ML21231A204

Technical Review Report Type I and II Tank SA PROJ0734 Enclosure

ML21231A204
Person / Time
Site:	PROJ0734
Issue date:	09/01/2021
From:	Cynthia Barr NRC/NMSS/DDUWP/LLWB
To:
Barr C
Shared Package
ML21231A202	List: ML21231A204
References
Download: ML21231A204 (76)

Text

Enclosure Technical Review of Documents Related to Type I and II Tanks Special Analysis at the H-Area Tank Farm at Savannah River Site Date: August 2021 Reviewers:

Cynthia Barr, U.S. Nuclear Regulatory Commission (NRC)

Cynthia Dinwiddie, Southwest Research Institute1 Stuart Stothoff, Center for Nuclear Waste Regulatory Analyses2 David Pickett, Center for Nuclear Waste Regulatory Analyses2 George Alexander, NRC Primary Document Reviewed:

SRR-CWDA-2016-00078. Type I and II Tanks Special Analysis for the Performance Assessment for the H-Tank Farm at the Savannah River Site. Revision 0. Aiken, South Carolina: Savannah River Remediation, LLC. August 2016. (ADAMS Accession No. ML19339E894)

Related H-Tank Farm Special Analysis Documents:

1.

SRNL-STI-2015-00339. M.E. Denham. Potential Solubility Controls for 129I in Residual Tank Waste. Revision 0. Aiken, South Carolina: Savannah River National Laboratory.

July 2015. (ADAMS Accession No. ML19318F721) 2.

SRNL-STI-2016-00224. G.A. Taylor and T. Hang. H-Area Tank Farm FY2016 Special Analysis Model Support. Revision 0. Aiken, South Carolina: Savannah River National Laboratory. May 2016. (ADAMS Accession No. ML20337A278) 3.

SRR-CWDA-2010-00023. H-Area Tank Farm Closure Inventory for use in Performance Assessment Modeling. Revision 6. Aiken, South Carolina: Savannah River Remediation, LLC. January 2016. (ADAMS Accession No. ML14027A069) 4.

SRR-CWDA-2014-00060. Updates to the H-Area Tank Farm Stochastic Fate and Transport Model. Revision 2. Aiken, South Carolina: Savannah River Remediation, LLC. July 2016. (ADAMS Accession No. ML21035A063) 5.

SRR-CWDA-2015-00158. H-Tank Farm Type I and Type II Tank Special Analysis Base Case Model Inputs (Interoffice Memorandum). Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC. 4 March 2016.

(ADAMS Accession No. ML21176A198) 6.

SRR-CWDA-2020-00018. Savannah River Site Liquid Waste Facilities Performance Assessment Maintenance Program - FY2020. Revision 1. Aiken, South Carolina:

Savannah River Remediation, LLC. August 2020. (ADAMS Accession No. ML20303A344) 1 The Southwest Research Institute is a not-for-profit research institute benefiting government, industry, and the public through innovative science and technology, located in San Antonio, Texas 78238.

2 The Center for Nuclear Waste Regulatory Analyses (CNWRA) is a federally funded research and development center, which was established in 1987 by the U.S. Nuclear Regulatory Commission. The CNWRA is part of the Chemistry and Chemical Engineering Division of Southwest Research Institute, San Antonio, Texas 78238.

1.0 Summary of Documents Reviewed Summaries of the primary documents listed above, which are related to Savannah River Site (SRS) HTF closure, are provided in Appendix A. Technical reviews of these documents, as well as related special analysis and inventory reports that NRC reviewed previously (ADAMS Accession Nos. ML13100A230, ML13273A299, ML15301A710, ML15301A830, and ML17277B235) are the basis for NRCs evaluation of DOEs HTF Type I and Type II Tanks Special Analysis, discussed herein.

2.0 Background and Summary of SA and Results The H-Tank Farm Type I and Type II Tank Special Analysis (hereafter, SA) updates the H-Area Farm (HTF) Performance Assessment (PA) model (SRR-CWDA-2010-00128, Revision 1). The SA also documents sensitivity analyses performed to provide additional information that can be used to inform decision-making. The sensitivity analyses provided information about the impact on peak dose of alternative tank configurations/fast-flow paths, variable waste volumes and inventories, iodine solubility variability, iodine and plutonium soil partitioning coefficient (Kd) variability, and grout/cement transition-time variability.

The Tank Types I and II SA focuses on two of four tank types at the HTF, although the cumulative dose associated with all HTF tanks is considered. These tank types do not have full secondary containments and are slated for early closure. At the HTF, submerged Tanks 9-12 are Type I tanks and partially submerged Tanks 13-16 are Type II tanks. Type I and Type II tanks generally received the same type of waste, and therefore are grouped together for purposes of estimating their inventories. Additionally, the tanks have leak sites and four of the tanks have a significant number of leak sites and are, therefore, assumed to have failed liners at the time of closure. Of these two tank types, only the Type II tanks were engineered to have 2.54-cm (1-in)-thick sand pads between the primary tank and the secondary liner, and between the secondary liner and the external tank vault. Tanks 12 and 16 have recently been closed; Tank 15 is the next scheduled for closure.

HTF PA modeling is supported by two sets of models. A set of deterministic PORFLOW models is used to demonstrate compliance with the performance objectives for low-level waste disposal, as documented in 10 CFR Part 61, Subpart C. Then, a set of stochastic GoldSim models is used to provide additional information to risk-inform the compliance decision. Details regarding the deterministic PORFLOW and stochastic GoldSim models are found in SRR-CWDA-2010-00128, Revision 1. The models are a set of coupled engineered barrier/natural barrier abstractions or simplifications. While the PORFLOW model is more complex (e.g., simulates multidimensional flow through the engineered barriers and natural system), both models are a simplified representation of the real system. However, because the PORFLOW models are more complex, outputs from the PORFLOW models are used as inputs to the GoldSim models.

The engineered barrier abstractions for both models involve representations of HTFs 29 waste tanks, 9 pump tanks, 3 evaporators, 2 concentrate transfer system (CTS) pump pits, and 4 areas of transfer line pipes. The unsaturated zone and SZ natural barriers are also represented in both sets of models. In the GoldSim models, if a tank is in the unsaturated zone, its unsaturated zone abstraction is run simultaneously with its engineered barrier abstraction; radionuclide releases from the unsaturated zone become the source term for each SZ abstraction. If any part of the engineered barrier is in the SZ, then the SZ abstraction is run simultaneously with the engineered barrier abstraction. A flow chart depicting the structure of the GoldSim model representing submerged Type I and partially submerged Type II tank system is shown in Figure 1. Figure 1 does not illustrate, however, the GoldSim models of AFZ pathways that are described later in more detail.

Figure 1 Flow chart depicting Type I and II Tanks as represented in GoldSim modeling (Image Credit: SRR-CWDA-2014-00060, Revision 2) 2.1 PORFLOW Groundwater Flow and Contaminant Transport Modeling Updates The HTF PORFLOW base case (i.e., Case A) conceptual model was not changed for this SA (SRR-CWDA-2016-00078), because an annular waste CZ with horizontal flow in the annulus had already been added to the material models of Type I and II tanks in the Tank 16 SA.

However, updated HTF Tank I and Tank II inventories and the updated I-129 cement Kd of 2 mL/g were revised inputs to the Case A PORFLOW modeling runs for this SA (SRR-CWDA-2016-00078). Case A modeling runs extend to 100,000 yrs post-closure. Additionally, to enable benchmarking between PORFLOW and GoldSim, a new set of PORFLOW output-generated flow-field files extending to 20,000 yrs post-closure was created for alternate tank configuration Cases B, C, D, and E, using the material model that includes the annular waste CZ with horizontal flow in the annulus (SRNL-STI-2016-00224). The HTF tank configuration cases are as follows (the simulation period is found in parentheses):

Case ANo fast-flow paths (i.e., base case tank configuration, 100k yrs)

Case BPartial fast-flow path; CZ transition based on grout/cement transition (20k yrs)

Case CPartial fast-flow path; more rapid CZ transition compared to Case B (20k yrs)

Case DFull fast-flow path; CZ transition based on grout/cement transition (20k yrs)

Case EFull fast-flow path; more rapid CZ transition compared to Case D (20k yrs)

Finally, a new set of PORFLOW output-generated flow field files, using six variable cementitious material hydraulic conductivities, was developed to enable variable cementitious material hydraulic conductivity sensitivity runs in GoldSim, both with and without early liner failure for the base-case scenario (Case A) and Type I and II tanks out to 20,000 yrs for doses at the 100-m facility boundary (SRR-CWDA-2014-00060).

PORFLOW vadose zone flow field information extracted for use as input to the GoldSim model included Darcy velocities, volumetric fluxes, flowrates, saturations, grout/cement transition times, and effective diffusion coefficient time series (SRNL-STI-2016-00224). Effective diffusion coefficient time series were generated for each material zone for Cases A-E and for each of the variable cementitious material hydraulic conductivity cases.

2.2 GoldSim Stochastic and Uncertainty Modeling Updates The Tank Types I and II SA documents updates made to the HTF GoldSim model (HTF_Transport_Model_v4.000_Rad; SRR-CWDA-2014-00060, Revision 2, Section 2).

Model updates include the following: (i) HTF waste tank radiological inventory values for Tanks 9, 10, 11, 13, 14, and 15 that incorporate waste tank cleaning inventory lessons learned from the final waste characterization results from Tanks 5F, 6F, and 12H; (ii) stochastic distributions used for inventories of Tanks 13, 14, and 15; (iii) replacing a 9 mL/g Reduced Region II cement iodine (I) Kd with 2 mL/g (based on information provided by Savannah River National Laboratory), which allows faster transport of I-129 through cement compared to 9 mL/g (SRR-CWDA-2015-00158, Revision 1), and replacing a log-normal probabilistic distribution for strontiums (Sr) saturated sandy soil partitioning coefficient with a triangular distribution with minimum, most likely, and maximum values of 4, 5, and 6 mL/g; (iv) addition of new points of assessment/theoretical well locations along the 1-m facility boundary to better replicate advective transport and inadvertent human intruder doses simulated by the HTF PORFLOW model; (v) the structure of the annulus CZ, mirroring changes in the PORFLOW model and the set of PORFLOW-generated effective diffusion coefficient input files; (vi) a new set of base case PORFLOW-generated flow field files for input to the GoldSim model that include horizontal flow in the annulus CZ, and a new set of PORFLOW-generated flow field files that also include horizontal flow in the annulus for alternate configurations Case B, Case C, Case D, and Case E to enable benchmarking between PORFLOW and GoldSim for these AFZ cases; (vii) addition of a set of model scenarios to evaluate the sensitivity of the modeled system to different quality grouts; and (viii) distributions for source-specific SZ Darcy velocities.

DOE noted in the SA that the only sorption coefficient that had been changed from a prior compilation of inputs was the iodine Kd for cement in RR_II, which was lowered from 9 to 2 mL/g (SRR-CWDA-2016-00078, Rev 0, p 54; SRR-CWDA-2015-00158). The specific basis for the new value was not provided. The NRC TRR that was cited as having been partially addressed by this change (ML15301A170) did not specifically discuss iodine Kd values in cementitious materials. This value was, however, included in an NRC TRR about the SDF PA (ML16342C575), which concluded that all DOE cementitious material iodine Kd values may be nonconservative. In addition, batch sorption studies conducted well after this SA was published showed no iodine sorption in concrete, nor in SRS soils at pH values >6 (SREL Doc. R 0004, Rev 0). Taken together, this information suggests that a technical basis is lacking for the cement iodine Kd value used in this SA.

2.3 Near-Field Transport Model Updates The HTF GoldSim model was updated to address annulus CZ structure for Cases B-E (SRR-CWDA-2014-00060), consistent with changes made to the HTF PORFLOW model for HTF Type I and Type II tanks to enable simulation of horizontal flow in the CZ. For Case A, the time-dependent effective diffusion coefficient for the annulus CZ is the same as for the rest of the annulus so that the GoldSim model continued to read PORFLOW time series of effective diffusion coefficient for the annulus. Because of the effort associated with generating new time-dependent effective diffusion coefficient files, the GoldSim model was updated to use 9.42 x 106 cm2/s [1.46 x 106 in2/s] for alternate configuration Cases C and E. In Cases C and E, the reducing capacity of the tank grout is not available to infiltrating water flowing through fast pathways, such that water chemistry is driven by the number of pore volumes that pass through the CZ, rather than the number of pore volumes of water that pass through the reducing grout; additionally, grout degradation is a slow process that follows a specified degradation curve (SRR-CWDA-2010-00128, Revision 1, Table 4.2-30). In contrast, for Cases B and D, the reducing capacity of the full volume of grout is available to chemically reduce infiltrating water, and grout degradation occurs instantaneously 501 yrs post-closure.

The HTF GoldSim model was also updated based on a new set of PORFLOW-generated flow-field files, to enable a set of six variable cementitious material hydraulic conductivity sensitivity runs, with and without early liner failure for the base case scenario (Case A) and Type I and II tanks out to 20,000 yrs, for doses at the 100-m facility boundary (SRR-CWDA-2014-00060).

The HTF GoldSim model was updated to add new points of assessment (i.e., theoretical well locations) along the 1-m HTF facility boundary to better replicate inadvertent human intruder doses simulated by the HTF PORFLOW model that was used to support the Types I and II Tanks SA. Previously, model values such as buffer distance, plume centerline distance, offset distance, and flow rate multiplier were modified for each waste tank until a relatively close match (or calibration) was achieved between GoldSims 1-m dose and PORFLOWs 1-m dose. For the recent HTF GoldSim model update, this qualitative approach was modified to allow only vertical mixing and plume divergence to be used as a calibration tool to account for differences between the PORFLOW and GoldSim models. The three-dimensional nature of the HTF PORFLOW model includes vertical components of flow (which influence mixing of plumes from different sources) and a flow divide that causes divergence of radionuclide releases from tanks in two directions. Plume divergence is especially prominent in Type II tanks (i.e., Tanks 13, 14, 15, and 16).

2.4 Far-field Transport Model Updates The HTF radionuclide fate and transport GoldSim model was also updated with respect to the stochastic distribution of SZ Darcy velocities used in sensitivity and uncertainty analyses.

The Darcy velocities are source-specific, based on PORFLOW transport simulations where the center of mass of a pulse released from a source is tracked until it crosses the 100-m facility boundary. The Darcy velocities are calculated from the time it takes for the center of mass to reach the 100-m facility boundary, the distance traveled, and the SZ porosity. The previous HTF GoldSim model used a uniform distribution with the minimum value set to 0.5x the source-specific Darcy velocity and the maximum value set to 1.5x the source-specific Darcy velocity. In the updated HTF GoldSim model, the SZ Darcy velocity distribution is a truncated normal distribution with the mean set to the tank-specific deterministic value, the standard deviation set to 0.25x the mean, and the minimum and maximum values set to 0.5x and 1.5x the source-specific Darcy velocities, respectively (SRR-CWDA-2014-00060).

2.5 Benchmarking GoldSim Modeling to PORFLOW Modeling The initial phase of benchmarking evaluated how well the GoldSim HTF radionuclide transport model for the base case flow configuration (Case A) approximated PORFLOW-modeled radionuclide mass releases from the tanks to the SZ. Variances of 50 percent for peak release rates during benchmarking were considered good, and differences between 50 and 100 percent were considered acceptable, as long as general trends were captured by the GoldSim model.

Occasionally, peak releases of Tc-99, Am-241, and Np-237 exceeded the 50 percent variance heuristic, yet were found acceptable (SRR-CWDA-2014-00060, Revision 2, Section 3).

The second phase of benchmarking of the GoldSim HTF radionuclide transport model evaluated how well it approximated PORFLOW-modeled radionuclide transport behavior in the SZ to the 100-m facility boundary, and member of the public (MOP) radionuclide doses (mrem/yr) from the major contributing radionuclide species, over 20,000 yrs. Radionuclide doses in five sectors (A, B, C, E, and F), were examined. In the GoldSim model, path lines from HTF tanks do not cross Sector D; therefore, this sector was omitted from benchmarking. Dose comparisons between the PORFLOW model and GoldSim abstractions were most consistent for sector A, due to its straight-line transport path between tanks and the facility boundary. The GoldSim model captured dilution and attenuation processes in the SZ for all sectors, and was shown to provide a computationally efficient approximation of 100-m facility boundary dose contributions.

A comparison of maximum dose-contribution curves from the major radionuclide species (i.e., Tc-99, Ra-226, I-129, and Np-237), irrespective of sector, demonstrated that the GoldSim model closely approximated the maximum dose to a MOP estimated by the PORFLOW model.

The third phase of benchmarking of the GoldSim HTF radionuclide transport model evaluated how well it approximated PORFLOW-modeled radionuclide transport behavior in the SZ to the 1-m facility boundary, and inadvertent human intruder dose estimates. Radionuclide doses in six sectors (including D), were examined. Solubility controls for Am-243, Cm-248, Th, and U were deactivated in GoldSim to simulate the PORFLOW model. The reported dose for each sector along the 1-m facility boundary is based upon the maximum concentration recorded at the simulated wells for each sector. Dose comparisons between the PORFLOW model and GoldSim abstractions were consistent for all sectors except D, because three-dimensional flow in Sector D modeled by PORFLOW was not well-represented by the one-dimensional GoldSim model. Nevertheless, DOE concluded that the GoldSim radionuclide transport approximation was satisfactory for Sector D. Dose curves resulting from PORFLOW and GoldSim for the major radionuclide species (i.e., Tc-99, Ra-226, I-129, and Np-237) were assessed.

Cm-248 and Am-243 were also important dose contributors in Sector D, which were modeled differently by the two simulators. Aquiclude layers, such as the Gordon Confining Unit, are modeled explicitly in PORFLOW, but not in GoldSim, leading to differences in the modeled breakthrough curves. However, the simplification to not model a storage zone for sorptive elements in GoldSim was not important to evaluating peak doses. The GoldSim abstraction model captured dilution and attenuation processes in the SZ for all sectors, and provided a computationally efficient approximation of 1-m facility boundary dose contributions.

The GoldSim benchmarking effort confirmed that the simulator could be used reliably to evaluate parameter sensitivity and uncertainty during stochastic studies in the effort to ascertain whether performance objectives will be met for HTF tank closure.

2.6 Dose Calculation Methodology The HTF radionuclide fate and transport stochastic GoldSim model contains a module for calculating receptor doses to the MOP or the inadvertent human intruder. Doses are evaluated at specified points of compliance, including sectors along the 100- and 1-m facility boundaries.

Concentrations used in the dose calculations may result from (i) the HTF GoldSim radionuclide transport abstraction module, (ii) the HTF PORFLOW model, and or (iii) be based on exposure to contaminated drill cuttings for the inadvertent human intruder. Dose calculations are abstracted from conceptualizations of possible exposure pathways. A complete description of the dose calculator and its Revision 1 update can be found in SRR-CWDA-2013-00058; NRC staffs assessment of the dose calculator was presented in ML16277A060. Revision 2 of this document was also reviewed recently (ML20254A003).

2.7 Results of the SA DOE indicates that the SA demonstrates that conclusions reached based on the HTF PA and subsequent SAs (Tanks 12H and 16H) remain valid. Further, DOE indicates that the SA process has confirmed that there continues to be reasonable assurance that the 10 CFR Part 61 Subpart C performance objectives will be met for HTF. DOE notes that consistent with DOE Manual 435.1-1, Chapter 4, a 1,000-yr period is used to assess compliance, while longer time periods help to risk-inform the decision.

The results from the H-Tank Farm Type I and Type II Tank SA show the projected peak annual all-pathways dose to a member of the public (MOP) within 10,000 yrs following HTF closure (Figure 2) using the updated inventories is approximately 0.07 mSv/yr (7 mrem/yr), occurring in Year 2,620. This peak dose within 10,000 yrs is primarily associated with Tank 12s I-129 residual inventory. Using the updated inventories, the peak annual all-pathways dose to a MOP within 20,000 yrs following HTF closure (Figure 3) is approximately 0.36 mSv/yr (36 mrem/yr),

occurring in Year 11,740. The peak dose within 20,000 yrs is primarily associated with the I-129 residual inventory of Tanks 9, 10, and 11. Using the updated inventories, the peak annual all-pathways dose to a MOP within 100,000 yrs following HTF closure (Figure 4) is approximately 1.63 mSv/yr (163 mrem/yr), occurring in approximately Year 61,000, with Ra-226 being the principal radionuclide contributor.

Figure 2. 100-m Member of the Public (MOP) groundwater pathway peak dose within 10,000 yrs, all HTF sources. Image credit: Figure ES-2 of SRR-CWDA-2016-00078.

Figure 3. 100-m Member of the Public (MOP) groundwater pathway peak dose within 20,000 yrs, all HTF sources. Image credit: Figure ES-3 of SRR-CWDA-2016-00078.

Figure 4. 100-m Member of the Public (MOP) groundwater pathway peak dose within 100,000 yrs, all HTF sources. Image credit: Figure ES-4 of SRR-CWDA-2016-00078.

3.0 NRC Staff Evaluation 3.1 Inventory Development 3.1.3 Primary and Secondary Liner Inventory DOE updated HTF Type I and II tank inventories based on information about waste volumes and concentrations and most notably informed by final characterization of other Type I and II tanks. The updates are focused on the primary liner inventories of Tanks 9-11 and 13-15 and the partial secondary liner inventories of Tanks 9, 10 and 14, which have more significant quantities of waste in their respective annuli. FTF Tanks 5 and 6 (Type I) and HTF Tanks 12 (Type I) and 16 (Type II) have undergone final waste characterization, and updated inventory estimates for the other Type I and II tanks at HTF were developed. Because Tank 16 waste retrieval methods are considered unique, residual Tank 16 waste in its primary is not expected to be representative of waste remaining in other tanks; thus, its inventory was not used to inform updated inventories.

The Type I and II Tanks SA assumes that key radionuclides in the primary liners and secondary liners (i.e., annuli) of Tanks 9, 10, 11, 13, 14, 15 will be removed to the maximum extent practical. A volume of 11,356 L (3,000 gal) of waste is assumed to remain in the primary tanks following waste retrieval activities. This volume is based on current waste volumes in Tanks 9 and 10 of 10,258 L (2,710 gal), remaining waste volumes in Tanks 5 [7,192 L (1900 gal)],

6 [11,356 L (3,000 gal)] and 12 [5,678 L (1,500 gal)] at the time of closure, which support 11,356 L (3,000 gal) as a conservative estimate.

The annulus inventories for Type I and II tanks are based on the waste concentrations reported in SRR-CWDA-2010-00023, Revision 4. Although Tanks 9, 10, and 14 currently have a few to 30 cm (12 in) of waste in their annuli, a trace volume of 379 L (100 gal) of waste is assumed to remain in the annuli of HTF Type I and II tanks yet to be cleaned following waste retrieval operations for the purposes of the SA3. The volume of waste is the same as what is assumed for Tanks 11, 13, and 15, which only ever had a trace quantity of waste in the secondary containment. Table 4.2-3 in SRR-CWDA-2016-00078, Revision. 0 provides annulus inventories for HTF radionuclides of concern. It will be important for DOE to sample the annuli of Tanks 9, 10, and 14, and to provide inventory estimates based on actual laboratory measurements and volume estimates following waste retrieval activities to verify that waste retrieval will be as successful as assumed and that the concentrations of key radionuclides (based in part on Tank 16 concentrations) are representative. Given the location of waste outside of primary containment, the risk associated with annular waste is expected to be higher than the risk associated with stabilized waste inside primary contaminant for a given key radionuclide activity in the waste.

Waste concentrations for Tanks 9-11 and 13-15 are based on Tanks 5, 6 and 12 waste concentrations for the assigned year-of-closure of HTF, which is 2032. HTF Type I and II tanks received bulk sludge waste from direct H-Canyon transfers of H-Modified (HM) and Plutonium Uranium Extraction (PUREX) waste transfers. Similarly, Tanks 5, 6, and 12 received direct transfers of HM and PUREX wastes from the F and H Canyons. Furthermore, the planned cleaning technologies for HTF Type I and II tanks are expected to be similar to Tanks 5, 6 and

12. The maximum radionuclide concentration calculated from four inventory options: (i) HTF PA (SRR-CWDA-2010-00023, Revision 4), (ii) Tank 5 inventory determination (SRR-CWDA-2012-00027, Revision 1), (iii) Tank 6 inventory determination (SRR-CWDA-2012-00075, Revision 1), and (iv) Tank 12 inventory determination (SRR-CWDA-2015-00075, Revision 1) was used to update the inventories for the remaining HTF Type I and II tanks yet to be cleaned and closed. Some further adjustments were made to the inventory assignments, as noted in the Comments Section of Table 4.2-1 of SRR-CWDA-2016-00078. Most notably, the inventories for Pu-238 and Tc-99 are assumed to be slightly lower than half the average of the HTF PA value and the Tank 12 and Tank 6 values, respectively.

SRR-CWDA-2010-00023 is periodically updated based on new characterization and other data collected at the HTF. The document also provides information on the screening process used to select a targeted list of radionuclides of concern for laboratory analysis. Revision 4 provides updated inventory data based on the final inventory for Tank 16 and new information on Tank 12. Additionally, the sand pad and annulus inventories for Type I and II tanks were updated based on characterization of Tank 16. The inventory adjustments made in Revision 4 were used in the Tank 16 SA. Revision 5 updates the inventory based on the final inventory for Tank 15. The inventory adjustments made in Revision 5 were used to support the Type I and II Tank SA. Table 1 has information about the more recent updates to SRR-CWDA-2010-00023.

3 Waste is also assumed to be present in the sand pads of Type II Tanks 13-16. Tanks 13 and 15 had only a small amount of waste in the annulus, therefore, 378.5 L [100 gal] of waste was assumed to be in the sand pads of these tanks. Tanks 14 and 16 had a more significant amount of waste in the annulus.

The assumed volume was based on the total void space of 4,921 L [1300 gal]. Additionally, Tank 16 leakage extended into the secondary sand pad located between the primary and secondary liner.

The total estimated amount of waste overtopping the secondary liner {98.4 L [26 gal]} was assumed to be in the secondary sand pad (SRR-CWDA-2010-00023, Revision 6).

SRR-CWDA-2010-00023 discusses the waste characterization system (WCS), which is an electronic information system that tracks waste tank data for 40 radionuclides.

Revision Supports Updates Rev. 4 Tank 16 SA Updated inventories for Tank 16 and Tank 12 Updated Type I and II tank annulus and sand pad inventories based on Tank 16 Editorial change in Pu-238 inventory in Tank 35 Rev. 5 Type I and II Tank SA Updated Tank 12 (final)

Updates to Tanks 9-11 and Tanks 13-15 Updated annulus inventory for Tanks 9, 10, and 14 Updates intruder inventory Rev. 6 Future SAs or PAs Updates Tanks 9-11 and 13-15 for five constituents (K-40, Co, Cu, SO4, and Sr)

Based on SRR-CWDA-2015-00166 Table 1 Summary of Inventory Revisions for HTF PA (SRR-CWDA-2010-00023)

WCS concentrations are based on process history, sample analysis, composition studies, and theoretical relationships. The radionuclides selected were based on their importance to safety (e.g., tank flammability and corrosion control), potential inhalation hazard, or waste acceptance criteria for E-Area vaults. WCS concentrations are expected to be conservative because they assume that reactor spent fuel processed in the canyons was at the maximum burn-up, while fission product concentrations may be present at significantly lower values and are based on dry sludge concentrations prior to chemical cleaning (e.g., chemical cleaning could preferentially removal certain radionuclides resulting in lower concentrations). Of the radionuclides of concern listed in Table 2, several are not included in the WCS, and inventory estimates for these radionuclides were developed based on supplemental data. These radionuclides are:

Ba-137m, Cl-36, H-3, K-40, Pd-107, Pt-193, Ra-226, Ra-228, Th-229, Th-230, Th-232, U-232, U-234, U-236, Y-90, and Zr-93. Table 2.2-1 in SRR-CWDA-2010-00023, Revision 6, provides information about the initial concentration estimate method for the radionuclides that were not included in the WCS. Of note, Ra-226 and Th-230 inventories were developed under the assumption that a minimal concentration is expected and therefore the minimum HTF tank concentration was used. No other information was provided to support the assumption regarding the representativeness of the minimum concentration for tanks yet to be cleaned.

The zirconium (Zr)-93 concentration is based on the ratio of Zr-93 to Sr-90 in sludge batches.

The Zr-93 concentration is set equal to Sr-90 sludge concentration divided by 58,000. It is significant to note that an unexpected, risk significant inventory of Zr-93 was determined following Tank 5 and 6 characterization activities. The final estimated Zr-93 inventory was 10,000x higher than originally assumed in the FTF PA. The higher than expected Zr-93 inventory led to an unexpected, risk significant peak dose from Nb-93m, daughter product of Zr-93 (ADAMS Accession No. ML13273A299). Due to the increased risk significance of Nb-93m, DOE made adjustments to model parameters to more realistically simulate Nb-93 m transport.

Nonetheless, given the high uncertainty in inventory estimates for certain radionuclides that are not well characterized in TFF waste, NRC staff will continue to monitor the final inventories for TFF tanks, and particularly for radionuclides such as Zr-93 that are supported by limited characterization data.

Based on the final volume estimates for cleaned Type I and Type II tanks (Table 3; primarily Tanks 5, 6 and 12), a projected residual waste volume of 11,356 L (3,000 gal) was assumed for other HTF Type I and Type II tanks yet to be cleaned. Updated concentrations are also based on Type I and Type II tanks that have been cleaned (see SRR-CWDA-2015-00166 for more detail). While higher annular inventories were initially assigned to HTF tanks with significant annular contamination, adjustments were made and all annuli were assumed to have trace quantities in the amount of 379 L (100 gal) for tank annuli that have yet to be cleaned based on the assumption that the annuli waste could be easily removed.4 Ac-227 Cl-36 Eu-152 Pa-231 Ra-226 Th-232 Al-26 Cm-243 Eu-154 Pd-107 Ra-228 U-232 Am-241 Cm-244 H-3 Pt-193 Se-79 U-233 Am-242 m Cm-245 I-129 Pu-238 Sm-151 U-234 Am-243 Cm-247 K-40 Pu-239 Sn-126 U-235 Ba-137 m Cm-248 Nb-94 Pu-240 Sr-90 U-236 C-14 Co-60 Ni-59 Pu-241 Tc-99 U-238 Cf-249 Cs-135 Ni-63 Pu-242 Th-229 Y-90 Cf-251 Cs-137 Np-237 Pu-244 Th-230 Zr-93 Table 2 Radionuclides of Concern at HTF (see Table 2.1-1 of SRR-CWDA-2010-00023, Revision 6) 3.1.2 Ancillary Equipment SRR-CWDA-2010-00023, Revision 6 also contains information about ancillary equipment (i.e., transfer lines, pump tanks, CTS tanks, and evaporator vessels) inventory estimates.

Inventories were not developed for other ancillary equipment, such as secondary containments, pump pits, catch tanks, diversion boxes and valve boxes, and therefore such equipment does not contribute to releases to groundwater in the HTF PA. Waste transfers within HTF and between FTF and HTF were used to determine the volume percent of all waste transfers attributable to each HTF waste tank. The concentration was determined by applying a weighted average of the tank radiological distributions. The slurry concentrations are constrained by technical safety requirements, which require less than or equal to 16.7 weight percent solids in waste transfers. The slurry concentrations were reduced to 20 percent of the dry sludge concentrations to account for this constraint. It is also significant to note that the dry sludge concentrations may overestimate actinide concentrations because dry sludge typically contains higher concentrations of actinides compared to solutions. While short-lived radionuclides may be concentrated in the supernate, DOE argues that short-lived radionuclides are expected to decay significantly and not drive long-term risk associated with the TFFs.

Transfer line inventory is calculated from data for waste concentrations and transfers.

The waste transfer lines are assumed to be flushed three times and transfer line concentrations are reduced accordingly using a continuously stirred tank reactor mass balance calculation.

The volumetric transfer line concentrations are then converted to surface area concentrations based on the pipe diameter (SRR-CWDA-2010-00023, Revision 6, Section 4.2). The gravity line drains are prone to plugging with salt waste; however, the lines are expected to be easily cleaned and thus, the gravity drains lines are treated similarly to the transfer lines with respect to inventory calculations. The total surface area of the transfer lines is approximately 6,884 m2 (74,100 ft2). The calculated surface area concentrations are multiplied by the total surface area to estimate the transfer line inventory, which was presented in Table 4.2-3 of SRR-CWDA-4 It is thought that the Tank 16 annular inventory is higher due to sand blasting activities that caused the radioactivity in the annulus to be less soluble.

2010-00023, Revision 6. It is important to note that diffusion of radioactivity into the metal piping and waste residue associated with the oxide film were previously calculated and added to the inventory associated with the waste transfers. However, only the waste transfer calculated inventory contributes significantly (99 percent or greater) to the total transfer line inventory and thus, only this inventory is considered in the PA.

Tank Residual Volume (gal*)

5 1,900 6

3,000 12 1,500 16 220 18 3,900 19 2,000

• 1 gal = 3.79 L 400 gal of waste is associated with the cooling coils in Tank 12, for a total of 1,900 gal of waste (SRR-CWDA-2010-00023, Rev. 6)

Table 3 Final Estimates of Residual Volumes in Cleaned Tanks With respect to pump tanks HPT-2 to HPT-10 (HPT-1 was removed from service), DOE assumes that the residual waste remaining in the tank will be similar to the waste tank residual depth of 1.6 mm (0.0625 in). The diameter of the base of the pump tanks is 3.7 m (12 ft).

The residue for each pump tank was therefore estimated at 16.7 L (4.4 gal) each. The basis for this value is not clear, because many tanks have waste heights greater than 1.6 mm (0.0625 in) and waste retrieval from pump tanks may differ from waste retrieval from HLW tanks due to differences in size, geometry, access, and waste retrieval methods. Some of these factors may lead to greater success in cleaning pump pits and other factors may be detrimental to waste retrieval. The HTF also has two CTS tanks that are comparable in capacity to pump tanks.

CTS tanks are assigned the same inventory as the pump tanks.

Sample data from an FTF evaporator (FTF 242-F) was used to estimate the inventory for three evaporators located at HTF. Samples taken from overheads during the residual waste removal campaign showed low activity and volume, and therefore were not included in the inventory.

Measured radionuclide concentrations were provided in Table 4.4-1 of SRR-CWDA-2010-00023, Revision 6; however, only 18 radionuclides are listed. It is unclear how the rest of the radionuclide inventories and concentrations are calculated. Comparison of the evaporator sample results to sludge slurry sample (1/5 of dry sludge sample concentrations) results reveals significantly lower concentrations of Se-79 and Sr-90 in the evaporator samples listed in Table 4.1-1 of SRR-CWDA-2010-00023, Revision 6 (reproduced here in Table 4). The total volume of waste assumed to remain in the evaporators based on the 1.5875-mm (0.0625-in) depth of waste is 1.14 L (0.3 gal) of sludge with a dry sludge density of 815 kg/m3 (6.8 lb/gal). The inventories of the 18 radionuclides listed in Table 3 are all <1 Ci (<3.7 x 1010 Bq).

DOE assumes that pump pits, catch tanks, diversion boxes, and valve boxes contain no significant contamination and therefore, no inventory is estimated for these components for the purposes of the HTF PA (SRR-CWDA-2010-00023, Rev. 6)5. Pump pits are shielded, 5 Although no inventory was developed for diversion boxes for the purposes of the HTF PA, SRR-CWDA-2020-00011 indicates that an inventory was developed to support closure of F-Area diversion boxes (FDB) 5 and 6. NRC staff may review the FDB-5 and FDB-6 inventory report (SRR-CWDA-2020-00029) and special analysis (SRR-CWDA-2020-00055) in a separate technical review report.

reinforced concrete structures lined with stainless steel that are located at transfer line low points. Pump pits house pump tanks and are accessible for cleaning at the time of closure.

Catch tanks collect drainage from Type I tank transfer line encasements, as well as in-leakage from groundwater. Diversion boxes are shielded, reinforced concrete structures that contain transfer line nozzles to which jumpers are connected to direct waste transfers. Most diversion boxes are below ground and lined with stainless steel or sealed with water-proofing compounds; additionally, they are accessible for cleaning at the time of closure. Valve boxes provide secondary containment and facilitate waste transfers that are conducted frequently. The valves are generally manual ball valves in removable jumpers and have flush water connections on the transfer lines. Valve boxes are also accessible for cleaning. NRC staff will verify that no significant radioactivity remains in these ancillary components at the time of closure.

Radionuclide Evaporator Sample FTF 287 (µCi/g)

Am-241 4.4 Co-60 1.33 Cs-137 1.02E+03 H-3

<1.62E02 Np-237 3.89E03 Pu-238 5.74 Pu-239 1.5E+01 Pu-240 3.32 Pu-241 4.73E+01 Pu-242

<4.8E03 Se-79 8.27E06 Sr-90 5.95E+01 Tc-99 1.37 U-233

<1.2E02 U-234

<7.6E03 U-235 8.72E05 U-236 1.47E04 U-238 8.05E03 Table 4 Measured Radionuclide Concentrations in the 242-F Evaporator Sample Although not specific to HTF, DOE has performed some limited characterization of diversion boxes FDB-5 and FDB-6 at FTF (Figure 5) and concluded that because transfer lines were typically flushed several times with clean water after each transfer, only minimal waste might be present inside the jumpers. Camera inspections show only a small deposit of salt waste in FDB-6 that is thought to be associated with leakage due to an inadequate seal between a wall nozzle and jumper connection. Otherwise, there are no significant accumulated solids on the floors of FDB-5 and FDB-6. DOE SRS estimates that the volume of waste in FDB-6 is 1.1 L (0.3 gal) with an uncertainty range of 0.4 to 2.3 L (0.1 to 0.6 gal).

Although no inventory estimates were developed for diversion boxes, such as FDB-5 and FDB-6 for the PA, DOE developed an inventory for these diversion boxes for the purpose of an SA.

The approach used was similar to that used to develop the inventory for the HTF transfer lines, as described in preceding paragraphs. The inventories associated with residual material: (i) in the jumpers, (ii) on the floor of the diversion boxes and sump, and (iii) in the sump drain piping were thought to bound the inventory associated with diffusion into the metal or material associated with an oxide film coating. For FDB-6, the 1.1 L (0.3 gal) salt waste volume was expected to be bounded by the inventory assumed to be associated with the waste residue on the diversion box floor. Furthermore, the same average representative FTF radionuclide concentration based on tank concentration and waste transfer data from WCS was also calculated (SRR-CWDA-2010-00023, Rev. 6) to determine the FDB-5 and FDB-6 waste inventories (SRR-CWDA-2020-00011). The conservatism of the approach was supported by higher concentrations of 1.63E+10 Bq/m2 (0.44 Ci/m2) for Sr-90 (the primary beta-gamma source) and 9.47E+07 Bq/m2 (2.56E03 Ci/m2) for Pu-238 (the primary alpha source) than would be predicted by the smear sample results of 1.99E+08 Bq/m2 (5.38E03 Ci/m2)

FDB-5 FDB-6 Figure 5 Plan View Maps of the FDB-5 (Top) and FDB-6 (Bottom). Image Credit:

Figures 2.1-1 and 2.5-3 in SRR-CWDA-2020-00011.

beta-gamma and 9,950 Bq/m2 (2.69E07 Ci/m2) alpha. DOE comparisons against relatively short-lived, high activity radionuclide concentrations is helpful; however, NRC staff note that while these radionuclides may dominate the activity at early times, these short-lived radionuclides may not be driving risk over longer performance periods if engineered barriers are effective in mitigating releases for a sufficient period of time to allow these radionuclides to decay to insignificant levels. Furthermore, differences in mobility and dose conversion factors may also cause longer-lived, lower activity key radionuclides to dominate risk. Stronger support for the assumed conservatism of inventory estimates for ancillary equipment or PA estimates is needed.

NRC staff note that the assumed low risk of transfer lines (and other ancillary equipment) is based on DOEs inventory estimates, which have not yet been validated with field characterization data. As stated in staffs FTF and HTF Monitoring Plan for the TFFs (Monitoring Factor 1.4, Ancillary Equipment Inventory), DOE should validate the theoretical inventory calculations through field characterization and sampling activities to confirm the assumed low risk. Additionally, the transfer line inventory (and other ancillary equipment) is more easily accessible than HLW tanks, and therefore, could be more risk significant to the inadvertent human intruder. DOE should also clarify the state of knowledge regarding the location and status of any abandoned transfer lines and any efforts to clean or characterize waste within these lines. NRC staff will monitor DOEs efforts to validate the transfer line and other potentially risk significant ancillary components under Monitoring Factor 1.4, Ancillary Equipment Inventory.

A separate inadvertent intruder inventory was developed with Tank 13 representing the Type I/II grouting, Tank 24 representing Type IV tanks, and with a 7.6-cm (3-in) and 10.2-cm (4-in) transfer line diameter. The residual material is assumed to be spread over the bottom of the tank floor. A well diameter of 20.3 cm (8 in) is assumed. The inventory is based on the ratio of the 20.3 cm (8 in)-diameter well to the tank area (e.g., Type II tank diameters are 26 m (85 ft)).

For transfer lines, the curved surface area of the cylinder removed from the transfer line during well drilling is 2rh where r is the well radius [10.2 cm (4 in)] and h is the transfer line diameter.

To study the impact of inventory uncertainty on the results, DOE performed a probabilistic analysis using GoldSim. A generic log uniform distribution with multipliers between 0.01 and 10 were applied to the inventories of radionuclides. A comparison between PA inventory values and those determined with actual characterization data showed that 75 percent of the values overestimated or closely matched the measured inventory values. In other words, less than 25 percent of the estimated PA inventory values (prior to tank closure) were lower than the measured inventory values. Notable underestimates include Zr-93 in Tanks 5 and 6, and K-40 in Tank 6; these species were underestimated by more than a factor of 10. Most recently, the maximum underestimate for Tank 12 was less than a factor of 10. Because of uncertainty in the parameter distributions for ancillary equipment and sand pad initial inventory, uncertainty in these components was not considered.

It is precisely because there are no data to develop parameter distributions for ancillary equipment and sand pad inventories that the impact of inventory uncertainty for these components should have been evaluated. NRC staff expects DOE to collect data about ancillary equipment inventories to confirm the expected low risk. DOE should also comment on historical plugging of the transfer lines and jumpers and records of any abandonment of ancillary components (e.g., historical plugging of jumpers in FDB-5 occurred and jumpers are known to have been abandoned within FDB-5). NRC staff also thinks DOE should consider indirect methods for estimating inventories of sand pads if they cannot sample the sand pads directly (e.g., analysis of liquid samples in contact with the sand pad to estimate the inventory associated with the aqueous phase and inventory associated with the solid phase [with the assumed distribution coefficient of the sand pad]).

Previous findings with respect to inventory development that are still pertinent today include the following:

1.

DOE should consider various forms of uncertainty and explicitly consider components of uncertainty in its sample design and analysis (ML13085A291).

2.

DOE should improve estimation methods for waste volume uncertainty (ML13085A291);

there is limited validation of the methods previously used to estimate waste areas and heights for calculation of residual waste volumes (e.g., estimates of waste areas and heights are based on video and photographic evidence, use of landmarks and professional judgment). Estimation of waste heights in tanks with significant residual waste in tank annuli also presents its own challenges that should be considered in obtaining more accurate estimates of waste volumes and associated uncertainty (ML15301A830).

3.

DOE should explicitly consider uncertainty in sample density, in addition to volume uncertainty. Currently, DOE determines mass fractions of increments comprising the composite samples based upon a single, deterministic density and considering uncertainty in volume measurements using a volumetric proportional compositing method, which samples just a few values from a triangular distribution (ML17277B235). A more optimal sampling design might be to composite samples within segments (e.g., segments include cooling coil samples, dispersed floor samples, and accumulated waste solids/mounds in less accessible portions of the tank) to preserve information on composite sample variability, which appears to dominate the uncertainty in inventory estimates. Volume and density uncertainty of segments could be considered in a probabilistic analysis calculating total inventory based on radionuclide concentrations determined by compositing sampling within segments.

4.

DOE should consider calculating the 95th percent upper confidence level of the mean based on the average and standard deviation of three composite samples analyzed in triplicate rather than treating all nine samples as independent based on the results of the F-test.

The F-test determines if random error needs to be considered (i.e., if there is a statistically significant variance between the three composite samples, then the nine samples are not treated as independent). (ML13085A291) 5.

DOE should continue to update inventory and associated uncertainty estimates based on characterization of final tank/annulus activities to ensure that risk to members of the public is not underestimated. (ML13085A291) 6.

DOE should consider methods to ensure representativeness of samples. For example, unwashed samples behind the cooling coil fences are almost always underrepresented in preliminary estimates of waste volumes with fewer samples collected. Vertical and horizontal heterogeneity of mound samples and under-sampling of mound samples could lead to underestimates of radionuclide concentration in higher concentration wastes associated with relatively unwashed mounds (ML17277B235). Issues associated with sampling annulus waste are discussed in more detail in the Tank 16H inventory TRR (ML15301A830).

NRC staff will continue to monitor DOEs radionuclide inventory development for the purpose of PA calculations under Monitoring Factors 1.1, Final Inventory and Risk Estimates, 1.2, Residual Waste Sampling, 1.3, Residual Waste Volume, and 1.4 Ancillary Equipment Inventory listed in NRC staffs plan for monitoring at the TFFs (ML15238A761).

3.2 Sensitivity Analyses The SA includes a series of sensitivity analyses that provide additional information to inform decisions regarding HTF Type I and Type II tank closure operations. These sensitivity analyses can be used individually and in various combinations to better understand the impact of different tank conditions and situations that might be encountered during waste tank closure.

Chapter 5 of the SA presents information on the base case tank configuration (i.e., Case A).

Results are provided for doses at the following locations (Figure 6):

100-m member of the public dose

1-m inadvertent human intruder dose

Intruder wells.

Figure 6 Hypothetical intruder well locations where dose results are computed at the 1-m (3-ft) boundary (inner boundary grouped in model evaluation sectors A-F) and inside the 1-m (3-ft) boundary (yellow squares). 100-m (300-ft) facility boundary sectors (outer boundary) and PORFLOW particle tracking pathlines from the center of each tank also illustrated. Image Credit: Figure 4-25 of SRR-CWDA-2010-00128, Revision 1.

The projected peak dose within 1,000 yrs is 0.002 mSv/yr (0.2 mrem/yr) from Tc-99 in Sectors A, B, and C (see Figure 6 for locations of sectors). The peak dose within 10,000 yrs is 0.07 mSv/yr (7 mrem/yr) from I-129 in Sector A (see Figures 2 and 6). The peak dose within 100,000 yrs is 1.63 mSv/yr (163 mrem/yr) from Ra-226 in Sector B (see Figures 4 and 6). The primary pathway of exposure is drinking water ingestion. It is noteworthy to emphasize that the 100,000-yr peak dose is significantly higher than the 1,000- and 10,000-yr peak doses due to a correction in the initial Pu-238 inventory assigned to Tank 35, as noted in the Tank 16 SA quality assurance report (SRR-CWDA-2014-00134). Additionally, the increased I-129 inventory assumed for Type I and II tanks leads to a higher dose in the 10,000-yr period, while a decreased inventory of Tc-99 associated with annuli of Type I and II tanks leads to a reduction in the Tc-99 dose. Not captured in these results is a peak of approximately 0.3 mSv/yr (30 mrem/yr) from I-129 after 10,000 yrs.

The peak dose at the 1-m facility boundary within 1,000 yrs is from Sr-90 in all sectors at a value of 0.084 mSv/yr (8.4 mrem/yr), but this dose is associated with the plant ingestion pathway and drill cuttings at the end of the institutional control period (i.e., 100 yrs). The peak dose within 10,000 yrs is from Ra-226, Sr-90 or I-129, with the highest in Sector D for Ra-226 at a value of 0.34 mSv/yr (34 mrem/yr) and with significant contributions from U-234. The highest dose within 100,000 yrs is from Ra-226 in Sector F with a value of 4.76 mSv/yr (476 mrem/yr).

The peak doses within 10,000 and 100,000 yrs are associated with the drinking water pathway.

A short list of radionuclides that were targeted for additional simulation includes: Am-241, Am-243, I-129, Ni-59, Np-237, Pa-231, Pb-210, Pu-238, Pu-239, Ra-226, Tc-99, Th-230, U-234, U-235, and Zr-93.

Sensitivity analyses described in Chapter 6 are for Cases B through E in Sections 6.1 through 6.4 of the SA, respectively. Results show that the performance objectives could be exceeded over longer performance periods or for alternative tank configurations. Peak doses from Case E occur earlier in time and with much higher magnitude compared to other cases. Case E represents both chemical transitions and steel liner failures that occur much earlier than in other cases. Table 5 provides the base case scenarios (i.e., Case A) chemical transition times.

Table 6 provides information about the alternative cases B through E. Table 7 provides a summary of the results of the sensitivity cases.

GoldSim was used to conduct sensitivity analyses. Results of the following sensitivity analyses are presented in Sections 6.2 through 6.6 of the SA, respectively:

Section 6.2: Waste tank inventory variability

Section 6.3: Iodine solubility variability

Section 6.4: Soil Kd (i.e., plutonium and iodine soil Kd) variability

Section 6.5: Chemical transition-time variability

Section 6.6: Grout hydraulic performance variability 3.2.1 Sensitivity to Inventory With regard to sensitivity related to tank inventory, inventory multipliers were applied to Type I and Type II Tanks 9-11 and 13-15 only (Tanks 12 and 16 inventories are based upon final characterizations of waste in the closed tanks and are more certain). The inventory multipliers were 0.5, 1, 1.5, and 2. The results of the sensitivity analysis revealed the importance of I-129, Ra-226, and Ni-59 to peak dose. The dose from I-129, in particular, was the key dose contributor from Type I and II tanks with Type II tanks 14-15 doses occurring earlier and within 10,000 yrs. The timing of steel liner failure influences the timing of the release and magnitude of dose (e.g., whether releases occur at the same time and therefore lead to a higher cumulative risk). Because of the assumption of no solubility control, the inventory was linearly related to the dose for I-129.

Table 5 Pore volumes to chemical transitions and chemical transition times.

Reproduced from Table 3-2 in the Tank Farms Monitoring Plan (ML15238A761).

3.2.2 Sensitivity to Iodine and Plutonium Solubility Limits DOE studied the impact of uncertainty on solubility limits for I-129, Tc-99, Pu-239, and U.

Of particular interest is iodine solubility sensitivity, discussed next.

Sensitivity to Iodine Solubility Limits In the HTF PA and current waste release modeling, DOE conservatively assumed that no controls limit the instantaneous release of I-129 from residual waste once tank liners are breached. However, DOE indicates that the dose associated with I-129 could be mitigated if its solubility is controlled by its association with solid phase metals in the residual waste, whereby metal salts may form, such as AgI, AuI, CuI, or Hg2I2. Iodine solubility within residual tank waste should, therefore, be an area of focus during future PA maintenance research activities.

To evaluate sensitivity to the iodine solubility limit, DOE conducted analyses comparing I-129 doses for the base case scenario (i.e., Case A)in which I-129 release was not solubility limitedagainst (i) the case of using the solubilities presented in SRNL-STI-2015-00339 (mechanistic case) and (ii) three non-mechanistic cases that acknowledge there is uncertainty (e.g., because I-129 has not been measured in waste release experiments). For the mechanistic cases, iodine solubility values 1.4E04 mol/L (M) were applied in all conditions Table 6 Evaluation case (A) and alternative case descriptions. Reproduced from Table 4-11 in NRC staffs Technical Evaluation Report for H-Area Tank Farm Facility (ML14094A514).

Case A mrem/yr (sector or intruder well)

Case B mrem/yr (sector or intruder well)

Case C mrem/yr (sector or intruder well)

Case D mrem/yr (sector or intruder well)

Case E mrem/yr (sector or intruder well) 100 m (1,000 yr) 0.2 (A, B, C) 14.9 (A) 6.8 (C) 10.4 (A) 52 (C) 100 m (10,000 yr) 7 (A) 32 (A) 17.5 (B) 32.5 (A) 299 (B) 100 m (20,000 yr*)

163 (B) 49 (B) 37 (B) 49 (B) 299 (B) 1 m (1,000 yr) 8.4 (all) 31 (A) 19.1 (D) 31.4 (A) 685 (D) 1 m (10,000 yr) 34 (D) 154 (D) 105 (D) 154 (D) 1220 (B) 1 m (20,000 yr*)

476 (F) 253 (B) 216 (B) 253 (B) 1220 (B)

Intruder (1,000 yr) 4 (Tank 12) 246 (Tank 12) 49 (Tank 15) 177 (Tank 12) 1672 (Tank 15)

Intruder (10,000 yr) 79 (Tank 12) 424 (Tank 15) 282 (Tank 15) 411 (Tank 15) 3349 (Tank 13)

Intruder (20,000 yr) 408 (Tank 15) 666 (Tank 35) 459 (Tank 35) 664 (Tank 35) 3349 (Tank 13)

• The peak dose within 100,000 yrs (instead of 20,000 yrs) is reported.

Note: Values in red are greater than the performance objectives of 0.25 mSv/yr (25 mrem/yr) for members of the public and 5 mSv/yr (500 mrem/yr) for inadvertent human intruders.

Table 7 Summary of DOE Type I and II Tank SA evaluation case and sensitivity analysis results except Reduced Region II (RR_II) and Submerged Region C (SubC) (see Table 8, discussed below). In the non-mechanistic cases, selected limits were applied also in RR_II and or SubC.

The only sensitivity cases in which the peak 10,000- and 20,000-yr I-129 doses decreased were those non-mechanistic cases in which iodine solubility was controlled in SubC (Table 9).

A notable implication of these results is that, in the SA model, the peak I-129 dose at the 100-m facility boundary is controlled by one or more of the submerged tanks (page 191). It is also notable that, even with solubility limits as low as 9.8E09 M, the 100-m peak I-129 doses up to 20,000 yrs are lowered by, at most, 70 percent relative to the case with no solubility control.

The mechanistic cases were developed based on an iodine solubility limit scoping modeling study (SRNL-STI-2015-00339) that proposed I-129 release from tank residual wastes could be controlled by silver iodide (AgI) or mercury iodide (Hg2I2). The models, based on observations of elevated concentrations of these metals in residual waste, used Geochemists Workbench and the Japan Atomic Energy Agency thermodynamic database. The models implied solubility-limited I-129 release concentrations under oxidizing conditions of 9.8E9 to 2.7E7 M for AgI control and 1.2E7 to 1.2E4 M for Hg2I2 control (Table 8). For modeled reducing conditions, I-129 concentrations were not solubility limited because the metals were not stable as cations (corresponding to RR_II and SubC). Therefore, assumed solubility control of I-129 in the AgI or Hg2I2 solubility-limiting phases could lead to lower dose, but only if solubility control is assumed for RR_ II (at OR_ II modeled levels for non-submerged tanks or Condition D levels for submerged tanks). Otherwise, all of the I-129 is released before chemical transition to OR_II or OR_III. However, the reduction in I-129 dose, while significant, is only a factor of 2 reduction, at most, as illustrated in Table 9.

While these iodine solubility models add to the body of relevant information on I-129 release, they have not been supported by site-specific experimental evidence and should, therefore, not be relied upon solely for PAs.

The solubility sensitivity analyses (SRR-CWDA-2016-00078) and scoping solubility model study (SRNL-STI-2015-00339) were also discussed in an earlier TRR (ML19298A092) that was chiefly concerned with (i) DOE-sponsored radionuclide release testing on actual residual tank waste (SRNL-STI-2018-00484, Revision 1) and (ii) a report relating these results to tank farm PA results (SRR-CWDA-2016-00086, Revision 1). In the aforementioned TRR, NRC staff noted DOEs observation that higher than expected I-129 concentrations in residual waste suggest that I-129 release must, to some extent, be controlled by solubility. NRC staff also noted that, in waste release leaching studies, I-129 was below the detection limit for all but two leachate subsamples. Two significant NRC staff comments in ML19298A092 were:

I-129 in the Tank 12H residual waste sample is relatively insoluble, under both reduced and oxidized conditions, compared to the PA waste release model that conservatively assumed no solubility control for I

Overall, the solubility testing results for RR_II, OR_II, and OR_III indicate iodine is much less soluble than assumed in the TF PAs Nevertheless, the aforementioned TRR did not conclude that the body of evidence was sufficient to support using the low modeled solubility limits for I-129 in PA modeling.

The conclusions on iodine solubility and recommendations for further waste release testing and geochemical modeling embodied in ML19298A092 have not been altered by staffs review of the Type I and II Tanks SA (SRR-CWDA-2016-00078). Because the dose associated with I-129 could be mitigated to some degree if its solubility were controlled by its association with solid phase metals in the residual waste, iodine solubility within residual tank waste should continue to be an area of focus during future PA maintenance research activities. I-129 solubility control would need to be supported by a systematic argument compiling all available data, including both model and experimental studies.

Tanks Above Water-Table Submerged Tanks Solid Phase Reduced Region II (RR_II)

Oxidized Region II (OR_II)

Oxidized Region III (OR_III)

A B

C D

AgI No Solubility Control 2.7E07 3.8E08 9.8E09 9.8E09 No Solubility Control 9.8E09 Hg2I2 No Solubility Control 2.0E05 1.2E04 1.2E07 1.8E07 No Solubility Control 1.8E07 Table 8 Calculated iodine solubilities (mol/L or M) for specific pore-water conditions (summarized from Table 6.3.1-2 of SRR-CWDA-2016-00078)

Table 9 Iodine solubility sensitivity analyses results (reproduced from Table 6.3.1-3 of SRR-CWDA-2016-00078)

These NRC staff positions are also consistent with more recent statements in a TRR on the topic of Tank 12H grouting operations (ML20296A550), which noted:

If DOE takes advantage of solubility control for I-129 for tank farm tanks in the future, a better understanding of the expected evolution of the geochemical conditions in the waste zone would be needed. Additional information to support the expected solubility of I-129 under the assumed geochemical conditions would also be needed, as discussed in more detail in ADAMS Accession No. ML19298A092 (e.g., the targeted Eh and pH in the Tank 12H waste release experiments were inconsistent with the reference case conditions assumed in DOEs H-Tank Farm PA).

Sensitivity to Plutonium Solubility Limits Results of sensitivity analyses for Pu-239, Tc-99, and U that were conducted for this SA revealed minor to negligible impacts of assumptions regarding solubility on the peak dose results. In particular, Section 6.3.2 briefly discussed analyses conducted of the groundwater pathway dose sensitivity to plutonium solubility. For these calculations, the plutonium solubility limit of 3E11 M used for RR_II, OR_II, and OR_III was increased two to three orders of magnitude for conservative and nominal cases that considered waste release testing.

The highest plutonium solubility limit evaluated (conservative) was 2E08 M in OR_III; the other values for all three regions were between 1E09 M and 1E08 M. Considering all radionuclides contributing to dose, there was essentially no impact of these higher plutonium solubilities on dose for all cases within 20,000 yrs, and there were relatively subdued impacts even at the peak doses at 75,000 to 80,000 yrs. The SA did not explain the plutonium results.

NRC staff infer that the lack of sensitivity of Pu solubility to dose was due to the presence of a more effective downgradient barrier (e.g., the concrete basemat or natural system Kd) controlling the release or groundwater concentrations and dose, as discussed in ML15238A761.

The NRC staff has discussed plutonium solubilities in a number of TRRs, both before and after the Type I and II Tanks SA was published. Uncertainty in plutonium isotope release behavior was a key motivation for the NRC staff to define Monitoring Factor 2.1, Solubility-Limiting Phases/Limits and Validation (ML15238A761). Subsequently, in reviewing DOE Tank 18 residual-waste-leaching tests and implications for PA (ML18242A259), NRC staff noted that the tests indicated a significant fraction of waste plutonium was released at concentrations much higher than assumed in PA models. A pre-leach wash solution contained plutonium at 3E07 M, and some of the leach tests also yielded concentrations far in excess of the 3E11 M value used in FTF and HTF PA modelswith the exception of an earlier FTF model that assumed 6E05 M for OR_III conditions. In addition, NRC staff noted that steady-state plutonium concentrations were not always achieved during tests. Furthermore, staff noted that DOE models did not account for the possibility of multiple plutonium-dissolved species with widely varying solubility limitsincluding the possibility of a mobile early release fraction. NRC staff recommended that DOE apply spectroscopic methods, such as Extended X-Ray Absorption Fine Structure, to better understand plutonium speciation. Staff concluded that a greater range of uncertainty in plutonium solubility was needed to reflect a lack of support for the base case and for the bounding nature of the conservative values.

In a later review of Tank 12 residual-waste-leaching tests (ML19298A092) and a revision of the earlier PA implications report, NRC staff noted that the Tank 12 plutonium-release concentrations were considerably lower than found for Tank 18 wastes and were both lower and higher than modeled values. The contrast between the two tanks was cited as contributing to the uncertainty of plutonium-release modeling and NRC staff stressed the importance of actual waste testing, with attendant characterization, to supplement geochemical models to better understand the nature of plutonium aqueous release. Furthermore, NRC staff pointed out that the targeted Eh and pH in the Tank 12 experiments were inconsistent with the reference case conditions assumed in DOEs HTF PA. Given current uncertainties, plutonium-release models should be continuously updated in response to waste release testing and should be tested against alternative conceptual models.

Previous TRRs have emphasized not only uncertainty in released plutonium concentrations, but also the implications of variability in plutonium mobility in the subsurface and biosphere.

In summary, the waste release testing reports and associated PA analyses released since the Tanks I and II SA show that plutonium-release models are subject to considerable uncertainty, should be evaluated against residual-waste testing from different tanks, and should consider multiple mobility fractions. Sensitivity analysis results, such as reported in this SA should be evaluated in a manner that helps elucidate the risk significance of the relevant models and parameters. Specific NRC staff recommendations regarding plutonium-release-modeling for PA are best summarized in TRRs ML18242A259 and ML19298A092.

3.2.3 Sensitivity to Iodine and Plutonium Soil Partition Coefficient, Kd Sensitivity to Iodine Soil Partition Coefficient, Kd To evaluate sensitivity of dose results to I-129 soil partition or sorption coefficients and to cement leachate impact factors (LIFs), DOE conducted analyses that compared I-129 doses for the base case (i.e., Case A or evaluation case)in which sand and clay Kds were 1 and 3 mL/g and the LIF was 0.1to cases with different Kd or LIF values (Table 10). LIF is a factor by which the soil Kd is multiplied to account for the chemical effect of interaction of percolating waters with cementitious materials.

Table 10 Iodine soil partition coefficient (Kd) sensitivity analyses (reproduced from Table 6.4-1 of SRR-CWDA-2016-00078)

Using a LIF of 0, 1, or 10 with Case A Kds had no effect on the timing or magnitude of 10,000-and 20,000-yr peak I-129 doses (Table 11) and had only a limited effect on a secondary peak at approximately 12,900 yrs. The observation that peak doses in the case with LIF of 0 are equivalent to Case A with LIF = 0.1and that even cases with LIF of 10 show a subdued effectis unexpected to the NRC staff, but may suggest that cement leachate is not important in how transport is modeled. NRC staff will seek a better understanding of why the LIF has limited impact on dose modeling for these tanks.

Table 11 Iodine Soil Partition Coefficient (Kd) Variability Sensitivity Analyses Results (from Table 6.4-2 of SRR-CWDA-2016-00078)

Changing the soil Kd values, however, had significant impacts on peak I-129 doses (Table 11).

Using conservative values of 0.3 and 0.9 mL/g for sand and clay, which NRC staff had not previously challenged (ML15301A710; ML16342C575), increased the peak 10,000-yr dose by 34 percent and more than doubled the peak 20,000-yr dose (Table 11). The overly conservative case with no I-129 sorption resulted in peak doses 48 and 216 percent higher than Case A for 10,000 and 20,000 yrs, respectively (Table 11). In no case did the 10,000-yr peak exceed the 0.25 mSv/yr (25 mrem/yr) performance objective, but the 20,000-yr peak of 0.82 mSv/yr (82 mrem/yr) did (Table 11). Therefore, uncertainty in the timing of peak dose, which is related to uncertainty in engineered barrier performance, could be important to the compliance demonstration. These sensitivities point to the importance of continued attention being placed on iodine sorption during NRC monitoring.

NRC staff has discussed iodine sorption in a number of TRRs. In 2015 (ML15301A710), the NRC staff concluded that the increases in Case A iodine sand and clay Kds to 1 mL/g and 3 mL/g, respectively, were not adequately supported. DOE based the changes on (i) studies of iodine groundwater speciation in F-Area Seepage Basin wells (e.g., Otosaka et al., 2011) that revealed the presence of iodate and organo-iodide species, in addition to iodide, and (ii) laboratory experiments on the sorption of iodide and iodate. The staff noted that there was not enough evidence of the expected iodine speciation in groundwater to use speciation to raise Kd values. The staff also objected to the use of a composite Kd value, which could lead to underestimating the transport rate and distance for more mobile iodine species that may be present.

The treatment of I-129 sorption in the SDF PA was evaluated in a 2017 TRR (ML16342C575).

This report noted that NRC staff had not objected to use of 0.3 mL/g for sandy soil and 0.9 mL/g for clayey soil in previous DOE analyses, but that the values used in a 2014 SA1 mL/g for sandy soil and 3 mL/g for clayey soilwere not adequately supported. Furthermore, NRC staff concluded that the technical bases for the LIFs applied to subsurface Kds were unclear.

Although a 2019 TRR (ML19298A092) was focused on Tank 12 waste release experiments and a DOE report on performance impacts, this TRR also discussed iodine SRS soil sorption tests conducted for the SDF program that simulated conditions during progressive cementitious material degradation (SREL Doc. R-17-0004 Version 1). The NRC staff stated that, in the DOE experiments, I-129 sorption was weak but measurable, ranging from 0.2 to 1.1 mL/g, under only three of the tested conditions: in oxic and anoxic groundwater (pH 4.66 and 5.22) and in oxic calcite water (pH 5.74). At higher pH conditions, including in concrete tests, iodine sorption was statistically indistinguishable from a zero Kd. NRC staff stated that more information was needed to support non-zero iodine Kds for pH values >6. This is consistent with earlier sorption reviews calling for more support for DOEs adopted base case Kd values for iodineand NRC staff is not aware of relevant newer data. DOE should also review the technical bases for the LIFs, if they can be shown to have a significant impact on PA results.

These observations suggest that the iodine Kds will continue to be an important monitoring factor, i.e., Monitoring Factor 4.1 Natural Attenuation of Key Radionuclides in the NRCs Tank Farms Monitoring Plan (ML15238A761).

The NRC staff notes that much has been learned, in large part from DOE-sponsored studies at SRS, regarding iodine aqueous speciation and its implications for I-129 environmental mobility (see recent summaries by Santschi et al., 2017, and Neeway et al., 2019). As discussed extensively in Yeager et al. (2017), microbial effects on aqueous iodine mobility can be profound and need to be considered in developing models for I-129 transport at SRS.

Sensitivity to Plutonium Soil Partition Coefficient, Kd Section 6.4.2 of the SA discussed analyses of the groundwater pathway dose sensitivity to plutonium soil Kds. Table 12 lists the sensitivity cases and Kd values, whereas Table 13 provides results. The analysis considered different values for sand and clay Kd, various fractions of oxidized versus reduced forms of plutonium, and different values for the cement LIF.

For reducing and oxidizing conditions in the conservative case, DOE lowered sand Kd from 650 to 290 mL/g and did not change the clay value. However, in the weighted reducing/oxidizing cases, sand Kd was assumed to be 16 mL/g, which differs considerably from the conservative case value of 290 mL/g. The LIF was varied between 0 and 10.

Table 12 Plutonium soil partition coefficient (Kd) sensitivity analyses (reproduced from Table 6.4-3 of SRR-CWDA-2016-00078)

Table 13 Plutonium soil partition coefficient (Kd) sensitivity analyses results (from Table 6.4-4 of SRR-CWDA-2016-00078)

While the peak doses within 10,000 yrs for the Kd sensitivity cases were not significantly higher than the evaluation case, the peak dose within 20,000 yrs could be significantly higher and greater than the performance objective if more conservative assumptions regarding Pu-239 sorption are considered. Furthermore, the effect of lower plutonium Kd can be seen within 10,000 yrs in Figures 6.4-7 and 6.4-8 of the SA, showing an up-to-5-to-6-fold increase in dose for cases more weighted to the oxidizing Kd values (note that these effects are not seen in the peak dose values of Table 13). Although DOE is correct that peak dose within 10,000 yrs was not affected significantly for the plutonium Kd cases, these results do not consider underperformance of other barriers to waste release in the engineered barriers that could hasten the time to peak dose and lead to higher concentrations and doses. Furthermore, there are still uncertainties and questions regarding plutonium sorption modeling that the NRC staff continues to track.

Regarding the Kd sensitivity analysis reported in this SA, NRC staff has the following comments, while acknowledging that the analysis was published in 2016 and there may have been subsequent developments in how plutonium is modeled by DOE.

DOE should explain the rationale for the difference between the oxidizing case sand Kd used in the conservative case (290 mL/g) and in the mixed redox cases (16 mL/g)

DOE should explain why the same Kd values were used for both reducing and oxidizing conditions in the evaluation and conservative cases

DOE should explain why the conservative clay Kd is the same as in the evaluation case and the oxidizing value used in the mixed redox cases is only slightly lower The LIF has no apparent effect on doses within 20,000 yrs when applied to the evaluation case plutonium Kd values (Figure 6.4-5 of the SA). This result is unexpected, given that (i) for LIF = 0, the effective plutonium Kd is 0 mL/g, (ii) for other cases with lowered but non-zero plutonium Kd values dose was affected within 20,000 yrs, and (iii) other Kd changes had effects within 20,000 yrs. DOE should explain the apparent lack of impact of the plutonium LIF in dose modeling. The NRC staff documented review findings regarding the LIF in previous reports, including the HTF Technical Review Report (ML14094A514). In the 2015 environmental monitoring TRR (ML12272A124), NRC staff stated that the LIF apparently is arbitrarily assigned to adjust Kd and specific LIF values lack technical support. NRC staff continues to monitor this technical issue.

More generally, with respect to DOE selections and treatment of plutonium Kd values, NRC staff continues to monitor a number of technical issues. ML12272A124 discussed how DOE lysimeter study results commonly exhibit evidence for a mobile dissolved plutonium fraction that could be risk significant if it is transported significant distances during the compliance period.

The possibility of a mobile plutonium fraction is consistent with some of the residual waste release experimental results discussed in the previous section on sensitivity to plutonium solubility limits. Subsequent environmental monitoring TRRs pointed to continued evidence for this mobile fraction, as well as for complex, evolving plutonium redox behavior in lysimeter studies. The 2019 environmental monitoring TRR (ML19280A059) emphasized evidence for the mobile fraction and questioned DOEs continued use of weighted average plutonium Kd values to represent multiple oxidation states. Weighted averaging is inappropriate because it will underestimate the transport rate of the mobile plutonium fraction, which can have a significant dose impact if sufficiently abundant. ML19280A059 also stated that desorption experiments, which DOE has used in some cases to support Kd values, may not detect a small mobile plutonium fraction.

These issues were further discussed in the 2021 environmental monitoring TRR (ML21119A316), which reviewed additional lysimeter studies, the results of which confirmed the presence of the mobile fraction and cyclic redox plutonium behavior during transport in the soil column, as well as the sensitivity of plutonium transport to specific geochemical conditions.

For example, NRC staff noted the transport of plutonium to the bottom of a lysimeter in only a few years (as reported in SRRA021685-000013, Revision A). Staff noted in the TRR that it was not aware of DOE efforts to update plutonium transport model treatment to be more technically defensible. While NRC staff acknowledges that the mobile fraction may be small and appreciates DOEs efforts to study the dose sensitivity of various plutonium oxidation states, a more complex model with multiple Kd values to simulate the oxidation and reduction of plutonium along the flow path may be needed if, given all model and data uncertainties, Pu-239 is potentially important to the compliance demonstration.

3.2.4 Sensitivity to Chemical Transition-Time Variability DOE also studied the impact of chemical transition-time variability on peak dose. Because the peak dose is dominated by I-129 at earlier times when there is no solubility control, the impact of chemical transition times is low (only a slight shift earlier in time and slightly lower peak dose occurs for fully oxidized conditions). Although not discussed in the SA, the shift in the peak dose may have resulted from a slight change (lowering) in the basemat Kd for I-129.

The slightly lower peak dose may be related to changes in the timing of releases from multiple sources. A deterministic sensitivity analysis provides limited information about the sensitivity of the results to changes in multiple parameters (i.e., multiple parameters would need to be varied to see an impact of chemical transition-time on peak dose).

3.2.5 Sensitivity to Grout and Concrete Hydraulic Performance Variability For these sensitivity analyses, which were addressed in Chapter 6 of the SA, Tanks 12, 16, and all HTF Type III, IIIA, and IV tanks were modeled using only the PA-assumed hydraulic conductivities, not the scenario-based variable cementitious material properties (i.e., Table 14)

Table 14 Concrete and grout hydraulic conductivity scenario sensitivity analyses (from Table 6.6-1 of SRR-CWDA-2016-00078) that were applied to other Type I and II tanks. Tanks 12, 14, 15 and 16 are always modeled without liners in the PA, given the assumption that their liner failure occurs at time zero. The results from the sensitivity studies of grout and concrete hydraulic conductivity (see Table 14 and Figures 7-9) revealed that early liner failure (sensitivity case no liner) had the most significant impact on the timing and magnitude of peak dose results or that early liner failure was needed to see an impact of grout hydraulic conductivity on the results. For tanks that are not assumed to be initially failed (i.e., Tanks 9-11 and 13), this result is expected. The most conservative conditions modeled (i.e., sensitivity case no liner) resulted in a peak dose of 0.24 mSv (24 mrem) at 2,530 yrs post-closure (Figure 8). The overall peak dose within 20,000 yrs

[0.27 mSv/yr (27 mrem/yr)] is controlled by I-129 released from Tanks 9, 10, and 11 after liner failure occurs in Year 11,397 (Table 14 and Figure 9).

Sensitivity analyses described in Chapter 7 of the SA studied the impact of filling individual tank primaries and annuli with an alternative, non-reducing fill grout (i.e., Controlled Low Strength Materials [CLSM]). Three Case A flow scenarios were considered: (i) average grout, (ii) backfill grout, and (iii) all backfill, where the all backfill case is used to assess potential changes to the integrity of the waste tank concrete vaults. To account for the lack of slag in an alternative CLSM fill grout, two chemical scenarios at time equal zero were considered: (i) SubC, where reduction capacity is equivalent to that provided by tank fill grout as defined in the HTF PA base case (Case A), and (ii) OR_III to represent zero reduction capacity and accelerated pH change provided by CLSM. Two Case C flow scenarios were also considered to compare the impact of CLSM on dose with a fast-flow pathway penetrating the alternative fill grout.

For Tanks 9, 10 and 11, if they are filled with backfill grout having OR_III initial chemistry, (i) the peak doses at the 100-m facility boundary change little with respect to timing or magnitude relative to the evaluation case, and (ii) the Ni-59 peaks shift earlier by ~300 yrs relative to the evaluation case, which has minimal effect on the timing of the secondary peak around Year 15,000. If the Tank 9, 10 and 11 liners remain intact, (i) the peak doses at the 100-m boundary change little with respect to timing and magnitude relative to the evaluation case, Figure 7 Hydraulic performance sensitivity analyses maximum total dose at the 100-m facility boundary for all HTF tanks through 1,000 yrs post-closure (reproduced from Figure 6.6-1 of SRR-CWDA-2016-00078)

Figure 8 Hydraulic performance sensitivity analyses maximum total dose at the 100-m facility boundary for all HTF tanks through 10,000 yrs post-closure (reproduced from Figure 6.6-2 of SRR-CWDA-2016-00078)

Figure 9 Hydraulic performance sensitivity analyses maximum total dose at the 100-m facility boundary for all HTF tanks through 20,000 yrs post-closure (from Figure 6.6-3 of SRR-CWDA-2016-00078) regardless of the hydraulic and chemical properties of the grout and concrete, (ii) the hydraulic properties have little impact on the 100-m dose curves, (iii) the chemical properties of grout/concrete have a minor impact with respect to the peak timing associated with Ni-59, and (iv) for the Case C tank configuration (i.e., partial fast-flow path through the grout; more rapid CZ transition compared to Case B), the 100-m peak doses within 10,000 yrs decrease but occur approximately 200 yrs sooner when the initial chemical environments are set to OR_III instead of SubC.

For Tank 13, if it is filled with backfill grout having OR_III initial chemistry, (i) the peak dose at the 100-m facility boundary changes little with respect to timing or magnitude relative to the evaluation case, and (ii) the Ni-59 peak shifts earlier by ~3,000 yrs relative to the evaluation case, which has minimal effect on the timing of the secondary peak around Year 17,000. If the Tank 13 liner remains intact, (i) the peak dose at the 100-m boundary changes little with respect to timing and magnitude relative to the evaluation case, regardless of the hydraulic and chemical properties of the grout and concrete, (ii) the hydraulic and chemical properties of grout/concrete have a minor impact on the 100-m dose curve with respect to the peak timing associated with Ni-59, and (iii) for the Case C tank configuration, the 100-m peak dose within 10,000 yrs decreases but occurs approximately 100 yrs sooner when the initial chemical environment is set to OR_III instead of SubC.

For Tank 14, if it is filled with backfill grout having OR_III initial chemistry, (i) the peak dose at the 100-m facility boundary increases in magnitude and shifts to occur earlier in time relative to the evaluation case, (ii) the Ni-59 peak shifts earlier by ~7,000 yrs relative to the evaluation case, which has the secondary peak around Year 15,000, and (iii) A small Tc-99 peak becomes noticeable at approximately Year 600. The timing of the peak dose from I-129 for the Case A backfill grout and all backfill sensitivity cases is similar to the timing of the peak dose for the Case C tank configuration. A small Tc-99 peak is observed for all sensitivity cases other than Case A subC and Case A OR_III. Finally, the absence of slag appreciably impacts (i) the I-129 dose-contribution only for the Case A subC sensitivity case, (ii) the timing of the Ni-59 dose for the Case A subC, Case A average grout, Case A backfill grout, and Case A all backfill sensitivity cases; and (iii) for the Case C tank configuration, the 100-m peak dose within 10,000 yrs decreases when the initial chemical environment is set to OR_III instead of SubC.

For Tank 15, if it is filled with backfill grout having OR_III initial chemistry, (i) the peak dose at the 100-m facility boundary increases in magnitude and shifts to occur earlier in time relative to the evaluation case, and (ii) the Ni-59 peak shifts earlier by ~8,000 yrs relative to the evaluation case, which has the secondary peak around Year 14,000. The timing of the peak dose from I-129 for the Case A backfill grout and all backfill sensitivity cases is similar to the timing of the peak dose for the Case C tank configuration. Finally, the absence of slag appreciably impacts (i) the I-129 dose-contribution only for the Case A subC sensitivity case, (ii) the timing of the Ni-59 dose for the Case A subC, Case A average grout, Case A backfill grout, and Case A all backfill sensitivity cases; and (iii) for the Case C tank configuration, the 100-m peak dose within 10,000 yrs decreases when the initial chemical environment is set to OR_III instead of SubC.

3.2.6 Sensitivity to Alternative Fast Zone (AFZ) Pathways Through Tank Vault, Grouted Annuli, and Sand Pads DOE indicates that doses at the 1-m facility boundary associated with flow through a construction joint in a tanks concrete side walls (i.e., the vault) are very high if the distribution coefficient (Kd) for I-129 is assumed to be zero [or a retardation factor (Rf) of 1 is assumed].

DOE also indicates the risk is primarily associated with Type II tanks, which have sand pads that serve as a continuous horizontal flow path between the primary and secondary liners, and likens this simulation to one in which I-129 in the waste is placed directly into the SZ. DOE also indicates that the peak 10,000-yr dose is driven by Type I and II tanks that are assumed to have no liners or annular pans at time zero. Peak doses could exceed performance objectives if the liners are assumed to fail completely and simultaneously at the estimated I-129 inventories and no attenuation. DOE indicated that a patch model was more realistic and would be a focus area during future PA maintenance activities.

An AFZ sensitivity analysis for Type I and Type II tanks previously was performed for the Tank 16 SA (SRR-CWDA-2014-00106) to address the impact on dose at the 1-m facility boundary of short-lived radionuclides released early from the tank annuli CZ. In this previous sensitivity analysis, the grout placed above the CZ was modeled as a waste release barrier that provided a resistance to flow through the CZ. In an updated sensitivity analysis conducted for this SA, the waste release barrier influence of annulus grout was negated (SRR-CWDA-2016-00078) to evaluate whether a high-conductivity flow path could accelerate the early release of short-lived radionuclides (e.g., Sr-90 and Cs-137), highly sorbing species (Np-237 and Pu-239), and influence dose calculations. The two tank types considered each have their own distinct conceptual model, as Type I primary tanks do not sit on a sand pad, while Type II primary tanks sit directly on a sand pad.

The Type I and II Tanks SA considered AFZ models to:

Address concerns related to preferential release pathways on short-lived radionuclides

Reflect the potential for flow through vault walls

Account for effect of grout as a barrier to release of short-lived radionuclides

Examine the influence of fast zones of relatively more highly sorbing, long-lived radionuclides The primary focus of the AFZ analysis was on (i) short-lived radionuclides Sr-90 and Cs-137 and (ii) long-lived radionuclides Np-237 and Pu-239, which leaked from the primary liner of Type I and II tanks during HTF operations, but were captured by the secondary liner inside the tank annuli. The AFZ analysis considered release and transport consequences for these radionuclides due to groundwater flow passing between the primary and secondary liners.

Tank Construction Details Relevant to Flow Within Type I and II Tanks The Type I and II tank primary and secondary liners have a base plate constructed from 0.0127-m (0.5-in)-thick rectangular steel plates. Construction plans specified through-welded butt joints and a backing strip on the underside of all bottom plate seams the Type I primary liner (W145379) and the Type I and II secondary liners (W145367). Construction plans for the Type II tank allow either double-sided welds or single-sided welds with a backup strip (W162672). The vertical wall of the secondary liner extends 1.52 m [5 ft] above the concrete working slab, leaving an annulus between the primary and secondary liners. The exterior concrete vault wall for the tanks was cast using the secondary liner as part of the interior form.

In Type I vaults, the secondary liner rests on a 7.6-cm (3-in)-thick grout layer poured on top of the basemat and the primary liner rests on a 7.6-cm (3-in)-thick grout layer poured on the secondary liner (SRR-CWDA-2010-00128, Revision 1, Figure 3.2-3). In Type II vaults, the grout layers are both replaced with a 2.5-cm (1-in)-thick compacted sand layer (SRR-CWDA-2010-00128, Revision 1, Figure 3.2-16). Given specifications provided in the PA, the top of the secondary liner is at elevation 75.9 m [249 ft] (Tanks 9 and 10), 75.6 m [248 ft] (Tanks 11 and 12), and 85 m [279 ft] (Tanks 13-16). The AFZ analysis does not discuss the contact between the primary liner and the underlying grout or sand layer. A gap may exist beneath the Type I primary and secondary liners because of the backup strips below the welded seams.

The backup strip thickness is not specified, but MIL-STD-22 D requires a minimum thickness of 0.635 cm [0.25 in] if the plate thickness is >0.79 cm [>5/16 in]. Assuming a radius of 10.36 m

[34 ft] and backup strip thickness of 0.635 cm [0.25 in], the gap volume beneath the primary liner may be 2.14 m3 (566 gal). The Type II liners may have a similar gap, a partial gap, or minimal gap, depending on (i) whether backup strips were used and (ii) the extent to which the underlying sand layer was rearranged during and after emplacement to accommodate irregularities.

Concrete vault construction details for Type I tanks (W145225) indicate that the walls were joined to the basemat using a keyed horizontal joint with a vertical copper waterstop (the wall rests on the basemat), but the detail does not indicate other joints in the walls or describe the concrete pour strategy. Concrete vault construction details for Type II tanks (W163018) specify that the vault walls were joined to the basemat using a vertical keyed joint with a horizontal copper waterstop (the wall is adjacent to the end of the basemat), and the concrete was poured in a single monolithic pour.

AFZ Source Specification The AFZ analysis assumed that leaks through the primary liner resulted in highly soluble material deposited within the tank annuli. The AFZ analysis assumed that (i) some leakage from Tanks 13-16 also leaked into the sand pads below the primary liner, (ii) overflow from the Tank 16 secondary liner in 1960 migrated into the sand pad under the secondary liner, (iii) cleaning activities remove most inventory, and (iv) all annuli will be grouted before closure.

Table 15 lists the radionuclide inventories and liner failure times assumed for this SA, as obtained from the GoldSim model. All annulus inventories are based on estimates of material radionuclide concentrations scaled by material volume, assuming trace material volume remains after annulus cleaning, except for Tank 16. Updated inventory estimates for this SA accounts for ongoing waste removal with a trace volume of 0.379 m3 (100 gal). For comparison, leakage filling a hypothetical 0.635 cm [0.25 in] gap under a Type I primary liner would result in a 2.14 m3 (566 gal) primary pad residual volume, compared to the assumed 0.379 m3 (100 gal) trace volume in an annulus.

Type I tanks have most of their residual inventory inside the primary tank and only a small fraction of their residual inventory in the annulus. Type II tanks have different combinations of inventories and liner failure times based on operational history.

GoldSim AFZ Flow and Transport Model Most HTF tanks are located above the water-table, thus are expected to have predominantly vertical flow paths through the tanks. Type I and II tanks in the HTF are partially to completely submerged, leading to potential lateral flow in the portions of the tank below the water-table.

For Type I tanks, the 2016 SA AFZ analysis considers flow zones that pass over or through the base of the annulus, directly contacting the annulus residual volume. For Type II tanks, the AFZ analysis considers lateral flow that (i) passes through the sand layer between the primary and secondary liners, and (ii) exits the tank through the annulus and wall.

Tank Item Units 9

10 11 12 13 14 15 16 All radionuclides Tank 229.2 229.2 229.2 82.4 229.2 229.2 229.2 9.4 Annulus 0.77 0.77 0.77 0.18 0.77 0.77 0.77 13.75 Primary pad/sand 0

0 0

0 0.77 9.53 0.77 13.75 Secondary pad/sand kCi 0

0 0

0 0

0 0

0.19 Sr-90 + Cs-137 Tank 206.7 206.7 206.7 81.1 206.7 206.7 206.7 9.4 Annulus 0.73 0.73 0.73 0.18 0.73 0.73 0.73 13.70 Primary pad/sand 0

0 0

0 0.73 9.4 0.73 13.70 Secondary pad/sand kCi 0

0 0

0 0

0 0

0.19 Np-237 + Pu-239 Tank 130.5 130.5 130.5 43.2 130.5 130.5 130.5 0.22 Annulus 0.25 0.25 0.25 0.06 0.25 0.25 0.25 4.72 Primary pad/sand 0

0 0

0 0.25 3.21 0.25 4.72 Secondary pad/sand Ci 0

0 0

0 0

0 0

0.06 Liner failure time (yrs)

Diffusion control 11,397 11,397 11,397 0.01 12,687 0.01 0.01 0.01 Groundwater contact*

Year 1,142 1,142 1,142 0.01 2,506 0.01 0.01 0.01

• Early failure mode, not used in AFZ analysis Table 15 Source characteristics/inventories/liner failure times for Type I and II tanks The PA (SRR-CWDA-2010-00128, Revision 1) uses PORFLOW simulations to perform detailed flow and transport numerical calculations. It used 2D PORFLOW models in cylindrical coordinates (i.e., radial, r, and vertical, z) for the near-field containing the engineered system and a 3D PORFLOW model in cartesian coordinates (i.e., x, y, z) for farfield simulations; the 3D model uses 2D model outputs as a source term. This near-field model design choice was made to improve model resolution compared to a fully 3D PORFLOW model. PA GoldSim models were abstracted (or simplified) versions of the PORFLOW near-field models that were benchmarked against PORFLOW simulation results. The GoldSim models are essentially 1 D in the vertical direction with a small amount of radial exchange, thus they are also conceptually representative of cylindrical tank systems.

Figure 10 illustrates the GoldSim conceptual approach for this SAs near-field models without lateral flow. The AFZ analysis considers lateral flow in Type I and II HTF tanks by adding a series of domain cells aligned with an assumed unidirectional lateral flow zone. These domain cells interact with the previously implemented cells used to represent vertical flow. In Type I tank models, the lateral flow zone wraps around the outside of the annulus (i.e., just inside the vertical wall of the secondary liner) but flow does not go directly under the primary liner.

In Type II tank models, the lateral flow zone runs in a sand layer under the primary liner and no-flow goes around the perimeter of the primary liner in the annulus bottom. The SA appears to base these modeling choices for the AFZ pathway on comparisons of transmissivity between two pathways in parallel, assuming (i) a shrinkage crack is much more transmissive than a grout pad with small hydraulic conductivity under the primary liner (Type I tanks) and (ii) a sand layer under the entire primary liner is much more transmissive than any annulus pathway (Type II tanks).

GoldSim AFZ Flow and Transport Model for Type I Tanks The Type I tank AFZ analysis is based on a conceptual model of groundwater (i) entering the annulus between the secondary and primary steel liners through a horizontal construction joint in the wall at the top of the secondary liner, (ii) flowing vertically in the annulus to the base of the annulus, (iii) flowing horizontally around the annulus in a shrinkage crack along the vertical wall of the secondary liner, (iv) rising vertically back to a horizontal construction joint above the secondary liner, and (v) exiting the tank. The GoldSim model provides several options to represent flow moving within the annulus from the construction joint to the shrinkage crack:

(i) undegraded grout, (ii) degraded grout, and (iii) a shrinkage crack along the vertical wall of the secondary liner. The flow rate through the fast path is based on the overall path hydraulic conductivity and the assumed lateral hydraulic gradient across the tank. The SA assumes that the flow path within the annulus is entirely in a series of three shrinkage cracks (vertical on entry and exit, and horizontal around the perimeter of the grout), which generates the largest flow rate among the three options.

The Type I tank AFZ analysis considers two CZs: (i) above the bottom plate of the primary liner (tank CZ) and (ii) at the base of the annulus grout immediately above the secondary liner (annulus CZ). Figure 11 illustrates release pathways from these CZs that are related to the AFZ. The GoldSim implementation allows multiple additional release pathways through the secondary liner and the basemat for both CZs (as shown in Figure 10); these additional pathways were deactivated for the AFZ analysis.

Figure 10 Simplified representation of GoldSim radionuclide transport model for Type I and II tanks without an AFZ. The primary and secondary pads are (i) grout for Type I tanks and (ii) a sand layer for Type II tanks. Flow rates used for advective exchanges are based on PORFLOW simulations. Initial inventories are tank-specific. Image is redrawn from Figure 1.

Figure 11 Simplified representation of the GoldSim transport model implementation exchanges between initial inventory and the AFZ for Type I and II tanks. The Type I tank model (top) segments the annulus domain into a sequence of cells to represent radial diffusion through the annulus to the AFZ shrinkage crack along the inside of the secondary liner vertical wall. The Type II tank model (bottom) adds a separate AFZ domain consisting of a series of cells to represent lateral flow through the primary sand pad and across the annulus. Initial inventories are tank-specific. Liner cells are used to organize flow paths.

The annulus bottom is represented by variably sized radial concentric grout shells between the vertical walls of the primary and secondary liners, and the AFZ domain is represented by a series of ten uniformly sized cells representing a shrinkage crack wrapping around the inside perimeter of the secondary liner. Radionuclides are allowed to diffuse between (i) the primary liner cell and the inner annulus shell, (ii) adjacent annulus shells, (iii) each annulus shell and the underlying secondary liner (after liner failure) and (iv) the outer annulus shell and each of the AFZ cells. As the AFZ removes mass from the outer shell, a radionuclide concentration gradient develops that allows the radionuclides to diffuse within the grout toward the AFZ.

An additional series of five cells represent the vertical outflow transport from the downstream end of the annulus base to the joint in the wall where flow exits (i.e., the vertical leg of the AFZ crack on the downgradient end). The model assumes no transport occurs in the inflow vertical leg of the shrinkage crack (i.e., the water entering the first upgradient AFZ shrinkage crack cell is clean).

The tank CZ is not directly connected to the AFZ pathway. Instead, the model assumes vertical flow through the tank grout within the primary liner will carry dissolved radionuclides past the bottom of the primary liner once the primary liner fails. The flow rate through the grout was calculated from PORFLOW simulations. This flow collects in the primary liner cell domain and is partitioned into several pathways. The only modeled pathway for releases from the primary liner CZ to reach the AFZ requires advection from the tank grout to the primary liner, then diffusion within the annulus grout to the AFZ.

The AFZ domain is initiated with the entire annulus CZ inventory. Releases from the CZ are delayed until the primary liner fails (i.e., 11,397 yrs post-closure for Tanks 9-11, and 0.01 yr post-closure for Tank 12) and flow begins through the tank grout. The primary liner is assumed to fail due to corrosion, with a precalculated failure time that does not consider the presence of an AFZ. The degradation model assumes ingress of carbon dioxide, chloride and oxygen, which lead to carbonation, corrosion and eventual failure of the liners. The steel liner failure model assumes diffusion through grout controls the rate of ingress of deleterious species and therefore the failure time of the steel liners. The secondary liner is assumed to fail as soon as the primary liner fails.

GoldSim AFZ Flow and Transport Model for Type II Tanks The Type II tank conceptual model differs from the Type I model (Figure 11) because the sand layers under the primary and secondary liners represent continuous, wide flow paths with much larger hydraulic conductivity than grout, thus flow will preferentially enter the sand layers.

The Type II tank model assumes that fully connected flow paths always exist, even though measured water-table elevations have been observed to occasionally drop below the base of the secondary liner.

The Type II tank AFZ analyses also assume fast-flow through the concrete vault wall via construction joints. The analyses are similar to the Type I approach, except that the entry and exit flow paths between the exterior environment and the top lip of the secondary liner are through the wall/floor joint in sequence with a shrinkage crack along the exterior of the secondary liner, instead of through a horizontal joint at the top of the secondary liner.

The model includes three options to represent the connection between the top lip of the secondary liner and sand pad on entry and exit: (i) flow from the liner to the primary sand pad through intact annulus grout (the partial fast flowpath case), (ii) flow from the joint to the sand pad through a shrinkage crack and a seam at the bottom of the annulus (the complete fast flowpath case), and (iii) an intermediate case with flow from the joint to the primary sand pad through degrading annulus grout.

The Type II tank AFZ analyses consider both tank and annulus CZs, a CZ in the primary sand layer (the primary sand layer CZ), and, for Tank 16, an additional CZ with a small inventory in the secondary sand layer (the secondary sand layer CZ). The primary sand layer AFZ is modeled with a sequence of 10 cells that are initiated with the entire primary sand layer CZ inventory and that can receive released radionuclides from the tank CZ. An additional series of five cells represents flow from the downstream end of the sand layer to the joint in the wall where flow exits. The AFZ flowpath through the sand pad is approximated by a rectangular strip. Instead of a series of radial cells, the annulus is represented by a single cell upstream of the sand pad and a single cell downstream of the sand pad, each of which are assigned half the annulus CZ inventory. Figure 11 illustrates release pathways from these CZs that are related to the AFZ; multiple additional release pathways through the tank components that bypass the AFZ are omitted, for simplicity.

The Type II tank AFZ release pathways are connected differently than the Type I tank pathways. Once the primary liner fails, vertical flow through the tank passes through the tank CZ and picks up radionuclides, passes to the AFZ cells, and passes back out to the secondary liner cell and down through the secondary sand layer to the basemat and out of the tank structure. Within the sand layer AFZ cells, transported radionuclides partition into the vertical and AFZ flow, with no diffusive transport. The small inventory in the Tank 16 secondary sand layer CZ can only exit via the basemat (i.e., it cannot leave through the AFZ).

Like the Type I tank AFZ model, releases from the Type II tank CZ are delayed until the primary liner fails (i.e., 12,687 yrs post-closure for Tank 13, and 0.01 yr post-closure for Tanks 14-16) and flow through the tank grout initiates. Liner failures are calculated the same way for Type I and II tanks.

Discussion of AFZ Implementation Staff considered the implementation of the AFZ analysis by examining the following risk-significant aspects of the modeling approach:

Internal consistency and conservatism of the AFZ implementation o

CZ inventories o

Liner performance and failure time o

Releases from the CZ o

Partitioning of released radionuclides between slow and fast-flow paths

Groundwater factors that may reduce peak concentration at the 1-m well o

The representation of velocity and spatial diffusion o

The representation of temporal diffusion Internal Consistency and Conservatism Assessment of the AFZ Implementation for Type I Tanks The Type I tank AFZ analyses do not address the contact between the liners and the grout that they rest on. Liner seam backup strips are called for in construction details. Such strips presumably would lift the liner off the underlying grout, forming a grid of open gaps under both the primary liner and secondary liner. Over the large expanse of the liner bottom, it may be that there is irregular contact between steel and grout that would form thin gaps, so a continuous network of gaps may form over the expanse of the liner bottom; this would be the case even without backup strips. Several consequences are envisioned:

The initial inventory outside the primary liner may be several times larger than the assumed trace amounts in the annulus CZ, if the gaps are a reservoir for leaked supernate.

Such gaps would create a low-resistance pathway, fostering enhanced flow below both the primary and secondary liners.

Lateral flow under the primary liner implies that the entire tank CZ inventory is directly exposed to the AFZ after the liner fails, in the same way as the Type II tanks.

The assumed tank inventory has 30x larger volume for uncleaned tanks than the assumed annulus inventory, although the mix of radionuclides differs between the tank and annulus, and key radionuclides in the tank inventory are expected to be less soluble than key radionuclides in the annulus inventory.

The liner failure time model is based on diffusion of carbon dioxide, oxygen and chloride through the concrete vault exterior (i.e., flow does not affect concrete degradation or liner corrosion). Lateral flow under the liners for Tanks 9-11 would bring groundwater past the base of the liners, short-circuiting the delivery of these constituents to the liners (Tank 12 is assumed to be initially failed at time equal zero). Life estimation calculations of HTF steel by SRNL-STI-2010-00047 suggest that pitting corrosion caused by groundwater contacting tank steel will penetrate the thickness of the liner within a few decades. Even a two-stage failure model, where first the more accessible secondary liner is breached and then the primary liner is attacked, may greatly reduce the lifetime of the initially intact primary liners.

Internal Consistency and Conservatism of the AFZ Implementation for Type II tanks The Type II tank AFZ analyses consider a fast pathway in the sand layer beneath the primary liner, but not in the sand layer beneath the secondary liner. Three of the four Type II tank liners (Tank 14, 15, and 16) are assumed to fail essentially immediately after closure.

The fast pathway through the primary sand layer is assumed to always be active. Flow through the assumed flow path requires that water levels are above the lip of the secondary liner pan, but observed water levels have not always risen to this elevation (Figure 12). Although not discussed in the SA, continual flow may lead to overall higher release rates of key radionuclides through the AFZ pathway than would be released in the scenario with intermittent flow, all else equal, although the release concentrations could be higher if intermittent flow is considered.

The AFZ implementation appears to have several nonconservative inconsistencies, however:

The AFZ implementation assumes no lateral flow passes through the secondary sand layer. Flow through the secondary sand layer would be likely to have a larger magnitude than flow in the assumed AFZ pathway, given (i) the Type II tank construction configuration has the secondary sand pad terminating adjacent to the vertical joint between wall and basemat, providing a relatively short pathway (joint to sand layer to joint) for groundwater to migrate from the external environment to the secondary sand layer and (ii) any cracks in the basemat would allow additional pathways for groundwater flow to penetrate to and from the secondary sand layer. Observed water levels have often risen above the base of the Type II primary liners but less frequently above the secondary liner pan lips (Figure 12), implying that a pathway that includes the secondary sand layer would be able to flow much more frequently than a pathway constrained by the secondary liner lip elevation.

Once the secondary liner fails (i.e., at the same time as the primary liner fails), flow in the primary and secondary sand layers would merge into a single flow pathway occupying essentially the entire space under the primary and secondary liners, including the secondary sand pad under most of the annulus. Overall resistance to flow would likely be smaller for the merged pathway than for the model-assumed pathway, so flow rates may be considerably larger. The primary and secondary sand layers would intercept essentially all of the released radionuclide mass in the tank CZ, which may result in substantial mass bypassing the basemat pathway.

The Tank 13 primary liner is assumed to remain intact for 12,687 yrs post-closure.

The diffusion-constrained transport model used to calculate failure time may no longer constrain failure, because flow in the secondary sand layer would bring groundwater in contact with the secondary liner. Once the Tank 13 secondary liner fails, the merged primary and secondary sand layers would allow the same transport process to act on the Tank 13 primary liner.

Internal Consistency and Conservatism of the AFZ Implementation for Both Tank Types

The tank CZ is assumed to release radionuclides across the primary liner solely due to vertical advective transport, nonconservatively neglecting diffusive transport. This assumption may have small consequences for release rates with sufficiently large vertical flow rates, but may significantly underestimate release rates with smaller flow rates. Given that (i) the vertical flow rate is controlled by the tank grout hydraulic conductivity, (ii) the entire release pathway is saturated for Type I and II tanks, and (iii) vertical hydraulic gradients may be very small for the submerged Type I tanks, including diffusion may result in significantly larger release rates from the tank CZ immediately after liner failure. Incorporating diffusive releases would tend to increase concentrations in the initial pulse after the liner fails, which may be the most consequential period for calculating peak dose. Underestimating the initial pulse is especially consequential when the AFZ flow pathway is immediately below the primary liner.

Flow rates through the AFZ pathway are calculated based on an assumed gradient of 0.01. The historical record of observations suggests that water-table gradients across the Type I and II tanks were frequently several times larger than this, and the flow direction was not always consistent over time (Figure 12). Calculating an appropriate magnitude is difficult for these tanks, given the curvature of the water-table mound due to recharge, the ever-evolving set of wells with observations, the different rates of water-level changes across the site, and the unknown influence of a closure cap on recharge rates and flow directions. Nevertheless, it may be appropriately conservative to calculate AFZ flow rates using gradients that are several times larger.

Figure 12. Water levels in the UTRA-UZ at tank group centers (left) and the corresponding horizontal gradient at the same location (right). The rows are ordered from west to east, and the dots are colored by time. Water levels and gradients represent values over a 21-day interval, calculated using trilinear interpolation among just the wells with observations during the snapshot. In the water-level plots, blue lines are the assumed water level from Table 4.2-15 of SRR-CWDA-2010-00128, Revision 1, and gray lines indicate the top of the working slab/bottom of the basemat. For Type I and Type II tanks, the pink, orange, and gray zones are the elevation range of the primary steel liner above the secondary steel liner, secondary steel liner, and basemat. Type II tanks are the only ones with the secondary liner and basemat in the depth range of water-table fluctuations. In the gradient plots, positive values represent flow to the east and north, and the concentric circles indicate gradient intervals of 0.01.

Changing water-table gradients over time may change flow paths within the tank, which would tend to increase dispersion within the AFZ flow path, yet dispersion is not modeled. It is not clear whether incorporating dispersion along the AFZ flow path would tend to increase peak dose consequences by enhancing early releases of rapidly decaying radionuclides or decrease peak dose consequences by smearing the arrival times. It may be worthwhile to perform a sensitivity case for a representative tank to confirm that neglecting dispersion within the AFZ flow path is conservative.

If future steel liner degradation modeling considers the influence of advective inflow of deleterious species on degradation, it is recommended that variation in the AFZ flow path is considered. Unidirectional flow may lead to degradation proceeding from the upstream end.

With variable flow, degradation would be more uniform, which would tend to create more uniform failure and release conditions. A single failure event tends to generate the largest release pulse, all else equal.

Summary of Findings Related to the AFZ Implementation The NRC staff recognize that the AFZ model shows the importance of fast pathways with respect to dose consequences. The Type I tank AFZ analysis clearly demonstrates the importance of intact grout, even if degraded, in limiting releases by reducing flow rates. The NRC staff suggests that the Type I tank AFZ models may underestimate release rates due to the following factors:

The Type I AFZ models do not consider a steel backing strip specified below all bottom plate welds in design drawings, which may have created a significant gap beneath both the primary and secondary liners, caused by steel backing strips under all welds. A gap would provide both (i) a potential reservoir (which is not accessible for cleaning) for overflow from the annulus and (ii) a fast pathway similar to a sand layer. Accordingly, it may be appropriate to consider a Type I tank AFZ model similar to the Type II model.

Even if the backing strips are determined to not be present, the Type I tank AFZ model tacitly assumes that the fast pathway from a shrinkage crack dominates lateral flow through the tank. An irregular contact between liners and their underlying grout pads may allow fast pathways under the liners that would intercept releases from the primary CZ.

The liner failure model for the AFZ analysis does not consider the role of the AFZ pathway on liner degradation for Tanks 9-11 (liners for Tank 12 are assumed to be initially failed). The failure model assumes that failure is mediated by diffusion of oxygen, carbon dioxide, and chloride through the tank walls. Liner failure analyses considering the effect of groundwater contact on liners reduce the failure time by

>10,000 yrs. An AFZ pathway passing the bottom plate of a liner would bring groundwater in contact with the liner, which may facilitate early liner failure.

Further, the Type II tank AFZ model may underestimate release rates for Type II tanks because:

There is a likely fast flow pathway for the Type II tanks that is distinct from the hypothesized AFZ pathways. The secondary sand layer would be readily accessible to vertical flow through cracks in the basemat and the vertical joints between the tank wall and basemat. Once the secondary liner fails, the primary sand layer would join the pathway. A similar fast flow pathway would be likely for Type I tanks if there is a gap or poor contact between the liners and underlying grout.

The liner failure model for the AFZ analysis does not consider the role of the AFZ pathway on liner degradation for Tank 13 (the other three tanks are assumed to fail early). An AFZ pathway passing the bottom plate of a liner would bring groundwater in contact with the liner, which may facilitate early liner failure.

Finally, the AFZ models may underestimate release rates for both Type I and II tanks because the transfer of radionuclides from the primary tank CZ is assumed to occur by advection alone, with the flow rate controlled by tank grout. For a period after failure, diffusion is likely to provide faster release rates to the AFZ, which may significantly increase peak dose equivalent concentrations of rapidly decaying radionuclides.

The SA assumed that groundwater flow rates are steady within the AFZ and in the SZ. The observed variation in groundwater levels and inferred groundwater flow directions within the HTF vicinity may influence the AFZ analyses in various ways, such as:

Representing the flow rates as steady and neglecting the influence of fluctuations on dispersion of released radionuclides would tend to increase peak concentrations at compliance points, so these modeling choices for representing the effects of groundwater flow on transport are conservative both within the tanks and in the groundwater system.

The representation of the AFZ pathway is most risk significant for (i) rapidly decaying radionuclides that are also weakly sorbing (because of the competition between transport and decay) and (ii) radionuclides that sorb much more strongly to cementitious materials than the natural environment (because sorption to the cementitious materials is the transport-limiting step).

Evaluation of Groundwater Levels and Flow Paths Near HTF Type I and II Tanks The shallow groundwater system consists of, in descending order, the Upper Zone (UZ) and Lower Zones (LZ) of the Upper Three Runs Aquifer (UTRA) and the Gordon Aquifer.

The UTRA-UZ is the unconfined surficial aquifer; lower units are confined. Over portions of the UTRA away from groundwater divides, flow is predominantly horizontal with a smaller vertically downward component; flow is predominantly vertical near groundwater divides, however, because the horizontal gradient is small. Discharge from the UTRA-UZ is largely to nearby streams or seeps, combined with vertical losses to the UTRA-LZ. The H-Area straddles the groundwater divide between the Upper Three Runs watershed (to the north and east) and the Fourmile Branch watershed (to the west and south) (Figure 13). The configuration of the stream channels near the HTF, combined with the local topographic relief, results in UTRA-UZ flow from the HTF tending to be captured by Crouch Branch to the northwest and McQueen Branch to the northeast, and Fourmile Branch to the southwest and south. The UTRA-LZ is less affected by small streams, thus the flow directions in the UTRA-LZ and UTRA-UZ differ and release trajectories change direction as water transfers vertically from the UTRA-UZ to the UTRA-LZ (Figure 6).

A set of UTRA-UZ monitoring wells drilled in the 1970s and 1980s (HTF wells) were located around the tanks within the HTF; a more recent set around the perimeter of the HTF (HAA wells) were drilled in the 1990s. Starting in 2012, data collection was based on a revised monitoring plan, with less frequent acquisitions and well data was no longer collected from most HTF wells. Tank and well locations are shown in Figure 13.

The PORFLOW and GoldSim models used for transport calculations approximate groundwater flow as if it were a steady-state system. PORFLOW groundwater simulations are performed for the regional system. PORFLOW transport calculations use a smaller domain with a finer grid to reduce numerical dispersion, and interpolate the regional flow to the finer grid. Local particle tracking pathlines within the HTF from the center of each tank (Figure 6; Figure 3.4-1 of this SA) show (i) the Type I tanks (Tanks 9-12) and Type II Tank 14 exiting to the NW; (ii) the other Type II tanks (Tanks 13, 15, and 16), the Type III (29-32) tanks, the Type IV tanks (21-24), and the western Type IIIA tanks (35-37) exiting generally to the W; and (iii) the eastern Type IIIA tanks (38-43 and 48-50) exiting to the N. These pathways are generally reflective of a three-way groundwater divide. A later version of the groundwater model (SRNL-STI-2017-00445 Figures 12 and 14) shows pathways exiting the 100-m boundary to the northwest and north for all HTF tanks, consistent with flow primarily to Crouch Branch and a few of the eastern Type IIIA tanks possibly captured by McQueen Branch.

Available water-level data from the vicinity of Type I and II tanks (SRNL-STI-2010-00148 and following monitoring reports) suggest that (i) observed water levels in HTF wells surrounding the Type I tanks were between 80 and 86 m [262 and 282 ft] elevation and (ii) observed water levels in HTF wells surrounding the Type II tanks (HTF-5 through HTF-8 and HTF-34) were between 76 and 91 m [250 and 299 ft] elevation. Certain extremely low values at HTF-34 and extremely high values at HTF-5 and HTF-6 are not mirrored by data from the other wells, thus may not be representative of expected future behavior.

All observed water levels near the Type I tanks were always at least [13 ft] above the top of the Type I tank secondary liner. Most observed water levels near the Type II tanks were below the top of the secondary liner {85 m [279 ft]}. For both tank types, the water-level behavior in the HTF wells appeared to change after a large drawdown occurred in 1986, with the water levels somewhat lower, and with steeper gradients in the tank area. Water levels may have been affected by pump-and-treat activities from 1997 through 2003 (SRNL-STI-2015-00351).

These systemic changes appeared to persist until the HTF wells were removed from DOEs environmental monitoring plan in 2012.

Figure 13 (a) Location map with watershed boundaries and surface water features acting as shallow groundwater discharge locations. (b) Position of HTF tanks and wells.

The available water-level data can be used to estimate the groundwater level and gradient at the Type I and II tanks (Figure 12) using linear interpolation between the triangulated wells and data that were measured within a few weeks of each other6. The centroid coordinate of the various tank groups is approximated using linear interpolation between monitoring wells (e.g.,

the Type I tank reference location is approximately at the midpoint between HTF-2 and HTF-4).

Computations of groundwater levels and gradients at the tank group centroid use trilinear interpolation on a Voronoi triangulation for the monitoring wells; the triangulation changes between observation periods to include just monitoring wells with observations in the period.

The water levels and gradients are color-coded by date in Figure 12.

The interpolated groundwater levels at the tank locations appear to respond roughly in unison, reflecting overall wet and dry periods. However, the estimated gradients at the tank locations exhibit more variability in time, especially in the years prior to 2012, which included water-level data from HTF wells (i.e., blue and green dots in Figure 12). Estimated gradients at the Type I and II tanks range from <0.01 to >0.06. At different times, each tank location has gradients that are consistent with ultimate capture by the three surface-water discharge controls (i.e.,

Crouch Branch, McQueen Branch, and Fourmile Branch). The most recent water-level data based on the circumferential HAA wells suggest relatively milder gradients to the northwest for the Type I tanks and to the north and northeast for the Type II tanks.

When groundwater level data are contoured, the earlier data indicate higher water levels in the HTF area than in the surrounding unpaved areas, implying that the HTF is a recharge zone to the aquifer, despite the impermeable cover over most of it. Typically, individual wells had water levels indicating local highs, suggestive of focused recharge. Wells HTF-5 and, to a lesser extent HTF-6 (east and west of Type II Tank 14, respectively), in particular, tended to have anomalously-high water levels suggestive of recharge near Tank 14. Gradients calculated using wells impacted by focused but local recharge may be unrepresentative of the overall flow regime.

The groundwater flow direction and water-table elevation near the Type I and II tanks has historically appeared to respond to perturbations in a complex way. Part of this complexity is due to varying responses to precipitation and recharge across the site, especially in areas with a large fraction of impermeable cover, and the resulting apparent migration of the water-table mound back and forth near the tanks. Some part of the complexity is also due to the changing network of wells with water-level observations; for example, wider spacings between wells result in shallower gradients in this area.

The uncertainty in water levels is conservatively addressed in the AFZ analyses by assuming that the AFZ flow paths through the tanks are always active. The gradient used to calculate the flow rate, however, may be unrepresentative, because the observed water-table gradient appears to be several times larger than the value that was used.

The fluctuating groundwater flow field implies that tank releases will follow different trajectories at different times. The SA assumes that the groundwater flowfield is steady in time, and the 6 Although SRNL-L3200-2015-00008 evaluates the hydraulic gradient near HTF based on a fourth quarter 1995 water table map, the TRR analysis considers areas local to HTF tanks at specific times over a long period of time and evaluates both the magnitude and direction of the hydraulic gradient.

diffusion coefficient is calculated based on steady flow. One way to address fluctuating flow fields is to increase the dispersion coefficient, but this option is not exercised in this SA.

It is not clear whether it is conservative to assume a smaller dispersion coefficient. For a single release plume, using the smaller dispersion coefficient based on a steady flow field should be a conservative assumption because increasing dispersion tends to reduce peak plume concentrations. For release plumes from multiple sources, increasing dispersion may blend plumes together in a way that generates a larger overall peak.

Implications of Groundwater Flow on Dose at 1-m Wells Type I tanks and Cs-137 and Sr-90 doses Radionuclide release breakthrough curves at the 1-m boundary for Type I tanks are shown in Figures 6.7-6 through 6.7-10 of the SA. These indicate a peak before 500 yrs and generally less than 1 mrem/yr for the rapidly decaying Cs-137 and Sr-90 radionuclides, compared to compliance limits of 0.25 mSv/yr (25 mrem/yr) [e.g., Figure 14(a)]. For these radionuclides, radioactive decay reduces the concentration by approximately an order of magnitude every 100 yrs.

The Tank 12 results appear anomalous, because the peak is at least four orders of magnitude smaller than for Tanks 9-11, although the inventory is only a factor of five smaller and the liner is assumed to fail immediately. The modeled path length from Tank 12 to the 1-m boundary is longer than for Tanks 9-11, and the longer travel time may allow the rapidly decaying radionuclides to decay prior to reaching the 1-m boundary.

The early Tank 12 liner failure time should have negligible influence on the calculated peak concentration in the GoldSim model, because the diffusion path through the grout to the AFZ delays tank releases enough to fall out of the decay window for transport to carry released radionuclides to the boundary before decay. If a gap exists between the liners and grout for Type I tanks, however, an AFZ may be appropriate, passing through the gaps and through cracks in the basemat. This may allow more rapid travel through the tank, and allow immediate release from the much larger tank inventory as well as diffusive releases from the annulus CZ.

In this conceptualization, early diffusion from the tank waste to the alternative AFZ may dominate the current release rates calculated based on advection through the tank grout.

The combination of the AFZ with diffusive releases from the tank CZ may result in a much larger peak. Nevertheless, consequences from AFZ conceptualizations for Tank 12 are likely to be small compared to the Type II consequences.

Type II tanks and Cs-137 and Sr-90 doses Radionuclide release breakthrough curves at the 1-m boundary for Type II tanks are shown in Figures 6.7-11 through 6.7-14 of the SA, assuming a continuous pathway through a shrinkage crack within the annulus [Figure 14(b) reproduces Figure 6.7-13 of the SA]. Type II tanks are closer to the 1-m boundary than Type I tanks; the Type II peak is earlier by ~100 yrs and orders of magnitude larger than the Type I peak [(200 to 20,000 mrem/yr) versus (105 to <2 mrem/yr) for Sr-90, and (2 to ~6,000 mrem/yr) versus (<106 to <0.002 mrem/yr) for Cs-137).

However, all breakthrough rates rapidly drop off to (<106 mrem/yr) within 500 yrs (consistent with rapid radionuclide decay), and peaks at the 100-m boundary are <0.001 mSv/yr

(<0.1 mrem/yr) for Tank 16 (i.e., the tank with the largest 1-m peak dose). Therefore, reducing the uncertainty regarding the mobility of short-lived radionuclides such as Sr-90 could be important to the compliance demonstration.

The SA shows that dose contributions are negligible for the Type II tanks when the vertical leg of the AFZ flow pathway within the annulus is through degraded grout, and argues that the continuous shrinkage crack assumption is highly conservative.

The assumed AFZ flow path, however, may not be representative of a system with failed liners.

The model assumes that the secondary liner remains a barrier to AFZ flow even after the liner fails. If the secondary liner is not a barrier after failure, an AFZ consistent with this observation entirely bypasses the vertical wall of the secondary liner and the annulus grout, and would offer less hydraulic resistance than already assumed for the continuous path. In this scenario, the presented results may underestimate releases, especially if the model enables diffusive releases from the tank CZ.

Type I tanks and Np-237 and Pu-239 doses The AFZ calculations show that Np-237 and Pu-239 initially arrive at the 1-m boundary at approximately the same time as Cs-137 and Sr-90, but with very low dose contributions.

The breakthrough curve shows that annulus releases decline once the liners fail, followed by contributions from the tank CZ (e.g., Figure 14(a)). The dose contributions from Pu-239 at the 1-m boundary are generally orders of magnitude smaller than the dose contributions from Np-237. At the end of the 20,000-yr simulation, the Np-237 dose contributions are generally

~0.001 mSv/yr (~0.1 mrem/yr) and increasing by approximately an order of magnitude every 2,000 to 2,500 yrs.

(a)

(b)

Figure 14 AFZ analysis radionuclide dose contributions at the 1-m facility boundary for (a) Type I Tank 10 and (b) Type II Tank 15 with the complete fast-flow AFZ path (reproduced from Figures 6.7-8 and 6.7-13 of SRR-CWDA-2016-00078).

For these long-lived radionuclides, the peak dose-contribution is strongly tied to the liner failure time. If a gap exists beneath the liners due to the weld backup strips, an alternative AFZ similar to the Type II AFZ may be appropriate. The increased flow is likely to increase corrosion rates for the liners, which would cause the liners to fail sooner. Corrosion calculations for steel contacting groundwater indicate that the liners may fail as much as 10,000 yrs earlier (Table 15), which would initiate releases much earlier and may allow peak concentrations to become of concern. Furthermore, once the secondary liner fails, the AFZ model may have shorter pathways that pass through cementitious materials, which may reduce the influence of sorption on delaying releases for these radionuclides.

Type II tanks and Np-237 and Pu-239 doses The presented AFZ calculations for Type II tanks in Section 6.7.2.2 of the SA describe a variety of scenarios. In general, Np 237 arrives earlier than Pu-239 at the 1-m boundary but the Pu-239 peak dose-contribution is orders of magnitude larger. Calculated radionuclide breakthrough curves at the 100-m boundary for Tank 16 (Figure 6.7-15 of the SA) show peak dose contributions from Np-237 that are <0.001 mSv/yr (<0.1 mrem/yr) and no Pu-239 within the figure limits. Almost the entire inventory is initially in the AFZ for Tank 16; Tanks 13-15 have approximately an order of magnitude greater total inventory than Tank 16, but mostly in the primary, rather than in the annulus.

Both radionuclides sorb to the soil and cementitious materials, which slows transport and reduces dissolved concentrations. In the SA, both are represented as strongly sorbing in cementitious materials, but neptunium is represented as much less sorbing than plutonium in soil materials. Accordingly, the AFZ pathway has a roughly equivalent effect on both radionuclides within the tank, but Np-237 transports much more rapidly in the natural environment. Consideration of more mobile fractions of Pu in the natural system should also be considered.

Comparing the Tank 16 Pu-239 dose contributions for the 1-and 100-m boundaries, plutonium sorption in the soil dominates transport to the 100-m boundary relative to any influence of cementitious sorption. Even though the sorption coefficient is smaller in the soil, the path length in the natural system is orders of magnitude longer than the path length through cementitious material. Therefore, for Pu-239, the representation of the AFZ pathway is primarily important for the 1-m boundary within the 20,000 yr timeframe.

An alternative AFZ pathway through failed liners may significantly reduce Pu-239 contact with cementitious materials, increasing peak dose contributions at the 1-m boundary. The same alternative AFZ pathway would have a larger influence on Np-237 peak dose equivalent concentrations, because for neptunium the sorption coefficient is orders of magnitude larger for cementitious materials than for soil.

Summary of Findings Related to the Effect of Groundwater Flow on AFZ Dose Contributions The SA assumed that groundwater flow rates are steady within the AFZ and in the SZ. The observed variation in groundwater levels and inferred groundwater flow directions within the HTF vicinity may influence the AFZ analyses in various ways.

Representing the flow rates as steady and neglecting the influence of fluctuations on dispersion of released radionuclides would tend to increase peak concentrations at compliance points, so these modeling choices for representing the effects of groundwater flow on transport are conservative both within the tanks and in the groundwater system

The representation of the AFZ pathway is most risk significant for (i) rapidly decaying radionuclides that are also weakly sorbing (because of the competition between transport and decay) and (ii) radionuclides that sorb much more strongly to cementitious materials than the natural environment (because sorption to the cementitious materials is the transport-limiting step) 3.2.7 Sensitivity to Water-Table Fluctuations Water-table fluctuations have the potential to accelerate the release of radionuclides from the CZ of submerged and partially submerged Type I and II tanks. Historical observations suggest that water-table elevations have fluctuated by approximately 3 to 6 m (10 to 20 ft) within the HTF area. DOE considered a simplified model of the saturation of a cementitious cylinder (i.e.,

grout monolith), showed that the monolith would be expected to have a thick, fully saturated capillary fringe due to the air entry pressure of grout being 100 to 1000 m even after 20,000 years, and deduced that little to no water would drain from the cementitious structure pores during a typical water-table decline. DOE concluded that although radionuclides within the CZ would be subject to downward advective transport due to the percolation rate of meteoric water through the monolith, water-table decline would have little effect on release rates into the environment.

The DOE-provided basis for release of water from storage held in intact grout materials is reasonable. However, neither the analysis on AFZs nor the analysis on water-table fluctuations consider the influence of water-table fluctuations with respect to drainage from gaps in the grouted-tank system and cracks in the grout. For Type I tanks, changes in water-table gradients may have a more significant influence on release rates than changes in vertical fluxes because the CZs and all release pathways are likely to remain submerged during water-table fluctuations. For Type II tanks, however, observed fluctuations in the water-table elevation pass through the CZ and release pathway elevations. As described in the previous section, fluctuating water levels may allow pulses of higher radionuclide concentrations to be released, depending on the potential release pathway. Depending on the frequency of water table fluctuations and the saturated zone hydraulic gradient, the release rates associated with potential water table fluctuations for Type II tanks could be higher than release rates in the base case. NRC staff continues to recommend that DOE evaluate the impact of drainage of inventory from shrinkage gaps in the tank systems for Type II tanks within the zone of water table fluctuation due to water table rise and fall.

3.2.8 Individual Tank Sensitivity Analyses DOE evaluates the waste storage, transfer and retrieval history of each Type I and II tank in Chapter 6 of the SA and provides results of a sensitivity analysis on inventory, and other parameters such as iodine solubility limits, for these individual tanks. The results show that for certain cases (e.g, Cases C and E) and inventory multipliers, the performance objectives can be challenged earlier in time and different radionuclides may become more risk significant (e.g., Np-237, K-40, Cs-135, Tc-99). The individual tank histories and sensitivity analyses provide useful and interesting information. However, due to the limited evaluation of parameter space in deterministic sensitivity analysis, the results can be viewed as providing general risk insights, but should only be used semi-quantitatively to assess uncertainty in dose projections.

Furthermore, some non-intuitive results were presented without interpretation, making it difficult to draw definitive conclusions (e.g., lower I-129 doses associated with higher inventories in Cases C and E, perhaps due to greater attenuation in the concrete basemats early in the simulation period; higher Np-237 doses, perhaps due to basement bypass; similar or lower doses associated with different grout types [similar grouts] and use of oxidized grout [lower doses]). Although the non-intuitive results are likely valid, lack of transparency in simulation descriptions and interpretation of results limits the usefulness of the information provided.

Given the large number of simulations presented, a more comprehensive synthesis of key results would have strengthened analysis conclusions.

Summary of Findings While the results of the deterministic sensitivity analysis provide interesting information about the impact of single parameter uncertainty on peak dose, one-off analyses do not provide insights to more complex responses to changes in multiple parameters and impacts to the peak dose results. Results of probabilistic analyses are needed to better understand impacts of uncertainty over a larger range of parameter space (see review of DOEs probabilistic analysis).

Furthermore, interpretation of the results of the sensitivity analyses was lacking in some instances, and further interrogation of the results is necessary to better understand the cause of some non-intuitive results.

NRC staff recognize DOEs effort to study the impact of Pu natural system Kd on the results, given evidence of more mobile forms of Pu in the natural system, consistent with NRC staff recommendations in ML14094A514, ML12272A124, ML19280A059, and ML21119A316.

The sensitivity analyses reveal the importance of explicit consideration of multiple oxidation states of Pu on dose results, although more complex modeling of oxidation and reduction reactions along the flow pathway may be more realistic.

3.3 Probabilistic Assessment: Summary and Evaluation DOE conducted a probabilistic assessment using the GoldSim model. GoldSim stochastic and uncertainty modeling updates were discussed previously. The primary update was to the inventory, based on data collected from Tanks 12 and 16. The results of the probabilistic analysis show lower doses compared to the results presented in the HTF PA. DOE also compared the peak of the mean with the mean of the peaks as a method to evaluate the potential for risk dilution. The results of the mean of the peaks is about a factor of 5x higher than the results of the peak of the mean.

DOE used two different approaches to determine the total dose. Picking the highest dose for each realization, regardless of sector, and averaging those doses together to arrive at the total dose (more conservative), or the alternative total dose approach of averaging the realizations for each of 6 sectors (A-F) and picking the sector with the highest average for each time step.

A total of 3,000 realizations were run for Case A and another 3,000 realizations were run for all cases including Case A (6,000 cases total) at the assigned probabilities for the cases. Tables 16 and 17 provide mean of the peak and peak of the mean results and related statistics from the SA and HTF PA probabilistic analyses.

Table 16 Mean of the peak dose and other statistics (reproduced from Table 8.1-1 of SRR-CWDA-2016-00078)

Table 17 Peak of the mean and other statistics (reproduced from Table 8.1-2 of SRR-CWDA-2016-00078)

Seventy-two (72) flow cases discussed in Section 4.4.4. of the HTF PA were used in the probabilistic assessment, with the first 24 flow cases representing Case A. Expert judgment was used to screen-out a portion of the 2,539 uncertain parameters, arriving at a limited set of 1,040 parameters. The drinking water ingestion rate parameter distribution was changed from a triangular distribution to a truncated gamma distribution with a higher mean spanning a wider range of values (see Section B.1 of SRR-CWDA-2013-00058). Results are presented for a 1,000-, 10,000- and 20,000-yr period. Over the 20,000-yr period, the peak of the mean dose is 0.49 mSv/yr (49 mrem/yr), the peak of the median is 0.14 mSv/yr (14 mrem/yr), and the peak of the 95th percentile dose is 2.07 mSv/yr (207 mrem/yr). Dose results are presented for Tc-99, Np-237, Ra-226, I-129, and Sr-90. The next two highest dose contributors are Ni-59 and Pb-210.

Figure 15 illustrates the mean dose over a 10,000-yr period organized by key radionuclide for all cases. The risk associated with Ra-226 in-growth increases over time, and leads to the highest dose over the 10,000-yr period (dose is still increasing at 10,000 yrs), while the dose associated with Tc-99 is estimated to be higher than the dose of all other radionuclides and peaks earlier at approximately 2,000 yrs. Assumptions regarding the inventory multiplier for Tc-99 is probably the reason for this behavior (I-129 inventory may not have been varied). On the other hand, I-129 dose dominates Case A, as illustrated in Figure 16 and the dose from Tc-99 is significantly lower, suggesting that a fast pathway is necessary to realize the potential dose from Tc-99.

Figure 15 Mean dose contributions from select radionuclides, all cases probabilistic results (10,000 yrs). Image credit: Figure 8.4-2 of SRR-CWDA-2016-00078 Figure 16 Mean dose contributions from select radionuclides, Case A probabilistic results (10,000 yrs). Image credit: Figure 8.4-4 of SRR-CWDA-2016-00078 Various methods are used to perform probabilistic sensitivity analysis with the following parameters identified as important to various end points. The end points, however, are unclear and DOE does not explicitly indicate what end points are used in the tables, noting that GoldSim only considers certain end points available in the model, such as dose at 10,000 yrs and does not consider time steps that correspond to peak values. DOE should create a metric, such as peak dose, to study the importance of parameters to key end points. The top sensitivity indices (i.e., important parameters) are listed in Tables 8.5-1 and 8.5-2 of the SA (SRR-CWDA-2016-00078) and include the following:

Well depth (determines aquifer that a well is completed in)

Water-ingestion multiplier

Liner failure time for Tanks 32 and 35

Tank 15 inventory multiplier for Pu-238

Tank 10 inventory multiplier for I-129

Kd of Ra-226 in sandy soil While some of these parameters were identified as important in the HTF PA (e.g., well depth/aquifer), new parameters include the following:

Tc solubility in OR III

Inventory multipliers for certain radionuclides (e.g., Tc-99, Pa-231, U-232, U-235, and Am-243) and certain tanks (e.g., Tanks 10, 12, 35, and 39)

Till depth DOE presents histograms of numbers of realizations with peak dose in each sector, with Sector A having the greatest number of realizations with peak dose. Partial rank correlation coefficients (PRCCs) over time for various parameters are also presented for all cases, and Case A. For all cases, DOE indicates that, between 500 to 600 yrs post-closure, water ingestion and well depth are the most important parameters. Between 400 to 1,500 yrs post-closure, flow configurations for Type II tanks with initially failed liners (i.e., Tanks 14, 15, and 16) are important (higher flow configurations with continuous fast-flow path through the grout, CZ, and basemat). Between 1,600 and 2,800 yrs post-closure, Tank 12 flow configuration is important, with early releases being driven by Cases D and E full fast pathways. Early release, however, would drive down dose in later times, and so the PRCC actually goes from strongly positive to strongly negative. DOE also notes the importance of Ra Kd increasing between 3,000 to 5,000 yrs (negative correlation) and Tc OR_II solubility at approximately 15,000 yrs (strong negative correlation) near the time of transition to OR_III. Chapter 8 of the SA includes results for (i) the time dependence of parameters that are important to dose for Case A, (ii) various sectors, and (iii) certain radionuclides.

Of note, additional parameters that may be important are:

Darcy velocity in the SZ

Kd, solubility limits, and soil-to-plant uptake factors for radionuclides that are the focus of the sensitivity analysis (i.e., I-129, Ra-226, Np-237, Tc-99)

Flow configurations and unsaturated zone soil hydraulic conductivity curves for various tanks Finally, DOE studied the highest risk realizations in more detail in Section 8.6 of SRR-CWDA-2016-00078. Table 18 presents the 5 highest dose realization cases and associated key parameters.

After the well depth and water-ingestion multiplier, the tank-specific flow configuration variables are most important, if all cases (i.e., Cases A-E) are considered. DOE indicates that when fast-flow path configurations are sampled, particularly Case E with a complete flow path through the tank, annulus, and basemat, the resulting doses can be significant. Under Case E, Np-237 in particular can mobilize and lead to a significant dose within the first 5,000 yrs post-closure.

For Case A, the next-most important parameters, after well depth and water-ingestion multiplier, are the tank-specific liner failure times for Tanks 9, 10, and 11. Earlier steel liner failure times allow I-129 to become mobilized and produce a significant dose within the first 10,000 yrs.

DOE also found that inventory multipliers and parameters related to the release and transport of key radionuclides (e.g., the sandy soil Kd for radium) are important to dose.

Summary of Findings NRC finds that DOEs probabilistic sensitivity analysis study provides useful information about the potential for risk dilution, and interesting results related to the change in parameters of importance over time and for different configurations and radionuclides. DOE should continue to use probabilistic analysis to better understand uncertainty in the results, as well as to study the impact of changes in multiple parameters on peak dose. While additional work on parameter distribution development would be needed to use the results to demonstrate compliance with performance objectives, these analyses provide useful information regarding the importance of model parameters to dose. Obtaining additional information and refining modeling could help address uncertainty in areas of greatest risk significance. To improve the presentation of information, NRC staff suggests DOE could:

Better explain the dose metric used to identify the key parameters in Tables 8.5-1 and 8.5-2 of the SA

Develop a peak of the mean calculation that can be used as an end point in lieu of looking at dose over time; this approach would help streamline the probabilistic sensitivity analysis approach

Use the peak of the mean as an end point to study the importance of model parameters on dose

Evaluate the importance of model parameters on the timing of peak dose 4.0 Teleconference On July 1, 2021, the U.S. Nuclear Regulatory Commission (NRC) and U.S. Department of Energy (DOE) and their contractors held a teleconference call to discuss NRC questions on the Type I and II Tank SA GoldSim model. Specifically, NRC had questions about the GoldSim simulation of AFZ cases in the SA. The list of questions discussed and preliminary responses to the questions is available in a Note to File (ML21194A069). Following the teleconference call, DOE also provided example GoldSim files (SRR-CWDA-2021-00059) that were requested by NRC staff during the teleconference.

Table 18 Parameter values associated with realizations with highest peak doses within 10,000 yrs (all cases) (reproduced from Table 8.6-2 of SRR-CWDA-2016-00078) 5.0 Follow-Up Actions NRC staff will continue to monitor DOEs parameterization of final radionuclide inventories in SAs under Monitoring Factor 1.1, Final Inventory and Risk Estimates listed in NRC staffs plan for monitoring at the TFFs (ML15238A761).

Under Monitoring Factor 3.1, Hydraulic Performance of Concrete Vault and Annulus (As It Relates to Steel Liner Corrosion and Waste Release) and Monitoring Factor 3.5, Vault and Annulus Sorption, NRC staff will continue to monitor DOEs analysis of the potential release of radionuclides from the annulus and sand pads, including: (i) a continuous preferential pathway, (ii) a fluctuating water-table, and (iii) long-lived radionuclides. NRC staff will also evaluate DOEs implicit assumption that annulus waste and overlying grout are well-mixed.

Although not specifically listed under Monitoring Factor 4.1, Natural Attenuation of Key Radionuclides, NRC staff will monitor DOEs development of information to support the sorption of iodine in the natural environment based on its risk significance.

6.0 Conclusions The NRC staff concludes that:

1.

DOEs Special Analysis provides useful information on engineered and natural system performance.

2.

DOEs Special Analysis is an improvement over the HTF PA and associated special analyses providing updated inventory information and extensive uncertainty and sensitivity analysis.

3.

Many of NRC staffs previous findings are still appliable as noted in more detail in the preceding sections of this report.

4.

DOEs future PA documentation could be improved with additional information and analyses as indicated in Table 19 and discussed in more detail in the preceding sections of this report.

Table 19 Summary of Findings MF Description of Finding TRR Section 1

3.5 Technical basis for the cement iodine Kd sorption coefficient value in RR_II and explanation for lack of importance of iodine leachate impact factors on dose.

2.2, 3.2.3 2

4.1 Technical issues associated with iodine Kd sorption coefficient values for soils.

3.2.3 3

2.1 Technical support for iodine solubility values and explanation for lack of importance of iodine solubility in submerged region C on dose.

3.2.2 4

4.1 Technical basis for selection of Pu Kd values in soils.

3.2.3 5

2.1 and 3.5 Lack of sensitivity of Pu solubility and leachate impact factors on dose.

3.2.2 6

1.1, 1.2, and 1.3 Sampling and inventory development for tanks with waste in secondary containment.

3.1.1 7

1.1 Consideration of uncertainty in primary tank inventory.

3.1.1 8

1.4 Validation of ancillary equipment inventory.

3.1.2 9

1.4 Consideration of uncertainty in ancillary equipment inventory.

3.1.2 10 3.1, 3.2, 3.3, 6.1, and 6.2 Modeling treatment of the alternative fast zone (AFZ) case for Type I tanks (e.g., AFZ pathway under tanks).

3.2.6 11 1.1 Potential missed inventory under Type I tank backing strips at bottom plate welds.

3.2.6 12 3.1 Lack of consideration of impact of AFZ pathways on enhancing steel liner corrosion (Type I and II tanks).

3.2.6 13 6.1 Modeling treatment of AFZ case for Type II tanks (e.g., joining of primary and secondary sand pads after secondary liner failure).

3.2.6 14 6.1 Modeling treatment of AFZ case for Type I and II tanks (e.g., lack of consideration of diffusion, which may be important at early times; advection is the only transfer mechanism from the tank grout to the fast zone).

3.2.6 15 6.1 Simulation of release associated with water table rise and fall.

3.2.7 16 6.2 Need for more extensive interpretation of sensitivity analysis results, and technical issues associated with use of one-off analyses.

3.2 17 6.2 Recommendations for probabilistic analysis metrics and presentation of results.

3.3 In this report, there is no significant change to the NRC staffs overall conclusions from the F-and HTF Technical Evaluation Reports (TERs) regarding compliance of DOE disposal actions with the 10 CFR Part 61 performance objectives.

7.0 REFERENCES

MIL-STD-22 D. Design Criteria: Welded Joint Design. Department of Defense. 1979.

Accessed at <<http://www.everyspec.com>> on 14 June 2021.

ML12272A124. Barr, C.S. Technical Review of Environmental Monitoring and Site-Specific Distribution Coefficient Reports. Memorandum to Gregory Suber. Rockville, Maryland: U.S.

Nuclear Regulatory Commission. March 2015. (ADAMS Accession No. ML12272A124)

ML13100A230. Barr, C.S. Technical Review of Tank 18 and 19 Special Analysis for the Performance Assessment for the F-Tank Farm at the Savannah River Site. Memorandum to Gregory Suber. Rockville, Maryland: U.S. Nuclear Regulatory Commission. April 2013.

(ADAMS Accession No. ML13100A230)

ML13085A291. Barr, C.S. et al. Technical Review of Final Inventory Documentation for Tanks 5 and 6 at F-Tank Farm, Savannah River Site (Project No. PROJ0734). Memorandum to Gregory Suber. Rockville, Maryland: U.S. Nuclear Regulatory Commission. 30 September 2013. (ADAMS Accession No. ML13085A291)

ML13273A299. Barr, C.S. Technical Review of Tanks 5 and 6 Special Analysis at F-Tank Farm Facility, Savannah River Site. Memorandum to Gregory Suber. Rockville, Maryland:

U.S. Nuclear Regulatory Commission. September 2013. (ADAMS Accession No. ML13273A299)

ML14094A514. Technical Evaluation Report for Draft Waste Determination for Savannah River Site H-Area Tank Farm. Rockville, Maryland: U.S. Nuclear Regulatory Commission.

June 2014. (ADAMS Accession No. ML14094A514)

ML15301A710. Alexander, G. et al. Technical Review of Special Analysis for Tank 16H.

Rockville, Maryland: U.S. Nuclear Regulatory Commission. October 2015. (ADAMS Accession No. ML15301A710)

ML15238A761. U.S. Nuclear Regulatory Commission Plan for Monitoring Disposal Actions Taken by the U.S. Department of Energy at the Savannah River Site F-Area and H-Area Tank Farm Facilities in Accordance with the National Defense Authorization Act for Fiscal Year 2005.

Rockville, Maryland: U.S. Nuclear Regulatory Commission. October 2015. (ADAMS Accession No. ML15238A761)

ML15301A830. Barr, C.S. & L. Parks. Technical Review of Final Inventory Documentation for Tank 16H at Savannah River Site (Docket No. PROJ0734). Memorandum to Douglas Mandeville. Washington, DC: U.S. Nuclear Regulatory Commission. 30 October 2015.

(ADAMS Accession No. ML15301A830)

ML16231A444. Barr, C.S. Technical Review: U.S. Department of Energy Documentation Related to Tanks 16H and 12H Grouting Operations with Emphases on Specifications, Testing, Recommendations and Placement Procedures (Project No. PRO0734). Memorandum to Richard Chang. Washington, DC: U.S. Nuclear Regulatory Commission. 6 September 2016.

(ADAMS Accession No. ML16231A444)

ML16277A060. Grossman, C. & K. Pinkston. Technical Review of Dose Calculation Methodology for Liquid Waste Performance Assessments at the Savannah River Site (SRR-CWDA-2013-00058, Rev. 1, July 2014, Docket No. PROJ0734). Memorandum to Gregory Suber. Washington, DC: U.S. Nuclear Regulatory Commission. 23 December 2016.

(ADAMS Accession No. ML16277A060)

ML16342C575. Pinkston, K. Technical Review of the Sorption of Iodine at the Saltstone Disposal Facility at the Savannah River Site (Docket No. PROJ0734). Washington, DC:

U.S. Nuclear Regulatory Commission. 5 January 2017. (ADAMS Accession No. ML16342C575)

ML17277B235. Barr, C.S. Technical Review of Final Inventory and Special Analysis Documentation for Tank 12H at Savannah River Site (Docket No. PROJ0734). Memorandum to Gregory Suber. Washington, DC: U.S. Nuclear Regulatory Commission. 8 January 2018.

(ADAMS Accession No. ML17277B235)

ML18242A259. Barr, C.S. Technical Review of Real Waste Release Testing Results for Tank 18F and Associated Performance Assessment Documentation for F-Area and H-Area Tank Farm Facilities at Savannah River Site, Aiken, South Carolina (Docket No. PROJ0734).

Memorandum to Richard Chang. Washington, DC: U.S. Nuclear Regulatory Commission.

21 September 2018. (ADAMS Accession No. ML18242A259)

ML19280A059. Barr, C.S. Technical Review of Environmental Monitoring Reports for F-Area and H-Area Tank Farm Facilities (Docket No. PROJ0734). Memorandum to Stephen S.

Koenick. Washington, DC: U.S. Nuclear Regulatory Commission. 17 December 2019.

(ADAMS Accession No. ML19280A059)

ML19298A092. Barr, C. et al. Technical Review of Tank 12H Waste Release Reports and Associated Impact Analysis. Memorandum to Stephen S. Koenick. Washington, DC:

U.S. Nuclear Regulatory Commission. 20 December 2019. (ADAMS Accession No. ML19298A092)

ML20254A003. Preliminary Review of the U.S. Department of Energys Submittal of the 2020 Savannah River Site Saltstone Disposal Facility Performance Assessment (

Enclosure:

Request for Supplemental Information for the 2020 U.S. Department of Energy Savannah River Site Saltstone Disposal Facility Performance Assessment). Washington, DC: U.S. Nuclear Regulatory Commission. 5 October 2020. (ADAMS Accession No. ML20254A003)

ML20296A550. Barr, C. Technical Review of Documents Related to Tank 12H Grout Formulations, Grout Testing, Procedures, and Grouting Operations at the H-Area Tank Farm at Savannah River Site (Supplement to Technical Review Report on Tank 16H and 12H

[ML16231A444]). Memorandum to Stephen Koenick. Washington, DC: U.S. Nuclear Regulatory Commission. 18 November 2020. (ADAMS Accession No. ML20296A550)

ML21119A316. Barr, C. Technical Review: U.S. Department of Energy Documentation Related to Environmental Monitoring And Natural Attenuation of Key Radionuclides at the F-Tank Farm And H-Tank Farm Facilities at the Savannah River Site (Project No. PROJ0734).

Memorandum to Stephen Koenick. Washington, DC: U.S. Nuclear Regulatory Commission.

14 May 2021. (ADAMS Accession Nos. ML21119A316 and ML21119A317)

Neeway, J.J, D.I. Kaplan, C.E. Bagwell, M.L. Rockhold, J.E. Szecsody, M.J. Truex, and N.P. Qafoku. A Review of the Behavior of Radioiodine in the Subsurface at Two DOE Sites.

Science of the Total Environment. Vol. 691. pp. 466-475. 2019. doi:

10.1016/j.scitotenv.2019.07.146.

Otosaka, S., K.A. Schwehr, D.I. Kaplan, K.A. Roberts, S. Zhang, C. Xu, S.-P. Li, Y.-F. Ho, R. Brinkmeyer, C.M. Yeager, and P.H. Santschi. Transformation and Transport Processes of 127I and 129I Species in an Acidic Groundwater Plume at the Savannah River Site. Science of the Total Environment. Vol. 409. pp. 3,857-3,865. 2011. doi:

10.1016/j.scitotenv.2011.05.018.

Q-SQP-A-00008. G.P. Flach and B.T. Butcher. Software Quality Assurance Plan for PORFLOW Flow-Field Extraction Tool (GoldSimFlows). Revision 0. Aiken, South Carolina:

Savannah River National Laboratory. June 2013. [ADAMS Accession No. ML14007A669]

Santschi, P.H., C. Xu, S. Zhang, K.A. Schwehr, R. Grandbois, D.I. Kaplan, and C.M. Yeager.

Iodine and Plutonium Association with Natural Organic Matter: A Review of Recent Advances.

Applied Geochemistry. Vol. 85. pp. 121-127. 2017. doi: 10.1016/j.apgeochem.2016.11.009.

SREL-R-21-0001. J.C. Seaman et al. Aqueous and Solid Phase Characterization of Potential Tank Fill Materials. Aiken, South Carolina: The University of Georgia Savannah River Ecology Laboratory. 20 August 2020. [ADAMS Accession No. ML20303A339]

SRNL-ESB-2008-00023. M.A. Phifer. H-Area Tank Farm Closure Cap and Infiltration.

Revision 0. Aiken, South Carolina: Savannah River National Laboratory. 21 May 2008.

SRNL-L3200-2015-00008. G.P. Flach. Nominal Water Table Gradient in the Vicinity of the H-Area Tank Farm. Aiken, South Carolina: Savannah River National Laboratory.

29 January 2015. [ADAMS Accession No. ML16057A521]

SRNL-RP-2015-01008. G. A. Taylor. Task Technical and Quality Assurance Plan for H-Area Tank Farm FY2016 Special Analysis Model Support. Revision 0. Aiken, South Carolina:

Savannah River National Laboratory. December 2015.

SRNL-STI-2010-00047. B.L. Garcia Diaz. Life Estimation of High Level waste Tank Steel for H tank Farm Closure Performance Assessment. Revision 0. Aiken, South Carolina:

Savannah River National Laboratory. March 2010. (ADAMS Accession No. ML111220407)

SRNL-STI-2010-00148. W.E. Jones et al. Hydrogeologic Data Summary in Support of the H-Area Tank Farm Performance Assessment. Revision 0. Aiken, South Carolina: Savannah River National Laboratory. February 2015. (ADAMS Accession No. ML111220411)

SRNL-STI-2015-00339. M.E. Denham. Potential Solubility Controls for 129I in Residual Tank Waste. Revision 0. Aiken, South Carolina: Savannah River National Laboratory. July 2015.

(ADAMS Accession No. ML19318F721)

SRNL-STI-2015-00351. R.A. Hiergesell. et al. General Separations Area (GSA) Groundwater Model Calibration Targets. Revision 0. Aiken, SC: Savannah River National Laboratory.

August 2015. (ADAMS Accession No. ML18107A071).

SRNL-STI-2016-00224. G.A. Taylor and T. Hang. H-Area Tank Farm FY2016 Special Analysis Model Support. Revision 0. Aiken, South Carolina: Savannah River National Laboratory. May 2016. (ADAMS Accession No. ML20337A278)

SRNL-STI-2017-00445. G.P. Flach and T. Hang. Impacts of Updated GSA Groundwater Flow Models on the FTF, HTF and SDF PAs. Revision 0. Aiken, South Carolina: Savannah River National Laboratory. September 2017. (ADAMS Accession No. ML18081A308)

SRNL-STI-2018-00484. W.D. King. Determining the Release of Radionuclides from SRS Tank 12H Waste Residual Solids Following Tank Closure. Revision 1. Aiken, South Carolina:

Savannah River National Laboratory. October 2018. (ADAMS Accession No. ML19024A505)

SRNL-STI-2018-00643. G.P. Flach. Updated Groundwater Flow Simulations of the Savannah River Site General Separations Area. Revision 0. Aiken, South Carolina: Savannah River National Laboratory. 15 January 2019. (ADAMS Accession No. ML19053A383)

SRRA107772-000009. C.H. Benson and J.M. Benavides. Predicting Long-term Percolation from the SDF Closure Cap. Charlottesville, Virginia: University of Virginia School of Engineering (Report No. GENV-18-05). 23 April 2018. (ADAMS Accession No. ML18215A277)

SRRA021685-000013. Parker, C. et al. Determination of Constituent Concentrations in Field Lysimeter Effluents (FY19 Report). Revision A. Clemson, South Carolina: Clemson University. 7 August 2020. (ADAMS Accession No. ML20303A343)

SRR-CWDA-2010-00023. H-Area Tank Farm Closure Inventory for Use in Performance Assessment Modeling. Revision 4. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. 11 November 2014.

SRR-CWDA-2010-00023. H-Area Tank Farm Closure Inventory for Use in Performance Assessment Modeling. Revision 6. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. 13 January 2016. (ADAMS Accession No. ML20206L195)

SRR-CWDA-2010-00128. Performance Assessment for the H-Area Tank Farm at the Savannah River Site. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. November 2012. (ADAMS Accession No. ML13045A499)

SRR-CWDA-2012-00027. Tank 5 Inventory Determination. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

15 August 2012. (ADAMS Accession No. ML21180A447)

SRR-CWDA-2012-00070. Performance Assessment for the H-Area Tank Farm at the Savannah River Site: Quality Assurance Report. Revision 0. Aiken, South Carolina:

Savannah River Remediation, LLC. 30 August 2012. (ADAMS Accession No. ML13078A208)

SRR-CWDA-2012-00075. Tank 6 Inventory Determination. Revision 0. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

August 2012. (ADAMS Accession No. ML21180A449)

SRR-CWDA-2012-00075. Tank 6 Inventory Determination. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. August 2012.

SRR-CWDA-2013-00058. Dose Calculation Methodology for Liquid Waste Performance Assessments at the Savannah River Site. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. July 2014. (ADAMS Accession No. ML16167A295)

SRR-CWDA-2014-00060. B. Lester. Updates to the H-Area Tank Farm Stochastic Fate and Transport Model. Revision 2. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. July 2016. (ADAMS Accession No. ML21035A063)

SRR-CWDA-2014-00106. Tank 16 Special Analysis for the Performance Assessment for the H-Tank Farm at the Savannah River Site. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. 27 February 2015. (ADAMS Accession No. ML15058A397)

SRR-CWDA-2014-00134. Tank 16 Special Analysis for the Performance Assessment for the H-Area Tank Farm at the Savannah River Site: Quality Assurance Report. Revision 0. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

2 February 2015.

SRR-CWDA-2015-00075. Tank 12 Inventory Determination. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

August 2015. (ADAMS Accession No. ML21180A453)

SRR-CWDA-2015-00158. H-Tank Farm Type I and Type II Tank Special Analysis Base Case Model Inputs (Interoffice Memorandum). Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. 4 March 2016. (ADAMS Accession No. ML21176A198)

SRR-CWDA-2015-00166. D.B. Dixon. Updated Assigned Radionuclide and Chemical Inventories in Tanks 9, 10, 11, 13, 14, and 15. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. January 2016. (ADAMS Accession No. ML21209A516)

SRR-CWDA-2016-00061. Mangold, J.E. Evaluation of the Probabilistic Distribution used for Modeling Strontium Sorption to Sandy Soils. Revision 0. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. May 2016.

SRR-CWDA-2016-00078. Type I and II Tanks Special Analysis for the Performance Assessment for the H-Tank Farm at the Savannah River Site. Revision 0. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

August 2016. (ADAMS Accession No. ML19339E894)

SRR-CWDA-2016-00086. Evaluation of Waste Release Testing Results Against the Tank Farm Performance Assessment Waste Release Model. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

August 2016. (ADAMS Accession No. ML19024A494)

SRR-CWDA-2019-00104. Strategy for Updating the SRS Tank Farm Performance Assessments. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. December 2019.

SRR-CWDA-2020-00011. Industrial Wastewater Closure Module for F-Area Diversion Boxes 5 and 6, F-Area Tank Farm Savannah River Site, Industrial Wastewater Construction Permit No.

17,424-IW. Revision 0. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. February 2021. (ADAMS Accession No. ML21181A394)

SRR-CWDA-2020-00018. Savannah River Site Liquid Waste Facilities Performance Assessment Maintenance Program - FY2020. Revision 1. Aiken, South Carolina:

Savannah River Remediation, LLC, Closure and Waste Disposal Authority. August 2020.

(ADAMS Accession No. ML20303A344)

SRR-CWDA-2020-00029. Inventory Assignment at Closure for FDB-5 and FDB-6.

Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. 23 February 2021. (ADAMS Accession No. ML21180A428)

SRR-CWDA-2020-00055. FDB-5 and FDB-6 Special Analysis for the Performance Assessment for the F-Tank Farm at the Savannah River Site. Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

23 February 2021. (ADAMS Accession No. ML21180A433)

SRR-CWDA-2020-00061. G.P. Flach. Application of Characterization of the Aqueous and Solid Phase Chemistry of Closure Grouts (Interoffice Memorandum). Revision 0. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority.

25 August 2020. (ADAMS Accession No. ML20303A345)

SRR-CWDA-2021-00024. Savannah River Site Liquid Waste Facilities Performance Assessment Maintenance Program - FY2021. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. May 2021. (ADAMS Accession No. ML21148A005)

SRR-CWDA-2021-00059. Dataset for the Type I and II Tanks Special Analysis Modelling for H-Area Tank Farm. Aiken, South Carolina: Savannah River Remediation, LLC, Closure and Waste Disposal Authority. July 2021. (ADAMS Accession No. ML21190A193)

SRR-ESH-2021-00012. Status of F/H Area Radioactive Liquid Waste Tanks Being Removed from Service. (CY2020 Annual Report). Revision 0. Aiken, South Carolina: Savannah River Remediation. March 2021.

SRR-UWMQE-2017-00005. UWMQE to Evaluate Impacts to FTF PA Doses Due to the Update of the GSA Model. Revision 0. Aiken, South Carolina: Savannah River Site.

9 November 2017. (ADAMS Accession No. ML18081A327)

SRR-UWMQE-2017-00005. UWMQE to Evaluate Impacts to FTF PA Doses Due to the Update of the GSA Model. Revision 1. Aiken, South Carolina: Savannah River Site. FY2019.

SRR-UWMQE-2017-00006. UWMQE to Evaluate Impacts to HTF PA Doses Due to the Update of the GSA Model. Revision 0. Aiken, South Carolina: Savannah River Site.

9 November 2017. (ADAMS Accession No. ML18081A658)

SRR-UWMQE-2017-00006. UWMQE to Evaluate Impacts to HTF PA Doses Due to the Update of the GSA Model. Revision 1. Aiken, South Carolina: Savannah River Site. FY2019.

W145225. 200 Area Waste Storage Tanks - 241 F & H, Design of Concrete Tank Concrete (Construction Drawing). Revision 4. Aiken, South Carolina: Savannah River Site.

7 July 1954. (ADAMS Accession No. ML111230202)

W145367. 200 Area Waste Storage Tanks - 241 F & H Steel Pan Plate Details Steel.

Revision 1 (Construction Drawing). Aiken, South Carolina: Savannah River Site. 7 July 1954.

(ADAMS Accession No. ML111230210)

W145379. 200 Area Type I Tanks 1-8 and 9-12 Waste Storage Tanks 241 F & H 75-0 Dia.

Steel Tank Details, Steel (Construction Drawing). Revision 4. Aiken, South Carolina:

Savannah River Site. April 2004. [ADAMS Accession No. ML111230239].

W162672. 200 Area-Type II Tanks 13-16, Waste Storage Tanks 241-H, 85-0 Dia. Steel Tank Details Steel (Construction Drawing). Revision 30. Aiken, South Carolina: Savannah River Site. 6 April 2004. (ADAMS Accession No. ML111230594)

W163018. 200 Area - Bldg. 241H, 85-0 Dia. Waste Storage Tanks General Arrangement &

Construction Details Concrete & Steel (Construction Drawing). Revision 28. Aiken, South Carolina: Savannah River Site. 7 November 1955. (ADAMS Accession No. ML111230605)

WSRC-STI-2007-00184. M.A. Phifer et al. FTF Closure Cap Concept and Infiltration Estimates. Revision 2. Aiken, South Carolina: Savannah River National Laboratory.

October 2007.

Yeager, C.M., S. Amachix, R. Grandbois, D.I. Kaplan, C. Xu, K.A. Schwehr, and P.H. Santschi.

Microbial Transformation of Iodine: From Radioisotopes to Iodine Deficiency. Chapter 3 in Advances in Applied Microbiology. Vol. 101. pp 83-136. 2017.

Appendix A Review Status of Documents:

SRR-CWDA-2010-00023. H-Area Tank Farm Closure Inventory for Use in Performance Assessment Modeling. Revision 6. Aiken, South Carolina: Savannah River Remediation, LLC. January 2016. (ADAMS Accession No. ML20206L195)

The purpose of this document is to report updated inventories of radiological and chemical constituents in the residual material in the 29 HTF tanks and ancillary equipment at the HTF closure date of 2032 to support SA and PA modeling. Revision 6 updates the primary tank radionuclide and chemical inventory values for Tanks 9-11 and Tanks 13-15 for K-40, Co, Cu, SO4, and Sr, per a revision to reference document SRR-CWDA-2015-00166. These inventory updates were not included in the HTF modeling for the HTF Type I and II Tanks SA, but will be included in subsequent PA modeling. NRC staff reviewed Revision 6 in support of this TRR.

SRR-CWDA-2014-00060. Updates to the H-Area Tank Farm Stochastic Fate and Transport Model. Revision 2. Aiken, South Carolina: Savannah River Remediation, LLC. July 2016.

[ADAMS Accession No. ML21035A063]

This purpose of this report revision is to describe changes to the HTF radionuclide stochastic fate and transport model, which was developed using GoldSim to support the HTF PA, the Tank 16 SA, and the Tank 12 SA and related decision-making. In preparation for use of the GoldSim radionuclide advective-dispersive transport model to support the HTF Types I and II Tanks SA, several updates to the GoldSim model were implemented. Updates to the GoldSim

v. 10.5, HTF_Transport_Model_v4.000_Rad radionuclide transport model, described in Section 2 of the report, include: (i) inventory values for Tanks 9, 10, 11, 13, 14, and 15; (ii) stochastic distributions used for inventories for Tanks 13, 14, and 15; (iii) iodine (I) Kd and strontium (Sr) Kd distribution; (iv) addition of new points of assessment/well locations along the 1-m facility boundary to better replicate inadvertent human intruder doses simulated by the HTF PORFLOW model; (v) the structure of the annulus CZ, reflecting changes in the PORFLOW model and the set of PORFLOW-generated diffusion coefficient input files; (vi) a new set of basecase PORFLOW output-generated flowfield files for input to GoldSim that include horizontal flow in the annulus CZ, and a new set of PORFLOW output-generated flowfield files that also include horizontal flow in the annulus for alternate configurations Case B, Case C, Case D, and Case E to enable benchmarking between PORFLOW and GoldSim for these AFZ cases; (vii) addition of a set of model scenarios to evaluate the sensitivity of the modeled system to different quality grouts; and (viii) distributions for source-specific SZ Darcy velocities.

Section 3 of the report documents benchmark testing of the stochastic GoldSim radionuclide transport model to demonstrate that it represents valid abstractions of the processes simulated by the deterministic HTF PORFLOW model.

SRR-CWDA-2015-00158. H-Tank Farm Type I and Type II Tank Special Analysis Base Case Model Inputs (Interoffice Memorandum). Revision 1. Aiken, South Carolina: Savannah River Remediation, LLC. 4 March 2016. (ADAMS Accession No. ML21176A198)

The purpose of this memo was to document proposed HTF base case model input revisions for incorporation into the HTF conceptual model. Individual inputs identified in the Dose Calculator document (SRR-CWDA-2013-00058, Revision 1) were considered and reference documents were evaluated to identify any dose calculation inputs that should be revised; however, no dose calculation inputs were identified that required revision. Additionally, the Savannah River Site Liquid Waste Facilities Performance Assessment Maintenance Program - FY2016 Implementation Plan (SRR-CWDA-2015-00152, Revision 0) was reviewed to identify inputs that should be revised. Parameters and models considered were (i) inventories at closure; (ii) soil and cementitious material distribution coefficients; (iii) liner and cementitious material degradation modeling; (iv) solubility and transition-time modeling; (v) grout hydraulic conductivity, and (vi) closure cap modeling. Revisions to the HTF Type I and Type II tank inventories were the primary changes proposed for the HTF conceptual model (SRR-CWDA-2010-00023, Revision 6). Additionally, updating the Reduced Region II cement Kd value from 9 mL/g to 2 mL/g for iodine was proposed, which conservatively allows more rapid transport of iodine through cementitious materials and addresses an NRC-expressed concern about the uncertainty associated with the value assumed in the HTF PA (ADAMS Accession No. ML15301A710).

SRR-CWDA-2020-00018. Savannah River Site Liquid Waste Facilities Performance Assessment Maintenance Program - FY2020. Revision 1. Aiken, South Carolina:

Savannah River Remediation, LLC. August 2020. (ADAMS Accession No. ML20303A344)

The purpose of this document is to provide an annual update for the SRS Liquid Waste PA maintenance program, including activities completed during the last fiscal year (FY) or earlier, those planned for the current FY, and those anticipated as out-year activities. PA maintenance consists of activities performed on an ongoing or annual basis. The PA maintenance programs purpose is to confirm the adequacy of the current PA and SAs and increase confidence in inputs, assumptions, and results of analyses. SAs are performed to evaluate the impact of new information or new analytical methods on results and conclusions of a given PA. No new SAs are planned through the end of FY 2024, during the period when PA revisions are the focus of related efforts. Elements of the PA maintenance program are (i) testing and applied research; (ii) monitoring, (iii) unreviewed waste management questions/question evaluations, (iv) special analyses, and (v) PA revisions. Two unreviewed waste management question evaluations (UWMQEs) were completed during FY 2019 (SRR-UWMQE-2017-00005, Revision 1 for the FTF PA, and SRR-UWMQE-2017-00006, Revision 1 for the HTF PA) to evaluate the doses resulting from the 2018 model update of the General Separations Area PORFLOW database (SRNL-STI-2018-00643). The two evaluations concluded that performance objectives continued to be met, requiring no operational or design changes. Section 3.2 of the report addresses out-year HTF and FTF PA revisions, scheduled to be completed during FY 2022 and 2024, respectively. A document describing the activities and data needed to revise the FTF and HTF PAs was issued in December 2019 (SRR-CWDA-2019-00104, Strategy for Updating the SRS Tank Farm Performance Assessments). The PA revisions will include analyses and results from all SAs and UWMQEs completed by the date of their preparation, consideration of Low-Level Waste Disposal Facility Federal Review Group feedback, NRC recommendations, and new information generated by research and development, changes to site future land use plans/closure plans, changes to PA guidance documents requirements, and updated modeling improvements (e.g., SRR-CWDA-2012-00070). Section 3.3 of the report addresses PA-related testing and research activities applicable to both the HTF and FTF sites, including (i) tank residual characterization, waste release testing, and waste tank water chemistry dynamics testing; (ii) CLSM testing; and (iii) tank farm closure cap design. Tank 18F and 12H waste release solubility data for Np, Pu, Tc, U and I (SRR-CWDA-2016-00086, Revision 1; SRNL-STI-2018-00484, Revision 1) should be integrated into revisions of the tank farm Waste Release Models for improved realism. Tank grout water-conditioning information should be developed for all proposed alternative grout types intended for use during tank farm closure to estimate how pH and Eh of closure grout pore water and grout mineralogy would evolve through time, and the anticipated performance of proposed alternative grout formulas, including CLSM. Both oxic and anoxic dynamic grout column experiments simulating meteoric water infiltration and interaction with proposed tank grout variants were undertaken by SREL, along with static, batch grout-water interaction experiments (SRR-CWDA-2020-00018; SREL-R-21-0001; SRR-CWDA-2020-00061). It was anticipated that column and batch experiments would be monitored for a minimum of 20 weeks. Results of such experiments are anticipated to be used in Geochemists Workbench simulations to improve existing definitions of tank grout chemistry conditions after tank farm closure, to refine Waste Release Models, and reduce uncertainty in PA models. Tank grout testing anticipated to be undertaken during FY 2020 was intended to yield: (i) information about grout attributes that affect performance (e.g., slump flow, bleed water); (ii) definition of performance metrics and associated requirements and goals; (iii) CLSM characterization needed to support PA revisions; (iv) an assessment of the pros and cons of reducing tank grout LP#8-16 and candidate CLSM mixes on an attribute-by-attribute basis, and (v) recommendations of next steps to be taken in the selection process for a bulk fill grout for the next tank closure (i.e., Tank 15H). It remains unclear whether the intended test technical report and SRR Waste Disposal Authority evaluation associated with the planned CLSM testing was completed as scheduled, because NRC has not received related documentation. Finally, the tank farm engineered closure cap concepts used in the initial PAs (WSRC-STI-2007-00184 for FTF and SRNL-ESB-2008-00023 for HTF) are being updated from having a 2 percent slope to having a 4 percent slope in the revised PAs to comply with South Carolina Department of Health and Environmental Control requirements of 3 to 5 percent slopes to promote drainage.

Additionally, lessons learned about high density polyethylene/geosynthetic clay liner degradation from a SDF closure cap document (i.e., SRRA107772-000009) should be incorporated into the updated tank farm closure cap concepts. PA-related testing and research activities during FY 2021 and beyond had not yet been defined at the time of report completion.

Some SDF PA-related testing and research activities will also inform the tank farm PAs, especially those that address site-specific soil Kd values for Pu, Ra, Cs, Co, Ba, and Eu, retardation coefficient for Np, and evidence for or against colloidal radionuclide transport of Pu, as well as cementitious material degradation and crack formation (see Section 2.3.2 of the report). Three new lysimeters containing new radionuclide sources (i.e., a radium source and two saltstone pucks spiked with iodine) were planned to be installed during FY 2020. Lysimeter leachate reports are prepared annually through the completion date, which is set for FY 2030.

A literature search is planned to improve understanding of the potential degradation of cementitious materials exposed to radiation, but no completion date for this work has been set, and no funding has been assigned to the effort through 2024.

SRNL-STI-2016-00224. G.A. Taylor and T. Hang. H-Area Tank Farm FY2016 Special Analysis Model Support. Revision 0. Aiken, South Carolina: Savannah River National Laboratory. May 2016. (ADAMS Accession No. ML20337A278)

This brief report summarizes PORFLOW fate and transport modeling analyses to support an HTF SA for Type I and Type II waste tanks. The PORFLOW model was revised from the Tank 16 SA model (SRNL-STI-2014-00612) with new inventories; I-129 Kd changed to 2.0 mL/g for middle-aged reducing grout; and vadose zone simulations for Cases B through E were updated to include a 0.5-in-and 1.0-in-thick annular CZ in the annulus of Type I and II tanks, similar to what had been done previously for Case A in the Tank 16 SA.

Six sensitivity analyses, extending to 20 ky, were performed based on the base case scenario (Case A), but with variable tank grout hydraulic conductivities representative of possible hydraulic performance scenarios. Study 7 of the sensitivity study was identical to Study 1, and thus was not repeated.

Vadose zone PORFLOW flow fields and related information, including Darcy velocities, volumetric fluxes, flowrates, saturations, effective diffusion coefficient time series for each material, and Eh and pH transition times were extracted using the GoldSimFlow program (Q-SQP-A-00008) for use in HTF GoldSim models (SRR-CWDA-2014-00106).

A Task Technical and Quality Assurance Plan (SRNL-RP-2015-01008) for this work was issued, and PORFLOW analyses were design-checked (SRNL-L3200-2016-00044).