water

Article Evaluation of LNAPL Behavior in Water Table Inter-Fluctuate Zone under Groundwater Drawdown Condition

Reza Azimi 1, Abdorreza Vaezihir 1,* , Robert J. Lenhard 2 and S. Majid Hassanizadeh 3 1 Department of Earth Sciences, University of Tabriz, 29 Bahman Blvd., Tabriz 5166616471, Iran; [email protected] 2 San Antonio, TX 78253, USA; [email protected] 3 Department of Earth Sciences, Utrecht University, 3584 CD Utrecht, The Netherlands; [email protected] * Correspondence: [email protected]



Received: 20 July 2020; Accepted: 15 August 2020; Published: 20 August 2020


Abstract: We investigate the movement of LNAPL (light non-aqueous phase liquid) into and out of monitoring wells in an immediate-scale experimental cell. Aquifer material grain size and LNAPL viscosity are two factors that are varied in three experiments involving lowering and rising water levels. There are six monitoring wells at varying distances from a LNAPL injection point and a water pumping well. We established steady water flow through the aquifer materials prior to LNAPL injection. Water pumping lowered the water levels in the aquifer materials. Terminating water pumping raised the water levels in the aquifer materials. Our focus was to record the LNAPL thickness in the monitoring wells under transient conditions. Throughout the experiments, we measured the elevations of the air-LNAPL and LNAPL-water interfaces in the monitoring wells to obtain the LNAPL thicknesses in the wells. We analyze the results and give plausible explanations. The data presented can be employed to test multiphase flow numerical models.

Keywords: groundwater level fluctuations; LNAPL; sand tank

1. Introduction Groundwater contamination by petroleum products is a common problem in many industrialized areas as well as in oil producing countries. Gasoline (petro), diesel fuel, and crude oil are potential groundwater pollutants. As a group, they are commonly called light non-aqueous phase liquids (LNAPLs), which may exist in the subsurface in four forms: vapor phase in vadose zone pore spaces, vadose zone liquid residual LNAPL, free phase LNAPL in the water-unsaturated zone, and entrapped LNAPL globules above and below the water table [1–3]. Understanding the volume and underground distribution of these forms is essential for predicting potential groundwater contamination and for designing associated remediation cleanup programs. Remediation of groundwater at LNAPL contaminated sites is still a complicated and challenging issue. Poor planning can increase the potential risk to public health and the environment. Over time, engineers and scientists have increased their knowledge of subsurface LNAPL behavior from laboratory experiments, field observations, and models. From older conceptions that considered LNAPL as pancakes on the surface of the water-saturated region to the current understanding that regard LNAPL as free, residual, and entrapped phases with varying degrees of saturation, significant progress has beenmade [4]. Understanding how all forms of LNAPL are distributed underground [5], along with its possible risk to human health and the environment [6], is essential for defining and developing an effective LNAPL cleanup strategy.

Water 2020, 12, 2337; doi:10.3390/w12092337 www.mdpi.com/journal/water Water 2020, 12, 2337 2 of 15

Groundwater table fluctuations can change the subsurface distribution of the different LNAPL phases [7–9]. As the water table rises or falls, free LNAPL may become entrapped or entrapped LNAPL may become free LNAPL, respectfully [10]. Further, in the dry seasons as the groundwater level drops, free LNAPL may become a residual form in the unsaturated zone as the bulk of the free LNAPL migrates downwards with the water table. Initially, various researchers have predicted the relationship between the thickness of LNAPL observed in a monitoring well to its actual thickness in groundwater aquifers [11–17]. They found that the actual thickness of LNAPL in the subsurface is less than that in monitoring wells. They calculated the volume of LNAPL in the subsurface from the predicted LNAPL thickness in the subsurface and the subsurface porosity. Some assumed the LNAPL saturation was one minus the residual water saturation instead of one (completely LNAPL saturated). In either case, the LNAPL saturation was assumed to be constant with depth—the pancake model. Later, Farr et al. and Lenhard, Parker [18,19] showed that LNAPL saturations in the subsurface vary with depth and is a function of the LNAPL-water capillary pressure. They predicted the LNAPL-water capillary pressures from the depths (or elevations) of the air-LNAPL and LNAPL-water interfaces in a monitoring well, assuming vertical LNAPL movement has terminated, i.e., vertical equilibrium. From the LNAPL-water capillary pressures, LNAPL saturations as a function of depth were predicted using approaches by Parker et al. [20]. When the LNAPL saturations are summed over a vertical slice of the subsurface near the well, the LNAPL volume of the subsurface vertical slice (specific LNAPL volume) is predicted. Parker, Lenhard [21] developed an approach to estimate the LNAPL transmissivity, which predicts the rate at which LNAPL in a vertical slice may move into a monitoring well utilizing relations by Parker et al. [20]. The approaches in Farr et al., Lenhard, Parker and Parker, Lenhard [18,19,21] assume all LNAPL is mobile and capillary pressure-saturation relations are single valued (i.e., non-hysteretic). Parker, Lenhard. [21] further assume saturation-relative permeability relations are single valued (i.e., non-hysteretic). The equations developed by Farr et al. and Lenhard, Parker [18,19] are analytical expressions to calculate the specific LNAPL volume. The equations developed by Parker, Lenhard [21] are vertically integrated capillary pressure-saturation-relative permeability relations (constitutive relations) for implementation in two-dimensional numerical models for predicting quasi three-dimensional subsurface LNAPL behavior. Parker, Lenhard and Lenhard, Parker [22,23] also modelled hysteretic capillary pressure-saturation-relative permeability relations for porous media containing LNAPL. They considered LNAPL entrapment by water imbibition saturation paths, which can occur wherever LNAPL is displaced by water. Parker et al. [24,25] modified the [21] vertically integrated constitutive relations to consider LNAPL entrapment by water and residual LNAPL following LNAPL drainage in the vadose zone. They developed equations for predicting the specific total, free, entrapped, and residual LNAPL saturations as well as for the LNAPL transmissivity of the free LNAPL. Because their equations only predicted the various specific LNAPL volumes, the distribution of the different LNAPL forms as a function of depth (elevation) was unknown. Later, Waddill et al. [26] developed equations for calculating the average entrapped and residual LNAPL saturations in the unsaturated and saturated zones. The average saturations were constant values over the respective elevations. Still later, Lenhard et al. [27] modified the hysteretic [22,23] constitutive relations to include elevation-specific residual LNAPL saturation. Charbeneau, Charbeneau et al. and Jeong, Charbeneau [28–30] utilized the earlier works of [21,24–26] in their predictive models. To provide insights concerning subsurface LNAPL behavior and test predictive models, investigators conducted small- and intermediate-scale laboratory experiments and field investigations [7,8,10,31–40]. The works involved entrapment of LNAPL, the effect of water table fluctuations, assessing LNAPL recoverability, transient effects, and other aspects. In addition, investigators conducted multidimensional, multiphase simulations considering free, residual, and entrapped LNAPL [28,30,41–43] to predict subsurface LNAPL behavior. Water 2020, 12, 2337 3 of 15

The purpose of this paper is to provide additional insights concerning the movement of LNAPL Waterinto monitoring2020, 12, x FOR wells PEER byREVIEW experimentally investigating LNAPL behavior subject to a fluctuating water3 of 16 table in both coarse- and fine-grained aquifer materials. We also consider effects of LNAPL viscosity. Although some investigators have considered considered effects effects of water water table table fluctuations fluctuations previously, previously, our focus focus isis on on characterizing characterizing transient transient aspects aspects of of changing changing LNAPL LNAPL thickness thickness in in monitoring monitoring wells. wells. We We measure measure thethe time-dependent time-dependent changes of monitoring well LNAP LNAPLL thickness in a laboratory 3D model. The data data presented can be employed to test multiphase flow flow numerical models. In In addition to the measured changes ofof waterwater and and NAPL NAPL elevations elevations in monitoringin monitori wellsng wells with timewith attime various at various positions, positions, we provide we provideLNAPL andLNAPL porous and medium porous properties, medium properties, initial conditions, initial andconditions, boundary and conditions. boundary Modelers conditions. can Modelersutilize this can wealth utilize of datathis towealth conduct of data simulations to conduct with simulations multiphase with flow numericalmultiphase models. flow numerical Relevant models.van Genuchten Relevant or van Brooks-Corey Genuchten parameters or Brooks-Corey can be parameters found in the can literature. be found in the literature.

2. Material Material and and Methods Methods

2.1. Tank Tank Description The experiment experiment involved involved a glass a glass tank tank with with appr approximateoximate inner inner dimensions dimensions of 50 cm of 50high, cm 70 high, cm wide,70 cm and wide 50, andcm deep 50 cm (see deep Figure (see Figure1). The1). packed The packed aquifer aquifer material material had approximate had approximate dimensions dimensions of 40 cmof 40 high, cm high50 cm, 50 wide, cm wide, and 50 and cm 50 deep cm deep(i.e., the (i.e., aquife the aquiferr material material packing packing was to was a 40 to cm a 40 height cm height with thewith inlet the water inlet waterchamber chamber being 12.5 being cm 12.5 wide cm and wide the and outlet the water outlet chamber water chamber being 7.5 being cm wide). 7.5 cm A wide).water pumpingA water pumping well was welllocated was at located the center at the of the center tank of and the six tank observation and six observation wells were wells located, were aslocated, shown inas Figure shown 1. in All Figure wells1. Allpenetrated wells penetrated the aquifer the material aquifers materials to the base to theof the base tank, of the had tank, a diameter had a diameter of 2 cm, andof 2 cm,were and screened were screened 35 cm 35to cmprevent to prevent entry entry of aq ofuifer aquifer material. material. Screening Screening also also prevented prevented aquifer aquifer material from entering the inlet and outlet flowflow chambers of the tank.

Figure 1. TopTop view of wells and inlet and outlet chamberschambers of the tank. Distances of the monitoring wells from the pumping well (PW) are given in millimeters (mm). Adjustable floats controlled the water levels in the inlet chamber, thereby manipulating the water Adjustable floats controlled the water levels in the inlet chamber, thereby manipulating the flow rate through the sand. A 1-cm diameter drainage tube was used to control the water level in water flow rate through the sand. A 1-cm diameter drainage tube was used to control the water level the outlet chamber. Further, an adjustable pump controlled pumping rates from the pumping well. in the outlet chamber. Further, an adjustable pump controlled pumping rates from the pumping well. To measure LNAPL (light non-aqueous phase liquid) thickness in the wells, a HERON Oil-Water To measure LNAPL (light non-aqueous phase liquid) thickness in the wells, a HERON Oil-Water Interface meter was used, which required the wells to have the 2 cm diameter. The interface meter Interface meter was used, which required the wells to have the 2 cm diameter. The interface meter measures air-water, air-LNAPL and LNAPL-water interface elevations with a precision of 0.1 cm. measures air-water, air-LNAPL and LNAPL-water interface elevations with a precision of 0.1 cm. A drop in water level near the pumping well was accomplished by pumping from the well. During the pumping period, the valve to the inlet chamber was kept open to maintain a constant water level in the inlet chamber with aid of the float. To restore the water level near the pumping

Water 2020, 12, 2337 4 of 15

A drop in water level near the pumping well was accomplished by pumping from the well. During the pumping period, the valve to the inlet chamber was kept open to maintain a constant water level in the inlet chamber with aid of the float. To restore the water level near the pumping well to its original level, pumping was stopped and the water flow from the inlet chamber was maintained.

2.2. Aquifer Material We used glass beads as aquifer material to minimize effects of potential heterogeneities and anisotropy from packing the tank. Further, aquifer material grain size is more easily controlled with using glass beads. We used glass beads with a size range of 400 to 750 µm in diameter, for what we refer to as our coarse-grained aquifer material, and glass beads with a size range of <125 µm in diameter, for what we refer to as our fine-grained aquifer material. During our experiments with fine-grained aquifer material, we used glass beads with a size range of 400 to 750 µm as a ‘gravel pack’ around the wells. The water saturated hydraulic conductivities for the coarse- and fine-grained aquifer materials are 2.5 and 0.79 m/day, respectively. The dry bulk densities for the coarse- and fine-grained aquifer materials are 1620 and 1400 kg/m3, respectively.

2.3. LNAPL Properties In order to study the effect of viscosity, we employed two different LNAPLs (crude oil and gas oil). The physical properties of these LNAPLs are listed in Table1. The crude oil is Iranian Crude Oil and the gas oil is a refined product from the Shazand Refinery in central Iran. In order to visualize the gas oil in our experiments, we mixed the gas oil with “Blue Dispers BL” dye made by Alwan Company in Iran.

Table 1. Physical properties of LNAPLs.

LNAPL API Viscosity (Pa s) Specific Weight (kg/m3) · 0.0230 at 10 C Crude Oil 31.5 ◦ 868.2 0.0140 at 20 ◦C Gas Oil 39.2 0.0017 at 50 ◦C 829.0

2.4. Injection of LNAPLs LNAPL was injected into the aquifer materials between the observation well M1 and the pumping well (PW) (see Figure1) with a 1-cm diameter tube. We used di fferent protocols for the LNAPL injection based on the aquifer material. For the coarse-grained aquifer material, we injected the LNAPLs continuously at about 5 cm below aquifer material surface at a pressure head of approximately 10 cm. The LNAPLs migrated downwards spreading over and depressing the water capillary fringe. For the fine-grained aquifer material, we injected the LNAPL at the water table at a pressure head of approximately 25 cm. The volume of LNAPL injected in all cases was approximately 1500 cm3 (1.5 L).

3. Experiments We performed three different experiments to study the behavior of LNAPL (light non-aqueous phase liquid) thickness in observation wells (Table2). Two main factors were LNAPL type and grain size of the aquifer materials.

Table 2. Experiments.

Exp. No. LNAPL Grain Size (µm) E1 Gas Oil Coarse grain (400–750) E2 Crude Oil Coarse grain (400–750) E3 Gas Oil Fine grain (<125) Water 2020, 12, 2337 5 of 15 Water 2020, 12, x FOR PEER REVIEW 5 of 16

During the experiments, we measured the elevations of the air-LNAPL and LNAPL-water interfaces inin thethe observation observation wells. wells. For For ease ease of of following following discussions, discussions, we employwe employ abbreviations abbreviations to refer to refer to the thickness of LNAPL in the aquifer materials and wells, which are shown in Figure 2. to the thickness of LNAPL in the aquifer materials and wells, which are shown in Figure2.

Figure 2. AbbreviationsAbbreviations used for elevations and thickne thicknessessses of LNAPL and fluid fluid interfaces, where Ho is the thicknessthickness of LNAPLLNAPL contamination in the subsurface,subsurface, Hao is the elevation of the air-LNAPL interfaces inin thethe subsurfacesubsurface above above a a datum, datum, How How is is the the elevation elevation of theof the LNAPL-water LNAPL-water interfaces interfaces in the in subsurfacethe subsurface above above a datum, a datum, Haw Haw is theis the elevation elevation of of the the water water table table above above a a datum, datum, ZoZo isis thethe LNAPL thickness in a monitoring well, Zao is the elevationelevation of the air-LNAPL interfaceinterface in a monitoringmonitoring well above aa datum,datum, Zow Zow is is the the elevation elevation of theof LNAPL-waterthe LNAPL-water interface interface in a monitoring in a monitoring well above well aabove datum. a 3.1. Experimentdatum. E1: Coarse Grain, Gas Oil

3.1. ExperimentPrior to LNAPL E1: Coarse injection, Grain, the Gas water Oil elevations in the inlet and outlet chambers were set at 23.1 and 22.1 cm, respectively. LNAPL injection was initiated after steady-state water flow conditions were Prior to LNAPL injection, the water elevations in the inlet and outlet chambers were set at 23.1 established. The flow of water through the packed aquifer material was approximately 1.25 10 6 m3/s. and 22.1 cm, respectively. LNAPL injection was initiated after steady-state water flow× conditions− 24 h after LNAPL injection was terminated, water was pumped from the pumping well at a rate of were established.6 3 The flow of water through the packed aquifer material was approximately 1.25 × 5.6 10− m /s for 102 min to cause a rapid drop in the groundwater table. After LNAPL injection 10−6× m3/s. 24 h after LNAPL injection was terminated, water was pumped from the pumping well at and before water pumping, some LNAPL passed from the aquifer material into both the inlet and a rate of 5.6 10 m3/s for 102 min to cause a rapid drop in the groundwater table. After LNAPL outlet chambers. There was a greater amount of LNAPL in the outlet chamber than the inlet chamber, injection and before water pumping, some LNAPL passed from the aquifer material into both the which only had a minor amount. After water pumping was terminated, the groundwater level returned inlet and outlet chambers. There was a greater amount of LNAPL in the outlet chamber than the inlet to its original level. Air-LNAPL (Zao) and LNAPL-water (Zow) interfaces were measured in the chamber, which only had a minor amount. After water pumping was terminated, the groundwater monitoring wells. The change of LNAPL thickness (Zo) in the monitoring wells following the start level returned to its original level. Air-LNAPL (Zao) and LNAPL-water (Zow) interfaces were of water pumping is shown in Figure3. For a better understanding of how LNAPL thickness (Zo) measured in the monitoring wells. The change of LNAPL thickness (Zo) in the monitoring wells changed, Zao and Zow in the monitoring wells following the start of water pumping are shown in following the start of water pumping is shown in Figure 3. For a better understanding of how LNAPL Figures4 and5, respectively. Figure6 schematically shows Zo in each monitoring well before pumping, thickness (Zo) changed, Zao and Zow in the monitoring wells following the start of water pumping immediately after pumping, and at the end of test (experiment). are shown in Figures 4 and 5, respectively. Figure 6 schematically shows Zo in each monitoring well before pumping, immediately after pumping, and at the end of test (experiment).

Water 2020, 12, 2337 6 of 15 Water 2020,, 12,, xx FORFOR PEERPEER REVIEWREVIEW 6 of 16 Water 2020, 12, x FOR PEER REVIEW 6 of 16

FigureFigure 3. 3. LNAPLLNAPL LNAPL thickness thickness (Zo) (Zo) variations variations inin monitoringmonitoring wells wells for for E1. E1.

Figure 4. Air-LNAPL interface elevation (Zao) variations in E1. Figure 4. Air-LNAPLAir-LNAPL interface elevatio elevationn (Zao) variations in E1.

Figure 5. LNAPL-water interface elevation (Zow) variations in E1.

Figure 5. LNAPL-waterLNAPL-water interface elevation (Zow) variations in E1.

WaterWater 20202020,, 1212,, 2337x FOR PEER REVIEW 7 of 1516

Figure 6.6. LNAPLLNAPL thickness thickness (Zo) (Zo) in monitoringin monitoring wells wells at the at beginning the beginning of pumping of pumping (BP), end (BP), of pumping end of (EP),pumping and end(EP), of and test end (ET) of for test E1. (ET) for E1.

At the onset of pumping (see Figure3, time = 0), M1 and M6 had approximately the same Zo with At the onset of pumping (see Figure 3, time = 0), M1 and M6 had approximately the same Zo M3 having a slightly greater thickness, which suggests there may be some heterogeneity in the packing with M3 having a slightly greater thickness, which suggests there may be some heterogeneity in the because both M1 and M6 are closer to the LNAPL injection point than M3 (see Figure1). Transient packing because both M1 and M6 are closer to the LNAPL injection point than M3 (see Figure 1). conditions existed during the experiment and ideally M3 should have a smaller LNAPL thickness than Transient conditions existed during the experiment and ideally M3 should have a smaller LNAPL M1 and M6. Farther from the LNAPL injection point, M2 and M4 had similar LNAPL thicknesses, thickness than M1 and M6. Farther from the LNAPL injection point, M2 and M4 had similar LNAPL which is reasonable. The very small LNAPL thickness at M5 is also reasonable because it lies some thicknesses, which is reasonable. The very small LNAPL thickness at M5 is also reasonable because distance upstream of the injection point. The greater the LNAPL thickness in a well typically indicates it lies some distance upstream of the injection point. The greater the LNAPL thickness in a well thetypically LNAPL indicates saturation the willLNAPL be higher saturation in the will aquifer be hi nextgher toin itthe and aquifer some LNAPLnext to it will and be some contained LNAPL in largerwill be pore contained spaces. in larger pore spaces. During water pumping, the produced cone of depressiondepression isis expectedexpected toto bebe nonsymmetricalnonsymmetrical because of water flowflow fromfrom thethe inletinlet toto outletoutlet chambers.chambers. One would expect LNAPL to drain toward thethe bottombottom of of the the water water depression. depression. From From Figure Figure4, it shows 4, it Zao shows decreased Zao decreased similarly for similarly all monitoring for all wells,monitoring except wells, for M1 except which for did M1 not which decline did as not much decli asne the as other much monitoring as the other wells. monitoring The explanation wells. The for theexplanation M1 Zao behaviorfor the M1 is unclear,Zao behavior but may is unclear, be attributed but may to thebe nonsymmetricalattributed to the waternonsymmetrical depression water from pumping.depression From from Figure pumping.5, Zow From was Figure di fferent 5, Zow for each was monitoring different for well each because monitoring of the diwellfferent because Zo in of each the monitoringdifferent Zo well in each initially monitoring and the ewellffects initially from the an pumping.d the effects From from about the pumping. 50 min to theFrom end about of pumping, 50 min theto the rate end of declineof pumping, in Zow the was rate less of for decline M1, M3, in Zo andw M6,was whichless for are M1, the M3, closer and wells M6, towhich the pumping are the closer well, thanwells for to M2,the pumping M4, and M5.well, The than rate for of M2, decline M4, inand Zow M5. for The M3 rate and of M6, decline however, in Zow is slightly for M3 lessand thanM6, thehowever, decline is in slightly Zao, which less than means the Zodecline is decreasing in Zao, which for those means wells. Zo Theis decreasing rate of decline for those in Zow wells. for The M2, M4,rate andof decline M5, conversely, in Zow for is M2, slightly M4, greaterand M5, than converse the declinely, is slightly in Zao, whichgreater means than the Zo isdecline increasing in Zao, in thosewhich monitoring means Zo wells.is increasing Since the in ratethose of monitoring decline in Zao wells. for M1Since is eventhe rate less of than decline the rate in Zao in other for M1 wells is andeven the less rate than of the decline rate inin Zowother is wells less thanand the rate Zao of rate, decline Zo is in increasing. Zow is less Again, than thethe behaviorZao rate,at Zo M1 is isincreasing. unclear. TransientAgain, the fluid behavior flow isat complicated. M1 is unclear. Because Transient of the fluid rates flow at is which complicated. Zao and Because Zow declined of the duerates to at water which pumping, Zao and ZoZow slightly declined decreased due to water in M3 pumping, and M6, but Zo increased, slightly decreased sometimes in significantly,M3 and M6, inbut M1, increased, M2, M4, sometimes and M5 at thesignific terminationantly, in of M1, pumping. M2, M4, A and plausible M5 at explanation the termination for the of increase pumping. in Zo A atplausible the end explanation of pumping for the M2, increase M4, and in M5 Zo could at the be end the of migration pumping of for LNAPL M2, M4, from and the M5 inlet could and be outlet the chambers.migration of The LNAPL cone of from depression the inlet from and pumping outlet chambers. would produce The cone a gradient of depression whereby from LNAPL pumping under slightwould positive produce pressures a gradient in thewhereby chambers LNAPL would under migrate slight to positive the aquifer pressures material. in the chambers would migrateAfter to pumpingthe aquifer was material. terminated, Zao and Zow increased, as seen in Figures4 and5, respectively, indicatingAfter pumping rising water was table. terminated, During Zao pumping, and Zow LNAPL increase wasd, as migrating seen in Figures to the 4 centerand 5, respectively, of the water coneindicating of depression rising water at the table. pumping During well. pumping, Following LNAPL pumping, was migrating the mass to of the LNAPL center near of the the water pumping cone wellof depression will move at upwards the pumping and outwardswell. Following as the pumping, water table the rises. mass As of seenLNAPL in Figure near the5, the pumping increase well in Zowwill move appears upwards to be approaching and outwards a stable as the level water in alltable of therises. monitoring As seen in wells Figure when 5, the the increase experiment in Zow was terminated,appears to be but approaching Zao was still a increasing, stable level indicating in all of the the LNAPL monitoring was still wells redistributing when the experiment at the end of was the experiment.terminated, but Figure Zao3 wasshows still that increasing, the increase indicating in Zo in the the LNAPL monitoring was wellsstill redistributing is greater for at wells the closerend of tothe the experiment. pumping well.Figure At 3 theshows end that of the the experiment, increase in theZo LNAPLin the monitoring thickness inwells all monitoringis greater for wells wells is greatercloser to than thethat pumping at the onsetwell. ofAt pumping the end of (Figure the ex6).periment, The di fference the LNAPL in LNAPL thickness thicknesses in all inmonitoring the wells wells is greater than that at the onset of pumping (Figure 6). The difference in LNAPL thicknesses in

Water 2020, 12, 2337 8 of 15 Water 2020, 12, x FOR PEER REVIEW 8 of 16

the wells at the end of pumping and the end of the experiment may be also partly attributed to at the end of pumping and the end of the experiment may be also partly attributed to hysteresis in hysteresis in capillary pressure-saturation relations. capillary pressure-saturation relations.

3.2.3.2. ExperimentExperiment E2:E2: Coarse Grain, Crude Oil After repacking the tank with coarse-grained aquifer material and establishing water flow from AfterAfter repackingrepacking thethe tanktank withwith coarse-grained aquifer material material and and establishing establishing water water flow flow from from the inlet to outlet chambers, we injected crude oil as the LNAPL. Packing the tank and establishing thethe inlet inlet toto outletoutlet chambers,chambers, wewe injectedinjected crude oil as the LNAPL. Packing Packing the the tank tank and and establishing establishing water flow followed similar procedures as for E1. The water elevation in the inlet and outlet chambers water flow followed similar procedures as for E1. The water elevation in the inlet and outlet chambers were set at 30.3 cm and 29.5 cm, respectively. The flow of water through the packed aquifer material were set at 30.3 cm and 29.5 cm, respectively. The flow of water through the packed aquifer material −6 3 was approximately 1 × 10 6 m 3/s. Roughly 7 days after LNAPL injection, the water level was lowered was approximately 1 10− m /s. Roughly 7 days after LNAPL injection, the water level was lowered via water pumping from× the pumping well. Pumping continued for 110 min at a discharge rate of via water pumping from the pumping well. Pumping continued for 110 min at a discharge rate of 4 10 m3/s. The pump was then turned off so that the water levels could return to pre- 4 10 6 m3/s. The pump was then turned off so that the water levels could return to pre-pumping pumping× − elevations. At the beginning of pumping, M2, M4, and M5 did not have any LNAPL elevations. At the beginning of pumping, M2, M4, and M5 did not have any LNAPL because they were because they were farther from the injection point than M1, M3, and M6. In fact, no LNAPL entered farther from the injection point than M1, M3, and M6. In fact, no LNAPL entered M2, M4, and M5 at M2, M4, and M5 at any time during E2. The higher viscosity of the crude oil relative to the gas oil any time during E2. The higher viscosity of the crude oil relative to the gas oil caused the crude oil to caused the crude oil to migrate at a much slower rate than the gas oil. The LNAPL thicknesses (Zo) migrate at a much slower rate than the gas oil. The LNAPL thicknesses (Zo) in monitoring wells M1, in monitoring wells M1, M3, and M6 during the experiment (E2) are shown in Figure 7. Figures 8 and M3, and M6 during the experiment (E2) are shown in Figure7. Figures8 and9 show the elevation 9 show the elevation changes of Zao and Zow, respectively. In addition, we show Zo in M1, M3, and changes of Zao and Zow, respectively. In addition, we show Zo in M1, M3, and M6 before water M6 before water pumping, immediately after pumping, and at the end of the experiment in Figure pumping, immediately after pumping, and at the end of the experiment in Figure 10. 10.

FigureFigure 7.7. LNAPL thickness (Zo) variations in in monitoring monitoring wells wells for for E2. E2.

FigureFigure 8.8. Air-LNAPL interface elevatio elevationn (Zao) (Zao) variations variations in in E2. E2.

Water 2020, 12, 2337 9 of 15 Water 2020, 12, x FOR PEER REVIEW 9 of 16 Water 2020, 12, x FOR PEER REVIEW 9 of 16

FigureFigureFigure 9. 9. LNAPL-water9. LNAPL-water LNAPL-water interface interfaceinterface elevation (Zow)(Zow) (Zow) variations variations variations in in E2. inE2. E2.

Figure 10. LNAPL thickness (Zo) in monitoring wells at the beginning of pumping (BP), end of Figure 10. LNAPL thickness (Zo) in monitoring wells at the beginning of pumping (BP), end of pumping (EP), and end of test (ET) for E2. Figurepumping 10. LNAPL(EP), and thickness end of test (Zo) (ET) in for monitoring E2. wells at the beginning of pumping (BP), end of pumping (EP), and end of test (ET) for E2. Prior to pumping (Figure7, time = 0), LNAPL had entered only M1, M3, and M6. Although Prior to pumping (Figure 7, time = 0), LNAPL had entered only M1, M3, and M6. Although M1 M1 isis closer closerPrior tototo thepumpingthe LNAPLLNAPL (Figure injection 7, time point point = 0), than thanLNAPL M3 M3 and had and M6, entered M6, ZoZo is onlygreater is greater M1, in M3, M3 in and M3and M6. and M6. Although M6.A possible A possible M1 explanationis explanationcloser to is thatthe is LNAPLthat M1 isM1 upgradient isinjection upgradient point to theto thanthe water water M3 flow, andflow, whereasM6, whereas Zo is M3 M3 greater and and M6M6 in areareM3 downgradientand M6. A possible from from the injectionexplanationthe injection point is (Figure pointthat M1 (Figure1). is The upgradient 1). larger The larger Zo to in theZo M6 inwater M6 is likely is flow, likely duewhereas due to to the the M3 fact and that that M6 it it isare iscloser closerdowngradient to the to theinjection injection from pointthepoint thaninjection than M3. pointM3. Even Even (Figure after after 71). days7 The days larger of of LNAPL LNAPL Zo in redistribution,M6redist is likelyribution, due transient to transient the fact flow that flow conditions it conditionsis closer likelyto the likely existedinjection existed becausepointbecause ofthan the of M3. highthe Evenhigh LNAPL LNAPLafter viscosity.7 daysviscosity. of LNAPL redistribution, transient flow conditions likely existed becauseShortlyShortly of after the afterhigh water water LNAPL pumping pumping viscosity. from from the the pumping pumping well well started, started, Figures 8 and 9 show Zao Zao and and Zow Zow in M1, M3,in ShortlyM1, and M3, M6 after and rapidly water M6 decreasingrapidly pumping decreasing from due tothe thedue pumpinglowering to the welllowering of started, the waterof Figuresthe table. water 8 andThe table. 9 Zo show The in these ZaoZo in and monitoring these Zow monitoring wells (Figure 7), however, did not decline as rapidly because of the high LNAPL viscosity. wellsin (FigureM1, M3,7), and however, M6 rapidly did not decreasing decline as due rapidly to the because lowering of theof the high water LNAPL table. viscosity. The Zo The in these flow of monitoringThe flow wellsof LNAPL (Figure toward 7), however, the bottom did not of declinethe cone as of rapidly depression because was of slow. the high Toward LNAPL the viscosity. end of LNAPLpumping toward (near the 110 bottom min), ofthe the relatively cone of flat depression curvatures was in Figures slow. Toward 7–9 suggest the endLNAPL of pumping is moving (near 110The min), flow the of relatively LNAPL toward flat curvatures the bottom in Figuresof the cone7–9 suggest of depression LNAPL was is movingslow. Toward closely the toward end of the pumpingclosely toward(near 110 the bottommin), the of therelatively cone of flatdepression. curvatures At thein endFigures of pumping, 7–9 suggest there LNAPL was decreases is moving in bottomZo ofin theall monitoring cone of depression. wells because At thethe endLNAPL of pumping, is moving theretoward was the decreasespumping well. in Zo The in allrate monitoring of Zo closely toward the bottom of the cone of depression. At the end of pumping, there was decreases in wellsdecrease because varied the LNAPL among isthe moving three monitoring toward the wells pumping with the well. rate being The rate greater of Zo fordecrease wells with varied a higher among Zo in all monitoring wells because the LNAPL is moving toward the pumping well. The rate of Zo the threeZo initially. monitoring A higher wells Zo within a monitoring the rate being well greaterinfers a forlarger wells LNAPL with saturation a higher and Zo initially.some LNAPL A higher decrease varied among the three monitoring wells with the rate being greater for wells with a higher will be in larger pores, which will have greater mobility. The results are different than in E1 because ZoZo in ainitially. monitoring A higher well Zo infers in a amonitoring larger LNAPL well infers saturation a larger and LNAPL some LNAPLsaturation will and be some in larger LNAPL pores, of the higher LNAPL viscosity. In E1, some monitoring wells had a greater Zo at the end of pumping whichwill willbe inhave larger greater pores, mobility.which will The have results greater are mobility. different The than results in E1are because different of than the in higher E1 because LNAPL while others had less Zo. viscosity.of the higher In E1, LNAPL some monitoringviscosity. In E1, wells some had monito a greaterring wells Zo at had the a endgreater of pumpingZo at the end while of pumping others had lesswhile Zo. others had less Zo.

After pumping was terminated, Zao and Zow increased as water levels were returning to their pre-pumping elevations. Zow increased at a greater rate than Zao. Consequently, Zo continued to Water 2020, 12, 2337 10 of 15 decrease even after the termination of pumping for an extended period. Only toward the end of the experiment did Zo in the monitoring wells begin to increase. Because of the slow mobility of the LNAPL, not as much LNAPL accumulated at the bottom of the cone of depression during pumping. Therefore, when the water elevations were returning to their pre-pumping elevations, there was less LNAPL available to redistribute upwards and outwards from the pumping well. Together with the slowWater mobility 2020, 12, x of FOR the PEER LNAPL, REVIEW the increase in Zo did not occur until later in the experiment (Figure10 of 16 7) when the rate of Zow increases became less than the rate of Zao increases. The effect was greater for After pumping was terminated, Zao and Zow increased as water levels were returning to their M1 and M6 than M3 because M1 and M6 are closer to the pumping well than M3. Hence, there was no pre-pumping elevations. Zow increased at a greater rate than Zao. Consequently, Zo continued to increase in Zo from the end of pumping and the end of the experiment (Figure 10). decrease even after the termination of pumping for an extended period. Only toward the end of the 3.3.experiment Experiment did E3: Zo Fine in Grain,the monitoring Gas Oil wells begin to increase. Because of the slow mobility of the LNAPL, not as much LNAPL accumulated at the bottom of the cone of depression during pumping. Therefore,To investigate when the potential water changeselevations in were LNAPL returning behavior to their due topre-pumping differences inelevations, grain size there when was subjected less toLNAPL a fluctuating available water to redistribute table, we packed upwards the and tank outw withards the from fine-grained the pumping aquifer well. material. Together Packing with the the tankslow and mobility establishing of the waterLNAPL, flow the followed increase thein Zo same did proceduresnot occur until as forlater E1 in and the E2. experiment The water (Figure elevations 7) inwhen the inlet the andrate outletof Zow chamber increases were became set at less 26.4 than cm the and rate 22.8 of cm, Zao respectively. increases. The The effect flow was of water greater through for theM1 packed and M6 aquifer than M3 material because was M1 approximately and M6 are closer 1.7 to10 the8 pumpingm3/s. After well establishing than M3. Hence, steady there water was flow, × − LNAPLno increase (gas oil)in Zo was from injected the end into of thepumping aquifer and material. the end LNAPL of the experiment injection was (Figure diff erent10). than during E1 and E2 (see Material and Methods). The injection was directly on the water table with a 25 cm LNAPL pressure3.3. Experiment head at E3: the Fine injection Grain, point. Gas Oil About 7 days after LNAPL injection, we pumped water from the pumpingTo wellinvestigate for 34 minpotential at the changes rate of 3in LNAPL10 7 m3 /behaviors, which due is more to differences than an order in grain of magnitude size when less × − thansubjected E1 and to E2. a fluctuating Thereafter, water the pump table, waswe packed turned the off sotank water with could the fine-grained return to pre-pumping aquifer material. levels. ZoPacking in the monitoringthe tank and wells establishing during water the experiment flow followed is shown the same in Figure procedures 11. Zao as for and E1 Zow and areE2. shownThe inwater Figures elevations 12 and 13in ,the respectively. inlet and outlet Additionally, chamber we Zore in set wells at 26.4 is givencm and at 22.8 the beginningcm, respectively. of pumping, The endflow of pumping,of water through and end the of thepacked experiment aquiferin material Figure 14was. Data approximately for M2 are not1.7 shown× 10−8 m in3/s. the After figures becauseestablishing there wassteady no measurablewater flow, LNAPLLNAPL in(gas M2 oil) at anywas timeinjected during into the the experiment, aquifer material. possibly LNAPL because ofinjection packing was heterogeneities. different than during E1 and E2 (see Material and Methods). The injection was directly on Priorthe water to pumping table with (Figure a 25 cm11, LNAPL time = 0),pressure M1 had head the at largest the inje Zoction followed point. byAbout M6. 7 Bothdays ofafter these 310 3 monitoringLNAPL injection, wells were we pumped the closest water to the from injection the pumping point, withwell for M1 34 being min closerat the rate than of M6. Zo in M3,m/s, M4, which is more than an order of magnitude less than E1 and E2. Thereafter, the pump was turned off and M5 was relatively small, with less than 1 cm. M3 is closer to the injection point than M5, but M5 so water could return to pre-pumping levels. Zo in the monitoring wells during the experiment is had more Zo than M3. Further, M5 is upgradient to water flow from the injection point and M3 is shown in Figure 11. Zao and Zow are shown in Figures 12 and 13, respectively. Additionally, Zo in downgradient. In addition, M2 is closer to the injection point than M4, but M4 had more Zo than M2, wells is given at the beginning of pumping, end of pumping, and end of the experiment in Figure 14. which had no LNAPL during the experiment. So, there was likely some heterogeneities in the aquifer Data for M2 are not shown in the figures because there was no measurable LNAPL in M2 at any time materialduring packing.the experiment, possibly because of packing heterogeneities.

FigureFigure 11. 11.LNAPL LNAPL thicknessthickness (Zo) variations variations in in monitoring monitoring wells wells in inE3. E3.

WaterWater2020 2020, ,12 12,, 2337 x FOR PEER REVIEW 1111 of of 16 15 Water 2020, 12, x FOR PEER REVIEW 11 of 16

Figure 12. Air-LNAPL interface elevation (Zao) variations in E3. FigureFigure 12.12. Air-LNAPL interface elevation (Zao) (Zao) variations variations in in E3. E3.

Figure 13. LNAPL-water interface elevation (Zow) variations in E3. FigureFigure 13.13. LNAPL-water interface elevation (Zow) (Zow) variations variations in in E3. E3.

FigureFigure 14.14. LNAPLLNAPL thicknessthickness in monitoring wells at the beginning of of pumping pumping (BP), (BP), end end of of pumping pumping (EP),(EP),Figure and and 14. end end LNAPL of of testtest thickness (ET)(ET) forfor scenarioinscenario monitoring inin E3.E3. wells at the beginning of pumping (BP), end of pumping (EP), and end of test (ET) for scenario in E3. Immediately following pumping, Zao and Zow decreased rapidly (Figures 12 and 13, respectively). Prior to pumping (Figure 11, time = 0), M1 had the largest Zo followed by M6. Both of these The lowering of the water pressures from pumping caused the water elevation in the monitoring monitoringPrior to wells pumping were the(Figure closest 11, to time the injection= 0), M1 point,had the with largest M1 being Zo followed closer than by M6.M6. ZoBoth in M3,of these M4, wells to immediately drop, but the water table in the aquifer material did not respond as quickly andmonitoring M5 was wells relatively were small,the closest with to less the than injection 1 cm. point, M3 is withcloser M1 to beingthe injection closer than point M6. than Zo M5, in M3, but M4,M5 and M5 was relatively small, with less than 1 cm. M3 is closer to the injection point than M5, but M5

Water 2020, 12, 2337 12 of 15 because of its low permeability. With the drop of the water elevations in the monitoring wells (Zow), LNAPL in the wells (Zao) also dropped at the same rate. During the drop of Zow and Zao, however, Zo increased. The Zo increase is attributed to the packing (i.e., gravel pack) around the monitoring wells, which was the coarse-grained aquifer material. The permeability of the coarse-grained aquifer material is significantly higher than the permeability of the fine-grained aquifer material. Therefore, LNAPL in the packing around the monitoring wells was able to move into the monitoring wells much faster when Zao and Zow dropped than any LNAPL in the fine-grained aquifer material could move into the packing material. Consequently, immediately after the start of pumping, Zo increased when Zao and Zow dropped, but LNAPL did not come from the aquifer material—instead, the LNAPL came from the packing around the monitoring wells. After the initial rapid drop of Zao and Zow with the onset of pumping, Zao and Zow continued to drop at a slower pace (Figures 12 and 13, respectively). The rate of the Zao decline was generally greater than the rate of the Zow decline, which yielded Zo to decrease. From the onset of pumping to the end of pumping, Zo became less in all monitoring wells (Figure 11). In fact, Zo disappeared in M3 and M4, and there was only a very minor amount in M5. Even though the fine-grained aquifer material permeability was low, some LNAPL should have migrated toward the bottom of the cone of depression at the pumping well during pumping. Following termination of pumping, there was a rapid rise in Zao and Zow as the water table started to return to its initial elevations. As a result, there was a minor rapid Zo increase in M1 and M6. As both Zao and Zow rose at M5, the LNAPL in M5 entered the packing around the well, which previously was drained from LNAPL when Zao and Zow were decreasing in elevation. Because there was only a minor amount of LNAPL in M5 at the end of pumping, the upward movement of Zao and Zow in M5 caused all of the LNAPL to enter the packing around the well so no LNAPL was measured in the well at the end of the experiment. We presume the same phenomena occurred for M1 and M6, but both of them had larger amounts of LNAPL at the end of pumping. Further, we reason LNAPL that accumulated near the pumping well during pumping would move upwards and outwards as water levels raised from cessation of pumping. The nearest monitoring well to the pumping well would benefit first from the upward and outward movement of LNAPL from the pumping well. We contemplate that M6, the closest monitoring well to the pumping well, received an influx of LNAPL from this upward and outward movement so its Zo increased from the end of pumping to the end of the experiment, similar to what occurred in E1. M1 is farther from the pumping well and upgradient to the steady water flow. At the end of the experiment, we hypothesize M1 has yet to receive any influx of LNAPL; therefore, its Zo decreased from the end of pumping to the end of the experiment. At some point, if the experiment would have continued, we would expect some influx of LNAPL at M1 although it might not be as much as M6 because it is upgradient from the pumping well.

4. Discussion We presented experimental results providing insights concerning the movement of LNAPL (light non-aqueous phase liquid) into monitoring wells subject to varying water table elevations in both coarse- and fine-grained aquifer materials. We also considered the effects of LNAPL viscosity. We measured the time-dependent changes of monitoring well LNAPL thickness in a laboratory 3D model. In the case of a low viscosity LNAPL in coarse-grained aquifer material, initiating water drawdown via pumping caused LNAPL to migrate toward the bottom of the cone of depression. Any premature stopping of the pumping before significant LNAPL is removed could cause LNAPL to move upwards and outwards from rising water elevations. The result may be more LNAPL in monitoring wells than what may have been initially present. It is also possible that a larger area of LNAPL contamination could result because the gradient LNAPL movement may be greater than that which occurred during the initial contamination event. In another case with a high viscosity LNAPL, the results may be different if pumping is prematurely stopped. The rate of LNAPL movement, because of its high viscosity, would be slow toward the bottom of the cone of depression, which would yield less LNAPL Water 2020, 12, 2337 13 of 15 migrating to the bottom of the cone of depression. So, when water levels return to their pre-pumping elevations because pumping was prematurely stopped, monitoring wells may not receive an influx of LNAPL. In the third case with a low viscosity LNAPL in fine-grained aquifer material where packing of coarser material is used as packing around monitoring wells, the packing material can impact the amount of LNAPL entering and existing monitoring wells, which will be unrepresentative of LNAPL behavior in the aquifer. The results from the three cases suggest remediation engineers/practitioners should avoid prematurely stopping water drawdown operations to extract LNAPL or otherwise they may inadvertently expand the area of LNAPL contamination. The results also suggest that to fully understand the behavior and the amount of LNAPL in the subsurface the multiphase flow physics of materials used as packs for groundwater wells need to be known. The results additionally show that transient flow behavior may suggest potential flow behavior outcomes, but they will not be those corresponding to quasi-equilibrium conditions.

5. Conclusions The movement of LNAPL (light non-aqueous phase liquid) into monitoring wells, especially when subjected to changing water table elevations, is complex. To understand the LNAPL distribution in the subsurface, there are many factors to consider such as aquifer grain sizes, LNAPL properties, and materials used to construct the monitoring wells. It is seldom that equilibrium conditions will prevail in the field, so remediation engineers/practitioners need to understand how transient flow phenomena can affect what is observed/measured in monitoring wells and apply this knowledge for understanding existing subsurface conditions. The data presented in this manuscript can help to better understand transient multiphase flow and can be used to test and, possibly, improve predictive models. Except for parameters to predict multiphase flow (i.e., van Genuchten or Brooks-Corey parameters), we provide LNAPL and porous medium properties, initial conditions, and boundary conditions that modelers can utilize to conduct simulations with multiphase flow numerical models. Relevant van Genuchten or Brooks-Corey parameters can be found in the literature. The data are particularly useful in a data assimilation framework, when there is a significant uncertainty on model error, as discussed (e.g., [44,45]). Only by conducting experiments and comparing the results to predictive models can better and more accurate multiphase flow models be developed.

Author Contributions: R.A. planned and carried out the experiments and contributed to the interpretation of the results. A.V. conceived the original idea, planned and carried out the experiments, contributed to the interpretation of the results and supervised the project. R.J.L. contributed to the interpretation of the results, took the lead in writing the manuscript and assisted in supervising the project. S.M.H. contributed to the interpretation of the results and assisted in supervising the project. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by Iran National Science Foundation (INSF) grant number 97014255. Conflicts of Interest: The authors declare that there are no known conflicts of interest associated with this publication and confirm that the manuscript has been read and approved by all named authors. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.

References

1. ASTM E2531-06(2014). In Standard Guide for Development of Conceptual Site Models and Remediation Strategies for Light Nonaqueous-Phase Liquids Released to the Subsurface; ASTM International: West Conshohocken, PA, USA, 2014. 2. CL: AIRE. An Illustrated Handbook of Lnapl Transport and Fate in the Subsurface; CL: AIRE: London, UK, 2014; ISBN 978-1-905046-24-9. 3. Fetter, C.W.; Thomas, B.; Boving, T.B.; Kreamer, D. Contaminant Hydrogeology, 3rd ed.; Waveland Press: Long Grove, IL, USA, 1997; 647p. Water 2020, 12, 2337 14 of 15

4. Lenhard, R.J.; Sookhak Lari, K.; Rayner, J.L.; Davis, G.B. Evaluating an Analytical Model to Predict Subsurface LNAPL Distributions and Transmissivity from Current and Historic Fluid Levels in Groundwater Wells: Comparing Results to Numerical Simulations. Groundw. Monitor. Remed. 2018, 38, 75–84. [CrossRef] 5. US EPA. A Decision-Making Framework for Cleanup of Sites Impacted with Light Non-Aqueous Phase Liquids (LNAPL). EPA 542-R-04-011. 2004. Available online: https://www.epa.gov/remedytech/ decisionmaking-framework-cleanup-sites-impacted-light-non-aqueousphase-liquids-lnapl (accessed on 15 June 2020). 6. CRC CARE. A Practitioner’s Guide for the Analysis, Management and Remediation of LNAPL; CRC CARE Technical Report 34; CRC for Contamination Assessment and Remediation of the Environment: Adelaide, Australia, 2015. 7. Atteia, O.; Palmier, C.; Schäfer, G. On the influence of groundwater table fluctuations on oil thickness in a well related to an LNAPL contaminated aquifer. J. Contam. Hydrol. 2019, 223, 103476. [CrossRef][PubMed] 8. Gatsios, E.; García-Rincón, J.; Rayner, J.L.; McLaughlan, R.G.; Davis, G.B. LNAPL transmissivity as a remediation metric in complex sites under water table fluctuations. J. Environ. Manag. 2018, 215, 40–48. [CrossRef][PubMed] 9. Lenhard, R.J.; Rayner, J.L.; Davis, G.B. A practical tool for estimating subsurface LNAPL distributions and transmissivity using current and historical fluid levels in groundwater wells: Effects of entrapped and residual LNAPL. J. Contam. Hydrol. 2017, 205, 1–11. [CrossRef][PubMed] 10. Kemblowski, M.W.; Chiang, C.Y. Hydrocarbon Thickness Fluctuations in Monitoring Wells. Ground Water 1990, 28, 244–252. [CrossRef] 11. Ballestero, T.P.; Fiedler, F.R.; Kinner, N.E. An investigation of the relationship between actual and apparent gasoline thickness in a uniform sand aquifer. Ground Water 1994, 32, 708–718. [CrossRef] 12. De Pastrovich, T.L.; Baradat, Y.; Barthel, R.; Chiarelli, A.; Fussell, D.R. Protection of groundwater from oil pollution. In CONCAWE Report 3/79; CONCAWE: The Hague, The Netherlands, 1979; p. 61. 13. Hall, R.A.; Blake, S.B.; Champlin, S.C., Jr. Determination of hydrocarbon thicknesses in sediments using borehole data. In The Fourth National Symposium and Exposition on Aquifer Restoration and Ground Water Monitoring; National Ground Water Association: Dublin, OH, USA, 1984; pp. 300–304. 14. Hampton, D.R.; Miller, P.D.G. Laboratory Investigation of the Relationship between Actual and Apparent Product Thickness in Sands; National Ground Water Association: Dublin, OH, USA, 1988; pp. 157–181. 15. Schwille, F. Petroleum contamination of the subsoil: A hydrological problem. In The Joint Problems of the Oil and Water Industries: Proceedings of a Symposium Held at the Hotel Metropole, Brighton, UK, 18–20 January 1967; Institute of Petroleum: London, UK, 1967; pp. 23–54. 16. Van Dam, J. The migration of hydrocarbons in a water-bearing stratum. In The Joint Problems of the Oil and Water Industries: Proceedings of a Symposium; Institute of Petroleum: London, UK, 1967; pp. 55–96. 17. Zilliox, L.; Muntzer, P. Effects of hydrodynamic processes on the development of ground-water pollution: Study on physical models in a saturated porous medium. Prog. Water Technol. 1975, 7, 561–568. 18. Farr, A.M.; Houghtalen, R.J.; McWhorter, D.B. Volume estimation of light nonaqueous phase liquids in porous media. Ground Water 1990, 28, 48–56. [CrossRef] 19. Lenhard, R.J.; Parker, J.C. Estimation of free hydrocarbon volume from fluid levels in monitoring wells. Ground Water 1990, 28, 57–67. [CrossRef] 20. Parker, J.C.; Lenhard, R.J.; Kuppusamy, T. A parametric model for constitutive properties governing multiphase flow in porous media. Water Resour. Res. 1987, 23, 618–624. [CrossRef] 21. Parker, J.C.; Lenhard, R.J. Vertical integration of three-phase flow equations for analysis of light hydrocarbon plume movement. Trans. Porous Media 1989, 5, 187–206. [CrossRef] 22. Parker, J.C.; Lenhard, R.J. A model for hysteretic constitutive relations governing multiphase flow. 1. Saturation-pressure relations. Water Resour. Res. 1987, 23, 2187–2196. [CrossRef] 23. Lenhard, R.J.; Parker, J.C. A model for hysteretic constitutive relations governing multiphase flow. 2. Permeability-saturation relations. Water Resour. Res. 1987, 23, 2197–2206. [CrossRef] 24. Parker, J.C.; Kaluarachchi, J.J.; Kremesec, V.J.; Hockman, E.L. Modeling free product recovery at hydrocarbon spill sites. In Conference Petroleum Hydrocarbons and Organic Chemicals in Ground Water: Prevention, Detection and Restoration; National Water Well Association: Dublin, OH, USA, 1990; pp. 641–655. 25. Parker, J.C.; Zhu, J.L.; Johnson, T.G.; Kremesec, V.J.; Hockman, E.L. Modeling free product migration and recovery at hydrocarbon spill sites. Ground Water 1994, 32, 119–128. [CrossRef] Water 2020, 12, 2337 15 of 15

26. Waddill, D.W.; Parker, J.C. Simulated recovery of light, nonaqueous phase liquid from unconfined heterogeneous aquifers. Ground Water 1997, 35, 938–947. [CrossRef] 27. Lenhard, R.J.; Oostrom, M.; Dane, J.H. A constitutive model for air-NAPL-water flow in the vadose zone accounting for immobile, non-occluded (residual) NAPL in strongly water-wet porous media. J. Contam. Hydrol. 2004, 73, 283–304. [CrossRef] 28. Charbeneau, R.J. LNAPL distribution and recovery model (LDRM). In Volume 1: Distribution and Recovery of Petroleum Hydrocarbon Liquids in Porous Media; API Publ. No. 4760; American Petroleum Institute: Washington, DC, USA, 2007. 29. Charbeneau, R.J.; Johns, R.T.; Lake, R.W.; McAdams, M.J. Free-product recovery of petroleum hydrocarbon liquids. Ground. Water Monit. Rem. 2000, 20, 147–158. [CrossRef] 30. Jeong, J.; Charbeneau, R.J. An analytical model for predicting LNAPL distribution and recovery from multi-layered soils. J. Contam. Hydrol. 2014, 156, 52–61. [CrossRef] 31. Aral, M.M.; Liao, B. Effect of groundwater table fluctuations on LNAPL thickness in monitoring wells. Environ. Geol. 2002, 42, 151–161. [CrossRef] 32. Deska, I.; Ociepa, E. Impact of the Water Table Fluctuations on the Apparent Thickness of Light Non-Aqueous Phase Liquids. Ecol. Chem. Eng. A 2013, 20, 771–778. 33. Johnston, C.D.; Adamski, M. Relationship between Initial and Residual LNAPL Saturation for Different Soil Types; National Ground Water Association: Dublin, OH, USA, 2005; pp. 29–42. 34. Lenhard, R.J.; Parker, J.C. Measurement and prediction of saturation-pressure relationships in three phase-porous media systems. J. Contam. Hydrol. 1987, 1, 407–424. [CrossRef] 35. Liao, B.; Aral, M. Interpretation of LNAPL Thickness Measurements Under Unsteady Conditions. J. Hydrol. Eng. 1999, 4, 125–134. [CrossRef] 36. Marinelli, F.; Durnford, D.S. LNAPL thickness in monitoring wells considering hysteresis and entrapment. Ground Water 1996, 34, 405–414. [CrossRef] 37. Rayner, J.L.; Johnston, C.D.; Rao, P.S.C. Characterising residual NAPL using partitioning and interfacial tracers and implications for interphase mass transfer. In Contaminated Site Remediation: From Source Zones to Ecosystems, Proceedings of the 2000 Contaminated Site Remediation Conference, Melbourne, Australia, 4–8 December 2000; Johnston, C.D., Ed.; CSIRO Land and Water: Wembley, Australia, 2000; pp. 613–620. 38. Steffy, D.A.; Johnston, C.D.; Barry, D.A. A field study of the vertical immiscible displacement of LNAPL associated with a fluctuating water table. In Groundwater Quality: Remediation and Protection; IAHS: Wallingford, UK, 1995; pp. 49–57. 39. Steffy, D.A.; Johnston, C.D.; Barry, D.A. Numerical simulations and long-column tests of LNAPL displacement and trapping by a fluctuating water table. J. Soil Contam. 1998, 7, 325–356. [CrossRef] 40. Vaezihir, A.; Bayanlou, M.B.; Ahmadnezhad, Z.; Barzegari, G. Remediation of BTEX plume in a continuous flow model using zeolite-PRB. J. Contam. Hydrol. 2020, 230, 103604. [CrossRef] 41. Matos de Souza, M.; Oostrom, M.; White, M.D.; da Silva, G.C., Jr.; Barbosa, M.C. Simulation of subsurface multiphase contaminant extraction using a bioslurping well model. Trans. Porous Media 2016, 114, 649–673. [CrossRef] 42. Sookhak Lari, K.; Johnston, C.D.; Davis, G.B. Gasoline multi-phase and multicomponent partitioning in the vadose zone: Dynamics and risk longevity. Vadose Zone J. 2016, 15.[CrossRef] 43. White, M.D.; Oostrom, M.; Lenhard, R.J. Apractical model for mobile, residual and entrapped NAPL in water-wet porous media. Ground Water 2004, 42, 734–746. [CrossRef] 44. Berardi, M.; Andrisani, A.; Lopez, L.; Vurro, M. A new data assimilation technique based on ememble Kalman filter and Brownian bridges: An application to Richards’ equation. Comput. Phys. Commun. 2016, 208, 43–53. [CrossRef] 45. Zhang, Q.; Shi, L.; Holzman, M.; Ye, M.; Wang, Y.; Carmona, F.; Zha, Y. A dynamic data-driven method for dealing with model structural error in soil moisture data assimilation. Adv. Water Resour. 2019, 132, 103407. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).