., water and vapor phases: Implications for monitoring, biodegradation and remedia- tion," NWWA/API Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground Water, Houston, TX, November 17-19. Lapidus,L., and. N.R. Amundson. 1952. "Mathematics of Adsorption in Beds, VI. The Effects of Longitudinal Diffusion in Ion Exchange and Chromatographic Columns," J. of Physical Chemistry, Vol. 56, 984-988. Marley, M.C. and G.E. Hoag. 1984. "Induced soil venting for the recovery/restoration of gasoline hydrocarbons in the vadose zone," NWWA/API Conference on Petroleum Hydrocar- bons and Organic Chemicals in Ground Water, Houston, TX, November 5-7. Miller, C.T., M.M. Poirier-McNeill and A.S. Mayer. 1990. "Dissolution of Trapped Nonaqueous Phase Liquids: Mass Transfer Characteristics," Water Resources Research, Vol. 26, pp. 2783-2796. Nielson, D.R., M.Th. van Genuchten, and J.W. Biggar. 1986. "Water flow and solute transport processes in the unsatu- rated zone," Water Resources Research, Vol. 22, pp. 89S- 108$. Pfannkuch, H.O. 1984. "Determination of the Contaminant Source Strength from.Mass Exchange Processes at the Petro- leum-Ground-Water Interface in Shallow Aquifer Systems", Proceedings of the NWWA/API Conference on Petroleum Hydrocarbons and Organic Chemicals in Ground Water Preven- tion, Detention and Restoration, Published by National Water Well Association. Prill R.C., A.I. Johnson and D.A. Morris. 1965. "Specific Yield-Laboratory Experiments Showing the Effect of Time on Column Drainage," U.S. Geol. Survey Water-Supply Paper 1662-8, 55 pp. 166 Prill R.C., and A.I. Johnson. 1967. "Specific yield - moisture tension techniques," U.S. Geol. Survey, open-file report pp. 65. Schwille. F. 1988. 0W ' e ' , pp. 103-112. Translat- ed by J.F. Pankow, Lewis Publications. Smith, W.O. 1933. "The final distribution of retained liquid in an ideal uniform soil," Physics, Vol. 4, pp.425-438. Stiver W., and D. Mackay. 1984. "Evaporation rate of spills of hydrocarbons and petroleum mixtures," Environ. Sci. Technol., Vol. 18, pp. 834-840. Texas Research Institute , Inc., "Forced Venting to Remove Gasoline Vapor From a Large-Scale Model Aquifer. " Submitted to: American Petroleum Institute, 1983. van der Waarden, M., A.L.A.M. Bridie and W.M. Groenewoud. 1971. Transport of Mineral Oil Components to Groundwater, I. Model Experiments of the Transfer of Hydrocarbons from a Residual Oil Zone to Trickling Water", Water Research, Vol. 5, pp. 213-226. van Genuchten, M.Th. and Wierenga, P.J. 1976. "Mass Transfer Studies in Sorbing porous Media I. Analytical Solutions," Soil Science Soc. Amer., Vol. 40, no.4, Pages 473-480. van.Genuchten, M.Th. and.W.J} Alves. 1982. "Solutions of the one-dimensional convective-dispersive solute transport equation," U.S. Dept. of Agriculture Bulletin, No. 1661, pp. 151. van Genuchten, M.Th. and P.J. Wierenga. 1986. "Chapter 44, Solute Dispersion Coefficients and Retardation Factors," In Methods of Soil Analysis," Ed. by A.Klute, American Society of Agronomy, Inc. & Soil Science Society of America,Inc. Valocchi, A.J. 1985. "Validity of the local equilibruim assumption for modeling sorbing solute transport through homogeneous soils," Water Resources Research, Vol. 21, pp. 808-820. Versuluys, J., 1917. "Die Kappillaritat der Boden," Inst. Mitt. Bodenk., Vol. 7, pp. 117. Voice, T.C., N.J. Hayden and Y.L. Pan. 1990. "Headspace Gas Chromatography of Petroleum-Derived Contaminants in Soils and Groundwater," Presented at the 1990 Pittsburgh Confer- ence on Analytical Chemistry and Applied Spectroscopy, New York. 167 Weber W.J., P.M. McGinley, and L.E. Katz. 1991. "Sorption phenomena in subsurface systems: Concepts, models and effects on contaminant fate and transport," Water Re- search, Vol. 25, No. 5, pp. 499-528. Wierenga, P.J., R.J. Black and P. Manz. 1973. "A Multichan- nel Syringe Pump for Steady-State Flow in Soil Columns", Soil Sci. Soc. Amer. PrOc., Vol. 37, pp. 133-135. Wilson, D.J., A.N. Clarke and.J.H. Clarke. 1988. "Soil clean up by in-situ aeration. I. Mathematical modeling," Separ. Sci. Tech., Vol. 23, pp. 991-1037. Wilson, D.E., R.E. Montgomery, and M.R. Sheller. 1987. "A Mathematical model for removing volatile subsurface hydrocarbons by miscible displacement," Water, Air, and Soil Pollution, Vol. 33, pp. 213-255. Wilson, J.L., S.H. Conrad, W.R. Mason, W. Peplinski, and E. Hagan. 1990. "Laboratory investigation of residual liquid organics form spills, leaks, and the disposal of hazardous wastes in groundwater," EPA/600/6-90/004, EPA CR-81357l. Wootan W.L., and T. Voynick. 1984. "Forced venting to remove gasoline vapor from a large-scale model aquifer," Texas Research Institute, Final Report to American Petroleum Institute. Yang, W.C., and H. Wang. 1977. "modeling of Oil evaporation in aqueous environment," Water Research, Vol. 11, pp. 879- 887. Zalidis, G. 1987. "Experimental and Theoretical Analysis of Capillary Rise through Porous Media", Ph.D. Dissertation, Michigan State University, East Lansing, MI. Zalidis, C.C., R.B. Wallace and T.C. Voice. 1991 "Influence of initial water saturation on residual gasoline saturation in the vadose zone, " Submitted to J. Contaminant Hydrology.