An Experimental Approach to Assessing Material Corrosion Rates in a Reactor Containment Sump Following a Loss of Coolant Accident

Thesis

Presented in Partial Fulfillment of the Requirements for the Degree Master of Science in the Graduate School of The Ohio State University

By

Erik Anders Lahti, B.S.

Graduate Program in Nuclear Engineering

The Ohio State University

2013

Thesis Committee:

Jinsuo Zhang, Advisor

Tunc Aldemir

Richard Denning

Copyrighted by

Erik Anders Lahti

2013 Abstract

The present study conducts an extensive a review of relevant research that pertains to Generic Safety Issue 191 (GSI-191) and the chemistry and corrosion behavior of the various materials present in the containment sump of a pressurized water reactor (PWR). Facility designs are described for two experimental systems that would examine corrosion effects. Thermodynamic simulations, and integrated and benchtop tests have determined the structure of the potential precipitates that may clog the sump strainer and cause a failure of the emergency core cooling system (ECCS). Based on this research, it was determined that the available research has been insufficient in terms of simulating the transient temperature and pressure behavior in the containment sump in the post loss of coolant environment. Research gaps are identified, and recommendations for future research are presented that would be performed in the proposed experimental facilities.

ii Dedication

This thesis is dedicated to my family and friends. Thank you for all your love and support!

iii Acknowledgements

I would like to acknowledge Doctor Jinsuo Zhang for his guidance and tutelage. Without it, this document would not have been possible.

iv Vita

June 2008 ...... Diploma, Sylvania Southview High School

June 2010 to August 2010...... Intern, Los Alamos National Laboratory

January 2011 to August 2012 ...... Intern, Ohio Emergency Management Agency

March 2012 to present ...... Intern, Pacific Northwest National Laboratory

June 2012 ...... B.S. Mechanical Engineering, The Ohio State University

August 2012 to present ...... Graduate Research Assistant, Nuclear Engineering Program, The Ohio State University

Fields of Study

Major Field: Nuclear Engineering

v Table of Contents

Abstract...... ii

Dedication...... iii

Acknowledgments...... iv

Vita...... v

List of Tables ...... viii

List of Figures...... ix

List of Equations...... x

Chapter 1: Introduction...... 1

Chapter 2: Background ...... 3

2.1 Relevant Results from GSI-191 Research ...... 4

2.2 Sump Materials ...... 7

2.2.1 Zinc ...... 11

2.2.2 Aluminum ...... 14

2.2.3 Carbon Steel...... 17

2.2.4 Copper...... 20

2.2.5 Concrete ...... 23

2.2.6 Fiber Insulation ...... 26

Chapter 3: Experimental Setup and Procedure for Autoclave Tests...... 32

vi 3.1 Experimental Setup...... 33

3.1.1 Autoclave ...... 33

3.1.2 Potentiostat; Working, Counter and Reference Electrodes; and Wiring ...... 36

3.1.3 Hydrogen Canisters, Heater, Thermocouples and Temperature Controller ...... 39

3.2 Experimental Procedure...... 40

Chapter 4: Experimental Setup and Procedure for a Non-isothermal Test Loop ...... 43

4.1 Experimental Setup...... 43

4.1.1 Pressure Vessel ...... 45

4.1.2 Heat Exchanger and Pre-Heater...... 47

4.1.3 Test Columns ...... 50

4.2 Experimental Procedure...... 51

Chapter 5: Conclusions and Future Research ...... 53

5.1 Conclusions...... 53

5.2 Future Research ...... 54

References...... 55

vii List of Tables

Table 2.1: Expected Precipitates in Alkaline Solutions at Various Temperatures ....5

Table 2.2: Expected Precipitates in Neutral Solutions at Various Temperatures...... 6

Table 2.3: Calculated Corrosion Potentials for Containment Sump Materials...... 9

Table 2.4: Simulated Zinc Corrosion Product Production...... 13

Table 2.5: Simulated Aluminum Corrosion Product Production...... 16

Table 2.6: Simulated Carbon Steel Corrosion Product Production ...... 19

Table 2.7: Simulated Copper Corrosion Product Production ...... 22

Table 2.8: Composition of Portland Cement ...... 24

Table 2.9: Oxide Composition of Portland Cement...... 24

Table 2.10: Simulated Concrete (Non-Particulate) Corrosion Product Production...25

Table 2.11: Simulated Concrete (Particulate) Corrosion Product Production...... 25

Table 2.12: Measured Leaching Rates of Major Components of Concrete...... 26

Table 2.13: Composition of Nukon Fiber Insulation...... 27

Table 2.14: Composition Limits of E-Glass Fiber...... 27

Table 2.15: Estimation of Fiber Insulation Concentration...... 29

Table 2.16: Simulated Nukon Insulation Corrosion Product Production ...... 30

Table 3.1: Test Matrix for Autoclave Experiments ...... 42

viii List of Figures

Figure 2.1: Pourbaix Diagram for Zinc at 25 °C ...... 12

Figure 2.2: Pourbaix Diagram for Aluminum at 25 °C ...... 15

Figure 2.3: Pourbaix Diagram for Iron at 150 °C ...... 18

Figure 2.4: Pourbaix Diagram for Copper at 100 °C...... 21

Figure 3.1: EZE-Seal Autoclave from Parker Hannifin Corporation ...... 35

Figure 3.2: Packing Gland from Conax Technologies...... 36

Figure 3.3: Power Lead Gland from Conax Teachnologies ...... 36

Figure 3.4: High Pressure Ag/AgCl Reference Electrode from Corr Instruments ....38

Figure 3.5: Model SP-300 Potentiostat from Bio-Logic Science Instruments ...... 39

Figure 4.1: Non-isothermal Test Loop Concept Drawing ...... 44

Figure 4.2: Immersion Heater from Chromalox ...... 46

Figure 4.3: Spiral Heat Exchanger from Sentry Equipment...... 49

Figure 4.4: Heat Tape from BriskHeat ...... 50

ix List of Equations

Equation 2.1: Calculating the Reversible Potential ...... 8

Equation 2.2: Nernst Equation...... 8

Equation 2.3: Corrosion Product Formation Relation ...... 10

Equation 2.4 Effect of Temperature on the Solubility of Silicon Dioxide ...... 29

Equation 4.1: Heat Exchanger Design Equation...... 47

Equation 4.2: Calculating the Logarithmic Mean Temperature Difference ...... 47

x Chapter 1: Introduction

In 1992, a loss-of-coolant accident (LOCA) occurred at the Barsebäck nuclear power plant in Sweden that led to degradation of the emergency core cooling system

(ECCS) due to plugging of the sump pump [1]. This accident led the Nuclear Regulatory

Commission (NRC) to examine sump strainer performance in reactors in the United

States. Initial research was conducted for boiling water reactors (BWRs) because the unit at Barsebäck was a BWR. It was found that the problem could be avoided by simply removing any fibrous material not rated to withstand LOCA conditions and enlarging the strainer surface area [2].

Although BWRs were able to quickly address the problem, pressurized water reactors (PWRs) could not do so because of the major differences between BWR and

PWR containment designs. This led the NRC to issue Generic Safety Issue 191 (GSI-

191), Assessment of Debris Accumulation on Pressurized Water Reactor Sump

Performance, several years later. Although GSI-191 was closed in 2008, the NRC did issue additional guidance as to how power plants should ensure further confirmatory research into the findings of GSI-191 [3]. This was due largely in part to the complexity of the problem. This confirmatory research will help establish an adequate technical basis so that the corrective actions can be deemed complete and sufficient for each affected power plant.

1 This confirmatory research included several integral tests aimed at validating thermodynamic simulations and truly describing the chemical effects present in this environment. The most in depth of these tests was the Integral Chemical Effects Test

(ICET) loop, which was constructed at the University of New Mexico (UNM) and overseen by Los Alamos National Laboratory. This apparatus was later utilized for the

Chemical Effects Head Loss Experiment (CHLE) [4].

Although the integral tests were imperative in generating experimental data, very little experimental work has been done to understand and validate the corrosion rates and chemical stabilities produced from thermodynamic simulations for each common containment sump material. In the present study, the status of GSI-191, including its relevant supporting research to understand the corrosion and chemistry of the various containment sump materials in the environment of the sump following a LOCA, is extensively reviewed. Based on the identification of a research gap, a large test loop that can simulate a more realistic post-LOCA environment than the ICET loop is designed. In addition to this large test loop, an autoclave apparatus and a corresponding experimental procedure are then designed so that online measurements of corrosion rates in a simulated environment similar to the containment sump can be accurately performed. Both systems will be able to produce meaningful results so that the desired adequate technical basis may be formed such that the complex problem of GSI-191 may be solved in a risk- informed manner.

2 Chapter 2: Background

Corrosion, in its various forms, has been studied since the early 1900s, with most interest coming during and after the 1950s and 1960s. Thus, the basic thermodynamics of corrosion are well understood. However, corrosion that takes place in extreme and complex environments, like that of the containment sump following a LOCA, is not well known. In fact, many of the materials present in the containment sump have little experimental data for the post-LOCA environment; most of the current data are obtained from running thermodynamic simulations.

As the materials in the containment sump corrode, it is possible for the corrosion products to chemically react in the solution and form precipitates. These precipitates, if enough were produced, could, in conjunction with fibrous debris materials, clog the filtration screen protecting the sump pump from debris, and cause a head loss leading to less suction, and possibly a failure of the ECCS to provide adequate cooling water to the core.

Most of the experimental data generated from GSI-191 were from integrated tests, in which the entire environment is simulated in one test. From these tests, the general behavior of sump screen clogging could be observed, but not the detailed corrosion and precipitation processes for single materials, or the possible interactions of a few materials at once. In this section, a detailed review of the relevant findings of GSI-191 is presented.

3 In addition, the general chemical and corrosion behaviors of each common containment sump material are described.

2.1 Relevant Results from GSI-191 Research

The first major details of relevant data related to GSI-191 regard the description of the containment environment following a LOCA. Water from the primary systems of the reactor is ejected at temperatures of a maximum of 315 °C and at pressures of 2200 psi. In the transient process, the maximum sump temperatures can approach 130 °C and the maximum gauge pressures can reach 36 psi. Once the containment environment has reached equilibrium, temperatures of up to 55 °C and atmospheric pressures are observed

[5]. As for the water itself, PWRs utilize borated water in the primary loop for reactivity control. This boron is in the form of boric acid, so the initial water is slightly acidic. The pH of the primary water is balanced by the rate of boric acid and lithium hydroxide injection through the chemical shim system. In the event of a LOCA, the chemical shim system will not be functional, so the acidity needs to be controlled in a different manner within the containment. This containment pH control chemistry is plant specific, and may contain either lithium hydroxide or sodium hydroxide as the major pH controller, which is introduced via spray systems, or have buffers of either hydrated trisodium phosphate

(TSP), or hydrated sodium tetraborate that exist as dissolvable powders in the sump [6].

This harsh environment is difficult to incorporate into experiments.

Various thermodynamic simulations were run to support GSI-191 research. These simulations were summarized by Jain, et al. [7] [8] and McMurray, et al. [9]. In general, the thermodynamic simulations were able to perform several calculations:

• Aqueous speciation and saturation calculations

4 • Precipitation reactions

• Calculations at standard and elevated temperatures

• Aqueous processes at ionic strengths up to 0.5

• Maintain fixed conditions if deemed necessary

These types of calculations can be performed by several different thermodynamic codes, so runs in each code were made and compared. Although there were differences in the code outputs, the order of magnitude for each output was usually similar. From these various simulations, the expected species expected to precipitate out of solution were determined. In addition to determining the type of solids produced, the amount of precipitate that could be dissolved in solution at a pH level of 10 was also calculated. The results are provided in Table 2.1. For plants using non-alkaline buffers, such as TSP, the species of precipitates was similar. The results for these plant types are provided in Table

2.2.

Table 2.1: Expected Precipitates in Alkaline Solutions at Various Temperatures [8] [9] Precipitate Percent in Solid Percent in Solid Percent in Solid Chemical Formula Phase at 60 degrees Phase at 90 degrees Phase at 110 Celsius (%) Celsius (%) degrees Celsius (%) NaAlSi3O8 90.2 84.2 79.7 Ca2Mg5Si8O22(OH)2 8.7 15.6 16.8 Fe3Si2O5(OH)4 26 N/A N/A ZnOFe2O3 0.03 N/A N/A Fe3O4 N/A 12 N/A Zn2SiO4 0.36 0.09 0.66 Mg 100 100 100 Al 100 100 100 Si 35 32 32 Continued

5 Table 2.1: Continued Ca 8 17 19 Zn 63 39 81 Fe 75 87 98

Table 2.2: Expected Precipitates in Neutral Solutions at Various Temperatures [8] [9] Precipitate Percent in Solid Percent in Solid Percent in Solid Chemical Formula Phase at 60 degrees Phase at 90 degrees Phase at 130 Celsius (%) Celsius (%) degrees Celsius (%) NaAlSi3O8 28 26 25 Ca5(OH)(PO4)3 67 64 63 Fe3Si4O10(OH)2 N/A 9 1 Fe3(PO4)28H2O 5 N/A N/A ZnOFe2O3 N/A N/A 4 Fe3O4 N/A N/A 10 Mg 0 0 0 Al 100 100 100 Si 29 36 29 Ca 96 99 99 Zn 0 22 72 Fe 38 82 96

Table 2.1 and Table 2.2 indicate that the major contributors to potential sump clogging are due to silicates, phosphates and hydroxides.

To validate these thermodynamic simulations experimentally, a number of integral chemical effects tests were run. Several of the most in depth of these tests were run using the ICET loop at UNM. These tests were able to recreate the containment sump environment at the post-LOCA steady state conditions of 55 °C and atmospheric pressure. Corrosion was facilitated and various precipitates could be identified. In fact, a gel-like substance was found at the bottom of the reaction tank following tests that incorporated TSP as the main pH buffer. It was determined that this gel was mainly

6 composed of Ca3(PO4)2 [6]. In general, the majority of the precipitates that were found were from the fiberglass insulation itself, or from the aluminum. It was also determined that the silicon and calcium from the fiberglass inhibited corrosion in the aluminum, thus limiting precipitate formation in the high-pH tests [6]. In addition to characterizing the chemistry of the solution following 30 days of tests, the particle size distributions in solution were also analyzed. The distributions were in the range of 1 to 100 µm, and the mean particle size varied with each test.

In addition to these large-scale loops, tests were also conducted in small benchtop loops to measure the potential head loss across the sump strainer. Pacific Northwest

National Lab (PNNL) operated multiple tests to relate any chemical effects to the measured pressure drop [10]. At most, the pressure drop was measured to be 2.2 psi, with minimum drops of 0.3 psi. It was found that these pressure drops were extremely dependent on the material preparation, what order the debris material is placed on the screen, and the amount of material placed on the screen. The approach velocity also played a large role in the variability of the pressure drop. Although a pressure drop was measured, the drop (2.2 psi) is very small, and may not lead to an ECCS failure like that seen at the Barsebäck plant.

Based on the results, or rather lack thereof, from the ICET and PNNL experiments, some major factors could have been overlooked in the past NRC funded reports. Since most of the tests have been isothermal, it is postulated that the temperature transient seen during a LOCA could be such a missing factor.

2.2 Sump Materials

7 Several different materials are common in the containment sump. These are aluminum, concrete, copper, carbon steel, fiber insulation and zinc. These materials can be divided into two different groups: those that corrode by general electrochemical processes, and those that degrade primarily through a phenomenon called leaching.

For those materials that corrode in the most common, electrochemical form, the electrochemical potentials can be described by a set of two equations given in Equations

2.1 and 2.2. The first of these equations calculates the potential at which a reversible reaction takes place. The second equation, the Nernst equation, describes how the potential changes with differing concentrations and temperatures.

−ΔG0 E 0 = (2.1) nF

€

 a  0 RT ∏ p E = E − ln  (2.2) nF    ∏ar 

€

In these equations, –ΔG0 is the change in Gibb’s free energy for the standard reversible reaction, n is the number of equivalents of electrons transferred, F is Faraday’s constant

(96485 C/equivalent), E0 is the standard potential at which the reversible reaction takes place, R is the universal gas constant (8.314 kJ/kmol/K), T is the absolute temperature at

8 which the reaction is taking place, ap is the activity of the products of the reaction, ar is the activity of the reactants of the reaction, and E is the potential at which a non- reversible reaction takes place.

From these equations, the corrosion potential can be calculated for any temperature. The potentials for the maximum sump temperature, given in Table 2.3, were calculated using the standard assumptions that proton reduction occurs at zero volts (also known as the standard hydrogen electrode or SHE), and that corrosion occurs when the ionic molar fraction of the corrosion product is 10-6.

Table 2.3: Calculated Corrosion Potentials for Containment Sump Materials Material (Reduced Ion) Corrosion Potential at 130 Degrees Celsius (VSHE) Zinc (Zn2+) -1.000 Aluminum (Al3+) -1.820 Carbon Steel (Fe2+) -0.650 Copper (Cu+) -0.120

In addition, using these equations, diagrams can be made which describe how various materials react in a water solution at varying pH levels. These diagrams are known as

Pourbaix diagrams, and form the backbone of determining what corrosion processes are taking place in a solution. However, most of these diagrams are for materials at room temperature. In addition to this shortcoming, these diagrams also do not give any information on the rate at which the various reactions occur.

For those materials that do not corrode in the electrochemical sense, these diagrams do not apply. Thus, different corrosion processes, in the forms of leaching and

9 hydration, are the primary forms of degradation for concrete and fiber insulation [11]; the ceramic constituents of their structures do not readily take place in electrochemical reactions due to their strong bonds. In general, leaching is the process by which soluble compounds are dissolved out of a host material. Hydration is the process by which the material in question will absorb water to form hydrates.

After either corrosion process has taken place, the major issue becomes the possible chemical reactions that will occur between the ions in solution, particularly those that form precipitates that could potentially clog the sump strainer. The rate of the production of these precipitates is closely related to how much corrosion product is generates. Jain, et al. [7] found that the amount of corrosion products is related to the time of exposure and can be expressed by:

CP = ESA × CR × t (2.3)

€

Here, CP is the amount of corrosion product that is produced, ESA is the exposed surface area, CR is the corrosion rate and t is the time the material is exposed to the corrosive environment.

Equation 2.3 indicates that the surface area of each sump material is almost as vital a part in determining the potential amount of precipitates as the corrosion rate. Thus, a good estimate as to the amount of material present in the containment sump is integral to calculating the true amount of corrosion products.

10 In the following sections, each common containment sump material, their corrosion behaviors, the amount of corrosion products that are produced and their solubility in solution are discussed.

2.2.1 Zinc

Zinc is found in the containment sump mainly in the form of galvanized steel, but is also present in various protective coatings [8]. The Pourbaix diagram for zinc is given in Figure 2.1. The figure shows that zinc will corrode in neutral, room temperature water, and that increasing the pH to a weak basic level will lead to passivation, which implies a sharp drop in the corrosion rate. However, if the solution becomes too basic, corrosion will again take place. At high temperatures, the diagram is expected to have a general shift to lower potentials, as well as changes in the slope of the line separating the immune, pure zinc from the passive and basic corrosion domains.

11

Figure 2.1: Pourbaix Diagram for Zinc at 25 °C [12]

The positive zinc ion can form a stable, solid hydroxide precipitate in the expected containment sump environment following a LOCA [13]. However, at significantly high pH, this zinc hydroxide will form the soluble tetrahydroxozincate ion,

2- Zn(OH)4 [14].

The amount of the potential precipitate, zinc hydroxide, is directly linked to the amount produced by zinc corrosion, and thus its corrosion rate. From Piippo, et al. [15];

Griess and Bacarella [16]; and Niyogi, et al. [17], the corrosion rate was estimated to be

0.163 g/m2/h (7.9 mil/year). A simulation was run using this estimate to determine the

12 amount of corrosion products released to the sump environment [8]. Table 2.4 contains the findings from the aforementioned simulation, as well as the initial conditions, including the estimated surface area ratios for a general plant.

Table 2.4: Simulated Zinc Corrosion Product Production [8] Surface Submerged Submerged Time Corrosion Product Area to Fraction Surface Area to (h) Released per Liter Volume Volume Ratio (g) [mol] Ratio (m2/m3) [ft2/ft3] (m2/m3) [ft2/ft3] 26.2 [8.0] 0.05 1.31 [0.4] 0.5 1.1×10−4 [1.6 ×10−6]

€ € This estimated corrosion rate was later determined for several temperatures by

Jain, et al. [7] by using glass cells, an autoclave, and polarization resistance measurements. At 60 °C, the corrosion rate was measured to be 0.0357 g/m2/h (1.7 mil/year), at 90 °C, the corrosion rate was measured to be 0.0405 g/m2/h (2.0 mil/yr), and at 110 °C, the corrosion rate was measured to be 0.234 g/m2/h (11.3 mil/year). While these measurements verify the estimated corrosion rate, these tests were all run in isothermal systems at low pressure, which may not accurately describe the true corrosion processes that may be seen in the containment sump following a LOCA.

Although high pH seems to limit the formation of solid precipitates, some reactions could take place to create other solids. Pandya, et al. [18] observed no evidence

2- 2+ that Zn(OH)4 . Instead, they found soluble Zn ions arranged tetrahedrally in

13 coordination with hydroxide ions. However, when taking multiple scatter angles into account, the tetrahydroxozincate ion was identified, and that it may react with alkali metals, such as potassium, to create insoluble zinc oxide and alkali metal hydroxides

[18].

2.2.2 Aluminum

Aluminum, though its amount is generally restricted because of its high hydrogen generation rate, can still be present in the containment sump in the form of stored scaffolding [8]. The Pourbaix diagram for aluminum, given in Figure 2.2, is similar to that of zinc’s diagram. However, aluminum is passive in neutral water; it forms a very stable oxide layer on its exposed surfaces. Adjustments in pH in either the acidic or the basic direction will lead to corrosion. At high temperatures, the only expected changes in the Pourbaix diagram are a change in the slope of the line separating the immune, pure aluminum, and the passive and basic corrosion domains, and a general shift of the diagram to lower potentials.

14

Figure 2.2: Pourbaix Diagram for Aluminum at 25 °C [12]

In addition to the corrosion product forms seen in Figure 2.2, the passive oxide

2+ + - layer itself may corrode to additionally form Al(OH) , Al(OH)2 and Al(OH)4 [19]. The majority of these corrosion products can form stable, solid aluminum hydroxide or aluminum oxyhydroxide in the basic solutions expected in the sump following a LOCA

[13]. These compounds would precipitate out of solution and could potentially clog the sump strainer.

The amount of aluminum hydroxides or aluminum oxyhydroides in the sump is directly linked to the amount of aluminum-based ions produced by corrosion, and thusly the corrosion rate. Again, from data produced by Piippo, et al. [15]; Griess and Bacarella

15 [16]; and Niyogi, et al. [17], the corrosion rate was estimated to be 1.45 g/m2/h (186 mil/year). Using this estimate, a simulation was run to determine the amount of corrosion products released to the sump environment [8]. Table 2.5 contains the findings from the aforementioned simulation, as well as their initial conditions.

Table 2.5: Simulated Aluminum Corrosion Product Production [8] Surface Submerged Submerged Surface Time Corrosion Product Area to Fraction Area to Volume (h) Released per Liter (g) Volume Ratio (m2/m3) [mol] Ratio [ft2/ft3] (m2/m3) [ft2/ft3] 11.5 [3.5] 0.34 0.57 [0.17] 0.5 4.2 ×10−4 [1.5 ×10−5]

€ € As with zinc, Jain, et al. [7] measured corrosion rates for aluminum. At 60 °C, the corrosion rate was measured to be 0.986 g/m2/h (126 mil/year), at 90 °C, the corrosion rate was measured to be 1.89 g/m2/h (242 mil/year), and at 110 °C, the corrosion rate was measured to be 2.20 g/m2/h (282 mil/year). Again, while these measurements verify the estimated corrosion rate, these tests were all run in isothermal systems at low pressure, which may not accurately describe the true corrosion processes that may be seen in the containment sump following a LOCA.

Certain conditions can affect the dissociation of aluminum hydroxide and aluminum oxyhyrdroxide, and thus whether the precipitate would eventually lead to strainer clogging. From Goldberg, et al. [20], aluminum hydroxide forms in four different

16 ways. The first is an amorphous form, while the other three are crystalline forms known as gibbsite, bayerite, or nordstrandite, which only differ by their layer packing arrangements. These layers are held together with hydrogen bonds. It is on the subtle differences between these solid phases that the dissociation process hinges; those crystal structures with stronger hydrogen bonds will be less likely to dissociate. The aluminum oxyhydroxides, of the form AlOOH, form two crystalline structures called boehmite and pseudoboehmite. The only difference between these two structures is the amount of water molecules between the layers of their octahedral crystals. Of the compounds most likely to form in this type of environment, the amorphous form is most likely to dissociate, followed by pseudoboehmite, then bayerite, and lastly, the least likely to dissociate, gibbsite. The formation of these various structures is strongly dependent on the solution’s temperature and pH [21] [22]. It was found that, below a pH of ten, only microcrystalline structures were observed. In addition, it was found that, below a temperature of 60 °C, the structure of the hydroxides was either amorphous or poorly crystallized bayerite.

From this analysis, it can be expected that the formation of large precipitates in the sump following a LOCA would only occur in the initial stages of the break, when the temperatures would be highest. After the transient, the containment temperature is expected to be below the 60 degree Celsius threshold cited by Vermeulen [21], thus limiting the creation of phases that are unable to be dissolved.

2.2.3 Carbon Steel

Carbon steel is the main structural component within the containment [8]. The majority of the steel is coated in a design basis accident proofed coating that protects it against the high temperature water jet that may occur during a LOCA. Despite this,

17 corrosion may still occur. It can be safely assumed that carbon steels will behave similarly to pure iron because the only specified alloying material present is carbon, and it is only present as 0.12 to 2.0 percent by weight [23]. Carbon steel is also known as mild steel or low-alloy steel because of this composition. Thus, the Pourbaix diagram for pure iron, shown in Figure 2.3, is accurate enough for this alloy. In general, iron corrodes in acidic solutions, and is passive in neutral or basic solutions. However, as shown in

Figure 2.3, at high temperatures, it is possible for iron to corrode in basic solutions to

- form HFeO2 ions. However, at high enough potentials, such as those seen in water’s thermodynamically stable region, the iron will still form a protective oxide layer. This behavior is also known as active-passive behavior [24].

Figure 2.3: Pourbaix Diagram for Iron at 150 °C [25]

18

Similar to the other metals previously discussed, the iron ions produced by acidic corrosion before the pH is adjusted will form insoluble iron hydroxides when the pH in the containment sump increases. This hydroxide can form an unstable passive layer. This passive film itself will corrode producing the stable ferric oxide passive film, but also producing protons. These protons could decrease the local pH and produce more iron ions that will form further iron hydroxides, thus creating an autocatalytic reaction until the ferric oxide layer is too thick to allow bulk water transport [26]. Pitting corrosion, a similar autocatalytic corrosion process, may also occur, especially in the presence of chloride or nitrite ions. This form of corrosion would also lead to the creation of iron hydroxides at a rapid rate [27].

Once again, these precipitates are linked to the amount of corrosion products that are generated. Based on data from Griess and Bacarella [16] and Hall [28], the corrosion rate was estimated to be 19.0 g/m2/h (833 mil/year). Again looking at Jain, et al.’s thermodynamic simulation [8], the amount of corrosion products produced from the corrosion of carbon steel was estimated. The results are given in Table 2.6.

Table 2.6: Simulated Carbon Steel Corrosion Product Production [8] Surface Area Submerged Submerged Time Corrosion Product to Volume Fraction Surface Area to (h) Released per Liter (g) Ratio (m2/m3) Volume Ratio [mol] [ft2/ft3] (m2/m3) [ft2/ft3] 0.49 [0.15] 0.05 0.17 [0.05] 0.5 1.6 ×10−4 [2.8 ×10−5]

19 € €

Jain, et. al [7] later attempted to confirm these rates through measurements similar to the measurements used for zinc and aluminum. At 60 °C, the corrosion rate was measured to be 0.0135 g/m2/h (0.59 mil/year), at 90 °C, the corrosion rate was measured to be 0.0295 g/m2/h (1.29 mil/year), and at 110 °C, the corrosion rate was measured to be

0.0821 g/m2/h (3.60 mil/year). In a solution with a pH level of seven instead of ten, the results changed. At 60 °C, the corrosion rate was measured to be 0.127 g/m2/h (5.57 mil/year), at 90 °C, the corrosion rate was measured to be 0.0936 g/m2/h (4.10 mil/year), and at 110 °C, the corrosion rate was measured to be 0.0214 g/m2/h (0.94 mil/year).

These corrosion rates differ greatly from the estimated rate. This is because Hall [28] found that carbon steel was attacked much more aggressively by the spray than by the borated water. Since Jain’s experiments were run in borated water, these rates are expected to be different.

Despite the difference, several important trends are seen in the data that has also been observed by Hall [28]. The main trend is the decrease in the corrosion rate as temperature increases at lower pH levels, which is due to a result of the decreasing availability of oxygen, because the solubility of oxygen decreases as temperature increases.

2.2.4 Copper

Copper is mainly present in the containment in air lines and fan coolers [8]. It may also be present in scaffolding or insulation jackets. Unlike the previous metals that have been discussed, pure copper is thermodynamically stable in deaerated water,

20 regardless of whether the solution is acidic or basic. However, in aerated conditions, copper behaves similarly to zinc and aluminum: it corrodes at high and low pH levels, and is passive at neutral pH levels. The Pourbaix diagram detailing this aeration and pH dependence is provided in Figure 2.4. Figure 2.4 also contains the stable corrosion products and passive oxide layer compositions.

Figure 2.4: Pourbaix Diagram for Copper at 100 °C [29]

Copper ions are produced by the initially acidic solution, which will form copper hydroxides as the pH increases. These copper hydroxides are insoluble like most metal hydroxides. Despite this initial acidic corrosion, the major form of corrosion expected in 21 the containment sump will be corrosion of the passive film, which corrodes with a mechanism similar to that of carbon steel. The passive layer’s composition depends greatly on the potential of the solution [30]. At low potentials, a copper (I) oxide layer is formed. With increasing potential, a copper (II) oxide and copper (I) oxide mixture is formed, and at higher potentials still, a copper (II) hydroxide layer develops. The copper

(I) oxide layer may react with the hydroxide ions in solution, forming additional copper hydroxides [31]. In addition to the corrosion of the passive film, an accelerated form of corrosion is also possible. Similarly to the pitting corrosion seen in passive carbon steel, this type of accelerated attack on copper is caused by the presence of chloride ions [31].

The resulting copper chlorides are insoluble. Sulfide ions also cause a higher corrosion rate [32].

Once again, the amount of precipitates is reliant on the amount of corrosion product that is released. Griess and Bacarella [16] estimated the corrosion rate of copper to be 0.03 g/m2/h (1.2 mil/year). Using this corrosion rate estimate, Jain et al. [8] estimated the amount of copper corrosion products produced using a thermodynamic simulation. The results are provided in Table 2.7.

Table 2.7: Simulated Copper Corrosion Product Production [8] Surface Submerged Submerged Time Corrosion Product Area to Fraction Surface Area (h) Released per Liter (g) Volume to Volume [mol] Ratio Ratio (m2/m3) (m2/m3) [ft2/ft3] [ft2/ft3] 19.7 [6.0] 0.25 4.9 [1.5] 0.5 7.6 ×10−5 [1.2 ×10−6]

22 € €

Jain, et al. [7] later measured corrosion rates for copper. At 60 °C, the corrosion rate was measured to be 0.00478 g/m2/h (0.19 mil/year), at 90 °C, the corrosion rate was measured to be 0.0519 g/m2/h (2.1 mil/year), and at 110 °C, the corrosion rate was measured to be 0.0991 g/m2/h (4.0 mil/year). Again, while these measurements verify the estimated corrosion rate, these tests were all run in isothermal systems at low pressure, which may not accurately describe the true corrosion processes that may be seen in the containment sump following a LOCA.

2.2.5 Concrete

Concrete is the material that the containment itself is made from. It can be present in the containment sump either as an exposed surface, or as particles eroded away from the bulk material by the steam jet from the break causing the LOCA [8]. It was found that the eroded particles were much more prevalent than the exposed surfaces [8].

Concrete is made up of two parts, cement and an aggregate material. This aggregate material can vary with geography, but are generally naturally occurring rocks, such as limestone, or sand. These typically have similar compositions to the cement, but are inert and will not take place in the hydration reaction that makes concrete. However, there with exposure to carbon dioxide in air, various aggregates can react to produce carbonates, which can produce carbonic acid. Portland cement, the most common cement type used for reactor containments, is a mixture of tricalcium silicate, dicalcium silicate, tricalcium aluminate and tetracalcium aluminoferrite [33]. The overall percentage of these materials is given in Table 2.8.

23

Table 2.8: Composition of Portland Cement [34] Compound Chemical Formula Weight Percent (%) Tricalcium silicate Ca3SiO5 or 3CaO•SiO2 50 Dicalcium silicate Ca2SiO4 or 2CaOSiO2 25 Tricalcium aluminate Ca3Al2O6 or 3CaOAl2O3 10 Tetracalcium aluminoferrite Ca4Al2Fe2O10 or 4CaOAl2O3Fe2O3 10 Gypsum CaSO42H2O 5

This composition can be further broken down to the oxides that make up the compounds in Table 2.8. The oxide composition limits of Portland cement are given in

Table 2.9.

Table 2.9: Oxide Composition of Portland Cement [33] Oxide Weight Percent (%) CaO 60-67 SiO2 17-25 Al2O3 3-8 Fe2O3 0.5-6.0 MgO 0.5-4.0 Alkalis (as Na2O) 0.3-1.2 SO3 2.0-3.5

Different from the previous materials, concrete does not corrode in the traditional, electrochemical sense. Concrete can degrade either by hydration or by leaching. In either case, the main component of the cement, calcium oxide, is diffused from the concrete

24 matrix, and dissolved in the solution, producing calcium ions [35]. The secondary component of Portland concrete, silicon dioxide, may also be leached into solution. This behavior will be examined in detail in the fiber insulation section, as silicon dioxide is the main component of glass fiber.

The production of these corrosion products was simulated by Jain, et al. [8], using the initial corrosion rate estimate from Jantzen of 0.13 g/m2/h (14.7 mil/year) [36]. The results of the simulation are presented in Tables 2.10 and 2.11.

Table 2.10: Simulated Concrete (Non-Particulate) Corrosion Product Production [8] Surface Submerged Submerged Time Corrosion Product Area to Fraction Surface Area to (h) Released per Liter (g) Volume Volume Ratio [mol] Ratio (m2/m3) [ft2/ft3] (m2/m3) [ft2/ft3] 0.15 [0.045] 0.34 0.05 [0.015] 0.5 3.3×10−6 [1.9 ×10−8]

€ € Table 2.11: Simulated Concrete (Particulate) Corrosion Product Production [8] Surface Area Specific Submerged Time Corrosion Product to Volume Surface Area Surface (h) Released per Liter (g) Ratio (m2/g) Area to [mol] (m2/m3) [ft2/lbm] Volume [ft2/ft3] Ratio (m2/m3) [ft2/ft3] 22.25 [0.014] 15 [7.3×104 ] 333 [101] 0.5 1.0 ×10−2 [1.3×10−4 ]

€ € €

25 It can be seen from Tables 2.10 and 2.11 that the overall corrosion product contribution from concrete is due to the eroded particulates. Thus, only particulate concrete needs to be used for the proposed tests.

Jain, et al. also later measured the leaching rates of concrete samples [7]. Rather than use the glass cells or autoclave, polytetrafluoroethylene vessels were utilized. In addition, rather than using resistance polarization, a standard leaching method was used.

The results were separated by the leached oxide being measured, and are provided in

Table 2.12.

Table 2.12: Measured Leaching Rates of Major Components of Concrete [7] Oxide Leaching Rate at 60 Leaching Rate at 90 Leaching Rate at °C (g/m2/h) °C (g/m2/h) 110 °C (g/m2/h) CaO 1.02 1.65 3.08 Al2O3 0.057 0.111 0.168 SiO2 0.081 0.236 0.321

While these measurements verify the estimated corrosion rate, these tests were all run in isothermal systems at low pressure, which may not accurately describe the true corrosion processes that may be seen in the containment sump following a LOCA

The calcium ions produced from leaching and hydration can then form insoluble hydroxides like those seen in the other discussed materials. Many studies have shown that calcium hydroxide’s solubility decreases drastically with increasing pH [36].

2.2.6 Fiber Insulation

26 There are two principal types of thermal insulation used in RCS piping, fibrous insulation and expanded metallic insulation. Fibrous insulation has generally been preferred. Most fibrous insulation used in nuclear power reactors in the U.S. is called

Nukon, which is a brand of insulation produced by the PCI Company using an Owens-

Corning fiberglass [8]. The composition of Nukon is given in Table 2.13. Nukon, like many other fiber insulators, use a curing agent to bond the fibers together. This curing agent, normally phenol formaldehyde, converts to an insoluble polymer.

Table 2.13: Composition of Nukon Fiber Insulation [8] Oxide Weight Percent (%) SiO2 62.5 Al2O3 3.6 CaO 8.2 MgO 3.45 Na2O 15.8 B2O3 5.0

There is little material regarding the corrosion of this specific fiber insulation, but there are several papers regarding the corrosion of E-glass fiber. E-glass fiber is similar in composition to Nukon, as shown in Table 2.14.

Table 2.14: Composition Limits of E-Glass Fiber [37] Oxide Weight Percent (%) SiO2 52-58 Continued

27 Table 2.14: Continued CaO 16-25 Al2O3 12-16 B2O3 5-10 MgO 0-5 Na2O and K2O 0-2 TiO2 0-0.8 Fe2O3 0.05-0.4 CaF2 0-1.0

Like concrete, fiber insulation does not corrode in the traditional, electrochemical sense either; it also degrades by leaching or by hydration. For fiber insulation, these processes are highly dependent on the pH of the solution. In acidic leaching, the most common corrosion products are calcium and aluminum ions. In essence, these ions are depleted from the original fiber matrix and replaced by protons [37]. It can be assumed that more ions will be leached in the presence of more protons, as this process is largely driven by the ionic concentrations within the matrix and the solution [38]. This same type of process is observed in basic solutions [38]. The main silicon dioxide component is largely unaffected by this process at low temperatures. This is due to the extremely low solubility of silicon dioxide in water solutions, and the preferential dissolution of the calcium and aluminum oxides. However, at high temperatures, the dissolution of silicon dioxide is more likely to take place [39]. A relation of expected silicon dioxide concentration in water versus temperature at the corresponding saturation pressure was fit to data collected by Fournier and Rowe [40] and is given in Equation 2.4.

28 731 log(C) = 4.52 − (2.4) T

€

In this equation, C is the concentration of silicon dioxide in an aqueous solution measured in milligrams of silicon dioxide per kilogram of water and T is the absolute temperature of the solution. This equation holds for temperatures ranging from 0 to 250

°C (273 to 523 K). From this equation, it is expected that a maximum of approximately

509 mg/kg could be dissolved in the sump at the maximum temperature seen after a

LOCA. As the temperature gradually decreases, it is expected that the silicon dioxide will precipitate out of solution in large amounts.

Jain’s calculated estimate of fiberglass surface area, the parameters of which are given in Table 2.15, was used as the initial condition for the thermodynamic simulations

[8]. Table 2.16 details the simulated amounts of each specific corrosion product released into the solution, which were based on the corrosion rate measurements by Pan, et al., which found a corrosion rate of 0.025 g/m2/h (3.4 mil/year) [41].

Table 2.15: Estimation of Fiber Insulation Concentration [8] Parameter (SI Units) [English Units] Value Nukon Volume to Water Volume Ratio 0.137 [0.137] (m3/m3) [ft3/ft3] Immersed Fraction 0.75 Immersed Nukon Volume to Water 0.103 [0.103] Volume Ratio (m3/m3) [ft3/ft3] Density of Nukon (kg/m3) [lbf/ft3] 38.1 [2.4] Continued

29 Table 2.15: Continued Weight of Nukon to Water Volume Ratio 3.9 [0.25] (kg/m3) [lbf/ft3] Density of Fiber Glass (kg/m3) [lbf/ft3] 2500 [157] Suface Area to Volume Ratio of Fiber 4.3×105 [1.3×105 ] Glass (m2/m3) [ft2/ft3] Fiber Glass Surface Area to Water Volume 678 [207] Ratio (m2/m3) [ft2/ft3] € Time (h) € 0.5 Amount of Fiber Glass in Insulation (g/L) 8.5 ×10−3 [1.4 ×10−5] [mol/L]

€ €

Table 2.16 Simulated Nukon Insulation Corrosion Product Production [8] Oxide Amount Released per Liter (mol) −5 SiO2 9.0 ×10 −6 Al2O3 3.0 ×10 CaO 1.3×10−5 MgO 7.4 ×10−6 € −5 Na2O 2.2 ×10 € −6 B2O3 6.2 ×10 € € € € Jain, et al. [7] later attempted to confirm these rates through measurements similar to the measurements used for concrete. At 60 °C, the corrosion rate was measured to be

0.0486 g/m2/h (6.6 mil/year), at 90 °C, the corrosion rate was measured to be 0.232 g/m2/h (31.6 mil/year), and at 110 °C, the corrosion rate was measured to be 0.453 g/m2/h

(61.6 mil/year). In a solution with a pH level of seven instead of ten, the results changed.

At 60 °C, the corrosion rate was measured to be 0.0109 g/m2/h (1.5 mil/year) , at 90 °C, the corrosion rate was measured to be 0.075 g/m2/h (10.2 mil/year), and at 110 °C, the corrosion rate was measured to be 0.143 g/m2/h (19.4 mil/year). Again, while these measurements verify the estimated corrosion rate, these tests were all run in isothermal

30 systems at low pressure, which may not accurately describe the true corrosion processes that may be seen in the containment sump following a LOCA.

31 Chapter 3: Experimental Setup and Procedure for Autoclave Tests

The environment of the sump of a containment following a LOCA is chemically aggressive. As described previously, during the transient process, maximum sump temperatures can approach130 °C and maximum gauge pressures can reach 36 psi. Once the containment environment has reached equilibrium, temperatures of up to 55 °C and atmospheric pressures are observed [5].

To measure corrosion rates in these conditions is exceptionally difficult. Thus, for these tests, the baseline environment is used. That is, the environment will be that of the sump at steady state, with no flow. In addition to this baseline, which is similar to the experiments run by Jain, et al. [7], this system will also be capable of measuring corrosion rates in simulations of the harshest environment seen in the sump: temperatures of 130 °C and gauge pressures of 36 psi. Modeling the pressure transient allows for high temperatures without the solution coming to a boil. In addition to this necessity, the added molecular energy from the pressure could have some effect on the corrosion rates, especially for samples that are exposed to the spray. Although this may be negligible, no research has been found to suggest that it actually is so.

The following sections detail the equipment that will be used for the experiments, and the procedures that will be used to generate the corrosion rates for each material

32 under several different conditions including differing water chemistry and possible inhibition effects.

3.1 Experimental Setup

To measure the corrosion rates of the various materials in an environment similar to that of the sump of the containment following a LOCA, several pieces of specialized equipment were needed. A small pressure vessel, known as an autoclave, was required to run the tests in the high-temperature, high-pressure environment. In addition to the vessel itself, the measurement instrumentation needed to be able to survive this environment.

Other equipment required to create the environment include a heater and a heater controller, and several hydrogen gas canisters.

This section describes each component, and its intended use.

3.1.1 Autoclave

To prevent water from flashing at high temperature, a pressure vessel is needed.

However, since the proposed experiments will only test one or two small samples at a time, a two-liter autoclave was determined to be sufficient.

Other than the requirements that the autoclave be suited for temperatures of up to at least 130 °C and pressures up to 36 psi, the autoclave needed to be corrosion resistant so it would not interfere with the measurements. The material selected for the autoclave is alloy 316 stainless steel. For the possibility of future research, the autoclave should also be able to handle maximum temperatures of 210 °C, and maximum pressures of

2200 psi.

It was determined that several penetrations through the autoclave were needed so that measurements may be taken. The first of these penetrations, for a pressure gauge, is

33 also a main component for safe operation. A pressure gauge is necessary to measure the pressure of the system both to properly maintain the environment needed for the experiments, but also to ensure that, in the event the pressure rises, proper actions can be taken to bleed the autoclave in accordance with the regulations of the American Society of Mechanical Engineers. Another penetration, a thermowell, is necessary for temperature measurements and control. For these experiments, a type-K thermocouple will be used in the thermowell. To allow for pressure control and deaeration of the water, a blowpipe was added so that hydrogen gas may be bubbled through the water. With the lid closed, this hydrogen mixture will be allowed to build and form a gas bubble, thus pressurizing the system. The last set of penetrations allow for access of the electrodes required for meaningful corrosion measurements to be made. One NPT connection was created to allow wires to be fed through the autoclave to be connected to the working and counter electrodes. The other NPT connection allowed a reference electrode to enter the autoclave. The reasoning behind these connections will be discussed with the electrodes and the potentiostat.

A suitable autoclave system was supplied by the Parker Hannifin Corporation through their Autoclave Engineers Operations division. The company was selected for the high amount of customization they offered in terms of the various penetrations, as well as other supplies that were made specifically for the autoclave, such as a heater and a bench-top stand. Figure 3.1 shows the basic autoclave before any customization.

34

Figure 3.1: EZE-Seal Autoclave from Parker Hannifin Corporation [42]

For the NPT connections required for the electrode access, Conax Technologies was chosen to be the supplier because they were recommended by the Autoclave

Engineers Operations division due to prior experience between the two companies. One packing gland was selected for the reference electrode penetration, and one power lead gland was chosen to allow wires into the system for online measurements. These connections are shown in Figures 3.2 and 3.3. In addition to the prior experience between the two companies, these glands were also selected because of the large selection of seals that allow for use in various environments.

35

Figure 3.2: Packing Gland from Conax Technologies [43]

Figure 3.3: Power Lead Gland from Conax Teachnologies [44]

This system is able to operate at 3300 psi at 204.4 °C, both above what is required for the proposed experiments.

3.1.2 Potentiostat; Working, Counter and Reference Electrodes; and Wiring

Corrosion measurements are generally taken using a potentiostat. This instrument is able to vary the electrical potential that drives the corrosion reactions, and measure the currents generated by the electrochemical corrosion processes so that a plot of the corrosion potential as a function of the corrosion current may be created. After this plot is 36 created, Tafel analysis may be used to determine the proper corrosion potential and corrosion rate for the material in question.

In order for the potentiostat to accomplish this, three types of electrodes are needed [45]. The first type of electrode is the working electrode. This electrode is the specimen itself. As the specimen corrodes, it will produce electrons that will be used in a cathodic reaction, such as the reduction of protons to hydrogen gas. For these electrons to flow to generate a current, a counter electrode is required. A counter electrode is generally a noble metal, like platinum, so that it acts as a catalyst, breaking down the products formed from the electrochemical reactions of the working electrode and taking the electrons for itself, or directly using the electrons from the working electrode. Used together, they can create an electric circuit. Thus, both the working and counter electrode must be electrically connected to the leads of the potentiostat.

For any meaning to be derived from the current generated from the working and counter electrodes’ interactions, a reference electrode must be used. A reference electrode uses a known half-cell reaction to establish the datum of zero potential. This half-cell reaction must be able to transfer electrons to the solution so a full electrochemical reaction is made. This is generally accomplished through use of a salt bridge. By convention, the datum is usually the reduction of hydrogen. This type of electrode is known as a standard hydrogen electrode (SHE). For these high-temperature and high- pressure experiments, SHEs were not able to function. Thus, a silver and silver chloride

(Ag/AgCl) reference electrode was selected.

37 Corr Instruments was the vendor that could supply an Ag/AgCl reference electrode that fit the necessary requirements. Their model works up to 2000 psi and up to

305 °C. This electrode is shown in Figure 3.4.

Figure 3.4: High Pressure Ag/AgCl Reference Electrode from Corr Instruments [46]

For the working and counter electrodes, simple test coupons can be used. These test coupons were selected from Metal Samples, a branch of Alabama Specialty Products.

For the working and counter electrodes to be electrically connected to the leads of the potentiostat, electrically conducting wires need to be used. In addition to the wires needing to carry an electrical current, they should be corrosion resistant so that they do not add to the measured corrosion current. Thus, molybdenum or tantalum wire should be used. Tantalum is much more corrosion resistant at high temperatures, but is more expensive.

Lastly, the potentiostat selected was the SP-300 model supplied by Bio-Logic

Science Instruments, shown in Figure 3.5.

38

Figure 3.5: Model SP-300 Potentiostat from Bio-Logic Science Instruments [47]

3.1.3 Hydrogen Canisters, Heater, Thermocouples and Temperature Controller

In order to create the environment of the sump of a containment following a

LOCA, several pieces of equipment were needed. In general the environment needed was one of high pressure, high temperature, all submerged in a corrosive water solution.

The part to supply the high-pressure environment is the hydrogen canister. This canister allows for the injection of an inert hydrogen and helium mixture into the autoclave. This serves a dual purpose. Not only is the hydrogen used to pressurize the autoclave; it is also used to deaerate the water. Hydrogen is highly flammable, so a mixture of hydrogen and helium should be used to prevent explosions. Praxair was chosen as the supplier for the gas mixture, which was selected to be 3.9 percent hydrogen, which is near the upper bound of, but well within, the safe operation limit.

39 To heat the water to the high temperatures seen after a LOCA, a special heater was supplied by the Autoclave Engineers Operations division. This heater is a blanket- type heater, and can be wrapped around the autoclave both supplying heat to the water and insulating the system. This heater is controlled by a universal controller, also supplied by the Autoclave Engineers Operations division.

3.2 Experimental Procedure

The first step for these experiments is to prepare the water. The water chemistry needs to be that of the water in the containment sump. This mixture can be accomplished by standard chemistry procedures on a bench-top as the general balance of the buffer chemicals, primarily sodium hydroxide or TSP, is known, and the pH level needs to be 10 for sodium hydroxide systems and 7 for TSP systems [6].

There needs to be some airspace in the autoclave, so less than two liters should be prepared for each test. However, since the water will need to be replaced with each test, it is prudent to make several batches at once. A sample should be taken of this water as a control sample for mass spectroscopy measurements. The water should then be poured into the autoclave.

The next step is to prepare the samples and the electrodes. The working and counter electrodes should be small coupons, with areas on the order of a few square centimeters. The working electrode and the counter electrode must be hooked up to the potentiostat leads with corrosion resistant wires, made from materials such as molybdenum or tungsten. These connections must be very stable so that they do not break off the sample during a test. As such, it may be necessary to solder them in place.

The electrodes should then be added to the autoclave.

40 With the cover not tightened, but on the autoclave and in place for tightening, the inert hydrogen and helium mixture should be bubbled through the water to deaerate the solution. Having the autoclave not tightened allows for the air in the autoclave to be evacuated by the inert hydrogen and helium mixture. After suitable water chemistry is attained, the cover should be tightened. At this time, the reference electrode and wire feed through glands should also be tightened.

The heater should then be set to bring the autoclave up to the desired temperature.

After the desired temperature is attained, the pressure should be brought up to the desired level by use of the hydrogen canisters. Since hydrogen and helium naturally leak out of systems, the pressure will need to be maintained throughout the duration of a test.

A potentiodynamic scan can be run now that the environment within the autoclave is representative of the containment sump. Several scans of each material should be run so that good statistics will be achieved. For each test, the water and the working electrode should be replaced, and the autoclave and other electrodes cleaned.

Once a test is finished, the heater should be turned off so that the water may return to room temperature conditions. Water samples may be taken for use in mass spectroscopy so that corrosion products, precipitates, and any other important chemical species. These concentrations can then be compared to the initial water samples.

To assure that all materials can be tested, and the interactions between all materials can be observed, 21 tests need to be run according to the test matrix provided in

Table 3.1.

41 Table 3.1: Test Matrix for Autoclave System Test Zinc Aluminum Carbon Steel Copper Concrete Fiber Insulation 1 X 2 X 3 X 4 X 5 X 6 X 7 X X 8 X X 9 X X 10 X X 11 X X 12 X X 13 X X 14 X X 15 X X 16 X X 17 X X 18 X X 19 X X 20 X X 21 X X

Both containment water chemistries (sodium hydroxide and TSP) need to be tested. Thus,

42 tests will insure that each material’s corrosion rate is measured both on its own, and with each other material to observe any potential corrosion inhibition effects in both possible water chemistries.

42 Chapter 4: Experimental Setup and Procedure for a Non-isothermal Test Loop

Once the underlying physics for each material has been observed with the proposed autoclave system, an integrated test can be run to fully simulate the containment sump environment. This test will not only run with each sump material with proper surface area to water volume ratios; it will also be able to fully simulate the temperature transient described by Rao, et al. [5]. It is this ability to simulate the transient behavior that distinguishes this design from UNM’s ICET and CHLE loops.

The following sections detail the equipment that will be used for the experiments, and the procedures that will be used to run the integrated tests.

4.1 Experimental Setup

The integrated tests will need to be run in a non-isothermal environment. To accomplish this, several key components are required. These components include a pressure vessel, a heat exchanger, and several test columns. Several minor components, while just as important, will not be covered because they are standard parts with low degrees of customization. These parts include, but are not limited to, piping and pumps.

Although the minor parts will not be discussed, it is important to cover the desired parameters that the overall system must provide so that the requirements of these parts may be defined. From Rao, et al. [5], all components must be able to withstand the maximum temperature and pressure in the containment sump: 130 °C and 36 psi. In

43 addition to these maxima, the materials used must not be sensitive to large changes in either temperature or pressure, as the minimum values are 55 °C and atmospheric pressure. These environmental factors are not the only parameters that the components must satisfy. The water flow velocity over the coupons should simulate the flow rates in the containment that result from ECCS and containment spray following a LOCA.

Dallman, et al. [6] and Bahn, et al. [48] both suggest that the maximum flow velocity seen in the sump is three cm/s, although this is highly variable between different plants.

This flow velocity implies that the average volumetric flow rate that must be supplied is

25 gallons per minute, which should be able to be varied slightly for different plants’ approach velocities.

The initial design and the layout of the loop are presented in Figure 4.1.

Figure 4.1: Non-isothermal Test Loop Concept Drawing

44

This section details each major structural part and their intended use for the experiments.

4.1.1 Pressure Vessel

Just as an autoclave was the major component of the autoclave system, a pressure vessel is the major component for the non-isothermal test loop. This pressure vessel allows for the safe use of the high-temperature and high-pressure water that is needed to simulate the containment sump following a LOCA. As stated previously, the maximum temperature and gauge pressure expected in the sump are 130 °C and 36 psi. The pressure vessel should be rated for higher temperatures and pressures to allow for a small factor of safety.

In addition to being able to simulate and hold the high-temperature and high- pressure environment, the pressure vessel should also be able to simulate the duality of the sump environment. Not all materials are submerged under water in the containment, and water can be injected into the sump either by spray or by direct injection. Thus, the vessel should have shelves at varying heights that will be able to hold test coupon racks both below and above the waterline. The vessel should also have several penetrations in order to simulate both spray and direct injection.

Aside from penetrations for the piping, there should also be penetrations for several immersion heaters. A typical immersion heater is given in Figure 4.2.

45

Figure 4.2: Immersion Heater from Chromalox [49]

These heaters will be able to sustain the temperature in the vessel and add to the overall heating effect from the heat tape pre-heater.

There should also be several sight glasses so that the fiber bed at the bottom of the tank may be observed. These windows must be placed in a way such that the entirety of the fiber bed can be observed.

In order to load the test coupons into the vessel, the lid of the vessel should be removable. This lid should also have several penetrations for various sensors and safety features including a pressure relief valve, a hydrogen sensor and a thermowell.

In addition to the main pressure vessel, there should also be a preparation tank.

This tank will not need a removable lid, but will need an inlet so that water and the necessary pH control chemicals may be added. There should be a few penetrations for

46 immersion heaters so that the water may be heated up to temperature before it is pumped

into the main vessel and a thermowell.

4.1.2 Heat Exchanger and Pre-Heater

To have a non-isothermal system, and thus truly simulate the ECCS, a heat

exchanger and a pre-heater must be incorporated into the flow path. The design of heat

exchangers follows a guess and check method by assuming several values (generally

those that can be controlled easily like mass flow rates) to be fixed so that the unknown

values can be found. This process utilizes Equations 4.1 and 4.2 to calculate the total heat

transfer rate [50].

Q˙ = UAT = m˙ c ΔT = m˙ c ΔT (4.1) T LMD ( p )hot ( p )cold

€

(T1,hot − T2,cold ) − (T2,hot − T1,cold ) TLMD = (4.2)  T − T  ln 1,hot 2,cold   T2,hot − T1,cold 

€

˙ Here, QT represents the total heat transfer rate, U is the total heat transfer coefficient, A is

the heat transfer area, TLMD is the logarithmic mean temperature difference, m˙ is the mass

€ flow rate, cp is the specific heat capacity at constant pressure and ΔT represents the

temperature change. The subscripts hot and cold refer for which working€ fluid these

47 parameters should be calculated, with the hot subscript referring to the primary test water in the loop and the cold subscript referring to the coolant. In addition, the subscripts 1 and

2 refer to either the inlet or the outlet conditions of the working fluid, with 1 being the inlet and 2 being the outlet.

From Equation 4.1, the unknown parameters of the coolant mass flow rate and total heat transfer area can be found using the known hot side parameters and by assuming that the coolant can be rejected at a fixed temperature of 50 °C and that the total heat transfer coefficient is known to be approximately 15 kW/m2. It was determined that the maximum temperature drop across the heat exchanger (or temperature rise across the pre-heater) for the primary test water should be 20 °C. Based on the maximum sump flow velocity of three cm/s, the volumetric flow rate through the hot side of system should be 25 gallons per minute. Multiplying this volumetric flow rate by the average density of the fluid calculates the mass flow rate of the fluid through the piping. Also using the average heat capacity, the total heat transfer rate could be calculated. This total heat transfer rate was determined to be 130 kW. From this number, the coolant flow rate should be 13 gallons per minute, and the heat transfer area should be on the order of 0.2 m2.

Based on these numbers and the need to conserve space in the overall loop footprint, a spiral heat exchanger was determined to be best.

48

Figure 4.3: Spiral Hear Exchanger from Sentry Equipment [51]

Rather than use another counter-flow setup for the pre-heater, it was determined that wrapping a section of the piping with heat tape would be the best solution. This section of heat tape covered pipe would need to supply the 130 kW that were removed by the heat exchanger. This preheating would be supplemented by the immersion heaters that are in the pressure vessel. An example of the application of heat tape is shown in

Figure 4.4. This may be slightly unrealistic, as several tens of feet of heat tape may be needed to assure proper preheating.

49

Figure 4.4: Heat Tape from BriskHeat [52]

4.1.3 Test Columns

The other major components are the test columns. These columns will hold small fiberglass beds across which the pressure differential will be monitored. It is important that the fiberglass bed is removable, and that it can be seen during operation. UNM’s

CHLE loop incorporated several columns that had a translucent polymer section. Based on the successes of the CHLE loop, the proposed non-isothermal loop will have at least two columns of similar design to those of the CHLE loop.

The columns must both be able to withstand the high-temperature, high-pressure environment of the containment sump. Thus, the flanged portions of the column will be made from stainless steel. The major problem is the polymer section of the column because of potential swelling or cracking issues. It is highly important that the selected polymer be able to withstand up to 130 °C and up to 36 psi. In addition, it must also have a low glass transition temperature so that it does not crack at room temperature when the

50 loop is not running an experiment or when the loop is being run at the low temperatures toward the end of an experiment.

The test column must also be designed such that the flow at the fiberglass bed is fully developed. As the test column represents a sudden expansion in the flow area, there will be some head loss due to the loss of velocity and gain in pressure as described by the

Bernoulli Equation [53]. However, there should not be a significant entrance or reattachment length associated with the expected eddy currents. As such, the test columns should be several feet in length, but do not need to be any more than six feet in length.

It is important to note that one of the test columns will be positioned after the heat exchanger, and thus runs colder than the first test column. This allows for the temperature effects on precipitation out of solution to be observed. The test columns should also be isolatable from the main loop.

4.2 Experimental Procedure

The first step in the integrated experiments is to prepare the solid samples.

Fiberglass insulation beds should also be prepared and placed at the bottom of the main pressure vessel and in both test columns. The test coupons will be one by one foot squares. These coupons should then be loaded in the coupon racks, which should then be hoisted into the pressure vessel.

As with the autoclave experiments, the water chemistry is a large portion of the experimental procedure. In these tests, the initial water chemistry should be that of the primary system water of a PWR: it should contain boric acid with some lithium hydroxide to balance the pH. The TSP or sodium hydroxide buffer solutions that are within containment should be placed in the main pressure vessel, or sprayed into the

51 vessel during the initial stages of the experiment so that they will dissolve in solution as the water from the preparation tank is pumped in, thus simulating a LOCA. Once this final preparation is completed, the pressure vessel should be sealed. It is important that the initial borated water is prepared in the preparation tank and that the pressure vessel is initially dry. The water in the preparation should then be heated to the maximum sump temperature. Once the borated water is up to temperature, it should be pumped into the main pressure vessel.

Once the pressure vessel has been filled to the desired volume, the water can then be pumped throughout the entire test loop. The chilled water flow rate through the heat exchanger should be controlled so that the temperature profile of the sump after a LOCA is maintained. Once the flow through the system has stabilized at its steady state, samples should be taken periodically so that several parameters of the solution, such as pH and turbidity, can be measured. In addition, the test columns should be monitored closely for both the differential pressure across the fiberglass bed and any visible precipitation effects at the fiberglass bed. The tests can run up to 30 days.

After an experiment has been finished, the test coupons and fiberglass beds need to be analyzed for corrosion product deposition or passive layer growth. This analysis will be accomplished using several tests including x-ray diffraction, mass spectrometry and electron microscopy.

Tests can be run to simulate the conditions of any PWR, so there will be many different tests from which to generate data.

52 Chapter 5: Conclusions and Future Research

It is evident from currently available research that the problem regarding a potential pressure drop across the strainer in the containment sump of a nuclear reactor caused by corrosion product buildup is still a long way from being solved. Much progress has been made since the issuing of GSI-191 in the form of thermodynamic simulations and several integrated and benchtop tests.

This section reviews the major findings of this paper, as well as suggests ways to continue research in this subject matter.

5.1 Conclusions

Thermodynamic simulations predict that various precipitates will form.

Depending on the pH of the solution (and thusly which buffer is used), complex silicates, phosphates or hydroxides may be formed. These complex compounds were generally not observed during integrated or benchtop tests. However, many types of precipitates were found including a gel-like calcium phosphate.

Corrosion rates for each common sump material have been researched and presented in various, currently available studies. Although these rates were measured, they were measured in an environment much simpler than that of the containment sump; it was at high temperature and in a solution similar to the water in the sump, but it was not at high pressure. The measurements also did not consider any transient effects, or

53 look at any potential interactions between materials. As such, the proposed autoclave system could verify whether these are minor issues in terms of their effects on the corrosion rate.

5.2 Future Research

To continue this research, it is recommended that both of the proposed systems be built. With the autoclave system, the corrosion rates of each material can be observed in a simulated containment environment. This system also makes it possible to observe any inhibition or acceleration effects that may be caused by interaction between several of the materials. Once these tests have been run, the non-isothermal test loop should also be built. With this system, the overall effects of sump material corrosion on head loss can be observed. Together, these systems will insure a risk informed solution of GSI-191.

However, these systems still do not address some issues that may have an effect on the overall chemical effects. These issues, brought to light by Torres [54], include the simulation of radiation and carbon dioxide uptake effects. Future systems should build on the proposed non-isothermal system, and should have the ability to simulate these problem areas to observe whether their effects are significant.

54 References

[1] “Potential Plugging of Emergency Core Cooling Suction Strainers by Debris in Boiling Water Reactors”. Nuclear Regulatory Commission Bulletin, NRCB 96-03 (1996).

[2] Zigler, G., et al. “Parametric Study of the Potential for BWR ECCS Strainer Blockage Due to LOCA Generated Debris”. NUREG/CR-6224 (1995).

[3] Delegard, C. H., et al. “Final Report – Evaluation of Chemical Effects Phenomena in Post-LOCA Coolant”. NUREG/CR-6988 (2009).

[4] Howe, Kerry and Cody Williams. “Debris Bed Requirements and Preparation Procedures”. Report to South Texas Project Nuclear Operating Company, CHLE- 007, Revision 3 (2012).

[5] Rao, D. V., et al. “GSI-191 Technical Assessment: Parametric Evaluations for Pressurized Water Reactor Recirculation Pump Performance”. NUREG/CR-6762, Vol. 1. (2002).

[6] Dallman, J., et al. “Integrated Chemical Effects Test Project: Consolidated Data Report”. NUREG/CR-6914, Vol. 1 (2006).

[7] Jain, V., et al. “Corrosion Rate Measurements and Chemical Speciation of Corrosion Products using Thermodynamic Modeling of Debris Components to Support GSI- 191”. NUREG/CR-6873 (2005).

[8] Jain, V., et al. “Chemical Speciation, Using Thermodynamic Modeling, During a Representative Loss-of-coolant Accident Event”. Report to the Nuclear Regulatory Commission, CWNRA 2004-07. (2004).

[9] McMurray, J., et al. “GSI-191 PWR Sump Screen Blockage Chemical Effects Tests: Thermodynamic Simulations”. NUREG/CR-6912 (2006).

[10] Enderlin, C. W., et al. “Experimental Measurements of Pressure Drop Across Sump Screen Debris Beds in Support of Generic Safety Issue 191”. NUREG/CR-6917 (2006).

[11] McCauley, Ronald A. Corrosion of Ceramic Materials. 3rd ed. New York: CRC

55 Press, 2013.

[12] Pourbaix, Marcel. Atlas of electrochemical equilibria in aqueous solutions. Houston: National Association of Corrosion Engineers, 1974.

[13] Klasky, M., et al. “Integrated chemical effects tests: Data analysis report and literature review”. Report to Los Alamos National Laboratory, LA-UR-05-4881. (2005).

[14] Holleman, Arnold F., et al. “Zink”. Lehrbuch der Anoranischen Chemie. Walter de Gruyter, 1985. 1034-1041.

[15] Piipo, J., et al. “Corrosion Behavior of Zinc and Aluminum in Simulated Nuclear Accident Environments”. Report to the Finnish Center for Radiation and Nuclear Safety, STUK-YTO-TR 123. (1997).

[16] Griess, J. C. and A. L. Bacarella. “Design of Reactor Containment Spray Systems – Part III: The Corrosion of Materials in Spray Solutions”. Report to Oak Ridge National Laboratory, ORNL-TM-2412, Part III. (1969).

[17] Niyogi, K. K., et al. “Corrosion of Aluminum and Zinc in Containment Following a LOCA and Potential for Precipitation of Corrosion Products in the Sump”. Proceedings of the Second International Conference on the Impact of Hydrogen on Water Reactor Safety, Albuquerque, New Mexico, October 3-7, 1982. Rockville, Maryland: Nuclear Regulatory Commission, 1982.

[18] Pandya, Kaumudi I., et al. “EXAFS Investigations of Zn(II) in Concentrated Aqueous Hydroxide Solutions”. Journal of Physical Chemistry 99 (1995): 11967- 11973.

[19] Zhang, J., et al. “The aluminum chemistry and corrosion in alkaline solutions”. Journal of Nuclear Materials 384 (2009): 175-189.

[20] Goldberg, S., et al. The Environmental Chemistry of Aluminum. 2nd ed. Ed. G. Sposito. CRC Press, 1996.

[21] Vermeulen, A. C., et al. “Hydrolysis-Precipitation Studies of Aluminum (III) Solutions – Part I. Titration of Acidified Aluminum Nitrate Solutions”. Journal of Colloid and Interface Science 51.3 (1975): 449-458.

[22] Bahn, C. B., et al. “Evaluation of precipitates used in strainer head loss testing: Part III. Long-term aluminum hydroxide precipitation tests in borated water”. Nuclear Engineering and Design 241 (2011): 1914-1925.

[23] “Classification of Carbon and Low-Alloy Steels”. Key to Metals AG, n.d.

56

[24] Corrosion: Understanding the Basics. Ed. Joseph R. Davis. Materials Park, OH: ASM International, 2000.

[25] Townsend, H. E. “Potential-pH Diagrams at Elevated Temperature for the System Fe-H2O”. Corrosion Science 10 (1970): 343-358.

[26] Haupt, S. and Strehblow, H. H. “Corrosion, Layer Formation, and Oxide Reduction of Passive Iron in Alkaline Solution: A Combined Electrochemical and Surface Analytical Study”. Langmuir 3.6 (1987): 873-885.

[27] Valcare, V. B. and Vázquez, M. “Carbon steel passivity examined in alkaline solutions: The effect of chloride and nitrite ions”. Electrochimica Acta 53.15 (2008): 5007-5015.

[28] Hall, J. F. “Corrosion of Low Alloy Steel Fastener Materials Exposed to Borated Water”. Proceedings of the Third International Symposium on Environmental Degradation of Materials in Nuclear Power Systems – Water Reactors. G. J. Theuss and J. R. Weeks, eds. La Grange Park, Illinois: American Nuclear Society, 1988: 711-722.

[29] Beverskog, B. and Puigdomenech, I. “Revised Pourbaix Diagrams for Copper at 25 to 300 °C”. Journal of the Electrochemical Society 144.10 (1997): 3476-3483.

[30] Strehblow, H. H. and Speckmann, H. D. “Corrosion and layer formation of passive copper in alkaline solutions”. Materials and Corrosion 35.11 (1984): 512-519.

[31] Kwok, C. T., et al. “Effect of pH on corrosion behavior of CuCrZr in solution without and with NaCl”. Journal of Nuclear Materials 394 (2009): 52-62.

[32] Taniguchi, Naoki and Kawasaki, Manabu. “Influence of sulfide concentration on the corrosion behavior of pure copper in synthetic seawater”. Journal of Nuclear Materials 379 (2008): 154-161.

[33] Neville, A. M. Properties of Concrete. 5th ed. New York: Prentice Hall, 2012.

[34] “Concrete: Scientific Principles”. MAST Teacher’s Workshop. University of Illinois, n. d.

[35] Rechenburg, W. and S. Sprung. “Composition of the Solution in the Hydration of Cement”. Cement and Concrete Research 13 (1983): 119-126.

[36] Jantzen, C. M. “Radioactive Waste – Portland Cement Systems: II, Leaching Characteristics”. Journal of the American Ceramic Society 67.10 (1984): 674-678.

57 [37] Qiu, Q. and M. Kumosa. “Corrosion of E-glass Fibers in Acidic Environments”. Composites Science and Technology 57 (1997): 497-507.

[38] Helfferich, F. Ion Exchange. New York: McGraw-Hill, 1962.

[39] Al Cheikh, A. and M. Murat. “Kinetics of Non-Congruent Dissolution of E-Glass Fiber in Saturated Calcium Hydroxide Solution”. Cement and Concrete Research 18 (1988): 943-950.

[40] Fournier, Robert O. and Jack J. Rowe. “The solubility of amorphous silica in water at high temperatures and high pressures”. American Mineralogist 62 (1977): 1052-1056

[41] Pan, Y., et al. “Degradation of High-Level Waste Glass Under Simulated Repository Conditions”. Journal of Non-Crystalline Solids 319 (2003): 74-88.

[42] “EZE-Seal Closure Pressure Vessels”. Autoclave Engineers. Parker Hannifin Corporation, 2013.

[43] “Packing Glands (PG)”. Conax Technologies, 2013.

[44] “Insulated Wire: Power Lead Glands (PL)”. Conax Technologies, 2013.

[45] McCafferty, E. Introduction to Corrosion Science. New York: Springer, 2010.

[46] “High-Pressure Ag/AgCl Reference Electrodes”. Corr Instruments, 2012.

[47] “SP-300 Modular Research Grade Potentiostat/Galvanostat/FRA”. BioLogic Science Instruments, 2013.

[48] Bahn, C. B., et al. “Evaluation of precipitates used in strainer head loss testing. Part I. Chemically generated precipitates”. Nuclear Engineering and Design 239 (2009): 2981-2991.

[49] “Immersion Heaters”. Chromalox Precision Heat and Control, 2013.

[50] Bell, K. J. and A. C. Mueller. Wolverine Tube Heat Transfer Data Book. Wolverine Division of UOP Inc., 1984.

[51] “Spiral Tube Heat Exchanger”. Sentry Equipment Corporation, 2012.

[52] “HSTAT Silicone Rubber Heating Tape with Adjustable Thermostat Control”. BriskHeat, 2013.

[53] Munson, Bruce R., et al. Fundamentals of Fluid Mechanics. 6th ed. John Wiley &

58 Sons, 2009.

[54] Torres, P. A. “Peer Review of GSI-191 Chemical Effects Research Program”. NUREG-1861 (2006).

59