R. A. Kloppl J. F. Haasbeekz P. B. Bedient3 A. A. Biehle4

Transcription

1 Advanced Technologies for Pollutant Detection, Monitoring, and Remediation in Ground Water R. A. Kloppl J. F. Haasbeekz P. B. Bedient3 A. A. Biehle4 Abstract Advances in technology for detection, monitoring, and remediation of hazardous waste constituents both in soil and ground water are rapidly changing. Conventional drilling and sampling for pollutant detection is giving way to sophisticated in situ technologies. One of the most promising is the use of cone penetration testing. The cone penetration test has been useful in the siting of waste disposal facilities and in the design of remedial action alternatives. Various types of cone penetrometers have been used to conduct a number of in situ tests where accurate information on soil stratigraphy and variability is essential for the consideration and evaluation of various hazardous waste disposal technologies. lntroductior, During site investigations, the owner/operator is interested in obtaining data toward the ultimate god of ground water rzmediation. The data of primary concern are the vertical and iateral extent of the contaminants and their respective concentrations, as well as subsurface geological data. Conventional sources available for the gathering of this data include soil borings, installation of monitoring wells, geophysical methods, and aquifer testing. One method of collecting site data is the cone penetrometer test (CPT). This method can be used to define such factors as sand geometry, hydraulic conductivities, detailed stratigraphy, and other soil properties. Once these data are collected, the remediation options can be examined to determine which is best suited for the project at hand. Equipment and test procedures The cone penetration test involves pushing a cone-shaped instrument into the soil and measuring its resistance to penetration. The two basic types of instruments are mechanical and electrical cone penetrometers. The eloctronic cone penetrometer is shown in Figure 1. As it penetrates, sensitive strain gauges transmit electronic measurements of resistance to penetration of the cone tip and friction sleeve to an automatic data acquisition system. Measurements of the lateral drift of the cone are also obtained during the sounding via a built-in inclinometer. The test was recently standardized by the American Society of Testing and Materials (ASTM, D ). An alternate design which incorporates a piezoelement in the cone tip is often used for environmental studies. The piezoelement has the ability to measure the pore water pressures in the soil that are developed during penetration. Ground water sampling cones (Figure 2) are also capable of monitoring pore pressure and are discussed in greater detail below. 1 Principle, Terra Technologies, Houston, Texas 2 Graduate Student, Rice University, Houston, Texas 3 Professor, Rice University, Houston, Texas 4 Consultant, Houston, Texas 67

2 Digital output of soil measurements are printed continuously during the sounding. All data are stored on a magnetic medium for future processing, and a graphical presentation of the data is immediately available for i n-fie Id stratigraphic correlation and evaluation. This feature -allows testing to be concentrated in critical areas. Field plots consist of cone tip resistance, sleeve friction resistance, friction ratio, pore pressure, and differential pore pressure ratio. The interpretive data are compatible with most personal computer systems for further data reduction, including stratigraphic cross sections and contouring. Ground water sampling cones are used to obtain selected samples of in situ gasses and liquids, retaining volatile components. Samples are encapsulated for easy and accurate lab analysis, mi n i mizi ng external co ntam i nation. The equipment is ideal for determination of the vertical and lateral extent of contamination. Tracer tests are often conducted where the system is used for the controlled injection of a tracer fluid. The spread of the fluid can then be observed by repeated sampling at various distances from the point of injection (Torstensson, 1984; Torstensson, et al., 1986). The major advantages of the electronic cone penetrometer over conventional sampling and testing are speed, continuous data measurements, economy, and reliability. A typical field program consists of an initial site survey with the cone penetrometer, followed by a limited number of soil borings. One of the disadvantages of CPT equipment is the lack of soil samples for testing. Although near-surface samples are easily obtainable via a sampling adapter, its depth capability is limited. The cone penetrometer will, however, define the oflimum depth and location for samples to be taken, thereby reducing the high cost normally associated with soil borings. With the incorporation of the piezoelement into the system, the cone penetrometer has become an accepted, cost effective tool for environmental site assessments. Additional information on the use of piezocone data including discussions on corrections for pore pressure effects and normalization of data can be found in Robertson, et al., Use of the Cone Penetrometer: Case History A large truck mounted electronic cone penetrometer was used to collect data at a site in the southwestern United States. Based on limited data from a few monitoring wells at the site, an extensive survey of the subsurface was conducted using the CPT equipment. Data from the survey included detailed stratigraphy, soil types and properties, estimates of hydraulic parameters such as transmissivities and potentiometric surfaces, and the vertical and lateral extent of contamination. These data were then used to simulate contaminant infiltration and transport at the site, and to begin design of a remedial strategy for the aquifer. The objectives of the modeling effort were to simulate the extent of the contaminant plume, and to examine the effectiveness of various remedial schemes. The CPT data showed the stratigraphy at the site to consist of an upper clay layer extending 30 feet (ft) below the ground surface. This clay was underlain by a sandy aquifer approximately 40 ft thick, which was in turn underlain by another clay layer of undetermined thickness. Transport Concepts - Upper Zone The upper clay at the site was relatively tight, with hydraulic conductivities averaging 10-6 centimeters per second (cmkec), and ranging from 63

3 10-5 to 10-9 cm/sec. The rates of vertical migration were governed mainly by rainfall and infiltration processes within the unsaturated and saturated zones of the upper clay unit, and were computed analytically. The average water table elevation in the clay was approximately 5 ft msl, whereas the average elevation of the piezometric surface in the sand below was approximately 3 ft msl. Darcy's equation was used to calculate velocity and travel time of contaminants through the upper clay to the lower sand. The computations showed that for average hydraulic conductivities at the site, contaminants may require from 22 to 220 years to reach the sandy aquifer. Once the contaminants in the clay travel to the base of the clay layer, they enter the sand, in which horizontal transport dominates. Transport Concepts - Lower Zone The sandy aquifer at the site comprised the lower zone for transport. The general direction of flow in this zone was horizontally to the west. Velocities in the sand, as indicated by observed perchlorethylene (PCE) values, appeared to be relatively low. Measured values indicated that the contaminants had migrated about 300 ft from the source area. In modeling this site, several transport mechanisms which may also affect the movement of the contaminants through the sands were not considered including adsorption, chemical reactions, and biodegradation. The exclusion of these mechanisms normally produces a conservative result. Description of the USGS Ground Water Model One of the most widely used twodimensional ground water transport models is by Konikow and Bredehoeft (Konikow, 1978). The USGS method of characteristics (MOC) model simulates solute transport in flowing ground water and can be applied to a wide range of problem types involving steady-state or transient flow. The model computes concentration changes over a grid with time caused by advection and mixing (dispersion) from fluid sources. The model allows the solute to adsorb or degrade linearly, and assumes that the gradients of fluid density, viscosity, and temperature do not effect the velocity distribution. However, the aquifer may be heterogeneous and/or anisotropic with variable pump rates or head values. The program uses an iterative alternating direction implicit (ADI) procedure to solve a finite-difference approximation to the ground water flow equation. After the head distribution is computed, the velocity of the ground water flow is computed for each node using an explicit finite-difference form of Darcy's equation. The model then uses a particletracking procedure to represent advective transport and a two-step explicit procedure to solve a finite-difference equation that describes the effects of hydrodynamic dispersion, fluid sources and sinks, and divergence of velocity. Input parameters for the model include: transmissivity, porosity, storage coefficient, longitudinal and transverse dispersivity, grid system and overall geometry of boundary conditions, locations of sources and sinks, wells, boundary conditions, and initial conditions for heads and background concentrations. The model output can include 2-D head distributions, the x and y velocity distributions, the drawdown depths, the concentration values at each grid location, and detailed breakthrough curves at selected observation wells. In order to analyze the model output, a graphical array package was used. This package, Biograph, is being developed at Rice University (Newell, 1988) as part of a Decision Support System for ground water 69

4 modeling. It displays numerical data over the model grid as patterns varying in dar kn ess : dar ke r patterns rep resent higher values. The program also allows the user to graph values versus time (breakthrough curves), or over a cross section across the grid. Many of the figures in this paper are printed output from this package. Initial Model Setup A 19 by 25 grid of cells 50 ft on a side was set up and located in an east-west direction. The source area is located toward the eastern end of the grid. The overall length and width of the grid were 1250 ft and 950 ft, respectively, giving a simulation area of approximately 27.3 acres. The grid setup is shown in Figure 3.. The hydraulic conductivity used in the base run simulation was cm/sec, based on aquifer tests and CPT data at the site, and was assumed to be constant and isotropic throughout the aquifer. The longitudinal and transverse dispersivity values used in the base run were selected from the literature and set to 10 ft and 1 ft, respectively. An early investigation of the effects of variable thickness showed little effect, so a constant sand thickness of 40 ft was used. Finally, the background concentrations in the region were assumed to be negligible. The source area was simulated in the model using injection wells in cells designated as source areas. The source cells used are shown in Figure 3. The leakage rate from the source into the aquifer was estimated using Darcy's law, and the extent of observed PCE. The final estimate was cubic feet per second (cfs), or 0.27 gallons per minute (GPM). The contaminant concentrations were calculated as a percentage of the source concentration, which was taken to be approximately 66,000 micrograms per liter (pg/i), based on observations in a monitoring well near the source and solubility data for the chemicals involved. The hydraulic conditions in the aquifer were assumed to be at steady state in all of the simulations presented here. This assumption greatly reduces run times, and seldom produces any effect on model results. The validity of this approach was verified using the model in a hydraulically transient mode. The period of time since the contaminants first reached the sand layer was difficult to determine. A simulation time of 5 years was chosen based on records of source activity, and on initial model results which indicated that approximately 5 years of contaminant movement were required to match observed data points. Sensitivity Analysis The model was run using the parameters described above to produce a base run to be used as a base line for comparison in the sensitivity analysis. Three additional runs were made in order to investigate the sensitivity of the contaminant plume to changes in model parameters. The results of these runs are compared to the base run in Table 1. The table shows that the model is most sensitive to the source leakage rate, and shows little response to changes in other para met e rs. Model Calibration The goals of the hydraulic calibration effort were to simulate the general direction and velocity of the ground water in the area of interest. Water level contours, plume extent, and the geologic nature of the aquifer system were all used to develop a satisfactory "match" between reality and simulation. The final calibration was achieved using constant head cells around the grid perimeter, and three additional cells representing local peaks and depressions 70

5 (Figure 3). The head values from the model output were entered into a contouring package, and the resulting contours were overlaid on the contour map created from observed head data. The final match presented a close correlation, as shown in Figure 4. The observed data for PCE were chosen as the data to which the contaminant transport results of the model would be compared. This was only a preliminary Cali bration process, as normally at least 6 to 10 observed data points are required within the plume to calibrate a 2-dimensional model. Measurements from one monitoring well and two CPT points were used for comparison. Assuming a source concentration of 66,000 pg/i, the observed PCE concentrations as percentages of the source concentration in three model cells would be: Cell o/o of source concentration 7, , , The results of the final model run are shown in Figure 5. The concentrations in the above three cells are 0.5%, O.8%, and 0% of source concentration, respectively. The run shows good areal agreement, and is satisfactory based on the available data. The parameters used in this final run were: Hydraulic Conductivity : cm/sec Sand Thickness : 4011 Leakage rate : cfs Time : 5 years Longitudinal Dispersivity : 5 ft In addition, the source area in the model was slightly reduced. The leakage of contaminants into cells (8,B) (9,B) and (9,lO) was eliminated, effectively reducing the source area by 7500 square feet. Remedial Design Criteria Based on the CPT data collected at the site, several remedial techniques were found to be infeasible because of the depth of the contamination. These included a collector trench for contaminant removal, and slurry walls or sheet piling for plume containment. It was determined that the installation of a recovery well field was the only economically feasible option. A recovery field may include the installation of extraction and injection wells: extraction wells to reverse the gradient, and injection wells to increase the gradient and expedite the remediation time. The number of wells and the pumping rates required for a 5-year cleanup operation were investigated by numerical modeling. Remedial Modeling - Basic Scenario As an initial design of the extraction well field, three pumping wells were simulated approximately 75 ft downgradient of the source area, spaced 150 ft apart. These locations in the model grid are shown in Figure 3. The wells were operated for 5 years. The results of operating these wells at various pumping rates are shown in Table 2. A longitudinal cross section of the plume through time for the run labeled "Low Rate 2" is shown in Figure 6. The table shows that the 3- well scheme is probably inadequate for a 5-year cleanup operation. Additional Wells Due to the lack of response from the 3- well system, four additional wells were placed in grid cells (8,9)(10,7) (12,5) and (1 1,8) and the pumping rate in each of the wells was increased to 3.5 GPM, which produced drawdowns of up to 20 ft. The source area was assumed to have been removed. After 5 years of pumping, the maximum concentration in the grid was reduced to 0.1 Yo, and the plume area was less than 0.3 acres. A longitudinal cross 71

6 section of the plume through time is shown in Figure 7. The plume after 5 years is shown in Figure 8. Conclusions After reviewing all available hydrogeology, CPT and monitoring well data.for the site, it was possible to successfully model the migration of contaminants in the sand. Although the calibration of the model was based on limited data, the results of the remedial design effort are indicative of the scale of the remedial systems to be examined. In addition, the modeling effort at the site has exposed several areas where data were inadequate, and has shown which design options must be investigated further. The availability of the CPT equipment has.proven extremely useful in the investigation of this site, and can be used to collect the additional data required in order to complete the design of the remediation scheme. In order to fully utilize the power of the CPT method, communication must be increased between modeling and data collection stages. After the initial site survey, the investigation should iterate between modeling and data collection steps, with each step directing the efforts of the next. This combination of two powerful technologies can provide cost savings through increased productivity and efficiency, and is a concept which applies in all engineering disciplines. Disclaimer 'Ihe work described in this paper was not funded by the U.S. Environmental Protection Agency. The contents do not necessarily reflect the views of the Agency and no official endorsement should be inferred. References ASTM, 1979, "Standard Method for Deep, Quasi-Static Cone and Friction-Cone Penetration Tests of Soi I", Designation 0344 I - 79, pp Konikow, L. F., and J. D. Bredehoeft, 1978, "Computer Model of Two-Dimensional Solute Transport and Dispersion in Ground Water", U. S. Geological Survey Techniques of Water Resources Investigation, Book 7, Chapter C2. Newell, C. J., P. B. Bedient, and J. F. Haasbeek, 1988, "Oasis: A Graphical Hypertext Decision Support System for Ground Water Contaminant Modeling", Submitted for Publication Robertson, P. K., R. G. Campanella, D. Gillespie, and J. Greig, 1986, "Use of Piezometer Cone Data", ASCE Specialty Conference on the Use of In-Situ Tests in Geotechnical Engineering, pp Torstensson, B., 1984, "A New System for Ground Water Monitoring", Publication of BAT Envitech, Inc., pp Torstensson, B., and A. M. Petsonk, 1986, "A Hermitically Isolated Sampling Method for Ground Water I n v es t i g at i on s", A S TM Symposium on Field Methods for Ground Water Contamination Studies and their Standardization, Cocoa Beach, Florida. 72

7 Table 1. Sensitivity Analysis Run Number Parameter Changed (units) (base run value) Maximum Concentration Plume Area (modified value) ("10) (Acres) 1 Hydraulic Conductivity (cm/sec) Source Leakage Rate (cfs) Longitudinal Dispersivity (ft) a Table 2. 3-Well Remedial Scenario 'Run Source Pumping Maximum Plume Description Removed Rate Concentration Area (GPM) (%) (Acres) No Action 1 No No Action 2 Yes High Rate No Low Rate 1 No Low Rate 2 Yes Figure 1 : Electric Friction Cone (after ASTM) 73

8 North Figure 2: Ground Water Sampling Probe (courtesy of BAT Envitech) Legend 300' feet Source Cell Constant Head Cell 0 Pumping Well Location Figure 3: Grid Setup I I Legend 300 feet Thick contours based on computed values Thin contours based on observed values Contour values in ft msl Contour interval ft Figure 4: Head Calibration 74

9 - - s Y C.- 0 E c. c) C a 0 c s Direction of Flow n V 0 East - West Distance (ft) feet Figure 5: Calibrated Plume Figure 6: Cross Section through Plume - 3-Well Pumping Scenario (run description: Low Rate 2) h c c Q, 0 c 0 Direction of Flow n U East - West Distance (ft) feet Figure 7: Cross Section through Plume - 7-Well Pumping Scenario Figure 8: 7-Well Pumping Scenario - Plume after 5 Years 75

10 USE OF ABANDONED COAL/LIGNITE OPEN PITS FOR WASTE DISPOSAL IN SELECTED EUROPEAN COUNTRIES Jacek S. Libicki Central Research and Design Institute for Surface Mining POLTEGOR Powstancow S1. 95, Wroclaw, Poland ABSTRACT The use of aban coal/l ignite pits as sal sites for solid waste appears to be a reas roblem, especially if they are located close to t te source. How a potential for groundwater and soil pollution exi This issue was sed by a "Group of Experts on Opencast Mining of the or Europe" because most of the sites are operated by mini paper contains the major topics of discussion includ he problem, legal aspects, characteristics of t ended for disposal, investigations required to obtain a sposal techniques, protection measures, monitoring ts, and research trends. A few countries are used as examples. INTRODUCTION discussion was to compare legision and regulations, current pro- posal sites. Their adv that an excavated area no new land would be d the waste can be pla original land level even be improved) fugitive dust is disadvantages a be in contact or leachate f hus result in e use of abandoned pits for disposal s of waste was dis- ni ted Kingdom (UK), Pol and, open pits for w disposal is subjected to diffe laws and regula- poisonous, noxious, or substances which can give rise to an 76