ML13323B202

Attachment 8: CHLE-015, Rev. 3, Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal.

ML13323B202
Person / Time
Site:	South Texas
Issue date:	01/15/2013
From:	Howe K Southern Nuclear Operating Co
To:	Office of Nuclear Reactor Regulation
Shared Package
ML13323A673	List: ML13323B187 ML13323B191 ML13323B194 ML13323B196 ML13323B198 ML13323B199 ML13323B200 ML13323B201 ML13323B202 ML13323B204 ML13323B207 ML13323B208 ML13323B209 NOC-AE-13003040, Attachment 1: CHLE-005, Rev. 1, Determination of the Initial Pool Chemistry for the Chle Test.
References
GSI-191, NOC-AE13003040, STI 33762096, TAC MF2400, TAC MF2401 CHLE-015, Rev 3
Download: ML13323B202 (46)

Text

INOC-AE-1 3003040Attachment 8CHLE-01 5: Summary of Chemical Effects Testing in 2012 for STP GSI-1 91 License Submittal PROJECT DOCUMENTATION COVER PAGEDocument No: CHLE-015 I Revision:

3 1 Page 1 of 45Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Project:

Corrosion/Head Loss Experiment (CHLE) Program Date: 21 Jan 2013Client: South Texas Project Nuclear Operating CompanySummary/Purpose of Analysis or Calculation:

Corrosion/Head Loss Experiment (CHLE) tests were performed to support the risk-informed resolution ofGSI-191 at the South Texas Project Nuclear Operating Company (STP). This document presents theresults of testing, calculations, and analysis conducted during 2012 to evaluate the extent to whichchemical precipitates may contribute to additional blockage on emergency core cooling system (ECCS)strainers beyond what is caused by non-chemical debris.The results of this analysis suggest that it is reasonable to conclude that chemical precipitates areunlikely to have an additional contribution to strainer blockage following ruptures on pipes up to 6 inchesin diameter.

Although the results are more preliminary, this outcome may also be extended to breaks onpipes up to 15 inches in diameter.

Role: Name: Signature:

Date:Prepared by: Kerry Howe < signed electronically

> 11/14/2012 UNM review: Janet Leavitt < signed electronically

> 1/22/2013 STP review: Ernie Kee < signed electronically

> 11/20/2012 Soteria review: Zahra Mohaghegh

< signed electronically

> 11/26/2012 Revision Date Description 1 11/14/2012 Draft document for internal reviewAddressed internal review comments.

Submitted to NRC for public2 11/26/2012 meigI meeting.3 1/15/2013 Addressed internal review comments, added material from LBLOCA testand changed name of document.

____ I. ____ I _____________________

I. +

Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Table of ContentsIntroduction

.....................................................................................................................................

3Review of ECCS Sure-Flowrm Strainer Module Testing for the STP Plant ..............................

4Calculated Release of Aluminum,

Silicon, and Calcium in a Medium Break LOCA underR ealistic C onditions

........................................................................................................................

6General Experimental Approach for the CHLE Tests .............................................................

10Corrosion and Precipitation in a Medium Break LOCA in the 30-day CHLE Test .................

14C orro sion ...................................................................................................................................

16Concentration of Metals in Solution and Evidence for Precipitation

...................................

18H ead L oss ..................................................................................................................................

25Corrosion and Precipitation in a Large Break LOCA in the 30-day CHLE Test ....................

27Evaluation of Debris Bed Response to WCAP Precipitate Addition

.......................................

33Evaluation of Long-Term Aluminum Nitrate Addition to Debris Beds ...................................

38Sum m ary and Conclusions

.....................................................................................................

41R eferences

.....................................................................................................................................

44Document No: CHLE-015, Rev 3 Page 2 of 45Document No: CHLE-01 5, Rev 3Page 2 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Introduction The purpose of this report is to describe the results from calculations, experiments, and analysisconducted as part of the Corrosion/Head Loss Experimental (CHLE) program.

The CHLEtesting is being conducted at the University of New Mexico (UNM) to investigate the effects ofchemical precipitates on Emergency Core Cooling System (ECCS) strainer blockage underrealistic conditions for the South Texas Project (STP) in support of the risk-informed resolution of Nuclear Regulatory Commission (NRC) Generic Safety Issue (GSI) 191. Three objectives within the CHLE program are to determine, for realistic conditions, (1) the rates of corrosion andrelease of chemical reactants and products from materials present in the containment building ata nuclear power plant (2) whether or not chemical precipitates can form in the post loss-of-coolant accident (LOCA) environment, and (3) the effect on ECCS strainer debris bed head lossfrom any observed chemical products.

Four sets of tests are described in this report:1. A series of tests were performed to examine the characteristics and performance of twotypes of fiberglass debris beds when prepared with varying amounts of debris and whensubjected to chemical precipitants that are formed using the WCAP- 16530-NP protocol[1]. The two types of fiber bed preparation methods included disaggregating fibers bypressure-washing using the protocol advocated by Nuclear Energy Institute (NEI) [2] orchopping fibers in a blender.

This testing program included 14 tests and was conducted from 22 May 2012 to 07 June 2012.2. Two multi-day tests that evaluated the response of two types of debris beds whenexposed to aluminum nitrate injected directly into the tank solution to simulateprecipitation that may occur when aluminum enters the solution at a slow rate, as it woulddo via corrosion.

These tests were conducted from 28 June 2012 to 24 July 2012.3. A 30-day experimental simulation of a LOCA on a 6-inch pipe under conditions that arerepresentative of the STP plant. For the purposes of the risk-informed resolution of GSI-191, pipe breaks have been divided into small break LOCAs (SBLOCA, 2-inch breaksand smaller),

medium break LOCAs (MBLOCA, 2-inch to 6-inch breaks),

and largebreak LOCAs (LBLOCA, greater than 6-inch breaks).

This test simulated the upperbound of the MBLOCA class. The test included aluminum specimens in a corrosion tankand fiberglass debris in three parallel head loss columns.

This test was conducted from22 Aug 2012 to 25 Sept 2012.4. A 30-day experimental simulation of a LOCA on a 15-inch pipe (an intermediate-sized LBLOCA) under conditions that are representative of the STP plant. In addition to thematerials present in the medium break LOCA, this test also included a concrete

specimen, galvanized steel coupons, and zinc granules.

This test was conducted from 5 Oct 2012 to8 Nov 2012.In addition, this report summarizes earlier testing of the ECCS Sure-FlowTM strainer modulesubjected to prototypical debris from the STP plant by Alden in 2008 [3,4]. The Alden testingwas conducted using bounding conditions representative of a design basis LOCA, based oncalculations performed for STP [5] for this condition.

The Alden testing forms the prior basis forDocument No: CHLE-015, Rev 3 Page 3 of 45Document No: CHLE-015, Rev 3Page 3 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal the effect of chemical precipitates on head loss at the STP plant and is a useful point ofcomparison for the results from the CHLE program.The 30-day MBLOCA and LBLOCA tests are the primary tests with regard to the effect ofchemical precipitates on the resolution of GSI-191 under risk-informed conditions.

The resultsof the other two testing campaigns provide background information on the utility of usingvarious types of beds for chemical effects testing and are discussed at the end of this report.Review of ECCS Sure-FIowTM Strainer Module Testing for the STPPlantTesting was conducted at Alden Research Laboratory, Inc. over a 3-day period in July 2008 toevaluate the head loss across a Performance Contracting, Inc (PCI) Sure-FlowTM strainer modulesubjected to prototypical debris from the STP plant [3,4]. The test module had a screen area of91.44 ft2. The tank had a volume of 1830 gallons and flow was recirculated at 353 gpm, whichprovided a screen velocity of 0.0086 ft/s and a circulation time of 5.2 minutes.

Two tests wereconducted, one with no debris that established the baseline head loss through a clean strainer, and a second one with a combination of non-chemical debris and chemical precipitates.

Thetarget temperature for the tests was 120 'F but the measured temperature was somewhat lower.At 0.0086 ft/s, the normalized head loss (normalized to 116.3 'F) through the clean strainer was0.092 ft (1.1 inches).

During the test with debris addition, the water temperature ranged from102 'F to 112 'F (38.8 to 44.4 °C).The quantity of debris was scaled to the surface area of the screens in the STP plant with twoECCS trains in operation.

The ECCS screens in the STP plant have a surface area of 1815.5 ft2each, for a total area of 3,631 ft2.Thus, the quantity of debris in the Alden test was scaled to be2.52 percent of the quantity in a design-basis LOCA at STP. The quantity of non-chemical debris in the Alden test included 6.6 Ibm of Nukon fiberglass in various sizes, 8.55 Ibm of Knaufthermal wrap fiberglass insulation in various sizes, 0.95 Ibm of Microtherm powder, 4.65 Ibm ofMarinite Board powder, 3.65 Ibm of latent particulate dirt and dust, 34.45 Ibm of tin, 14.95 Ibmof pulverized acrylic coatings, and 2.75 Ibm of acrylic chips. These materials were added to thetest system in 4 batches over about 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> and water was allowed to circulate overnight beforechemical precipitates were added to the system. The head loss peaked at about 5 ft of watercolumn after the non-chemical debris was added and declined to about 4 ft by the following morning.The quantity of chemical precipitates added during the test was determined using calculations fora design-basis LOCA at the STP plant [5]. Calculations based on the WCAP-16530-NP protocolpredict the release of 63.06 kg of calcium, 208.85 kg of silicon, and 82.92 kg of aluminum whenthe water in the pool is at the maximum level and the spray systems are operated continuously for 30 days in a design basis LOCA [5]. The design basis for the quantity of precipitates issummarized in Table 1. The amount of silicon and aluminum released into solution is reducedwhen the water is at the minimum level, which decreases the amount of sodium aluminumsilicate (NaAISi3Os) and increases the amount of aluminum oxyhydroxide (AIOOH) that isformed. By selecting the maximum water level to determine NaAlSi3O8 formation and theminimum water level to determine A1OOH formation, the test evaluated a quantity of aluminumDocument No: CHLE-01 5, Rev 3Page 4 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal precipitates that were greater than could be generated from the aluminum available under anysingle condition.

As permitted in WCAP-16530-NP, Alden substituted AlOOH precipitate for NaA1Si308precipitate (on a mass basis). Using the recipes in WCAP-16530-NP, the AIOOH was preparedusing 117.2 kg of aluminum nitrate [AI(NO3)3 9H20] and 37.6 kg of sodium hydroxide (NaOH)added into 450.6 gallons of tap water. The calcium phosphate

[Ca3(PO4)2] precipitate wasprepared using 7.27 kg of calcium acetate [Ca(C2H302)2zH20] and 10.48 kg of trisodium phosphate (Na3PO4" 12H20) added into 226 gallons of tap water.Precipitate addition started after water had been circulating through the strainer covered with thenon-chemical debris for approximately 18 hours2.083333e-4 days <br />0.005 hours <br />2.97619e-5 weeks <br />6.849e-6 months <br />. The AIOOH precipitate was added in 20batches and the Ca3(PO4)2 precipitate was added in 18 batches.

The test plan called for theAIOOH and Ca3(PO4)2 precipitates to be added separately with typically 1 or 2 pool turnovers between one batch and the next. The first 23 batches of precipitates were added to the tank overabout 5.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> on Day 2 of the test. Water continued to circulate overnight (about 17 hours1.967593e-4 days <br />0.00472 hours <br />2.810847e-5 weeks <br />6.4685e-6 months <br />),and then the remaining 15 batches of precipitates were added over about 5.3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> on Day 3 ofthe test.The chemical addition resulted in the ECCS strainer module being thoroughly coated withchemical precipitates, as shown in Figure 1 [4]. The maximum normalized head loss (excluding clean strainer head loss and piping losses) at 0.0086 ft/s with chemical precipitates was 8.75 feet(normalized to 116.5 'F). Immediately prior to chemical

addition, the head loss through thestrainer loaded with non-chemical debris was about 4.0 feet. Thus, the additional head loss dueto chemical precipitates was 4.75 ft or an increase of 2.2 times over the head loss withoutchemical precipitates.

Table 1 -Quantities of Precipitates used for Preparing the Prototype Strainer ModuleTesting for STP ____________

____________

Peci&tat QuianWv (k Water 0~ 1 IIINaAISi308 649.5 MaximumAIOOH 64.9 MinimumCa3(PO4)2162.7 MaximumDocument No: CHLE-015, Rev 3 Page 5 of 45Document No: CHLE-01 5, Rev 3Page 5 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Figure 1 -Precipitates covering a PCI Sure-FlowTM strainer module after testing for adesign basis LOCA at the STP. In this test, the loading rate on the strainer was 990 g/m2for aluminum and 190 g/m2 for calcium (reproduced from [41).To compare the results of experimental testing programs to full-scale performance, the chemicalquantities can be scaled based on either pool volume and strainer area. Dividing chemicalquantities by pool volume defines the concentration of chemicals in solution, which is relevantbecause the quantity of precipitates that form is affected by the relationship between actualconcentration and the solubility limitations of the relevant precipitates.

Dividing chemicalquantities by the screen area is relevant because the head loss will be influenced by the actualprecipitate loading on the screen. Based on pool volume, the chemical additions in the Aldentest resulted in an aluminum concentration of 1,220 mg/L as Al and a calcium concentration of240 mg/L as Ca. Based on screen loading, the chemical quantities resulted in a screen loading of990 g/m2 of Al and of 190 g/m2 of Ca.Calculated Release of Aluminum,

Silicon, and Calcium in a MediumBreak LOCA under Realistic Conditions The quantity of precipitates used in the Alden testing as discussed in the previous section wasbased on a design basis LOCA and several conservative assumptions.

The calculated amounts ofDocument No: CHLE-01 5, Rev 3Page 6 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal precipitates change when the quantities of materials in containment and assumptions are basedon realistic conditions.

A summary of the quantities of aluminum,

silicon, and calcium predicted by the WCAP-16530-NP calculation procedure to be released into solution is shown in Table 2.The results presented in this table demonstrate that a more realistic assessment would generatelower concentrations of chemicals in solution.

Comparing Case 0 and Case 4 demonstrates thatusing realistic values for the water and aluminum in containment, running the ECCS spraysystems for the proper duration, and adjusting the calculation to reflect the fact that the marinitehas been removed from containment results in 70 percent less aluminum, 20 percent less silicon,and 65 percent less calcium than was predicted for a design basis LOCA. Furthermore, applyingthe quantities of concrete and fiberglass exposed by a LOCA on a 6-inch cold leg break (Case 7)results in a 97 percent reduction in aluminum, 93 percent reduction in silicon, and 96 percentreduction in calcium from the baseline condition.

As an addition reduction in debris generation, little or no Microtherm debris is generated for most 6-inch pipe breaks because this insulation material is only present in the bioshield wall penetrations.

A pipe break would have to be in thevicinity of a wall penetration to generate any Microtherm debris, and the quantity of Microtherm in any one penetration is minimal.

An important cause of the reduced release of chemicals is thechange in the temperature profile.

The temperature profile of the design basis LOCA isconsiderably higher than for a MBLOCA, and as an additional conservatism the temperature profile was increased by an additional 5 'F when the corrosion product quantities were calculated for the design basis LOCA [5]. The temperature profile of the original STP calculation and theprofile for a MBLOCA are shown in Figure 2.The quantities of chemicals released into solution during a MBLOCA, based on calculations from the WCAP-16530-NP procedure, are 1.70 kg of Al, 2.36 kg of Ca and 13.16 kg of Si. Theconcentration of aluminum and calcium in solution and loading rate onto a strainer issummarized in Table 3, along with the corresponding values that were tested at Alden asdescribed in the previous section.

As is evident in this table, the concentrations and loading ratesfor calcium and aluminum in a MBLOCA are substantially lower than the values for a designbasis LOCA or the values that were tested at Alden.The concentrations of aluminum and calcium during a test are significant if the concentrations are near the saturation concentrations of the relevant precipitates.

If the concentrations aresignificantly above the saturation limits, the concentrations will not necessarily affect the totalhead loss achieved during a head loss test but may affect the rate at which head loss accumulates.

The concentrations of aluminum and calcium in the Alden tests were higher than predicted for adesign basis LOCA because the water volume in the test system was not scaled to the watervolume in the STP containment pool during a LOCA. The water volume in the Alden test was1830 gal, whereas the water volume that would correspond to the ratio of ECCS screen areas inthe STP plant and Alden test would be 16,830 gal. This difference in water volumes resulted inconcentrations in solution that were 9.2 times greater and a pool turnover time that was 9.2 timesshorter in the Alden tests than would be calculated to be present in a design basis LOCA at theSTP. While the difference in concentration may not have a significant difference in total headloss for a design basis LOCA, a longer pool turnover time would have slowed the rate at whichprecipitates were deposited on the strainer and would have greatly increased the amount of timenecessary to conduct the tests.Document No: CHLE-015, Rev 3 Page 7 of 45Document No: CHLE-01 5, Rev 3Page 7 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Table 2 -Summary of quantities of Aluminum,

Silicon, and Calcium Released from theSTP Plant during a variety of LOCA conditions.

082.92208.8563.06Baseline calculation for a design basis LOCA [5].Same as baseline but eliminates 15.2 ft3 of calsil insulation fromthe calculation due to the removal of marinite from the STPcontainment building.

Marinite was originally installed on thereactor vessel nozzles but has been removed.Same as Case 1 but adjusts the aluminum quantity from 7,000 ft22 67.54 184.91 28.58 to 5,567 ft2 to be consistent with the best estimate of the quantityof aluminum in containment

[6].Same as Case 2 but adjusts the spray duration to 6.5 hour5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br />s3 25.41 185.07 28.51 instead of assuming that the sprays run for the entire 30 dayduration.

Same as Case 3 but adjusts the pool water volumeA from 89,3504 24.56 169.48 23.16 ft3 to 73,400 ft3, which is the best estimate of the pool watervolume during MBLOCAs and LBLOCAs [7].Same as Case 4 but adjusts the quantity of concrete from 5,7005 24.56 169.45 23.09 ft2 to 1,446 ft2, based on the best estimate of exposed concretein a LBLOCA [8]. Note that the contribution of chemicals fromconcrete dissolution is minor compared to other materials.

Same as Case 5 but adjusts the quantity of Nukon from 2,385.16 19.72 21.76 2.38 ft3 to 60 ft3, based on the best estimate of NUKON fiberglass debris generated in a 6-inch cold-leg break MBLOCA [9].Same as Case 6 but adjusts the pool temperature profile fromthe base condition to a 6-inch break on a cold-leg pipe [10] andadjusts the pH after TSP addition is complete from 7.49 to 7.20,7 17 1 which is the average pH measured over 30 days in the CHLEMBLOCA test. This case is a prediction of the quantities ofmaterials that would be released into solution due to corrosion and dissolution during a 6-inch cold-leg break MBLOCA at theSTP plant using realistic conditions and assumptions.

Same as Case 7 but eliminates the concrete (1,446 ft2) andmicrotherm (1.8 ft3). This case is a prediction of the quantities ofmaterials that would be released into solution due to corrosion under the conditions tested in the CHLE test, which did notinclude concrete and microtherm.

The calculated contribution from the concrete is 0.01 kg of Ca and 0.007 kg of Si.Based on density of the containment sump solution at 185 OF, 60.957 Ibm/ft3,Document No: CHLE-01 5, Rev 3Page 8 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 300-Design Basis LOCA250 -MBLOCA-200S150EI- 100500 5 10 15 20 25 30Time (days)Figure 2 -Temperature profiles of (a) the design basis LOCA used to determine thequantities of precipitates for prototypical strainer testing and (b) a MBLOCA of a 6-inchcold leg break.The more important factor in head loss testing is the screen loading rate. The screen loading ratein the Alden test was the same as calculated for a design basis LOCA for calcium, but higherthan calculated for a design basis LOCA for aluminum.

The higher screen loading rate foraluminum resulted from two factors.

First, the design condition for the Alden tests used thequantity of NaAlSi308 precipitate generated for the maximum water level condition and theAIOOH precipitate generated for the minimum water level condition.

The increase of AIOOHprecipitate in the minimum water condition is caused by a reduction in the amount of siliconavailable to form NaAlSi308 precipitate, not by an increase in the amount of aluminum available.

Thus, the same aluminum in solution contributes to one precipitate in the maximum watercondition and the other precipitate in the minimum water condition, but the two precipitate quantities were added together in determining the total amount of aluminum precipitates to usefor testing.

Second, AIOOH precipitate was used as a substitute for NaAISi308 precipitate (asallowed in the WCAP-16530-NP protocol),

but the result of substituting AIOOH for NaAlSi308precipitate on a mass basis increases the total amount of aluminum nitrate that must be added tothe system. As given in the recipes in WCAP-16530-NP

[1], is necessary to add 625 g ofaluminum nitrate for every 100 g of AlOOH, but only 143 g of aluminum nitrate for every 100 gof NaAISi3O8.By substituting AIOOH for NaAISi308, the total amount of aluminum that wasadded to the system was increased.

Document No: CHLE-015, Rev 3 Page 9 of 45Document No: CHLE-015, Rev 3Page 9 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Table 3 -Concentration and loading rates of aluminum and calcium under design andmedium break LOCAs at STP.Concentration of aluminumadded (mg/L as Al)0.8432.91,220Loading rate of aluminum onECCS strainer (g/m2 as Al) 5.0 250 990Concentration of calcium added 1.16 24.9 240(mg/L as Ca)Loading rate of calcium onECCS strainer (g/m2as Ca) 7.0 190 190General Experimental Approach for the CHLE TestsThe CHLE program was designed to experimentally evaluate the impact of chemical precipitates on the risk-informed resolution of GSI-191 under realistic conditions at STP. The overall testprogram was envisioned to include several components, including (1) a limited set of 30-daytests to investigate the overall chemical effect scenario in an integrated fashion under realistic conditions, (2) bench-scale tests to investigate variability in corrosion and release rates as a resultof variability in chemical concentrations and other factors during a LOCA, and (3) vertical-column head loss tests to identify the extent of head loss under varying debris bed conditions using precipitates generated in-situ based on information generated during the 30-day tests. Atest plan [ 11] for the experimental program was developed and modified several times as theprogram evolved, as results were generated, and as additional information became available.

Notall components of the test program were completed during 2012. Key tests that were completed included two 30-day tests, which simulated a MBLOCA (6-inch break) and LBLOCA (15-inchbreak), along with preliminary tests that evaluated the performance of two types of fiberglass debris beds.The CHLE test equipment for the 30-day tests consists of a material corrosion tank and threevertical head loss modules.

The 304 stainless steel tank, as shown on the left in Figure 3, isnominally 4 ft x 4 ft x 6.6 ft and has a bottom that slopes to a centrally-located discharge port.Although the inside of the tank is not physically divided into different compartments, the tankcan be visualized as containing upper and lower sections.

The upper section of the tank isdesigned to accommodate all vapor space materials in containment that contribute to chemicaleffects through exposure to containment sprays. The lower section of the tank is designed toaccommodate solution and materials that may be submerged in containment.

A removable coverand gantry crane allow for placing and removing samples into the desired areas of the tank. Thetank is not pressurized and is therefore limited to testing below the normal boiling point of water.Document No: CHLE-015, Rev 3Page 10 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal The tank is insulated and contains two titanium-jacketed 3.5-kW rod-type heaters in the tankpool to maintain the temperature of the solution at a maximum of 185 OF (85 °C) +/- 5 OF.Solution can enter the tank through injection headers in the lower portion of the tank (submerged area) and/or through spray nozzles in the upper portion of the tank (non-submerged area).Additional details of the experimental equipment are described in [12].The three identical vertical head loss modules, as shown in Figure 4, are designed to operate inparallel when connected to the corrosion tank or individually when isolated from the corrosion tank. Each has an independent pump so that the flow rate to each column and velocity to thescreen area can be separately controlled.

The nominal velocity through the screen area duringtesting was 0.01 ft/s. The upper and lower portions of the 6-in diameter vertical head lossmodules are constructed of stainless steel and are sealed at the top with a blind flange. The blindflanges can be removed to introduce debris into the head loss assembly.

The upper section ofpiping also has an air vent to allow gas to be vented from the head loss assembly, if necessary.

The middle section of the assembly is constructed of 1/4-in thick polycarbonate to allow view ofthe debris bed (6 inches below and 16 inches above) with a perforated plate supported by a ringlocated 6-in from the bottom. The perforated plate is constructed of stainless steel and contains0.094-inch holes in the same pattern as on the STP strainers.

There are vents above and belowthe screen support to assist in post-test activities.

A differential pressure (DP) transducer is pipedto ports above and below the screen support to measure the pressure loss through the debris bed.Each module has a dedicated flow meter to monitor flow to the column.Figure 3 -CHLE tank.Document No: CHLE-015, Rev 3Page 11 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Figure 4 -CHLE Head Loss Columns.As noted earlier, tests to simulate strainer head loss can be scaled to full-scale installations basedeither on containment pool volume or on screen area. Head loss testing is appropriately scaledon the basis of screen area so that screen loading rates can be maintained.

For the effects ofchemistry,

however, it is necessary to scale the quantities of materials in the system on the basisof containment pool volume, so that the concentrations of species in solution in the CHLEsystem would properly simulate the concentrations that would occur in the full-scale installations.

Matching concentrations is important because the kinetics and outcome of chemical reactions depends on the concentrations of the reactants present.

However, it was also necessary toconsider the ratio of screen area to ensure that the CHLE tests did not result in a screen loadingrate that was below the rate that would occur in the STP plant. Based on pool volumes, thevolume of water in the STP containment during a LOCA is 1,790 times greater than the volumein the CHLE system, and all materials used during testing were scaled to this ratio. However, thescreen area in the STP plant when two ECCS trains are in operation is 3,631 ft2, compared to0.59 ft2 in the C1HLE system with three 6-inch diameter

screens, resulting in a ratio of 6,160.Thus, the CHLE system was conservatively designed with less screen area than would beproperly scaled to the STP plant to provide a screen loading rate greater than that which wouldoccur with the same quantity of precipitates in the STP plant. The increase in screen loading ratebased on the design of the CHLE head loss columns is a factor of 3.4.An important consideration is that the fiber debris beds that are used in the chemical effecttesting be suitable surrogates for debris that would be formed during a LOCA. Attributes thataffect the suitability of a particular debris bed design include the stability of the debris bed, theDocument No: CHLE-015, Rev 3Page 12 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal reproducibility of the results, and the ability of the debris bed to participate in chemicalinteractions under a variety of conditions.

Debris beds used in some previous GSI-191 work arenot necessarily applicable to the current study for three reasons.

First, the approach velocities historically used in head loss testing were more than an order of magnitude higher than the STPstrainer design and testing was often conducted with tap water and at room temperature.

Second,the historical observations were typically for short periods compared to the CHLE investigations.

Third, the NEI recently developed a debris preparation method [2] that is believed to beprototypical of debris formed during a LOCA, and most previous head loss testing have usedother debris preparation methods.

The Nuclear Regulatory Commission (NRC) reviewed theNEI plan and noted it is generically an acceptable way of producing debris, but declined toofficially endorse it as the only way to produce acceptable debris because of the dependence onhuman actions [13].Two types of fiber bed preparation methods were evaluated for possible use within the CHLEprogram.

First, the recent debris formulation advocated by NEI for strainer testing involvesbaking fiber blankets on one side at 300 °C for 6 to 8 hours9.259259e-5 days <br />0.00222 hours <br />1.322751e-5 weeks <br />3.044e-6 months <br />, followed by disaggregation with acommercial pressure-washer; this method is referred to as the NEI pressure-washing method inthis report. Second, fiber blankets were subjected to the same baking procedure, but wereseparated by fine chopping of fibers in a blender.

The NEI pressure-washed fiber, whenintroduced into a head loss column, results in a light and fluffy debris bed. The blender-processed fiber results in a denser bed with a more uniform top surface.

Photographs of fibersafter the two preparation methods are shown in Figure 5, and the respective debris beds areshown in Figure 6. Performance of the debris beds with respect to head loss and the ability tocapture precipitates is discussed later in this report.A BFigure 5 -Examination of debris on a light table from (A) blended fiber preparation, and(B) NEI pressure-washed fiber preparation.

Document No: CHLE-015, Rev 3Page 13 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal AFigure 6 -Debris beds in head loss columns at the beginning of test operation.

(A) Column2 blended fiber debris bed and (B) Column 2 NEI pressure-washed fiber debris bed.Chemicals concentrations representative of the containment pool at STP during a LOCA wereused throughout the testing program.

Water was deionized to a conductivity less than 50 ýLS/cmusing a reverse osmosis system. Chemicals included 250.5 mM boric acid (2,710 mg/L as B),8.87 mM trisodium phosphate (3,370 mg/L as Na3PO4" 12H20), and 0.061 mM lithium hydroxide (0.42 mg/L as Li). Hydrochloric and nitric acids were added periodically over the duration of the30-day tests to simulate acid production from radiolysis, with maximum concentrations of 0.812mM HCI and 0.229 mM HNO3.The basis for the chemical concentrations is described inadditional detail in [14].Corrosion and Precipitation in a Medium Break LOCA in the 30-dayCHLE TestA long-term test was conducted from 22 Aug 2012 to 25 Sept 2012 to simulate the interaction between materials in containment and the containment pool solution during a 6-inch break on acold-leg pipe in the STP containment building.

The test included materials in the corrosion tankand solution that continuously circulated through the tank and through fiberglass debris beds onthe screens in the three vertical head loss columns.The overall duration of the test was 34 days. The NEI pressure-washed debris beds were presentin all 3 columns during the first 30-day period and the blender-prepared debris beds were presentin all 3 columns for the last 4 days. The NEI pressure-washed debris bed was selected for the30-day test because the blender-prepared debris beds were not sufficiently reproducible in aprevious test series to allow chemical precipitates to be adequately detected (described later inthis report).

However, preliminary tests also demonstrated that the blender-prepared debris bedswere more effective at capturing chemical precipitates circulating in the pool solution, and therewere concerns that the NEI pressure-washed debris beds may not be sufficiently sensitive.

OnDay 30, the columns were isolated from the tank, the NEI debris beds were removed, andblender-processed beds were installed.

Fresh TSP-buffered borated solution was circulated Document No: CHLE-015, Rev 3Page 14 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal through the blender-processed debris beds (isolated from the solution in the corrosion tank) toestablish baseline behavior for 2 days. The columns were then linked to the corrosion tank andthe test solution was allowed to circulate through the blender-processed debris beds for anadditional 2 days. Thus, if precipitates had been formed and were circulating in the pool solutionafter 30 days of simulated LOCA duration undetected by the NEI pressure-washed debris beds,the blender-processed debris beds would provide an additional opportunity to detect the presenceof precipitates in the pool solution.

During the 30-day test, the head loss columns each contained 18 g of fiberglass debris tomaintain consistency with the debris beds from previous tests. The column approach velocitywas maintained at 0.01 ft/s. An analysis using Containment Accident Stochastic Analysis(CASA) determined that the appropriate amount of fiberglass debris scaled to the volume of theCHLE pool volume was 36.5 g; thus, the 54 g (total) of fiber present in the columns was about50 percent greater than the quantity required to realistically simulate a 6-inch break. To maintainthe proper ratio of aluminum to silicon for corrosion and dissolution

purposes, the portion of thealuminum that was allocated to the submerged portion of the tank was also increased by 50percent.

Thus, while the total amount of aluminum was correctly scaled for a 6-inch break,increasing the fiberglass and the submerged portion of the aluminum provided somewhatconservative test conditions with respect to realistic conditions for a 6-inch break.The material in the tank consisted of 3.11 ft2 of aluminum scaffolding that was obtained from theSTP plant, with 15 percent submerged and 85 percent exposed to the containment sprays. Thealuminum scaffolding had been in use at the STP plant for a long time and therefore wasrepresentative of actual conditions at the STP plant, including being made of the appropriate aluminum alloy and containing any relevant pre-existing scale and surface coatings.

About 90percent of the aluminum in STP's containment is scaffolding.

The aluminum scaffolding waswashed to try to remove any latent debris or coatings that could become dislodged during the 30-day test and interfere with weight-loss measurements for aluminum corrosion, but withoutremoving any corrosion products that could affect the rate of corrosion.

The test was run using only aluminum and fiberglass so that if precipitates did form, then thecharacteristics of the precipitates could be compared to precipitates from a previous test in whichaluminum nitrate was injected into the tank solution at a slow rate, which is described later inthis report.The temperature profile for the test was determined by running MELCOR and RELAP-5 andwas shown previously in Figure 2 [10]. The chemicals in the containment solution were asdescribed in the previous section; the boric acid and lithium hydroxide were added before the teststarted, the TSP was added over a 65 minute period near the start of the test to simulate thedissolving of TSP in baskets in the containment

building, and the hydrochloric and nitric acidswere added periodically during the 30 day period. The sprays were operated for 6.5 hours5.787037e-5 days <br />0.00139 hours <br />8.267196e-6 weeks <br />1.9025e-6 months <br /> at thebeginning of the test. Additional details of the test conditions are available in [15] and completetest results are available in [23].Document No: CHLE-015, Rev 3Page 15 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Corrosion The rate of aluminum corrosion can be assessed by measuring the weight change of thealuminum specimens before and after the test. The weight of specimens both before and afterthe test consists of aluminum metal and corroded scale layers. An ASTM standard acid-washing procedure

[16] was used to assess the quantity of scale present on the specimens.

Acomplication is that the chemical formula of the scale is not known, so it is therefore not trivialto determine how much of the scale weight is due to aluminum and how much is due to otherelements.

The aluminum scaffolding provided by STP is non-homogenous with unknownconstituents from years of use, which remained after cleaning.

The specimens were cut to sizefor testing, cleaned with mild laboratory soap to remove latent debris, and allowed to dry, asshown in Figure 7. It should be noted that is it not possible to determine whether the pre-testwashing procedure removed all latent debris and other materials that may have been washedfrom the surface and contributed to weight loss during the test. The samples that were to besubmerged were from the side of the scaffolding and had a different texture and appearance thanthe samples cut for the vapor space. Therefore, unused samples taken from locations similar tothat shown in Figure 7 (vapor space and submerged samples) were analyzed using X-rayphotoelectron spectroscopy (XPS) and scanning electron microscope (SEM) with energydispersive X-ray spectroscopy (EDX) in efforts to evaluate the original scale composition.

XPS analysis of unused pieces of the pre-test samples detected the presence of multiplechemical species.

Interpretation of the analysis determined that two scale types, aluminumphosphate and aluminum oxide/aluminum hydroxide, were present on pre-test samplescorresponding to both the vapor space and submerged specimens.

Aluminum oxide andaluminum hydroxide have the same binding energy so it is not possible using XPS to tell whichspecies is present.

Using the pre-test and post-test information for masses of the aluminumsamples, XPS results, and weight measurements of the scale layer that were removed per ASTMstandards

[ 16], an experimental aluminum corrosion mass under STP conditions was calculated.

The mass was calculated assuming two different scenarios for the composition of the scale,based on the presence of aluminum phosphate in combination with either aluminum oxide oraluminum hydroxide.

The weight measurements are summarized in Table 4.Figure 7 -STP aluminum scaffolding used in this test.Document No: CHLE-015, Rev 3 Page 16 of45Document No: CHLE-015, Rev 3Page 16 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Table 4 -Change in distribution of scales between post- and pre-test sample. Case 1 wascalculated assuming aluminum phosphate-aluminum hydroxide scale. Case 2 wascalculated assuming an aluminum phosphate-aluminum oxide scale.Vapor834.111.411.59833.521.611.780.380.400.200.19Submerged 1 63.36 0.10 0.13 63.21 0.13 0.17 0.13 0.12 0.03 0.04Submerged 2 63.72 0.10 0.13 63.56 0.13 0.18 0.13 0.12 0.03 0.04Total Mass 961.19 1.62 1.86 960.29 1.87 2.21 0.56 0.55 0.26 0.27Table 5 -Summary of measured aluminum weight lossISubmerged 1 7.1 1 4.818245IExposed to sprayT 2.4 N/A 1.4 103The total aluminum metal loss due to corrosion was calculated to be 0.59 g as Al from thespecimens exposed to containment spray and 0.31 g as Al from the submerged specimens.

Thealuminum metal lost as a function of surface area is summarized in Table 5. For comparison, corresponding values from ICET Test #2 and WCAP calculations are also shown in Table 5.The submerged specimens in the CHLE MBLOCA test had somewhat more aluminum weightloss than in ICET Test #2 [17]. However, considering that the ICET test used fresh aluminumcoupons of a known alloy rather than old scaffolding, was conducted at a constant temperature of 140 'F for 30 days, and did not include a scale removal procedure during the corrosion weightloss measurements, the difference in corrosion rate is not dramatic.

The submerged coupons inthe CHLE and ICET tests has a lower rate of corrosion than predicted for a MBLOCA by theWCAP calculation, indicating that a small amount of passivation or inhibition of the aluminumsurface may be operative.

The specimen exposed to spray in the CHLE test had a somewhat higher corrosion rate thancalculated by the WCAP protocol for a MBLOCA, but several factors may have contributed tolow accuracy on the vapor specimen calculations.

First, it was not possible to execute the acid-Document No: CHLE-01 5, Rev 3Page 17 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal washing procedure on a specimen of the size of the vapor space specimens, so the procedure wasperformed on a small piece and the percent weight loss due to scale removal was assumed to bethe same for the entire vapor space specimen, even though the corrosion products present on thescaffolding clearly were not homogeneous.

Second, the measured differences in weight due tocorrosion and due to scale removal were very small so that relative errors in measurements couldbe significant.

Third, other materials (latent debris) may have been present on the specimen andsubsequently washed off during the spray period, which would have been measured as a weightloss due to aluminum corrosion even though it would in fact have been due to other materials.

Because of these factors, the difference in the corrosion rates for the aluminum exposed to sprayis not considered significant.

A significant outcome of this analysis is the measured rates in the CHLE are less than 3 percentof the expected corrosion rates in a design basis LOCA, indicating that considerably lessaluminum would be available to contribute to potential chemical precipitates.

Concentration of Metals in Solution and Evidence for Precipitation Several parameters were measured to determine whether precipitates formed in solution.

First,turbidity of the pool water was measured on a daily basis. The trend over the course of the test isshown in Figure 8. The turbidity was the highest,

-0.6 NTU, on the first day of testing andgradually decreased over the thirty days of testing.

This trend is similar to tests that did notcontain corrosion specimens in the tank. The turbidity increased when the blender-prepared debris beds were linked to the tank on Day 32 of the test, likely due to fiber shedding from thenew debris beds. Previous testing showed a good correlation between turbidity and aluminumhydroxide precipitates, as shown in Figure 9. The lack of increase in turbidity over the testsuggests that precipitates did not form in the solution.

1.0&Tank0.8 ePre-Heat Exchanger 0*Post-Heat Exchanger 0.6 AAA2o0. A4 '*AII 1 4A ,I* A I A h A S A=A0.4 -., I~A A A0.20.0 ** ,l* ** ' ,;. .0 5 10 15 20 25 30 35Time (day)Figure 8 -Turbidity over the duration of the 30-day MBLOCA CHLE test.Document No: CHLE-015, Rev 3Page 18 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal A second indicator of precipitation is the difference between filtered and total concentrations ofaluminum,

silicon, and calcium.

The concentrations in solution are measured by inductively coupled plasma optical emission spectrometry (ICP-OES),

with and without filtration throughmembrane filters with a nominal pore size of 0.45 Vim. The concentrations over the test areshown in Figures 10, 11, and 12, respectively.

The similarity between filtered and totalconcentrations suggests that no precipitates larger than 0.45 gtm in diameter were circulating insolution.

3025S201510500 5 10 15 20 25 30 35 40 45Aluminum Added (mg/L)Figure 9 -Correlation between measured turbidity in solution and amount of aluminumadded to the CHLE tank in the form of aluminum nitrate.1.00.8E 0.6 *Filtered Aluminum0 O=Total Aluminum0.4Q 0.20.0 ....0 5 10 15 20 25 30 35Time (day)Figure 10 -Total and filtered concentration of aluminum over the duration of the 30-dayMBLOCA CHLE test.Document No: CHLE-015, Rev 3Page 19 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 7.06.05.0C" 4.0 1 l1l9l i * *

• C 3.00 *Filtered Silicon0mTotal Silicon1.00.0 , .i I, I I0 5 10 15 20 25 30 35Time (day)Figure 11 -Total and filtered concentration of silicon over the duration of the 30-dayMBLOCA CHLE test.3.0*Filtered Calcium2.5 ETotal Calciumj2.0 m -*C. 1.51.00Co 0.50 .0 ......,0 5 10 15 20 25 30 35Time (day)Figure 12 -Total and filtered concentration of calcium over the duration of the 30-dayMBLOCA CHLE test.In each case, the concentration became relatively constant after a few days of testing.

Theaverage concentration of each species over the last 20 days of the test is summarized in Table 6.For comparison, the concentration predicted by the WCAP calculation procedure for aMBLOCA are also presented in Table 6. The concentrations of aluminum and silicon aresomewhat lower than predicted by the WCAP corrosion rate calculation procedure, and theDocument No: CHLE-015, Rev 3Page 20 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal calcium concentration is marginally higher. In all cases, however, the measured concentrations are not dramatically different from the concentrations predicted by the WCAP corrosion ratecalculation procedure.

The lack of precipitation products can be assessed by comparing the concentrations measured insolution to the saturation concentration of the relevant precipitates.

Work conducted at ArgonneNational Laboratory (ANL), reproduced in Figure 13, suggests that the precipitation boundaryfor amorphous aluminum hydroxide is a good indicator of the ability of precipitates to form inpast GSI-191 head loss testing.

Using Visual MINTEQ [ 18], the precipitation boundaries foraluminum hydroxide and calcium phosphate as a function of temperature and pH are shown inFigures 14 through 17. Figures 14 and 15 show the concentration of soluble aluminum in asolution that is in equilibrium with amorphous AI(OH)3 solid, and Figures 16 and 17 show theconcentration of soluble calcium in a solution that is in equilibrium with amorphous Ca3(PO4)2solid (included in the Visual MINTEQ database as species Ca3(PO4)2 am-2). The saturation concentrations were calculated for initial conditions of 250.5 mM H3B03, 8.87 mM P04-3, 26.61mM Na+1, 0.061 mM Li+', 0.334 mM C1-', and 0.812 M NO3+1 to reflect the chemicals that wereadded in the MBLOCA test. The figures demonstrate that the concentrations measured in theCHLE tests were below calculated saturation concentrations for these precipitates, except whencalcium phosphate was above 70 'C at pH = 7.2, which was the measured pH in the MBLOCAtests. The corresponding value of the 'pH + p[Al]-' value in the CHLE MBLOCA is 12.2 and isalso shown on Figure 13.Although equilibrium speciation models such as Visual MINTEQ can be helpful inunderstanding solution

behavior, it should be noted that the chemistry of the containment poolsolution during a LOCA is extremely complex.

The concentrations of the relative speciesdepends on the presence of other species in solution (boron, sodium, etc.), pH, temperature, ionicstrength, and a host of other factors including nonidealities, and that many of these factorsinteract with each other. Thus, thermodynamic modeling can be used as a general indication ofthe state of a chemical systerm under specified conditions, but should be used cautiously andshould not be construed as being an absolute indicator of the concentrations that would causecertain species to precipitate.

Nevertheless, the thermodynamic modeling predictions aregenerally supportive of the results obtained in this test.Table 6 -Summary of measured and calculated concentration in a MBLOCA at STP.I Aluminum0.280.82Calcium 1.8 1.13Silicon 4.7 6.14Document No: CHLE-01 5, Rev 3Page 21 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Tempeture

('FM60 s0 100 120 140 160 10 200 22012.51211.5z0. 10.5109.5920 30 40 50 W0 70 W0 90Tempwatre CC)100Figure 13 -"Al hydroxide precipitation map in the 'pH + p[AI]T' vs. temperature domainbased on ANL's bench top and loop test data and literature data." Reproduced from Bahn,et al., 2011 [191. The red line shows the pH + p[AIIT value corresponding to the resultsfrom the CHLE MBLOCA test.Document No: CHLE-015, Rev 3Page 22 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 1000:rE01001010.14 5 6 7 8 9 10pH11Figure 14 -Equilibrium saturation concentration of aluminum for the formation ofamorphous aluminum hydroxide

[AI(OH)31 as a function of pH. Values given for WCAPand CHLE are the concentrations in solution at the end of a 30-day event as a result ofcorrosion.

1000a00Cc0Ei1001010.1020 40 60 80Temperature (C)100Figure 15 -Equilibrium saturation concentration of aluminum for the formation ofamorphous aluminum hydroxide

[AI(OH)31 as a function of temperature.

Values given forWCAP and CHLE are the concentrations in solution at the end of a 30-day event as aresult of corrosion (temperature varies during the event).Document No: CHLE-015, Rev 3Page 23 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 100-Am-2 T=40-Am-2 T=60-A--2 T=85E a Predicted by WCAP= 10.o

• Measured in CHLE.2C0E4 5 6 7 8 9 10 11pHFigure 16 -Equilibrium saturation concentration of calcium for the formation of calciumphosphate

[Ca3(P04)21 as a function of pH. Values given for WCAP and CHLE are theconcentrations in solution at the end of a 30-day event as a result of corrosion.

100-pH = 7.0-pH = 7.22 -pH =7.5E0000_ Measured in CHLE MBLOCA testPredicted by WCAP-16530-NP 0.1 ...0 20 40 60 80 100Temperature (C)Figure 17 -Equilibrium saturation concentration of calcium for the formation of calciumphosphate

[Ca3(PO4)2J as a function of temperature.

Values given for WCAP and CHLEare the concentrations in solution at the end of a 30-day event as a result of corrosion (temperature varies during the event).Document No: CHLE-015, Rev 3Page 24 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Head LossThe head loss through the debris beds over the 30-day test is shown in Figure 18. Head loss wasnormalized to a temperature of 20 'C using corrections for both the density and viscosity ofwater. The normalized head loss was less than 1 inch over the duration of the test, and did notincrease during the 30-day period. As will be discussed later in this report, however, the NEIpressure-washed beds would be unlikely to detect precipitates at the concentration that occurredin this test.The ability for the available aluminum to cause head loss in a debris bed can be assessed bycomparing the results of this test to previous head loss testing.

Based on the measured aluminumconcentration of 0.28 mg/L, the maximum loading that would occur on the CHLE strainers would be 5.8 g/m2 as Al. As noted earlier, the CHLE strainers were designed with less surfacearea than would be proportional to the STP plant to ensure a conservative loading rate. If thecorresponding concentration were applied to the STP strainers, the loading rate would 1.7 g/m2as Al.2.0-Column 1S1.5 -Column 2-Column 3MI1.oo 0.50.00 5 10 15 20 25 30Time (day)Figure 18 -Normalized head loss through NEI pressure-washed debris beds during the 30-day MBLOCA CHLE test.Document No: CHLE-015, Rev 3 Page 25 of 45Document No: CHLE-015, Rev 3Page 25 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal

-0.035I 0.03MPAtApproximate maximum AlS20 -loading rate measured in 0.02 ~CHLE MBLOCA test10 NUKON AddOM Ils~~1050 l(D0 1602 21 3WThre (mini)Figure 19 -"Pressure drop and screen approach velocity versus time history in a loop testusing the WCAP aluminum hydroxide surrogates" reproduced from Bahn et al, 2009 [201.The quantity marked with the arrow represents approximately the maximum loading ratethat can be achieved in a MBLOCA at the STP plant, using either the measured corrosion quantities in the CHLE test or the calculated corrosion quantities in the WCAP protocol.

This loading rate can be put in context by comparing it to tests conducted at ANL. Results fromthe ANL testing are reproduced in Figure 19. The point on the graph labeled as 0.5 ppmcorresponds to a loading rate of 4.4 g/m2 as Al based on the surface area of the screens in theANL test loop. Rapid head loss did not occur until the loading rate reached 13.3 g/m2 as Al.Similarly, in small-scale chemical effect testing described in NUREG/CR-6868, additional headloss due to chemical precipitates did not begin to occur until the aluminum concentration was 2.7mg/L and the loading rate was 3.5 g/m2 as Al [22]. Thus, the loading rate, even if all aluminummeasured in the CHLE MBLOCA test were to have precipitated, would be below the quantities that were necessary to cause significant head loss in the ANL testing.

The fiber for the ANLdebris beds was prepared by coarse shredding, followed by blending in a household blender onhigh-ice crush mode for 11 seconds, followed by mixing in a beaker with a magnetic stirrer for10 minutes.The loading rates can also be compared to strainer module testing conducted at Alden described earlier in this report. The total loading rates during the Alden testing were substantially higherthat those described here, reaching a maximum aluminum loading rate of 990 g/m2 as Al.However, it is also important to note that a significant increase in head loss was observed in theAlden testing as soon as the first batch of aluminum precipitates was added, suggesting thatsmall amounts of precipitate (compared to the total amount) can cause significant head loss. Theloading rate for the first batch of aluminum precipitates in the Alden testing,

however, was 77g/m2. Since the loading rate for the first precipitate batch was higher than that which causedDocument No: CHLE-015, Rev 3 Page 26 of 45Document No: CHLE-015, Rev 3Page 26 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal rapid head loss in the ANL tests, an increase in head loss in the first batch in Alden test isconsistent with the other results reported here. The amount of aluminum released in the CHLEMBLOCA test corresponds to less than one-tenth of the loading of the first batch of aluminumprecipitates in the Alden test, so the Alden strainer module testing cannot be used to predict theamount of head loss that may occur under MBLOCA conditions.

Corrosion and Precipitation in a Large Break LOCA in the 30-dayCHLE TestA second 30-day test was conducted from 5 Oct 2012 to 8 Nov 2012. This test simulated a largebreak LOCA (a 15-inch pipe break, which could be considered an intermediate-sized LBLOCA)under conditions that are representative of the STP plant. Like the previous long-term test, thetest was conducted for 34 days, with the NEI pressure-washed debris beds in the three head losscolumns for the first 30 days and the blender-prepared debris beds in the columns for the last 4days. Complete results are available in [24].In addition to the materials present in the medium break LOCA, this test included a concretespecimen, galvanized steel coupons, and zinc granules that simulated failed inorganic coatings.

The test also included a greater amount of fiber debris and a different temperature profile basedon predictions for a break of this size.The temperature profile used for the test is shown in Figure 20. The temperature profile includeda higher peak temperature than the MBLOCA test (219 'F versus 193 'F), but the time above thetest equipment maximum temperature of 185 'F was only 1.3 minutes.

Because of this shortduration, the additional corrosion that could occur at temperatures above the maximum testtemperature was accommodated by operating the CHLE system at above the LBLOCA profilefor a sufficient period of time to cause additional corrosion.

250200a. 150ECLE 100500 ., ....' .., ..0 5 10 15 20 25 30Time (day)Figure 20 -temperature profile for the CHLE LBLOCA Test.Document No: CHLE-015, Rev 3Page 27 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 1.2ATank1.0 0 Pre-Heat Exchanger

• Post-Heat Exchanger 5" 0.8.0.6A0.4 S&AAIA A AA0.2 1A A A**8* G A ftAAAG AA0.0 i i i .0 5 10 15 20 25 30 35Time (day)Figure 21 -Turbidity over the duration of the 30-day LBLOCA CHLE test.The turbidity results from the LBLOCA test are shown in Figure 21. The turbidity follows thesame general trend as that which was observed in the MBLOCA test; in this test, turbidity peaked just above 1 NTU after the spray phase was complete and declined to near 0.13 NTUnear the end of the 30-day period. The higher turbidity at the beginning of the test was likely dueto the additional latent debris introduced into the tank because of the inclusion of additional materials present in the tank. The lack of increase in turbidity over the test suggests thatprecipitates did not form in the solution.

The total and filtered concentrations of aluminum,

calcium, silicon, and zinc over the test areshown in Figures 22 to 25. In each cases, the total and filtered concentration for each constituent are similar to each other, which suggests that no precipitates larger than 0.45 pam in diameterwere circulating in solution.

Throughout the LBLCOA test, the aluminum concentration waslower than in the MBLOCA test, with a concentration that exceeded 0.1 mg/L in only onesample near the beginning of the test. It should be noted that these results are below thequantification limit for these analyses (which is 0.2 mg/L), so it would not be appropriate to infertrends of increasing or decreasing concentration over the duration of the test based on theseresults.

The calcium concentration was around 1.7 mg/L, which is similar to the results obtainedin the MBLOCA test. The similarity of the calcium concentration between the two tests isconsistent with information that indicates that the leaching from concrete is a small source ofcalcium compared to other materials in containment (cal sil and fiberglass).

The maximumsilicon concentration in the LBLOCA test was 2.7 mg/L, which, like aluminum, is lower than inthe MBLOCA test. Finally, the zinc concentration increased over the first 10 days of the test andthen remained relatively constant at 0.6 mg/L for the duration of the test.Document No: CHLE-015, Rev 3Page 28 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 0.15* Al TotalMAl FilteredCCa,U0.10 +.moo*EII. j*SU0.05 +a0.000 5 10 1! 20 25 30 3505101520Time (day)2530 35Figure 22 -Total and filtered concentration of aluminum over the duration of the 30-dayLBLOCA CHLE test.2.0-- 1.5C.2CU 0.5 *Ca TotalECa Filtered0.0 ....0 5 10 15 20 25 30 35 40Time (day)Figure 23 -Total and filtered concentration of calcium over the duration of the 30-dayLBLOCA CHLE test.Document No: CHLE-015, Rev 3Page 29 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 3.02.5 ."-aE 2.0.2 1.5Q 1.0Uo 0 0Si TotalESi Filtered0.0 1 ..i0 5 10 15 20 25 30 35Time (day)Figure 24 -Total and filtered concentration of silicon over the duration of the 30-dayLBLOCA CHLE test.0.80.6 mfmalag *piiP me *~0.4U Zn TotalCo *lZn FilteredU0.0 .l ... 1 ., ... .0 5 10 15 20 25 30 35 40Time (day)Figure 25 -Total and filtered concentration of zinc over the duration of the 30-dayLBLOCA CHLE test.In contrast to the MBLOCA test, an increase in head loss was observed in the LBLOCA test.The increase in head loss was similar in magnitude for both the NEI pressure-washed debris bedsand the blender-processed debris beds. The head loss trend for the NEI pressure-washed debrisbeds is shown in Figure 26. All columns experienced an increase of head loss; Column 2 had thelargest final head loss final value of -2.5 inches of water, while Column 1 had the smallest headloss value of 0.4 inches of water. Column 3 had a final head loss measurement of 1.2 inches ofwater. The increase in head loss in the blender-processed debris beds over a 2-day period, asshown in Figure 27, was similar to the increase in head loss during the first 2 days of operation Document No: CHLE-015, Rev 3Page 30 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 3.02.5 -Column 1-Column 22.0 -Column 31.5S 0.50.00 5 10 15 20 25 30Time (day)Figure 26 -Normalized head loss through NEI pressure-washed debris beds during the 30-day LBLOCA CHLE test.2.0-Column 11.6i .-Column 2F -Column 3i 1.20.8a OA0.0)30 31 32 33 34 35Time (day)Figure 27 -Normalized head loss through blender-prepated debris beds during the 4-dayperiod at the end of the LBLOCA CHLE test.Document No: CHLE-015, Rev 3Page 31 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal of the NEI pressure-washed debris beds. The base-line head loss measurements of all blender-processed debris beds experienced little to no change during the 2 days of isolation (shown as thetime between 30 and 32 days in Figure 27), suggesting that a constituent in the water in the tankwas responsible of causing the increase in head loss. For both types of beds, however, theoverall increase in head loss was small, measuring in just inches of head loss at the end of thetesting.The constituents that contributed to the increase in head loss cannot be identified with certainty.

One possibility is the zinc granules that were used to simulate the failed inorganic zinc coatings.

The zinc granules were enclosed in a stainless steel mesh bag with the intention of allowingwater to circulate through the zinc material without allowing the zinc to circulate through thesystem, which could contribute to head loss. However, some of the zinc granules did escapefrom the mesh bag and were evidently circulating with the tank solution, as evidenced byobservation of zinc granules on in-line filters that were used to filter solution from the tankperiodically, as shown in Figure 28.Zinc corrosion products were observed upon the conclusion of the LBLOCA test. Whitematerials were observed on discrete areas of the submerged galvanized steel coupons, as shownin Figure 29-A. Similar materials were observed on the tank floor immediately below thesubmerged galvanized steel coupons.

After testing, a similar material was also observed onsmall areas of the mesh enveloped which contained the zinc granules and the zinc granulesthemselves.

This material was not visibly evident on the fiber debris beds. It should be notedthat similar deposition of white materials were observed on the inorganic zinc coated coupons inICET Test 2, which had chemical conditions similar to these tests.A BAFigure 28 -SEM images of (A) zinc granules contained within stainless steel mesh bags,and (B) product captured on an in-line membrane filter after 24 hours2.777778e-4 days <br />0.00667 hours <br />3.968254e-5 weeks <br />9.132e-6 months <br /> of testing.

EDXanalysis of capture material indicates it is likely zinc granules.

Document No: CHLE-015, Rev 3 Page 32 of 45Document No: CHLE-01 5, Rev 3Page 32 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal The white material was scraped from the galvanized steel plates and subjected to scanningelectron microscopy (SEM) with energy dispersive X-ray spectrometry (EDX), X-rayphotoelectron spectrometry, inductively-coupled plasma spectrometry (ICP) following aciddigestion, and X-ray diffraction (XRD) analysis.

The analyses indicate that the material is acrystalline form of a zinc phosphate solid similar to the mineral hopeite with the chemicalcomposition Zn3(PO4)2"4H20. A SEM image of the crystalline materials is shown in Figure 29-B. The thermodynamic equilibrium dissolution coefficient (log Ksp value) present in the VisualMINTEQ database for this material suggests it is very insoluble.

Evaluation of Debris Bed Response to WCAP Precipitate AdditionThe primary objective of this CHLE preliminary test series was to select a bed preparation protocol suitable for initiating the 30-day baseline performance tests. Important attributes inselecting a debris bed included mechanical integrity, reproducibility, chemical detection threshold sensitivity, and time-response sensitivity.

The tests evaluated the NEI pressure-washed and the blender-prepared debris beds at 0.1 and 0.01 ft/s approach velocities.

The tests wereconducted with circulation solutions of either deionized water or deionized water with boric acidand trisodium phosphate (TSP). Some preliminary tests also considered the use of green siliconcarbide as particulate debris, the use of a double leaf-shredded fiber debris bed, and deionized water with boric acid and sodium tetraborate.

WCAP precipitates were added directly in thehead loss columns or in the CHLE corrosion tank.Testing found that the NEI fiber preparation method resulted in very consistent debris beds.Incremental additions of fiber resulted in a linear increase in bed thickness and head loss.Repeated tests resulted in similar increases in bed thickness and head loss, as shown in Figures30, 31, and 32. The blender method resulted in less consistent debris beds. The increase in headloss with additions of fiber was non-linear.

In addition, changing the approach velocity andreturning it to the original velocity resulted in significant increases in head loss. After athreshold quantity of fiber was introduced into the column, the head loss through the fiber-only bed increased rapidly to over 50 inches at both 0.1 and 0.01 ft/s, as shown in Figures 33 and 34.Figure 29 -(A) White material present on the submerged galvanized coupons at theconclusion of the 30-day LBLOCA test, and (B) SEM images of the same white material.

Document No: CHLE-015, Rev 3Page 33 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 20.1jis* Test 2 y=0.1427x-0.286

1. Test 5 R2 910.y = 0.1282x -0.66912= 0.998700 20 40 60 80 100 120 140 160Fiber quantity (g)Figure 30 -Head loss through NEI pressure-washed debris beds at approach velocity of0.093 ft/s in Test 2 and 0.1 ft/s in Test 5.2.5ýr 2.0'M Test 2* Test5 5, .S1.5* y= 0.0143x + 0.159 y U.0145X--+-U.UM8 R, R 0.9804d R2 0.978t=C1.0" 0.50 20 40 60 80 100 120 140 160Fiber quantity (g)Figure 31 -Head loss through NEI pressure-washed debris beds at approach velocity of0.01 ft/s in Tests 2 and 5.10Z mTest 2 Y = 0.0475x + 0.9W # Test 5=R=0"C4.,-ffi y = 0.04x + 1.250 20 40 60 80 100 120 140 160Fiber quantity (g)Figure 32 -Fiber bed thickness of NEI pressure-washed debris beds at approach velocityof 0.1 ft/s in Tests 2 and 5.Document No: CHLE-015, Rev 3Page 34 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 60*0 Test 950 #Test 10; -40W-c 300 S.;-- 20z 100 10 20 30 40 50 60 70Fiber quantity (g)Figure 33 -Head loss through blender-processed debris beds at approach velocity of 0.1ft/s in fiber beds prepared with a blender (Tests 9 and 10).8060 *Test 1040* u 2000 20 40 60 80 100 120Fiber quantity (g)Figure 34 -Head loss through blender-processed debris beds at approach velocity of 0.01ft/s in fiber beds prepared with a blender (Tests 9 and 10).The increase in head loss caused by cycling the velocity low and high through the blender-prepared debris bed was as much as 80 percent, as shown in Figure 35.The NEI and blender fiber preparation methods were both effective at capturing WCAPprecipitates, but the blender-prepared fiber was more effective.

The addition of WCAPprecipitate in small increments resulted in non-linear head loss behavior.

The first additions, ifsmall, resulted in small increases in head loss. Once a threshold quantity that coated the leadingsurface of the bed was reached,

however, head loss increased rapidly until the test had to beterminated.

For both debris beds, the WCAP precipitates were added in batches corresponding to 0.45 g ofAl directly to the recirculating head loss column, which resulted in a bed loading rate of 24.7g/m2 as Al for each batch. As shown in Figure 36 for the NEI bed, 4 batches of WCAPprecipitate were added with only marginal increases in head loss, resulting in an aluminumloading rate into the bed of 98.8 g/m2 as Al. An additional loading of 12.3 g/m2 as Al for a totalloading of 111 g/m2 as Al, resulted in a rapid increase in head loss. As is demonstrated in Figure37, these results were reproducible over multiple tests.Document No: CHLE-015.

Rev 3Page 35 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 9080 60 g fiber after lowering v to 0.01 ft/sand returning to 0.1 ft/s7060ogfiberatO.1ft/s 60.5 50(A.2 40 40 g fiber after lowering v to 0.01 ft/sand returning to 0.1 ft/s: 3020 40gfiberat0.1ft/s

__,__1003:00 PM 3:30 PM 4:00 PM 4:30 PM 5:00 PM 5:30 PMTimeFigure 35 -Head loss through a blender-prepared fiber debris bed in Test 10.1008060o Fiber added at 0.1 ft/s 5 g ppt40x Velocity changed to 0.01 ft/s202g p 00-3 gppI 4 g pipt .g1:00 PM 1:30 PM 2:00 PM 2:30 PM 3:00 PM 3:30 PM 4:00 PM 4:30 PM 5:00 PMTimeFigure 36 -Head loss after addition of 1 g increments of WCAP precipitate introduced into a NEI pressure-washed debris bed (Test 6). Quantities refer to amount of AIOOHprecipitate prepared according to the WCAP recipe.Document No: CHLE-015, Rev 3Page 36 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 7060" 50q-U- Test 8u --#--Test 6.S 40(Ao.23010x20100 ..........,0 1 23 4 5 6WCAP Precipitate (g)Figure 37 -Comparison of head loss after addition of WCAP precipitate introduced into aNEI pressure-washed debris bed in Tests 6 and 8.For the blender-processed bed, the first batch of precipitate caused a rapid increase in head loss,indicating that the threshold for causing head loss in this type of bed is 24.7 g/m2 as Al or less.In tests conducted at ANL, head loss increased rapidly when the aluminum loading reached 13.3g/m2 as Al. The results of these tests are consistent with the ANL work for the blender-prepared bed, but demonstrate that the NEI pressure-washed debris bed is less sensitive to the presence ofchemical precipitates than the blender bed under these conditions.

Additional tests were conducted in which the WCAP precipitates were added into the CHLEcorrosion tank at high temperature instead of directly into the head loss columns at roomtemperature.

An important consideration in the addition of precipitates into the CHLE tank isthat the circulation time is much greater.

The circulation time in the closed-loop head losscolumns was 8.5 minutes, whereas the circulation time when the CHLE corrosion tank was inoperation was 110 minutes.

This difference in circulation time was not taken into account whenthe tests were conducted.

Two to three pool turnovers were allowed between precipitate additions in the closed-loop head loss column tests, but multiple batches of precipitate wereadded in less than one pool turn-over in the tank tests. Thus, it is not possible to determine thequantity of precipitate that would cause head loss to increase rapidly in the tank tests from thesetests.The results described in this section demonstrate that the NEI pressure-washing method resultedin more stable and reproducible debris beds than the blended bed method in relatively short(several hour) head loss tests. However, the blended debris beds experienced greater head losswhen precipitates prepared according to the WCAP protocol were introduced directly into thehead loss assemblies, leading to the conclusion that the blended fiber debris beds are moresensitive detectors for the presence of precipitates.

Document No: CHLE-01 5, Rev 3Page 37 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Evaluation of Long-Term Aluminum Nitrate Addition to Debris BedsThe original objective of this test series was to establish the baseline performance (without theaddition of corrosion materials) of the debris bed to be used for the 30-day CHLE tests withcorrosion materials in place. Based on discussions with the NRC, the selected debris bed typefor the 30-day testing was the blender-processed bed. Within a few days of operation, however,the blender-prepared debris beds exhibited instabilities and inconsistent performance betweenparallel

columns, as detailed later in this section.

Thus, the objective of the test series wasmodified to additionally evaluate the performance of the debris bed with the addition ofaluminum nitrate at a slow rate to simulate the introduction of aluminum into the solution as theresult of corrosion processes.

In addition, a second test was conducted using NEI-processed debris beds so that the two beds could be compared.

The testing program was conducted from28 June 2012 to 24 July 2012.Throughout the tests, the chemical system in the tank was prototypical of the post-LOCA chemical environment at STP; the chemicals included boric acid, trisodium phosphate, lithiumhydroxide, hydrochloric acid, and nitric acid. The support screen used was prototypical of theECCS strainers at the STP plant. A temperature profile characteristic of a medium-break LOCAas predicted by MELCOR and RELAP-5 was used. The approach velocity to the debris bedswas 0.01 ft/s to be consistent with the strainers at STP.As noted above, the blender-processed debris beds were not reproducible between columns.After 6 days of operation, the normalized head loss (normalized to 20 'C based on both viscosity and density corrections) varied from 1.6 inches of water to 100 inches of water through debrisbeds, as shown in Figure 38. It is important to note that the columns were in parallel operation so that each column was circulating the same water at the same rate. This degree ofinconsistency between columns was determined to be too excessive to allow assessment of theimpact of chemical precipitates, since it was clear that factors other than chemical precipitates could lead to significant differences in head loss.An inspection of the debris beds after the test was complete revealed that the blender-prepared beds formed small, dense nodules of fiber at the base of the fiber bed, immediately adjacent tothe perforated support plate, as shown in Figure 39. The nodules formed a dimple pattern thatmatched the pattern of holes in the perforated plate, indicating that the smaller fibers formed bythe blending process were able to form a more dense fiber mat in a localized area. Thedifference in head loss among the three columns with the blended fiber preparation appears to bedue to a trace amount of dirt or other material that collected in the small, dense fiber nodules atthe base of the fiber bed. The debris bed with the highest head loss visually had the greatestamount of darker material present in the nodules.

The fiber beds were visibly clean through therest of the depth, suggesting that little or no head loss occurred through the bulk of the depth ofthe fiber bed and that nearly all of the head loss occurred in the fiber where it contacted theperforated plate, indicating significant nonhomogeneity to the porosity and head losscharacteristics of the bed.Document No: CHLE-015, Rev 3Page 38 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 1601401201001 80(A-60q 4000 1 2 3 4 5 6 7 8 9Time (days)Figure 38 -Normalized head loss through fiberglass debris beds prepared by chopping in ablender (normalized to 20 °C). Aluminum nitrate addition started at about 6.75 days.A BCFigure 39 -Dimple pattern from the support plate on the bottom side of the blended fiberdebris beds in (A) Column 1, (B) Column 2, and (C) Column 3. The dimples in Column 3are darker than in Column 1. Also, the cross section of the debris bed, most evident onColumn 3, appears clean throughout the entire depth of the debris bed.The fiber beds prepared with the NEI pressure-washing method were reproducible betweencolumns.

After 6 days of operation, the normalized head loss remained less than 0.5 inches ofwater, as shown in Figure 40. Similar behavior continued until the test was terminated after 12days. The fiber beds prepared with the NEI pressure-washing method did not form the densenodules of fiber that were observed with the blended fiber beds. The absence of these nodulesand the reproducible behavior of the NEI beds lends further credibility to the conclusion thatthese nodules were responsible for the non-reproducible behavior of the blended fiber beds.Document No: CHLE-015, Rev 3Page 39 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal 2.01.51-- --Column 10CD) -Column 2* 10 -Column 31.000.50 1 2 3 4 5 6 7 8 9 10 11 12 13 14Time (days)Figure 40 -Normalized head loss through fiberglass debris beds prepared using the NEIpressure-washing method (normalized to 20 QC).The blended fiber preparation method resulted in shorter fibers (often called "shards" or"fragments")

than with the NEI pressure-washing method, which may have led to the ability ofthe bed to form the dense nodules at the holes in the perforated plate. The shorter fibers led to amore compact debris bed, as shown earlier in Figure 6. Low porosity nodules are formed bylocal bed compaction, enabled by the mobility of short fiber shards formed during the choppingprocedure.

Local compaction in regions of flow acceleration near the strainer penetrations isfurther enabled by loosely aggregated beds formed under very low approach velocity.

The debris beds did not change thickness significantly over the course of the test, which lastedover 8 days for the blended fiber debris bed and over 12 days for the NEI fiber debris bed.Further, minimal differences in thickness were observed between beds despite the wide variation in measured pressure loss. The lack of change in bed thickness, coupled with the visually cleannature of the debris beds, lends credibility to the conclusion that the head loss associated with theblended fiber debris bed was due to localized conditions at the perforated plate.After 6.75 days of operation, aluminum nitrate solution was bled into the system at a slow rate tosimulate the addition of aluminum via corrosion processes.

It was intended to add the aluminumnitrate at a sufficiently low rate to prevent the formation of precipitates until the entire solutionreached the saturation concentration, but insufficient mixing at the point of addition allowedlocalized higher concentration conditions that caused precipitation to occur. A significant fraction of the aluminum that was added to solution formed a precipitate, as indicated byturbidity measurement and supported by total and filtered aluminum analyses.

The addition of 1mg/L of aluminum in solution caused the formation of in-situ precipitates that caused head lossthrough the blender-processed debris beds. Since the bed in column 3 was already at 100 inchesDocument No: CHLE-015, Rev 3Page 40 of 45I Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal of normalized head loss (Figure 38), this addition was sufficient to reach the terminal head losscondition in that column. Column 2, which had been at about 38 inches of normalized head loss,increased to about 100 inches of head loss and appeared to be leveling off. Column 1 increased from 1.6 inches to about 18 inches. After 23 hours2.662037e-4 days <br />0.00639 hours <br />3.80291e-5 weeks <br />8.7515e-6 months <br /> (12.5 pool turnovers),

an additional 5 mg/L(6 mg/L total) caused sufficient head loss to terminate the test.In contrast to the blender-processed debris bed, the addition of up to 40 mg/L of aluminum overa period of 6 days was not sufficient to cause head loss through the NEI fiber debris beds, asshown in Figure 34. However, in tests reported in the previous

section, the same amount ofaluminum in the form of pre-formed WCAP precipitates caused extensive head loss (Figure 36)that terminated a similar test within 3 hours3.472222e-5 days <br />8.333333e-4 hours <br />4.960317e-6 weeks <br />1.1415e-6 months <br /> (1.6 pool turnovers).

These results indicate thatprecipitates formed in-situ through the addition of aluminum nitrate at a slow rate havesignificantly different characteristics from those of the pre-formed WCAP precipitates.

Particle size analyses indicate that the size of precipitates formed in-situ are up to 10 timessmaller than the pre-formed WCAP precipitates with similar solution conditions (0.17 pm versus1.6 [im in diameter, as shown in Table 7). This significant difference in size appears to besufficient to cause the pre-formed WCAP precipitates to be retained by the NEI fiber debris bedsbut allow the in-situ precipitates to pass through the NEI fiber debris beds. The results indicatethat head loss may be less significant than indicated by the use of pre-formed WCAP precipitates, depending on the solution chemistry and the filtration characteristics of the debris bed. Thesmall size of the colloids formed in-situ is consistent with the constituent particle diameterreported in NUREG/CR-6915

[21 ] (which reported a diameter of 0.13 Pim).Similar results were also observed in testing at ANL [19]. In tests conducted with blended-fiber beds, aluminum added at a concentration of 1.5 ppm (filter loading rate of 13.3 g/m2 as Al) wassufficient to plug the debris bed when added as pre-formed WCAP precipitates.

The researchers at ANL found that aluminum precipitates formed using a different recipe, in water that contained boron and silicon, form particles in the range of 0.1 to 0.3 pm in diameter, considerably smallerthan the particles observed when precipitates were formed using the standard WCAP protocol.

When the alternate precipitates were introduced into a head loss loop containing a blended-fiber bed, no head loss occurred even at aluminum concentrations as high as 33 ppm (filter loadingrate of 260 g/m2 as Al) The ANL researchers concluded that "particles less than 0.5 ptm in sizedo not get trapped in ANL's reference NUKON bed and are therefore not effective at producing head loss across the ANL fiber-only debris bed." [ 19]. Based on the work conducted at ANL,the ability of colloidal aluminum precipitates to pass through a debris bed without being captureddoes not appear to be limited to the NEI-processed debris bed, but can also happen with blender-processed beds under appropriate conditions.

Summary and Conclusions The primary conclusions of the CHLE tests in 2012 are related to the outcome of the 30-dayMBLOCA test. The significant conclusions of the testing program can be summarized asfollows:Document No: CHLE-015, Rev 3Page 41 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal Table 7 -Particle size and zeta potential of aluminum hydroxide precipitates IPre-formed WCAP Precipitates In boron/TSP water, 30 mg/L dilution 1.6 -28In boron/TSP water, 40 mg/L dilution 1.6 -25In boron/TSP water, 10 mg/L dilution 1.6 -27In deionized water, 40 mg/L dilution 1.8 29In deionized water, 10 mg/L dilution 2.5 30In tap water 1.2 -3In-situ Precipitates (in boron/TSP water)At end of test 0.17 -31Two weeks later 0.18 -321. Each value given is the average of three measurements

1. Weight loss measurements and calculations indicated that a total of 0.90 g of aluminummetal corroded during the CHLE 30-day MBLOCA test. This quantity of material lostcorresponded to corrosion rates of 7.1 g/m2 as Al for submerged aluminum and 2.4 g/m2 asAl for aluminum exposed to containment sprays over a 30 day period. The observed resultsare in general agreement with the results from ICET Test #2, which was conducted undersimilar chemistry and pH conditions, but with different aluminum material and temperature profile.

The corrosion rates are less than 3 percent of the rates calculated for a design basisLOCA using the WCAP-16530-NP spreadsheet, indicating that significantly less aluminumwill be available to form chemical products in a MBLOCA than predicted for a design basisLOCA.2. The concentration of aluminum measured in solution in the CHLE tank during the MBLOCAtest was 0.28 mg/L, in general agreement with the low rate of corrosion.

As a point ofreference, the concentration of aluminum predicted by the WCAP- 16530-NP spreadsheet forthe design basis LOCA was 32.9 mg/L.3. The turbidity in the tank solution peaked at the beginning of the MBLOCA test atapproximately 0.6 NTU and gradually decreased to about 0.3 NTU at the end of the test. Asimilar trend was observed in tests that did not have corrosion coupons in the tank, indicating that this level of turbidity was due to particles present in the tank at the beginning of the test.The fact that turbidity did not increase supports a conclusion that precipitation of corrosion products in the bulk solution did not occur.Document No: CHLE-015, Rev 3 Page 42 of 45Document No: CHLE-01 5, Rev 3Page 42 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal

4. Measurements of filtered and total aluminum in solution were essentially identical to eachother over the duration of the MBLOCA test. The fact that there was no difference betweenthe filtered and unfiltered samples supports a conclusion that there was no precipitated aluminum greater than 0.45 gm (the pore size of the filter) circulating in the solution.

Similar results were obtained for calcium and silicon.5. The measured aluminum and calcium concentrations were below the solubility limits foraluminum hydroxide and calcium phosphate, respectively, as calculated by Visual MINTEQfor the conditions representative of STP. Work conducted at ANL has indicated thatsolubility of amorphous aluminum hydroxide calculated by Visual MINTEQ was a goodindicator of the boundary at which precipitation occurs. These results provide a rationale forwhy no precipitates formed and are generally supportive of the observed results.6. The normalized head loss through all three of the NEI-processed debris beds remained below0.5 inches of water column for the duration of the 30-day test. The normalized head lossthrough the blender-processed debris beds was marginally higher than the head loss throughthe NEI-processed debris beds in all 3 columns when exposed to the tank solution for 2 daysat the end of the 30-day MBLOCA test. However, the normalized head loss was still below 1inch in all 3 columns.

While questions have been raised regarding the appropriate debris bedto use in these experiments, these results are consistent with experiments conducted at ANL,which used blender-prepared debris beds. Based on the measured aluminum concentration insolution, the aluminum loading rate on the CHLE debris beds would have been 5.8 g/m2 asAl if precipitates had formed and would be 1.7 g/m2 as Al based on the surface area of thestrainers at the STP plant. In testing at ANL, introduction of pre-formed precipitates at aconcentration of 0.5 ppm of Al and aluminum loading rate of 4.4 g/m2 as Al caused only aminor increase in head loss. In the ANL work, plugging of the debris bed did not occur untilthe aluminum loading reached 13.3 g/m2 as Al. Similarly, in small-scale chemical effecttesting described in NUREG/CR-6868, additional head loss due to chemical precipitates didnot begin to occur until the aluminum concentration was 2.7 mg/L and the loading rate was3.5 g/m2 as Al. Comparison to these previous results supports a conclusion that little or noincrease in head loss would likely have occurred in these tests regardless of what debris bedwas used, even if precipitates had formed.7. The testing of the ECCS Sure-Flow TM strainer module subjected to prototypical debris fromthe STP plant by Alden in 2008 resulted in additional head loss due to chemical precipitates of 4.75 ft, an increase of 2.2 times the head loss due to non-chemical debris. The chemicalprecipitate loading to the system that caused this head loss was screen loading rate of 990g/m2 of Al and of 190 g/m of Ca. These quantities are substantially higher than theprecipitate loading rates observed in the CHLE work for a 6-inch cold leg MBLOCA. Headloss was observed as soon as the first batch of precipitates was added in the Alden test, butthat first addition corresponded to a loading rate of 77 g/m2 of Al. The substantial difference in concentrations and loading rates provides an explanation for why an increase in head losswas observed in the Alden testing and not in the CHLE tests.8. The results of the LBLOCA test was similar to the MBLOCA test in many respects, such asgenerally low concentration of metal ions in solution, low and gradually declining turbidity Document No: CHLE-015, Rev 3Page 43 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal over the duration of the test, and minimal levels of head loss. However, some differences were observed.

The head loss did increase during the LBLOCA test, whereas it did not in theMBLOCA test. The constituents that contributed to the increase in head loss could not beidentified, but may have been zinc granules circulating in solution or zinc phosphate thatformed during the test. The concentrations of metal ions were generally lower in theLBLOCA test even though predictions based on corrosion rates suggest that theconcentrations should have been higher. The presence of zinc in solution in the LBLOCAtest may have had an inhibitory effect on the corrosion and release rates of other materials and should be explored in further testing.

The zinc phosphate material that formed wasidentified as a crystalline material with the chemical formula Zn3(P04)2"4H20.Based on the low quantities of aluminum and calcium released into solution and the fact that theresulting concentrations appear to be below the saturation concentrations for aluminumhydroxide and calcium phosphate, it is expected that no precipitates will form for a LOCA in a 6-inch or smaller break at the STP plant. Furthermore, if precipitates were to form, they would beunlikely to cause a significant increase in head loss. As a result, chemical precipitates areexpected to have no effect on the ECCS strainers at the STP plant for all LOCAs smaller than a6-inch break. Although additional testing is needed, the similarity of the results between theMBLOCA and LBLOCA tests suggests that it may be appropriate to extend the conclusions ofthe MBLOCA test with respect to the effect of chemical precipitates on ECCS strainer head lossto include breaks on pipes up to 15 inches in diameter.

References

1. Lane, A.E., et al., Evaluation of Post-Accident Chemical Effects in Containment SumpFluids to Support GSI-191, 2006, Westinghouse Electric Company:

Pittsburg, PA.2. Nuclear Energy Institute (NEI). ZOI Fibrous Debris Preparation:

Processing, Storage,and Handling, Revision 1, Jan 2012.3. Areva, NP, Inc. South Texas Project Test Plan for ECCS Strainer Performance Testing,Doc. 63-9086408-001, Aug, 2008.4. Areva, NP, Inc. South Texas Project Test Report for ECCS Strainer

Testing, Doc. 66-9088089-000, Aug, 2008.5. Westinghouse Electric Company LLC, South Texas Project GSI-191 Chemistry EffectsEvaluation, Doc. CN-CSA-06-6, Aug, 2007.6. STP, Added Commodities Inside The RCB, 2003, South Texas Project.7. Alion, STP Post-LOCA Water Volume Analysis, Calc. No. ALION-CAL-STP-8511-01, Rev. 1, Alion Science and Technology:

Albuquerque, NM., Aug, 2012.8. Alion, STP Concrete Surface Area Analysis, Calc. No. ALION-CAL-STP-8511-04, Rev.0., Alion Science and Technology:

Albuquerque, NM, Aug, 2012.9. Letellier, Bruce. Calculation of Nukon Fiberglass Debris Quantities in a 6-inch break.10. TAMU, 6-Inch Cold Leg Break Scenario 30-Day Containment

Response, Texas A&MUniversity, College Station, TX, Nov. 2012.11. UNM, STP Corrosion Head Loss Experiment (CHLE) Test Plan, Doc. CHLE-003, Rev.1.4, University of New Mexico, Albuquerque, NM. Aug, 2012.Document No: CHLE-015, Rev 3 Page 44 of 45Document No: CHLE-01 5, Rev 3Page 44 of 45 Title: Summary of Chemical Effects Testing in 2012 for STP GSI-191 License Submittal

12. UNM, CHLE Equipment Description and Specifications, Doe. CHLE-004, Rev. 1,University of New Mexico, Albuquerque, NM. Aug. 2012.13. Ruland, W.H. Letter to John Butler of the Nuclear Energy Institute with the subject line"Fibrous Debris Preparation procedure for Emergency Core Cooling SystemRecirculation Sump Strainer

Testing, Revision 1" dated April 26, 2012.14. UNM, Determination of the Initial Pool Chemistry for the CHLE Test, Doc. CHLE-O05, Rev. 1, University of New Mexico, Albuquerque, NM. Aug. 2012.15. UNM, Test 2: Medium Break LOCA Tank Test Parameter

Summary, Doc. CHLE-011, University of New Mexico, Albuquerque, NM. Aug. 2012.16. ASTM Standard G1 -03: Standard Practice for Preparing,

Cleaning, and Evaluating Corrosion Test Specimens, 2011.17. Chen, D. Howe, K.J., Dallman, J., Letellier, B.C., Klasky, M., Leavitt, J., and Jain, B."Experimental analysis of the aqueous chemical environment following a loss-of-coolant

accident, Nuclear Engineering and Design, vol. 237, pp. 2126-2136, 2007.18. Gustafsson, J.P., Visual M1NTEQ Ver. 3.0 Software, KTH Royal Institute of Technology, Sweden, Accessed at <www2.lwr.kth.se/english/OurSoftware/vminteq/download.html>,

2012.19. Bahn, C.B., Kasza, K.E., Shack, W.J., Natesan, K, and Klein, P. "Evaluation ofprecipitates used in strainer head loss testing:

Part III. Long-term aluminum hydroxide precipitation tests in borated water," Nuclear Engineering and Design, vol. 241, pp.1914-1925, 2011.20. Bahn, C.B., Kasza, K.E., Shack, W.J., Natesan, K, and Klein, P. "Evaluation ofprecipitates used in strainer head loss testing.

Part I. Chemically generated precipitates,"

Nuclear Engineering and Design, vol. 239, pp. 2981-2991, 2009.21. Klasky, M., Zhang, J., Ding, M., Letellier, B., Chen, D., and Howe, K. AluminumChemistry in a Prototypical Post-Loss-of-Coolant-Accident, Pressurized-Water-Reactor Containment Environment, NUREG/CR-6915, 2006.22. Johns, R.C., Letellier, B.C., Howe, K.J., and Ghosh, A.K. Small-Scale Experiments:

Effects of Chemical Reactions on Debris-Bed Head Loss: A Subtask of GSI-191.NUREG/CR-6868, 2005.23. UNM, Medium Break LOCA Test Report, Doc. CHLE-012, University of New Mexico,Albuquerque, NM. Jan. 2013.24. UNM, Large Break LOCA Test Report, Doc. CHLE-014, University of New Mexico,Albuquerque, NM. Jan. 2013.Document No: CHLE-015, Rev 3 Page 45 of 45Document No: CHLE-01 5, Rev 3Page 45 of 45