ML083170346

Response to Request for Additional Information Related to NRC Generic Letter 2004-02, Potential Impact of Debris Blockage on Emergency Recirculation During Design Basis Accidents at Pressurized-Water Reactors

ML083170346
Person / Time
Site:	Three Mile Island
Issue date:	11/10/2008
From:	Cowan P B AmerGen Energy Co
To:	Document Control Desk, Office of Nuclear Reactor Regulation
References
5928-08-20203, GL-04-002
Download: ML083170346 (45)

Text

10 CFR 50.54(f)5928-08-20203 November 10, 2008 United States Nucle.ar Regulatory Commission ATTN: Document Control Desk Washington, DC 20555-0001 Three Mile Island Nuclear Station, Unit 1 Facility Operating License No.DPR-50 NRC Docket No.50-289

Subject:

Three Mile Island Unit 1 Response to Request for Additional Information Related to NRC Generic Letter 2004-02,"Potential Impact of Debris Blockage on Emergency Recirculation during Design Basis Accidents at Pressurized-Water Reactors"

References:

(1)Generic Letter 2004-02,"Potential Impact of Debris Blockage on Emergency Recirculation During Design Basis Accidents at Water Reactors," dated September 13, 2004 (2)Letter from K.R.Jury (Exelon Generation Company, LLC and AmerGen Energy Company, LLC)to U.S.Nuclear Regulatory Commission"Exelon/AmerGen Response to NRC Generic Letter 2004-02, Potential Impact of Debris Blockage on Emergency Recirculation During Design Basis Accidents at Pressurized-Water Reactors," dated March 7, 2005 (3)Letter from P.B.Cowan (Exelon Generation Company, LLC and AmerGen Energy Company, LLC)to U.S.Nuclear Regulatory Commission"Exelon/AmerGen Response to NRC GenericLetter2004-02, Potential Impact of Debris Blockage on Emergency Recirculation During Design Basis Accidents at Pressurized-Water Reactors," dated September 1,2005 (4)Three Mile Island Unit 1 Supplemental Response to NRC Generic Letter 2004-02,"Potential Impact of Debris Blockage on Emergency Recirculation during Design Basis Accidents at Pressurized-Water Reactors," dated December 28, 2007 (5)Letter from P.B.Cowan (Exelon Generation Company, LLC and AmerGen Energy Company, LLC)to U.S.Nuclear Regulatory Commission"Response to Request for Additional Information Regarding NRC Generic Letter 2004-02, Potential Impact of Debris Blockage on Emergency Recirculation During Design Basis Accidents at Pressurized-Water Reactors," dated July 27,2005 (6)Letter from P.Bamford (U.S.Nuclear Regulatory Commission) to C.G.Pardee (AmerGen Energy Company, LLC),"Three Mile Island Nuclear Power Station, Unit 1-Request for Additional Information Related to Generic Letter 2004-02," dated August 12, 2008 U.S.Nuclear Regulatory Commission November 10, 2008 Page 2 The U.S.Nuclear Regulatory Commission (USNRC)issued Generic Letter (GL)2004-02 (Reference 1)on September 13, 2004, requesting that addressees perform an evaluation of the emergency core cooling system (ECCS)and building spray system (BSS)recirculation functions in light of the information provided in the GL and, if appropriate, take additional actions to ensure system function.Additionally, the GL requested addressees to provide the USNRC with a written response in accordance with 10 CFR 50.54(f).The request was based on identified potential susceptibility of the pressurized water reactor recirculation sump screens to debris blockage during design basis accidents requiring recirculation operation of ECCS or BSS and on the potential for additional adverse effects due to debris blockage of flowpaths necessary for ECCS and BSS recirculation and containment drainage.Reference 2 provided the initial AmerGen Energy Company, LLC (AmerGen)response to the GL followed by supplemental responses in References 3 and 4.Reference 5 responded to a request for additional information regarding the Reference 2 response to the GL.During the review of the Reference 4 submittal, the NRC identified various issues that require clarification (Reference 6).The AmerGen responses to these questions are provided in Attachment 1 to this letter.This information is being provided in accordance with 10 CFR 50.54(f)There is one regulatory commitment provided in this submittal.

The commitment states that TMI Unit 1 will report how it has addressed the in-vessel downstream effects issue within 90 days of issuance of the final NRC staff SE on WCAP-16793.

If you have any questions or require additional information, please contact Wendi Croft at (610)765-5726.I declare under penalty of perjury that the foregoing is true and correct.Executed on the 10 th day of November 2008.Respectfully,Director-Licensing and Regulatory Affairs AmerGen Energy Company, LLC Attachment (1)Three Mile Island Unit 1, Response to Request for Additional Information Related to USNRC Generic Letter 2004-02 (2)Three Mile Island Unit 1, Summary of Regulatory Commitments cc: Regional Administrator, USNRC Region I Project Manager, NRR, USNRC-Three Mile Island, Unit 1 Senior Resident Inspector, USNRC-Three Mile IslandR.R.Janati, Commonwealth of Pennsylvania File No.05049 ATTACHMENT 1 Three Mile Island Unit 1 Response to Request for Additional Information Related to USNRC Generic Letter 2004-02

Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 ACRONYMS ANSI American National Standards Institute NPSH Net Positive Suction Head ASTM American Society for Testing and Materials NPSHa Net Positive Suction Head available BS Building Spray NPSHm Net Positive Suction Head margin BWST Borated Water Storage Tank NPSHr Net Positive Suction Head required CAD Computer Aided Drafting MIL Department of Defense Military

Standard CF Core Flood MFWLB Main Feedwater Line Break CFT Core Flood Tank MSLB Main Steam Line Break CFD Computational Fluid Dynamics OTSG Once Through Steam Generator CSHL Clean Strainer Head Loss PWR Pressurized Water Reactor DH Decay Heat PZR Pressurizer ECCS Emergency Core Cooling Sy stem QA Quality Assurance EOP Emergency Operating Procedure RAI Request for Additional Information EQ Environmentally Qualified RB Reactor Building FTC Fuel Transfer Canal RCP Reactor Coolant Pumps GL Generic Letter RCS Reactor Coolant System GL 2004-02 SR TMI Unit 1 GL 2004-02 Supplemental Response dated 12/28/07 RNG Renormalized Group Theory GR Guidance Report RMI Reflective Metal Insulation HELB High Energy Line Break SB LOCA Small Break LOCA HP High Pressure SE Safety Evaluation ICP Inductively coupled plasma mass spectrometry TKE Turbulent Kinetic Energy LB LOCA Large Break LOCA TMI Unit 1 Three Mile Island Unit 1 LDFG Low-Density Fiberglass Insulation TSP Tri-Sodium Phosphate LHR Linear Heat Rate USNRC United States Nuclear Regulatory Commission LOCA Loss-of-Coolant Accid ent WC Water Column LPI Low Pressure Injection ZOI Zone-of-Influence Page 1 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 TMI Unit 1 Response Overview:

The recirculation functions for the ECCS and the BS System for TMI Unit 1 continue to comply with the regulatory requirements listed in the applicable Regulatory Requirements section of the subject GL under debris loading conditions. The response to the GL 2004-02 SR for TMI Unit 1 describes the completed corrective actions that ensure this compliance.

Listed below are the conservatisms, detailed throughout the GL 2004-02 SR, which TMI Unit 1 incorporated into its methodology for meeting GL 2004-02:

1. TMI Unit 1 utilizes a bounded debris loading strategy for testing inputs. The debris load utilized is a combination of the bounding fiber loads from the East D-Ring break and the bounding particulate loads from the West D-Ring break.

2. TMI Unit 1 utilizes a 7D ZOI on jacketed LDFG insulation for sump design calculations. WCAP-16710-P testing confirmed ZOI could be reduced further to 5D.

3. TMI Unit 1 utilizes a latent debris load of 300 lbs versus a walkdown determined value of 192.65 lbs

4. TMI Unit 1 utilizes a tags and labels loading of 400 ft 2 versus a walkdown determined value of 332.3 ft 2 5. TMI Unit 1's minimum 15" submergence of the top hat modules at minimum credited water level is greater than that used in the testing. Testing was conducted at an initial submergence of approximately 6" above the top hat modules at prototypical plant conditions, and no vortexing was observed for the postulated operating conditions of the

TMI Unit 1 sump strainer design.

NRC Question 1

Please describe the approach to the break selection process used (e.g., incrementing the break location along the potential high pressure lines) and explain how it is systematic and effective in bounding the amounts of debris generated from the various potential Loss of Coolant Accident (LOCA) locations.

TMI Unit 1 Response:

The break selection process evaluated a number of break locations in order to identify the location that is likely to present the greatest challenge to post-accident sump performance. The debris inventory and the transport path were both considered when making this determination.

The SE (NEI 04-07, Volume 2, Revision 0, NRC SE, "Safety Evaluation by the Office of Nuclear

Reactor Regulation Related to NRC GL 2004-02, Nuclear Energy Institute GR 'Pressurized Water Reactor Sump Performance Methodology'", December 2004) discusses a systematic approach to the break selection process where an initial location is selected at a convenient location and break locations are evaluated at 5-foot intervals in order to evaluate all break locations. Section 4 of the GR (NEI 04-07, Volume 1, Revision 0, NEI PWR Sump Performance Task Force, "PWR Sump Performance Evaluation Methodology," December, 2004) allows for Page 2 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 the development of target-based ZOIs, taking advantage of materials with greater destruction pressures. The Alion series of debris generation calculations utilized multiple ZOIs at specific break locations dependent upon the target debris.

Section 3.3.4.1 in the GR recommends that a sufficient number of breaks in each high-pressure system that rely on recirculation be considered to ensure that the breaks that bound variations in debris generation by the size, quantity, and type of debris are identified. The following five (5) breaks were considered in the TMI Unit 1 analysis:

Break No. 1: Breaks in the RCS with the largest potential for debris Break No. 2: Large breaks with two or more different types of debris Break No. 3: Breaks in the most direct path to the sump Break No. 4: Large breaks with the largest potential particulate debris-to-insulation ratio by weight Break No. 5: Breaks that generate a "thin bed" - high particulate with 1/8" fiber bed

The TMI Unit 1 Debris Generation Calculation determines which accidents required sump operation. LOCAs and certain SB LOCAs require sump operation. Other HELBs were considered and it was determined that sump operation was not required.

The following LB LOCA lines were evaluated. Breaks are considered up to the first normally closed isolation valve. The LB LOCA lines are:

36" RCS Hot Leg (East and West D-Ring) 28" RCS Cold Leg (East and West D-Ring) 14" CF Line (East and West D-Ring) 12" DH Line (West D-Ring)

All of the lines and their associated boundary valves discussed above are located within the

D-Ring. The SB LOCA lines evaluated at TMI Unit 1 are:

10" PZR Surge Line (East D-Ring Compartment) 2 1/2" PZR Spray Line (East D-Ring Compartment) 2 1/2" RCS Letdown (West D-Ring Compartment, Letdown Heat Exchanger Room, Containment Basement Over Containment Sump) 2 1/2" HP Injection (East and West D-Ring)

Other HELB Scenarios, MSLB, MFWLB, and RB Secondary Systems were individually evaluated to determine the specific method of DH removal for each scenario. In these cases, DH removal is accomplished by means other than operation of the ECCS sump recirculation system.

The break with the largest potential for debris generation is the break in an area with the largest concentration of debris source material. A comprehensive comparison of debris source term was performed to determine where the greatest quantity of material was installed within the East and West D-Rings. Three possible break locations were identified that have the potential to generate the largest quantity of debris based on the amount of each debris material within the Page 3 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 zone. One additional break was considered for the 2-1/2" letdown line as the break that could occur closest to the sump. These breaks are as follows:

RCS Hot Leg break in East D-Ring RCS Hot Leg break in West D-Ring Nozzle Break in Reactor Cavity Letdown Line Break The location of each type of material (fibrous, particulate, RMI and Latent Debris) was then evaluated to determine the specific ZOI and break location that would produce the greatest

quantity of debris.

NUKON LDFG The TMI Unit 1 East and West D-Rings, inside the secondary bio-shield wall but outside the primary bio-shield wall, are identical except that the East D-Ring contains the PZR and associated piping. The PZR is insulated with NUKON. Therefore, the RCS Hot-Leg Break in the East D-Ring compartment is the HELB with the largest potential for debris (more NUKON debris is generated).

RCS Hot Leg breaks in East D-Ring and RCS Hot Leg breaks in West D-Ring produce NUKON debris. The break in the Reactor Cavity and Letdown Line Break do not generate any NUKON/Thermal-Wrap. Summarizing and comparing the total amount of NUKON destroyed in the debris analysis, it was determined that a break in the East D-Ring Hot Leg is limiting in terms of the amount of NUKON/Thermal-Wrap insulation generated.

The specific break location on the RCS East D-Ring Hot Leg was determined by application of the ZOI criteria described in Table 2 of the GL 2004-02 SR.

The maximum destroyed quantity of NUKON generated by an RCS East D-Ring Hot Leg break was calculated by associating the specific TMI Unit 1 NUKON jacketing with the ZOIs and destruction characteristics as described in Table 2 of the GL 2004-02 SR. This was accomplished by positioning the break at the location that produces the maximum quantity of transportable NUKON fines. The critical areas to consider in this assessment are the top of the steam generator and PZR, the side of the PZR, and the NUKON on the candy cane piping. The lower elevations of RCS piping and the OTSG sides are insulated with RMI. Several break locations on the RCS Hot Leg were evaluated from the base of the Hot Leg near the exit of the reactor vessel to the top of the candy cane. The maximum quantity of transportable fiber is produced by an RCS pipe break near the candy cane, because a break at this location will include the largest quantity of NUKON on RCS piping within a 5D ZOI. A significant quantity of NUKON vulnerable at distances of 7D from the break (e.g., PZR) is also affected, including all of the NUKON on the top of the OTSG and PZR. Finally, the remaining NUKON is affected by the 17D ZOI as the 17D zone extends throughout most of the D-Ring. It can be observed from Figure 1 the 5D ZOI that breaks below the chosen location reduce the quantity of debris produced as the ZOI moves away from the components insulated with NUKON. Breaks above the chosen location reduce the quantity of NUKON destroyed on the PZR at the 7D and 17D

ZOIs. Page 4 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 Figure 1. TMI Unit 1 RCS Hot Leg at the base of the Hot Leg near the exit of the reactor vessel to the top of the candy cane.

RMI Prior to testing, the Debris Generation, Transport and Head Loss Calculations were performed to determine the head loss due to RMI and other debris constituents. All four (4) breaks taken into consideration produce RMI debris. NEI 04-07, Volume 2, Revision 0, identifies the radius of the ZOI for RMI is 2D. Summarizing and comparing the total amount destroyed, it is observed that the break at the nozzle in the Reactor Cavity generates a larger quantity of RMI than the

other three breaks.

Head Loss was analytically determined for the maximum quantity of RMI calculated. The analytical value of head loss associated with the quantity of transported RMI is defined by the following equation:

H = 0.108 U 2 A foil/A c where: H is the head loss, (feet-of-water) U is the sump screen approach velocity, (ft/sec)

A foil is the RMI foil surface area, (ft

2) A c is the effective screen surface area, (ft

2)

When the actual maximum value of RMI transported to the TMI Unit 1 sump is substituted into this equation, the resulting head loss is calculated to be <0.01 ft. H

20. Since the head loss associated with RMI was determined to be negligible, RMI was not utilized in testing. Page 5 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 Coatings For sump assessment, the quantities of coatings within the ZOI and outside the ZOI are determined. The failed coating debris source term was determined by considering the following:

1. All qualified coatings within the coating ZOI are postulated to fail.

2. All uncovered unqualified coatings both inside and outside the coating ZOI are postulated to fail.

For the postulated break, the ZOI was placed in several locations along the 28" and 36" RCS piping in order to determine the most limiting case for coating failure. From this evaluation it was determined that the greatest quantity of particulate debris is generated by a LOCA in the West D-Ring.

All unqualified coatings were postulated to produce debris regardless of their proximity to the

ZOI.

Testing Protocol The greatest amount of transported fibrous debris was determined to occur with a LOCA in the East D-Ring. The largest quantity of particulate debris was determined to occur with a LOCA in the West D-Ring. In order to bound the quantities of all debris types when establishing test protocol for TMI Unit 1 the maximum transported fiber from the East D-Ring was combined with the maximum particulate from the West D-Ring.

NRC Question 2 Please provide justification to support the characteristically smaller size distribution of destroyed fibrous insulation within the 7 diameter (7D) zone-of-influence (ZOI) for jacketed low-density fiberglass versus the size distribution which would exist for a larger ZOI. Include an explanation of how Table 2 (on page 8 of 65) of the GL supplemental response is consistent with the 7D ZOI assumed size distribution of 60 percent small fines and 40 percent large pieces.

TMI Unit 1 Response:

TMI Unit 1 established two zones for the fiberglass insulation debris within a 7D ZOI. The results of testing documented in WCAP-16710-P indicate that the blowdown forces at distances greater than 7D do not damage stainless steel jacketed NUKON insulation and cause a release of fibrous insulation. Therefore, TMI Unit 1 established 7D as the ZOI boundary for jacketed NUKON insulation. A test conducted at a distance equivalent to 5D resulted in minor damage to an insulation blanket when the jacketing was ejected, but the fiberglass blanket remained essentially intact and only a very small amount of fibrous insulation was released. Although the WCAP-16710-P test demonstrated that jacketed NUKON destruction levels are negligible at 5D, TMI Unit 1 assumed that some level of destruction occurs for insulation located at least 5D from the break location out to 7D.

All jacketed LDFG insulation located within 5D of the break is assumed to produce small fines. This is a conservative assumption based on the WCAP-16710-P test results. The insulation blanket positioned 5D from the break sustained only minor damage and was largely intact after the test. The debris generation analysis assumed that jacketed insulation located between 5D Page 6 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 and 7D of the break produces fibrous debris consisting of 60 percent small fines and 40 percent large pieces. The testing documented in WCAP-16710-P did not quantify the debris size distribution, but a review of the damage to the insulation blanket positioned 5D from the break concluded that significantly less than 60 percent of the fibrous insulation in the blankets was released. TMI Unit 1 conservatively applied the recommendation in NEI 04-07 of 60 percent small fines and 40 percent large pieces to insulation between 5D and 7D of the break.

NRC Question 3 Please provide the assumed size distribution for reflective metal insulation (RMI) debris.

TMI Unit 1 Response:

The destroyed RMI size distribution developed for TMI Unit 1 is based on steam jet test data in NUREG/CR-6808.

Table 1- The size distribution used in the debris generation analysis Debris Size (in.)

Percentage 1/4 4.3% 1/2 20.2% 1 20.9% 2 25.6% 4 16.8% 6 12.2% NRC Question 4 Please provide the post-transport size distributions for the RMI, and jacketed and unjacketed Nukon insulation debris with justifications for the transport fractions (e.g., erosion effects).

TMI Unit 1 Response:

NUKON Multiple ZOIs were considered to determine NUKON fiber loads. A ZOI of 5D is identified to produce 100 percent fines from all NUKON applications. Jacketed NUKON insulation located beyond 5D up to 7D from the break location is reported to produce 60 percent transportable fines and 40 percent large pieces. Other NUKON applications are reported to produce 60 percent transportable fines and 40 percent large pieces within a ZOI of 17D. For additional conservatism, the transport fraction of 15 percent for large intact pieces identified in the Debris Transport calculation is applied to the 40 percent destroyed as large pieces. For the fiber destruction methodology applied to this evaluation, it can be observed from Table 2 that the overall transport fraction is 84 percent (199.26/237.41 cu. ft.). Page 7 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 Table 2- NUKON Fiber Locations, Characteristics and Transport Fractions Total Fiber Generated NUKON Transported to the Sump Break Location w/top of Hot Leg boundary ZOI placed at Break Location Components Affected (ft 3) Size % of Total NUKON Destroyed Transport Fraction (ft 3) 5D 5D OTSG Top Head, Hot Leg Top Loop 125.19 Fines 100% 100% 125.19 Large Pieces 40% 15% 1.96 5D 7D - 5D PZR middle section (Shadowed by RCP) 32.7 Fines 60% 100% 19.62 Large Pieces 40% 15% 2.27 5D 17D PZR Top and Bottom Heads (no shadowing credited) 37.91 Fines 60% 100% 22.75 Large Pieces 40% 15% 0.43 5D 17D PZR Spray Line 1.08+1.30+4.81 Fines 60% 100% 4.31 Large Pieces 40% 15% 0.27 5D 17D OTSG "A" Manway 2.24+2.24 Fines 60% 100% 2.69 Large Pieces 40% 15% 0.09 5D 17D OTSG "A" Handhole 0.77+0.77 Fines 60% 100% 0.92 Large Pieces 40% 15% 1.64 5D 17D Hot Leg "A" Blanket 27.37 Fines 60% 100% 16.42 Large Pieces 40% 15% 0.06 5D 17D PZR Surge Line 1.03 Fines 60% 100% 0.62 Total NUKON Generated (ft

3) 237.41 Total NUKON Transported to the Sump (ft

3) 199.26 Table 2 Notes

1. NUKON Jacketed means the stainless steel jacket is supported in bearing by the NUKON fiberglass insulation blanket.

2. Small fines are transported to the sump. Large pieces are not transported to the sump, and there is no further breakdown or erosion of large pieces.

3. An obstructed object is in the shadow of a component that blocks the jet. An obstructed object is not the back side of the object that blocks the jet. Page 8 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 RMI The transport fractions for RMI were determined analytically by developing a Flow 3D model. The program utilized has been validated for use in safety-related analysis. Transport fractions are determined by the program based on pool velocities and TKE in each break case. Although RMI was not represented in the test, quantities and transport fractions were determined for each postulated break case as follows:

Case 1 - Break in the Loop 1 Hot Leg The post-transport size distribution for RMI debris was derived from the results of the debris transport calculation and is presented in the table below:

Table 3- Case 1 Post-Transport RMI Size Distribution Debris Size Debris Quantity Generated Debris Transport Fraction Debris Quantity at Sump Post-Transport Size Distribution Small Pieces (<4")

8,720 ft 2 48% 4,186 ft 2 87% Large Pieces (>4")

3,562 ft 2 17% 606 ft 2 13% The transport calculation concluded that both the small and large pieces of RMI could conservatively transport by tumbling.

Case 2 - Break in the Loop 2 Hot Leg The post-transport size distribution for RMI debris was derived from the results of the debris transport calculation and is presented in the table below:

Table 4- Case 2 Post-Transport RMI Size Distribution Debris Size Debris Quantity Generated Debris Transport Fraction Debris Quantity at Sump Post-Transport Size Distribution Small Pieces (<4")

8,217 ft 2 48% 3,944 ft 2 87% Large Pieces (>4")

3,356 ft 2 17% 571 ft 2 13% Both the small and large pieces of RMI would transport by tumbling.

Case 3 - Break in the Reactor Cavity The post-transport size distribution for RMI debris was derived from the results of the debris transport calculation and is presented in the table below:

Table 5- Case 3 Post-Transport RMI Size Distribution Debris Size Debris Quantity Generated Debris Transport Fraction Debris Quantity at Sump Post-Transport Size Distribution Small Pieces (<4")

18,197 ft 2 100% 18,197 ft 2 79% Large Pieces (>4")

7,433 ft 2 65% 4,831 ft 2 21% Both the small and large pieces of RMI would transport by tumbling.

Page 9 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 The debris generation calculation reported that the reactor cavity break would not generate any jacketed or unjacketed quantity of NUKON.

Case 4 - Break in RCS Letdown Line Outside Bioshield Wall The post-transport size distribution for RMI debris was derived from the results of the debris transport calculation and is presented in the table below:

Table 6- Case 4 Post-Transport RMI Size Distribution Debris Size Debris Quantity Generated Debris Transport Fraction Debris Quantity at Sump Post-Transport Size Distribution Small Pieces (<4")

865 ft 2 100% 865 ft 2 71% Large Pieces (>4")

353 ft 2 100% 353 ft 2 29% It was conservatively assumed in the debris transport calculation that all debris destroyed by the letdown line break cases postulated in the debris generation calculation would be transported to the sump screen.

The debris generation calculation reported that the RCS letdown line break would not generate any jacketed or unjacketed quantity of NUKON.

The Debris Transport analysis assumed that RM I would not break down into smaller pieces following the initial generation. This is a reasonable assumption since RMI is a metallic insulation that would not be subject to erosion by the flow of water.

NRC Question 5 Please provide a detailed description of any methods or assumptions used by the transport evaluation vendor for the refined transport analysis as discussed in Section 3.e.2 of the GL supplemental response. This would specifically include any items that are not consistent with the approved guidance (Safety Evaluation by the Office of Nuclear Reactor Regulation related to NRC Generic Letter 2004-02, Nuclear Energy Institute GR (Proposed Document Number NEI 04-07), ADAMS Accession No. ML043280007) for debris transport.

TMI Unit 1 Response:

The following assumptions apply to the debris transport calculation and the NUKON transport methodology:

General Assumptions

1. NUKON Jacketed means the stainless steel jacket is supported in bearing by the NUKON fiberglass insulation blanket. Small fines are transported to the sump. Large pieces are not transported to the sump, and there is no further break down or erosion of large pieces (See response to RAI No. 4 for details on transported fiber quantities). For conservatism, an additional 15 percent of the large piece quantity was assumed to transport to the sump.

2. It was assumed that 1/4"-4" pieces of RMI debris can be treated as 1/2" pieces and 4"-6" pieces can be conservatively treated as 2" pieces for transport purposes. This is a conservative assumption since smaller pieces of RMI tend to transport more easily. Page 10 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02

3. It was assumed that RMI would not break down into smaller pieces following the initial generation. This is a reasonable assumption since RMI is a metallic insulation that would not be subject to erosion by the flow of water.

4. It was assumed that the settling velocity of fine debris (insulation, dirt/dust, and paint particulate) could be calculated using Stokes' Law. This is a reasonable assumption since particulate debris is generally spherical and would settle slowly (within the applicability of Stokes' Law).

5. Due to a lack of data, it was conservatively assumed that the transportable miscellaneous debris addressed in the debris generation calculation including tags, labels, etc. would be transported to the emergency sump during recirculation.

Logic Trees

1. It was assumed that the fines generated by the LOCA blast would be transported to upper containment in proportion to the relative volume of upper containment compared to the entire volume. This is a reasonable assumption since fine debris generated by the LOCA jet would be easily entrained and carried with the blowdown flow.

2. It was assumed that a fraction of small piece debris would also be transported to upper containment in proportion to the relative volume. However, since there is grating between the break locations and upper containment, no large piece debris was assumed to be blown to upper containment.

3. It was conservatively assumed that all debris blown upward would be subsequently washed back down by the BS flow. The fraction of debris washed down to various locations was determined based on the spray flow split determined in Section 5.8.3 of the debris transport calculation.

4. During pool fill-up, it was assumed that a fraction of the debris would be transported to inactive areas, as well as some debris transported directly to the sump screen as the sump cavity fills with water. These fractions were determined based on the ratio of the cavity volumes to the pool volume at the point when the cavities are filled.

5. It was conservatively assumed that all of the debris generated by the Case 4 RCS letdown line break would be transported to the sump screen.

Debris Distribution at the Beginning of Recirculation

1. With the exception of latent debris washed to the sump and inactive cavities during pool fill-up, it was conservatively assumed that all latent debris is in lower containment, and would be uniformly distributed in the containment pool at the beginning of recirculation. This is a conservative assumption since no credit is taken for debris remaining on structures and equipment above the pool water level.

2. It was assumed that the unqualified coatings in lower containment would enter the recirculation pool in the vicinity of the locations where they are applied. This is a reasonable assumption since unqualified coatings outside the ZOI would break down gradually, and would be likely to fail after recirculation has been initiated.

3. It was assumed that the debris washed down from upper containment by the spray flow would be washed down during recirculation, or if washed down during pool fill-up would remain in the general vicinity of the location where it was washed down until recirculation begins. This is a reasonable assumption since there is no preferential pool flow direction during pool fill-up after the inactive and sump cavities have been filled. Also, this assumption is somewhat conservative since the local turbulence caused by the sprays would increase the potential for debris to transport from these locations. Page 11 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02

4. With the exception of debris washed directly to the sump screen or to inactive areas, it was assumed that the fine debris that is not blown to upper containment would be uniformly distributed in the recirculation pool at the beginning of recirculation. This is a reasonable assumption, since the initial shallow flow at the beginning of pool fill-up would carry the fine debris to all regions of the pool.

5. It was assumed that small and large piece debris that is not blown to upper containment would be uniformly distributed between the locations where it is destroyed and the sump. This is a conservative assumption since it neglects the fact that some debris would be blown or washed to areas farther away from the sump during the blowdown and pool fill-

up phases.

CFD Model

1. The coherent water stream falling from the RCS breach was assumed to disassociate into drops as it falls, with these drops reaching a terminal velocity. This is a reasonable assumption given the height of the hot leg above the pool surface, and the fact that the break flow would be broken up by equipment and piping at lower elevations.

2. For the CFD model, it was assumed that potential upstream blockage points (e.g., drains, fences, grating, etc.) would not inhibit the flow of water through these areas. The potential blockage of various points is addressed qualitatively in the TMI Unit 1 walkdown report, and in Section 5.2 of the debris transport calculation.

Note: Each of the assumptions listed above are consistent with the approved guidance provided in the SE and the GR.

The methodology that was used in the debris transport calculation is described below:

Methodology Debris transport is the estimation of the fraction of debris that is transported from debris sources (break location) to the sump screen. The four major debris transport modes are:

1. Blowdown transport - the vertical and horizontal transport of debris to all areas of containment by the break jet.

2. Washdown transport - the vertical (downward) transport of debris by the BS and break flow. 3. Pool fill-up transport - the transport of debris by break and BS flows from the BWST to regions that may be active or inactive during recirculation.

4. Recirculation transport - the horizontal transport of debris from the active portions of the recirculation pool to the sump screen by the flow through the ECCS.

The methodology used in this analysis of transport is based on the NEI 04-07 GR for refined analyses as modified by the USNRC's SE, as well as the refined methodologies suggested by the SE in Appendices III, IV, and VI. The specific effect of each transport mode was analyzed, and a logic tree was developed for each debris type generated (with the exception of NUKON) to determine the total transport to the sump screen. The purpose of this approach is to break a complicated transport problem down into specif ic smaller problems that can be more easily analyzed. A generic transport logic tree for a four-category size distribution is shown in Figure 2 below.

Page 12 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 The size distribution and characterization for the specific debris types comes from the debris generation calculation. The logic tree shown in Figure 2 is somewhat different from the baseline logic tree provided in the GR. This departure was made to account for certain non-conservative assumptions identified by the SE including the transport of large pieces, erosion of small and large pieces, the potential for washdown debris to enter the pool after inactive areas have been filled, and the direct transport of debris to the sump screen during pool fill-up. Also, the generic logic tree was expanded to account for a more refined debris size distribution. Page 13 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 DebrisGenerationDebris SizeFinesUpperContainmentLowerContainmentRetained onStructuresWashed DownBlowdownTransportWashdownTransportPool FillTransport CFDRecirculationTransportFraction of Debrisat SumpInactive PoolActive PoolSump ScreensSedimentTransportErosionSedimentTransportSmallPiecesUpperContainmentLowerContainmentRetained onStructuresWashed DownInactive PoolActive PoolSump ScreensSedimentTransportSedimentTransportRemains intactErodes to FinesLargePiecesRemains intactErodes to FinesLargePieces withJacketingIntactUpperContainmentLowerContainmentRetained onStructuresWashed DownInactive PoolActive PoolSump ScreensSedimentTransportSedimentTransportRemains intactErodes to FinesUpperContainmentLowerContainmentRetained onStructuresWashed DownInactive PoolActive PoolSump ScreensSedimentTransportSedimentTransportRemains intactErodes to FinesRemains intactErodes to FinesRemains intactErodes to Fines Figure 2 - Generic debris transport logic tree Page 14 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 The basic methodology used for the TMI Unit 1 transport analysis is documented below:

1. Based on many of the containment building drawings, a three-dimensional model was built using CAD software.

2. A review was made of the drawings and CAD model to determine transport flow paths. Potential upstream blockage points including screens, fences, grating, drains, etc. that could lead to water holdup were addressed.

3. Debris types and size distributions were gathered from the debris generation calculation for each postulated break location.

4. The fraction of debris blown into upper containment was determined based on the relative volumes of upper and lower containment.

5. The quantity of debris washed down by spray flow was conservatively determined. During the washdown phase, debris in upper containment could be washed down by BS. Since all of the debris blown to upper containment was determined to be fines and small pieces in the debris transport calculation, it was conservatively assumed that all of it would be washed back to lower containment. It was also assumed that failed coatings in upper containment would be washed down by the BS.

6. The quantity of debris transported to inactive areas or directly to the sump screen was calculated based on the volume of the inactive and sump cavities proportional to the water volume at the time these cavities are filled.

7. The location of each type/size of debris at the beginning of recirculation was determined as follows: The latent debris, unqualified coatings, and fines in lower containment were conservatively assumed to be uniformly distributed throughout the recirculation pool at the beginning of recirculation. The unqualified coatings, fines, and small piece debris washed down from upper containment were conservatively assumed to be distributed in the general vicinity of the washdown locations until recirculation starts. Small and large pieces of insulation debris (RMI) not blown to upper containment were conservatively assumed to be uniformly distributed between the locations where it would be destroyed and the sump screen. This is a conservative assumption since the blowdown and the majority of the pool fill-up phases are multi-directional flows that would tend to disburse debris around containment (including areas with lower transport potential).

8. A CFD model was developed to simulate the flow patterns that would occur during recirculation.

a. The mesh in the CFD model was nodalized to sufficiently resolve the features of the CAD model, but still keep the cell count low enough for the simulation to run in a reasonable amount of time.

b. The boundary conditions for the CFD model were set based on the configuration of TMI Unit 1 during the recirculation phase.

c. The BS flow was included in the CFD calculation with the appropriate flow rate and kinetic energy to accurately model the effects on the containment pool.

d. At the postulated LOCA break location, a mass source was added to the model to introduce the appropriate flow rate and kinetic energy associated with the break flow.

e. A negative mass source was added at the sump location with a total flow rate equal to the sum of the break flow and spray flow with the exception of the refueling canal spray flow.

f. The RNG turbulence model was judged to be the most appropriate for the CFD analysis due to the large spectrum of length scales that would likely exist in a containment pool during emergency recirculation. The RNG approach applies statistical methods in a derivation of the averaged equations for turbulence Page 15 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 quantities (such as TKE and its dissipation rate). RNG-based turbulence schemes rely less on empirical constants while setting a framework for the derivation of a range of models at different scales.

g. After running the CFD calculations, the mean kinetic energy was checked to verify that the model had been run long enough to reach steady-state conditions. Checks were also made of the velocity and TKE patterns in the pool to verify that steady-state conditions were reached.

h. Transport metrics were determined based on relevant tests and calculations (NUREG/CR-6772, NUREG/CR-6808, and calculations per Stokes' Law) for each significant debris type present in the TMI Unit 1 containment building.

9. A graphical determination of the transport fraction of each type of debris was made using the velocity and TKE profiles from the CFD model output, along with the determined initial distribution of debris.

10. The recirculation transport fractions from the CFD analysis were gathered to input into the logic trees.

11. The overall transport fraction for each type of debris was determined by combining each of the previous steps in logic trees.

NRC Question 6

The graph of head loss vs. time in Section 3o.2.17.1 of the GL supplemental response indicates that a large vortex occurred during testing. The vortex resulted in significantly decreased head loss that did not recover. However, in Section 3f.3 it is stated that no significant vortex

formations occurred during testing with non-chemical debris. Please address whether the vortex was prototypical for strainer operations under LOCA conditions. Further, please address whether the vortex that occurred during testing potentially disturbed the debris bed. Such a disturbance could result in non-conservative head loss results or non-conservative temperature scaling (due to turbulent flow through a discontinuity in the debris bed).

TMI Unit 1 Response:

The debris bed is postulated to have yielded and failed drawing a portion of the deposited debris through the strainer creating an area of clean screen and a resulting decrease in head loss.

The high shear which is associated with vortex formation almost certainly did impact the debris bed, resulting in a 21.3 ft WC stable head loss dropping to 12.5 ft WC during vortexing and then recovering to approximately 15 ft WC when the vortexing was eliminated (all at flows scaled from 8800 gpm ECCS flow). That the debris bed head loss did not recover to its pre-vortex value indicates that the vortex did indeed impact the head loss. While the test was completed, no credit is taken for data collected after the vortex formation that occurred with 96.5% of the chemical debris having been added.

The disturbed debris bed generated a reduced head loss, but the higher obtained head loss prior to the formation of the vortex was conservatively credited for the design qualification without temperature correction (therefore, at 85 degrees F). The maximum measured head loss was corrected for reduction in flow that would occur from securing BS flow (a higher turbulent credit than was taken would result in a lower head loss when correcting for a flow reduction) resulting in a credited head loss at the end of the 30 day cycle of 14.3 ft WC (at 6440 gpm).

TMI Unit 1 has implemented procedural actions to secure the BS pumps to maintain the debris bed head loss at values less than the tested and analyzed limits.

Page 16 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 Vortexing is not expected to occur under LOCA conditions with the noted procedural actions implemented. This is because the flow rate and debris bed head loss will be maintained significantly less than the peak values observed in the TMI Unit 1 testing. Additionally, the actual submergence of the sump strainer under LOCA conditions is at least 15" versus the above-described WCAP test setup, which had 10-1/2 " of water level above the top of the strainers. A further mitigating design feature is installation of a trash rack (made from horizontal grating) that will be 6" below the minimum water level and approximately 11" above the strainer serving to suppress the rotational shear that leads to vortexing. (These features are also discussed in the answer to Question 10 below.)

NRC Question 7 The response to Section 3f.6 states that the Enercon strainers were shown to resist the formation of a thin debris bed by testing. It is not clear that the strainers were tested with prototypically fine fibrous debris, introduced conservatively with prototypical flow conditions to ensure that a thin bed would not occur in the plant. The debris generation section of the GL supplemental response gives a breakdown of large pieces and small fines, but, for the purposes of a thin bed, fines are the important debris bed constituent. The debris characteristics section for a 17D NUKON ZOI in the GL supplemental response states that the size distribution is per the safety evaluation (SE) and has been reviewed by the NRC during audits. This is an insufficient rationale for debris sizing. Most NRC audits have identified issues relating to fibrous debris generation and debris size distribution. Please provide information to demonstrate that the fibrous debris sizes used for testing match the debris reaching the strainer as calculated in the debris transport calculation, differentiating between fine and small fibrous debris. Thin bed testing should be conducted with the finest debris predicted to reach the strainer. A specific reference to consider is staff review guidance, Enclosure 1 of a March 28, 2008, NRC letter to the Nuclear Energy Institute (ADAMS Accession No. ML080230038).

TMI Unit 1 Response:

Size Distribution for Test Material Testing of the TMI Unit 1 strainers was performed with the specific debris material found in the plant, or with a debris surrogate material. At TMI Unit 1, NUKON LDFG is installed in containment. For testing, NUKON insulation was consistently processed in accordance with established procedures for fiber preparation. This multi-step process requires all fiber be cut, shredded, boiled, and mixed with an electric paint mixer prior to being added to the test tank.

The received insulation blankets are cut into 12" pieces and processed through a leaf shredder.

The physical characteristics of the fiber are inspected and evaluated to compare the processed fiber to standard classifications described in the debris preparation procedure. The desired characteristics must be achieved before processing continues. Fibers are observed to exist in a range of sizes from single individual fibers to small (approx. <1") tufts of fibers before additional processing. After these initial characteristics are met, the shredded fiber is weighed in the desired quantities and boiled in accordance with the procedure. Boiled fiber is placed in containers with additional water and particulate debris, and mixed with an electric paint mixer until a homogeneous mixture is observed. The paint mixer is typically applied again immediately before each batch of debris is added to the test tank.

Page 17 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 Section 3.4.3.2 of NEI 04-07 provides a discussion of the debris size distributions and characteristics associated with each size. In this document, a two-size distribution is described for fiber destruction within the ZOI - small fines and large pieces. Small fines are defined as any material that could transport through gratings, trash racks, or radiological protection fences by blowdown, BS, or post accident pool flows. NEI 04-07 classifies the remaining material as that which cannot pass through gratings, trash racks, and radiological fences, as large pieces with a nominal size of 4".

Comparing the size distribution for the two categories described by NEI 04-07 with the debris used in TMI Unit 1 testing, the fiber introduced into the test tank provides the desired proportion of small fines. All of the initial dry processed fiber is observed to be substantially smaller than the category defined in NEI 04-07 as "large pieces." The remaining category of "small fines" as described in NEI 04-07 could be applied to all fiber used in the TMI Unit 1 testing based solely on the observed condition of the fiber at the end of the dry processing steps. The additional mixing of the fiber with electric paint mixers breaks the fiber down further. It should be noted that the fiber processing applied to the TMI Unit 1 test fiber yields a substantial quantity of base constituent (individual fibers). Fiber processed in this manner was used for all testing performed for TMI including thin bed and full load testing.

Additionally, in NEI 04-07, the nature of transport associated with small fines is discussed in terms of extended suspension time, as opposed to the large pieces that could transport to the strainer along the floor. The TMI Unit 1 tests included deliberate agitation of the test tank to maximize the amount of time the fibrous debris is suspended in the water. Agitation of the tank continued throughout the test.

NRC Question 8 Section 3f.13 of the supplemental response describes how the test results were scaled for strainer approach velocity. Please provide the tested velocity and velocities to which the test results were scaled. In addition, please justify that the scaling was conservative or prototypical. For example, scaling to a higher velocity would not be considered conservative because of the potential for effects such as bed compression.

TMI Unit 1 Response:

An approach velocity of 0.0086 ft/sec was used during strainer head loss testing. This value is based on a maximum sump flow rate of 8800 gpm and a net strainer surface area of 2280 sq. ft.

The maximum expected sump flow rate includes two trains of LPI at indicated flow of 3000 gpm per train and two trains of BS at 1180 gpm per train. Instrument error is included for the LPI

flow. During the debris head loss tests, flow sweeps were performed to determine the relative proportionality of the laminar and turbulent flow contribution to head loss. This test data was used to develop a correlation that can be used to scale the test result to various temperatures

and flow rates.

Although flow sweeps were performed to gather data for flow rates greater than 0.0086 ft/sec, actual recirculation flow rates for TMI Unit 1 will be less than the 8800 gpm flow rate used as the basis for the test flow rate. The NPSH calculation uses the correlation developed during testing to determine strainer head loss values for flow rates that are less than the tested value. Specific Page 18 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 cases evaluated in the NPSH analysis include total strainer flow rates of 8582 gpm (two trains LPI + two trains BS), 6222 gpm (two trains LPI), 3076 (one train LPI), and 4256 gpm (one train LPI + one train BS).

NRC Question 9 It appears that the final chemical effects test was run over a period of 2 days. The head loss trend appeared to be increasing slowly over the last 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> of the test. The response to Section 3o.2.17 says no extrapolation was conducted. Other long-term testing has shown a slow but significant increase in head loss over a matter of days after initial bed formation/head loss increases. This is particularly important since oneTMI-1 net positive suction head (NPSH) case shows only 0.1 ft. of margin (although it cannot be determined from the submittal when this occurs). Please provide an extrapolation of the data to the mission time or explain why one is not necessary.

TMI Unit 1 Response:

The minimum excess NPSH occurs immediately following switchover to the sump recirculation mode when the sump temperatures are still relati vely high. Available NPSH then increases slowly as the saturation pressure gradually lowers with decreasing sump liquid temperatures.

Below 140 degrees F, the additional precipitation of chemicals out of solution onto the sump strainer results in a significant increase in the strainer head loss. Proceduralized actions to secure the BS pumps and reduce LPI flow are in place to ensure the strainer differential pressure does not exceed the structural limits.

As noted in the response to RAI No. 6 above, the debris bed is postulated to have yielded and failed. This occurred approximately 18 hours2.083333e-4 days <br />0.005 hours <br />2.97619e-5 weeks <br />6.849e-6 months <br /> before the conclusion of the head loss test. The head loss prior to the disturbance was 21.3 ft. with 96.5 percent of the chemical debris having been added. This higher head loss value was conservatively credited for the design qualification without temperature correction. No credit was taken for the data collected after the step reduction in head loss occurred. Therefore, the last 12 hours1.388889e-4 days <br />0.00333 hours <br />1.984127e-5 weeks <br />4.566e-6 months <br /> of data was not used in subsequent head loss analyses.

Note: In the response to RAI No. 14 below, the test system head loss was observed to remain stable within 1 percent over the final 12-hour period.

NRC Question 10

Section 3f.8 of the supplemental response does not identify any conservatisms or margins for the head loss calculations. Please list any conservatisms or margins associated with the head loss and vortexing evaluation. Page 19 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 TMI Unit 1 Response:

Head Loss Evaluation Conservatism

1. The maximum expected BS pump flow rate is applied to ensure a minimum NPSH margin is achieved for all cases and scenarios.

2. By procedure, one of the operating BS pumps is secured after initiation of sump recirculation. The NPSH analysis allocates one hour to perform the action of securing the BS pump.

3. Flow sweeps were performed during the head loss test to determine the relative proportionality of the laminar and turbulent flow contribution to head loss. As a result of the analysis performed, it was determined that the flow through the debris bed was 97 percent laminar and 3 percent turbulent. To scale the measured head loss results to the actual sump temperature that results in a limiting available NPSH, a flow split of 70 percent laminar and 30 percent turbulent was conservatively utilized.

Vortex Evaluation Conservatism

1. Testing has shown that top hat strainer modules in a clean condition are not susceptible to drawing an air-core vortex from the water surface for a top hat flow rate up to 3.29 cu.ft./sec and a submergence as low as 4" above the top hat modules. The maximum top hat flow rate for the TMI Unit 1 modules (0.245 cu. ft./sec) is significantly less than that of the tested flow rate. Site specific testing was conducted at a depth of 6" and observed no air core vortex. Based on the minimum containment water level calculation, actual submergence of the sump strainer is at least 15".

2. Vortexing that occurred in the chemical effects head loss testing occurred at a head loss (21 feet) that exceeded the maximum head loss expected in the plant. By procedure, action is taken to secure BS pumps and reduce LPI flow to ensure that the strainer structural differential pressure limit of 16.15 feet is not exceeded.

3. A trash rack is installed over the top of the top hat modules to protect them from large debris. The trash rack grating will also provide vortex suppression although the grating is not credited as a vortex suppressor in the hydraulic evaluation.

NRC Question 11 Section 3f.14 of the supplemental response states that containment accident pressure was not credited to ensure that flashing would not occur. However, the response (Section 3f.10) states that above 140 degrees the head loss is 1.7 ft. This is more than the submergence of the strainer (15 inches). Please explain the apparent discrepancy, and describe and justify whether flashing occurs.

TMI Unit 1 Response:

Per 3f.10, the head loss is provided as 1.7 ft at 8800 gpm. The response was provided as a simplified reference case value only.

This 1.7 ft head loss at 8800 gpm is modified as a function of the flow rate and sump temperature for the various operational cases described in Section 3g.16 of the GL 2004-02 SR.

Because of the flow rate and temperature effect, the actual head loss is not constant at 1.7 ft. Page 20 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 In addition, at temperatures less than 208 degrees F, the liquid temperature is sub cooled relative to the pre-accident RB atmosphere. The minimum RB pressure pre-accident is (-1) psig, which corresponds to a saturated vapor temperature of 208 degrees F. When the temperature is less than 208 degrees F, the subcooled liquid condition is credited to preclude vapor formation (flashing). The cooler the sump water, the less prone the liquid is to flashing for

a given head loss.

Given these appropriate modifications and subcooling effect on flashing potential, the strainer head loss is modified from the 1.7 ft head loss reference case provided in the response.

Because of these factors, flashing is not expected.

NRC Question 12 The response to Section 3o.2.2.i is not sufficiently justified. The question asks if the test debris selection will result in the highest head loss. The AmerGen response was that the maximum amounts of fibrous and particulate debris were used in integrated chemical effects testing. This response does not address the potential for the formation of a chemical thin bed. Testing that searches for potential thin beds should use prototypically fine fiber introduced in graduated amounts. Please address the potential formation of a thin bed during integrated chemical

effects testing.

TMI Unit 1 Response:

Prior to performing the chemical hydraulic test for TMI Unit 1, Alion performed a series of prototype tests with only physical debris. These tests included a thin bed test. As observed in other strainer tests that utilized Enercon designed top-hat strainers, debris does not deposit uniformly across the screen surface. For the pr ototype thin bed test, particulate and fiber were scaled to the test array to deliberately form a 1/8" - 1/4" thick debris bed. The test protocol included intermediate review of the head loss generated during the testing to confirm that the screen was completely covered with the quantity of fiber and particulate introduced to the test system. At the conclusion of the test series, the thin bed head loss was observed to be less than the maximum head loss for all tests performed. The maximum head loss was found to occur with the full debris load.

As noted in the strainer head loss calculation, "Based on experience with testing of advanced screen designs, complex screens generally do not exhibit the thin bed effect. This is due to the non-uniform approach velocities along the screen surface developing non-uniform debris bed thickness. By the time the fibrous debris fully covers the screen in all areas, the debris bed is sufficiently thick in other areas to preclude the high head losses associated with thin fiber beds and large particulate loads."

The integrated chemical effects testing did not include a chemical thin bed since the maximum fiber and particulate load tested in non-chemical hydraulic testing exhibited a larger head loss than the thin bed tests.

Therefore, the most conservative debris bed to which chemical precipitate would be added (highest head loss) was tested with the chemical precipitates.

Page 21 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 NRC Question 13 The response to Section 3o.2.15.i stated that some debris settlement occurred during stirred integrated chemical effects testing, but concluded that the quantities were negligible and reasonable without apparent justification. Please provide a justification for why the settlement that occurred during integrated chemical effects testing did not result in non-conservative head loss values.

TMI Unit 1 Response:

The Alion Test Program has developed generic test procedures for debris preparation and addition. For all tests, the fiber and particulate was mixed together thoroughly with a paint mixer attached to an electric drill until a homogeneous slurry was formed. The complete contents of the buckets are then added to the test tank to ensure no loss of fine debris. The debris is introduced at the top of the tank and transported towards the strainer array by the water flow drawn through the strainers. The Alion Test Program allows for various methods of suspending the debris in the test tank and ensuring adequate presentation to the screen. These methods include the use of sparger systems as well as periodic manual stirring, and constant stirring with up to two electric propellers at each end of the tank, to keep the debris suspended in the fluid and prevent non-prototypical settling on the tank floor. The TMI Unit 1 tests included the use of a sparger as well as periodic manual stirring. The manual stir actions were continued until it was observed that subsequent stirs did not yield significant increases in head loss across the strainer. Any material that settled in the corners or on the floor was qualitatively documented in the test observations or with the use of test photographs. This amount of material when mixed back in with the test tank flow would re-accumulate and settle in the corners. The observation of re-settling debris combined with a lack of head loss increase after stirring suggests that there will be no more accumulation of debris on the screen at the tested approach velocity.

NRC Question 14

Please supplement the response to Section 3o.2.16.i to provide the test termination criteria.

TMI Unit 1 Response:

The head loss measurements for each test are recorded continuously throughout the test. The final head loss value is achieved when a stable differential pressure is achieved as determined by the test coordinator. This is determined to occur when the change in head loss is found to be less than 1 percent over a 1-hour period. The TMI Unit 1 test introduced individual batches of chemical debris after the initial physical bed was established and met the stabilization criteria. Prior to the end of the test sequence, the TMI Unit 1 test required a reduction in flow during the last of the chemical additions. Flow was adjusted to a specific predetermined value that correlates to plant conditions. After the flow was reduced, the test system head loss was observed to remain stable within 1 percent over an extended 12-hour period. The final test conditions were within the established termination criteria.

NRC Question 15

Sections 3o.2.19 and 3o.2.20 were evaluated as not applicable to TMI-1. Please provide a more comprehensive answer to these questions regarding representative testing, specifically addressing tank scaling, bed formation, and debris transport. Page 22 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 TMI Unit 1 Response:

Tank Scaling The prototype strainer consisted of a 2 X 2 top hat array representative of a section of the full plant strainer. For fiber, particulate, and chemical debris scaling, the prototype strainer screen surface area is divided by the total plant strainer area (with latent tag blockage considered) to determine the appropriate scaling factor. This scaling factor is also applied to determine the test flow rate to ensure that the approach velocity is identical to the plant installation. This methodology results in the same debris per square foot loading and screen approach velocity for the prototype and the plant strainer. In addition, the ratio of physical to chemical debris is consistent between the test conditions and plant configuration.

Bed Formation The test array consisted of several prototypical top hat modules that are expected to load similarly to the plant top hat modules. Upon establishing the required flow across a clean strainer, the physical (fiber and particulate) debris is added to the test tank. The fiber and particulate are mixed homogeneously with water outside of the tank in individual batches. Each batch is vigorously mixed with a paint mixer for several minutes immediately prior to being added to the tank. When all the fiber and particulate was added to the tank, the test system was allowed to run until the stabilization criteria was achieved. This process involved constant agitation of the tank water with an inlet sparger and constant mixing with up to two electric propellers at each end of the tank, supplemented with manual stirs. The manual stirring was repeated until the activity was observed to result in a negligible increase in head loss, and the stabilization criteria were met. Then, with the developed bed formed on the prototype array, chemical precipitants were added to the tank. The chemicals were prepared prior to the test in quantities determined by WCAP 16530, and segregated. The first chemical addition consisted of the total quantity of calcium phosphate. After this initial chemical addition, the tank was again frequently stirred until stabilization. Then the total quantity of sodium aluminum silicate was added. As before, the tank was again stirred until stabilization. Finally, the aluminum oxyhydroxide was added in two batches. This first batch quantity of aluminum oxyhydroxide was consistent with the predictions of the refined WCAP model (16785-NP). The second batch of aluminum was intended to increase the total aluminum oxyhydroxide quantity to that predicted by the original WCAP model (16530-NP). With identical approach velocities and the debris introduction methodology employed in testing, the bed is anticipated to form in a representative manner as it would under plant LOCA conditions.

Debris Transport The Debris Generation and Transport analyses determine the quantity of debris that arrives at the sump. With the exception of RMI and tags/labels (see Notes below), the total debris quantities from the analyses are used to determine the scaled debris quantities for testing. The prototype tank test uses various deliberate methods to ensure all of the debris is presented to the strainers for bed formation. Initially, debris is suspended in individual liquid batches before adding to the test. Each batch is stirred with a paint mixer when the water is added to the debris batch. The batch is again stirred with the paint mixer immediately prior to addition to the tank.

Once in the tank, the debris is suspended by continuously operating an internal sparger system.

This is further supplemented with manual agitation using a wooden oar or an electric propeller Page 23 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 as debris is observed to settle on the floor of the tank. The manual agitation activities are typically continued until it is observed that further agitation does not result in an increase in strainer head loss.

Note 1: The head loss associated with RMI was analytically shown to be negligible; RMI was not utilized in the test.

Note 2: The net available strainer surface area was adjusted based on the quantity of tags/labels assumed to transport to the sump.

NRC Question 16

Please provide a more detailed description of the NPSH margins calculation methodology, including a description of the time-dependent analysis specifying selected values for NPSHa (NPSH available) and NPSHr (NPSH required) throughout the mission time. This description should include significant time-dependent variables and how they change throughout the postulated event. For example, head loss changes due to chemical effects at different temperatures and changes in head loss and NPSHa due to sump temperature changes should be discussed.

TMI Unit 1 Response:

The GOTHIC computer program was used for the LPI and BS pump NPSHm analysis.

GOTHIC is a general purpose thermal-hydraulics computer program developed by EPRI. GOTHIC determined the post-LOCA reactor building sump conditions as a function of time, as well as pressure drop from the sump to the pump suction. Control variables were included in the model to account for sump strainer losses, including debris buildup and chemical effects.

Four cases are evaluated in the NPSH margin analysis to provide bounding NPSHa, NPSHr and NPSHm analyses for a rapid cooldown and a slow cooldown case:

Case I 2 LPI + 2 BS pumps available (max flow and max head loss case)

Case II 2 LPI pumps with no BS (limited RB cooling and max flow to most challenged NPSH pumps, coincident with elevated RB sump temperatures) Case III 1 LPI pump with max recirculation flow still operating (Max LPI pump flow)

Case IV 1 LPI pump plus 1 BS pump (maximum single train line losses)

Each of the above cases were evaluated using minimum cooldown (similar to "EQ" conditions) and maximum cooldown boundary conditions ("Maximum") to challenge the strainer head loss conditions.

The significant time-dependent variables in the NPSH analysis are:

1. Strainer Head Loss A step change in the strainer head loss function occurs when the sump temperature reaches 140 degrees F.

This is due to the impact of the aluminum based chemical precipitants as described in the response to RAI No. 27.

Page 24 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02

. 2. BS Pump Operation Each operating BS pump impacts the strainer head loss by increasing the flow rate through the strainer by 1180 gpm. To reduce the differential pressure across the strainer, the first BS pump is secured one hour after recirculation is established. The last running BS pump is secured 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> after recirculation. EOP guidance for operation of the BS pumps was revised to ensure that BS is shutdown within the limits assumed in the analysis.

3. LPI Throttling The NPSH analysis also evaluates the maximum differential pressure across the strainer to ensure the strainer design limits are not exceeded. Credit is taken for operator action to throttle LPI flow based on indications of high strainer differential pressure as described in the response to section 3f.10 of the GL 2004-02 SR.

NRC Question 17 Please clarify the NPSH result Tables 14 and 15, including a statement as to whether or not the screen and debris losses are included in all of the results. The clarifying information should specify the equipment qualification and maximum reactor building cooling scenarios, and provide an evaluation as to why these scenarios represent the most limiting cases for NPSH margin calculations throughout the mission time.

TMI Unit 1 Response:

The NPSH results provided in GL 2004-02 SR Tables 14 and 15 include the screen and debris

losses.

The maximum cool-down profiles were evaluated for maximum strainer differential pressure concerns. When evaluating the strainer head loss test results, it became apparent that under cold conditions with high flow rates, the differential pressure across the strainer could approach strainer structural design limitations. To quantify the impact of lower containment temperatures on the strainer head loss, each of the NPSH cases (listed in the response to RAI No. 16 above) was run with RB heat removal parameters adjusted to maximize the building cooldown rate. For example, initial containment temperature is assumed to be 80 degrees F and initial BWST temperature is assumed to be 40 degrees F. To maximize the heat removal equipment failures are not assumed and maximum cooling air/water flow rates and non-degraded cooler surface areas are used. By setting these parameters to maximize the cooldown rate, a conservative evaluation of the maximum head loss across the strainer was performed. Based on the results of these evaluations, operator actions to secure spray pumps and throttle flow were initiated as described in the GL 2004-02 SR to Issue 3f.10. These actions were analytically shown to be effective in maintaining strainer differential pressure below the strainer design limits.

NRC Question 18

Please provide a technical basis for Building Spray flow rate. In doing so, please describe whether it is a maximum/runout flow rate, a calculated flow, or a flow rate set by an operator under proceduralized criteria. If Building Spray rate is calculated, provide a description of the Page 25 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 system lineup, the plant conditions, the calculation method, and a list of assumptions and conservatisms.

TMI Unit 1 Response:

The maximum BS flow rate (1180 gpm per train) used in the Debris Transport and Head Loss/NPSH analyses is a calculated maximum flow rate. Orifices were installed in the TMI Unit 1 BS System to regulate BS flow between 1100 and 1180 gpm per train. A Pipe-Flo model was used to evaluate various combinations of ECCS and BS lineups to determine the maximum BS flow expected during any design basis accident.

The maximum system flow through the spray nozzles is nominally 1136 gpm for the "A" train and 1135 gpm for the "B" train. A bounding value of 1180 gpm has been used in other system analyses; therefore, this bounding value was also applied in the RB sump analyses.

For maximum BS pump flow, an RB pressure of 25 psig with a full BWST was used. This 25 psig value is a conservative value based on the plant actuation setpoint (30 psig) and instrument error. The corresponding LPI flow is considered to be the nominal value less instrument error. The BWST level was assumed to be 1 ft above the high level alarm setpoint.

A second case evaluated for maximum BS pump flow assumed an RB pressure of 0 psig with a low BWST level. This pressure bounds the lowest RB pressure with the BWST at minimum level. The LPI flow rate applied is as described above.

NRC Question 19

Please provide a discussion of how the single failure criterion was used in determining the bounding NPSH margin and why there is confidence that the worst-case single failure was identified and considered.

TMI Unit 1 Response:

In the sump recirculation mode, NPSH requirements are more limiting for the LPI pumps than for the BS pumps. The most limiting case for LPI NPSH requirements is any failure that results in operation of a single LPI pump injection through both discharge paths into the RCS using an open cross-connect header. The implied single failure is any event that disables the parallel train LPI pump. For this application, the flow rate through the single operating pump, and thus the NPSH requirement, is greater than when the system is operating with both pumps running and the cross-connect valves closed. The i ndividual pump flow rates for two LPI pump operation (Cases I and II) and single LPI pump operation (Cases III and IV) are provided in Table 14 of the GL 2004-02 SR response. The minimum excess NPSH condition occurs for Case IV, which is the single pump configuration with the same train RB Spray pump operating.

Operation of the same train RB spray maximizes flow in the LPI pump suction header and also the resulting piping friction losses, minimizing the NPSH available. (Additional discussions of the system lineups assumed for Cases I - IV is pr ovided in the response to RAI No. 16 above). Page 26 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 NRC Question 20

Please provide a description of the qualified sealant mentioned as being used in the Reactor Building Moisture Barrier and provide a brief summary of the testing performed to qualify the

sealant.

TMI Unit 1 Response:

The sealant used to replace the TMI Unit 1 RB Moisture Barrier is the Thiokol 2235M Industrial Polysulfide Joint Sealant, manufactured by Polyspec. Thiokol 2235M is a high performance, non-sag, NSF approved chemical resistant elastomeric joint sealant. Due to its high polysulfide polymer content, it is resistant to many chemicals, shrinkage, aging, thermal stress and the effects of outdoor exposure.

The product benefits include: Retains elasticity even as concrete moves; maintains flexibility over time Resists mild acids, alkalies and petroleum products Resists effects of aging, shrinkage and cyclic temperature changes, even after years of service Contains no volatile solvents

The product approvals include: Certified NSF Standard 61, Sec. 6 MIL TT-S-00227, Type II, non-sag ASTM C-920, Type M, Grade NS, Class 25, Use NT, M, G, A and O ANSI A116.1

The Korea Atomic Energy Research Institute tested the Thiokol 2235M Industrial Polysulfide Joint Sealant. The product was irradiation tested at 0.5 MRAD/hr (average) to total absorbed dose of 200 MRAD (2E8 RADS) with satisfactory results.

Also, the Thiokol TD-569N 5/69, "Radiation Resistance of LP Liquid Polysulfide Polymer Based Compounds" contains information concerning the base polysulfide resin used in the formulation of Thiokol 2235M Industrial Polysulfide Joint Sealant. This information indicates that the physical properties of the sealant, particularly tensile strength and elongation, will be satisfactory after irradiation to a total absorbed dose of 1E8 RADS.

NRC Question 21 Please summarize the evaluation of the flow paths from the postulated break locations and containment spray washdown to identify potential choke points in the flow field upstream of the sump.

TMI Unit 1 Response:

A walkdown of the RB and flow path assessment was performed during the refueling outage in

October 2005. This walkdown and flow path assessment was executed in accordance with guidance in NEI 02-01 Section 5.2.4.2.

Page 27 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 RB General Description

The TMI Unit 1 RB has a D-Ring containment design that houses a 2 loop B&W designed pressurized water reactor nuclear steam supply system.

1. Operating floor, FTC, and above from elevation 346'-0" to the top of the dome at elevation 468'-4 1/2" The BS System provides spray water through two redundant trains to give uniform spray coverage of the RB volume above the operating floor.

Spray water will pass to the elevations below through several paths: Through the top of the D-rings Into the FTC Onto the concrete and through the grating of the operating floor The top of the D-rings are partially covered with grating and partially open. The top of the D-Ring walls have a 6" curb around the outer periphery; therefore, all water falling onto the top of the D-rings would be routed downward into the D-rings. The partial grating would also prevent some large debris from ejecting out of the D-rings during the blowdown phase of a HELB.

Spray water falling into the FTC will collect in the shallow end and deep end of the canal.

Water retained in the shallow end will drain to the deep end before it reaches the height needed for water to enter the reactor vessel seal plate and into the reactor vessel cavity. All spray water entering the FTC will drain through the deep end 4" drain line that runs directly to the RB sump pit. The screen for this drain line was modified to prevent debris blockage and significant water hold up in the FTC (see response to RAI No. 23 below).

Water holdup in the FTC due to the drain piping stub has been accounted for in the water level calculation.

Spray water falling onto the operating floor will primarily land on 346'+0" elevation concrete. Some may also fall directly to the elevation below through the grated areas above the CFTs and above the equipment hatch, through the stairways, or onto the containment liner and through the open areas around the periphery of the operating floor. Water falling onto the concrete floor will drain to grated openings and the entrance/exit to the downward stairways that are open. In addition, the grated openings and stairs would prevent the transport of large debris to elevations below.

The floor is also equipped with 4" floor drains with slotted cover plates that might be susceptible to debris clogging. It is concluded, though, that the primary spray drainage flow paths from the operating floor are through the periphery around the floor, through the grated openings above the CFTs and above the equipment hatch, and through the

stairway openings.

2. Ground floor outside the D-rings from elevation 308'-0" to 346'-0" Spray water flows to the 308'+0" elevation from the operating floor above primarily through openings around the periphery of the 365'+0" floor elevation, the stairways, and the grated openings above the CFTs and above the equipment hatch. Water will begin to pool on the concrete floor and drain to penetrations and openings in the floor such as Page 28 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 the grated openings and the entrance/exit to the downward stairways. These flow paths would prevent the transport of large debris to elevations below.

As with the operating floor, the 308'+0' elevation floor is also equipped with 4" floor drains with slotted cover plates which flow to the RB sump pit in the basement. These drains could be susceptible to debris clogging. It is concluded that the primary spray drainage flow paths from the 308'+0' elevation are around the periphery of the floor, through the grated opening below the equipment hatch, and through the stairway openings.

3. "A" D-Ring (East) from elevation 281'-0" to 365'-6" and "B" D-Ring (West) from elevation 281'-0" to 365'-6" There are no concrete floors within the D-Rings to hold up water; however, the D-Rings are very congested with piping, equipment, and multiple grated platforms at varying elevations. Spray water entering the top of the D-Ring would impact and run off grating and equipment before reaching the recirculation pool below. Depending of the break location, it is likely that some large debris could be held up by the D-Ring grating.

While water can be held up in the RCP Lube Oil Collection System that uses a gutter arrangement to route oil to collection tanks, there were no other significant water hold ups identified in the D-Rings above the basement floor.

Once the spray and break water reaches the basement floor, there does not appear to be any water or debris hold ups. The only D-Ring exit is through the labyrinth pathway to the doorway through the secondary shield wall. Previously identified as a potential flow choke point, this doorway has been modified to prevent debris blockage and water holdup (See response to RAI No. 22 below).

4. Reactor cavity and incore instrumentation tunnel from elevation 279'-6" to 321'-0" Spray water drainage around the periphery of the 308'+0" elevation above, through the stairway openings, and through the grating below the equipment hatch will fall to the recirculation pool below. Obstructions located outside the D-Ring between the East stairway and the RB sump do not appear to provide any water or debris hold ups. The area is relatively open for flow.

NRC Question 22

The TMI Unit 1 GL 2004-02 response dated September 1, 2005, identified the doorways to the D-Ring and the incore chase areas as potential choke points for water flow to the sumps, whereas the GL 2004-02 supplemental response stated a replacement of the doors to the entrances to the D-rings was performed as a configuration change. Please describe the new door configuration and provide the basis for the statement that the D-ring and incore chase area doorways are no longer choke points based on plant modification or other considerations.

Page 29 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 TMI Unit 1 Response:

The main recirculation path for water from inside the D-rings to the RB sump is through the doorway on the 281' elevation of the RB. During the Fall 2007 TMI Unit 1 refueling outage (1R17), the door at this entrance to the D-rings was replaced. The existing hollow metal door with louvers was replaced with a jail cell type door. The jail cell door is fabricated from vertically oriented stainless tube steel spaced on 8" centers. The bottom of the door frame is 8" above the floor. Several short vertical bars extend 4" below the bottom frame as an additional barrier to unauthorized personnel access. The new door design provides a large open area for the passage of liquid and debris.

Note: There is only one entrance to the D-rings on the 281' elevation.

During the initial evaluations of potential flow paths in containment, it was believed that water could flow through the Incore Instrument Trench and return to the sump via the incore chase area doorway as discussed in the 2005 submittal. More detailed reviews performed as part of the debris transport analyses identified that a concrete incore tube support block restricts flow through the trench. There is no significant flow path through the incore chase area door and no door modifications were performed.

NRC Question 23

Please describe how potential blockage of the fuel transfer canal drain was evaluated, including likelihood of blockage. The response mentions the existence of a 4" drain line in the fuel transfer canal which drains to the recirculation sump. Please summarize how any temporary water holdup is integrated into the overall analysis.

TMI Unit 1 Response:

Given the large surface area of the transfer canal, the flow rate of water across the canal floor, and thus the ability to transport debris to the drain, is low. To provide additional assurance that the drain line would not become clogged with debris, a trash rack was installed over the top of the drain. The trash rack design includes an 8" stub pipe that extends above the FTC floor.

Given the low flow rate across the canal floor, and the presence of the trash rack and 8" stub pipe, the likelihood of blockage of the drain line is considered to be very low. The volume of water that would collect in the deep end of the FTC due to the presence of the 8" stub pipe is accounted for as a hold up volume in the calculation to determine the minimum

containment water level.

NRC Question 24

The supplemental response notes that the following drain lines were redirected to the new normal sumps: Four-inch FTC [Fuel Transfer Canal] drain downstream of valve SF-V-31 Two-inch Reactor cavity drain line discharging through WDL-V-520. Two other four-inch embedded RB [Reactor Building] floor drain lines One-half inch leak off drain line from SF-V-24

Please provide information on how the flow from these drain lines, and any other lines which drain to the normal sumps, is integrated into the overall sump water level analysis. If these Page 30 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 drain flows are credited on the sump water level evaluation, please provide the basis for ensuring that these drain lines will not become blocked by debris during a LOCA.

TMI Unit 1 Response:

1. 4" FTC drain downstream of valve SF-V-31

The 4" FTC drain line provides a flow path for water to return to the sump/recirculation pool from the deep end of the FTC. The discharge pipe downstream of valve SF-V-31 was directed to the normal sump as this line is used for controlled draining of the last few feet of refueling water from the FTC. Following refueling activities, SF-V-31 is locked open to provide the recirculation flow path from the FTC to the RB sump. Water returning to the sump via this drain line is available for recirculation as the containment water level is above the top of the normal sumps.

A trash rack with an 8" stub pipe was installed over the inlet to the FTC drain line to

ensure the flow path does not become blocked with debris. Given the low flow rate across the canal floor, and the presence of the trash rack and 8" stub pipe, the likelihood of blockage of the drain line is considered to be very low. The volume of water that would collect in the deep end of the FTC due to the presence of the 8" stub pipe is accounted for as a hold up volume in the calculation to determine the minimum

containment water level.

2. 2" reactor cavity drain line discharging through WDL-V-520.

This drain line provides a flow path from the area under the reactor vessel to the normal sumps. Valve WDL-V-520 is closed prior to plant heatup. Flow through this drain line is not credited in the sump water level analysis.

3. Two other 4" embedded RB floor drain lines.

The RB floor drain system returns water to the RB sump via two 4" drain lines embedded in the RB floor. The floor drain system is not credited in the sump water analysis. Water from the upper levels of containment can still return to the sump pool via stairwells, gratings, and other openings. The containment water level calculation does account for water that will pool on the floors at the 308' and 346' elevations as part of the total

holdup volume.

4. 1/2" leak off drain line from SF-V-24.

This 1/2" packing leak-off line for SF-V-24 does not factor in to the sump water level

analysis.

NRC Question 25 The supplemental response notes that a fuel and vessel downstream effects evaluation prepared by AREVA (using WCAP-16406-P) to estimate the effect of core blockage on core cooling results in a calculated cladding temperature of less than 904 degrees Fahrenheit (F).

The response also states that the core chemical effects evaluation using LOCADM spreadsheet software calculated a peak fuel cladding temperature of 439 degrees F. The acceptance Page 31 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 criterion for cladding temperature in WCAP-16793-NP, Revision 0, is 800 degrees F, as noted in the supplemental response. Please describe how these differing results are reconciled and if a calculated temperature above 800 degrees F is found to be acceptable, please provide the basis for this conclusion. Specifically, if the calculated temperature exceeds 800 degrees F, include cladding strength data for oxidized and pre-hydrided cladding material that exceeds this

temperature limit.

TMI Unit 1 Response:

In early 2005, AmerGen contracted with AREVA to evaluate the downstream effects of debris ingested from the containment sump on the fuel for TMI Unit 1. At the time, the only detailed guidance available for an evaluation of this type was provided in Chapter 9 of WCAP-16406-P, Revision 0. Based on this guidance the evaluation included the following areas of interest:

The fluid velocities in the RV lower plenum were evaluated to assess the potential for debris lift to the core inlet; A review of the flow paths from the ECCS injection location to the RCS through the reactor internals components (excluding the fuel) was performed; The maximum pressure drop that could be tolerated at the core inlet was calculated using conservative boundary conditions; and Alternate flow paths (around the core inlet) were assessed.

WCAP-16406-P did not include guidance for evaluating blockage of the internal spacer grids.

However, a calculation method was developed and included to proactively address this concern.

Although it is improbable that a buildup may occur that completely blocks a fluid sub-channel at the inlet to a spacer grid around a single rod, an evaluation was performed to demonstrate that even for this situation the clad would be successfully cooled. An assessment of a complete blockage was done by considering a solid plug around the limiting fuel pin at the peak power location as illustrated in Figure 3.

Fuel Clad Spacer GridBlockage 2L T s,2 T s,1 T q x q Figure 3- Illustration of a solid plug around the limiting fuel pin at the peak power

location Page 32 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 The calculation considered the blockage to be a one-dimensional solid with uniform energy generation per unit volume. The following assumptions were made in the calculation:

1. It was assumed that all of the energy originating in the fuel pellet was transferred radially to the clad. That is, axial conduction along the fuel pellet was ignored.

2. The clad is cooled by axial conduction only. That is, the surface of the plug is assumed to be totally insulated. Conduction was assumed to progress axially along the cladding to the edges of the plug.

3. The fluid temperatures above and below the blockage were the same such that Ts,1 =

Ts,2 = Ts.

4. Further, the maximum temperature was at the blockage centerline such that x=0.

The cladding temperature at the centerline of the blockage can then be determined by the conduction equation:

s.T kLqT)(T2 0 2 0 where = energy generation per unit volume

.q K = thermal conductivity of cladding L = half the length of blockage

T s = fluid sink temperature

For axial conduction along the cladding, the heat transfer area is the cross section of the cladding. The conduction length is half of the plug thickness. The energy deposited in the cladding inside the plug is calculated based on the core power at the peak power location and the core DH. The maximum analyzed LHR limit for the TMI Unit 1 core was used. The sink temperature is the fluid saturation temperature at the core pressure. The calculation was performed assuming a complete blockage of 1" in length, which resulted in a cladding temperature at the blockage centerline of 904 degrees F.

Subsequent to the completion of this calculation, additional guidance and calculations for blockage at intermediate spacer grids was developed by the PWROG and issued in WCAP-16793-NP, Revision 0. The approach for determining the cladding temperature at spacer grids is described in Section 2.2 and Section 4. A calculation method was also presented to determine the cladding temperature as chemical deposits built up on the cladding surface (LOCADM). These calculations are fundamentally different from the calculation described above. In particular, they considered radial conduction through the cladding and the debris to the core fluid with a conservative thermal conductivity for the debris buildup. The result is a demonstration that the calculation described above (which neglects radial conduction and only considers axial conduction) is overly conservative.

The calculation that determined a cladding temperature of 904 degrees F was an early attempt to proactively address a potential concern. The formulation was overly conservative. Subsequently, additional calculations were performed that conservatively determined the cladding temperature to be 439 degrees F. There is no expectation that the cladding temperature will exceed 800 F, and the cladding temperature of 439 degrees F calculated using the calculations described in WCAP-16793-NP take precedence over the calculations that produced a temperature of 904 degrees F. Page 33 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 NRC Question 26

The supplemental response states that aluminum-based precipitates do not form above 140 degrees F, but it does not provide data to support this assertion. Please provide the test data that forms the basis for this assertion.

TMI Unit 1 Response:

The statement on page 57 of the GL 2004-02 SR states, "The 30 day integrated testing also identified that aluminum based precipitates do not form until the post-LOCA environment has cooled to below 140 degrees F. The prototype testing used these results to sequence the WCAP-16530/16785 based precipitates."

The basis for this statement is the results of the 30-day integrated testing performed by Alion at the VUEZ test facility and a number of other collaborating studies/e xperiments performed by others in support of the chemical effects issue. It should be pointed out that the statement is specific to the conditions evaluated for TMI Unit 1 and is not generic in nature.

During the Alion VUEZ 30-day integrated chemical effects testing, the following data was obtained regarding head loss, temperature and time. Two experiments were performed at a pH of 8.5 and 8.0 (Figures 4 and 5, respectively). From these experiments it is evident that the increases in pressure drop significantly increased beyond those attributable to viscosity between 59 degrees C (138 degrees F) and 50 degrees C (122 degrees F) for the pH 8.5 but no increase in head loss was noted for the pH of 8.0. The differences between the two experiments are only pH. It has been shown through numerous studies that for aluminum based precipitates a reduction in temperature (and pH) can trigger precipitation. From the graph provided in Figure 4 below for the pH of 8.5, a chemical effect is occurring below 140 degrees F (60 degrees C).

The ICP analyses for these two (2) experiments are provided in Figures 6a/6b and 7a/7b below.

The differences in the corrosion rates for the aluminum in the experiment are clearly evident between the two solution pHs and consistent with the body of knowledge regarding the corrosion of aluminum versus pH. TMI Unit 1 has a sump pH of 8.0; the pH of 8.5 experiment was performed as a conservative sensitivity. The concentration of aluminum is approximately 10 ppm and silicon is between 13 - 16 ppm, respectively for the pH 8.0. It should be pointed out that both are decreasing slightly over time. The pH 8.5 shows an aluminum concentration of greater than 100 ppm and a silicon concentration of approximately 20 ppm.

The time dependent results would indicate that no chemical effects on head loss occur at these lower concentrations of aluminum (10 ppm) associated with the solution pH of 8.0 but did occur at concentrations greater than 100 ppm at temperature below 140 degrees F (60 degrees C).

To compare these results with other industry results, Alion reviewed the data from WCAP-

16785. Section 5.4, Solubility of Aluminum and Calcium Precipitates, specifically Section 5.4.1, provides additional test insight into the precipitation and solubility of sodium aluminum silicate. Page 34 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 These (WCAP-16785) tests were performed with sodium silicate and aluminum nitrate and varying concentration to verify the point of precipitation at temperatures between 200 degrees F and 140 degrees F.

"Long-term observation of TSP buffered solutions containing 79 ppm aluminum and 236 ppm silicon show precipitate formation within 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br />. Solutions containing 10, 20, 30, 40, and 50 ppm aluminum, with corresponding concentrations of silicon were then prepared for additional long-term observation.

No precipitates were observed to form in TSP buffered solution containing up to 40 ppm aluminum during a 30 day observation period-

Based on these results, the solubility limit of sodium aluminum silicate in TSP buffered solution at 140 degrees F to 200 degrees F is 40 ppm aluminum and 119 ppm silicon. Thus under these circumstances, sodium aluminum silicate will not precipitate until the aluminum concentration is above 40 ppm."

These WCAP results are consistent with the results obtained from the 30-day integrated chemical head loss experiments provided at VUEZ.

TMI Unit 1 performed testing with WCAP-16530 calculated chemical surrogates implementing a protocol that added these precipitates to simulate the chemical effects that would occur at 140 degrees F and below based on the results of both the WCAP and the VUEZ testing. The quantity of chemical precipitates was determined by the WCAP-16530 methodology for a pH of 8.0 without any refinements including phosphate inhibition that through numerous tests has been shown to occur at this pH. The output of WCAP-16530 for the TMI Unit 1 specific pH of

8.0 provides

an aluminum concentration of 22 ppm of aluminum and 28 ppm silicon which conservatively resulted in 327 lbs of gelatinous sodium aluminum silicate and 112 lbs of aluminum oxyhydroxide for the testing. Although the tank testing performed with the WCAP predicted precipitates produced a head loss that was considerable and found manageable, it is clear from the WCAP and the VUEZ testing, that no chemical effects would be expected to occur with aluminum levels associated with the TMI Unit 1 environment.

In summary, TMI Unit 1 only expects a potential impact from calcium phosphate and not from aluminum bearing precipitates as these are not in sufficient concentration to affect head loss based on testing performed by VUEZ and precipitation testing performed by Westinghouse.

TMI Unit 1 has been conservative in applying these precipitates at a temperature (140 degrees F) and operationally reduced the flow rate accordingly to minimize the impact these precipitates would have on the structural integrity of the sump strainer assembly.

Page 35 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 Figure 4- VUEZ Experiment for TMI Unit 1 Environment (pH 8.5)

Page 36 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 Page 37 of 40 Figure 5- VUEZ Experiment for TMI Unit 1 Environment (pH 8.0) 0100200300400500600700800 0 50100150200250300 weight concentration / ppmtime / hour L4 Al L5 Al L6 Al L6R Al 0100200300400500600700800 0 5 10 15 20 25 30 35 40 45 50 weight concentration / ppmtime / hour L4 Si L5 Si L6 Si L6R SiFigure 6a: Aluminum Concentration (pH 8.5)

Figure 6b: Silicon Concentration (pH 8.5)

Figure 7a: Aluminum Concentration (pH 8.0)

Figure 7b: Silicon Concentration (pH 8.0)

Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 NRC Question 27 Please explain what test parameters (e.g. visual, pressure drop) were measured to determine that no precipitates were formed above 140 degrees F, and explain whether it is possible that precipitates formed at temperatures above 140 degrees F but were not detected during the test.

TMI Unit 1 Response:

The test parameters that were measured were temperature, head loss and elemental concentration (ICP analysis) for the entire 30-day Vuez experiment. In addition to these parameters, visual observation was recorded over the entire experiment and no visual precipitates were noted. This was also independently confirmed in similar bench top experiments. The chemical effects analysis and test results did support calcium based precipitates formed at temperatures above 140 degrees F due to retrograde solubility. The aluminum precipitates in the pH 8.5 test were evident by a linear increase in head loss most likely associated with solubility, kinetics and filtration (see Figure 4 above in response to RAI No. 26). However, at the reduced pH of 8.0 for TMI Unit 1, the reduced aluminum corrosion, and the stability of the aluminum phosphate layer, supported the observation of no significant chemical effects in this environment (see Figure 5 above in response to RAI No. 26). It was concluded that calcium based precipitates did form at temperatures greater than 140 degrees F through ICP analysis and Scanning Electron Microscopy/Electron Dispersive Spectroscopy analysis. Head loss calculations used the head loss attributed to calcium phosphate to determine the head loss across the strainer at temperatures greater than 140 degrees F when the NPSH margin is limiting. Head loss calculations used the head loss attributed to aluminum and calcium precipitates at temperatures at or less than 140 degrees F when structural integrity of the screen is limiting. With regards to aluminum based precipitates, these may have formed but were insufficient to be seen visually or detected by pressure drop through the debris bed at a pH of 8.0. As explained in the response to RAI No. 26 above, these most likely did not

precipitate.

NRC Question 28 Please compare the test loop pH to the expected equilibrium containment pool pH following a LOCA. Discuss any differences between the pH values in terms of potential effects on aluminum solubility.

TMI Unit 1 Response:

A conservatively high representation of pool pH was used in the VUEZ test loops to determine the desired water chemistry of the test. Therefore, the pH was not manipulated in the chemical array testing.

The initial acidic conditions that result from borated water supplied through the building spray nozzles were identified to yield a pH of 4.5 for a duration of 4 hours4.62963e-5 days <br />0.00111 hours <br />6.613757e-6 weeks <br />1.522e-6 months <br />. For the VUEZ integrated tests, Alion performed an analysis to determine the necessary additional time that this pH must be maintained in order to compensate for the difference between the VUEZ test loop temperature and the actual containment temperature. From the Alion analysis it was determined that the initial pH of 4.5 must be maintained for a total of 7.7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br />. This length of time at a low pH ensures that the necessary conditions are established for chemical compounds to form and chemical processes, such as the oxidation of aluminum, to occur. At the end of this Page 38 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 initial test period, the pH of the test loop was adjusted in one step to achieve a target pH level of approximately 8.5 (Vuez Seq. 1) and 8.0 (Vuez Seq. 3). This value was selected to provide a slightly higher upper bound to the predicted plant pH value of less than 8.0. The upper limit pH occurs in the plant from the initial dissolution of the TSP buffer utilized in the modified plant configuration. The test apparatus closely matched this value by control of the test pool pH using the TSP buffer solution. The Vuez Seq. 3 test pH profile consisted of two pH values: 4.5 from 0 to 7.7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> after the postulated accident, and 8.0 from 7.7 hours8.101852e-5 days <br />0.00194 hours <br />1.157407e-5 weeks <br />2.6635e-6 months <br /> to the accident conclusion. It should be noted that the temperature compensat ion analysis performed by Alion also required that the experiment be extended by an additional 9.6 hours6.944444e-5 days <br />0.00167 hours <br />9.920635e-6 weeks <br />2.283e-6 months <br /> to compensate for the temperature differential between the predicted plant conditions and the test at the higher pH of 8.5 (Vuez Seq. 1). NRC Question 29 In the supplemental response, Table 19, sodium aluminum silicate settled volumes, do not meet the criteria in the USNRC's SE for WCAP-16530 for "aluminum containing precipitate" (1 hour1.157407e-5 days <br />2.777778e-4 hours <br />1.653439e-6 weeks <br />3.805e-7 months <br />, >6ml). Given this discrepancy, please explain why the settlement values used for TMI Unit 1 testing are acceptable.

TMI Unit 1 Response:

Table 7- The settling characteristics of the sodium aluminum silicate Sodium Aluminum Silicate Batch Settled Volume (mL) Date Tested Date Used Time from date tested to date used (days) Na 246 7.0 11/5/2007 11/16/2007 11 Na 247 8.0 11/5/2007 11/16/2007 11 Na 248 7.2 11/5/2007 11/16/2007 11 Na 249 9.2 11/5/2007 11/16/2007 11 Na 250 8.5 11/5/2007 11/16/2007 11 Na 251 9.5 11/5/2007 11/16/2007 11 Na 252 8.0 11/5/2007 11/16/2007 11 Na 253 8.5 11/5/2007 11/16/2007 11 Na 254 5.5 11/5/2007 11/16/2007 11 Na 255 6.0 11/5/2007 11/16/2007 11 Na 256 5.5 11/5/2007 11/16/2007 11 Na 257 5.5 11/5/2007 11/16/2007 11 Na 258 7.0 11/5/2007 11/16/2007 11 Na 259 7.1 11/5/2007 11/16/2007 11 Na 260 6.0 11/5/2007 11/16/2007 11 Na 261 5.5 11/5/2007 11/16/2007 11 Na 262 5.0 11/5/2007 11/16/2007 11 Na 263 6.1 11/5/2007 11/16/2007 11 Batches of precipitants are typically made in maximum volumes of 5 gallons (approximately 19L) when large quantities of precipitates are required for testing. All batches are combined when added to the test tank. Therefore, the final settling volume for all batches combined is the average of the individual settling volumes.

Taking the average of all the individual settled volumes added to the test tank yields an average settled volume of 6.95 ml if reported with two significant figures (>6ml).

Page 39 of 40 Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02 NRC Question 30 The NRC staff considers in-vessel downstream effe cts to be not fully addressed at TMI-1 as well as at other pressurized water reactors. The TMI-1 submittal refers to draft WCAP-16793-NP, "Evaluation of Long-Term Cooling Considering Particulate, Fibrous, and Chemical Debris in the Recirculating Fluid." The NRC staff has not issued a final SE for WCAP-16793-NP.

AmerGen may demonstrate that in-vessel downstream effects issues are resolved for TMI-1 by showing that the TMI-1 plant conditions are bounded by the final WCAP-16793-NP and the corresponding final NRC staff SE, and by addressing the conditions and limitations in the final

SE. AmerGen may also resolve this item by demonstrating, without reference to WCAP-16793 or the staff SE that in-vessel downstream effects have been addressed at TMI-1. In any event, AmerGen is requested to report how it has addressed the in-vessel downstream effects issue within 90 days of issuance of the final NRC staff SE on WCAP-16793.

TMI Unit 1 Response:

TMI Unit 1 will report how it has addressed the in-vessel downstream effects issue within 90 days of issuance of the final NRC staff SE on WCAP-16793. Page 40 of 40

ATTACHMENT 2 Three Mile Island Unit 1 Summary of Regulatory Commitments

Three Mile Island Unit 1 Response to Request for Additional Information Related to Generic Letter 2004-02

Summary of Regulatory Commitments The following table identifies commitments made in this document.

COMMITMENT TYPE COMMITMENT COMMITTED DATE OR "OUTAGE" ONE-TIME ACTION (Yes/No) PROGRAMMATIC (Yes/No) TMI Unit 1 will report how it

has addressed the in-vessel downstream effects issue Within 90 days of issuance of

the final NRC

staff SE on WCAP-16793.

Yes No Page 1 of 1