Loading...
HomeMy WebLinkAboutDRC-2009-006470 - 0901a0688014f6a8JON M h ."SMAN. JR GARY HERBERT Lieutenant Coventor State of Utah Department of Environmental Quality Richard W. Sprott Ejieciitive Director DIVISION OF RADIATION CONTROL Dant L. Finerfrock Director MEMORANDUM TO: Loren Morton, P.G., Manager FROM: Tom Rushing, P.G., Hydrogeologist DATE: January 13,2009 SUBJECT; Infiltration and Contaminant Transport Modeling Report, White Mesa Mill Site, Blanding, UT, Received November 21, 2007: DRC Review Memo Requirements Ground Water Permit No. UGW370004, Parts I.H.l J (Infiltration and Contaminant Transport Modeling Work Plan and Report), Part I.D.6 (Closed Cell Perfonnance Requirements) and Part I.C.I (Pemiit Limits), require Denison Mines U.S.A. to prepare an Infiltration and Contaminant Transport Model (ICTM.J Partl.H.U "11. Infiltration and Contaminant Transport Modeling Work. Plan and Report - the Permittee shall submit for Execulive Secretary approval an infiltration and contaminant transport modeling report that demonstrates the long-term ability of the tailings cells cover system to adequately contain and control tailings contaminants and protect nearby groundwater quality of the uppermost aquifer. Said report shall demonstrate how the tailings cell engineering design and specifications will comply with the minimum performance requirements of Part I.D.6 of this Permit. Within 180 days of Permit issuance, the Permittee shall submit a work plan for Executive Secretary approval, that; a) Identifies all applicable and pertinent historic studies and modeling reports relevant to tailings cell cover design and tailings cell system performance. b) Determines all information necessary for infiltration and contaminant transport modeling, including but not limited to representative input values for vadose zone and aquifer soil-water partitioning (Kd) coefficients, tailings source term concentrations, tailings waste leach rates, vadose zone and aquifer groundwater velocities, vadose zone and aquifer dispersivity, contaminant half-life or other rates of decay, etc. In the event 168 Nonh ! 050 West • PO Box 144850-Sail LjikeCit>. UT 84114-48.S0 • phone (801) 536-4250 • fax (801) 533 4097 TDD iZm) 5?.<}-AA\A ' ^viv^ .deq.umh.pfv Prtnicd on 1001. recycled p;ipei Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modehng Report Review Memo Page 2 that any required information is not currently available, the Permittee may select conservative assumptions for use in the required infiltration and contaminant transport models. Otherwise, the Permittee shall identify how information will be collected that is representative of actual field conditions, and a timetable by which said information will be submitted for Executive Secretary approval. c) Identifies all computer models that will be used to simulate long-term performance of the tailings cells cover system. All predictive models used shall be publicly available computer codes that adequately represent field characteristics at the tailings disposal site. d) Determines the conceptual model to be used and justifies why it is representative or conservative of actual field conditions at the site. Said concepmal model will identify the physical domain(s) and geometries to be simulated including the tailings cell design and construction, al) boundary and initial conditions to be assigned in the model(s), and the shallow aquifer locations where future potential contaminant concentrations will be predicted. e) Justifies how the infiltration and contaminant transport problem has been adequately conceptualized and planned to demonstrate compliance with the requirements of Part LD.6 of this Permit. Within 180 days after approval of the modehng work plan, the Permittee shall complete all modeling in accordance with the approved work plan and submit a final report for Executive Secretary approval. In the final report, the Permittee may include supplemental information to justify modification of certain Permit requirements, including, but not limited to: the ntmiber and types of groundwater compliance monitoring parameters, tailings cell cover system engineering design and construction specifications, tailings cell operational requirements, etc. Upon Executive Secretary approval of the fmal infiltration and contaminant Iransport report, the Reclamation Plan may be modified to accommodate necessary changes to protect public health and the environment." Pan I.D.6. *'6. Closed Cell Perfonnance Requirements - before reclamation and closure ofany tailings disposal cell, the Permittee shall ensure that the fmal design, construction, and operation of the cover system at each tailings cell will comply with all requirements of an approved Reclamation Plan, and will for a period of not less than 200 years meet the following minimum performance requirements: a) Minimize infiltration of precipitation or other surface water into the tailings, including, but not limited to the radon barrier, and b) Prevent the accumulation of leachate head within the tailings waste layer that could rise above or over-lop the maximiun FML hner elevation intemal to any disposal cell, i.e. create a "bathtub" effect. c) Ensure that groundwater quality at the compliance monitoring wells does not exceed the Ground Water Quality Standards or Oound Water Compliance Limits specified in Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 3 Part LCI and Table 2 of this Pennit. The Denison Mines USA ICTM for the White Mesa Mill Site in Blanding Utah was received by DRC on November 21, 2007 (Preparation Date November, 2007.) This memo summarizes the . review the plan." CeDl . White Mesa cell 1 is not included in the ICTM by reasoning that the cell will not contain tailings upon closure. The ICTM briefly summarizes some portions of the closure phasing of this cell which will include removal of evaporate crystals from current use as a dewatering cell and placement and cover of some demolition materials at the mill. Per Utah Administrative Code R313 and R317-6, Cell 1 falls under the requirements for modeling since wastes placed there will be contaminated with byproduct material. Alternatively, White Mesa needs to show that the wastes will have a de-minimus effect on ground water quality. Therefore, there has been a large oversight by Denison Mines U.S.A. in not including cell 1 in the modehng effort or alternatively showing cause for its omission. Overview The ICTM uses two flow modeling programs; HYDRUS ID for cover infiltration and unsaturated' flow, and MODFLOW for dewatering sinks in the^saturated tailings and flow modeling in the underlying aquifer. The ICTM additionally uses the contaminant transport model, MT3DMS (associated with MODFLOW), for chemical transport through saturated zones. PHREEQC is used as an associated model with HYDRUS 1-D to characterize geochemical process in the vadose zone (e.g. chemical reactions of leachate with the underlying soil matrix to estimate the Kd input values for Hydrus.) The tailing cell model design is highly dependent on large evapotranspiration (ET) sinks in the cover infiltrarion model to calculate the water balance and eventually achieve steady state for cells 2, 3 4A and 4B. Additionally, the model attempts to simulate long term dewatering of the tailings (14 year transient state) in cells 2 and 3 and associated pressure head drops within the cells and a shorter duration dewatering phase in cells 4A and 4B. Table 1. below summarizes the uses of MODFLOW and PTfDRUS-lD through the vertical domain ofthe model. Additionally, an accompanying flowchart is provided to better visualize the inter-dependency ofthe independent models. Per the flowchart, 5 distinct models have been used to characterize the flow and contaminant transport. Denison Mines White Mesa Uranium MiJi Infiltration and Contaminant Transport Modeling Report Review Memo Page 4 Table I. - Modeling Platforms Used throug Model No. 1 ' 2 3 4 5 Descriptive Depth in Domain Encompasses 7.5 ft thick cover system Encompasses saturated tailings zone to the top of the PVC or HDPE membrane. Geomembrane Liner Flux Encompasses the 3 Vadose Zones of the Dakota Fm. SS and Burro Canyon Fm. Encompasses the zone from flux out of the cover system to flux into the saturated Burro Canyon Fm. Model Platform *HYDRUS ID-Cover System MODFLOW-Tailings Dewatering Model, Calculates time to reach pseudo steady state head in the tailings cells Giroud & Bonaparte Anaiytical Equations PHRFRQC-Chemical Speciation and Complexation Model Used to Develop Kd and Retardation Coefficients in the Vadose Zone *HYDRUSID- Comprehensive Model Incorporates Inputs from Models 1, 2, and 3 to calculate flow and • chemical transport to the water table 1 the Vertical Domain Model Type PET Water Balance Dewatering Models Flux through the HDPE. (Cells 4A and 4B) and PVC (CeUs 2 and 3) Geomembrane Liners Ion Balance/ Chemical Speciation (geochemical . "beaker*' model) Flow and Chemical Transport—Solves for the Richards Equadon using modified VanGenuchten Fitting Parameters and numerical solutions for heat and solute transport Major Model Inputs Precipitation Evapotranspiration Infiltration Water Storage Loam and Clay Inputs (Cover Materials Specs.) Elevation Head Dependent Sink Term (Slimes Drain) Based on Manufacturing and Installation Defects and Decomposition of the Geomembrane Material. PIFO and Acid Neutralizing Data from Core MW-23 and MW-30 Analysis Dzombek and More! Diffuse Layer Ion Exchange Parameters -Model Layer Thicknesses -Residual Water Contents by Layer -Inverse Air Entry -Pore Size Distribution -Saturated K values -PVC and HDPE Membrane Inputs as Saturated K Values Parameters Predicted by Model Flux through cover into tailings zone (infiltration rate through the cover) Time lo reach pseudo steady state head in tailings cells after cover consfruction Flux through the bottom of the tailings cells. Vadose Zone K^ and Retardation Coefficients*** Water Balance Flux and Moisture Distribution** Breakthrough Curve ~ Uranium Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 5 6 Saturated ButTO Canyon Fm. (Horizontal aquifer flow &. transport) MODFLOW - Perched water table flow modeling. Burro Canyon Aquifer MT3DMS - Chemica] transport within the BuiTO Canyon saturated zone MODFLOW - Modular three- dimensional finite difference flow model MT3DMS - Multiple Contaminant Transport -Finite Difference Grid constant 50 ft spacing -Parameters for Saturated Flow Single Layer based on the perched aquifer -Hydrauhc Boundary Conditions -Hydraulic Head Distribution -Advective and Dispersive Processes -Breakthrough Curves - Chloride and Sulfate *Hydrus is not incorporated until pseudo steady stale conditions have been obiained Ci-c. - 3' head in cells 2 and 3 and 1' head in cells 4A and4B) **Flux calculaiions (HYDRUSJ arc based on pseudo steady state heads on the liner, 3" head in cells 2 and 3 and 1' head in cells 4A and 4B •"'Reiaidation coefficiem calculated using PHREEQC predicted Kd and Hydrus (Model 5) predicted moisture content. Flowchart Description of Models Used Model 1 — Conceptual Model for the Vegetated Cover Utilizing Hydrus ]-D, Conceptual Model Calculates Flux through the Cover Model 2 - Time to reach pseudo-steady state conditions in the tailings which equates to 3 ft of saturated conditions in cells 2 and 3 and 1 ft of saturated condidons in ceUs 4A and 4B using MODFLOW, Once Pseudo Steady State is achieved time 0 is set for further modeling Model 3 - Giroud & Bonapane analytical equations (predicts steady state infiltration through the geomembrane liners) Model 4 - PHREEQC Chemical Complexation (Beaker Chemistry) is modeled for 3 vadose zones using core data (inputs for neutralization potential and total HFO quantities) and existing tailings data. Model is limited to single ion reactions. Kd and Retardation coefficients are calculated using chemical speciation data in the PHREEQC Program for input into HYDRUS Model 5 Model 5 - Hydrus 1-D model is used for the modeling domain frora flux through the cover to fiux into the samrated Burro Canyon Formation, Values.for Model Layer Thicknesses, Residual Water Contents, Inverse Air Entry, Pore Size Distribution and Saturated Hydraulic Conductivity are entered into Hydrus Model 6 - MODFLOW and MT3DMS are used to model horizontal flow and transport through the saturated zone (advection and dispersion.) Chloride and Sulfate are modeled based on the findings of Model 3, no other ions are considered in the saturated zone model Denison Mines While Mesa Uranium Mil! Infiltration and Contaminant Transport Modeling Report Review Memo Page 6 Cover Design Discussioii The cover design presented in the plan is described as a shght modification to the NRC approved cover as detailed in the White Mesa Mill Reclamation Plan, Docket No. 40-8681, Revision 3, July 2000. Both top slope and side slope designs are detailed on Figure A-5.1-2 of the Reclamation Plan, "Reclamation Cover and Cross Sections." The ICTM discusses a change in philosophy at the White Mesa Mill. The current top-slope design standard, as outlined in the White Mesa Reclamation Plan includes a cobble layer cover system underlain by a frost barrier/water storage layer of 2 ft. thickness. The conceptual idea of DUSA is to alter this previously approved (NRC) cover design to a vegetated cover which will provide a larger ET sink term in the model. Note; the side slope design has not been altered and is not considered significant due to the relatively flat top-slope (0.2 %.) DRC did note that the side- slope is not considered in the HYDRUS 1-D model and based on the assumption of no runoff or intemal filter drainage from the top-slope area due to the relatively flat top-slope. The side-slope geometry is basically the same as the top slope design in the Reclamation Plan and has not been considered or redesigned in associarion with the ICTM. As discussed below, the assumption of no runoff or intemal filter drainage from the top-slope area may not be reasonable and may reflect a weakness in the selection of a one-dimensional model to simulate cover infiltiration performance. The top-slope cover designs listed in the Reclamarion Plan and ICTTM are summarized by layers in Table 2. below: Table 2. NRC Approved Tailings Cell Top-Slope Cover Design and ICTM Proposed Design by Layer Current Tailings Cell Cover, Reclamation Plan, Revision 3 Detail, Top to Bottom Tailings Cell Cover Proposed TCTM 6" Thick Riprap Layer (consisting of rock with a D50 minimum of 0.3 inches) Top Soil with Gravel (Vegetated with Grasses) 6", Saturated K = 6.3 cm/day (Per Table 3-1 ICTM) Frost Barrier, 2 ft 95 % compacrion, Average -7 Hydraulic Conductivity = 8.87 x 10 cm/sec = 0.077 cm/day (Per Page 3-19, Reclamation Plan) Frost Barrier 3 ft Sandy Silt and Silty Sand, Saturated K = 6.3 cm/day (Per Table 3-1 ICTM) Clay Layer (Radon Barrier), 1 ft 95 % compaction. Average Hydraulic Conductivity = 3.7 X 10"^ cm/sec = 0.003 cm/day (Per Page 3-19, Reclamation Plan) Clay Layer (Radon Barrier), 1ft, Saturated K = 0.041 cm/day (4.75 x 10"'' cm/sec) (Per Table 3-1 ICTM) Piatform Fill, 1 ft 90% compaction 3 ft Bridging Lift 80% compaction Average Hydraulic Conductivity = 8,87 x 10"^ cm/sec = 0.077 cm/day (Per Page 3-19, Reclamation Plan) Platform Fill, 3 ft., Saturated K = 63 cm/day (7.29 X 10"^ cm/sec) (Per Table 3-1 ICTM) Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Pase 7 *&* The alteration to the proposed cell cover has removed the 6" rip rap erosion protecrion layer and replaced it with a tops oil/gravel admixture surface layer vegetated with grasses. The redesigned top soil layer is proposed in order to greatly enhance the ET sink. However, the modified cover may also potentially cause penetration of deeper rooted species of plants onto the cover system and may increase long term permeability of the radon barrier. It was noted that the modified cover does not provide for any prevenrion of bio-intrusion into the radon barrier by plant roots or animals. The 3.5 ft total thickness of the top soil and frost barrier layers is questionable both from the perspective of bio-intrusion, as well as adequacy in protecting the radon barrier from frost/freeze damage. Per review of the ICTM, the issue frost depth analysis has not been adequately addressed, in that no frost penetration mode] or calculations were provided. Based on potenrial soil property differences and design differences between the Reclamation Plan and the ICTM, these analyses need to be re-evaluated for any approved changes of cover design. DRC reviewed the ET inputs into the Hydrus 1-D cover infiltration model (Partitioned into Potential Evaporation (PE) and Potential Transpiration (PT) for the modified cover design. It was noted that the ET calcularion was based on daily precipitation totals using the 3 "wettest" years in record from Blanding. However, it appears that this approach is not representative of actual site conditions in that the HYDRUS model assumes the daily precipitation is distributed evenly over 24 hours. In contrast, storm intensity is important for infiltration models for arid sites where the daily precipitation can arrive in a matter of minutes. Consequently, DUSA needs to explain and justify why the daily precipitation values used in the HYDRUS model are representative or conserv'ative of the storm intensities expected at the White Mesa site. This has imphcations for the DUSA use of a l-dimensional model to simulate embankment infiltration perfonnance, in that consideration of storm intensity will likely make mnoff and intemal filter drainage more important in the overall water budget. This may result in the need for a 2-dimensional infiltration model for the facility. It was noted that ET was calculated based on standard assumptions for grass cover and was not calculated based on field studies (e.g. lysimetry) or indigenous plant species. The ICTM states that the Feddes el al. (1978) water uptake model with water response funcrions for grass was selected in HYDRUS. The selection of this parameter is not well jusrified and no literature or local reference guides seem to have been used in choosing this parameter. DUSA needs to explain and jusrify why the grass water response functions from Feddes, et al. are representarive or conservative of the properties and characteristics of the actual White Mesa site. The cunent infiltration model assumptions are likely not calculating a representative water storage capacity since they don't account for short storm precipitation volume intensities (hourly precipitation runs), indigenous species transpiration properties, leaf-area indices, root volumes, growing season, rainfall season, or root structure development timeframes. It is suspected that the HYDRUS grass input is simulating a species such as Kentucky Blue Grass which is not at all representative of local narive vegetation and would not be sustainable on the tailings covers. The use of a standard grass input for rooting and canopy ET processes is not adequate given the large Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Pages dependence on this parameter to achieve water balance and the eventual establishment of steady slate flow to the perched aquifer. The ICTM also points to the construcrion of tailings cell covers per a study by Albright, and more specifically to the Monricello tailings cell cover system, and seems to infer that the White Mesa tailings cell cover proposed in the ICTM is a close analog of these other cover systems. Per DRC review of the Albright study and Monticello tailings cover it was noted that the White Mesa cover does not bear a close resemblance to these other cover system in location, design or in the amount of field data collected. The Monticello study utilized extensive site characteristic study and field lysimetry data prior to development of engineering design and also incorporates mnoff controls, a capillary barrier, thicker frost protection layer and bio-intmsion barrier and FML above the clay radon barrier. The Monticello site is also located at an elevation of about 7,000 ft. amsl, whereas, the White Mesa site is found at a lower elevation of about 5,500 ft amsl. DRC will request information from Denison USA regarding the omission of capillary barrier and filter layers (above the radon barrier) which appear to be fairly siandard in the ET cover designs per Albright. Justification also needs to be provided on why the higher Monticello sile is representative of White Mesa. Water Balance Per the ICTM summary of infiltration model results, section 4, flux rates at the water table below the taihngs cells (2, 3,4A, and 4B) are expected to reach steady state sometime after 200 years, as evidenced by steadily declining flux rate predictions over years 0 to 200 in the model (ICTM, figure 4-4). These declining flux rates at the water table appear to be a product of post dewatering flux rates through the bottom liner. DRC noted that the model was based on several simplifying assumptions given the 1-D aspect of the HYDRUS model and sparse site data. The model appears not to calculate steady state water contents in unsaturated zones of the tailings or geologic formations below the cells. Additional information needs to be provided, so as to verify steady- state moisture content profiles across the entire vadose zone. The ICTM does not present a consolidated calculation of water balance because of the use of various sequential and compartmentalized models (see Table 1. above). Per DRC review the values of average infiltration and flux to the water table was included in various sections. The model used to simulate infiltration conditions in the cover system and upper tailings reported duration of 228 years based pn a 57 year period daily rainfall average record which was run 4 consecutive times. It is questionable whether this is an adequate duration of modeling, even though the permit only requires the 200 year duration, given that the first mode] does not adequately display the achievement of hydrologic or chemical equilibrium. However, the infiltration model may be adequate and conservative ifthe FTYDRUS flux calculations consider the following: 1. The HYDRUS model (Model 5) uses early lime, higher vadose flux/velocity through the liner, based on Giroud and Bonaparte calculations (Model 3) and head on the liner Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 9 system which would cause more contaminant mass to enter the vadose zone (in the model) and provide a conservative 200 year contaminant mass estimate, or: 2. The HYDRUS flux prediction (Model 5) is directly linked to the transportation code, year by year. In this case the 200 years of prediction is based on declining flux rates. This would portray progressively less contaminant mass in future years. The ICTM does not clearly present the input criteria or methodology to insure that the 200 year modeling uses either of these 2 conditions, or is conservative in predicting contaminant mass at the groundwater monitoring wells. An overview ofthe water balance components is below: Infiltration Model Inputs: Precipitation (Model 1) - Precipitation is the only input into the water balance for the tailings cells (2,3,4A, 4B). The ICTM uses daily precipitation and air temperature measurements which were obtained from the Blanding Weather Station. Based on these precipitation volumes the model assumes direct infiltration into the cover based on a nearly flat top-slope cover (0.2%.) The modeling assumes, a unifonn infiltration through the cover. It does not address mnoff or intemal filter drainage from up gradient areas. DUSA needs to explain and justify why this is representative or conservative in light ofthe storm intensity issue described above. The ICTM is based on a 100 year, annual precipitation histogram which was created from the daily weather records for Blanding. Based on holes in the local precipitation record which were culled when the annual precipitation record was incomplete, White Mesa formulated a 57 year daily average precipitation input to the model. This approach may be conser\'ative in that it avoids bias that could be caused by including years with incomplete weather records which would represent under-reporting of actual precipitation (would amount to omission of precipitation due to no weather records.) No substantiation is given in the ISTM as to the suitability of a daily precipitation rate and its association with storm intensity impacts, nor is any explanation given to soil freeze and snow accumulation impacts. Infiltration Model Outputs: ET (Model 1) - As discussed in the cover section above, the water balance elements and conceptual model of the HY'DRUS simulation depend on a very large ET sink term. The model does not use site specific information to determine vegetation stmcture, leaf area index, growing season vs. wet season duration, planl mortality or revegetation with indigenous species. The model in effect uses a siandard input for grass which calculates these parameters according to the Feddes el al. (1978) water uptake model with water response functions for grass. Given the magnitude of this input to the water balance the development of the ET sink term needs to be based on local weather, soil, native plant data and related studies. At the very least, additional justification is needed. Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 10 Surface Runoff (Model 1) - No surface mnoff is considered in the model. All precipitation is assumed to be transferred through ET or to infiltrate (soil water storage and recharge.) Model 1 does not account for surface runoff or differential infiltration from surface ponding etc. In general the conceptual model does not adequately justify the omission of overland flow potentials or adequately describe hypothetical infiltration events orqtiantify the moisture capacity ofthe 6" topsoil layer in preventing surface runoff This would include an evaluation of catastrophic storm events (10 yr, 25 yr, 100 yr) and potential exceedences of the hydrauhc conductivity of the cover soil. Preferably this type of evaluation should be done in standard units (e.g. m/hr) over a short time duration to evaluate the effects of strong storm intensities. The HYDRUS cover model (Model 1) is not conservative in the respect that it can not consider overland fiow and/or differences in seepage due to potential pond areas on the cover. In tum, this omission ignores the effect of top-slope runoff and intemal filter drainage on the side-slope areas. Infiltration (Model 1): The model utihzes infiltration rates based on standard soil descriptions (loam) in the HYDRUS program. It is unclear Whether the input considers a measure of moisture storage capacity with respect to the cover materials or whether this parameter has been adequately characterized. A description of moisture capacity is necessary in order to completely evaluate runoff potential from the cell surface or ponding. The rainfall data needs to reficct hourly averages in order to simulate variances due to storm intensity as summarized in the Surface Runoff section above. Storage (Model 1) - Storage is not well specified in Model 1, as summarized in the Infiltration section above. The ICTM does not include a narrative of temporal changes in water storage and also does not explain whether these impacts are standard input assumptions in Model 1. Per review of the input decks, specific information conceming local climate in respect to seasonal variations, freeze/thaw effects, and potentia] snow accumulation was not identified in Model 1. RechargelFlux Rates Through the Cover (Model 1) - The tailings cover in Model 1 was run for 228 years, and repeats the same 57 year climatic record four times. The ICTM uses one ofthe 57 year periods to summarize the cover flux rate (Input as MAX precipitation.) It is reported that the minimum rate of flux ihrough the cover is 7.4 X 10"^ cm/day to a maximum rale of 2.0 X 10'^ cm/day, with an average flux rate through the cover system of 1.0 X 10^ cm/day. The ICTM additionally contains a curve depicting water flux rate cm/day plotted against time (years) encompassing the full 228 year model duration. For each of the 57 year precipitation cycles it was noted that the flux rates in the beginning of the plot is relatively low with a large spike of higher flux in the tail end of the record (from about year 42 until.year 57.) It was noted that the spike is evident in all plots depicting flux rates through the tailings cell and vadose zone, clearly this is an artifact of the rate of apphed precipitation in Model 1. Flux Rate Tiirough the Tailings (Model 2) - The hydraulic properties of the tailings were estimated based on comparison with the Canon City Mill where the properties were measured by aquifer testing. Based on the assumption that tailings hydraulic properties are.the same for Canon Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 11 City and White Mesa; MODFLOW simulations in Model 2 indicate that water levels in the tailings will rise on occasion during the transient phase (per the 57 year precipitation data plot) but will not overtop the liner. Again, this storage behavior predicted in Model 2 is strongly influenced by the ET predicted by Model I. For use in steady state modeling (Model 5), once transient dewatering of the tailings is done, the ICTM has made an assumption that the flux through the cover will match flux through the tailings which is reasonable since the infiltration flux is conserved in the model (assuming the predictions from Models 1, 2, and 3 are representative or conservative). The cmrent MODFLOW calculations (Model 2) for transient dewatering of the tailings is assumed to be gravity driven (head dependent.) As a result, the model sink term discharge rate differs based on the amount of head on the bottom liner for all four cells. In contrast, the actual slimes drain dewatering system is a pump out system where the pump is operated on an episodic basis and leachate is allowed to accumulate. Consequently, the elevation head in Model 2 is greater than the conesponding field condition because these fluids in the field are not removed instantiy bul allowed to build up as the pump cycles on and off. As a result, the head dependent boundary used to represent taihngs de-watering produces an artificially high sink term. Additional evaluation is needed to justify how the head dependent boundary and sink term iS:or can be representauve of field conditions. Since DUSA has never metered flows from slimes drain pumping it is unknown if flows have been continuous or have declined over the years. It is also unknown if cunent pumping rates can be sustained in the future. Model 5 does not begin simulation of infiltration and contaminant transport until the completion of: 1) the dewatering phase MODFLOW Model (Model 2) which is used solely to calculate a time duration until pseudo steady state water content in the tailings is achieved, and, 2) analytical infiltration flux calculations via model 3. Thus, all operational and post remediation phase water and contaminant fluxex on the system are not accounted for in Model 5. The model language indicates that, according to Denison Mines reasoning, the quantity of flow is minimal enough to discount these periods of discharges into the vadose zone, however, by DRC reasoning flows during the operational and transient conditions could potentially add a significant amount of unaccounted time (> 30 years) of liner seepage at a greater velocity due to higher heads in the tailings. This would affect the initial water content calculations within the unsaturated zone (cunentiy entered as 0% moisture) and could have a significant impact on chemical transport calculations over the required 200 year model period. Al the least, it is Denison Mines responsibility to show the claimed de-minimus impact through sensitivity testing or by other means such as comparisons of calculated seepage. Flux Rate Through the Liner (Models 1, 2 and 3) - Flux rates through the liner for all cells were calculated by use of the Giroud and Bonaparte equations separately for two of the models in the ICTM. As was mentioned above the MODFLOW model 2 and HYDRUS model 5 are independent. As such separate inputs (Model 2) for liner flux were used for the two models and are summarized in the table below: MODFLOW Liner Flux Inputs, Model 2 HTDRUS Liner Flux Inputs, Model 3 (Refers Denison Mines While Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 12 tailings dewatering to Flux inputs into Model 5) Assuming no-flow boundaries around the domain, the average flux through the liners of cells 2 and 3 calculated by MODFLOW (as a negative recharge rate) was *6.9 X 10"^ cm/day. As calculated by Geosyntec Consultants the potential flux rate through the liners of cells 4A and 4B per Giroud and Bonaparte analytic equations in Model 3 was 2.6 X 10"* cm/day. Per DRC review of the ICTM and cited document (Geosyntec 2007), it is unclear how this flux rate was derived. Therefore, DRC will request a full explanation as well as calculations used to derive the used flux rate. Per the ICTM the saturated hydraulic conductivity of the liner was selected as a fitting parameter, and was calibrated to match potential flux rates predicted by Knight-Piesold, based on this methodology a calibration target flux rate of *4.6 X 10"* cm/day was used for cells 2 and 3, and a calibration target flux rate of 2.6 X 10"^ cm/day was used for cells 4A and 4B. Per the saturated K method. Model 2 uses a flux rate analog input for pseudo steady state hydraulic conductivity of the liners (PVC and HDPE), this assumes a 4 ft saturated thickness on cells 2 and 3 and a 1 ft saturated thickness on cells 4A and 4B. Based on this, the actual HYDRUS inputs (saturated hych-aulic conductivity) are 7.3 X 10"^ cm/day for cells 2 and 3, and 9.2 X 10'^ cm/day for cells 4A and 4B. Per DRC review of Model 3, it appears that the values used could be appropriate for HDPE and PVC liners based on estimated flux calculations conducted by DRC engineering stafi". It was noted, however, that the range of hydraulic conductivity values is several orders of magnitude and the developmenl of a similar flux value, as that used in the ICTM, required the use of an assumed K value within a large range. DRC conducted a review of the GeoSyntec Computations (5/10/07) which were cited in the ICTM as the source ofthe used values, however, DRC was unable to locate the calculations corresponding to the used values for flux. Therefore, DUSA needs to provide the calculations for Model 3 to justify these values. *It is unclear why the flux values for cells 2 & 3 and 4A & 4B vary between the two models Flux to the Ground Water Table (Model 5) - Flux through the vadose zones of the Burro Canyon and Dakota sandstone formations in Model 5 was assumed to match the flux through the taihngs cell flexible membrane liners (Model 3.) This flux was then applied to the vadose zone in the Model 5, and moisture distribution was calculated. No information regarding the in sim residual water contents ofthe vadose sandstones was provided, hence model calibration was not possible. The use of'a 0% initial moisture content is not reasonable or conservative since the initial wetting front will take a long time to reach the water table based on low advective velocity in the vadose zone. Unsaturated Flow (Models 1 and 5) Denison Mines While Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 13 Goveming Equation — The HYDRUS-ID program solves for the Richards Equation using the VanGenuchten/Mualem equations and inputs for water retention parameters. The ICTTvI states that HYDRUS was selected because it is capable of simulating the dominant processes affecting infiltration and contaminant transport in semi-arid conditions and with multiple layer covers. The ICTM refers first to the Richards equation applicable to 1 dimensional (single dimension) vertical flow through unsaturated media: dQ/di=d/dz[K.(h)idh/dz+i)]~S Where z is the vertical coordinate positive upward [L], t is time fT], h is the pressure head [L], is the water content PL3 L-3], 5 is a sink term representing root water uptake or some other source or sink fT-ll, and K(h) is the unsaturated hydrauhc conductivity function, often given as the product ofthe relative hydraulic conductivity, Kr (dimensionless), and the saturated hydraulic conductivity, Ks [LT-1]. In solving the Richards equation the ICTM states that the Van Genuchten is used to specify the relationship of unsaturated hydraulic conductivity K(ev) vs water saturation (Sw) and of pressure head (h) vs water saturation (6). The ICTM lists the following equation (van Genuchten 1980) to describe the relationship of h vs e(h)=er + e„ 8^/(1+ |ah|")"' Or = residual water content L^L"^ Gs = saturated water content L^L'^ h = pressure head L a = inverse ofthe air-entry value (bubbling pressure) L"^ n = pore size distribution index (Dimensionless) m = 1-1/n, n>l (Mualem) The ICTM lists the following equation (Van Genuchten, 1980, Mualem 1976) to describe the relationship of K(h) vs. S*: As written in the ICTM: K(h) = KsS,'[l-(l-Se^'T]^ K(h,z) = unsaturated hydraulic conductivity function LT"' Ks= saturated hydraulic conductivity LT"^ Se = effective saturation (dimensionless) L = pore connectivity parameter (dimensionless) Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 14 M= 1-1/n, where n>l Tailings Cell Cover (Model 1) - Hydraulic properties for the tailings cell cover materials were estimated and taken from the soil-properties database in HYDRUS. The tailings cell cover was assumed in the model to be unaffected by hysteresis due to uncertainty ofthe soil properties. Tailings Model (Model 1) — Unsaturated hydrauhc properties for the tailings were estimated using grain size data for these materials and were taken from the soil-properties database in HYDRUS. The cover materials were simulated as loam for the top layer (topsoil and frost barrier), compacted clay for the radon barrier and loam for the platform fill. The ICTM states that the clay type for the radon barrier selected from the standard model inputs based on the saturated hydraulic conductivity ofthe clay (Table 1.) The loam types were predicted using an average grain size comprised of 43% sand, 41.5 % silt and 15,5% clay and a dry-bulk density of 1.71 g/cm^. The clay layer input seems reasonable and representative, however, the loam input used did not differentiate between the topsoil layer or compaction differences between the frost barrier and platform fill. The tailings were assmned to be unaffected by hysteresis due to uncertainty of the soil properties and as mentioned above were assumed to have the same hydraulic properties as the tailings ctW cover during pseudo-steady state and steady state flow. These states apply to all HYDRUS inputs into the model. Dakota Sandstone and Burro Canyon Unsaturated Zone (Model 5) — The characterization of the unsaturated zone of the Dakota Sandstone and Buno Canyon Formation was achieved for HYDRUS modeling in the ICTM by using field sample testing results from two locations. The cores were collected during the drilling for monitoring wells MW-23 and MW-30. The cores were taken from 4 different elevations for the MW-23 site and 2 different elevations for the MW- 30 site. The sample identification numbers include depth ranges for core samples. Per the modeling input table in the ICTM, the sample labeled MW-30 35.5-36.0 result was used to define the vadose zone top, the sample labeled MW-23 55.5-56 was used to define the vadose zone middle and MW-23 74.3-74.6 was used to define the vadose zone bottom. No additional explanation or foomole was included in the ICTM to justify the use of these particular samples or reasoning that there are three distinct layers within the vadose zone other than differences in hydraulic properties. The model layer thicknesses for each of these three zones are approximately 14 feet or 1/3 of the modeled total vadose domain. This seems to be an arbitrary thickness. The document titied "Perched Monitoring Well Installation and Testing at the White Mesa Uranium Mill, April through June, 2005" prepared by Hydro Geo Chem, Inc. for Intemational Uranium Corporation was reviewed by DRC to determine stratigraphy encountered in the drill holes as well as collection procedures and preservation of the core samples. The sample stratigraphy was reviewed to detennine whether the assumption of three distinct layers is appropriate. The stratigraphic summary for zones containing each ofthe three core samples is included in Table 3 below. Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 15 Table 3. - Stratigraphic Summary of Core Samples used to define Hydraulic Properties for the Dakota and Burro Canyon Unsaturated Zone (Used in Model 5) Sample No., Depth (ft) Geologic Formation Description of Core per the Drilhng Log (Hydro Geo Chem. 2005) MVv^-30, 35.5 - 36 Kd Core Log for depths 30.0-40.0 ft - Core recovery 100%, quartz sandstone, cross-bedded, medium - grit sized grains, tan, non calcareous, grit zone from 31.3 - 31.7, dark gray clay galls from 32.1 - 33.0 ft, no mineralized partings Per DRC review of the core photo, it appears to be well intact, the ''clay galls" are prevalent in the photo. Core Log for depths 49.0 - 59.5 ft - Core recovery 100%, 49.0-59.5 ft., quartz sandstone, fine — medium grained, tan, non calcareous cement, cross-bedded, very uniform, most partings occm: along cross beds and are mechanical (broken during drUling, no mineralized or weathered surfaces. Per DRC review of the core photo, the core was well intact and appears to be uniform sandstone in the 55.5 -56 ft range. MW-23 55.5-56 Kd BmroK^angiim'-iBca^taftlQa^JEbctop^ MW-23 74.3-74.6 Kbc Core Log for deptiis 70.0 - 80.0 ft - Core recovery 90 % 70.0 -70.5 ft no core recovered, 70.5 -73.5 ft, siltstone, ver)' light gray-green, soft core, low angle parting with limonite at 73.0 ft, 73.5-80 ft., quartz sandstone, light gray-tan to light pink-tan, limonite stained low angle parting at 73.7 ft, grit zone at 75.0 ft and from 75.5 - 76.5 ft, small limonite blebs after sulfides'at 77.5 -78.0 ft., some manganese dendrites from 78.5 - 79.5 ft., calcareous zone from 78.5 to 79.5 ft. Bnjsliy^asiii'J&iation. Jinb top contact is at^apprbxiinately: 105 ^ - '126*- Ijgsr^^f^T^'^ •:::! •Source - Titan Environmental, July 1994, Umeico Geophysical Log Data Per DRC review of the core logs it was determined that the unsaturated zone stratigraphy . represents a fairly uniform quartz sandstone layer, ranging from non calcareous to calcareous cement, with minor interbeds of silt stone and non-uniform clay blebs. It was noted that the two upper cores were within the Dakota SS formation which is a Paleozoic shoreline deposit, and* the deepest core was frorn the Burro Canyon SS which is a primarily a Paleozoic channel deposit. The Burro Canyon SS show a higher cementation with silica instead of kaolin, this varies from the Dakota where most of the cementation is kaolinitic. The core samples were analyzed by Daniel B. Stephens & Associates, Inc. According to the summary of tests performed, all core samples were analyzed for; inilial water content, dry bulk density, calculated porosity, saturated hydraulic conductivity (falling head), moisture characteristics (hanging column, pressure plate, water activity meter, relative humidity box), and unsaturated hydraulic conductivity (calculated.) A list of the laboratory methods used is included as Appendix A to this memo. Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 16 The validity of the laboralory residual water content used for model input in developing water saturation and unsaturated hydraulic conductivity is of concem. Table 2-1 in the ICTM summarizes the unsaturated hydraulic properties (Van Genuchten Parameters) for cores from While Mesa. It was noted that the residual water contents Hsted for all samples is 0.00 % vol. The values were cross checked with plots of Moisture Content vs. Pressure Head, using the lab data, included in the ICTM. Per review it was noted that the moisture content at a pressure head of 1 E+06 (-cm water) was approaching 0 but did not meet the 0 axis. It was noted that the head values at very low moisture contents contained a plot point detennined by Relative Humidity (RH) Box analysis. The methodology for this test was not included in the ICTM and it is unknown if the RH test provided a reliable result. Tables of data indicate tiiat a maximum of 851,293 -cm pressure was applied to the sample during analysis. Per Table 3-1 of the ICTM, residual water contents used in the HYDRUS model were shghtiy above 0. Per the JCTM language it is stated that residual water content in the vadose zone has not been determined. This would also indicate that the Van Genuchten parameters for matching of laboratory data curves and characteristic curves could not be achieved. The assumption, therefore, appears to have been made that the residual water content in the sandstones was 0% and that flux through this zone will initially match the hner fluxes. The ICTM does not clarify vadose zone fiow beyond this assumption. Although Daniel B. Stephens & Associates reported the residual water contents as 0, which DRC assumes to be an attempt on the part of the laboratory to be conservative, Denison Mines used non 0 values. The use of the non 0 values will be slightly less conservative due to an increased-K through the sandstone vadose layer and more pore dilution capabihty of the sandstone matrix. Layers and Nodes ~ The unsaturated zones are broken up inlo slightly different layers for cells 2 and 3, and 4A and 4B based on different liner materials and configurations. The HYDRUS Model 5 also depicts saturated layers in order to more accurately determine total flux ihrough overlying layers to the water table. Table 4. depicts the layers in order to display differences between the Cell 3 & 4 and 4A & 4B Model 5 inputs. These layers are depicted for Model 5 only, other models use individualized inputs and layers. Table 4. Model 5 (HYDRUS) Layers, Comparison Table for Cells 2 & 3 and 4A & 4B K used for Cells 2 and 3, Thickness ft \^^^]^ , (cm/day) HYDRUS Loam, 3.5* Compacted Clay, 1 Loam, 3 Tailings, 30 Sandy Loam, 1, Slimes Drain Loam, ] Geomembrane Liner, I*' Loam, 0.5 6.3 0.041 6.3 146.3 106.1 6.3 7.3 X 10"' 24.96 K used for Cells4A and4B,Thickness ft 1^^^!, , (cm/day) . HYDRUS Loam, 3.5'' Compacted Clay, 1 Loam, 3 Tailings, 30 Loamy Sand, 0.1, Slimes Drain Geomembrane Liner, 1 " Geosynthetic Clay Liner, 0.03 Loam, 0.5 6.3 0.041 6.3 146.3 350.2 9.2x10"' 0.024 24.96 Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 17 Vadose Zone Top, 13.8 Vadose Zone Middle. 14.1 Vadose Zone Bottom, 14.1 69.81 9.33 2.47 Vadose Zone Top, 13.2 Vadose Zone Middle, 13.5 Vadose Zone Bottom, 13.3 69.81 9.33 2.47 model assumes that the top two cover layers are combined into one equivalent layer (topsoil with gravel [0.5 ft] + frost barrier [3.0 ft]). ^ note, the geomembrane iiner layer is reflected as a 1 ft. layer and seems to correspond to the composite thickness of the liner protective blanket and 30 ML PVC Flexible Membrane Liner in cells 2 and 3 and a composite of the 2 (60 ML) HDPE GeoTTiembrafies and Geonct for cells 4A and 4B. The model inputs are based on saturated hydraulic K as an analog ofthe seepage flux. Several intermediate boundary conditions were also included in the various models (Table 1): 1. An atmospheric upper boundary condition was applied across the top ofthe model representing the tailings cell cover to simulate meteorological conditions and as a function of precipitation and potential evapotranspiration in Model 1, 2. Free drainage was assumed for the lower boundary condition of the model representing the tailings cell cover in Model 1, 3. Specified fiuxes were applied to the upper boundary at the top of the tailings in Model 4, and, 4. The lower boundary at the base of the domain was assumed to be fully saturated. It is unclear why FTYDRUS Model 5 uses layers above the flux into the vadose zone, given that at this point, through evaluation of Model 2, the head in the tailings should be pseudo steady state; ^ DUSA will be asked to explain and justify the need forthe top 7 layers listed in Table 4. Model 5 does not include a cross section spacing of nodes within the layers or at layer boundaries. The plan does state that the nodes are most concentrated within the ceil cover systems, varied spacing between 0.1 and 1 cm. The plan also states that node spacing is tighter in the syslem where changes in hydraulic properties are expected. The layers and nodes used in each Model need to be presented graphically, preferably showing nodes as applied to water contents in the layered systems where interface reactions will occur. Saturated Flow -MODFLOW (Model 6) Burro Canyon - Per geologic descriptions in the Reclamation Plan, the perched aquifer is bound on the bottom by the Brushy Basin Shale Member of the Jurassic Morrison Fonnation. Overlying strata includes the Cretaceous Dakota Sandstone, the Buno Canyon Formation, and includes "random discontinuous shale layers." T^e perched aquifer is commonly referred to as the Burro Canyon Formation Saturated Zone. Evaluation of saturated hydraulic conductivity refers to aquifer pump test, slug tests, packer tests and laboratory analysis of core samples. The summary refers to the Reclamation Plan, Revision 3.0 (Inlemational Uranium (USA) Corporation, 2000, for detailed information regarding hydraulic conducfivity measurements. It was noted that the sources of information and field tests used to calculate hydraulic conductivity were taken from tests at multiple times and from other report data, which includes; 1. Dames cS: Moore, "Environmental Report, White Mesa Uranium Project," January, 1978 (Laboratory Tests & Injection), 2. Peel Environmental Services, UMETCO Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 18 Minerals Corp., "Ground Water Study, White Mesa Facility," June 1994 (Single Well Drawdown & Injection Tests), 3. Hydro-Engineering "Ground Water Hydrology at the White Mesa Tailings Facility," July, 1991 (Single Well Drawdown and Recovery Tests), 4. D'Appolonia, "Assessment of the Water Supply System", White Mesa Project, Feb. 1981 (Multi-Well Drawdown & Recovery Tests.) The summary states that horizontal hydraulic gradient in the perched aquifer ranges from approximately 0.01 to 0.04 ft/ft and generally moves south and southwest with local variations in magnitude and direction. DRC findings of review of the water table contour maps included in the December 1, 2004 Statement of Basis (Attachment 1) confirm the direction of groundwater flow. According to the Statement of Basis pp 23 "recenl water table contour maps of the shallow aquifer have identified a significant westerly component to groundwater flow at the White Mesa facility... this change appears to be the result of wildlife pond seepage and groundwater mounding." Also, per DRC review of the July, 1994 Titan Environmental Corp. Documenl titied "Hydrogeologic Evaluation of White Mesa Uranium Mill" the investigation of the perched aqiiifer indicated an average hydraulic gradient from the center of the site to the edge of the site as 0.015 ft/ft. Although gradient calculations and potential ground water travel times seem consistent with past studies and documentation, DRC did note that according to recent investigations the highest gradients beneath the tailings cells (between wells MW-4 and TW4-8) are approx. 0.071, Therefore, DUSA will be asked to verify the sotu-ces of their current gradient inputs and provide a review of data which may be more current. The summary slates that discharge from the perched aquifer "is believed" to be springs and seeps along Westwater Creek Canyon and Cottonwood Wash to the west-southwest and along Corral Canyon to the east of the site but does not attempt to confirm the hydraulic connection. The summary does include a known surface discharge point to be Ruin Spring in Westwater Creek Canyon located 2 miles from the taihngs cells. The ICTM includes a summary of the volumetric flow budget used for the Model 6 inputs. The ICTM states that water fluxes into the groundwater flow model, include vertical areal recharge from precipitation, vertical fluxes from wildlife ponds, vertical potential fluxes from the tailings cells and water entering through upgradieni specified head boundaries (upgradient groundwater domain.) Additionally, water leaves the ground water flow model through specified head boundaries, which includes simulation of horizontal groundwater flow to the aquifer beyond the model domain and discharge to seeps and springs where the aquifer outcrops (e.g. Ruin Spring.) The overall mass balance error for the flow Model 6 was 0. ] % which amounts to 2 ft^/day with a total flow through the model of 2,244 ft^/day. Areal recharge from precipitation accounted for 26% of water entering the model, while fluxes from the wildlife ponds and potential fluxes from the tailings cells (2, 3, 4A and 4B) accounted for 67% and 6% respectively. The specified head boundaries represented less than 2% of the flow entering the model, however, specified head boundaries accounted for 100% of the flow leaving the model. Per DRC review of geologic and hydrogeologic descriptions and data of the area, these boundary conditions seem appropriate given the isolated conditions of the perched aquifer. Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 19 The ICTM MODFLOW stmcture uses predicted piezometric surface contours for the perched aquifer to test the model calibration and assumptions. Hydraulic investigation of the saturated zones has been fairly comprehensive in relation to the developmenl of the monitoring well network and subsurface investigation associated with engineering approvals. Contaminant Transport Modeling and Breakthrough Curves (Models 4. 5 and 6) Evaluation of Tailings Wastewater Quality ~ According to the ICTM, tailings wastewater chemistry is based on data collected by Denison Mines between September 1980 and March 2003. The ICTM states that the wastewater analysis used is summarized and presented in the DRC Statement of Basis for the Groundwater Permit. The ICTM states thai average values for the data are used. Table B-5 of the ICTM summarizes the concentrations used for the wastewater quality. Per DRC review it was noted that the concentrations used were taken directiy from the December 1, 2004 Statement of Basis for the White Mesa Facility (jround Water Permit. The table below summarizes the tailings wastewater concentrafions used in the model. - Table 5 - Summary of Tailings Wastewater Concentrations Used in Models 4 Concentrations as per the December 1, 2004 DRC While Mesa Ground Water of Basis") , 5, and 6 (Average Permit "Statement Analyte Aluminum Ammonia Arsenic Barium Beryllium Boron Cadmium Calcium Chloride Chromium Cobalt Copper Fluoride Concentration mg/L 1827 3131 149 0.048 0.5 6.9 3.4 368 4608 6.2 60.7 234.4 • 1695 Analyte Iron Lead Magnesium Manganese Mereur>' Molybdenum Nickel Nitrate Phosphorous Potassium Selenium Sihcon Silver Concentration mg/L 2212 3 4774 146 3.5 52.8 82.6 24 273 433 1.4 210 0.1 Analyte Sodium Strontium Sulfate Thallium TOC Uranium Vanadium Zinc Analvte pH pe Temp. Concentration mg/L 5809 7 64914 16 78.5 94 263.1 641 Concentration 1.83 S.U. 20.2 9.9 "C The use of average concentrations is not conservative (max. concentrations should be used) and also the use of a non-weighted data sel as found in the Statement of Basis table is not valid since the slimes drain samples are consistently much lower in concentration than samples laken from the tailings cells. Denison Mines While Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 20 Furthermore, DRC has noted, per comparison of the Statement of Basis data with data results found in the University ofUtah Study, ''Summary of work completed, data results, interpretations and recommendations For the July 2007 Sampling Event At the Denison Mines, USA, White Mesa Uranium Mill Near Blanding, Utah," submitted to DRC May 2008, that the Statement of Basis tailings maximum concentration data result is significantiy lower than the University of Utah data. The table below summarizes the analysis result for the 2007 University of Utah SampUng Event. The study included a limited number of results since the primary goal ofthe study was isotopic ground water age dating. Table 6 - Summary of Results for the University of Utah July 2007 Sampling Evenl per the May 2008 Study Document Site Tailings CeU 1 Tailings Cell 2 Slimes Drain Tailings Cell 3 Site Tailings Cell 1 Tailings Cell 2 Slimes Drain Tailings Cell 3 Site Tailings Cell 1 Taihngs Cell 2 Slimes Drain Tailings Cell 3 N02+N03, N (mg/L) 113 5.19 19.6 Selenium (pg/L) 16.200 <400.0 1,550 ""U(Mg/L) 581.000 . 23,700 68.100 Sulfate (mg/L) 2,500,000 666,000 107,000. Manganese (ug/L) 869.000 139,000 248.000 DRC expects that Denison Mines will justify the use of current average tailings wastewater concentration data, amend the ctnrent data to include more recent data, or use maximum concentration values. The 2004 ''Statement of Basis" data set has an autocorrelation problem since 6 of the 17 samples in the set were collected in March 2003 by DUSA. It was also noted that the sampling procediu-es for these grab samples are unknown. These issues will be included in correspondence to Denison Mines. The use of the 2004 "Statement of Basis" data will need to be justified as valid statistically and as conservative. Model 3 PHREEQC Development ofKd Coefficients and Rj Factors used in Model 4 - Model 4 is a precursor lo the development of Hydrus Model 5 and is used solely for the development of Kd Coefficients and R Factors. Core samples from the 2005 drilling of wells MW-23 and MW-30, the same cores described in Table 3 were used for analysis of the mass of hydrous-ferric oxide (HFO) present in bedrock underlying the White Mesa Mill. The samples included additional depths as summarized Table 7 Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 21 below. The analysis was conducted by ACZ Laboratories, Inc., report date August 27, 2007. Results were submitted for the following parameters: Dissolved Aluminum, Dissolved Calcium, Dissolved Iron, Dissolved Magnesium, Dissolved Manganese, and Dissolved Uranium. It was noted that the reports from ACZ Laboratories list the concentrations in units of mass/volume, mg/L. Foomote a) of Table B-5 in the ICTM states that "PHREEQC adjusts tiie initial concentrations according to the total mass of solutes; as a result, all input values were converted from mg/l to mol/kgw (moles per kg of water.)" Denison Mines U.S.A. will need to explain the conversion from liquid concentrarion (ACZ report) to the inputs into PHREEQC as moles per kg of water. Per review it is not clear exactiy when or how this conversion is being made. Table 7 - Core Depths used for HFO Analysis Core Sample ID lMW-30 MW-30 MW-30 MW-23 • MW-23 MW-23 MW-23 MW.23 Formation Kd* Kd* Kd* Kd* Kbc* Kbc* Kbc* Kbc* Depth bgs (feet) 37.5-38 43^3.2 43.2-43.5 53-53.5 74-74.3 82.5-82.7 99.8-100 103-103.3 'Source - Titan Environniental, July 1994, Umctco Geophysical Log Data Samples were also analyzed (ACZ Laboratories, Inc., report date August 10, 2007) to detenhme the acid-neutralization potential of the bedrock. The ACZ reports include results for; Acid Neutralization Potential (calc) and Neutralization Potential as CaC03. Per DRC review ofthe core samples used for acid-neutralization tests, the samples were not labeled according to well # and depth. Denison needs to clarify the source and applicability of the data. Dz-ombak and Morel Surface Complexation Model (PHREEQC Model 4)- Appendix B of the ICTM is titied "Speciation and Surface Complexation Modeling ofthe Tailings Porewater." This appendix presents the methodology used to develop Kd Coefficients and Rf Factors for the Vadose Zone Model. Appendix B of the ICTM references the book by Dzombak and Morel "Surface Complexation Modeling, Hydrous Ferric Oxide" (1990) which discusses surface chemistry of hydrous oxides (e.g. Fe, AL Mg, Silica.) This chemical modeling is strongly dependent on solution pH, ionic strength and competing ion effects. Dzombak and Morel ascertained that acid base chemistry and oxides would lend themselves to ion sorption models (which could then be translated into Kd and Rf variables.) Specifically the following bullets were noted in the introduction ofthe Dzombak and Morel book which have been utilized to develop a thermodynamic database for sorption reactions: Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 22 L Sorption on oxides lakes place at specified coordinate sites, 2. Sorption reaction on oxides can be described quantitatively via mass law equations, 3. Surface charge results from the sorption reactions themselves, and, 4. The effect of surface charge on sorption can be taken into account by applying a conection factor derived from the "Electronic Double Layer (EDL)" theory to mass law constants for surface reactions. One of the objectives of this type of modeling was to develop thermodynamic sorption databases for ion reactions based on laboratory tests and statistical evaluation. It appears that multiple databases which have been developed are part of the PHREEQC modeling code. Core analysis results as described above were used as inputs to the PHREEQC Model 4 in order to mimic tbe vadose zone mineralogy. Basically, it appears that the obtained data was used to quantify the number of hydrous oxide ion sorption sites available. The vadose zone was broken into 3 layers within Model 4 and as consistent with HYDRUS Model 5. Each layer represents a different receiving mineralogy (for the tailings water) and PHREEQC has computed different Kd and Rf values based on the Offering inputs from the ion sorption data. The PHREEQC model looks at available chemical species present at different pH and oxidative states thus the neutrahzation polential ofthe receiving matrix was also entered into Model 4 to mimic the chemical species evolution present in solution at these differing states. It appears that the ICTM is calculating an almost immediate neutralization upon tailings water release to the vadose zone, based on the evaluation of the core samples for acid neutralization potential due to CaC03 measurement. This is particularly tme for vadose zone layer 2. However, based on DRC review of the DUSA acid neutrahzation potential data, 2 of the 5 samples showed no neutralization capability, 2 other samples showed very low neutralization capability with only 0.1 % CaC03 and 1 of 5 samples showed a potentially low neutralization capabihty as 0.4 % CaC03 (See Appendix C "Core Neutrahzation Data" for laboratory data sheets.) The ICTM additionally refers to water quality data of the infiltrating porewaters which indicates high buffering capacity, however, there is no data supplied with the ICTM tp support these claims. Although zones of high neutrahzation are claimed, Denison needs to supply more information to support these claims. The vadose zone mineralogy has not been well characterized in the ICTM. The ICTM includes copies of laboratory reports for core analysis (recovered from well bores — the same bores summarized in Table 3 - for wells MW-30 & MW-23), per review^ the analysis included 6 dissolved metals: Aluminum, Calcium, Iron, Magnesium, Manganese, and Uranium; and at depths for MW-30 of 37.5-38 ft bgs, 43-43.2 ft bgs, and 43.2^3.5 ft bgs; and at depths for MW-23 of 53.0-53.5 ft bgs, 74.0-74.3 ft bgs, 82.5-82.7 ft bgs, 99.8-100.0 ft bgs, and, 103-103.3 ft bgs. The laboratory analysis results were then separated for each of the three layers in this zone in Model 5 (the zones were discussed above.) It is not clear how this information was used to provide a clear Denison Mines White Mesa Uranium Mil) Infiltration and Contaminant Transport Modeling Report Review Memo Page 23 understanding ofthe mineralogy ofthe vadose zone, however, it was noted that the concentrations have been converted to mass of HFO in the receiving mineral matrix. The ICTM states that model 5 is conservative given that it only calculates ion sorption based on a single iron-oxyhydroxide phase and is not considering the sorption of inorganic ions to specific mineralogical matrices, e.g. goethite.. From DRC perspective, to show that the model is truly conservative, the outputs of several different model assimiptions should have been included and compared. This is a particular concem since the downstream modeling (MODFLOW model 6) is dependent on the assumptions of model 5. Table 8 below presents the Kd Coefficients and Retardation Factors that were generated by PHREEQC for use in the Hydrus Model 5. Note that based on the core analysis, the Kd and Rf factors differ for the 3 vadose zone lavers. Table 8 - PHREEQC Model 3 Generated Kd Coefficients and Rf Factors T"te3ose3 :m^-sS I IMS "*:--.Li .-•'.. • ' ••i ,:* w^^ -.rr-.'.:.* Top 7.19 82.1 0.001 0.557 0.000 4.13 0.000 0.000 9.48 0.001 Middle 7094 72140 1.033 4.90 0.115 1220 0.0003 0.000 2197 0.901 Bottom 0.119 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000-0-000 ^s^tv-?i m^ .t>-.'i •Se^iJ. M. ms .4. -_. •* Top 0.000 0.014 0.005 0.015 0.000 0.002 0.000 8.47 0.000 0.009 Middle 0.000 0.663 1.380 0.015 0.000 0.003 0.0001 0.4 559 11.3 Bottom 0.000 0.000 0.000 0.001 0.000 0.000 0.000 0.000 0.000 0.000 fVadose .Zone. ;Layer As-. .Rf (l/kg) Cd Rf (l/kg) R-f •.- -(l/kgji '^ gGii/:. (l/kg)- ?5ie';^H-Pb Top 213. 2428 1.02 17 LOO 123 1.00 LOO 281 1.02 Middle 161804 1645434 25 113 3.63 27822 LOI LOO 50105 22 Botlom 3.30 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 Vadose Zoiie Layer Hg .. Rf " (l/kg) Mo Rf Ni • Rf' (l/kg) Se . R:f : (l/kg) •Ag Rf S04 Ri:'- --(l/kg). Ti Rf •:(-l/kg): u . •Rf'.- .(•l/kg) Rf . (I/kg). Zn- Top 1.00 1.41 1.14 1.46 LOO 1.07 1.00 251 LOO 1.26 Middle LOO 16 32 1.34 1.00 1.07 1.00 239 12744 260 Bottom 1.00 1.00 1.00 1.01 1.00 1.00 LOO 1.00 1.00 1.00 Model 5 Scenarios — The model assumed that for the analytes chloride and sulfate there was no retardafion of movement through the vadose zone. The transport for these two analytes calculates transport from the tailings liner through these zones based on average calculated hydraulic Denison Mines While Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 24 conductivity assumptions. /Uthough there are no state water quality standards associated with sulfate and chloride these analytes are modeled more extensively since they arc considered completely mobile and conservative. The natural background concentrations in the receiving rnedia for chloride and sulfate were set at zero and not considered in the transport calculation. Model 5 Predicted Breakthrough Curves ~ Four plotted breakthrough curves were submitted with the ICTM. It was noted that the plotted breakthrough curves only relate to predicted concentrations in the vadose zone from tailings pore waier within 200 years and only included single ion analysis for Cl, S04 and U. The plots are included in Appendix B. of this memo. Table 5 below summarizes the infonnation provided on the plots. Table 5 - Summary of submitted breakthrough curve plots, HYDRUS (Model 5) - Note: Plots included in Appendix B Plot Figure No. & Titie Figure 4-5 - Model 5 Predicted Chloride Concentrations in Vadose Zone Pore Water Immediately Above the Perched Aquifer Beneath Cells 2 and 3 Figure 4-6 - Model 5 Predicted Chloride Concentrations in Vadose Zone Pore Water Immediately Above the Perched Aquifer Beneath Cells 4A and 4B Figure 4-7 - Model 5 Predicted Sulfate Concentrations in Vadose Zone Pore Water Immediately Above the Perched Aquifer Beneath Cells 2 and 3 (Note: S04 plot was not included in the ICTM for concentrations below cells 4A and4B) Figure 4-8 - Model 5 Predicted Dissolved Uranium Concentrations in Vadose Zone Pore Water at 200 Years Nanative Per the ICTM description the chloride concentration in the HYDRUS model 4 was initially set to 0 mg/L at the lower model boundary. The plot shows that (with cunent inputs), chloride will not arrive at the top of the perched aquifer until 130 years after model start time below cells 2 and 3, The chloride concentration then rises from 0 mg/L lo 0.38 mg/L in 70 years. The modeled flux rate beneath cells 2 and 3 upon steady state is 2.5 X 10"^ cm/day. The plot shows that chloride concentrations will not arrive at the top of the perched aquifer until 145 years after the model start time below cells 4A and 4B. The chloride concentration at 200 years is 0.012 mg/L. The modeled flux rate beneath cells 4A and 4B upon steady state (approx. 200 years) is 1.4 X 10*^ cm/day. As with chloride, the sulfate concentration in the HYDRUS model was initially set to 0 mg/l at the lower model boundary: The plot show that sulfate will not migrate to the top of the perched aquifer beneath cells 2 and 3 for approx 155 years. The concentration in this zone in 200 years is 0.08 mg/L. The plot represents Dissolved Uranium Concentration (mg/L) against Depth Below Tailings.Cell Liner (cm.) The plot depicts that (with current modeling inputs) uranium concentration at 200 years will not be deeper than approx 23 feet in cells 4A, 4B, 2 and 3, however, it was noted by DRC that the scale on the figure is not adequate to show U concentrations at the Utah Water Quality Standard (0.030 mg/L). The figure needs to be rescaled to Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport ModeUng Report Review Memo Page 25 show the depth ofthe 0.030 mg/L concentration. The plot also depicts approx -10 mg/l concentration differential under cells 4A and 4B from cells 2 and 3 to a depth of 10 ft then a narrowing of concentration differentia] to 0. It was noted that the dissolved U concentrations in cells 4A and 4B showed a lower concentration curve (raised) down to a depth of approximately 23 feet. (See Appendix B Uranium Plot, Figure 4-8) Model 5 Kd Coefficient and R Factor Assumption Weaknesses - After DRC review of the model 5 inputs and assumptions to derive contaminant transport through the vadose zone, it was recognized that the modehng platform chosen is not adequate. Specifically, the development of Kd Coefficients and R Factors is all contrived through modeling assumptions which may not be reflective of tme site conditions. DRC delermination is that the development of Kd Coefficients and Rf Factors through empirical procedures is completely feasible and will provide a much more realistic determination ofthe mineral matrix within the vadose zone. Additional field data needs to be collected and analyzed. Sensitivity Analysis - The issue of sensitivity analysis was briefly mentioned above in terms of Model 2 inputs and claims that the modeling inputs are conservative. DRC noted that the actual output files should be included with the ICTM to show that this is the case, especially in terms of derived mode] inputs and assumptions. Per DRC review, and as per the section above, the model 5 assumptions are not appropriate since empirical data will supersede the current model 5 assumptions. The redevelopment/refining of Model 5 will also require that Model 6 be reconstmcted based on the new Model 5 output. It is DRC's recommendation that DUSA prepare a robust discussion of sensitivity analysis for Model 5 and Model 6. Ultimately, from DRC perspective, sensitivity analysis should determine whether sufficient data has been collected to deem the model output as credible. If the choice is to not collect additional data, then the sensitivity analysis needs to detennine that, based on what has been collected, the model output is the most conservative (e.g. is based on conservative flow and transport inputs.) In many cases, it is expected that conservative modeling will guide the need for additional empirical data. As noted, this is the case for Models 5 and 6. Conclusion Generally, DRC review concludes that empirical data is lacking to develop a representative model of infiltration, flow and chemical transport. This is primarily due to a lack of site specific data or local data reference regarding long term vegetation establishment inputs (e.g. plant density, leaf index, and rooting structure, etc.), long term degradation of the system (e.g.' cover flow impacts from frost heaving, erosion, intrusion, etc.), tailings wastewater pollutant concentrations based on sound statistical evaluation and representative sampling, initial vadose zone water content (based Denison Mines White Mesa Uranium Mill infiltration and Contaminant Transport Modeling Report Review Memo Page 26 on empirical core evaluation), vadose zone mineralogy, and geochemical Kd and R characteristics (based on empirical data,) RFI Requirements The following additiona] information will be required and will be hsted on an RFI letter to Denison Mines U.S.A. White Mesa Cell 1 RFI Per the ICTM, White Mesa states that Cell 1 will not contain mill taihngs upon decommissioning of the sile. It appears that Cell 1 has.not been included in the ICTM based on the detennination by Denison Mines U.S.A. that it does not meet the permit language requirement of a closed tailings cell. DRC does not agree with this interpretation for the following reasons: 1. The demolition and decommissioning wastes in question will be contaminated witii by- product material (See Utah Administrative Code (UAC) R313-12-3.) 2. Denison Mines U.S.A. has not made a demonstration that the wastes in cell 1 meet the definition of "de-minimus" effect on local ground water quality, pursuant to UAC R317-6-6.2(A)(25), nor has the Executive Secretary approved such a demonsti-ation. As a result, Cell 1 does not qualify for Permit-by Rule status. Therefore, because contaminated materials from mill site decommissioning, including by- product materia] will be disposed in Cell 1, Cell 1 is required to be modeled, and included in the ICTM. Alternatively, Denison Mines U.S.A. may submit and secure Executive Secretary approval of an apphcation that demonstrates how Cell 1 will have a "de-minunus" effect on local ground water quality pursuant to UAC R-317-6-6.2(A)(25.) Model Platforms RFI DUSA needs to justify the use of independent sequenced modeling platforms. DRC noted that input errors from one model to the next could result in trickle down errors of great magnitude. Table 1 and the accompanying flowchart display the weakness ofthis approach. 1. Denison Mines U.S.A. has not provided adequate justification that the conceptual approach of using several sequenced, compartmentalized models (HYDRUS, PHREEQC, and MODFLOW) provides a representative or conservative simulation of tailings cell performance. 2. Provide verification that the use of a variable saturated coupled flow and transport model (VSCF model) was considered in the conceptualized modeling phase and • justification why the VSCF modeling platform was not appropriate or necessary. Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 27 Cover Design RFI (Model 1) 1. The modified cover design eliminates the cobble surface layer. Please justify that long term surface erosion will not occur with the new surface layer. Specify topsoil gradations (e.g. admixture components) which will be used to replace the rip-rap layer. 2. The modified cover includes a top vegetated layer. The ICTM only describes the vegetation type as grasses. Please specify vegetation grass species and justify long term growth on the cover including reasoning for not including long term sustainability and potential for in growth of other natural species or plant succession (e.g. indigenous brush.) Also, provide additional information regarding root structures and plant type inputs entered into the HYDRUS 1-D surface layer model regarding long term establishment of vegetation. Provide the basis for any assumptions made, including documented studies of local vegetation/soil models (e.g. lysimetry) to support claims. 3. The HYDRUS inputs for ET from plant surfaces are unclear. Per appendix C, a grass layer was entered into the HYDRUS model. Please provide information regarding the details of this entry. Examples of needed information include the specific vegetation type or species, cover density and rooting depths and density, leaf area index, andra description of calculations performed in the FTYDRUS model and self-sustainability of the vegetation at the White Mesa Facility. Provide specific reference to studies used to support the long term vegetation estabhshment and assumed ET component. 4. The modified cover does not appear lo include design to avoid biointrusion into underlying layers (e.g. the compacted clay, radon barrier.) Please justify why this is not a concem for long term degradafion, including justification that the 3 ft frost barrier layer provides adequate protection from intmsion by roots, animals, etc. into the underlying layers. Alternatively, modify the design to include a biointrusion layer at an appropriate location above the radon barrier. 5. Provide documentation of frost depth analysis and maximum projected cover frost penetration depth in the ICTM to justify that the 3 ft frost barrier thickness is adequate. Provide reference to local and/or regional studies to support all claims. 6. Provide justification why the daily average precipitation rates used for the modeling assumption are representative (or conservative) of true field condirions and storm intensity effects. Specifically, the HYDRUS model distribution of a daily average precipitation over a 24 hour period negates storm intensity effects which are pertinent to the semi-arid environment at White Mesa. This assumpfion substantially changes soil saturation and hydrologic effects due to surface infiltration variations. Explain how the effects of these processes may change hydrologic properties of the cover and Denison Mines While Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 28 justify how these 2 and 3-dimensional phenomena can be modeled with a 1- dimensional model (HYDRUS 1-D.) 7. It was noted that the ICTM references ET cover design studies and articles in which the majority of the top cover designs include a capillary barrier and intemal drainage layer. Please provide discussion and justification in the ICTM regarding the omission of a capillary barrier/drain layer and surface runoff layer in the top cover design at White Mesa. 8. Include discussion regarding how the HYDRUS 1-D modeling inputs account for any long term changes of saturated hydrauhc conductivity in the vegetated suiface layer or increases in saturated hydraulic conductivity within the compacted radon-barrier clay layer due to long term degradation (e.g. root penetration, freeze-thaw damage, etc.) 9. Provide additional infonnation ifegarding seasonal variations of ET and water storage in the cover and how the HYDRUS model compensates for such seasonal differences (e.g. frozen soil and snow accumulation, spring snow-melt, etc.) 10. Explain and justify why the HYDRUS model assumes die FML to be 1 fool thick. Water Balance RFI (Models 1,2,5, and 6) 1. Provide a cross section(s) depicting: I) The model layers and node geometry used in • each model, and, 2) A plot of steady state water saturation with depUi from the top of the cell cover surfaces (cell's 2, 3, 4,4A) vertically downward lo the top of die water table. Identify on the cross section(s) which models were used to simulate the hydrauhc performance of each specific layer. 2. Provide graphs to demonstrate steady state water content at representative depth intervals for each layer in the model to represent flux through the Uner system (recharge component), flux through bottom hner for each cell, and flux through vadose zone to the perched aquifer for each cell (based on steady state saturation) and the modeled time (years) needed to reach steady state. The ICTM includes average flux through the cover system and anticipated fluxes through the vadose zone, however, these flux rates are not clearly associated with steady state predictions of water content. 3. For the unsaturated tailings and vadose zone please justify why the initial saturation in the PTYDRUS model (Model 5) was entered as 0%, 4. Please provide and justify the Van Genuchten/Mualem fitting parameters used in the HYDRUS model for both the unsaturated tailings (Model 1) and vadose zone materials (Model 5). Please ensure this justification accounts for grain size distribution and corrections for differential deposition within the tailings cell (e.g. shore deposits vs. mid ceU deposition.) Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 29 5. The slimes drain layer shown in cross section for cells 2 and 3 depicts a uiuform sand layer, per cell design and constmction specificafions this layer is not uniform across the bottom of each cell. Explain potential input differences based on anon-uniform sand layer. Emphasis is needed in explaining how and why a gravity driven, head dependent sink term in the MODFLOW model (Model 6)is representative of actual slimes drain construction and operation, that includes but is not limited to episodic operation of the slimes drain pump. 6. Provide calculations and plots of data reflecting expected/estimated leakage rates from the single FML (via factory defects) into tailings cells 2 and 3 foundations (e.g. specific calculations of Giroud and Bonaparte (Model 3) as representative of White Mesa materials and installation, and/or leakage rales estimated by field study,) Also, provide details of Giroud and Bonaparte Model for cells 4A and 4B (Model 3) as approved by DRC and appropriately applied to the ICTM modeling.- Flow Modeling RFI 1. Provide additional clarification regarding the 3 vadose zone intervals used in the .-. HYDRUS model (Model 5), including specific descriptions of each to discuss the different hydrauhc characteristics of each zone. Provide additional justification regarding the thickness ofthe zones. Provide a stratigraphic cross section ofthe vadose zone intervals in the ICTM, based on core log mineralogical evaluation and/or other data. For ease of review, provide the vadose zone interval descriptions, summary of data, and stratigraphic representation in a single section within the report. 2. Provide information regarding enor flags on the flow modehng runs (HYDRUS Models 1 and 5, and MODFLOW Models 2 and 6), including error messages for potential random data based on statistic values and the impact of these errors on the model output. DRC noted that several different types of error messages were logged on the HYDRUS output decks. It would be useful for Denison to include a table in the ICTM summarizing the location of these enor flags and explaining why the messages occurred, their impact on model output results, and potential enors which may be present in the output data. 3. Provide clarification regarding HYDRUS (Vadose Zone Model 5) and MODFLOW (Dewatering Phase Model 6) interface during transient dewatering and corresponding flux rates through the bottom liner. DRC needs to understand how the flux through the bottom liner during operations.and during slimes drain dewatering (transient flow period) is accounted for. According to language in the ICTM, it appears that wastewater entering the vadose zone during this time period is not included based on the assumption that it will have a minimal effect on long term model outcomes. Please jusrify this position arid explain why contaminant mass released to the foundation during the operational phase ofthe tailings cells can be ignored in the transport Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 30 predictions. This is particularly important in light of the fact that the highest driving heads are likely found during the operation of the taihngs cells. 4. Provide a graph depicting model layers associated with the HYDRUS Vadose Zone flow model (Model 5) to depict steady state water contents in each layer. 5. Provide a specific discussion regarding differential flow characteristics which define the layers in tiie model (e.g. differences in K,. flow direction, head distributions, etc.) This could be achieved through graphical representation and additional explanation of Table 3-1 in tiie ICTM. 6. Justify the use of a head dependent sink term in the MODFLOW model for tailings dewatering (Model 2). Explain and quantitatively justify how the simulation is representative of pmnp dependent slimes dewatering systems at each of the tailings cells. 7. . Explain and Justify why HYDRUS Model 5 breakthrough curves were based on a lower flux rate than predicted for other upstream models (Model 3.) 8. Provide specific calculations used to determine the geomembrane flux rates for cells 2 and 3. Contaminant Transport Modeling RFI DRC has significant concems with the current contaminant transport modehng, models 5 and 6 per the flow chart above. Two primary issues were evident through DRC review; 1. Tailings wastewater chemistry calculations are not conservative, and 2. The PHREEQC ion sorption model is based largely on assumption and sketchy field data which does not adequately represent the receiving mineralogical matrix or geochemical environment. The ICTM HYDRUS model 5 fails to provide a reasonably conservative evaluation of the physical system. This is primarily due to a lack of empirical data to represent chemical partitioning and retardation through the vadose zone. The failure of HYDRUS model 5 is further exasperated downstream in MODFUDW Model 6 where non-conservative HYDRUS inputs are used for the development of Cl and S04 breakthrough curves. Reguest for Information 1. Regarding the evaluation ofthe tailings cell wastewater chemistry, the use of average concentrations is not conservative (max. concentrations should be used) and also the use of a non-weighted data set as found in the Statement of Basis table is not valid since the slimes drain samples are consistently much lower in concentration than recent Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 31 University of Utah samples taken from the tailings cells. DRC also noted that no tailings cell wastewater data newer than the 2004 Statement of Basis (such as data included in the University of Utah Study document, "Summary of work completed, data results, interpretations and recommendations For the July 2007 Sampling Event At the Denison Mines, USA, White Mesa Uranium Mill Near Blanding, Utah," submitted to DRC May 2008) has been incorporated into the data evaluation. DRC is also concemed that auto-conelation is present in the data set used for the ICTM. Please revise the evaluation of the taihngs wastewater and include conservative estimates of the wastewater contaminants along with a full statistical review and justification of the data set used. 2. Clarify whether flux rates from the tailings cell liners, used to evaluate the 200 year contaminant iransport breakthrough curves include differentiating flux during transient phases of dewatering heads on the bottom liner. Include references to specific files/input decks showing the differential heads if applicable. If the transient dewatering phase is not included in transport modeling provide justification why this would not impact the result of the 200 year breakthrough curve modeling. 3- Per the ICTM, p. B-6, "water quality data for the White Mesa Mill tailings porewaters and leach extraction data for the underlying bedrock was examined to calculate - adsorption of dissolved species under varying geochemical conditions." Provide spreadsheets summarizing the data used as well as a copy of the laboratory data sheets including QA/QC verification of results and statistical analysis ofthe data set used.' Provide concentration data which was used for adsorption calculations and an explanation of the calculations made. Please demonstrate how these laboralory tests and calculations were performed using standardized methods recognized by the regulatory and relevant technical communities. 4. Provide additional justification that the laboralory results of digested core data are representative of the mineralogy of the vadose zone. Also, please prepare spreadsheets summarizing laboralory data results for the core samples encompassing all results characterizing the mineralogy. 5. Addifional clarification is needed regarding the modehng of tailings leachates and bulk reactions with the receiving aquifer matrix. Please perform laboratory tests for Kd using representarive soils/rock in the presence of multiple tailings leachate sampies with a range of contaminant concentrations to better characterize the geochemical relarionships. DRC anricipates that the development of empirical Kd coefficients and Retardafion Factors will supersede the use ofthe PFCREEQC ion sorption model. 6. Figure 4-8 of the ICTM, "Model-Predicted Dissolved Uranium Concentrations in Vadose Zone Pore Water at 200 Years" does not allow review of the migration of dissolved U below the tailings impoundments to the State Ground Water Quality Standard (GWQS) concentration, 0.030 mg/L. The figure needs to be magnified (re- Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 32 scaled) to allow review of U migration to tiie GWQS of 0.030 mg/L. Also, please insure that all breakthrough curves which depict parameter (pollutant) concentrations (based on revised modehng) which are subject lo a State Water Quality Standard, are drawn to scale lo show the State Standard. 7. The current geochemical model does not appear to anticipate the effects of changes in neutrahzation and HFO capabihty as minerals are consumed. Provide explanation of how this factor was considered in the model. 8. Per DRC review it was noted that 11 potential contaminants are hsted with very low Kd values predicted by PHREEQC. Although the breakthrough curves for CL and S04 are considered conservative by the model justification that they are modeled as un-retarded, they also do not have associated water quality standards for comparison of predicted concentrations with time. In association with the additional breakthrough curves please justify why the Kd and Rf used for Uranium is conservative in relation to the geochemical environment. References Albright, Wilham H., Benson, Craig H., Gee, Glendon W., Roesler, Arthur C, Abichon, Tarek, Apiwantragoon, Preecha, Lyles, Bradley F., and. Rock Steven, A., 2004 "Field Water Balance of Landfill Final Covers", Journal of Environmental Quality 33:2317-2332. Denison Mines (U.S.A.) Corp., Modeling Report, While Mesa Mill Site, Blanding, Utah, November 2007 . Dzombak, David A. and Morel, Francois M. M., 1990, "Surface Complexation Modeling, Hydrous Ferric Oxide," John Wiley & Sons Inc. Environmental Protection Agency, Understanding Variation in Partition Coefficient, K<j Values, EPA 402-R-99-004B. August 1999 R. Allan Freeze & John A. Cherry, "Groundwater". 1979 Hydro Geo Chem Inc., prepared for Intemational Uranium Corporation, August 3,2005; "Perched Monitoring Well Installation and Testing at the White Mesa Uranium Mill April through June 2005 Intemational Uranium U.S.A. Corp., White Mesa Mill Reclamation Plan, Revision 3.0, July 2000 Meyer, P.D., Rockhold, M.L., Nichols, W.K, Gee, G.W., Pacific Northwest laboratory (Battelle Memorial Institute), 1996, "Hydrologic Evaluation Methodology for Estimating Water Movement Through the Unsaturated Zone and Commercial Low-Level Radioactive Waste Disposal Sites" prepared for the U.S. Nuclear Regulatory Commission, NUREG/CR-6346 PNL-10843, 1996 Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Page 33 Smith, Gregory M., Weston, Roy F., U.S. Department of Energy Grand Junction Office, 1999, "Evolution of Disposal Cell Cover Design Used for Uranium Mill Tailings Long-Term Containment" Van Genuchten, M. Th, 1980. A Closed-form Equation for Predicting the Hydraulic Conductivity of Unsaturated Soils Waugh, W.J., 2002 for U.S. Department of Energy Grand Junction Office, "Monticello Field Lysimetry: Design and Monitoring of an Altemative Cover" APPENDIX A - Core Sample Analysis Denison Mines While Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Appendix A. ASTM Core Analysis Methods Summary Page 1 Core Samples for MW-23, and MW-30 (White Mesa Uranium Mill) were analyzed by Daniel B. Stephens & Associates, March 21, 2007 ASTM Methods Used and summary:' DRY BULK DENSITY ASTM D453] - "Standard Test Methods for Bulk Density of Peat and Peat Products", The method covers the measure of bulk density of samples taken with a piston sample or appropriate core sampler to obtain an undisturbed core, or by the paraffin wax method using undisturbed irregular pieces of clods of wet peat and compressed peal. The methods correspond to direct wet mass and mass by water displacement respectively. DRY BULK DENSITY AND MOISTURE CONTENT ASTM D6836 - "Standard Test Metiiods for Determination of the Soil Water Characteristic Curve for Desorption Using a Hanging Column, Pressure Extractor, Chilled Mirror Hygrometer and/or Centrifuge" Includes methods A. hanging columns suitable for making determinations for suctions in the range of 0 to 80 kPa, B. Pressure Chamber with Volumetric Measurement and C. Pressure Chamber with Gravimetric Measurement suitable for suctions in the range of 0 to 1500 kPa, D. Chilled Mirror Hygrometer suitable for determination for suctions in the range of 500 kPa to 100 MPa and E. Centrifuge Method suitable for determinations in the range 0 to 120 kPa. Method A. - coarse soils with little fines that drain readily Methods B. and C. - finer soils which retain water more tightiy Method D. ~ used when suctions near samration are not required and commonly is employed to define the dry end of the soil water characteristic curve (>1000 kPa) Method E. - typically used for coarser soils where an appreciable amount of water can be extracted with suctions up to 120 kPa Methods A-C - Yield soil water characteristic curves in terms of matric suction. Various suctions are applied to the soil and the conesponding water contents are measured. Two different procedures are used to apply the suction, Method A - tbe matric suction is applied by reducing the pore water pressure while maintaining the pore gas pressure at the atmospheric condition, Methods B and C - the pore water pressure is maintained at atmospheric pressure and the pore gas pressure is raised lo apply the suction via the axis translation principal. For all three methods, saturated soil specimens are placed in contact with water saturated porous plate or membrane. The matric suction is applied by one of the two aforementioned procedures. Application ofthe matric suction causes water to flow from the specimen until the equilibrium water content corresponding to the Denison Mines White Mesa Uranium Mill Infiltration and Contaminant Transport Modeling Report Review Memo Appendix A. ASTM Core Analysis Methods Summary Page 2 apphed suction ids reached. Equilibrium is established by monitoring when water ceases to flow from the specimen. Several equilibria are established at successive matric suctions to construct a soil water characteristic curve. The water content corresponding to the applied suction is determined in one of two ways. For A and B, the volume of water expelled is measured using a capillary tube. The water content is then determined based on the known initial water content of the specimen and the volume of water expelled. Method C. the waler content is measured gravimetrically by weighing the specimen, after removal from the apparatus. Method D - Yields soil water characteristic curve in terms of total suction. Water content of the soil is controlled and the corresponding suctions are measured. Based on this, two approaches are used, In one approach a set of specimens are prepared that are essentially identical, but have different water contents. Water contents are selected that span the range of water contents that will be used to define the soil water characteristic curve. In the other approach a single specimen is used. The specimen is tested, dried to a lower water content and then tested again, this process is repeated until suctions have been measured at all ofthe desired water contents. Method E. - Yields a soil water characteristic curve in terms of matric suction or capillary pressure. Different matric suctions are appUed by varying the angular velocity - of the sample and measuring volmne of waler displaced from the soil MOISTURE COISTENT ASTM D2216 - "Standard Test Methods for Laboratory Delermination of Water Content of Soil and Rock by Mass" Two test method which cover the laboratory determination of the water content by mass of soil, rock and similar materials where the reduction in mass by drying is due to loss of water. Due to drying difficulties with materials such as gypsum or dissolved solids the test may require different drying temperatures or humidity conditions. Assumptions are made that mass lost in oven drying is water content. CALCULATED POROSITY ASTM D 2435 - "Standard Test Methods for One-Dimensional Consolidation Properties of Soils Using Incremental Loading" In this test method a soil specimen is restrained laterally and loaded axially with total stress increments. Each stress increment is maintained until excess pore waler pressures are completely dissipated. During the consolidation process, measurements are made of change in the specimen height and the data are used to delermine the relationship between the effective stress and void ratio or strain, and the rate at which consolidation can occur by evaluating the coefficient of consoUdation. Denison Mines White Mesa Uranium Mil! Infiltration and Contaminant Transport Modeling Report Review Memo Appendix A. ASTM Core Analysis Methods Summary Page 3 PRESSURE PLATE ASTM D2325 - "Standard Test Metiiod for Capillary-Moisttire Relationships for Coarse and Medium Textured Soils by Porous Plate Apparatus" This test method covers the detennination of capillary moisture relationships for coarse and medium textured soils as indicated by the soil-moisture tension relations for tensions between 10 and 101 kPa (.1 and } atm. Under equilibrium conditions, moisture tension is defined as tbe equivalent negative gage pressure or suction corresponding lo soil moisture content. This test method determines the equilibrium moisture content retained in a soil subjected to a given soil water tension. This test method is not suitable for fine textured materials. This test measures water content at different pressures across a range. SATURATED HYDRAULIC CONDUCTIVITY FALLING HEAD RISING TAIL ASTM D 5084 "Standard Test Methods for Measurement of HydrauUc Conductivity of Saturated Porous Materials Using a Flexible Wall Permeameter** Metiiod A - Constant Head Method B - Falling Head, constant tailwater elevation Method C - Falhng Head, rising tailwater elevation Metiiod D - Constant Rate of Flow Method E - Constant Volume - Constant Head Method F - Constant Volume Falling Head Test methods per Darcys law at saturated pressure head greater than or equal to 0 (atmospheric.) As noted in the ASTM - The correlation between small sample size K measurements are seldom applicable lo field scale measurements. The systems are designed to determine the hydraulic conductivity of a material as rapidly as possible. .APPENDIX B - ICTM Plots of Flux Rates and Breakthrough Curves DRC NOTE: Figure 4-1 Plotted from HYDRUS Model 1 Output DRC NOTE: Figure 4-2 Plotted from MODFLOW Model 2 Output o in 050 lu —t <<? W —(-1 « UJK _| n t*3 =: _l m K E (1113) pv3H uncsajj DRC NOTE: Figure 4-3 Plotted from MODFLOW Model 2 Output o: o Si Si CO oil'" • IMZU IU US'- i ^' • I CO ^ at"' — ^ ^ *^ ^ o5 aP< (^p/taa) ajry xnij jajiA\ DRC NOTE: Figure 4-4 Plotted from HYDRUS Model 4 Output o O •A O "^ >* E a: o <-^ _i ifl-'d SJ z * o $ z UJ D cn S K O UJ P t E o z S X • si>: MO D E L - P R E D I C T E D C H L O IN V A D O S E Z O N IM M E D I A T E L Y A B O V E T BE N E A T H C E in tu 2 Ik (q/Sm) goi)Kjju»RO:> spuoiq;;) DRC NOTE: Figure 4-4 Plotted from HYDRUS Model 4 Output DRC NOTE: Figure 4-4 Plotted from HYDRUS Model 4 Output d.< m m oi z IU S cr KKO: zuJo IU tm*^ zSoz Ouiii:< OQ:UJ„ yoQ-uj " *"' [1/ Ul a: oissS< UD>-UJ [tl •> p -J UJ UJ ^ o 5 (l/Sni) UOT]EJ)ua3un3 nt^in^ DRC NOTE: Figure 4-4 Plotted from HYDRUS Model 4 Output APPENDDC C - CORE NEUTRALIZATION DATA Laboratories, Inc. 2773 Downtiill Drive Sleamboat Springs. CO 80487(800) 33'i-5493 Inorganic Analytical Results MWH America's Inc. Projecl ID: Sampte ID: L61917-01 ACZ Sample ID: L64240-01 Date Sampled: 08/03/07 09:55 Date Received; 08/03/07 Sample Matrix: Soil Soil AnalysiE Parainel9r Acid NeutralizatJon Potential (calc) Neutralization PotenLal as CaCOS EPA Method MB00/2-7B-054 1.3 MeOO/2-7a-05A 3.2.3 Soil Preparaiton Crush and Pulverize USDA No. 1,1972 Result .Qual XQ Units MDL PQL .Oate Analyst 1 0.1 t CaC03/Kt 1 5 08/08/07 16:22 calc % 0.1 0.5 08/04/07 9:36 Iwt EPAMctlipd Result- r. Qual'XQ Unitsr —MDL "PQL " " rDale --:. Analyst 08/03/07 14:00 Iwt REPIN.02.06.05.01 Please refer to Qualifier Reports for defaiV. L62140: Pa?e2of 17 Laboratories, Inc. 2773 Downtiill Dnvf SteBmboel Sorings, CO 30487(800) 334-5493 MWH America's Inc. Project ID; Sample ID: L61917-02 Inorganic Analytical Results ACZ Sample ID Date Sampled Date Received Sampie Matrix L64240-02 0&03/07 09:55 08)03107 Soil Soil Analysis Parameter Acid Neutralczalion Potential (calc) Neutraltzation Potential as CaC03 .EPA Mettiod M600/2-7B-0&4 1.3 M600/2-7B-054 3.2.3 Soil Preparstign Crush and Pulverize USDA No. 1,1972 Result /..QuDl. XQ Units WIDL.PQL. Date Analyst 1 0.1 1 CaC03/Kt 1 5 08/08/0716:22 calc % 0.1 0.5 08/04/07 10:07 Iwt EPA Method Result . Qual XQ Units- WIDL' PQL • • Dale" " AnnlyM 08/03A)7 14:03 Iwt REPIN.02.06.05 01 ' Please refer to Qualifier Reports for detail- Lei) 40: Page 3 of 17 Laboratories, Inc. 2773 Downtiill Dnve Steamboel Springs, CO 80487(800} 3S'i-5493 Inorganic Analytical Results MWH Amenca's Inc. Project ID: Sample ID: L61917-03 ACZ Sample ID: L64240-03 Dale Sampled: 08/03/07 09:55 Date Received: 08/03/07 Sample Matrix: Soil Soil AnalyBiE Parameter • . i. . EPA Method Acid Neiitralizatlon Potential (calc) Neutralization Polential as CaC03 M60D/2-78-05'} 1.3 M600/2-78-054 3.2.3 Soil Preparation Crush and Pulverize USDA No. 1, 1972 Result .. QiiBl XQ" . Units . .lUlDL PQL Date . Analyst 1 CBC03/Kt 1 5 08/08/07 16:22 calc V 0.1 0.5 08/04/07 10:39 M .•EPATJlclhod ~r Result- iCQualXQ"- Units". .MDL PQL . "^ Dale" 08/03^7 14:07 Iwt REPIN.02.06.05.01 * Please refer lo Qualifier Reports for detail. L62)40: Paee 4 of 17 Laboratories, Inc. 2773 Downhill Drive Steamboat Springs. CO 80487(800) 33^-5493 Inorganic Analytical Results MWH America's Inc, Project ID: Sample ID: L61917-04 ACZ Sample ID: L64240-04 Date Sampled: 08/03/07 09:55 Date Received: 08/03/07 Sample Matrix: Soil Soil Anaiyais Parameter Acid Neutralization Polential (calc) NBUtraiization Potential as CaC03 EPA Method M600/2-76-054 1.3 M600/2-7e-Q54 3.2.3 Soil Preparation Crush and Pulverize USDA No. 1.1972 EPA Method Result Qual Xq Units . MDL PQL Date Analyst A tCaC03/Kt 1 5 08/08/07 16:23 calc 0.4 B * % 0.1 0.5 08/04/0711;11 Iwt Result Qual XQ Units MDL -PQL . Date 1 Analyst 08/03/07 14:11 Iwl REPlN.02.0e .05.01 * Piease refer to Qualifier Reports for detail L6:]40; Page 5 of 17 Laboratories, Inc. 2773 Downhill Drive SlaamboBl Springs. CO 80487(800) 334-5493 Inorganic Analytical Results MWH America's Inc. Project ID: Sample ID: L61917-05 ACZ Sample ID Date Sampled Date Received Sample Matrix L64240-05 08/03/07 09:55 08/03/07 Soil Soil Analysis Parameter Acid Neutralizabon Potential (calc) Neutralization Potential as CaCOS EPA Method M600/2-78-054 1.3 M60O/2-7e-054 3.2.3 Soil Preparation Crush and Pulverize USDA No. 1, 1972 Result Qual XO. Units . MDL PQL . Dale . Analyst 0 t CaC03/Kt 1 5 08/08/07 16:23 calc % 0.1 0.5 08/04/07 11:43 Iwt EPA Method Result Qual XQ- Units MDL PQL Date . Analyst 08/03/07 14:15 iwt REPIN.02.06.05.01 * PIsase refer ro Qualifier Reports for detail L62J40: Page 6 of 17