DRC-2021-005546 - 0901a06880e6d464

Radioactive Material License Application / Federal Cell Facility

Page P-1 Appendix P April 9, 2021
Revision 0

APPENDIX P

NEPTUNE COVER INFILTRATION ANALYSIS

(Neptune, 2021b)

Radioactive Material License Application / Federal Cell Facility

Page P-2 Appendix P April 9, 2021
Revision 0

5.6.1.2 EnergySolutions’ Response to FPL Construction Specifications

Federal Cell Cover Interrogatories, Comment 6 – Frost Protection Layer:
“[D]iscuss the inherent difficulties of constructing a uniform material from such a specification,
and how consistency of layer properties will be maintained spatially and throughout time so that
the conditions inherent in the PA model are realized in the actual cover system over the service
life and compliance period of the proposed Federal Cell.”
EnergySolutions’ Response:
The Frost Protection Layer material property specification in Drawing 14004-C05 states that the material
is well graded bank run cobble/gravel/soil material with a maximum rock size of 16-inches. The
consistency of layers properties over time is addressed in the Neptune response to this interrogatory dated
December 3, 2020. As-built consistency is addressed in the FCF CQA/QC Manual work element for
Frost Protection Layer Placement (Specifications 123 thru 127). Specifically, Specification 125 requires
gradation testing be performed on the material using ASTM D5519 (“Standard Test Methods for Particle
Size Analysis of Natural and Man-Made Riprap Materials”) or C136 (“Standard Test Methods for Sieve
Analysis of Fine and Coarse Aggregates”) to ensure the material is well graded. The quality control
inspector will note any deficiencies or abnormalities and will notify the Project Manager to have the
material reworked to attain a more uniform gradation. Further, Specification 126 has quality control
observing placement of the Frost Protection Layer to ensure that fines are not concentrated in localized
areas. Again, if the quality control inspector notices any discrepancies, they will notify the Project
Manager and have operations re-distribute the material so that no localized concentration of finer material
is present. In both of these instances, the material will be re-inspected after operations.

NAC-0165_R0

Clive DU PA Model—Response
to DWMRC 12-3-2020
Comments
31 March 2021

Prepared by

NEPTUNE AND COMPANY, INC.
1435 Garrison St, Suite 201, Lakewood, CO 80215

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 ii

1. Title: Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
2. Filename: Clive DU PA Model - Response to DWMRC 12-3-2020 Comments final.docx
3. Description: Responses to UDEQ Letter “Comments on EnergySolutions Cover System
Described in the DU PA, Draft Federal Cell License Application,” dated December 3, 2020.
Name Date
4. Originator Dan Levitt, Paul Duffy, Gregg Occhiogrosso,
Dylan Boyle, and Matthew Bowers
31 March 2021
5. Reviewer Paul Black and Sean McCandless 31 March 2021
6. Remarks

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 iii
CONTENTS
CONTENTS ................................................................................................................................... iii
FIGURES ........................................................................................................................................ v
TABLES ....................................................................................................................................... viii
ACRONYMS AND ABBREVIATIONS ....................................................................................... ix
Executive Summary ......................................................................................................................... 1
1.0 Introduction ............................................................................................................................ 6
2.0 Revised Federal Cell Design .................................................................................................. 7
2.1 Embankment Footprint ..................................................................................................... 7
2.2 Top Slope Surface Layer Thickness ................................................................................. 9
2.3 Transition Zone and Side Slope Frost Protection Layer ................................................. 10
3.0 Results from the DU PA v1.4 Model ................................................................................... 11
3.1.1 Target Percolation Threshold .................................................................................... 13
4.0 Hybrid Cover Performance—HYDRUS-2D Modeling ....................................................... 14
4.1.1 Modeling Domain ..................................................................................................... 14
4.1.2 Layering .................................................................................................................... 16
4.1.3 Boundary Conditions ................................................................................................ 17
4.1.4 Mesh .......................................................................................................................... 18
4.1.5 Rooting Parameters ................................................................................................... 19
4.1.6 Atmospheric Input ..................................................................................................... 20
4.1.7 Initial Conditions ....................................................................................................... 20
4.2 Material Hydraulic Properties ........................................................................................ 21
4.3 Results ............................................................................................................................ 22
4.4 Discussion ....................................................................................................................... 28
5.0 UDEQ Comments and Responses ........................................................................................ 29
5.1 UDEQ Comment 1: Hybrid Cover Percolation Model .................................................. 29
5.1.1 Comment 1 Response ................................................................................................ 30
5.2 UDEQ Comment 2: HYDRUS Snowmelt Algorithm .................................................... 30
5.2.1 Comment 2 Response ................................................................................................ 31
5.2.1.1 Literature Review ................................................................................................ 31
5.2.1.2 Cover Test Cell Model ........................................................................................ 31
5.2.1.3 Comparison with Regional Snowpack Data ....................................................... 36
5.3 UDEQ Comment 3: Applying Cover Test Cell Data ..................................................... 42
5.3.1 Comment 3 Response ................................................................................................ 42
5.4 UDEQ Comment 4: Regression Model .......................................................................... 45
5.4.1 Comment 4 Response ................................................................................................ 45
5.4.1.1 Variation of Ksat of the Radon Barrier (Layers 4 and 5) in Model v1.4 .............. 46
5.4.1.2 Exploration of Ksat Variation in Upper Cover Soils Layers 1 and 2 ................... 46
5.5 UDEQ Comment 5: Hydraulic Properties ...................................................................... 50
5.5.1 Comment 5 Response ................................................................................................ 50
5.6 UDEQ Comment 6: FPL Properties ............................................................................... 50
5.6.1 Comment 6 Response ................................................................................................ 50
5.6.1.1 FPL Properties ..................................................................................................... 50
5.6.1.2 FPL Construction Specifications ......................................................................... 51
5.6.1.3 Long-Term Durability of the Frost Protection Layer .......................................... 51
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 iv
5.7 UDEQ Comment 7: Capillary Break .............................................................................. 55
5.7.1 Comment 7 Response ................................................................................................ 55
5.8 UDEQ Comment 8: Water Balance Graphs ................................................................... 62
5.8.1 Comment 8 Response ................................................................................................ 62
5.9 UDEQ Comment 9: Abstraction Model ......................................................................... 71
5.9.1 Comment 9 Response ................................................................................................ 71
5.10 UDEQ Comment 10: Characterizing Uncertainty .......................................................... 74
5.10.1 Comment 10 Response .............................................................................................. 74
5.11 UDEQ Comment 11: Tails of the Distribution ............................................................... 77
5.11.1 Comment 11 Response .............................................................................................. 77
5.12 UDEQ Comment 12: Climate Record and Comparison With Other Sites ..................... 80
5.12.1 Comment 12 Response .............................................................................................. 80
5.12.1.1 Climate Record .................................................................................................... 80
5.12.1.2 Comparison Across Sites .................................................................................... 89
6.0 Conclusion ............................................................................................................................ 96
7.0 Attachments .......................................................................................................................... 96
8.0 References ............................................................................................................................ 96

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 v
FIGURES
Figure ES-1. DU PA v1.4 dose results, R313-25-20 dose limit, and typical background dose. ..... 4
Figure 1. Revised (2021) Federal Cell footprint (from drawing 14004-C01, rev 2). ...................... 8
Figure 2. Former (2020) Federal Cell footprint (from drawing 14004-C-01, rev. 0). ..................... 9
Figure 3. Top Slope Detail (from drawing 14004-C05, rev. 1). .................................................... 10
Figure 4. Transition Zone (2021) Detail (from drawing 14004-C05, rev. 1). ............................... 10
Figure 5. Former Transition Zone (2020) Detail (from drawing 14004-C05, rev. 0). .................. 11
Figure 6. DU PA v1.4 dose results, R313-25-20 dose limit, and typical background dose. ......... 13
Figure 7. Cross section of federal cell, including a 135 m (448.4 ft) of top slope, and detail of the
transition zone of the hybrid cover design. ................................................................. 15
Figure 8. Full length cover design and abbreviated hybrid cover model. ..................................... 16
Figure 9. Model layering and materials for the 2D hybrid cover. ................................................. 17
Figure 10. Boundary conditions used in the 2D hybrid cover model. ........................................... 18
Figure 11. Node spacing detail shown for a portion of the top slope. ........................................... 19
Figure 12. Rooting depths shown for the top slope, transition zone, and side slope. ................... 20
Figure 13. Sorted results from the v1.4 HYDRUS modeling with the maximum, 2nd highest,
median, and minimum percolation results indicated. .................................................. 21
Figure 14. Percolation out of the lower radon barrier, as a function of lateral distance in the 2D
hybrid cover model. ..................................................................................................... 24
Figure 15. Tension at the bottom of the lower radon barrier, as a function of lateral distance in
the 2D hybrid cover model. ......................................................................................... 25
Figure 16. Water content at the bottom of the lower radon barrier, as a function of lateral
distance in the 2D hybrid cover model. ....................................................................... 26
Figure 17. Annual percolation out of the lower radon barrier for the entire 135 m length of
proposed ET cover (shown in Figure 8). Results for the Maximum and 2nd highest
show a minor amount of coarseness in the results after 300 years. This is due to the
precision of the model output in HYDRUS, and not a change in behavior of the
model. .......................................................................................................................... 27
Figure 18. Test Cell (reprinted from EnergySolutions (2020)). .................................................... 32
Figure 19. Layers in the 1D Test Cell model. ............................................................................... 33
Figure 20. Monthly tip data collected from the Cover Test Cell. .................................................. 35
Figure 21. 14-day periods of calculated vs observed snow layers at Dugway, UT in temporal
order of occurrence. ..................................................................................................... 38
Figure 22. 14-day periods of calculated vs observed snow layers at Dugway, UT in temporal
order of occurrence. ..................................................................................................... 39
Figure 23. Average daily calculated and observed snow layers at Dugway, UT. ......................... 40
Figure 24. Daily Mean Temperature at Dugway, UT averaged across the period of record. ....... 41
Figure 25. Percolation as a function of Ksat for 45 simulations. .................................................... 47
Figure 26. Percolation vs Ksat with h50 = 1500 cm for the same 45 model runs presented above. 49
Figure 27. Reproduced from A.1 Clive Pit Wall Interpretation (C. G. Oviatt, unpublished data)
and stratigraphic comparison with quarry wall studies from Neptune (2020b). ......... 54
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 vi
Figure 28. SWCC for the FPL as modeled in v1.4. ....................................................................... 57
Figure 29. Unsaturated conductivity as a function of pressure head for a coarse and fine material.
..................................................................................................................................... 58
Figure 30. Specific moisture capacity for the FPL as modeled in v1.4. ........................................ 59
Figure 31. Evaporation zone pressure head and flux through the FPL vs time for a wet period in
Simulation 20. The red dotted line is drawn at a pressure head of -250 cm. Flux
values are negative for downward flow. ...................................................................... 60
Figure 32. SWCCs for the FPL and two realizations of the evaporation zone layer. The tan
horizontal line is drawn at a tension of 250 cm, while the dotted lines indicate the
corresponding water content for the evaporative zone curves. ................................... 61
Figure 33. Additional nodes were added to the model domain to improve detail in the output for
each layer. Red dots represent observation nodes used in DU PA v1.4; black circles
represent the additional nodes. .................................................................................... 64
Figure 34. Water content at observations nodes on selected days around a large 2.7 cm storm
event. ........................................................................................................................... 65
Figure 35. Water content at observations nodes following a period of high precipitation
frequency that results in flow through the frost protection layer. ............................... 66
Figure 36. Ten years of daily water content in the v1.4 simulation #20 model. ........................... 67
Figure 37. Ten years of daily tension in the v1.4 simulation #20 model. ..................................... 68
Figure 38. Three years of daily water content in the v1.4 simulation #20 model. ........................ 69
Figure 39. Three years of daily tension in the v1.4 simulation #20 model. .................................. 69
Figure 40. Three years of daily (upward) fluxes in the v1.4 simulation #20 model. .................... 70
Figure 41. Three years of daily (downward) fluxes in the v1.4 simulation #20 model. ............... 70
Figure 42. Scatterplot of percolation values computed from both the regression model and
HYDRUS using the same pairs of α and n that were randomly generated. ................ 74
Figure 43. Comparison of Bingham Environmental (1991) water content data with water content
calculated using the regression equation for the DU PA GoldSim model and with the
results of the 20 HYDRUS simulations. Figure 26 of EnergySolutions (2018). ......... 78
Figure 44. Water content in the evaporation zone from 50 HYDRUS simulations used in DU PA
v1.4. ............................................................................................................................. 79
Figure 45. Water content in the evaporation zone from 50 HYDRUS simulations used in DU PA
v1.4. ............................................................................................................................. 80
Figure 46. Total precipitation over various time periods for the 1000y and 100y records. .......... 82
Figure 47. Flux at the bottom of the cover and precipitation for both the 100-year (top) and 1000-
year (bottom) meteorological records for Simulation 1 of 50. Only the last cycle of
the meteorological record is shown. Vertical scales are the same for both plots. ....... 84
Figure 48. Water stress models. .................................................................................................... 85
Figure 49. Histograms of sets of 50 simulations using the 100-year meteorological record (top),
the 1000-year meteorological record with Model v1.4 root water uptake parameters
(middle), and the 1000-year meteorological record with h50 set to 1500 cm (bottom).
..................................................................................................................................... 87
Figure 50. Simulation by simulation comparison of percolations derived from scenarios with h50
set to 200 cm (blue) and with h50 set to 1500 cm (green). ........................................... 88
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 vii
Figure 51. Layering of ET cover systems at Clive, Monticello, and Blanding, Utah. .................. 90

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 viii
TABLES
Table 1. Tension (pressure head) used for initial conditions in each model. Values are shown in
cm. ............................................................................................................................... 20
Table 2. Material hydraulic parameters used in DU PA v1.4 modeling. ...................................... 22
Table 3. Parameter sets for α, n, and Ksat used in the 2D hybrid cover models. ........................... 22
Table 4. Average annual percolation out of the radon barrier (model years 400–500); v1.4 results
vs 2D hybrid cover results. .......................................................................................... 28
Table 5. Comparison of the layering on the top and side slopes of the original Federal Cell design
and the revised design. ................................................................................................ 30
Table 6. Material hydraulic properties. ......................................................................................... 34
Table 7. Results of the 1D Test Cell model. .................................................................................. 35
Table 8. Engineering properties of cover layers for DU PA v1.4, the Cover Test Cell, and
NUREG/CR-7028. ....................................................................................................... 44
Table 9. Random sample from trimmed set of values for α, n, and percolation from both
HYDRUS and the regression model. ........................................................................... 73
Table 10. Summary statistics for 100-year and 1000-year climate records. ................................. 81
Table 11. Engineering properties of cover layers in the Clive Federal Cell, DU PA v1.4. ........... 91
Table 12. Engineering properties of cover layers in the Monticello disposal facility. .................. 92
Table 13. Engineering properties of cover layers in the Blanding White Mesa Mill Tailings
Facility. ........................................................................................................................ 93
Table 14. Precipitation and percolation data for the Clive Cover Test Cell, Monticello, and
Blanding facilities. Clive precipitation average calculated for the years 2002–2016 to
match with Cover Test Cell period of service; site average across 28-year
meteorological record is 217.41 mm/yr. ...................................................................... 94

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 ix
ACRONYMS AND ABBREVIATIONS
bgs below ground surface
CQA/QC Construction Quality Assurance/Quality Control
CWCB Colorado Water Conservation Board
DEQ (Utah) Department of Environmental Quality
DU depleted uranium
DWMRC Division of Waste Management and Radiation Control
ET evapotranspiration
FPL Frost Protection Layer
GWPL groundwater protection limits
HELP Hydrologic Evaluation of Landfill Performance model
LLRW low-level radioactive waste
MOP member of the public
MPV maximum permissible velocity
NRC (United States) Nuclear Regulatory Commission
PA performance assessment
PAWG Performance Assessment Working Group
PE potential evaporation
PET potential evapotranspiration
QA/QC quality assurance/quality control
SCS Soil Conservation Service
SER Safety Evaluation Report
SWAT Soil and Water Assessment Tool
SWCC soil water characteristic curve
SWE snow water equivalent
TEDE total effective dose equivalent
UDEQ Utah Department of Environmental Quality

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 1
Executive Summary
The Clive depleted uranium (DU) performance assessment (PA) evaluates the range of likely
impacts of disposal of DU in a new Federal Cell to be located in the southwest corner of the
licensed area. The DU PA is created as a systems-level model using the GoldSim probabilistic
modeling platform and is currently at version 1.4. The DU PA v1.4 model and supporting
documentation have been evaluated by the Utah Department of Environmental Quality (UDEQ)
and their contractor, SC&A Inc., for a number of years since its initial publication in 2015
(Neptune 2015e).
The current round of comments (Utah DEQ 2020) ask that the “hybrid” cover design introduced
in the 2020 response to interrogatories (Neptune 2020a) be subject to additional verification. The
hybrid cover features an evapotranspiration (ET) cover system of native soils and vegetation on
the large top slope area; and rip rap armoring of the steeper side slope area. The ET cover has
been selected for its superior performance in minimizing percolation of atmospheric precipitation
into the waste; while the rock armor cover has been selected for its improved assurance for
minimizing the potential for erosion of the steeper side slopes.
Additional hydrological modeling of this hybrid cover system has been performed using the
HYDRUS-2D modeling platform. As detailed below, this modeling projects that the rip rap side
slopes are expected to have higher percolation than the ET cover top slope; and that the impact
of this limited area of higher percolation remains within the bounds of previous analyses that
demonstrate acceptable embankment performance.
It is a truism when modeling complex systems such as radioactive waste disposal sites that no
model is perfect, but some models are useful. “Useful,” in this context, means that the model is a
reasonable representation of the system as currently understood and conceptualized, with the
acknowledgement that uncertainties will always remain. Important uncertainties are captured in
the probability distributions of the input parameters.
Decisions can and should be made based on the current model and its results. Standard PA
practice calls for the model to be routinely reviewed and updated as new information and data
from monitoring programs or new relevant research becomes available. This could include new
information about site characteristics, the waste itself, and the process models that have been
abstracted into the systems-level probabilistic model.
Updates to the model can lead to adaptive decision making if new model results indicate a need
to change a current decision. For the Clive site over the next few decades before final closure,
this could simply result, for example, in a change in cover design or procedures governing waste
placement. EnergySolutions is required to provide a surety fund that would accommodate
changes under such an adaptive management program. Alternatively, adaptive updates to the PA
could also demonstrate that initial constraints may safely be relaxed, such as that requiring DU
waste be to placed at an elevation below current native grade. What is described here might be
called a PA maintenance program, the details of which would normally be captured by License
condition outlining the schedule and expectations for routine updates to the PA.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 2
This concept is important in the context of DU PA v1.4 because a number of the outstanding
questions rest all or in part on research that has emerged; and continues to emerge, since
completion of this version of the model in 2015. For example, updated information on the
performance of a somewhat similar ET cover design in Monticello, UT is anticipated to be
available concurrent with preparation of this response document. This emerging research is
certainly of interest to the model and appropriate to incorporate in PA maintenance; but is shown
in the attached responses to support rather than change the fundamental conclusions of DU PA
v1.4.
The Final Report for the Clive DU PA Model, Clive DU PA Model v1.4 (Neptune 2015e)
provides the following summary of DU PA v1.4 results for the quantitative compliance period of
10,000 years. Additional work preparing interrogatory and comment responses after creation of
version 1.4 have not changed the principal analysis and reported conclusions.
Compliance with the performance objectives for the inadvertent intruder dose of 500 mrem
in a year and for the MOP of 25 mrem in a year is clearly established for all three types of
potential future receptors. This indicates that for the disposal configuration where DU wastes
are placed below grade, doses are expected to remain well below applicable dose
thresholds…
Results are also available for the offsite (MOP) receptors. None of the 95th percentile dose
estimates for these receptors exceeds 1 mrem in a year, and all of the peak mean dose
estimates are at or below 0.1 mrem in a year.
Table ES-1. Peak TEDE: statistical summary
peak TEDE (mrem in a yr) within 10,000 yr
receptor mean median
(50th %ile) 95th %ile
ranch worker 6.2E-2 5.1E-2 1.5E-1
hunter 4.5E-3 3.8E-3 9.9E-3
OHV enthusiast 8.4E-3 7.5E-3 1.8E-2
Results are based on 10,000 realizations of the Model.
TEDE: Total effective dose equivalent
For those radionuclides for which GWPLs exist, as specified in the facility’s permit (UWQB
2009), results are shown in Table ES-2. For all such radionuclides compliance with the
GWPLs is clearly demonstrated…
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 3
Table ES-2. Peak groundwater activity concentrations within 500 yr, compared to
GWPLs
peak activity concentration within 500 yr (pCi/L)
radionuclide GWPL1
(pCi/L) mean median
(50th %ile) 95th %ile
90Sr 42 0 0 0
99Tc 3790 26 4.3E-2 150
129I 21 1.7E-2 4.3E-11 1.1E-1
230Th 83 2.2E-28 0 0
232Th 92 1.4E-34 0 0
237Np 7 1.5E-19 0 3.7E-27
233U 26 5.6E-24 0 3.9E-28
234U 26 2.1E-23 0 2.2E-28
235U 27 1.6E-24 0 2.0E-29
236U 27 2.7E-24 0 3.3E-29
238U 26 1.5E-22 0 1.8E-27
1GWPLs are from UWQB (2009) Table 1A.
Results are based on 10,000 realizations of the Model.

Figure ES-1 displays Table ES-1 dose results graphically in context with the dose limit of 25
mrem/year for members of the public under R313-25-20. Typical background radiation dose is
also provided on this figure as a point of reference. DU PA v1.4 results are 2 to 3 orders of
magnitude below the dose limit.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 4

Figure ES-1. DU PA v1.4 dose results, R313-25-20 dose limit, and typical background dose.
It is worth noting that many of the open questions regarding DU PA v1.4 reflect new or revised
scenarios asked of the modeling. What if snowmelt behaves differently than the HYDRUS
percolation model predicts? What if the climate record is different from that initially modeled?
What if the material properties of the cover layers vary in this or that direction? Throughout the
review process, these and similar questions have been asked in a way that challenges the
modeling and its overall result that very limited percolation is expected through the
embankment; and thus, very limited impact to groundwater beneath the embankment is
projected.
At the same time, there are many scenarios to suggest that the percolation modeling has been
performed in a way that over-predicts potential impact to groundwater resources. What if desert
plants, adapted to arid conditions, are much more effective at removing water from surface soils
than modeled? What if the basic model structure of HYDRUS, which assumes essentially no
surface runoff of precipitation, overstates the amount of water available to infiltrate? What if this
or that material property understates its ability to minimize percolation?
The results reported above are based on 10,000 realizations of the DU PA model—each of these
realizations could be considered to reflect a unique what if scenario for the model. Through such
modeling, the central tendency of the system is evaluated; and the evaluation indicates very low
potential for radiological doses or groundwater quality impacts.
1.E-03
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
1.E+03
Mean Median 95th %ile
Do
s
e

(
m
r
e
m
/
y
e
a
r
)
Ranch worker (DU PA v1.4)
Hunter (DU PA v1.4)
OHV enthusiast (DU PA v1.4)
Dose limit
Typical radiation dose in U.S.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 5
Ultimately, the DU PA v1.4 evaluates an above-grade embankment located in a terminal desert
basin. Annual potential evapotranspiration in Utah’s west desert far exceeds precipitation; and
the site is located above a largely stagnant aquifer in an area with limited natural groundwater
recharge (Bingham Environmental 1994; Stantec 2020). Projections of negligible percolation
into and through an above-grade embankment in this context are not only reasonable, they are to
be expected.
It is appropriate to ask a number of what if questions when considering performance assessment
for radioactive waste disposal. However, at the end of the day it is also vital to recognize that the
performance assessment does not pretend to predict the future; rather, it projects performance
within a formalized scenario derived by NRC in establishing regulations for the safety of
radioactive waste disposal. The scenario itself is understood to be an artifact. NRC (2000)
guidance on performance assessment methodology cautions that “…consideration given to the
issue of evaluating site conditions that may arise from changes in climate or the influences of
human behavior should be limited so as to avoid unnecessary speculation.”
DU PA v1.4 demonstrates compliance with the dose and groundwater protection requirements of
Utah regulations relating to DU disposal. The interrogatory and response process has added to
the record supporting these conclusions but has not caused the quantitative model itself to require
revision. Accordingly, DU PA v1.4 remains a reasonable basis for demonstrating compliance of
the disposal facility.
Compliance with UAC R313-25-9(5)(a) is affirmed by DU PA v1.4, together with the supporting
documentation as supplemented by the interrogatory/response cycle.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 6
1.0 Introduction
Beginning in 2009, EnergySolutions contracted Neptune to create a probabilistic performance
assessment (PA) for the disposal of large quantities of depleted uranium (DU) at their Clive,
Utah low-level radioactive waste (LLRW) disposal facility.
The initial model was submitted as version 1.0 on June 1, 2011 (Neptune 2011) and was revised
to version 1.2 on June 5, 2014 (Neptune 2014). A Safety Evaluation Report (SER) based on
review of version 1.2 was issued by UDEQ in April 2015 (SC&A 2015).
On November 25, 2015, EnergySolutions submitted Radioactive Material License UT2300249:
Safety Evaluation Report for Condition 35.B Performance Assessment; Response to Issues
Raised in the April 2015 Draft Safety Evaluation Report (EnergySolutions 2015). This document
included version 1.4 of the DU PA (Neptune 2015e), prepared in response to open primary and
new interrogatories raised after development and DWMRC review of version 1.0; included in
Appendix C and Appendix B, respectively, of the SER.
On May 11, 2017, UDEQ provided Amended and New Interrogatories Related to Clive DU PA
Modeling Report Version 1.4 Dated November 2015 (Utah DEQ 2017). This document contains
clarifications to the original interrogatories from DU PA version 1.0 that remained open,
clarifications to the interrogatories newly raised with version 1.2 and new interrogatories
introduced with version 1.4 of the DU PA.
On April 2, 2018, EnergySolutions submitted Radioactive Material License UT2300249:
Responses to Amended and New Interrogatories Related to Clive DU PA Modeling Report
Version 1.4 Dated November 2015 (EnergySolutions 2018). As suggested by UDEQ, this
document included seven topical reports organized consistently with the themes expressed in the
interrogatory package (Utah DEQ 2017).
On July 25, 2019, UDEQ provided Depleted Uranium Performance Assessment (DU PA); Clive
Facility; Model Version 1.4 Amended Interrogatories; Radioactive Materials License #2300249
(Utah DEQ 2019). This document contains amended interrogatories of open issues regarding
version 1.4 of the DU PA model, closes several interrogatories, and introduces two more new
interrogatories. Neptune responded to these interrogatories on April 24, 2020 (Neptune 2020a).
In the 2020 response to interrogatories, a new “hybrid” cover design was introduced. This cover
design incorporates an evapotranspiration cover on the top slope; and a rock armor cover on the
side slope.
On December 3, 2020, UDEQ provided “Comments on EnergySolutions Cover Design System
Described in the DU PA, Draft Federal Cell License Application” (Utah DEQ 2020). This letter
poses 12 technical questions relating to the hybrid cover design.
Full text of each comment is quoted using blue text in Arial font, size 10.5 pt, and is indented to
visually distinguish the comment from the response. An example is shown below:
Sample format for quoting comment text.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 7
This response document does not comment on additional UDEQ comments provided under
separate cover “RE: Technical Report” dated January 28, 2021 (Utah DEQ 2021). This second
comment document includes concerns relating to erosion and embankment stability. Neptune is
preparing a response to those issues as a separate report (Neptune 2021).
Section 2.0 summarizes changes to the Federal Cell design. Section 3.0 discusses results from
the DU PA v1.4 model, and how these results relate to the revised design and additional analyses
performed. Section 4.0 presents a new HYDRUS 2D model of the hybrid cover. Section 5.0
responds point by point to the 12 comments presented in Utah DEQ 2020. Conclusions are found
in Section 6.0
2.0 Revised Federal Cell Design
In the 2020 response to interrogatories, the Federal Cell cover design was revised to adopt a rock
armor cover for the side slopes (Neptune 2020a). The ET cover previously analyzed for the top
slopes is retained. The ET cover has been selected for its superior performance in minimizing
percolation of atmospheric precipitation into the waste, while the rock armor cover has been
selected for its improved assurance in minimizing the potential for erosion of the steeper side
slopes.
The 2020 design has been further revised as discussed below. These revisions have been carried
through new and updated modeling as applicable. Updated drawings 14004-C01 through 14004-
C05 are included as Attachment 1.
2.1 Embankment Footprint
EnergySolutions has revised the embankment footprint in order to provide greater separation
between the Federal Cell and the 11e.(2) Cell to the east. The revised footprint is slightly
narrower east to west and slightly longer north to south than it was in prior drawings. The grade
of the top and side slope areas is unchanged; and the thickness of the cover layers are unchanged
from the drawings submitted previously.
The revised embankment footprint has slightly shorter top slope lengths; and a longer
embankment crest. Figure 1 shows the revised embankment footprint; Figure 2 shows the version
previously analyzed. Side slope lengths are unchanged. These changes result in a peak
embankment elevation at the crest that is one foot lower than that of the previous footprint1.
The current embankment footprint has been considered in HYDRUS 2D modeling performed to
address UDEQ comments. Prior hydrological modeling is unaffected by this change, since that
was conducted as one-dimensional modeling not connected to slope length.

1 Embankment thickness is considered in DU PA v1.4 in the context of radon emanation. This is modeled as an
average thickness of material between the DU waste and the surface, calculated in DU PA v1.4 to be 39.7 feet. The
revised embankment footprint changes this dimension to be 39.6 feet.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 8
The current embankment footprint has also been incorporated in SIBERIA modeling of erosion,
with results to be presented under separate cover.

Figure 1. Revised (2021) Federal Cell footprint (from drawing 14004-C01, rev 2).
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 9

Figure 2. Former (2020) Federal Cell footprint (from drawing 14004-C-01, rev. 0).
2.2 Top Slope Surface Layer Thickness
In the 2020 design change to utilize rip rap armoring on the side slopes, the top slope surface
layer thickness was increased from 6 inches to 12 inches. This change slightly increases the
storage capacity of the ET cover design. HYDRUS 2D evaluation of the full hybrid cover
includes this revision. Figure 3 provides the top slope layering.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 10

Figure 3. Top Slope Detail (from drawing 14004-C05, rev. 1).
2.3 Transition Zone and Side Slope Frost Protection Layer
The transition zone from the ET cover top slope to the rip rap cover side slope has been revised
from that presented in the 2020 design. The revision moves the transition zone to the shoulder of
the embankment and reduces its width. These changes were made to reduce the impact of
increased percolation through the rip rap portion of the cover. Figure 4 provides the transition
zone detail as currently modeled; Figure 5 displays the prior design.

Figure 4. Transition Zone (2021) Detail (from drawing 14004-C05, rev. 1).
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 11

Figure 5. Former Transition Zone (2020) Detail (from drawing 14004-C05, rev. 0).
Concurrent with this change, the material used as the Frost Protection Layer (FPL) for the side
slope has been changed. The current design specifies this material to be the same bank run as
used for the FPL on the top slope; where the 2020 design used native clay soils for the side slope
frost protection layer. This material was changed in order to ensure consistent drainage
properties from the top slope onto the side slope at this layer in the cover system. The change
also improves constructability of the transition zone.
As discussed in Section 4, HYDRUS 2D modeling of the full hybrid cover includes these
revisions.
3.0 Results from the DU PA v1.4 Model
Since initial submittal of DU PA v1.4 (Neptune 2015e), many technical issues have been
resolved relating to the probabilistic performance assessment (PA) model, through the
interrogatory/response process summarized in Section 1.0. In this report, Neptune presents
analyses of the hybrid cover design performance in relation to the percolation assumptions and
results embedded in v1.4 of the DU PA. If the hybrid cover is demonstrated to have comparable
performance to that of the ET cover modeled in DU PA v1.4, then the results of DU PA v1.4 can
be considered to hold as well.
The Final Report for the Clive DU PA Model, Clive DU PA Model v1.4 (Neptune 2015e)
provides the following summary of DU PA v1.4 results for the quantitative compliance period of
10,000 years. Additional work preparing interrogatory and comment responses after creation of
version 1.4 (Neptune 2015e) have not changed the principal analysis and reported conclusions.
Compliance with the performance objectives for the inadvertent intruder dose of 500 mrem
in a year and for the MOP of 25 mrem in a year is clearly established for all three types of
potential future receptors. This indicates that for the disposal configuration where DU wastes
are placed below grade, doses are expected to remain well below applicable dose
thresholds…
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 12
Results are also available for the offsite (MOP) receptors. None of the 95th percentile dose
estimates for these receptors exceeds 1 mrem in a year, and all of the peak mean dose
estimates are at or below 0.1 mrem in a year.
Table ES-1. Peak TEDE: statistical summary
peak TEDE (mrem in a yr) within 10,000 yr
receptor mean median
(50th %ile) 95th %ile
ranch worker 6.2E-2 5.1E-2 1.5E-1
hunter 4.5E-3 3.8E-3 9.9E-3
OHV enthusiast 8.4E-3 7.5E-3 1.8E-2
Results are based on 10,000 realizations of the Model.
TEDE: Total effective dose equivalent
For those radionuclides for which GWPLs exist, as specified in the facility’s permit (UWQB
2009), results are shown in Table ES-2. For all such radionuclides compliance with the
GWPLs is clearly demonstrated.
Table ES-2. Peak groundwater activity concentrations within 500 yr, compared to
GWPLs
peak activity concentration within 500 yr (pCi/L)
radionuclide GWPL1
(pCi/L) mean median
(50th %ile) 95th %ile
90Sr 42 0 0 0
99Tc 3790 26 4.3E-2 150
129I 21 1.7E-2 4.3E-11 1.1E-1
230Th 83 2.2E-28 0 0
232Th 92 1.4E-34 0 0
237Np 7 1.5E-19 0 3.7E-27
233U 26 5.6E-24 0 3.9E-28
234U 26 2.1E-23 0 2.2E-28
235U 27 1.6E-24 0 2.0E-29
236U 27 2.7E-24 0 3.3E-29
238U 26 1.5E-22 0 1.8E-27
1GWPLs are from UWQB (2009) Table 1A.
Results are based on 10,000 realizations of the Model.
Figure 6 displays Table ES-1 dose results graphically in context with the dose limit of
25 mrem/year for members of the public under R313-25-20. Typical background radiation dose
is also provided on this figure as a point of reference. DU PA v1.4 results are 2 to 3 orders of
magnitude below the dose limit.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 13

Figure 6. DU PA v1.4 dose results, R313-25-20 dose limit, and typical background dose.
DU PA v1.4 demonstrates compliance with the dose and groundwater protection requirements of
Utah regulations relating to DU disposal. The interrogatory and response process has added to
the record supporting these conclusions but has not caused the quantitative model to require
revision. Accordingly, DU PA v1.4 remains the basis for demonstrating compliance of the
disposal facility.
Compliance with UAC R313-25-9(5)(a) is affirmed by DU PA v1.4, together with the supporting
documentation as supplemented by the interrogatory/response cycle.
Ultimately, the DU PA v1.4 evaluates an above-grade embankment located in a terminal desert
basin. Annual potential evapotranspiration in Utah’s west desert far exceeds precipitation; and
the site is located above a largely stagnant aquifer in an area with limited natural groundwater
recharge. Projections of negligible percolation into and through an above-grade embankment in
this context are not only reasonable, they are to be expected.
3.1.1 Target Percolation Threshold
Within this response document, scenarios from the comments are sometimes evaluated against a
percolation criteria of 1 mm/yr. This criteria refers to the results of the DU PA v1.4XXX
GoldSim model described in Neptune (2015a). ET cover percolation inputs to this GoldSim
model were generated using a HYDRUS 1D model with a monolayer ET cover. In other words,
1.E-03
1.E-02
1.E-01
1.E+00
1.E+01
1.E+02
1.E+03
Mean Median 95th %ile
Do
s
e

(
m
r
e
m
/
y
e
a
r
)
Ranch worker (DU PA v1.4)
Hunter (DU PA v1.4)
OHV enthusiast (DU PA v1.4)
Dose limit
Typical radiation dose in U.S.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 14
in v1.4XXX, all layers of the modeled cover have identical material properties. The average
annual percolation through the monolayer cover is 0.91 mm/yr using hydraulic properties
generated from the Hydraulic Props Calc.xls file provided by Dr. Benson, and using inputs for a
fine-grained material from Benson et al. (2011) (NUREG/CR-7028).
Results of the DU PA v1.4XXX GoldSim model indicate that the Tc-99 median concentration is
below the groundwater protection limit (GWPL) of 3790 pCi/L, while the mean and 95th
percentile results exceed the GWPL. Rancher doses are slightly lower in the v1.4XXX model
because the increased percolation suppresses upward radon flux.
This monolayer ET cover scenario is not physically plausible; however, it provides a useful
metric on the percolation limit at which the Tc-99 groundwater protection limit may be
exceeded. The v1.4XXX results indicate that when average annual percolation exceeds
0.91 mm/yr (rounded to 1 mm/yr), mean and 95th percentile concentrations of Tc-99 potentially
exceed the GWPL of 3790 pCi/L.
4.0 Hybrid Cover Performance—HYDRUS-2D Modeling
The percolation results presented in the v1.4 HYDRUS modeling are based upon a 1D model of
an ET cover (Neptune 2015b). This modeling effort reflects a previous design for the federal cell
that included an ET cover on both the top and side slopes (see Section 2.0 Revised Federal Cell
Design). It was demonstrated that in this system lateral flow would be negligible, and therefore a
1D modeling approach would be sufficient to assess cover performance (Neptune 2015f).
Since 2014, modifications have been made to the Federal Cell design (see Section 2.0). The top
layer of the side slope consists of 45.7 cm (18 in) riprap rock, underlain by 30.5 cm (12 in) of a
filter layer. The remaining layers are identical to those specified in the ET portion of the hybrid
cover (Figure 8). Additionally, the surface layer of the ET portion of the hybrid cover design
increased from 15.2 to 30.6 cm (6 to 12 inches). The main concern discussed in the current
comments (Utah DEQ 2020) is a need to determine whether the results presented in the v1.4
HYDRUS modeling are still sufficiently representative of cover performance, given the revised
design of the federal cell cover.
To investigate this concern, a 2D model of the updated hybrid cover design has been developed.
The primary goal of the 2D model is to calculate percolation out of the cover above the waste
zone, and to compare these results to the percolation results presented in the v1.4 HYDRUS
modeling.
4.1.1 Modeling Domain
A cross section of the federal cell is shown in Figure 7, including detail of the configuration of
the ET cover, transition zone, and side slope in the hybrid cover design. To improve
computational efficiency and reduce model run times, only a portion of the hybrid cover design
is modeled (Figure 8). Only the ET portion of the hybrid cover is shortened. In the 2D hybrid
cover model, 56 m of ET cover is included as well as the full 54 m extent of the side slope. The
choice of this abbreviated domain was supported by several model runs with a variety of domain
lengths to ensure that there were not significant boundary effects associated with the truncation.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 15

Figure 7. Cross section of federal cell, including a 135 m (448.4 ft) of top slope, and detail of
the transition zone of the hybrid cover design.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 16

Figure 8. Full length cover design and abbreviated hybrid cover model.
4.1.2 Layering
A representative view of the modeled geometry of each layer for the ET cover, transition zone,
and side slope is shown in Figure 9. Because extremely coarse materials (i.e., riprap) are
problematic to model when paired with an atmospheric boundary (typically requiring very small
mesh refinement and/or timesteps, both which are impractical at the scale of the 2D model and
length of simulation), and to simplify the model domain using only v1.4 modeling parameters,
the material hydraulic parameters for the FPL are used for the top 3 layers of the side slope, as
shown in Figure 9.
With this parameterization of the side slope, Ksat is high enough that no runoff occurs throughout
the simulation, and therefore all precipitation is accounted for in the model domain. In addition,
water content on the side slope layer generally stays below 10%, thus there is no significant
storage in the “riprap” or “filter” layers using this parameterization. Lastly, values of Ksat of
riprap materials are typically very high, however the large pore spaces of this material will be
filled in with fines over time. Therefore, any measured K for the riprap would no longer be
representative of long-term conditions. For the purposes of exploring the hybrid cover design,
the frost protection parameters are therefore considered appropriate for simulating the relatively
higher levels of percolation expected over this portion of the cover.
190 m total length with 136 m of ET cover and 54 m of side slope
110 m total length with 56 m of ET cover
Hybrid cover design
Abbreviated hybrid cover model
ET cover
136 m
Side slope
54 m
ET cover
56 m
Side slope
54 m
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 17

Figure 9. Model layering and materials for the 2D hybrid cover.
4.1.3 Boundary Conditions
An atmospheric boundary condition is used along the top surface of the model and a free
drainage boundary condition along the bottom of the model, as specified in the v1.4 HYDRUS
modeling (Neptune 2015b). A “No Flux” (i.e., no flow) boundary condition is used on the left
boundary of the model domain and along the top of the right side of the model domain as the
model pinches out, as shown in Figure 10. The HYDRUS interface includes a feature called a
“mesh line” which allows the user to define a continuous series of nodes in the domain; the water
flux across the surface is reported at every time step in the output. A mesh line, indicated in pink
dots along the bottom of the ET portion of the hybrid cover, is used in the model to calculate the
flux of water leaving the free drainage boundary condition (bottom of the lower radon barrier)
for the portion that lies directly under the ET component of the hybrid cover. This line indicates
the lateral extend to which waste will be placed in the federal cell (i.e., waste will not be placed
under the side slope).
ET COVER
SIDE
SLOPE
TRANSITION
ZONE
ET COVER TRANSITION ZONE
2.1 m (7 ()
SIDE SLOPE
30.5 cm (12 in)
30.5 cm (12 in)
30.5 cm (12 in)
30.5 cm (12 in)
45.7 cm (18 in)
45.7 cm (18 in)
30.5 cm (12 in)
30.5 cm (12 in)
30.5 cm (12 in)
45.7 cm (18 in)
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 18

Figure 10. Boundary conditions used in the 2D hybrid cover model.
4.1.4 Mesh
A global target mesh spacing of 5 cm is used throughout the 2D hybrid cover model domain.
Additionally, three mesh refinements are specified where the target mesh spacing is set to 1 cm:
1) along the top surface of the model (atmospheric boundary) for both the top and side slopes,
and 2) between the evaporative layer and the frost protection layer. This refinement is needed to
model the sharp pressure head gradients induced by precipitation and evaporation at the surface,
and to provide enough resolution in the model to capture the performance of the capillary break
between the evaporative and frost protection layers.
Due to the horizontal scale of the model, a horizontal stretching factor of 15 is used to reduce the
number of nodes in the domain. Figure 11 shows a small portion of the ET side of the hybrid
cover; higher densities of nodes are seen along the surface and between the evaporative and frost
protection layers, as described above. Figure 11 also shows how the stretching factor can
increase the horizontal distance between nodes, while maintaining the higher resolution in the
vertical direction.
A mesh line is used tocalculate the por2on of
the free drainage fluxes out of the lower radon
barrier under the ET por2on of the hybrid cover
ET COVER SIDE
SLOPE
TRANSITION
ZONE
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 19

Figure 11. Node spacing detail shown for a portion of the top slope.
4.1.5 Rooting Parameters
A rooting depth of 80 cm is used for the ET portion of the hybrid cover (Figure 12), the same as
used in the v1.4 HYDRUS modeling (Neptune 2015b). No rooting is assigned for the side slope
and transition zone. As in the v1.4 HYDRUS modeling, the van Genuchten S-shaped model is
used for the water stress response function for root water uptake, using the same
parameterization; the default HYDRUS value of 3 is specified for the p exponent, and a value
of -200 cm for the h50 parameter (the pressure head or tension at which water uptake is reduce by
50 percent) (Neptune 2015b). HYDRUS now also allows the user to specify the pressure head
below which transpiration stops, known as the permanent wilting point. A tension of 50,000 cm
was specified for the wilting point. This is higher than typically assigned, but it is not impactful
because the S-shape water stress function discussed above effectively cuts off root water uptake
at much lower tensions. For example, the water stress factor is about 0.001 at a pressure head of
only -2000 cm.
Since every attempt is made to make the 2D model as similar as possible to the v1.4 HYDRUS
model runs, the value of h50 is specified in the 2D hybrid cover model to be identical to the value
used v1.4 HYDRUS modeling (-200 cm, as indicated above). However, recent work has shown
that this value is conservative for the expected soils and vegetation at the site. Additional
discussion regarding the root water uptake model is provided in Section 5.12.1.1.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 20
The lack of rooting parameters on the side slope was specified in order to allow maximum
impact of the side slope and transition zone on the percolation out of the ET cover. In reality,
riprap covers, despite their poor soils at the surface, have been shown to be effective at
promoting growth of native vegetation, allowing for transpiration to occur, and ultimately
reducing percolation from the cover system. Benson (2021) showed that by allowing a riprap
cover to grow vegetation, percolation out of the cover was substantially reduced.

Figure 12. Rooting depths shown for the top slope, transition zone, and side slope.
4.1.6 Atmospheric Input
The same 100-year year repeating atmospheric record used in the v1.4 modeling is used for the
2D hybrid cover model, including daily precipitation and calculations of potential
evaporation/transportation. This 100-year record was repeated 5 times in the 2D hybrid cover
model in order to run the 2D models out to 500 years.
4.1.7 Initial Conditions
Unique initial conditions are assigned for each model to reduce the time needed for the models to
reach a quasi-steady state equilibrium (Table 1). Values are selected to allow each model to start
relatively dry and to increase in water content until reaching a quasi-steady state equilibrium.
The results shown in Section 4.3 indicate this state is achieved within approximately 100 years.
Table 1. Tension (pressure head) used for initial conditions in each model. Values are
shown in cm.
Model Surface Layer
Evaporative
Layer
Frost
Protection
Layer
Upper Radon
Barrier
Lower Radon
Barrier
Maximum -500 -500 -500 -6800 -6800
2nd Highest -500 -500 -500 -11400 -11400
Median -1000 -1000 -1000 -12000 -12000
Minimum -1500 -1500 -1750 -78000 -78000

80cm
ET COVER SIDE
SLOPE
TRANSITION
ZONE
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 21
4.2 Material Hydraulic Properties
The v1.4 HYDRUS modeling allows values for α, n, and saturated hydraulic conductivity (Ksat)
to vary over a total of 50 model runs. Values for α and n are varied for the surface and
evaporative layers, and Ksat is varied for the upper and lower radon barriers (Table 2). Values for
each of the variables are drawn from unique distributions developed for each parameter (Neptune
2015b). Percolation results from the 50 unique parameter sets range from 0.0067 to 0.183
mm/year (Figure 13).
To select hydraulic parameters for the 5 materials used in the 2d hybrid cover model (Figure 9),
sets of α, n, and Ksat are selected from the v1.4 HYDRUS modeling. Since the 2D hybrid cover
model requires substantially more computational resources than the original v1.4 1D modeling,
four parameter sets from the v1.4 modeling are used to parameterize the materials in the 2D
hybrid cover model. The values of α, n, and Ksat that produced the maximum (0.183 mm/yr), 2nd
highest (0.0814 mm/yr), median (0.015 mm/yr), and minimum (0.0067 mm/yr) percolation are
selected to parameterize the material properties of the 2D hybrid cover model (Table 3).

Figure 13. Sorted results from the v1.4 HYDRUS modeling with the maximum, 2nd
highest, median, and minimum percolation results indicated.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 22
Table 2. Material hydraulic parameters used in DU PA v1.4 modeling.
θr θs α n Ksat
(VWC %) (VWC %) (1/cm) (unitless) (cm/day)
Surface 0.111 0.4089 *Variable Variable 4.46
Evaporative 0.111 0.481 Variable Variable 4.46
Frost Protection 0.065 0.41 0.075 1.89 106.1
Upper Radon 0.1 0.432 0.003 1.172 Variable
Lower Radon 0.1 0.432 0.003 1.172 Variable
*In the v1.4 HYDRUS modeling, values are drawn from distributions developed for α, n,
and Ksat.

Table 3. Parameter sets for α, n, and Ksat used in the 2D hybrid cover models.
V1.4 run
Surface and
Evaporative
Layers
α
(unitless)
Surface and
Evaporative
Layers
n
(unitless)
Upper and
Lower Radon
Barriers
Ksat
(cm/day)
1D HYDRUS
Results
(mm/yr)
Maximum #20 0.028186 1.378016 3.643845 0.183
2nd Highest #30 0.024165 1.349583 7.758327 0.0814
Median #36 0.014343 1.383885 1.005712 0.0149
Minimum #34 0.014338 1.265236 66.50366 0.00667

4.3 Results
The 2D hybrid cover model results are identified using the naming convention shown in the first
column of Table 3. For example, the 2D hybrid cover model results labeled “Maximum”
indicates that the values of α, n, and Ksat that produce the maximum amount of percolation in the
1D simulations (i.e., 0.183 mm/yr) are used in that simulation. The same applies with the 2nd
highest, median, and minimum models.
Percolation out of the lower radon barrier, tension at the bottom of the lower radon barrier, and
water content at the bottom of the lower radon barrier, as a function of horizontal distance along
the hybrid cover, are shown in Figure 14 through Figure 16. The snapshots in time are shown at
20-year intervals for model years 120 through 220. A small representation of the abbreviated
model domain is included at the top of each figure to indicate where along the hybrid cover the
results reflect.
The estimated annual percolation out of the lower radon barrier, for the ET cover portion of the
hybrid cover design, is shown in Figure 17 through 500 years. As seen on the left side of the
charts in Figure 14 through Figure 16, the abbreviated model is sufficiently large that the
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 23
percolation, tension, and water content are spatially constant for approximately the leftmost 10 m
of the domain, which implies that the choice of domain is not creating any undesirable boundary
effects. As such, conditions under the majority of the ET cover are expected to be unaffected by
interactions with the transition zone and side slope. Where the slopes flatten out in Figure 14
through Figure 16, it is assumed that these conditions will continue for the remainder of the ET
portion of the cover (i.e., the full 135 m proposed top slope that continues to the “crest” of the
cell identified in Figure 7). Percolation out of the lower radon barrier in the unaffected portion of
the model is used to estimate the average percolation out of the full 135 m proposed top slope, of
which the results are presented in Figure 17.
The percolation results presented in the v1.4 HYDRUS modeling (last column in Figure 13) are
based on the average percolation observed during a 100-year period of time at the end of the
simulations. To compare the 2D hybrid cover model results to the v1.4 results, the average
percolation is computed over model years 400–500 in order to capture the range of behavior over
the 100-yr precipitation record, while avoiding any transient fluxes early in the simulation
associated with initial conditions. A comparison of the results is shown in Table 4.
It is important to reiterate that the average percolation results shown in Figure 17 do not include
the higher percolation conditions observed under the side slope portion of the 2D hybrid cover
model. DU waste is not designated to be placed under the side slopes, and is proposed to be
confined only to areas that are directly beneath the ET portion of the hybrid cover design.
Therefore, only percolation through the ET portion of the hybrid cover design is presented here
and compared to the v1.4 HYDRUS results in Table 4.
31

M
a
r
c
h

2
0
2
1

24

Cl
i
v
e

D
U

P
A

M
o
d
e
l
—Re
s
p
o
n
s
e

t
o

D
W
M
R
C

1
2
-3-20
2
0

C
o
m
m
e
n
t
s

Figure 14. Percolation out of the lower radon barrier, as a function of lateral distance in the 2D hybrid cover model.
ET
COVER
SIDE
SLOPE
years years years
years years years
31

M
a
r
c
h

2
0
2
1

25

Cl
i
v
e

D
U

P
A

M
o
d
e
l
—Re
s
p
o
n
s
e

t
o

D
W
M
R
C

1
2
-3-20
2
0

C
o
m
m
e
n
t
s

Figure 15. Tension at the bottom of the lower radon barrier, as a function of lateral distance in the 2D hybrid cover model.
ET
COVER
SIDE
SLOPE
years years years
years years years
31

M
a
r
c
h

2
0
2
1

26

Cl
i
v
e

D
U

P
A

M
o
d
e
l
—Re
s
p
o
n
s
e

t
o

D
W
M
R
C

1
2
-3-20
2
0

C
o
m
m
e
n
t
s

Figure 16. Water content at the bottom of the lower radon barrier, as a function of lateral distance in the 2D hybrid cover
model.
ET
COVER
SIDE
SLOPE
years years years
years years years
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 27

Figure 17. Annual percolation out of the lower radon barrier for the entire 135 m length of
proposed ET cover (shown in Figure 8). Results for the Maximum and 2nd highest
show a minor amount of coarseness in the results after 300 years. This is due to the
precision of the model output in HYDRUS, and not a change in behavior of the
model.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 28
Table 4. Average annual percolation out of the radon barrier (model years 400–500); v1.4
results vs 2D hybrid cover results.
V1.4 run
V1.4 results
(mm/yr)
2D hybrid
cover results
(mm/yr)
Maximum #20 0.183 0.395
2nd Highest #30 0.0814 0.319
Median #36 0.0149 0.111
Minimum #34 0.00667 0.202

4.4 Discussion
The results from the 2D model of the hybrid cover indicate that percolation through the riprap
side slope will be higher compared to the ET top slope portion of the cover. In Figure 14, the
right portions of the charts show that percolation out of the side slope exceed 10 mm/yr during
the model simulation. This is ultimately due to the configuration of the side slope, which lacks
features such as a storage and evaporative zone (e.g., the surface and evaporative layers in the
ET portion of the hybrid cover design) and a capillary break (interface between the evaporative
and frost protection layers in the ET portion of the hybrid cover design). With this configuration,
the side slope permits greater percolation, leading to lower tensions and higher water content at
the bottom of the lower radon barrier (Figure 15 and Figure 16), and ultimately higher
percolation near the transition zone (Figure 14).
Figure 15 and Figure 16 show pressure head and water content for the 2D hybrid cover model
domain at snapshots over a 120-year period. Throughout this period, the side slope maintains
relatively lower tension and higher water content in the radon barrier, whereas the radon barrier
below the ET portion of the hybrid cover shows much higher tensions and lower water content.
This is due to the difference in average percolation rate of the ET portion of the hybrid cover
design on the left side of the model compared to that of the side slope on the right side of the
model. This results in a lateral hydraulic gradient in the radon barrier, lateral flow from right to
left, and, consequently, increased percolation beneath the ET cover near the transition zone.
Despite the lateral hydraulic gradient, the low conductivity of the radon barrier material does not
permit enough lateral flow to cause a sizeable increase in the average percolation under the ET
portion of the hybrid cover. As seen in Figure 14, percolation out of the lower radon barrier
beneath the ET cover does increase near the transition zone. However, Figure 17 shows that the
average percolation out of the lower radon barrier over the proposed 135 m of ET cover in the
revised cover design ranges from 0.11 to 0.41 mm/yr for the four models explored.
While 1D modeling performed in support of Model v1.4 (Neptune 2015d) suggested that
percolation is insensitive to the saturated hydraulic conductivity of the radon barrier, in this
configuration, the radon barrier provides a link between the side slope and the ET cover, and as a
result, these simulations suggest that percolation associated with this flow path is quite sensitive
to the radon barrier properties. Consequently, the impact of the hybrid design on the ET portion
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 29
of the cover appears to be most pronounced in the model realization with the highest saturated
hydraulic conductivity in the radon barrier, which happened to be the “Minimum” case described
above. This can be observed in Figure 14 through Figure 16. Ksat for this model realization,
66.5 cm/day, is significantly higher than the other three cases, which ranged from 1.0 to
7.7 cm/day (Table 3).
All values for Ksat used in the v1.4 modeling for the radon barrier are 3-4 orders of magnitude
higher than what has since been measured from samples obtained from the Test Cell
deconstruction (EnergySolutions 2020). Three 5 × 10-8 clay samples (upper radon barrier) were
taken from the Test Cell and submitted for analysis; the results for Ksat ranged from 0.002 to
0.012 cm/day, with a mean of 0.005 cm/day. Five 1 × 10-6 clay samples (lower radon barrier)
were taken from the Test Cell and submitted for analysis; the results for Ksat ranged from 0.001
to 0.041 cm/day, with a mean of 0.011 cm/day. In comparison, the v1.4 modeling used values
that ranged from 0.748 to 66.5 cm/day, with a mean of 6.76 cm/day.
The steepness of the curves in Figure 14 through Figure 16 indicate that elevated percolation,
water content, and lower tensions are restricted to distances close to the transition zone, with the
impacts of the transition zone diminishing rapidly with distance. While the effects of the hybrid
cover design are apparent in these figures, the impact is grossly overestimated as a result of the
high Ksat values used to parameterize the radon barrier. The “Median” model shows the steepest
decline in percolation rates away from the transition zone due to it having the lowest Ksat value
of 1.0 cm/day (although this value is now considered very high in comparison to the site-specific
data obtained discussed above).
The results of the 2D modeling indicate that percolation rates through the ET cover are expected
to remain low (Figure 17), despite the overly conservative parameterization of the radon barrier.
Previous GoldSim modeling of the Federal Cell has indicated that performance criteria for the
site will be met when percolation through the cover is less than approximately 1 mm/yr (Section
3.1.1). Figure 17 shows that the influence of the side slope configuration on the overall
performance of the federal cell hybrid cover design is minimal, and that average percolation is
expected to fall well below this threshold. If site-specific values for radon barrier materials are
incorporated into the HYDRUS modeling, the influence of the hybrid cover on percolation rates
below the ET portion of the hybrid cover is expected to be reduced further, and therefore the
impact of the hybrid design on overall percolation is comparable to modeling used in DU PA
v1.4.
5.0 UDEQ Comments and Responses
Each comment is quoted in full followed by Neptune’s response.
5.1 UDEQ Comment 1: Hybrid Cover Percolation Model
1. A new hybrid-cover design is proposed and included in the Federal-Cell license application.
EnergySolutions and its contractor, Neptune and Company, Inc., need to submit a supplemental
document that describes and justifies with supportive analysis and calculations how results from
the modeling of an evapotranspiration (ET) cover as presented in Clive DU PA Model v1.4 are
applicable to this new hybrid-cover design. Within this supplemental document, identify,
describe, and differentiate cover components, concepts and terminology that are applicable to
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 30
the hybrid-cover design and those that are being newly introduced or adapted for the hybrid-
cover design.
5.1.1 Comment 1 Response
Table 5 below describes the basic layering of the original design of the federal cell compared to
the revised hybrid cover design. Additional detail regarding the revised federal cell design is
provided in Section 2.0. Section 4.0 provides details and results of a 2D model of the hybrid
cover design that compares the performance of the revised cell design to the v1.4 HYDRUS
results.
Table 5. Comparison of the layering on the top and side slopes of the original Federal Cell
design and the revised design.
Original design Hybrid cover
Layer Top slope Thickness
cm (in) Side slope Thickness
cm (in) Top slope Thickness
cm (in)
Side
slope
Thickness
cm (in)
1 Surface 15.2 (6) Surface 15.2 (6) Surface 30.5 (12) Riprap 47.7 (18)
2 Evaporative 30.5 (12) Evaporative 30.5 (12) Evaporative 30.5 (12) Filter 30.5 (12)
3 Frost
Protection 47.7 (18) Frost
Protection 47.7 (18) Frost
Protection 47.7 (18) Frost
Protection 47.7 (18)
4
Upper
Radon
Barrier
30.5 (12)
Upper
Radon
Barrier
30.5 (12)
Upper
Radon
Barrier
30.5 (12)
Upper
Radon
Barrier
30.5 (12)
5
Lower
Radon
Barrier
30.5 (12)
Lower
Radon
Barrier
30.5 (12)
Lower
Radon
Barrier
30.5 (12)
Lower
Radon
Barrier
30.5 (12)

5.2 UDEQ Comment 2: HYDRUS Snowmelt Algorithm
2. The efficacy of the snowmelt algorithm utilized by HYDRUS remains in question. A validation
of the snowmelt algorithm utilized by HYDRUS is required and has not been presented. A
validation reported in the literature for conditions representative of the Clive Facility needs to be
submitted, or a regional-specific validation study needs to be conducted. It is essential that the
results of the algorithm validation, literature-based or regional-specific, demonstrate the efficacy
of the snowmelt model in HYDRUS for predicting snow accumulation, snow melt, and infiltration
in the Clive region. The submittal of the validation document should present and explain the
validation analysis and justify each of the parameters used in the HYDRUS snowmelt model
based on information in the literature and/or the outcomes of the regional-specific validation.
Part of a regional-specific validation may potentially be obtained by simulating hydrologic
conditions with HYDRUS using meteorological data from a site in the region and/or the Clive
meteorological station, where snow is known to accumulate, hydraulic properties are known
(e.g., from the Cover Test Cell data), and conditions within the profile are monitored for a
comparison between predicted and observed conditions. This will help demonstrate that the
HYDRUS snowmelt sub-model properly predicts snow accumulation, snow melt, infiltration,
measured soil-water content and soil-water storage at various depths.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 31
5.2.1 Comment 2 Response
This comment is addressed in three parts: (1) a literature review; (2) an attempted site-specific
validation using Cover Test Cell data; and (3) correlation of snowpack data from Dugway, UT
against the snowpack predicted by applying the Dugway climate data to the HYDRUS snowmelt
algorithm.
5.2.1.1 Literature Review
A brief literature search was conducted to assess whether the snowmelt algorithm has been
validated for a comparable setting to Clive. The HYDRUS User Manual (Šimůnek et al. 2007)
describes this algorithm as follows:
When heat transport is simulated simultaneously with water flow and atmospheric boundary
conditions, then snow accumulation on top of the soil surface can be simulated. The code
then assumes that when the air temperature is below -2 C all precipitation is in the form of
snow. When the air temperature is above +2 C all precipitation is in the form of liquid, while
a linear transition is used between the two limiting temperatures (-2,2). The code further
assumes that when the air temperature is above zero, the existing snow layer (if it exists)
melts proportionally to the air temperature.
No discussion of validation of the snowmelt algorithm is provided in the HYDRUS User
Manual, however.
The HYDRUS snowmelt infiltration routine has been reported to accurately simulate snowmelt
infiltration rates when compared with measured soil water content during spring thaw for the
Bucegi Mountains (Dobre et al. 2017). However, this work does not include conditions
representative of the Clive facility.
Similarly, Zhao et al. (2016) evaluate soil moisture and temperature simulated in HYDRUS 1D
using both the snow routine and a frozen soil module against measured data for a grassland in
Inner Mongolia. In this work, both the snow routine and the frozen soil module match well
against measured data when the soil is not frozen; and the HYDRUS 1D frozen soil module
better matches the data, particularly at an hourly level, when the soil is frozen. During times the
soil is frozen, the snow routine appears to overstate soil moisture compared with the data and the
frozen soil module. Summary climate data for the Inner Mongolia site is not provided; though
this appears to be an arid location that could be comparable in ways to Clive.
5.2.1.2 Cover Test Cell Model
Given indeterminate results from a literature review, the Cover Test Cell deconstruction data was
looked to as a possible way to validate the HYDRUS snow routine on a site-specific basis. If
HYDRUS could reasonably replicate the Cover Test Cell percolation data when the snow model
was employed, the snow model would be considered sufficient for capturing the effects of snow
formation and melting at the Clive, UT site.
The Cover Test Cell was constructed just south of the proposed site for the Federal Cell, and was
therefore exposed to all the meteorological impacts at the site, including snowfall, over the
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 32
15-year period it was monitored. The measured percolation data out of the bottom of the Cover
Test Cell effectively integrated all the processes of snow accumulation and melting that would
have occurred at the site during this period. Figure 18 shows the design and layering of the Cover
Test Cell as it was constructed at the site.
A simplified 1D model of the Cover Test Cell is constructed in HYDRUS. The snow feature is
left off in one simulation, and is turned on in another. Percolation predictions from the two
models are compared to actual percolation collected below the Cover Test Cell via a lysimeter
tipping bucket.

Figure 18. Test Cell (reprinted from EnergySolutions (2020)).
Model Domain
A 1D column is used to model the Cover Test Cell. Seven layers are used in the 1D Test Cell
model, shown in Figure 19. Because extremely coarse materials (i.e., riprap) are problematic to
model when paired with an atmospheric boundary (typically requiring very small mesh
refinement and/or timesteps, both of which are impractical at the scale of the 1D model and
length of simulation), the material hydraulic parameters for the riprap layer are assigned the
default parameters available in HYDRUS for sandy loam in the first 15.2 cm (6 in), and sand in
the next 30.5 cm (12 in), as shown in Figure 19. Default parameters available in HYDRUS are
based on Carsel and Parrish (1988).
With this parameterization of the riprap material, Ksat is high enough that no runoff occurs
throughout the simulation, and therefore all precipitation is accounted for in the model domain.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 33
In addition, water content in these layers generally stays below 10%, and therefore no significant
storage occurs in these layers using this parameterization.
Table 6 summarizes hydraulic properties used in the simulation. Hydraulic properties for the
remaining layers are taken from EnergySolutions (2020). The Filter A, Sacrificial Soil, and Filter
B layers each had a single sample taken and submitted for material hydraulic property testing, of
which the results are shown in Table 1. For the clay layers, multiple samples are taken from the
5x10-8 cm/s (n=3) and 1x10-6 cm/s (n=5) layers of the Test Cell. A representative sample is
selected for each layer, based the sample’s overall tendency toward the mean of each of the 5
parameters shown in Table 1.
Root Water Uptake
No root water uptake is specified in the model. After passing through the “riprap” layer, water
continues downward through the subsequent layers and ultimately out of the bottom of the
model.

Figure 19. Layers in the 1D Test Cell model.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 34
Table 6. Material hydraulic properties.
Thickness θr θs α n Ksat
cm (in) (VWC %) (VWC %) (1/cm) (unitless) (cm/day)
Sandy Loam 15.2 (6) 0.065 0.41 0.075 1.89 106.1
Sand 30.5 (12) 0.045 0.43 0.145 2.68 712.8
Filter A 15.2 (6) 0.03 0.37 0.329592 2.79 1874880
Sacrificial Soil 30.5 (12) 0 0.3 0.0336735 1.18 24192
Filter B 15.2 (6) 0.07 0.37 0.0591837 4 907200
Upper Radon 30.5 (12) 0 0.39 0.000153061 1.39 0.00216
Lower 182.8 (72) 0 0.39 0.000132653 1.43 0.0020736

Node Resolution and Boundary Conditions
1001 nodes are used for the 320 cm (10.5 ft) 1D Test Cell model, with uniform spacing equating
to approximately 3.2 mm between nodes. An atmospheric boundary condition is specified at the
top node of the model and a free drainage boundary condition at the bottom node of the model.
Atmospheric Input
The same 100-year year repeating atmospheric record used in the v1.4 modeling is used for the
1D Test Cell model, including daily precipitation and calculations of potential evaporation. This
100-year record is repeated 10 times in order to run the 1D model out to 1000 years. Over this
period, the HYDRUS model obtains a quasi-steady state equilibrium, from which the percolation
results are based. Two simulations are performed, one with the snow model turned off, and
another with the snow model turned on.
Initial Conditions
A pressure head of -500 cm was specified for all layers as an initial condition.
Results
Percolation results are calculated as the average percolation out of the system for the last 100
years of the simulation to avoid any transient fluxes early in the simulation associated with initial
conditions, and are presented in Table 7. Actual percolation data from the Cover Test Cell was
collected via a tipping bucket lysimeter beneath the Cover Test Cell for a 15-year period from
2002 through 2016. Monthly total tips are shown in Figure 20, with 1 tip corresponding to
approximately 4.74 cm3 (0.289 in3) of water. After dividing the measured volumes by the area of
the Test Cell, an average percolation rate of 0.2 mm/yr was recorded by the lysimeter tipping
bucket.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 35
Table 7. Results of the 1D Test Cell model.
Model Percolation (mm/yr)
With snow turned off 2.14
With snow turned on 5.32

Figure 20. Monthly tip data collected from the Cover Test Cell.
Discussion
Compared with measured data, HYDRUS overpredicts percolation through the Cover Test Cell
by more than an order of magnitude in both models. However, there are several additional
mechanisms that occurred in the Test Cell that are not accounted for in the 1D Test Cell model.
Figure 18 indicates that the 5 × 10-8 cm/s and 1 × 10-6 cm/s clays are in direct contact with
adjacent subsurface native clays, and no membrane was placed at this boundary to prevent lateral
flow. The Cover Test Cell deconstruction report states that it was assumed that only vertical flow
of water would occur in the test cell. This is likely not a valid assumption, given how the Cover
Test Cell was constructed.
Potential evaporation in the high desert environment at Clive exceeds annual precipitation by a
large margin, leading to extremely dry conditions in the shallow soil environment. In contrast,
the buried portion of the Cover Test Cell (5 × 10-8 cm/s and 1 × 10-6 cm/s clays) likely had
higher water content compared to the surrounding soils because it was constructed of very coarse
materials at the surface, having little storage capacity to hold on to and return water to the
atmosphere via evaporation and resulting in higher levels of percolation.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 36
This is observed in the side slope portion of the hybrid cover model in Section 4.0. That is,
placing highly conductive materials with low storage capacity at the surface allows water to
quickly pass through the shallow depths, and continue to deeper portions of the column where it
is not subject to evaporative processes. This was likely the case with the Cover Test Cell, where
increased downward flux of water through the riprap ultimately lead to higher water content (and
lower tension) in the 5 × 10-8 cm/s and 1 × 10-6 cm/s clays, compared the surrounding native
clays. The difference in pressure heads and water content between the clays in the Cover Test
Cell and the surrounding native clays would result in a hydraulic gradient, creating lateral flow.
Simply put, one of apparent reasons that the Cover Test Cell data did not record as much
percolation per year as the 1D HYDRUS models predict is that a portion of the infiltrated water
was likely leaving the system laterally, and therefore unaccounted for by the tipping bucket
lysimeter.
Another reason that the percolation data from the Cover Test Cell is lower than predicted by the
1D HYDRUS model could be the result of the establishment of vegetation and minor rooting in
some of the layers of the Cover Test Cell. In the 1D HYDRUS models, no root water uptake is
specified in the model. However, during deconstruction of the Cover Test Cell, a small amount
of rooting was observed in the sacrificial soil layer (EnergySolutions 2020). Even a small amount
of roots in the system could lead to significant amounts of water being captured and returned to
the atmosphere via transpiration. Benson (2021) has shown the effectiveness of even minor
amounts of rooting on the performance of covers constructed with riprap at the surface.
Lastly, Figure 18 also shows a “collection drainage trough” on the left side of the diagram. The
aim of this trough was to capture any excess water that did not travel downward into the 5 × 10-8
cm/s clay (the Cover Test Cell diagram indicates a 2.8% slope, and therefore some lateral flow at
this boundary was expected). The data collected from this trough, however, was deemed to be of
poor quality and not reliable, and therefore could not be used in a mass balance calculation.
Despite the actual data not being useful in a quantitative sense, there was reasonable assurance
that this trough did collect some amount of water over the 15-year period of monitoring.
Therefore, it is likely that a portion of the water infiltrating through the Test Cell left the system
through the collection drainage trough, and was therefore unaccounted for by the lysimeter
tipping bucket data.
Accordingly, the Cover Test Cell system is poorly suited to model with a simple 1D column, and
attempting to address the snowmelt question via this pathway is inconclusive.
5.2.1.3 Comparison with Regional Snowpack Data
An analysis is presented that evaluates how well the HYDRUS snowmelt sub-model predicts
snowmelt compared with observed historical data. This analysis is presented in two parts:
snowmelt and snow accumulation.
Snowmelt
The following is a brief overview of how the HYDRUS snowmelt algorithm operates. HYDRUS
calculates snowmelt proportional to the air temperature using a snowmelt constant (M), which is
the amount of snow (given in length units, such as cm of water) that will melt during one day for
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 37
each oC. The HYDRUS model also employs a sublimation constant (S) that accounts for
vaporization that is mediated by potential evaporation.
HYDRUS uses the following equation for calculation of the snow layer at time t:
ℎ!=ℎ!"#+∆ℎ!−(')()!)−(*)(+!)
where
ht is Snow water equivalent (SWE) height of the snow layer at time t (cm)
ht-1 is SWE height of the snow layer at time t-1 (cm)
Δht is Change in SWE height of the snow layer at time t due to precipitation (cm)
S is Sublimation constant
Et is Potential evaporation at time t (cm)
M is Melting constant (cm/˚C)
Tt is Average daily temperature above 0˚C; set to zero if the daily temp is < 0˚C
The change in snow layer, Δht, is determined by the daily temperature and precipitation. If the
temperature is below -2 C, all of the precipitation is assumed to have fallen as snow and adds to
the snow layer. If the temperature is between -2C and 2 C, the precipitation is assumed to have
fallen as a mix of rain and snow, and Δht is reduced in a linear fashion. As noted above, the last
term is set to zero if the daily average temperature is less than zero, as no melting is assumed to
occur. Converting the SWE snow layer to actual snow depth requires an assumption of snow
density; the HYDRUS manual recommends a ratio 10:1 (i.e., 1cm of SWE is equal to 10 cm of
snow).
Daily potential evaporation (PE) is calculated with values for extraterrestrial radiation and daily
maximum, minimum and mean temperatures using the Hargreaves method (see Neptune (2015b)
for details). HYDRUS default values of 0.43 for M and 0.2 for S are used in the calculation. This
algorithm is applied to local climate data to evaluate the efficacy of this snowmelt algorithm that
is used in HYDRUS. Daily temperature, snowfall and snow layer data are obtained for the
nearby site of Dugway, UT (site id 422257) using the SC ACIS Tool (http://scacis.rcc-acis.org/).
Observed snowfall from the Dugway site consists of the daily recorded snowfall and allow for
the snow layer to increase during the 14-day period. Since HYDRUS calculates snow water
equivalent, the snow layer is calculated with the rules outlined in the HYDRUS manual which
uses the ratio of 1 cm snow water to 10 cm of snow.
The observed snow depth data are processed to identify the 25 deepest snow layers separated by
at least 14 days. The 14-day period is arbitrarily determined to balance both typical dynamics for
the melt of a snowfall event at the site as well as issues associated with data gaps in the station
record at Dugway. Shorter time periods enhance the impact of missing data and make
meaningful comparisons more challenging. The day corresponding to the deepest snow layer is
then used as a starting point to apply the HYDRUS snowmelt algorithm. Maximum daily
temperature from the observed Dugway record is used in the HYDRUS-based calculation of
snow melt as part of this regional specific validation exercise. Maximum daily temperatures are
applied to M since daily mean temperatures appeared to underestimate melting. This is likely due
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 38
to the nature of the algorithm, where mean daily temperatures at or below zero fail to capture
melting that occurs during warm periods of the day when temperatures exceed zero.
The snowmelt equation in HYDRUS is run with the data from the Dugway site and compared to
the corresponding observed data for the 14-day period following each of the identified top 25
observed snow depth events. Observed snow layers are plotted along with calculated snow layers
for the 14-day periods (Figure 21 and Figure 22). Two figures with multiple panels are used for
improved clarity of presentation. In these figures, the blue lines are the data for snow depth as
recorded by the Dugway UT station data. The red lines show the snowmelt calculated using the
Dugway UT station meteorological data fed into the HYDRUS snowmelt algorithm. Two of the
25 events were omitted due to what appear to be data quality issues. The snow layer record from
1-29-1980 through 2-12-1980 showed significant melting despite the fact that the mean
maximum temperature for the first 3 days of the record was -8.52 C. Similarly, the snow layer
record from 12-26-2012 through 1-9-2013 showed significant melting despite the mean
maximum temperature for the 14 days being -5.30 C.

Figure 21. 14-day periods of calculated vs observed snow layers at Dugway, UT in temporal
order of occurrence.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 39

Figure 22. 14-day periods of calculated vs observed snow layers at Dugway, UT in temporal
order of occurrence.
Although the Dugway record has some data quality issues including missing data, the HYDRUS
algorithm performs well with respect to the estimation of the snowmelt relative to the available
data (Figure 21 and Figure 22). This performance is based on both the timing associated with the
initial melting and the total duration of the melting events for both the observed and HYDRUS-
estimated snowmelt episodes. These figures show that for the largest melting events in the
historical record at Dugway, direct comparison of the observed data with projections using the
HYDRUS snowmelt algorithm driven by site-specific meteorological data show good agreement.
These largest melting events are critical with respect to the subsequent accurate depiction of
infiltration, measured soil-water content and soil-water storage at various depths.
Snow Accumulation
To evaluate snow accumulation, snowfall is calculated using HYDRUS applied to daily
precipitation and average daily temperatures from the historical record of meteorological data at
Dugway. Because HYDRUS calculates snow water equivalent, the snowfall is again calculated
with the rules outlined in the HYDRUS manual which uses the ratio of 1 cm snow water to 10
cm of snow. HYDRUS assumes for temperatures less than -2 oC, all precipitation is snow, and
for temperatures above +2 oC, all precipitation is rain. For temperatures between -2 oC and +2
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 40
oC, there is a linear conversion from snow to rain (0 oC would be half and half). In this manner,
the calculated snowfall is allowed to accumulate and reduce using the HYDRUS equations
outlined above with maximum daily temperatures applied to M.
Snow layers are calculated for the entire Dugway record (09/21/1950 – 07/08/2013) using the
HYDRUS algorithm and compared to the corresponding observed snow layer. Average daily
calculated snow layers are plotted along with average daily observed snow layers across the
entire year (Figure 23). The HYDRUS algorithm performed well with calculated snow layers
tracking along the observed values. HYDRUS does appear to overestimate compared with
observed snow layers in January and underestimate in February. These values balance out when
averaged from December through February with the calculated snow layer (1.21 cm) only
slightly exceeding the observed snow layer (1.17 cm).

Figure 23. Average daily calculated and observed snow layers at Dugway, UT.
Dugway receives a reasonable amount of snowfall in February when average daily temperatures
move above 2 oC (Figure 24). Under these conditions HYDRUS would not predict snowfall.
Discrepancies may also occur from differences between air and ground temperature, where
warmer ground temperatures should lead to more snowmelt and less snow accumulation. Colder
ground temperatures might limit snowmelt when air temperatures begin to increase. HYDRUS
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 41
does not include ground temperatures in its snow layer calculation. Seasonal lag in ground
temperatures theoretically would lead HYDRUS to overestimate snow accumulation in the early
part of the winter and underestimate snow accumulation in late winter/early spring, which is
consistent with the pattern seen in Figure 23.

Figure 24. Daily Mean Temperature at Dugway, UT averaged across the period of record.
Summary
A quantitative validation of the snowmelt algorithm in HYDRUS is achieved by comparing
HYDRUS-based snowmelt estimates with observed snowmelt and snow accumulation data from
Dugway. The analysis of the 25 largest snow depths on record for the Dugway site focuses
attention on the most critical time periods for the validation of the snowmelt model in HYDRUS.
In general, the snowmelt depicted by HYDRUS tracks the observed snowmelt from Dugway
quite well. Despite small discrepancies, average calculated snow accumulation matches the
average observed snow accumulation. These analyses provide strong support to demonstrate the
efficacy and sufficiency of the snowmelt algorithm utilized by HYDRUS. Collectively, these
analyses help demonstrate that the HYDRUS snowmelt sub-model properly predicts snow
accumulation, snowmelt, and hence infiltration, measured soil-water content and soil-water
storage at various depths.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 42
5.3 UDEQ Comment 3: Applying Cover Test Cell Data
Provide a comparison of the engineering properties determined for the individual components of
the rock-armored Cover Test Cell as studied in connection with its deconstruction to the
properties used in the current model of an evapotranspiration (ET) cover system in the Clive DU
PA Model v1.4. Report how the properties from the Cover Test Cell compare to the naturalized
parameters from NUREG/CR-7028 and to those used in the most recent HYDRUS model of the
evapotranspiration cover. Do an analysis using the Cover Test Cell deconstruction data in the
HYDRUS model for an evapotranspiration cover: i) in as-built initial conditions , (ii) then apply a
deterioration methodology that represents pedogenic evolution of the cover soils over long
periods of time (e.g., due to rooting of overlying plants, burrowing by mammals or insects, frost
activity, wet-dry cycling, differential settlement, etc.) representative of what is reported in the
literature and available from analog studies in the region, and (iii) modify the HYDRUS model as
needed to account for the hybrid cover (evapotranspiration cover on the top slope; rock-armor
cover on the side slope) as currently proposed in the Draft Federal Cell license application. For
each of these analyses, prepare diagrams that display predictions of soil-water content and soil-
water storage at various depths. Submit the results of the analysis including an explanation of
how the different cover types compare.
5.3.1 Comment 3 Response
Much interesting data has been generated with the Cover Test Cell deconstruction project
completed in 2019 (EnergySolutions 2020). However, the Cover Test Cell was constructed to an
earlier version of the rock armor cover design used at the Clive facility; therefore, only data for
the radon barrier clays is comparable between the Cover Test Cell and the ET cover proposed in
DU PA v1.4. Considering that similar material properties were used for the surface/evaporative
zone and radon barrier layers in DU PA v1.4, some comparison can be made with these layers as
well.
NUREG/CR-7028 (Benson et al. 2011) provides recommendations in Section 10.2 regarding
engineering properties for fine-textured earthen storage and barrier layers that can be used in
performance assessments in lieu of site-specific data. These recommendations are based on
covers studied 5 to 10 years after initial construction. The Cover Test Cell was de-constructed
and its material properties tested 18 years after it was placed into service; for a roughly
comparable, if longer, period of system stabilization and naturalization to that evaluated in
NUREG/CR-7028.
Table 8 summarizes engineering properties across the HYDRUS modeling performed in DU PA
v1.4 (Neptune (2015b), Tables 8 and 9); the Cover Test Cell; and NUREG/CR-7028. Average
values are calculated in any cases where a range of values are provided in the source
documentation and discussed in more detail below.
Specifically, in the HYDRUS modeling supporting DU PA v1.4, the parameters of α and n were
varied for the surface layers; and Ksat was varied for the upper and lower radon barrier layers.
Other parameters were deterministic (Neptune 2015b).
For the Cover Test Cell, all radon barrier values represent the average of three (for the upper
radon barrier) to five (for the lower radon barrier) data points, as reported in the EnergySolutions
(2020) appendix “Wisconsin Geotechnical Laboratory, Hydraulic Properties of Soils from a
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 43
Final Cover Test Section in Clive, Utah, Geotechnical Laboratory Report No. 20-17, 2020; Table
1.” The Cover Test Cell also returned data for that cover design’s riprap, filter zone, and
sacrificial soil layers; however, the material properties for these layers do not match the ET cover
design so these data are not included in Table 8.
Values for NUREG/CR-7028 reflect the average of the recommended range in Section 10.2, or
the recommended initial condition for PA modeling. NUREG/CR-7028 recommended values are
applied to both the surface/evaporation zone and radon barrier layers. NUREG/CR-7028 does
not suggest values to use for a layer such as the frost protection layer, which provides a capillary
break within the cover system.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 44
Table 8. Engineering properties of cover layers for DU PA v1.4, the Cover Test Cell, and
NUREG/CR-7028.
Design Basis
Layer Input
Parameter
ET Cover DU PA v1.4
(actual, 50 sims)
Cover Test Cell NUREG/CR-7028
Surface θr (unitless) 0.111 n/a 02
θs (unitless) 0.4089 n/a 0.4
α (1/cm) 0.0169 n/a 0.0204
n (unitless) 1.3 n/a 1.3
Ksat (cm/day) 4.46 n/a 22
Evaporative Zone θr (unitless) 0.111 n/a 0
θs (unitless) 0.481 n/a 0.4
α (1/cm) 0.0169 n/a 0.0204
n (unitless) 1.3 n/a 1.3
Ksat (cm/day) 4.46 n/a 22
Frost Protection θr (unitless) 0.065 n/a n/a
θs (unitless) 0.41 n/a n/a
α (1/cm) 0.075 n/a n/a
n (unitless) 1.89 n/a n/a
Ksat (cm/day) 106.1 n/a n/a
Upper Radon
Barrier
θr (unitless) 0.1 0 0
θs (unitless) 0.432 0.38 0.4
α (1/cm) 0.003 0.0002 0.0204
n (unitless) 1.172 1.39 1.3
Ksat (cm/day) 6.75 5.16E-03 22
Lower Radon
Barrier
θr (unitless) 0.1 0 0
θs (unitless) 0.432 0.38 0.4
α (1/cm) 0.003 0.0002 0.0204
n (unitless) 1.172 1.4 1.3
Ksat (cm/day) 6.75 1.13E-02 22

A number of observations can be made from Table 8. All comparisons are limited to the
surface/evaporative zone and radon barrier layers, since there is not comparable data in the test
cell or NUREG/CR-7028 for the frost protection layer.

2 Inferred from Table 6.3 of NUREG/CR-7028 for all clay layers, though not explicitly discussed in Section 10.2
therein.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 45
In DU PA v1.4, the value for θr is higher than that reported for the Cover Test Cell and inferred
from NUREG/CR-7028. Benson et al. (2011), commonly report a value of zero for θr for all soil
types. Benson’s laboratory also evaluated the soil properties from the Test Cell, and therefore, θr
is reported to be zero for these soils as well. However, other well-used and well-known soils
databases (e.g. Rosetta, Carsel and Parrish (1988)) do not report zero values of θr. The effect of
using a zero value for θr is to increase the storage capacity of the soil. In addition, a zero value of
θr will also affect the water content at which percolation occurs.
Values for θs and n are comparable across these approaches/data sets, while the value for α used
in DU PA v1.4 is comparable to that in NUREG/CR-7028 for the surface/evaporative zone
layers. For the radon barrier layers, the value for α used in DU PA v1.4 is bracketed by those
demonstrated by site-specific Cover Test Cell Data and those recommended by NUREG/CR-
7028, with the Cover Test Cell data being roughly an order of magnitude lower than modeled;
and NUREG/CR-7028 roughly an order of magnitude higher than modeled. Therefore, the α
used in DU PA v1.4 is nicely bounded by site-specific data, and the “generic” value of α
informed by multiple datasets in NUREG/CR-7028.
Finally, values for Ksat vary widely between those reported by site-specific Cover Test Cell data
and those used in DU PA v1.4 and recommended by NUREG/CR-7028. The latter two values
are roughly comparable, while the site-specific data is 3 to 4 orders of magnitude lower. The site-
specific Ksat is very low as a result of construction specifications; and demonstrates little change
from the as-built condition over the Cover Test Cell’s 18-year service life.
Refer to Section 5.2.1.2 for discussion of an attempt to replicate Cover Test Cell data in a
HYDRUS 1D model. Cover Test Cell data are used in that attempt; however, the results
ultimately are inconclusive. Refer to Section 4 for discussion of a HYDRUS 2D model of the
hybrid cover design. While Cover Test Cell data for the radon barrier layers are not directly used
in the HYDRUS 2D model, they are effectively bracketed by the parameters evaluated as
discussed above.
5.4 UDEQ Comment 4: Regression Model
Quantitatively and rationally explain why the regression model used for abstraction of HYDRUS
results into the GoldSim model is insensitive to Ksat of the cover soils. Ensure that the submitted
explanation is consistent with the principles of variably saturated flow and the formulation of
Richards’ Equation in HYDRUS.
5.4.1 Comment 4 Response
The cover system consists of an evaporative zone composed of Unit 4 silty clay and a radon
barrier composed of compacted Unit 4 silty clay, separated by the frost protection layer. The
important properties of the FPL as it relates to cover performance are those that define the soil
water characteristic curve, as the FPL should provide a material contrast when compared to the
overlying evaporative zone; this is explored in Section 5.7.1. While the question was interpreted
as referring to the top two model layers that comprise the evaporative zone, the response below
begins by summarizing the results of the v1.4 HYDRUS simulations, which included variation of
Ksat in the radon barrier, conceptualized as the bottom two model layers. The remainder of the
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 46
response addresses variation of Ksat in the evaporative zone, and describes a new set of
sensitivity simulations that was prepared to explore model sensitivity to this parameter.
5.4.1.1 Variation of Ksat of the Radon Barrier (Layers 4 and 5) in Model v1.4
The set of 50 HYDRUS runs presented with the DU PA Model v1.4 included variation of
saturated hydraulic conductivity (Ksat) of the radon barrier in the bottom two layers of the soil
column; the Ksat values for the other three layers were not varied.
The regression equations indicate that the water content in the radon barriers are sensitive to the
Ksat parameter, as expressed by non-zero ,1 values in Table 10 of Neptune (2015b). Therefore,
this HYDRUS model output is sensitive to the corresponding material Ksat, and this sensitivity is
incorporated into the GoldSim model via the regression equations for water content.
The regression equations indicated that percolation through the cover system is insensitive to the
Ksat of radon barrier. This is largely due to the fact that the percolation through the cover system
hinges on the performance of the capillary barrier interactions in the upper layers of the cover;
this idea is explored extensively in Section 5.7.1 based on the principles of variably saturated
flow. An example of conditions that would cause this capillary barrier to momentarily “break” in
the v1.4 modeling during wetting events is provided in Section 5.8.1. If water penetrates the
cover to the bottom portion of the frost protection layer, below the practical reach of
evapotranspiration, it will eventually percolate through the bottom of the cover.
The Ksat of the radon barrier governs the kinetics of the percolation in the lower portion of the
cover. As such, the average water content of the radon barrier is sensitive to Ksat. If Ksat is higher,
excess moisture drains more quickly through the radon barrier and the water content is elevated
for relatively less time. If Ksat is lower, the equilibration takes longer, and the water content is
elevated for a longer period of time; averaged over long periods (100 years in the v1.4
modeling), lower Ksat values produce a higher water content and higher Ksat values produce a
lower water content. A negative value for ,1 in the regression equations captures this inverse
relationship between Ksat of the radon barrier and the resulting water content of the radon barrier.
5.4.1.2 Exploration of Ksat Variation in Upper Cover Soils Layers 1 and 2
Sensitivity of Ksat in the upper portion of the cover was not evaluated in the v1.4 modeling. This
parameter was fixed at a value of 4.46 cm/day for all 50 simulations. A set of additional runs was
performed in the preparation of this response to explore sensitivity of the cover performance
(percolation) to this Ksat in the top portion of the model. Nine Ksat values were selected based on
the v1.4 distribution of Ksat of the radon barrier and applied to model layers 1 and 2. This was
deemed a reasonable range for exploration given that the radon barrier layers are composed of
the same source material as layers 1 and 2, Unit 4 silty clay. However, it is recognized that, for
materials near the surface, saturated conductivity values are expected to be elevated compared to
laboratory measurements or typical ranges associated with a given soil texture due to factors like
biotic turbation and freeze/thaw cycling (Benson et al. 2011). For example, NUREG/CR-7028
recommends a value of about 22 cm/day for cover soils.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 47
The nine values were selected based on a range spanning the 5th percentile to 95th
percentile values of the radon barrier Ksat distribution. They were, in cm/day: 0.49, 1.07, 1.68,
2.41, 3.37, 4.72, 6.80, 10.69, and 23.25. HYDRUS runs were carried out for five parameter sets
taken from the set of 50 presented in Neptune (2015b) to cover a range of percolation values in
the previous modeling. Other than varying Ksat in the top two model layers, all other model
parameters and structure (meteorological input, initial conditions, averaging procedure for
results, etc.) are identical to the v1.4 process. The simulations selected were numbers 12, 18, 29,
30, and 36. The results of these 45 runs are presented in Figure 25.

Figure 25. Percolation as a function of Ksat for 45 simulations.
The results of these runs show that the lower Ksat values are associated with higher percolation.
While this result may appear somewhat counterintuitive, it is important to note that in
complicated unsaturated zone problems, higher Ksat does not always necessarily imply high flow
through the system as it might in a saturated context. For example, in Section 5.7.1, it is noted
that the frost protection layer, despite having a very high Ksat, has very low unsaturated
conductivity for dry pressure head regimes, and therefore, it can act as an impediment to flow;
this is the operating principle of a capillary barrier. The context here is different, but the same
perspective applies: it is not appropriate to assume that higher Ksat necessarily leads to higher
flow rates through the cover simply because Ksat appears as a multiplicative term in Richard’s
Equation. If the system were such that the boundary conditions were, for example, fixed constant
head boundaries, then this logic may apply. However, as discussed below, in this context, Ksat
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 48
and the boundary fluxes are interdependent, and a more accurate description of the behavior
relies on understanding their relationship.
In this case, the sensitivity of percolation is related to the ability of the topmost layers of the
cover to convey water upward for release out of the model domain via surface evaporation. As
shown in Section 5.8.1, the flow in the top two model layers is usually upward toward the
surface because evaporation creates a down-to-up pressure head gradient. During days without
precipitation, the pressure gradient is established by the potential evaporation and the current
moisture conditions in the profile, and Ksat governs the rate at which water flows toward the
surface. If the Ksat is low, less water is able to move toward the evaporative surface (the topmost
node in the 1D model), and more water is retained in the profile. Subsequent precipitation events
would be met with a wetter profile, and therefore, a breakthrough of the capillary barrier is more
likely. The net result, in the case shown here for the ET cover design, is that lower Ksat in the
upper layers ultimately result in higher rates of percolation.
During and immediately after precipitation events, downward flow is also governed by Ksat of the
surface layers. If Ksat is very low, the soil profile will not imbibe all of the meteoric water and
runoff will occur. Runoff was indeed observed in the model output for the lowest three Ksat
scenarios, though runoff events were infrequent. For example, for Simulation 12, the run with the
lowest Ksat listed above, runoff was recorded for 62 days in the last 100 years, totaling about 31
cm of water, compared to approximately 900 cm of infiltration into the column. Conversely,
Simulation 12 with the highest Ksat showed no runoff and 931 cm of infiltration through the top
node of the model over the same 100-year period.
For water that has infiltrated, its ultimate fate depends on whether the capillary barrier can
maintain a pressure regime of low flow in the frost protection zone (Section 5.7.1), allowing
evapotranspiration processes to dry out the upper layers of the column. These model results
suggest that, for the cover soil layers, low Ksat throttles the release of water via
evapotranspiration to a greater degree than Ksat inhibits infiltration via runoff. Comparing the
same two model runs of Simulation 12, the low Ksat run recorded 683 cm of evaporation from the
top of the model, while the high Ksat run recorded 799 cm of evaporation. In summary, Ksat in the
top layers of the model is interdependent with the boundary flux at the top of the model. For very
low Ksat values, this causes a throttling of evaporation that results in higher percolation.
The discussion above centers on evaporation from the top node rather than transpiration through
the root zone. This is because, as parameterized in v1.4, outflow from the model is very
evaporation dependent due to the root water uptake parameters, which conservatively limit root
uptake to a relatively narrow range of pressure heads. This idea is explored thoroughly in Section
5.12.1.1, but warrants mentioning here, because the Ksat dependence on percolation is contingent
upon this conservatism. If the root water uptake parameters are relaxed as described in Section
5.12.1.1 (i.e., h50 is set to 1500 cm) for the 45 model runs, the sensitivity of percolation to Ksat
disappears entirely. The results of this second set of runs are shown in Figure 26. Percolation is
reduced dramatically to values resembling the lowest results from the v1.4 HYDRUS
simulations. This is due to the roots providing another viable mechanism for water to leave the
domain.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 49
Another conservatism that was not explored here is the limiting pressure on the top model node,
which HYDRUS calls “hCritA”. This is the minimum pressure head that HYDRUS allows at the
top of the model and is set to -15,000 cm for these simulations. For these materials, this typically
yields a volumetric water content around 14%, or about 30% saturation. This also limits
evaporation at the surface conservatively with respect to percolation, as soil in top few
millimeters of the cover will be much drier than this during periods of high temperature and
insolation.

Figure 26. Percolation vs Ksat with h50 = 1500 cm for the same 45 model runs presented
above.
To summarize, the sensitivity of percolation to Ksat in the evaporative zone of the model was
explored by preparing a set of 45 new simulations based on the parameterization and structure of
the runs presented in the v1.4 modeling. Increasing percolation was associated with low Ksat
values in the evaporative zone due to its interdependence with the surface boundary evaporative
flux. However, this low range of Ksat values that resulted in increased percolation do not
accurately characterize the hydraulic properties of the surface soils. As described in NUREG-
7028, near surface processes will ultimately increase the Ksat of materials in the upper layers of a
cover system over time. Further investigation of model conservatisms in the v1.4 model
structure, which tend to suppress both transpiration and evaporation, suggests that the range of
results presented in Model v1.4 is adequate to represent the range of expected behavior in the
cover system.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 50
5.5 UDEQ Comment 5: Hydraulic Properties
Some hydraulic properties (saturated hydraulic conductivity and soil-water characteristic curves
(SWCCs)) used in the PA of the evapotranspiration cover for the proposed Federal cell are
based on historic data (ca. 1980s, early 1990s) obtained from laboratory tests using small
specimens that do not represent larger-scale features that affect the shape and connectivity of
the pore spaces in soils exposed to long-term pedogenic processes, or that yield an unrealistic
representation of the SWCC at lower water contents. Since low-water-content conditions are
expected for Clive’s desert conditions, particularly in the summer, this issue is important.
EnergySolutions should collect and test samples of sufficiently large scale to generate
appropriate saturated hydraulic conductivities and SWCC data and submit these results. If this is
not possible at this stage of the project, EnergySolutions needs to incorporate the new snapshot-
in-time SWCC data obtained from the recent Cover Test Cell deconstruction, at least for the
radon barrier. Other relevant approaches will need to be used for overlying soil layers. Submit
the results of an evaluation of hydraulic properties in relation to the above considerations.
5.5.1 Comment 5 Response
The Cover Test Cell data provide contemporary snapshot-in-time SWCC data for Clive clay soils
as used in radon barrier layers. These data are summarized and explored in Section 5.3.1 above.
While Cover Test Cell data for the radon barrier layers are not directly used in DU PA v1.4 or
the HYDRUS 2D model, they are effectively bracketed by the parameters evaluated.
5.6 UDEQ Comment 6: FPL Properties
Show that the hydraulic properties assigned to the Frost Protection Layer of the
evapotranspiration cover, which were obtained from the Rosetta database, are representative of
long-term conditions naturally developing at the Clive site. Compare the hydraulic properties
assigned to the Frost Protection Layer with the measured and/or described properties of the
Sacrificial Soil Layer from the Cover Test Cell deconstruction. In answering the above questions,
describe how the hydraulic properties assigned to the Frost Protection Layer, a material that has
been defined with a wide-ranging gradational specification, will be consistent with the hydraulic
properties of the Frost Protection Layer that is expected to be constructed given the
aforementioned gradation specification. Also, discuss the inherent difficulties of constructing a
uniform material from such a specification, and how consistency of layer properties will be
maintained spatially and throughout time so that the conditions inherent in the PA model are
realized in the actual cover system over the service life and compliance period of the proposed
Federal Cell. Discuss how these properties are expected to change or degrade over time, e.g.,
due to extreme weather events or other phenomena.
5.6.1 Comment 6 Response
This comment is addressed in three parts: (1) properties of the Frost Protection Layer;
(2) construction specifications and controls needed to ensure that the as-built cover reflects the
modeled properties; and (3) long-term durability of the FPL as a distinct layer over the
compliance period of the Federal Cell.
5.6.1.1 FPL Properties
Material hydraulic parameters for the cover layers, including the FPL, used in DU PA v1.4 are
summarized in Table 2. There are two purposes for the FPL. One is to protect layers below the
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 51
FPL from freeze/thaw cycles, wetting/drying cycles, and to inhibit plant, animal, or human
intrusion (Neptune 2015b). The second purpose is to create a strong capillary barrier that will
deter downward infiltration from the surface and evaporative layer into the FPL. Smesrud and
Selker (2001) and many other studies have shown that a cover design with a fine-grained layer
over a coarse-grained layer provides a strong capillary barrier to restrict the flow of water.
The FPL consists of particles ranging in size from 16 inches to clay size particles. In the
HYDRUS model used with the DU PA v1.4 GoldSim model, the FPL was modeled as a sandy
loam because a sandy loam represents a coarse-grained material with some silt and clay
(Neptune 2015b). Hydraulic properties of sandy loam for the FPL were selected using the
HYDRUS hydraulic properties pull-down menu, which use properties from the Carsel and
Parrish (1988) database of hydraulic parameters.
In order to evaluate the efficacy of the FPL in the DU PA modeling, a what-if scenario
HYDRUS model was built using all properties from the Clive DU PA HYDRUS model, but with
the cover layering, cover thickness, and cover hydraulic properties taken from the tailings cover
design for the White Mesa Mill Site near Blanding Utah (MWH Americas 2007).
The cover design from MWH Americas (2007) was selected because the two sites (Clive and
White Mesa) are similar in climate and setting, but the White Mesa design does not have a strong
capillary barrier (with a fine layer over a coarse layer). Average annual precipitation at Blanding
and Clive is approximately 13.3 and 8.3 inches per year, respectively; the Blanding site has about
60% more precipitation that the Clive site.
MWH Americas (2007) report an average flux rate through the cover system of 1E-4 cm/day
(0.4 mm/yr). When the White Mesa cover layering, cover thickness, and cover hydraulic
properties are used with all other components of the Clive DU PA HYDRUS model, the
percolation out of the bottom of the cover is 1.2 mm/yr; two orders of magnitude higher than the
average annual percolation for the 50 DU PA v.1.4 HYDRUS simulations (0.02 mm/yr). It is
notable that even with less precipitation at the Clive site, percolation was higher at the Clive site
with White Mesa layering than that reported for the White Mesa study (0.4 m/yr).
The results of this what-if scenario model exercise demonstrate the effectiveness of the FPL in
the Clive DU PA cover design. In addition, this discussion is relevant to Comment 12 discussed
below, and the comparison of some western waste sites.
5.6.1.2 FPL Construction Specifications
EnergySolutions has informed Neptune that this will be addressed under separate cover.
5.6.1.3 Long-Term Durability of the Frost Protection Layer
Part 3 of this comment is addressed by considering the following question: Assuming distinct
layers are present at the time of construction, is it reasonable to expect that those layers will
persist over geologic time?
In short, given the environmental and geological conditions at the Clive site, it is likely the FPL
will persist. Note also that the bank run material to be used for the FPL is mined from a gravel
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 52
pit located at an outcrop a few miles north of the site. These materials have persisted as a distinct
deposit since the beginning of the current interglacial climate cycle some 11.6 thousand years
ago (Neptune 2020b).
An assessment of the stability of distinct layers in the cover at the Clive begins with a review of
a similar evaluation of an analog at the Hanford site. The Hanford and Clive sites share
important similarities that are relevant to the potential for soil layers to mix, hence making
Hanford a reasonable analog for the Clive site. At Hanford, Bjornstad and Teel (1993) found that
natural processes (deflation, compaction, illuviation, cryoturbation, bioturbation) did not pose a
significant threat over the next 1,000 years to the stability of engineered barriers. Their assertion
regarding the lack of threat to the stability of the layers within an engineered barrier was based
on: 1) the arid to semi-arid desert environment, and 2) the location of the site within a basin
where significant eolian deposition of fine-grained silt occurs. These are the critical features that
Clive and Hanford have in common with respect to the stability of geologic layers.
Both Clive and Hanford are situated in arid to semi-arid desert environments. The arid climates
correspond to low primary productivity, relatively sparse insect and mammal activity and hence,
minimal bioturbation. SWCA (2011) observed limited density of diversity of vegetation with
average plant species cover consisting of: 14.3% black greasewood, 5.9% Sandberg bluegrass,
3% cover each of shadscale saltbrush and gray molly. Importantly, observed ground cover was
dominated by 79.2% biological soil crust which provide an effective stabilization for the soil
surface. These studies also found that root densities were largely concentrated near the surface of
the soil, with few large, woody roots were encountered in deeper soils. Rooting depths were
shallow, with the maximum rooting depth of dominant woody plant species ranging from 16 to
28 inches. Consequently, plants have negligible impacts on soil turnover.
The density and diversity of burrowing animals including ants and mammals are also limited by
the environmental conditions at the Clive site. SWCA found a low density of ant and mammal
burrows with an average of 24 ant mounds per hectare (9.7 per acre), with anthills covering 4.6%
of the ground surface in field study sites (SWCA 2012). Most of the below ground ant nest
volume is within 24 inches (60 cm) of the soil surface due to the presence of compacted clay and
caliche layers. Ant nest volume and corridor densities generally decrease with depth with most of
the activity occurring in the upper layers.
Within the survey, four categories of mammal burrows were identified: ground squirrels,
kangaroo rats, mice/rats/voles and one badger. Kangaroo rats and the mice/rats/voles represented
the vast majority of identified burrows, with only 2 burrows associated with ground squirrels and
1 badger burrow identified. For the PA model, maximum burrow depth was set at 200 cm based
on best professional judgment (Neptune 2015e). This depth is consistent with that used at NNSS
by Wolf et al. (2005), and represents the likely average vertical extent of multiple badger
excavations (Kennedy et al. 1985). Mammal burrows on average are much shallower and at the
Hanford Site, small mammals generally do not burrow below 10 inches (25 cm) depth (Bjornstad
and Teel 1993). Although badgers are capable of burrowing to depths over 2 m, it is thought that
most “badger burrows” are enlargement of small mammal burrows that were further excavated in
pursuit of prey (Bylo et al. 2014). Collectively, the results of these site-specific plant and animal
surveys provide strong evidence that Clive, like Hanford, has little exposure to meaningful
mixing of soil layers from either plant or animal activity.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 53
Both Clive and Hanford are located in arid to semi-arid desert environments that reside in basins
with significant eolian deposition of fine-grained silt. Neptune evaluated the nature, thickness,
and thickness variations of eolian sediments at the upper part of the sedimentary section in 9
excavated sections at the Clive site on December 15 to December 17, 2014 (Neptune 2015c).
The degree of soil development in the eolian silt is gradational through the deposits indicating
soil formation contemporaneous with eolian deposition. The primary mode of eolian deposition
at the Clive site is deposition of fine-grained silt from suspension fallout during episodic
windstorms (Neptune 2015c). Well-developed soil horizons are not superimposed on the upper
part of the eolian section. This conceptual model is supported by an analysis of a near continuous
record of eolian deposition and a lack of soil formation preserved at the Clive site since the
regression of Lake Bonneville below the Clive elevation (approximately 13,500 years B.P.)
(Figure 27: taken from Neptune (2015c)). Ultimately, aeolian deposition acts fast enough that
true soil horizons cannot form.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 54

Figure 27. Reproduced from A.1 Clive Pit Wall Interpretation (C. G. Oviatt, unpublished
data) and stratigraphic comparison with quarry wall studies from Neptune (2020b).
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 55
In summary, survey data from the site suggest that there are no plant or animal mechanism to
disturb surface soil layers. This is consistent with analysis conducted for similar purposes at the
analog Hanford site. An additional similarity with the analog Hanford site is the significant
aeolian deposition that prevents formation of soil layers. Together these two factors strongly
preclude the existence of any plausible mechanism that would significantly disturb the layers of a
cover. This conceptual model is supported by the work shown from analyses performed on the
site (Figure 27). Ultimately, given the environmental and geological conditions at the Clive site,
it is considered likely that the FPL will persist.
5.7 UDEQ Comment 7: Capillary Break
The Division is concerned that the sharp contrast in hydraulic properties at the interface between
the Evaporative Zone and the Frost Protection Layer as well as at the Frost Protection Layer and
the Radon Barrier result in capillary breaks in the model that may not be consistent spatially or
temporally with actual physical conditions throughout the cover system for 10,000 years.
Document and explain mechanistically why the water content below the Evaporative Zone
appears insensitive to meteorological conditions, based on the HYDRUS simulation outputs.
Document and explain what is/are the controlling mechanism(s) responsible for the apparent lack
of flow across these interfaces, and how will these mechanisms be maintained or remain
operative throughout the required service life and the compliance period associated with the
cover. Perform sensitivity analyses with both sharper and softer contrasts in hydraulic properties
(e.g., Ksat and SWCC) between the layers by systematically varying Ksat, α, and n of the
Evaporative Zone and Frost Protection Layer. Document how predictions made for the cover
model change if the interface with the Frost Protection Layer is removed, damaged, made more
heterogeneous, or comprised of materials that soften or diminish the contrasts at the interfaces.
Discuss how, if a capillary break across the interface with the Frost Protection Layer is
responsible in the model for minimal downward flow, the actual variation in water contents above
and below the Frost Protection Layer will vary systematically with the sharpness of the break.
5.7.1 Comment 7 Response
Questions 7 and 8 center on the performance of the evapotranspiration cover system. These
responses address concerns regarding the sensitivity of the flow through the cover to
meteorological events, specifically describing how the cover's design functions to store and
release infiltrated water in the upper layers while maintaining a steady moisture condition in the
lower layers for all but the most intense storm events. The response to Question 7 lays out the
theoretical basis for this behavior based on the theory of unsaturated flow, and demonstrates how
model output comports with the theoretical understanding. The response to Question 8 delves
deeper into the model output to further support the ideas developed in the response to Question
7, including a detailed presentation of the moisture profile's evolution in response to precipitation
events.
The focus of both responses is to demonstrate that, owing to the semi-arid climate and the cover
design, changes in moisture content and flow are largely isolated to the topmost layers of the
cover system. A brief review of the theory of operation for capillary barrier style ET covers is
given below to provide context, and to build a foundation for a mechanistic understanding of the
modeled behavior.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 56
Capillary barriers take advantage of differences in soil water characteristic curves (SWCC) to
inhibit downward flow through the cover system. These differences arise from the pore size
distribution of the materials in the various cover layers. In ET covers, including the proposed
cover at the Clive site, a finer grained material is typically placed on top of a coarser grained
material, intentionally creating an interface of differing pore sizes. As noted in EPA (2011):
The discontinuity in pore sizes between the coarser-grained and finer-grained layers forms a
capillary break at the interface of the two layers. The break results in the wicking of water
into unsaturated pore space in the finer grained soil, which allows the finer grained layer to
retain more water than a monolithic cover system of equal thickness.
The purpose of the upper, fine grained material (known as the evaporation zone in the model) is
to maximize the potential for water storage to facilitate its removal by evapotranspiration, while
at the same time minimizing the downward flux into the underlying coarse material (known as
the FPL in the model). The fine-grained material should also have sufficient permeability to
accept precipitation into the soil, in order to limit other undesirable impacts such as excessive
runoff and allow for plant growth on the surface. The permeability of the material also impacts
the rate at which water is redistributed in the fine-grained material due to moisture gradients,
which can be important for moving water upward into the evaporative zone after a precipitation
event. The goal of the ET cover design is to store all precipitation in the upper fine grained layer
and make it available for removal upward out of the soil column by ET. If designed properly,
and site conditions allow, no water flows through the coarse material.
A lack of significant flux below the coarse layer, as can been seen in some of the HYDRUS
model simulations for the Clive cover, might superficially suggest that the results are insensitive
to meteorological conditions, but they are actually a product of both meteorology (e.g., potential
ET, precipitation amount and pattern) and the cover design. The key questions addressed below
are: 1) Under what circumstances can the evaporative zone material store and release the
incoming precipitation while limiting flow to the underlying coarse material, thereby making
flows in the lower cover seemingly unresponsive to precipitation? 2) Can this be explained
mechanistically using the principles of unsaturated flow? 3) What are the critical material
properties governing these mechanisms?
The mechanics of this interaction are best understood by examining the SWCCs of the materials
involved, which govern water content and hydraulic conductivity as a function of pressure head
(tension). As an example, the SWCCs and properties discussed below are taken from the v1.4
HYDRUS model simulation that produced the highest percolation through the cover (simulation
20 of 50). However, the analysis can and will be applied to a variety of parameter sets as
requested by the comment.
Under unsaturated conditions, a coarse material that lacks a significant fraction of fine pores will
not imbibe water under high water tension (negative pressure head, h) because of the small
capillary forces associated with a larger pore structure. Large pores cannot hold water or support
flow at high tensions because capillary forces vary as the inverse of pore radius, which implies
that large pores are empty at high tension. Figure 28 shows the SWCC for the FPL as
implemented in the v1.4 HYDRUS model, which used the Van Genuchten model with α and n
equal to 7.5 m-1 and 1.89, respectively. Under normal circumstances, this material will be quite
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 57
dry in a semi-arid environment as the specific retention (i.e., the water content under only gravity
drainage) is quite low. Specific retention can be approximated as the water content at a tension of
3 m (Stephens 1996), which equates to a water content only about 0.02 above the irreducible
water content (θr) for this material.

Figure 28. SWCC for the FPL as modeled in v1.4.
For this material, at higher tensions (h), even large changes in tension result in very small
changes in volumetric water content. For example, an order of magnitude change in tension from
100 m to 10 m is associated with a change of water content from 0.066 to 0.072 for the SWCC
shown in Figure 28. As stated above, this is due to the lack of small pore sizes in the material.
The importance of prevailing tensions in the system is also reflected in the unsaturated hydraulic
conductivity function for the materials, as shown in Figure 29. Though a coarse material
typically has a high hydraulic conductivity under saturated conditions (Ksat), under high tension
unsaturated conditions, the conductivity can be very low. The modeled frost protection layer, for
example, has a Ksat of over 100 cm/day, but at a tension of 10 m, the conductivity is virtually
zero (~3E-7 cm/day). Unsaturated conductivity vs pressure head is shown in Figure 29 for the
FPL and the evaporation zone as simulated in the same high percolation model simulation 20.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 58

Figure 29. Unsaturated conductivity as a function of pressure head for a coarse and fine
material.
Note that the conductivity for the evaporation layer exceeds that of the FPL in the high tension
regime down to pressure heads of about -0.6m, despite the fact that the Ksat of the FPL is nearly
two orders of magnitude greater than that of the evaporation zone.
At lower tensions, the larger pores in a coarse material can become saturated, and as a result the
water content and conductivity increase markedly. As can be seen in Figure 28, the SWCC bends
sharply at around tensions of 2 m to 10 m. An order of magnitude change in tension in this range,
for example from 3 m to 0.3 m, would result in a water content change from 0.087 to 0.218,
while the conductivity would change four orders of magnitude from about 4.5E-5 cm/day to
4.9E-1 cm/day. The dramatic increase in conductivity is due to the fact that frictional forces vary
inversely with the fourth power of the pore radius.
Thus, the range of tensions maintained in the frost protection zone and at the interface with the
evaporation zone will be of critical importance to the flow through it, as lower tensions would
engage larger pores in flow and storage. Water coming downward through the cover would have
to generate a large enough pressure pulse to overcome the difference in capillary forces between
the fine and coarse layers in order to fully penetrate the cover system. Stephens (1996)
summarized this effect as follows:
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 59
Owing to heterogeneity, the downward percolation of water or redistribution may virtually
cease where the infiltrated water migrating through a fine soil encounters a dry and relatively
uniform, coarse textured layer. This occurs when the pressure head in the water pulse is not
sufficiently great to force water to enter the large pores of the coarse soil.
This can also be observed through the lens of specific moisture capacity, which is defined as the
first derivative of the SWCC with respect to pressure head (dθ/dh), and interpreted as the volume
of water released or taken into storage per unit change in pressure head. The specific moisture
capacity for the FPL is shown in Figure 30. Note that for tensions above about 2.5 m, the specific
moisture capacity is essentially zero, meaning that little water is taken or released from storage
despite potentially large changes in pressure. The value of the pressure head associated with this
threshold depends, of course, on the SWCC for the material. However, any coarse material
suitable for a capillary barrier will have similar-shaped SWCC and specific moisture capacity.

Figure 30. Specific moisture capacity for the FPL as modeled in v1.4.
The cover performance thus hinges on whether the evaporative zone can store and release the
incoming precipitation while maintaining sufficiently high tensions such that the coarse material
cannot store or conduct a significant amount of water. If tension deviations occur at significantly
higher values, (i.e., to the right side of the specific moisture capacity curve in Figure 30), then we
should expect the flow through the FPL and below to be insensitive.
Figure 31 shows this behavior in the model output for Simulation 20. Fifty years are shown,
including a notably wet period beginning at around model year 890. The tension at the interface
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 60
of the evaporative zone and FPL (green curve) is constantly changing as water is moved in and
out of the model domain via precipitation and evapotranspiration. The flux through the FPL
(blue curve) is insensitive to most of the oscillations, except when the tension dips to around 250
cm or below (red line). For those periods, we see a response in the flow through the frost
protection layer, as predicted by examination of Figure 30. At around model year 890, there are
two significant drops in tension, with the second drop down to around 100 cm. This produces the
most significant downward flux in the period shown.

Figure 31. Evaporation zone pressure head and flux through the FPL vs time for a wet
period in Simulation 20. The red dotted line is drawn at a pressure head of -250 cm.
Flux values are negative for downward flow.
While the above discussion focuses on the SWCC for the coarse frost protection layer, the
SWCC of the finer evaporation zone material also warrants discussion. In fact, the regression
model in the DU PA Model v1.4 reflects that percolation is sensitive to the SWCC parameters in
the evaporation zone. This can be understood in the context of the threshold effect described
above for the coarse layer.
For the cover to perform effectively as a capillary barrier, the evaporation zone needs to store
and release infiltrated water while maintaining a range of pressure heads that preclude flow in
the underlying frost protection layer. Performance can thus be estimated by evaluating how much
water can be stored in the evaporative zone while keeping tensions below this critical range.
Figure 32 shows the SWCC for the FPL along with the evaporation zone layer for two different
simulations from the v1.4 modeling.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 61

Figure 32. SWCCs for the FPL and two realizations of the evaporation zone layer. The tan
horizontal line is drawn at a tension of 250 cm, while the dotted lines indicate the
corresponding water content for the evaporative zone curves.
Examination of Figure 32 shows that, at a tension of 250 cm, the volumetric water content of the
evaporation zone in simulation 6 (red) would be at around 38%, while the water content for the
evaporation zone in simulation 20 (blue) would be around 28%. Viewed through this lens, it is
not surprising that the resulting percolation in simulation 6 was much lower than in simulation
20 (7.9E-3 mm/yr and 1.8E-1 mm/yr, respectively), as the evaporation zone in simulation 6 can
store much more water before significant breakthrough of the FPL occurs. This is precisely the
effect mentioned in the question as the “sharpness of the break” in material properties.
The 50 simulations performed to support Model v1.4 each had a different SWCC for the
evaporation zone, and thus comprise a systematic set of simulations that vary the sharpness of
the material contrast at this interface. The regression equations for percolation and water content
incorporate these relationships. The parameters of the van Genuchten SWCC model are θs, θr, α,
and n. The first two specify the saturated and residual water contents, respectively, which define
the endpoints at either end of the SWCC as shown in Figure 32. The remaining parameters
define the shape of the intervening curve. Alpha can be simply interpreted as inversely
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 62
proportional to the pressure head at which the inflection point on the right side of Figure 32
occurs, commonly known as the air entry pressure head. Simulation 20 has a higher α
(0.028 1/cm) than Simulation 6 (0.011 1/cm), and, therefore, a lower air entry pressure. The n
parameter is related to the grain size distribution, and can be simply interpreted as inversely
related to the slope of the central portion of the SWCC. For example, the FPL has the highest n
value in Figure 32 (1.89, vs. 1.28 and 1.38 for simulations 6 and 20, respectively) and, therefore,
the least steep slope in the central portion of the curves pictured.
With these relationships in mind, close examination of Figure 32 would predict that the best
performing evaporative zone materials should have low α and low n, as these properties would
be most conducive to water storage while maintaining relatively high pressure heads (i.e.,
pushing the evaporation zone curves up and to the right in Figure 32). Conversely, based on
Figure 32, poor performance would be associated with high α and high n. These relationships
manifest in exactly this fashion the regression equations of Model v1.4, which associate higher α
and higher n in the evaporation zone with increased percolation. These relationships would be
reversed, of course, when discussing the properties of the frost protection layer, in which high α
and high n are desirable for limiting percolation.
In summary, a mechanistic explanation for the capillary barrier performance is presented based
on the theory of unsaturated flow. Insensitivity of conditions in the lower portion of the cover to
meteorological forcing is the expected and intended behavior in such a cover design, as the
evaporative zone stores and releases water while maintaining a pressure regime in which the
coarse underlying materials do not allow flow. An example of the predicted behavior in the
model output was presented. The sensitivity of the sharpness of the material contrast was
evaluated in the 50 simulations performed for v1.4. The relationships predicted by the regression
equations comport with the predictions of the analysis of the system dynamics based first
principles of unsaturated flow. As suggested by the question, the material contrast is fundamental
to the performance of the cover system.
5.8 UDEQ Comment 8: Water Balance Graphs
Provide annual water balance graphs over a 10-year period for each of the model layers, in
addition to water balance graphs provided earlier. Some graphs should lead up to and follow
extreme weather events, and all should have sufficient detail so that the mechanisms controlling
flow can be understood and validated.
5.8.1 Comment 8 Response
Building on the theory presented in Section 5.7.1, plots of tension, water content, and
upward/downward fluxes are provided for v1.4 simulation #20 to demonstrate the mechanisms
controlling flow through the top slope ET cover materials through time. As discussed in the
Section 5.7.1, a capillary barrier is created by placing a fine-grained material (evaporative zone
layers) on top of a coarse material (frost protection layer). This configuration is of critical
importance to the performance of the ET cover design for the site; the capillary barrier prevents
water from traveling downward into the radon barrier clays, allowing water to be stored in the
evaporative zone where it is subject to evapotranspiration processes. The capillary barrier is
formed due to the differences in soil water characteristic curves of the material that arise from
different pore size distribution of the materials. In this section, the mechanisms controlling flow
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 63
across this critical barrier are discussed using the output of observation nodes placed at key
locations within the model column. The model output shows that the capillary barrier is effective
at preventing water from passing through it, however during rare and extreme wetting events,
conditions may develop that result in a “break” of the capillary barrier.
A 10-year period is selected, from model years 390 to 400, over which one large (2.7 cm) storm
and several successive moderate (0.5 cm) precipitation events occur during the Fall/Winter cycle
of model years 390/391. This 2.7 cm storm is the 4th largest storm in the 100-year record,
surpassed only by three slightly larger storms approximately 2.77 cm in magnitude. The effects
of this single large storm, followed by several moderate storms, can be observed from the output
provided by observation nodes placed within each layer that output daily values of tension, water
content, and flux.
Figure 32 compares the SWCC for the FPL and evaporative layer and indicates that in order to
have communication between these layers, the tension must be lowered to approximately 250 cm
in the evaporative zone. It is only at this point that the large pores of the FPL become available to
transmit water, and the hydraulic conductivity of the FPL is high enough to transmit water.
During this 10-year period, the capillary break is “broken” through the reduction of tension in the
evaporative layers to less than 250 cm, resulting increases in water content and elevated
downward fluxes in the frost protection and radon barrier layers.
Previously, only 6 observation nodes were used in the v1.4 models, one at the center node of
each layer, and one additional observation node at the bottom of the model (indicated with red
squares in Figure 33). An additional 9 observation nodes were added in the domain, one
additional at the top and bottom nodes of each layer (indicated with black circles in Figure 33).
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 64

Figure 33. Additional nodes were added to the model domain to improve detail in the
output for each layer. Red dots represent observation nodes used in DU PA v1.4;
black circles represent the additional nodes.
Water content output from the observation nodes for selected days around the large 2.7 cm storm
is shown in Figure 34. Model day 142729, the blue line, is the day that precedes the storm. While
elevated water content can be observed in the lower portion of the evaporative zone
(approximately 20-45 cm below the surface) from previous precipitation events, water content is
comparatively lower at the surface. On model day 142730, when the 2.7 cm storm occurs, water
content suddenly increases to almost 40% at the surface of the model, while the lower nodes are
largely unchanged. Three days later (model day 142733) strong evaporative processes at the
surface have lowered the water content in the upper portion of the evaporative zone
(approximately 0-20 cm below the surface), and the pulse of water can be seen via elevated
water contents in the lower portion of the evaporative zone (approximately 20-45 cm below the
surface). After an additional 3 days (model day 142739) the surface has returned to a water
content similar to that before the storm, and the pulse of water can be seen lower into the
evaporative zone. Despite the magnitude of this large storm, elevated levels of water content are
only observed in the evaporative zone. This pattern of wetting and subsequent drying out of the
evaporative zone, without appreciable impact to water content in the frost protection layer, is
how the cover system has been designed to work, and this pattern repeats throughout the climate
record.
However, prolonged periods of high precipitation frequency and low evapotranspiration can
exceed the evaporation zone’s capacity to store and release water without inducing flow in the
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 65
FPL, which ultimately results in drainage out of the lower clay layers; the moisture profiles in
Figure 35 show one such instance. As Section 5.7.1 indicates, increased flow between the
evaporative and frost protection zones occurs when the tension in the evaporative zone is lower
than approximately 250 cm. On day 142870, water content at the top of the FPL is at its
maximum following a particularly wet period in the meteorological record, and output is shown
in 10-day increments thereafter. With each 10-day increment, water content increases in the
lower portion of the FPL as water moves downward from the evaporation zone. These two
moisture profiles show two distinct behavior patterns of the cover over short timescales. In
Figure 34, moisture fluctuations are limited to the evaporation zone, while in Figure 35, a
pressure pulse results in percolation through the frost protection layer. To appreciate how these
patterns manifest at the broader timescales, the figures and discussion that follow show similar
behavior playing out over periods of years, and contextualize them further by the inclusion of the
corresponding meteorological record.

Figure 34. Water content at observations nodes on selected days around a large 2.7 cm
storm event.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 66

Figure 35. Water content at observations nodes following a period of high precipitation
frequency that results in flow through the frost protection layer.
Daily output for the ten model years (model years 390 to 400) of water content in the top,
middle, and bottom nodes of each layer for v1.4 simulation #20 is shown in Figure 36. In the
upper portion of the figure, the horizontal bands of colors represent water content as it changes
through time. Red colors indicate approximately 10% volumetric water content, green colors
20%, blue 30%, and purple approaching 40%. In the lower half of the figure the daily
precipitation and calculated PET record input to the model is provided. Figure 37 provides daily
tension in each observation node through time. Tension is shown on a log10 scale due to the
wide range of tensions present across the cover materials at any particular time during the
simulation. Yellow colors represent tension of approximately 100 cm, blue colors represent
tensions greater than 1000 cm, and purple indicates tensions over 10,000 cm.
For a closer look at annual cycles in the model, the first three years are selected of the ten shown
in Figure 36 and Figure 37. Water content and tension are shown for this three-year period in
Figure 38 and Figure 39, respectively. Additionally, the upward and downward daily fluxes are
shown in Figure 40 and Figure 41.

31

M
a
r
c
h

2
0
2
1

67

Cl
i
v
e

D
U

P
A

M
o
d
e
l
—Re
s
p
o
n
s
e

t
o

D
W
M
R
C

1
2
-3-20
2
0

C
o
m
m
e
n
t
s

Figure 36. Ten years of daily water content in the v1.4 simulation #20 model.
31

M
a
r
c
h

2
0
2
1

68

Cl
i
v
e

D
U

P
A

M
o
d
e
l
—Re
s
p
o
n
s
e

t
o

D
W
M
R
C

1
2
-3-20
2
0

C
o
m
m
e
n
t
s

Figure 37. Ten years of daily tension in the v1.4 simulation #20 model.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 69

Figure 38. Three years of daily water content in the v1.4 simulation #20 model.

Figure 39. Three years of daily tension in the v1.4 simulation #20 model.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 70

Figure 40. Three years of daily (upward) fluxes in the v1.4 simulation #20 model.

Figure 41. Three years of daily (downward) fluxes in the v1.4 simulation #20 model.
The pink line in the lower portion of the figures show the calculated daily PET, which can be
used to orient the reader to where the seasons are along this time series. PET is highest in the
summer, and lowest in the winter. Therefore, as indicated on the charts, this three-year period
begins in the winter of model year 390, and after three years terminates at the end of the fall of
model year 392.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 71
Early in the precipitation record, a large 2.7 cm storm occurs over a single day in the early fall of
model year 390. Following this large event, water content increases and tension decreases in the
surface and evaporative layers as the water is taken into storage (Figure 38 and Figure 39).
Strong downward fluxes are observed in the shallow layers immediately following this large
precipitation event (Figure 41). However, despite this large addition of water to the evaporative
zone, the tension in the evaporative layer remains relatively high. Green colors indicate the
tension remains well above the 250 cm threshold needed to break the capillary barrier in this
configuration (Section 5.7.1); bright green colors correspond to approximately 600–800 cm.
While this storm is particularly large in magnitude, it takes place at a point in the year following
the warm and dry summer. The summer period has left the upper layers with higher tensions and
low water content, with the ability to absorb this large 2.5 cm precipitation event without
reducing the tensions sufficiently (i.e., below approximately 250 cm) to cause a “break” in the
capillary barrier.
Following the large 2.7 cm storm in the fall of model year 390, the winter of model year 391
contains a series of moderate storms (approximately 0.5 cm) that build upon the wet conditions
established by the large 2.7 cm storm, and cause a break in the capillary barrier. The capillary
break is identified in Figure 38 where a noticeable increase in water content occurs in the frost
protection layer, yellow colors corresponding to approximately 12%. This same instance is
indicated in Figure 39 where the yellow colors indicate tension has been reduced to around 100
cm, allowing communication between the layers as predicted by Figure 32.
Following the winter of model year 390, the surface and evaporative layers return to lower levels
of water content and high tension as PET increases and dries out these layers. During this period,
upward fluxes are observed in the evaporative zone (Figure 40). While conditions in the
evaporative zone return to tensions that prevent significant communication between the
evaporative and frost protection layers, elevated downward fluxes are observed in the radon
barrier for at least one year (Figure 41). The following winter of model year 391/392 is not as
intense as the previous winter. Water content in the evaporative zone remains low enough that
tensions are maintained well above the 250 cm threshold, and little communication is seen
between the evaporative zone and the frost protection zone.
5.9 UDEQ Comment 9: Abstraction Model
Demonstrate the efficacy of the abstraction model used to determine percolation rates used in
GoldSim by conducting an independent set of blind-forward simulations with HYDRUS over a
broader range of conditions to represent the range of percolation rates in the abstraction model.
Directly compare percolation rates from the independent forward simulations to those predicted
from the abstraction model for the same conditions.
5.9.1 Comment 9 Response
The purpose of this analysis is to demonstrate the efficacy of the abstracted regression model
used to determine percolation rates used in the GoldSim v1.4 PA model. This was accomplished
by running an independent set of blind-forward simulations in HYDRUS for a set of relevant
input values that represent the range of percolation rates in the abstracted regression model.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 72
Regression equation (41) from the unsaturated zone modeling white paper (Neptune 2015b) is
presented here:
!"#$%= −0.32921 +5.56826 ∗4 +0.19538 ∗"
Where Infil is the net infiltration through the cover (percolation) in mm/yr.
The percolation rates from the blind-forward HYDRUS runs were then compared to those
predicted from the abstracted regression model for the same conditions.
Regression models are only ever considered appropriate or relevant for the range of the data for
which they are fit. Extrapolation of a regression equation outside the range of the data is usually
discouraged or inappropriate. Even within the range the uncertainty is greater at both ends of a
regression model than in the middle, largely because there is more information (data) relevant to
fitting the middle as the tail data is usually sparser. In addition, the lack of correlation between
parameters in the HYDRUS model can cause implausible combinations of inputs, which can
result in outlier effects.
Consequently, 10 blind-forward runs were made to cross-validate the regression model used in
the GoldSim v1.4 PA model that come from within the range of the response (percolation)
realizations, and to also avoid the potential for outliers that might arise from unlikely
combinations of inputs to unduly influence the regression.
The regression in the equation above was based on 50 HYDRUS runs, 90 percent or more of
which have percolation that ranges from about 0.02 to 0.1 mm/yr. This was the focus of this
effort to demonstrate through cross-validation that the regression equation is reasonable. Section
5.11 addresses concerns about the tails of the distribution of percolation and water content in
each cover layer in HYDRUS compared to GoldSim. The simulated GoldSim model has longer
tails because thousands of realizations are obtained, in which case the realizations reach further
into the tails of the input distributions. The runs that are far into the tails of the inputs are much
more unreliable than those from the center and stretch the regression equations for percolation
and water content beyond the range of the inputs for the abstracted regressions. This response
focuses on results that are considered more reliable because they do not represent combinations
of inputs from the tails of their distributions.
Ten random draws from the distributions of the 4 and n parameters used in the regression were
generated from within the output range of percolation rates from 0.02 to 0.1 mm/yr. These were
then fed into both the HYDRUS model and the regression equation presented above. Table 9
presents the values for 4 and n, and the resulting percolations from both-HYDRUS and the
abstracted regression equation.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 73
Table 9. Random sample from trimmed set of values for α, n, and percolation from both
HYDRUS and the regression model.
α (1/cm) n Percolation
Regression
Percolation
HYDRUS
0.019924 1.385094 0.052351 0.0562
0.029998 1.278558 0.013219 0.0232
0.016467 1.281967 0.018487 0.0251
0.011261 1.402093 0.062961 0.0542
0.010965 1.295776 0.008608 0.0233
0.01874 1.315412 0.012952 0.0231
0.017969 1.382271 0.068538 0.0844
0.011956 1.323457 0.013646 0.0238
0.026246 1.399699 0.032545 0.0322
0.011679 1.320503 0.032144 0.0307

Figure 42 shows that the output from these 10 blind forward realizations from HYDRUS provide
general agreement with the values produced by the regression model. These blind forward runs
indicate that the abstracted regression equation is reasonable for the range of data that are
important in the GoldSim v1.4 PA model.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 74

Figure 42. Scatterplot of percolation values computed from both the regression model and
HYDRUS using the same pairs of α and n that were randomly generated.
5.10 UDEQ Comment 10: Characterizing Uncertainty
The standard error of the mean is used to account for uncertainty in input parameters in the PA
models. Provide the rational basis for the appropriateness of this approach to characterize
uncertainty, including appropriate documentation of supporting information from the hydrologic
literature specific to unsaturated flow and vadose-zone processes.
5.10.1 Comment 10 Response
A challenge with the development of probabilistic PA models is putting the right numbers in the
model. Despite the existence of information that can be used to characterize aspects of input
distributions, aggregating information from sources with varying spatial and temporal scales to
meet the needs of a PA model is non-trivial. PA models represent processes with varying
intrinsic temporal and spatial scales, yet PA models cover large spaces (volumes or areas) and
are run for thousands of years. Available data typically do not correspond to the spatial and
temporal resolution of a PA model. Data gleaned from literature review generally correspond to
points in time and space and as such, characterize variability associated with the underlying
populations. (Blöschl and Sivapalan 1995) address this issue using the framework that
distinguishes among the process scale, observation scale and modeling scale. We consider the
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 75
observation scale to be equivalent to the population distribution. In this context, informing PA
models with population distributions can grossly over-represent the variability of input
parameters within the context of the spatial and temporal resolution of a PA model.
Consequently, the development of input distributions for PA models involves scaling the
available data such that larger spatial and temporal domains are characterized.
A number of different approaches have been assessed to accommodate the depiction of processes
with varying spatial and temporal scales in modeled systems (Arora et al. 2019; de Rooij 2011;
Rödenbeck et al. 2001; Yang et al. 2017). This issue can be addressed by running PA models in
one of two ways: 1) by drawing random numbers from input distributions at every time step or,
2) by selecting a random realization from each input parameter distribution at the beginning of
time and applying that value throughout time. These two ways of modeling are sometimes called
fast and slow models in the literature on stochastic averaging, where it is understood that fast
models are approximations to slow models and are formed through averaging processes (cf.,
Thompson et al. (2015)).
Complex models that have disparate process, observation and modeling scales require either
downscaling or upscaling to ensure that the models properly characterize the data when forming
input distributions for the model. For PA models it is almost always the case that spatial and
temporal upscaling is needed. This issue is also addressed in NUREG/CR 6805 (Neuman et al.
2003).
As described in the literature, including the cited references herein, there are several reasons why
stochastic averaging is considered necessary and appropriate. Similarly, the Goldsim PA model
built for this site does not allow for option #1 for several somewhat related reasons: i. the relative
computational intractability; ii. these types of models are often called “systems-level” models
and are aimed at the needs of decision making under uncertainty, in which case it is uncertainty
in parameters that is of importance in the context of the decision to be made; and, iii. the
complications this approach causes for global (simultaneous) sensitivity analysis of a
probabilistic model. Consequently, #2 must be addressed. This problem is common to all
complex modeling of which PA modeling is only one.
With approach #2, the input distribution must reflect values that are plausible across the duration
of the simulation period. Clearly in this case, using the underlying distribution of data is
inappropriate. For example, when a value representing the 99th percentile of the underlying
distribution of the data is selected for use throughout the simulation period, this corresponds to a
highly improbable outcome. In fact, comparing this to case #1 where draws from input
distributions are selected randomly at each time step, this corresponds to a probability of:
(1-.99)^(length of simulation)
If 1,000 years are simulated, this corresponds to what is effectively a zero-probability event.
Hence, when approach #2 is used, input distributions need to be scaled in such a way that there is
lower variability relative to the distribution of the underlying data.
When assumptions of linearity and stationarity are applied, then simple averaging provides an
exact solution for this type of scaling. That is, characterizing the input distribution to a linear and
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 76
stationary model using the mean and standard deviation of the mean (i.e., standard error) of the
available data can be shown to be analytically correct. In this sense, correct implies that this
approach can be shown to provide unbiased estimates for the mean and variance of the output of
interest. This type of analytical analysis helps inform the approach for scaling input distributions
for parameters in PA models.
As more complex non-linear functions are considered, the results and determination of a best
approach become less clear. Exploring this with both analytical and simulation approaches for
non-linear functions shows that simple averaging can provide a biased estimate of the mean and
variance in the output (Black et al. 2019). For example, when a quadratic function is considered,
application of simple averaging results in an estimate of the mean of the process of interest was
biased by approximately 3%. Similarly, for this same case the estimate of the variance of the
process of interest was biased by approximately 2.5%. These results are consistent with the work
of Vogel et al. (1991), who considered the space-time variability of soil hydraulic properties.
Black et al. (2019), also explored the more general case where the function of interest consists of
the product of two independent variables. Similar results were found that using the simple
averaging approach for characterizing the input distributions results in a small bias in the
estimates of both the mean and the variance.
While this is an area of active research with respect to the implementation of scaling approaches
for input distributions for PA models, these results provide important context regarding the use
of the standard error of the mean is used to account for uncertainty in input parameters in the PA
models. There are many complex functions and relationships embedded within PA models.
Functions that are multivariate, highly non-linear, and involve differential equations, exist in PA
models. Ultimately, the aggregate impact of all the functions within a PA model is the function
of interest with respect to assessing the impact of different approaches to scaling.
Of comfort for the GoldSim DU PA model is that the form of the marginal relationship between
the most important/sensitive input parameters and a response of interest in a PA model lies
somewhere between a linear and quadratic (Neptune (2015e), Figures 7 and 8). That is, these
types of complex models are ultimately dependent on only a few input variables for a specific
endpoint. Given this, the impact of scaling with the standard error of the mean to account for
uncertainty in input parameters is a slight bias in the estimation of the mean and the variance.
Ultimately, scaling of some form must be performed to avoid adversely impacting decisions,
because, otherwise, uncertainty will almost certainly be over-estimated, perhaps severely, and
PA decisions risk being made based on values from the tails of the output distributions.
Ultimately, there are no analytical solutions for upscaling for non-linear non-additive models. PA
models tend to be highly non-linear and highly multiplicative. The issue has been recognized
more generally (see references as examples), and research in this area continues to be performed
in various institutions around the world. However, for now there are no simple solutions.
Neptune will continue to perform its own research in this area and has made some breakthroughs
with the use of a novel statistical approach, but this research is not complete. For the current
GoldSim PA model, completed more than five years ago, the best available methods were used,
and not much has changed since, especially with respect to PA in general.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 77
As noted in NUREG/CR 6805 (Neuman et al. 2003), upscaling is necessary. Also, simple
averaging tends to be a very powerful process, in which case the complexity of PA models might
include trade-offs in negative and positive bias so that the center of the probabilistic output is
reasonable.
Tail behavior in the output of non-linear multiplicative PA models is probably more adversely
affected than the centers of the output distributions, but there are other issues in PA modeling
related to lack of correlation structure that probably create more extreme tail effects. Lack of
correlation structure between input variables, and lack of autocorrelation across time almost
certainly leads to more extreme tail effects than the upscaling effect of non-linear and
multiplicative relationships. Consequently, it is reasonable to consider the tails, and especially
the upper tails, outside the range of reasonable results. In effect, the center of the output of PA
models is probably far more reliable than the tails.
A further issue that suggests that upscaling using simple averaging might often be sufficient is
that global sensitivity analysis of complex models always reveals only a few input parameters
that are sensitive for a given output of a PA model. Consequently, it is scaling of those few
inputs that matters the most. When addressing the few inputs parameters that are sensitive, the
lack of analytical solutions for scaling non-linear multiplicative models is not as severe as it
might first seem. Often the response variable of interest is explained in large part by a linear
relationship with the most sensitive input parameters.
For complex models such as PA models, scaling must be done to rectify spatial and temporal
scale differences between processes, observations and models. To not do so would result in far
more problems in a PA model than arise from using averaging as a simple approach to scaling.
5.11 UDEQ Comment 11: Tails of the Distribution
Explain mechanistically why tails of the distribution for water content predicted in GoldSim differ
from those predicted by HYDRUS. Demonstrate that the tails of the distribution for water content
are properly accounted for in GoldSim.
5.11.1 Comment 11 Response
It appears that this comment relates to Figure 26 provided in the 2018 response to interrogatories
(EnergySolutions 2018). Figure 43 below reproduces that figure.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 78

Figure 43. Comparison of Bingham Environmental (1991) water content data with water
content calculated using the regression equation for the DU PA GoldSim model and
with the results of the 20 HYDRUS simulations. Figure 26 of EnergySolutions (2018).
Figure 43 was created in context with an interrogatory relating to DU PA v1.2; and thus plotted
Bingham Environmental (1991) data against those HYDRUS and GoldSim simulations (i.e.,
associated with DU PA v1.2; which has since been superseded by DU PA v1.4). This figure does
not accurately reflect the state of DU PA v1.4 in terms of the relationship between the water
content predicted by GoldSim against that predicted by HYDRUS. Figure 44 presents water
content in the evaporation zone from the 50 HYDRUS simulations used in DU PA v1.4.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 79

Figure 44. Water content in the evaporation zone from 50 HYDRUS simulations used in
DU PA v1.4.
Figure 44 demonstrates that the tails of the distribution for water content are reasonably
accounted for in GoldSim in DU PA v1.4. Note that this figure is compared with 1000 runs of
DU PA v1.4.
Figure 45 compares water content in the evaporation zone from the 50 HYDRUS simulations
used in DU PA v1.4 with the results of the DU PA v1.4 for 50 runs rather than 1000 runs. Figure
45 demonstrates that the tails of the distribution for water content are very well accounted for in
GoldSim in DU PA v1.4.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 80

Figure 45. Water content in the evaporation zone from 50 HYDRUS simulations used in
DU PA v1.4.
5.12 UDEQ Comment 12: Climate Record and Comparison With Other
Sites
Explain mechanistically why the percolation rates predicted with the original DU PA, Model v1.4,
and those utilizing the 1000-year precipitation record differ. Compare the predictions from the
models (e.g., water-content records, fluxes, etc.), and provide a mechanistic reason for the
differences in percolation rate between the sets of predictions that is consistent with the
principles of variably saturated flow and soil-atmosphere interactions. Present, justify, and
document why the model predicts percolation rates that are 10x lower than those that are being
measured in field studies of similar covers in Blanding, and Monticello, Utah.
5.12.1 Comment 12 Response
This comment includes two topics, which will be addressed separately below.
5.12.1.1 Climate Record
The results presented for the 1000-year meteorological record presented in Neptune (2020a)
were in error. We appreciate the question from the Department, as it prompted a re-examination
of the results and discovery of the error, which occurred in the processing of the potential
evapotranspiration (PET) record for input to HYDRUS. This led to unrealistically high PET and,
as the question notes, low percolations. The PET record was revised and the 50 simulations were
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 81
run again. The parameterization of the simulations is the same as documented in Table 9 of
Neptune (2015b).
The 1000-year meteorological record is prepared using the same methodology as the Model v1.4
100-year record, such that the average behavior across the records is similar. Average annual
precipitation and PET for both records are summarized in Table 10. While averages are similar,
the 1000-year record contains higher maximum daily precipitation events, as intended. For
example, there are 48 daily precipitation events higher than 2.77 cm in the 1000-year record,
which was the maximum daily precipitation in the 100-year record.
Table 10. Summary statistics for 100-year and 1000-year climate records.
Meteorological Record Average Precipitation
(cm/yr) Average PET*
(cm/yr) Maximum Daily Precipitation
(cm)
100-year (Model v1.4) 21.4 128.7 2.77
1000-year 21.9 127.1 4.49
*PET computed via the Hargreaves equation as documented in the Model v1.4 Unsaturated Zone White Paper.

The 1000-yr and 100-yr precipitation records are further compared in Figure 46. At both annual
and quarterly scales, the 1000-yr record contains broader ranges of total precipitation than the
100-yr record, though both datasets have similar means. Additionally, the 1000-yr record
contains more periods with exceptionally high precipitation, which are shown in Figure 46 as
points. These values are characterized by being greater than the 3rd-quartile of each dataset by an
amount at least 1.5 times the interquartile range.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 82

Figure 46. Total precipitation over various time periods for the 1000y and 100y records.
Simulations are run for 2000 years, repeating the 1000-year meteorological record twice. Results
are computed by averaging the flux at the bottom of the model over the last 1000 years of the
simulation in order to capture the range of behavior over the record, while avoiding any transient
fluxes early in the simulation associated with initial conditions. On average, percolations with
the 1000-year record are about 10 times higher than the corresponding simulations in Model
v1.4, with an average value of 0.23 mm/yr, compared to 0.024 mm/yr in the Model v1.4
simulations. The maximum percolation with the 1000-year record is 1.3 mm/yr, whereas the
maximum in Model v1.4 is about 0.2 mm/yr. This maximum occurs for the same parameter set
using both meteorological records, simulation 20 of 50, which is discussed at length in Section
5.7.1 due to having the least favorable evaporation zone SWCC of the 50 simulations.
Histograms of the results from these sets of runs are shown below in the top two panels of Figure
49.
The higher percolations are due to the atmospheric conditions breaking the capillary barrier
pressure threshold more often; this idea is described at length Section 5.7.1. Water fluxes
through the cover respond in a non-linear fashion to extreme precipitation events. The new 1000-
year meteorological record contains precipitation patterns that are more intense on a variety of
timeframes. For example, years 214 and 796 have the wettest 60-day periods in the entire 1000-
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 83
year record, with frequent intense precipitation that causes tensions in the evaporation zone to
fall well below the capillary barrier threshold and remain in that range for a prolonged period in
nearly every model realization, regardless of the evaporative zone SWCC. This causes a notable
increase in the flux at the bottom of the cover system.
By contrast, many model realizations of Model v1.4 (100-year record) had no significant spikes
in flow through the bottom of the cover. Using the 1000-year record, these simulations with the
more favorable SWCCs in the evaporation zone exhibited relatively stable long-term trends in
percolation, punctuated by a few rapid increases when the capillary barrier was overcome by
intense periods of precipitation and/or periods of low potential evapotranspiration. Results from
both meteorological records are presented below for Simulation 1 of 50. In the 100-year record,
the flux at the bottom of the cover is not responsive to the meteorological forcings because the
capillary barrier is isolating the bottom of the cover from the top. Using the 1000-year record,
however, the capillary barrier is unable to provide full isolation, and the fluxes are persistently
higher with notable rises around years 214 and 796 (years 1214 and 1796 in the plot, as only the
second 1000-year cycle is shown).
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 84

Figure 47. Flux at the bottom of the cover and precipitation for both the 100-year (top) and
1000-year (bottom) meteorological records for Simulation 1 of 50. Only the last cycle
of the meteorological record is shown. Vertical scales are the same for both plots.
By contrast, for the realizations with least favorable evaporation zone SWCCs, the 1000-year
meteorological record keeps tensions very near the critical range (200–700 cm) throughout the
simulation. As a result, flow through the bottom of the cover is much more responsive to the
meteorological forcing, as high precipitation events readily push tensions below the threshold
pressure at which the FPL has significantly higher hydraulic conductivity, as described in
Section 5.7.1. Further investigation of the persistently high moisture conditions observed for
these simulations reveals that the ratio of actual transpiration to potential transpiration is rather
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 85
low, considering the relatively wet conditions. For example, actual transpiration is about 19% of
potential transpiration for Simulation 20, despite prevailing root zone tensions less than 1000 cm,
typically thought to be favorable for root water uptake.
This highlights a powerful conservatism in the root water uptake parameters. The v1.4 HYDRUS
model uses the S-shaped root water uptake efficiency curve (ɑ) to govern plant water uptake as a
function of water tension. Plants can struggle to draw water from soil if the water is very tightly
held in the soil pore spaces. In very dry soils with tension above the permanent wilting point
(usually regarded as on the order of 10,000 cm for most plants), transpiration ceases entirely.
The S-shaped curve is defined by two parameters, including one called h50; h50 defines the
pressure head at which root water uptake is reduced by 50% due to water stress. The S-shaped
curve used in the v1.4 Model is depicted by the blue line in Figure 48, below, with h50 equal to
200 cm. This greatly reduces the root water uptake in the model, as transpiration is reduced
drastically when soil moisture tensions exceed a few hundred centimeters in the root zone, which
is very common in the model output. Therefore, even in relatively wet conditions, the model is
conservatively throttling root water uptake, leading to low transpiration utilization, as mentioned
above. A second S-shaped curve, with h50 equal to 1500 cm, is also plotted in the figure for
comparison.
Two other water stress models are shown in Figure 48 that use the Feddes water stress model
(Feddes et al. 1978) rather than the S-shaped model. One is taken from the HYDRUS input files
for the Blanding model (MWH Americas 2007). The other is from Taylor and Ashcroft (1972), a
compilation of root water extraction parameters for a variety of food crops. For these two
models, root water uptake would be uninhibited by water stress at tensions below about 1500 cm,
and uptake would still be above 50% up to tensions of about 4000–5000 cm.

Figure 48. Water stress models.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 86
The impacts of this parameterization are not readily apparent when using the 100-year
meteorological record, as breakthrough of the capillary barrier is relatively rare and fleeting.
However, simulations performed using the 1000-year record show persistently wet periods
following intense precipitation that are slow to recover to a drier condition, and despite that, the
ratio of actual transpiration to potential transpiration is still low. This prompted further
examination of the root water uptake model. To appreciate the impacts of this conservatism,
another set of 50 simulations using the 1000-year meteorological record were run with h50 set to
1500 cm. All other parameters are identical to the previous sets of simulations, which varied
three parameters as described in Neptune (2015b).
Under this scenario, percolations are below 0.1 mm/yr for all simulations, and transpiration
efficiency increases dramatically (e.g., from 19% to 66% for Simulation 20). Figure 49 shows
histograms for the 100-year record (Model v1.4 simulations), the 1000-year record with h50 set to
200 cm, and the 1000-year record with h50 set to 1500 cm. Figure 50 shows a simulation-by-
simulation comparison of the results. It is notable that the maximum percolation value for the
1000-year simulations using h50=1500 cm is lower than the maximum from the Model v1.4
simulations that used the 100-year record. The average percolation, however, is about two times
higher than the v1.4 simulations (0.042 mm/yr vs. 0.024 mm/yr).
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 87

Figure 49. Histograms of sets of 50 simulations using the 100-year meteorological record
(top), the 1000-year meteorological record with Model v1.4 root water uptake
parameters (middle), and the 1000-year meteorological record with h50 set to 1500 cm
(bottom).
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 88

Figure 50. Simulation by simulation comparison of percolations derived from scenarios
with h50 set to 200 cm (blue) and with h50 set to 1500 cm (green).
The overall effect of increasing the root water uptake efficiency is a more stable soil moisture
profile that recovers more quickly from perturbations. As a result, more pore space is available to
accommodate infiltration from subsequent precipitation events. Daily evaporation is often lower
when the root water uptake is increased, as the root zone competes with the surface evaporation
for water, and as a result, the upward hydraulic gradient during dry periods is diminished. In
short, the root water uptake model used in the v1.4 HYDUS simulations is extremely
conservative. When the root water uptake parameters are made less conservative for the 1000-
year climate record, percolations are below 0.1 mm/yr for all simulations.
In summary, simulations performed with the v1.4 HYDRUS model structure and the 1000-year
meteorological record produce higher percolations than those predicted using the 100-year
record. A powerful conservatism in the root water uptake model contributes heavily to the
change in results. When this conservatism is relaxed, the predicted percolations are well within
the bounds of the v1.4 HYDRUS simulations.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 89
5.12.1.2 Comparison Across Sites
The second part of the comment asks that the hybrid cover design proposed for the Federal Cell
be compared with similar covers in Blanding and Monticello, Utah; and the estimated
percolation rates through the Federal Cell compared with percolation data collected from the
covers at the other two sites. Figure 51 depicts the basic layering of the cover designs.
Table 11, Table 12, and Table 13 summarize important engineering/hydraulic properties of the
various layers in the Clive, Monticello, and Blanding cover designs. Table 14 summarizes
precipitation and percolation data available across these sites, including monitoring from a Cover
Test Cell at Clive, UT that was constructed to evaluate a previous riprap cover design. See
Section 5.2.1.2 for more details on the design and material properties of the materials used in the
Clive Test Cell.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 90

Figure 51. Layering of ET cover systems at Clive, Monticello, and Blanding, Utah.

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 91
Table 11. Engineering properties of cover layers in the Clive Federal Cell, DU PA v1.4.
Layer Input Parameter3 ET Cover DU PA v1.4 (actual, 50 sims)
Surface
θr (unitless) 0.111
θs (unitless) 0.4089
α (1/cm) 0.0169
n (unitless) 1.3
Ksat (cm/day) 4.46
Evaporative Zone
θr (unitless) 0.111
θs (unitless) 0.481
α (1/cm) 0.0169
n (unitless) 1.3
Ksat (cm/day) 4.46
Frost Protection
θr (unitless) 0.065
θs (unitless) 0.41
α (1/cm) 0.075
n (unitless) 1.89
Ksat (cm/day) 106.1
Upper Radon Barrier
θr (unitless) 0.1
θs (unitless) 0.432
α (1/cm) 0.003
n (unitless) 1.172
Ksat (cm/day) 6.75
Lower Radon Barrier
θr (unitless) 0.1
θs (unitless) 0.432
α (1/cm) 0.003
n (unitless) 1.172
Ksat (cm/day) 6.75

3 Values for α and n in the surface and evaporative zone layer; and for Ks in the radon barrier layers were variable in
the 50 simulations run for v1.4; average value reported here.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 92
Table 12. Engineering properties of cover layers in the Monticello disposal facility.
Monticello Input Parameter Table
Layer Input Parameter Monticello
Gravel Amended θr (unitless) n/a
θs (unitless) n/a
α (1/cm) n/a
n (unitless) n/a
Ksat (cm/day) 30.24
Water Storage & Protection θr (unitless) 0
θs (unitless) 0.3
α (1/cm) 0.0011
n (unitless) 1.31
Ksat (cm/day) 3.6288
Biota Barrier θr (unitless) n/a
θs (unitless) n/a
α (1/cm) n/a
n (unitless) n/a
Ksat (cm/day) n/a
Water Storage & Protection θr (unitless) 0
θs (unitless) 0.29
α (1/cm) 0.001
n (unitless) 1.5
Ksat (cm/day) 0.0320
Sand θr (unitless) n/a
θs (unitless) n/a
α (1/cm) n/a
n (unitless) n/a
Ksat (cm/day) n/a
Tailings θr (unitless) n/a
θs (unitless) n/a
α (1/cm) n/a
n (unitless) n/a
Ksat (cm/day) n/a

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 93
Table 13. Engineering properties of cover layers in the Blanding White Mesa Mill Tailings
Facility.
Blanding Input Parameter Table
Layer Input Parameter Blanding
Gravel Amended
θr (unitless) 0.045
θs (unitless) 0.254
α (1/cm) 0.0145
n (unitless) 1.406
Ksat (cm/day) 5.6
Water Storage & Protection
θr (unitless) 0.055
θs (unitless) 0.404
α (1/cm) 0.0145
n (unitless) 1.406
Ksat (cm/day) 7.4
Compacted Cover (Clay Radon Barrier)
θr (unitless) 0.046
θs (unitless) 0.334
α (1/cm) 0.0229
n (unitless) 1.261
Ksat (cm/day) 3.6
Interim
θr (unitless) 0.059
θs (unitless) 0.439
α (1/cm) 0.0125
n (unitless) 1.461
Ksat (cm/day) 10.4
Tailings
θr (unitless) n/a
θs (unitless) n/a
α (1/cm) n/a
n (unitless) n/a
Ksat (cm/day) n/a

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 94
Table 14. Precipitation and percolation data for the Clive Cover Test Cell, Monticello, and
Blanding facilities. Clive precipitation average calculated for the years 2002–2016 to
match with Cover Test Cell period of service; site average across 28-year
meteorological record is 217.41 mm/yr.
Reference Trinity
Consultants
(2021)
EnergySolutions
(2017)
Benson et al. (2008) Energy Fuels (2021)
Study Clive Clive Test Cell Monticello Blanding
Year
Precipitation
(mm/yr)
Percolation
(mm/yr)
Precipitation
(mm/yr)
Percolation
(mm/yr)
Precipitation
(mm/yr)
Percolation
(mm/yr)
2001 172.5 -- 377.7 0 -- --
2002 147.9 0.652 228.6 0 -- --
2003 183.8 0.446 364 0 -- --
2004 230.1 0.178 442 0.2 -- --
2005 258.1 0.455 519.7 3.8 -- --
2006 187.7 0.256 447.3 0.2 -- --
2007 210.6 0.237 304.5 0 -- --
2008 81.3 0.229 99.8 0.4 -- --
2009 206.2 0.14 -- -- -- --
2010 227.8 0.164 -- -- -- --
2011 246.1 0.11 -- -- -- --
2012 167.4 0.023 -- -- -- --
2013 209.8 0.0001 -- -- -- --
2014 213.6 0 -- -- -- --
2015 247.7 0 -- -- -- --
2016 271.3 0.231 -- -- 59.9 0
2017 193.8 -- -- -- 222.7 0.65
2018 170.2 -- -- -- 163.4 0.9
2019 342.6 -- -- -- 307.9 1.01
2020 116.3 -- -- -- 127.8 0.89
Average 205.96 0.208 347.95 0.575 176.34 0.69

There are several important distinctions between the designs. The Clive and Monticello designs
both employ fine materials at the surface, creating an evaporative zone where water is stored and
released through evapotranspiration processes, combined with a layer of coarse material placed
below the fine materials that creates a capillary barrier to prevent downward flow, holding water
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 95
in the evaporative zone more effectively. From the Monticello design, it is surmised that the
biota barrier gravel likely creates an effective capillary barrier; however, parameterization for
this layer was not available in the references reviewed.
In contrast, the Clive Test Cell and the cover design for the Blanding site do not include a
capillary break. While fine materials are used in the upper layers of the Blanding cover to create
a water storage zone, the design lacks a capillary break that would allow this water to be held in
the storage zone more effectively. The Clive test cell lacks both a storage zone in the surface
layers (consisting of riprap and coarse filter materials), and a capillary break between the
evaporative zone and lower clay radon barrier layers. The hydrologic principles and mechanisms
that govern the performance of a capillary barrier are expanded upon in detail in Section 5.7.1.
The differences in cover design and average precipitation between the sites provides some
insight into the difference in percolation measured from the covers, and the predicted
performance for the Clive cover design from the v1.4 HYDRUS modeling.
Precipitation is higher for the Monticello site compared to the Clive Cover Test Cell and
Blanding covers. While measured percolation rates are higher for the Monticello site compared
to the Clive Test Cell, they are similar to those monitored at Blanding. At first glance, the higher
rates of percolation at Monticello, compared to the Clive Test Cell, may be intuitive since
Monticello has roughly double the precipitation. However, another way to look at the results is
that, despite Monticello receiving roughly double the amount of annual precipitation, only a very
small fraction of this additional water is able to successfully travel through the cover. As such,
despite the markedly higher precipitation at Monticello, the cover is effective at limiting
percolation. The rates of percolation through the Clive Test Cell and Blanding covers may reflect
the more arid climate at these sites, as these covers lack both the storage and capillary barrier
features employed in the Monticello cover design.
In fact, this is what the HYDRUS modeling results indicate when a similar design is subjected to
the climate at Clive, UT. As noted in Section 5.12.1.1 above, average percolation for the 50
simulations run in DU PA v1.4 is 0.024 mm/yr; about an order of magnitude lower than the
Cover Test Cell and slightly more than an order of magnitude lower than the reference
Monticello and Blanding sites.
Figure 51 offers some possible explanations for this predicted difference in percolation. The
Federal Cell design is similar to that of the high-performance Monticello cover, and includes an
evaporative zone in the surface layers to store and release water through time, and a strong
capillary break between the evaporative zone and lower radon barrier layers. By employing
similar features used in the Monticello cover, but subjecting such a cover to only half the amount
of precipitation, results very low predicted rates of infiltration. Additionally, it is also possible
that the Monticello design has shown increased performance in the years following those
reported in Table 14. It is our understanding that additional data is nearing publication. If the
Monticello cover continues to show improvement in its performance, despite the much higher
levels of precipitation, the results predicted by the v1.4 HYDRUS modeling become increasingly
justified.
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 96
6.0 Conclusion
DU PA v1.4 demonstrates compliance with the dose and groundwater protection requirements of
Utah regulations relating to DU disposal. The interrogatory and response process has added to
the record supporting these conclusions; but has not caused the quantitative model to require
revision. Accordingly, DU PA v1.4 remains the basis for demonstrating compliance of the
disposal facility.
Compliance with UAC R313-25-9(5)(a) is affirmed by DU PA v1.4 and Deep Time model v1.5,
together with their supporting documentation as supplemented by the interrogatory/response
cycle.
7.0 Attachments
1. Federal Waste Disposal Cell engineering drawings, series 14004
8.0 References
Arora, B., et al., 2019. Understanding and Predicting Vadose Zone Processes, Reviews in
Mineralogy & Geochemistry 85 (2019) 303–328
Benson, C.H., 2021. Facilitating Informed Decision-Making for Waste Containment Systems,
proceedings of the Waste Management Symposia 2021, March 7–11, Phoenix AZ, 2021
Benson, C.H., et al., 2011. Engineered Covers for Waste Containment: Changes in Engineering
Properties and Implications for Long-Term Performance Assessment, NUREG/CR-7028,
Volume 1, United States Nuclear Regulatory Commission, Washington DC, December
2011
Benson, C.H., et al., 2008. Hydraulic Properties and Geomorphology of the Earthen Component
of the Final Cover at the Monticello Uranium Mill Tailings Repository, Geo Engineering
Report No. 08-04, April 2008
Bingham Environmental, 1991. Hydrogeologic Report, Envirocare Waste Disposal Facility,
South Clive, Utah, prepared for Envirocare of Utah, Bingham Environmental, Salt Lake
City UT, October 1991
Bingham Environmental, 1994. Hydrogeologic Report, Mixed Waste Disposal Area, Envirocare
Waste Disposal Facility, South Clive, Utah, prepared for Envirocare of Utah, Bingham
Environmental, Salt Lake City UT, November 1994
Bjornstad, B.N., and S.S. Teel, 1993. Natural Analog Study of Engineered Protective Barriers at
the Hanford Site, PNL-8840, UC-510, prepared for United States Department of Energy,
Pacific Northwest Laboratory, Richland WA, September 1993
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 97
Black, P., et al., 2019. Scaling Input Distributions for Probabilistic Models - 19472, proceedings
of the Waste Management Symposia 2019, March 3-7, Phoenix AZ, 2019
Blöschl, G., and M. Sivapalan, 1995. Scale Issues in Hydrological Modelling: A Review,
Hydrological Processes 9 (1995) 251–290
Bylo, L.N., et al., 2014. Grazing Intensity Influences Ground Squirrel and American Badger
Habitat Use in Mixed-Grass Prairies, Rangeland Ecology and Management 67 (2014)
247–254 doi: 10.1016/j.rama.2016.07.001
Carsel, R.F., and R.S. Parrish, 1988. Developing Joint Probability Distributions of Soil Water
Retention Characteristics, Water Resources Research 24 (5) 755–769
de Rooij, G.H., 2011. Averaged Water Potentials in Soil Water and Groundwater, and Their
Connection to Menisci in Soil Pores, Field-Scale Flow Phenomena, and Simple
Groundwater Flows, Hydrology and Earth System Sciences 15 (2011) 1601–1614 doi:
10.5194/hess-15-1601-2011
Dobre, R.G., et al., 2017. Snowmelt Infiltration Using Hydrus-1D Based on a Snow Surface
Energy Balance Model for Bucegi Mountains, Romania, available from
https://www.sgem.org/index.php/elibrary-research-
areas?view=publication&task=show&id=3455
Energy Fuels, 2021. Q4-20 Data Quality Report for the Primary Test Section, White Mesa Mill—
Tailings Management Cell 2, Energy Fuels Resources (USA) Inc., San Juan County UT,
January 2021
EnergySolutions, 2015. Radioactive Material License UT2300249: Safety Evaluation Report for
Condition 35.B Performance Assessment; Response to Issues Raised in the April 2015
Draft Safety Evaluation Report, EnergySolutions LLC, Salt Lake City UT, November
2015
EnergySolutions, 2017. Radioactive Material License No. UT2300249, License Condition 28B:
2016 Cover Test Cell Annual Report, EnergySolutions LLC, Salt Lake City UT, February
2017
EnergySolutions, 2018. Radioactive Material License UT2300249: Responses to the Amended
and New Interrogatories Related to Clive DU PA Modeling Report Version 1.4 Dated
November 2015, EnergySolutions LLC, Salt Lake City UT, April 2018
EnergySolutions, 2020. Radioactive Material Licenses UT 2300249 Cover Test Cell
Deconstruction Study Final Report, CD20-0123, EnergySolutions LLC, Salt Lake City
UT, August 2020
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 98
EPA, 2011. Fact Sheet on Evapotranspiration Cover Systems for Waste Containment, EPA 542-
F-11-001, United States Environmental Protection Agency, Washington DC, February
2011
Feddes, R.A., et al., 1978. Simulation of Field Water Use and Crop Yield, Simulation
Monographs, Pudoc, Wageningen, Netherlands, 1978
Kennedy, W.E., et al., 1985. Biotic Transport of Radionuclides from a Low-level Radioactive
Waste Site, Health Physics, 49 (1) 11–24
MWH Americas, 2007. Infilitration and Contaminant Transport Modeling Report, White Mesa
Mill Site, Blanding, Utah, prepared for Denison Mines (USA) Corp., MWH Americas
Inc., Salt Lake City UT, November 2007
Neptune, 2011. Final Report for the Clive DU PA Model version 1.0, Neptune and Company
Inc., Los Alamos NM, June 2011
Neptune, 2014. Final Report for the Clive DU PA Model, Clive DU PA Model v1.2, NAC-
0024_R2, Neptune and Company, Inc., Los Alamos NM, August 2014
Neptune, 2015a. Safety Evaluation Report Response, NAC-0053_R0, prepared for
EnergySolutions, Neptune and Company Inc., Los Alamos NM, November 2015
Neptune, 2015b. Unsaturated Zone Modeling for the Clive PA, Clive DU PA Model v1.4, NAC-
0015_R4, prepared for EnergySolutions, Neptune and Company Inc., Los Alamos NM,
October 2015
Neptune, 2015c. Neptune Field Studies, December, 2014, Eolian Depositional History Clive
Disposal Site, NAC-0044_R0, Neptune and Company Inc., Los Alamos NM, March 2015
Neptune, 2015d. Saturated Zone Modeling for the Clive DU PA, Clive DU PA Model v1.4, NAC-
0016_R4, Neptune and Company Inc., Los Alamos NM, October 2015
Neptune, 2015e. Final Report for the Clive DU PA Model, Clive DU PA Model v1.4, NAC-
0024_R4, Neptune and Company Inc., Los Alamos NM, November 2015
Neptune, 2015f. Biologically Induced Transport Modeling for the Clive DU PA, Clive DU PA
Model v1.4, NAC-0022_R2, Neptune and Company Inc., Los Alamos NM, November
2015
Neptune, 2020a. Clive DU PA Model—Response to Model Version 1.4 Amended Interrogatories,
NAC-0147_R0, Neptune and Company Inc., Lakewood CO, April 2020
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 99
Neptune, 2020b. Deep Time Assessment for the Clive DU PA, Deep Time Assessment for the
Clive DU PA Model v1.5, NAC-0032_R5, Neptune and Company Inc., Los Alamos NM,
March 2020
Neptune, 2021. Clive DU PA Model—Response to DWMRC 1-28-2021 Comments, NAC-
0166_R0, Neptune and Company Inc., Lakewood CO, March 2021
Neuman, S.P., et al., 2003. A Comprehensive Strategy of Hydrogeologic Modeling and
Uncertainty Analysis for Nuclear Facilities and Sites, NUREG/CR-6805, United States
Nuclear Regulatory Commission, Washington DC, July 2003
NRC, 2000. A Performance Assessment Methodology for Low-Level Radioactive Waste Disposal
Facilities: Recommendations of NRC's Performance Assessment Working Group,
NUREG-1573, United States Nuclear Regulatory Commission, Washington DC, June
2000
Rödenbeck, C., et al., 2001. Dynamical Systems with Time Scale Separation: Averaging,
Stochastic Modelling, and Central Limit Theorems, Progress in Probability 49 (2001)
189–190
SC&A, 2015. Utah Division of Radiation Control, EnergySolutions Clive LLRW Disposal
Facility, License No: UT2300249; RML #UT 2300249, Condition 35 Compliance Report;
Appendix A: Final Report for the Clive DU PA Model, Safety Evaluation Report, Volume
1, prepared for Utah Department of Environmental Quality, SC&A Inc., Vienna VA,
April 2015
Šimůnek, J., et al., 2007. HYDRUS (2D/3D), Software Package for Simulating the Two- and
Three-Dimensional Movement of Water, Heat, and Multiple Solutes in Variably-
Saturated Media, User Manual Version 1.0, PC-Progress, Prague, Czech Republic,
January 2007
Smesrud, J.K., and J.S. Selker, 2001. Effect of Soil-Particle Size Contrast on Capillary Barrier
Performance, Journal of Geotechnical and Geoenvironmental Engineering 127 (10) 885–
888
Stantec, 2020. Phase 1 Basal-Depth Aquifer Study Report - Final, Clive Disposal Facility,
prepared for EnergySolutions LLC, Stantec Consulting Services Inc., Salt Lake City UT,
September 2020
Stephens, D.B., 1996. Vadose Zone Hydrology, CRC Press Inc., Boca Raton FL
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 100
SWCA, 2011. Field Sampling of Biotic Turbation of Soils at the Clive Site, Tooele County, Utah,
prepared for EnergySolutions, SWCA Environmental Consultants, Salt Lake City UT,
January 2011
SWCA, 2012. Vegetated Cover System for the EnergySolutions Clive Site: Literature Review,
Evaluation of Existing Data, and Field Studies Summary Report, prepared for
EnergySolutions, SWCA Environmental Consultants, Salt Lake City UT, August 2012
Taylor, S.A., and G.M. Ashcroft, 1972. Physical Edaphology, Freeman and Co., San Francisco
CA
Thompson, W.F., et al., 2015. Stochastic Averaging of Dynamical Systems with Multiple Time
Scales Forced with alpha-Stable Noise, U. of British Columbia, Vancouver, BC, Canada,
2015
Trinity Consultants, 2021. Summary Report of Meteorological Data Collected at
EnergySolutions’ Clive, Utah Facility, January 2020 through December 2020 and
January 1993 through December 2020, Trinity Consultants, Salt Lake City UT, February
2021
Utah DEQ, 2017. Division of Waste Management and Radiation Control, EnergySolutions Clive
LLRW Disposal Facility License No: UT2300249; RML #UT 2300249, Amended and
New Interrogatories Related to Clive DU PA Modeling Report Version 1.4 Dated
November 2015, Utah Department of Environmental Quality (DEQ), Salt Lake City UT,
May 2017
Utah DEQ, 2019. Depleted Uranium Performance Assessment (DU PA); Clive Facility; Model
Version 1.4 Amended Interrogatories; Radioactive Materials License #2300249, Utah
Department of Environmental Quality, Salt Lake City UT, July 2019
Utah DEQ, 2020. Comments on EnergySolutions Cover System Described in the DU PA, Draft
Federal Cell License Application, DRC-2020-DRC-019244, Utah Department of
Environmental Quality, Salt Lake City UT, December 2020
Utah DEQ, 2021. Technical Report, Performance Objective R313-25-23, Stability of the
Disposal Site after Closure, DRC-2021-001162, Utah Department of Environmental
Quality, Salt Lake City UT, January 2021
Vogel, T., et al., 1991. Porous Media with Linearly Variable Hydraulic Properties, Water
Resources Research 27 (10) 2735–2741
Wolf, M., et al., 2005. Mammal Parameter Specifications for the Area 5 and Area 3 RWMS
Models, Neptune and Company Inc., Los Alamos NM, September 2005
Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 101
Yang, C., et al., 2017. Analysis of Temporal Variation and Scaling of Hydrological Variables
Based on a Numerical Model of the Sagehen Creek Watershed, Stochastic Environmental
Research and Risk Assessment doi: 10.1007/s00477-017-1421-0
Zhao, Y., et al., 2016. Modeling of Coupled Water and Heat Transfer in Freezing and Thawing
Soils, Inner Mongolia, Water 8 (424) 1–18

Clive DU PA Model—Response to DWMRC 12-3-2020 Comments
31 March 2021 102

Attachment 1:
Federal Cell Drawings 14004-C01 through 14004-C05