Corrosion and deposition on the secondary circuit of steam generators

ABSTRACT

This study conducts a critical review on the studies of material corrosion and deposition on the secondary circuit of a pressurized water reactor, especially on the steam generators (SGs). Available knowledge has shown that the structural materials in the environment of the secondary circuit are susceptible to flow-accelerated corrosion and deposition-induced degradation. The deposition of the non-volatile impurities, especially the corrosion products, on the SG surfaces can be a primary cause of material degradations, including stress corrosion cracking. The review will analyze the fouling mechanisms and behaviors, the source of impurities, corrosion mechanisms, and the factors that affect the deposition and corrosion behaviors.

KEYWORDS:

• Steam generator
• corrosion
• flow-accelerated corrosion
• deposition
• pressurized water reactor
• fouling

1. Introduction

A common operational problem that occurs during a flow of aqueous solutions in a heat exchanger is the corrosion of structural material and the accumulation and deposition of unwanted solid materials on the heat transfer surface, the process of which is called fouling. Three-phase, solid, liquid, and vapor can simultaneously present in the flow and induce fouling [1]. Commercial pressurized water reactors (PWRs) steam generators (SGs) experienced reliability problems within the first decade of operation associated with material degradation, including corrosion, tube fretting, and wear [2]. Caused by tube fouling, every year amount of cost is spent on maintaining, repairing, and inspecting SGs.

The advanced and reliable power production system is required for the long lasting and continuous availability of electricity supplies as well as the competitive price. To eliminate the damage caused by fouling, secondary circuit chemistry treatments have been developed, such as phosphate chemistry treatment and all volatile treatment¹(AVT) [3]. However, the treatments do not perfectly work as to induce quick and severe degradation caused by the denting phenomenon [4]. Fouling may keep producing in the heat transfer crevices after treatment.

The observations as well as the modeling results [5,6], both demonstrated that fouling potentially forms in the free span region between tube support plates (TSPs), within the sludge pile on top of the tube sheet and close to the crevices formed between the tubes and TSPs. The in situ deposits can inhibit heat transfer, leading to thermal–hydraulic instabilities and creating crevice regions, where corrosive species can be accumulated. Secondary circuit corrosion damage of SG tubes is primarily caused by impurities in boiling regions, where high concentrations of impurities occur in occluded regions of the SG formed by corrosion product deposits [7].

The fouling characterization of SGs has been well summarized in the Electric Power Research Institute (EPRI) report [8]; the prediction of deposit thickness and distribution on SGs have been well developed [5,9]; the analysis of the potential impact of fouling on SG integrity and performance has been conducted; and the means of mechanical and chemical to remove the deposits also have been developed [4]. Previous studies shed some light on the underlying fouling formation mechanisms; however, the corrosion and degradation behaviors in the secondary circuit have not been thoroughly recognized. The purpose of this work is to investigate and analyze the corrosion and deposit behaviors on the secondary circuit of SGs. Based on available knowledge, the study focuses on (1) determination of the source of impurities, (2) assessment of the corrosion mechanism and factors, and (3) fouling formation and analysis.

2. Determination of the source of impurities

2.1. Coolant loops

The reactor coolant flows from the reactor core to the SG. Inside of the SG, the primary coolant flows inside of the many tubes. The secondary coolant flows around the outside of the tubes, where it picks up heat from the primary coolant. When the secondary coolant absorbs sufficient heat, it starts to boil and steam. Accordingly, the primary circuit is a single-phase loop while the secondary circuit is a two-phase (steam–water) loop. The coolant circuit parameters are shown in Table 1. The typical void fraction of secondary coolant in the SG ranges from 0 to 1.0. The efficiency of removing moisture from the steam is high [10]. For all SG designs, the coolant is routed to the main turbine, the suction of the pumps and finally circulate back to pick up the heat.

Table 1. Parameters of typical PWR coolant loops.

Loop	Inlet temperature (°C)	Outlet temperature (°C)	Pressure (MPa)	Saturated temperature (°C)	Loop flow rate (kg/s)
Primary coolant loop	287.7	324	15.2	343.3	About 4360
Secondary circuit	227	285	6.9	285	7100 (WWER-440) or 5889 (WWER-1000)

2.2. Structural materials

A large variety of structural materials present in the secondary circuit of PWRs suffer corrosion:

• Carbon steels are cheap iron-base metals with less than 1% of alloying element present. These materials exhibit poor resistance to corrosion, but their forming, machining, and welding are superior.

• Low-alloy steels are iron-base metals containing a few percent of nickel, chromium, molybdenum, vanadium, etc. They usually achieve better hardenability but have a limited resistance to corrosion.

• Stainless steels (SSs) are iron-base alloys containing at least 12% chromium. They can be divided by austenitic, martensitic, ferritic, duplex, precipitation hardening, and super alloys.

• Nickel-based alloys have at least 15% chromium, with a high resistance to uniform and pitting corrosion, but they are susceptible to stress corrosion cracking (SCC), except for those with the highest range of chromium content (e.g. 30% chromium).

Materials and methods that are used to fabricate SG components significantly affect their susceptibility to corrosion. Initially, the heat exchanger tubing in most of PWRs SGs placed in-service in the western countries (except Germany) was made from nickel-based alloy 600.²The German SGs designed by Siemens/ Kraftwerk Union (KWU) use alloy 800M³tubing. Recently, most SGs designed by Westinghouse, Framatome, Siemens/ Framatome, Babcock & Wilcox, and Mitsubishi-Heavy Industries are being fabricated with thermally treated alloy 690.⁴Siemens/ KWU and Babcock & Wilcox Canada are also supplying replacement SGs with alloy 800M tubing.

The early PWR plants connected the mill-annealed tubing through hard rolling into the bottom of the tube sheet, which left an approximately 0.2 mm wide radial crevice (a narrow gap around the tube). The most recent procedure used by most PWRs and CANDUs⁵(CANada Deuterium Uranium) SG manufactures is to perform a hydraulic expansion over nearly the entire tube sheet thickness, followed by a one or two step mechanical hard roll near the top and the bottom. The earlier tube sheets used carbon steels, whereas the later models have corrosion resistant type 405 ferritic SS.

The SG shell, including the feedwater nozzle, is made of low-alloy ferritic steel, typically SA-533 Type A, Class 1 or 2 for the Westinghouse SG shells and SA-508 C12 for the feedwater nozzle forgings. Some of the earlier SGs made by Westinghouse in their Lester plant used SA-302 Grade B for the plate material, but all the SGs built at the Tampa plant used SA-533. The thermal sleeve inside the feedwater nozzle is made of SA-106 Grade B carbon steel.

2.3. Non-volatile impurities

Non-volatile impurities are of concern because of implications for impurities deposition. The secondary circuit includes amount sources of non-volatile impurities, mainly corrosion products (dissolved iron hydroxide, suspended iron oxides, and copper ions [8]) and non-metallic materials.

Non-volatile impurities released from the secondary circuit can be transported, dissolved, or deposited by the feedwater flow by solubility differences [11]. They react with each other and then change the pH and electrochemical potential (ECP) of the coolant, inducing more materials dissolved. Alloys susceptible to intergranular attack (IGA) and SCC, such as sensitized SS 304 and nickel-based alloys are significantly impacted by the impurities.

Except for the corrosion products released from the coolant loop, the impurities of non-metallic materials can be introduced during the normal operation. For example, the leak from condenser tube, the oils and greases for equipment maintenance and repair, and the organics that are not removed by the make-up water treatment.

Table 2 is a summary of sources of deposit precursors and displays the typical weight of the sources. The types of non-volatile impurities can be thoroughly summarized as [7,12]:

• Particulate impurities: magnetite, lepidocrocite, maghemite, hematite, silicon dioxide, apatites, copper, copper oxides, and lead.

• Ionic species: sodium, carbonates, chloride, potassium, sulfates, aluminates, copper hydroxides, iron hydroxides, iron ions, and copper ions.

• Solute species: oxygen, carbon dioxide, nitrogen, siliceous acid, organics, and organic acids.

Table 2. Sources of deposit precursors in the secondary circuit (summarized from [12]).

			Species
Source	Type of deposit precursor	per year per SG	Major species	Minor species
Feedwater, condensate, and drain systems	Corrosion products	100–250 kg 50–150 kg	Fe3O4, Fe(OH)2, Fe(OH)3, Cu(OH)^+, Cu, Ni	Lead, sulfur, carbon, phosphorous, silicon, arsenic
Makeup water system	Gases	2 µg/L	Oxygen, silica, hardness species, organics	Metals
	Soluble species	5 kg or more
Condenser	Gases	2 µg/L	Silica, hardness species, oxygen, organics	Metals, carbon dioxide
	Soluble species	Variable
	Insoluble species	Variable
Condensate polishers	Resin fines	>25 kg	Organics, resin particles, resin breakdown products, acids, caustic	Metals, carbon dioxide, iron, hardness species
and blowdown	Treatment chemicals	Variable
demineralizers	Absorbed impurities	Variable

3. Corrosion mechanism and factors

3.1. Fouling layers

The structure of fouling can be classified into three layers. The layer that is formed on the bulk material is called metal corrosion layer, or diffusion layer due to the diffusion of elements and the combination with oxygen atoms. Caused by the chemical reaction, the metal corrosion layer thins the bulk material and changes the mechanical and hydraulic property. The potential damages by the corrosion layer include the formation of crevices and pits, inducing SCC and changing the heat transfer rate and fluid friction. Above the metal corrosion layer, oxide layer is formed due to the electrochemical reactions of metal and solution. The oxide layer can be divided into the inner oxide layer and the outer oxide layer (or adherent oxide layer) by the formation mechanisms that depend on the SG materials. Distinction of the inner layer and metal corrosion layer is unclear, in most cases being treated as the same thing. The microstructure characterization shows that the outer layer is loose and spongy while the inner one is compact.

The third layer that exposes to the solution is called deposition layer due to the deposition of non-volatile impurities that come from the secondary circuit. The three layers interact with each other during formation, which makes it difficult to identify the boundaries.

3.2. Characterization of oxides

The elemental compositions of oxide layers depend on the compositions of the bulk material.

A duplex passive film of magnetite (Fe₃O₄) on carbon steels in a typical secondary circuit environment⁶has been observed, that composite of two layers: an outer layer with loosely packed crystals and an inner layer that was compact and protective [13]. The thicknesses of the outer layers and the inner layers are shown in Figure 1. Maghemite (γ − Fe₂O₃), magnetite (Fe₃O₄), and trace amounts of lepidocrocite (γ − FeOOH) were also detected in the oxide layer on the inside surface of the small-bore carbon steel pipes of the extraction steam system by Fourier transform infrared spectrophotometry [14].

Figure 1. Average thicknesses of oxide layers of carbon steel A106B in exposures under pH (25 °C) 9.5–9.6, 260 °C and 5.1 MPa (data are from [13]).

The inner oxide layer of alloy 800⁷is highly compact and enriched in Cr that is more than Cr content in the bulk alloy, while Ni content of which is less than the bulk. The outer layer is porous, consisting of Cr–Ni–Fe oxide crystals. The average thicknesses of oxide layers (with/without deposits) under a typical secondary circuit environment⁸after long-term exposure are shown in Table 3 [15]. The oxide layer tended to be 0.70 μm thick after a long time while the deposits kept increasing. Besides, the films exhibited the conventional double-layer structure, but after long exposures the inner layer was hidden by the formation of outer layers and/or by the deposition of species inevitably present in high-temperature coolant. The oxide compounds were assigned to be CrOOH, FeOOH and Ni(OH)₂ by X-ray photoelectron spectroscopy. In-depth layers near to the metal/oxide interface, ratios of each metal (Ni/Cr, Ni/Fe, Cr/Fe) evolve closer to those of the base metal. Ni/Cr was greater than 1 as to the double oxide layer, which was also found by the McIntyre et al. [16]. There was Ni enrichment in the outer oxide layer and, as in-depth layers near to the metal/oxide interface were analyzed, the ratios of each metal closer to those of the base metal [15]. The ratios of Ni/Fe and Cr/Fe increase in the order of inner layer < base metal < outer layer. Lower compositions of Ni and Cr in the inner layer as compared with the base metal indicate that higher ratios of Ni and Cr were diffused out than that of Fe. Higher compositions of Ni and Cr in the outer layer as compared with the base metal suggest that amount of Fe was diffused out.

Table 3. Average thicknesses of oxide layers of alloy 800 grown in exposures under typical secondary circuit environment (pH (25 °C) 10.2–10.8, 220 °C, hydrogen content 3–10 cm³/L D₂O) [15].

Temperature (°C)	Time (d)	Oxide layer thickness (μm)	Oxide layer with deposits (μm)
220	82	0.14	0.15
265	720	0.68	0.78
265	1290	0.70	1.06

Ratios of each metal in the oxide layers of SS 304⁹under a typical secondary circuit environment were detected to be different from alloy 800. The composition of the base alloy influences the composition of the oxide films and has a strong effect on the ratios of each metal. Table 4 summarized the ratios of each metal in the oxide layers of alloy 800 and SS 304. The corresponding ratios for the base material indicate the lower Ni content of SS 304 (∼ 7.5 at.%) as compared to that of alloy 800 (∼ 30 at.%). Unlike alloy 800, the ratios of Ni/Fe and Cr/Fe increased from inner to outer, that is base metal < inner layer < outer layer, suggesting Fe largely diffused out from base metal. The ratios of Ni/Cr in two layers were higher than the base metal, because the diffusion rate of Ni is much higher than Cr in SS 304.

Table 4. A summary of the metal:metal ratios of alloy 800 and SS 304.

Material	Location	Composition			Test conditions
Alloy 800 [15]	Outer layer	Ni/Cr = 2.2	Ni/Fe = 1.8	Cr/Fe = 0.8	pH (25 °C) = 10.2–10.8, hydrogen content 3–10 cm^3/L D2O, 770 hours, 220 °C.
	Inner layer	Ni/Cr = 1.4	Ni/Fe = 0.4	Cr/Fe = 0.3
	Base metal	Ni/Cr = 1.4	Ni/Fe = 0.67	Cr/Fe = 0.46
SS 304 [15]	Outer layer	Ni/Cr = 0.68	Ni/Fe = 0.27	Cr/Fe = 0.39	The same as above
	Inner layer	Ni/Cr = 0.68	Ni/Fe = 0.22	Cr/Fe = 0.32
	Base metal	Ni/Cr = 0.38	Ni/Fe = 0.10	Cr/Fe = 0.27

3.3. Flow-accelerated corrosion (carbon steels)

For PWRs, the most significant circuit issue appears to be flow-accelerated corrosion (FAC) of carbon steel piping [17–19]. Dissolution of iron that was caused by FAC has been observed in CANDU 6 plants [20].

FAC is an electrochemical corrosion process that flowing stream of water or a water–steam mixture diffuses the protective oxide layer formed on carbon or low-alloy steel. FAC and the corresponding mass transfer are very complicated processes, involving chemical reactions, mass diffusions, and flow convections. Carbon steel piping develops a protective magnetite oxide layer at the start of operation with the reaction, $$3\mathrm{Fe}+4{\mathrm{H}}_2\mathrm{O}\to \mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_4+4{\mathrm{H}}_2\uparrow .$$

Many investigations confirmed that the solubility of magnetite is very low (10⁻⁶–10⁻⁷ moles/L) if the water is pure [21]. While if the potential and pH are not kept within the stable area of magnetite, it will either dissolve to form Fe²⁺ at low pH and lower potentials or transform to hematite, Fe₂O₃, at high pH and higher potentials [21]. Under secondary water chemistry, the magnetite layer thus dissolves into the water stream or wet steam during the FAC process; and it is further oxidized to a non-protective hematite layer by the existing oxygen in the loop [22], via $$2\mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_4+\frac{1}{2}{\mathrm{O}}_2\to 3\mathrm{F}{\mathrm{e}}_2{\mathrm{O}}_3.$$

The chemical reaction may occur either in the bulk flow homogeneously or on the solid/liquid interface heterogeneously, or both [23]. The mass transfer is caused by the unsaturation of dissolving species at the interface between a dissolving oxide layer and a flowing fluid [24]. After the breakdown of the oxide layer, corrosion can keep occurring. As the oxide layer becomes thinner and less protective, the corrosion rate increases. Eventually a steady state is reached, where the layer formation and dissolution rates are equal.

The necessary conditions for FAC occurrence are designated as overlapping conditions of each parameter staying in the FAC sensitive range, as shown in Figure 2 [25]. In other words, FAC can be prevented by controlling each parameter so that it stays outside each sensitive range. FAC affects carbon steel components in de-aerated and alkalized flowing water [12], being high risk within a temperature range of 120 °C–180 °C. It has been shown that the FAC rate in carbon steel and low-alloy steels peaks at around 150 °C [26,27] at pH 7 [25]. Wall thinning rates are normally less than 0.25 mm/year, but rates up to 3 mm/year have also been observed in operating plants [26]. The carbon steel with 0.019% Cr was detected to be more FAC resistant than the carbon steel with 0.001% Cr, as the FAC rate was reduced by a factor of 2.4 [28]. No FAC-related failure/thinning in SS components have been reported as these are inherently not prone to FAC [29]. The use of a better material that contains higher Cr has been identified as helpful to increase the resistance to FAC [30,31]. SS and alloy steels are resistant to FAC, however, their thermal expansion rates of which is 1.4 times greater than carbon steel, and the chloride stress corrosion of which is another susceptible concern.

Figure 2. High FAC risk zone indicated by major parameters [25].

3.3.1. Flow rate effect

Flow rates may change due to changes in geometry (e.g. elbows/bends, etc.) or local flow disturbances (e.g. flow obstruction by flow measurement devices). Changes of local flow rate lead to localized changes in thinning rate [32]. Increased flow velocity leads to decreased boundary layer thickness, increasing the limiting current density of cathodic reactions. Increased flow velocity also decreases the protectiveness of inner layer, increasing the anodic current density. At a low fluid velocity, FAC rates are controlled by the mass transfer, while, at a high flow velocity, FAC rates are governed by the chemical reactions in the metal–oxide and oxide–coolant interfaces.

A few attempts have been conducted to correlate ECP with FAC rates. ECP is the interface potential between the immersed material and the solution, reflecting the kinetics of corrosion. At 140 °C in neutral, low oxygen water, ECP decreased with increase in the flow velocity and correspondingly FAC increased [19].

3.3.2. Temperature effect

Temperature is a significant factor affecting FAC of carbon steels and low-alloy steels. The temperature range of 120 °C–180 °C is a high risk zone to FAC. The maximum FAC rate for single phase (water) is at temperature 150 °C while 180 °C for two-phase [33,34]. The simulating results of the EPRI-CH and the KWR-KR FAC models between 40 °C and 260 °C were shown in Figure 3 [35]. Around 150 °C, the FAC rate reaches the maximum. Under the secondary water chemistry, FAC rate of above 200 °C is a priority concern, and it is less than 0.2 mm/year [35].

Figure 3. Comparison of the EPRI-CH and the KWR-KR FAC models for cold pH variations (data are from [35]).

The solubility of Ni and Fe largely depends on temperature, as well as pH and redox potential, which varies with location around the non-isothermal system [7]. With the gradient of temperature on the tube surface, the solubility of Ni and Fe is varied and correspondingly the distribution of corrosion layer is varied. At a temperature lower than 200 °C, the reaction of Fe and water products Fe(OH)₂, while at a temperature above 200 °C, Fe directly transfers to magnetite, Fe₃O₄. In the secondary circuit environment, the corrosion product of Fe-contained alloy is Fe₃O₄, following the reaction: $$3\mathrm{Fe}+4{\mathrm{H}}_2\mathrm{O}\to \mathrm{F}{\mathrm{e}}_3{\mathrm{O}}_4+4{\mathrm{H}}_2\left(\mathrm{above}200{}^{\circ }\mathrm{C}\right).$$

3.3.3. pH effect

Water chemistry plays a critical role on FAC. The general corrosion in a SG proceeds via the reactions: $$\mathrm{F}{\mathrm{e}}^0\to \mathrm{F}{\mathrm{e}}^{2+}+2{\mathrm{e}}^{-}\left(\mathrm{anodic}\right),$$$$\begin{array}{ccc}& & \frac{1}{2}{\mathrm{O}}_2+{\mathrm{H}}_2\mathrm{O}+2{\mathrm{e}}^{-}\to 2\mathrm{O}{\mathrm{H}}^{-};2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}\to {\mathrm{H}}_2\hfill \\ & & \left(\mathrm{cathodic}\right).\hfill \end{array}$$

Accordingly, maintenance of an alkaline pH and reduction of dissolved oxygen is helpful in controlling corrosion of base metal [36]. It is generally accepted that certain amount of dissolved oxygen (e.g. 1 μg/L [25,37], or 40 μg/L [38]) is favorable to control FAC rate at an extremely low level.

Through the analysis of complex capacitance data that was obtained from the interfacial electrochemistry study, it has been observed that the double-layer capacitance of the metal in pH 10 solution was dramatically reduced with increasing particle concentration, especially when compared with the acidic and near-neutral solution, which represented that amount of particles attached to the metal (surface blockage by particles); the attachment can reduce the active surface of the metal [39].

The range of 9.3–9.6 has been proved effective in reducing long-range corrosion product transport on all-ferrous secondary circuits [12]. Even the pH larger than 9.8 is helpful in minimizing the corrosion products as the plants that have used AVT (addition of ammonia and hydrazine) with the high pH of >9.8 from the beginning did not show a significant decrease of the heat transfer in 15 years operation [3]. Fe concentration values even below 1 μg/L are achievable [40]. It has been found that an addition of ethanolamine (ETA) to ammonia was more efficient than ammonia alone, and the mixed treatment can reduce 50% of iron at pH 9.5 or higher [41].

However, if the secondary circuit components (e.g. turbine) are copper-contained, the dissolution of copper would increase once the pH was above 10, and the circuit prone to corrode in the presence of ammonia. To adequate protect carbon steels and copper alloys, the recommended pH range is 8.5–9.5 [42], some literature suggested to be below 9.3 [8]. Since German utilities replaced the copper-contained turbine with SS or titanium turbine and improved the pH to above 9.8, no damage with safety relevance occurred in German nuclear power plants due to FAC [43].

Figure 4 shows the iron solubility at different temperatures between pH 8.75 and pH 9.60. The temperature of the secondary circuit ranges from 227 °C to 285 °C; the marked area indicates the possible solubility of iron in the coolant at different areas with fluid flow. Above temperature 150 °C, the solubility increases with pH decreasing and temperature decreasing.

Figure 4. Iron solubility at different pH/NH₃ and temperature (data are from [26]).

Calculated and measured results for the pH dependence of wall thinning rates are shown in Figure 5. The calculation was based on the coupled models of a static electrochemistry model and a dynamic double oxide layer model, developed by Uchida et al. [25]. The measured results were provided by Heitmann and Schub [44], and Tsuruta et al. [45]. The results are in fair agreement as well as with the results in Figure 3 when comparing the pH at 8.7 and 9.4.

Figure 5. Results of wall thinning calculation depend on pH [25].

Besides FAC, corrosion of other types, including IGA/SCC and pitting are strongly affected by the local pH. High pH (alkaline conditions) and low pH (acidic conditions) accelerate the rates of IGA/SCC.

3.3.4. Impurity effects (lead and oxygen)

The presence of lead at low coolant concentrations promotes SCC of nickel-based alloys. Lead absorbs on oxides and, therefore, exposes metal surface. Up to 20% concentrated lead has been detected at the crack tips of SG tubes. The concentration of lead in feedwater is less than 10^{− 5} mg/L while may accumulate to 2 × 10^{− 4} mg/L after recirculating. The mechanism by which it promotes SCC has not been fully understood, its potency at such low concentrations being difficult to counteract [46].

Under PWR feedwater chemistry conditions, certain amount of dissolved oxygen is favorable to control FAC rate at an extremely low level [25,37,38], and concentrations of oxygen 1–2 mg/L were required to stifle the FAC of small-diameter tubes corroding at the rate of a fraction of 1 mm/year [47]. Calculated and measured results for oxygen concentration dependence of wall thinning rates under high pH conditions (pH = 9.2) are shown in Figure 6. As increasing oxygen concentration, a sudden drop in corrosion rate, and a sudden increase in ECP have been observed.

Figure 6. Results of wall thinning calculation dependent on oxygen concentration [25].

Oxygen injected into steam and moisture separator reheaters is used for some plants to suppress FAC. Oxygen concentrations in the mg/L range are sufficient for the very high mass transfer rates encountered in the systems to change the nature of the oxide layer from magnetite (Fe₃O₄) to more protective hematite (Fe₂O₃), and therefore reduce the dissolution rate. However, relatively high concentrations of hydrazine must be injected to remove dissolved oxygen just before the SG feedwater inlet to protect SGs from different forms of corrosion.

3.4. Corrosion behavior of the moisture steam areas

The fundamental recognition on the mechanism of steam-induced corrosion has been consistent by the previous studies [48–50]. Dual-phase FAC is an important concern of SG corrosion. The secondary coolant is subcooled liquid at the inlet of cold leg, and it starts to be heated to two-phase liquid with void fraction increasing from 0.1 to 0.9 (at the top of SG tubes). The liquid boils to steam at the upper part of SG vessel. Observations found that some specific locations in the pipeline suffer from corrosion problems while other sites do not.

The oxide dissolution mechanism at the moisture steam areas is similar to single phase FAC. The FAC rates are varied in void fractions. In the moisturized steam area containing more than 5% of moisture, the corrosion rate of carbon steel is fastest. This is affected by factors like the size of water drops, liquid film thickness, diffusion rates, and locally dissolved iron concentrations [26].

The liquid phase in a steam line flows in a thin layer near the wall while the vapor forms in the core of the flow and moves much faster than the liquid phase. The velocity difference creates shear forces at the liquid–vapor interface. If this force is greater than the surface tension force at the interface, some liquid will be sheared off the liquid layer and carrier over with the vapor. This liquid will form droplets and will be entrapped and accelerated in the vapor core. A fraction of this liquid will impinge on the oxide film on the inside surface and can crack the oxide film thus exposing under layers to a corrosive attack of the coolant.

4. Fouling formation and analysis

4.1. Process and mechanism

Approximately 90% of the deposits on the tube surfaces and free span regions come from the corrosion products and other impurities existing in the secondary circuit [51]. The corrosion products and other impurities can be removed by blowdown, carried by steam, removed by chemical or mechanical means or form permanent deposits once transported to SG [8]. The representative processes of the scale formation include incubation, initiation, growth, growth-limiting state, spalling, and re-deposition [8]. The process is largely determined by heat flux, species concentration, time, temperature, fluid velocity, surface chemistry and morphology, and pressure [24]. The accumulated corrosion products showed widely dissimilar constituents concerning composition, thickness, and porosity at the upper and lower part of the crevice as well as inside the crevice area [52]. The preferential region for relatively thick deposit is the hot-side, high-void or low-velocity zones on the tube surface, which was suggested in a three-dimensional thermal hydraulics particulate deposition model [5].

Dissolved chemistries and suspended particles deposit on the surface where boiling and evaporation occur. Dissolved chemistries, such as mineral salts which have inverse solubility behavior, become precipitations and deposit on the surface with boiling solution, the process of which being crystalline fouling [1]. Suspended particles carried by the coolant flow also deposit on SG tube surfaces or TSPs due to change in operational parameters, which is particulate fouling. Bubble generation as well as the increase of fluid velocity promotes particle deposition in crystallization fouling due to increased super-saturation in the vicinity of heat transfer surface while it can hinder particulate fouling due to increased pumping power and turbulences near the heat transfer surface [1].

Fluid mechanics can explain how a particle gets to a wall and estimate the deposit characterization; however, it cannot estimate if the particle can adhere to the wall [53]. Visser [54] suggested that the deposition process can be understood from the surface forces between a colloidal particle and a wall: the forces include London-van der Waals force and electrical double-layer interaction force. London-van der Waals force arises from the interaction of fluctuating dipole moments generated by the motion of electrons around the nuclei of neutral atoms in proximity to each other. In a single-phase fluid, the resulting forces between particles and between a particle and a surface are always attractive. Electrical double-layer interaction arises from the electrical charges commonly acquired by particles or surfaces immersed in an electrolyte, and the compensating diffuse layer of counter-ions in the liquid adjacent to these surfaces. If the particles and the wall have opposite charges, these forces are attractive; otherwise, the forces are repulsive.

4.2. Concentration of impurities at the packed crevice

To the location of SG tube support at the packed crevice, the temperature gradient across the SG tube coupled with a restricted flow rate can result in nucleate boiling. Water is continuously evaporated by this mechanism. A steam-blanketed region exists because of the nucleate boiling region. All non-volatile compounds enter the crevice with water flow and deposit on the region because the water is boiled to steam. On the other hand, as the concentration of impurity increases, the boiling temperature rises. When the boiling temperature matches the primary tube temperature, the heat flux between the tube and the fluid equals zero. As a result, the concentration of impurities on this region reaches a maximum concentration after the boiling terminating. The concentrations of these impurities at the packed crevice are expected to be few μg/L [4].

4.3. Deposition of various species

4.3.1. Iron oxides

The investigation of nearly all PWRs shows that the deposits are mainly composite of iron oxides. The study of the fouling behavior of the long-term operating SGs shows that fouling is correlated to the amount of iron ingress (feedwater/steam side) which accumulates in the SG and adheres to the tube surface [3]. The iron content of fouling is a symbolic reference on detecting the total mass of fouling [3]. The fouling compositions on carbon steel include magnetite, hematite, ferrous hydroxides, and hydrated iron species [12]. Some studies suggested that the average deposition rate of iron oxide during the initial deposition period, over a pressure range of 0.1–18 MPa, was proportional to the square of the heat flux and directly proportional to the iron concentration [55]. While others suggested that the initial deposition rate was directly proportional to the heat flux and iron concentration [56,57]. The deposition rate in the steady-state deposition period of both in-reactor and out-of-reactor boiling water loops was shown to be dominated by the heat flux and the iron concentration [58] and increases when pressure increasing [59].

Significant radial variations have been observed in the deposits composition across the crevice from the tube to the TSP [53]. Fe oxides were the primary composition of all the deposits on TSP while other elements were precipitated at the tube surface where the temperature is maximum, likely starting the precipitations by the most insoluble species, followed by the low solubility species and finally by the highly soluble constituents [52].

4.3.2. Nickle oxides

Under Ni ion concentrated coolant with Fe ions, the Ni deposit of SS surface has been determined to be NiO, Ni(OH)₂, and NiFe₂O₄ [60–62]. Ni ion deposition rate has been found to be influenced by Ni ion concentration and heat flux, displaying linear relationship (Figure 7). The study also found that Ni ion deposition rate was independent of iron oxide concentration when taking 1.0 mg/L Ni ion and 0–5 mg/L iron oxide under 280 °C, 7 MPa (boiling condition) [63].

Figure 7. Dependence of amount of Ni ion deposits (a) on their ion concentration at 7 MPa, temp. 280 °C, heat flux: 46 W/cm², time: 8 hours; (b) on heat flux of heated rod at 7 MPa, temp. 280 °C, Ni ion: 1.5 ppm, time: 8 hours (data are from [63]).

4.3.3. Sulfate ions

Sulfate ions are involved in the deposition as the corrosion of loops generates colloidal particles, due to the solubility limits of corrosion products or detachment of small pieces of material [64–66]. Previous studies found that the deposition on the SG tubes was mainly composed of magnetite and sulfate ions, especially in flow restricted areas they were likely to promote corrosion [67,68].

4.3.4. Copper

Because the Cu-contained alloys were used in the condensers and heat exchangers in some PWRs, they became a source of copper that has been typically found in the deposits [8]. Copper was found in the deposits of the secondary coolant of Ringhals 3, which contained no copper [52].

4.3.5. Titanium compounds

Some studies suggested that Ti compounds were only effective for open crevices but not in packed crevices, where Ti was not able to penetrate. Ti compounds arrested intergranular stress corrosion cracking (IGSCC) of alloy 600 in concentrated alkaline solutions [69]; however, the inhibitor effect has not been confirmed due to the difficulty of Ti compounds to penetrate into packed crevices and to reach the tube surface [70,71]. Probably it is because the solubility of Ti oxides was very small at the secondary circuit, which led to the compounds undelivered to the crevices. The examinations of the retired Ringhal 3 SG TSP found Ti was detected at the free span areas but not inside the crevice, suggesting Ti was not contained, dissolved or suspended in the water that enters in the crevice [52].

4.3.6. Silica compounds

Silica compounds were suggested to be one of the main factors of corrosion and that most of the IGA/SCC damage occurred under aluminosilicate deposits [72]. However, no clear relationship was found to relate IGA/IGSCC with silica compounds [52]. Other studies found that some plants with higher silica blowdown concentrations have experienced less outside diameter stress corrosion cracking [73]. The PWR plant Ringhals 4 has less IGSCC in the TSP regions than Ringhals 3, but has higher silica compounds in these areas [74]. Accordingly silica compounds may be necessary but not sufficient to produce IGA/SCC.

4.4. Decline of deposition rate

When the fine particles as well as the precipitated compounds fill the roughness cavities of a surface, deposition would then be affected. The declining rate of particle accumulation with time has been frequently observed in fouling [75]. The interfacial electrochemistry study revealed that the double-layer capacitance of the material surface significantly affect the particle deposition rate [39]. A progressive change in wall zeta potential is due to deposition of dielectric particles with zeta potentials of opposite charge to the point where originally attractive electrical double layers become repulsive [27]. The interaction forces largely retard the deposition rate when the material surface and particles are similarly charged. When a stable suspension of relatively homogeneous particles is involved in the fouling, they restrict the rest particles to accommodate on the material surface [76]. As the particle layers build up and deposit thickens, the absorption of oppositely charged corrosion ions is reduced. The suppression of attachment is also increased by the increase of the deposit-scouring velocity due to channel blockage [27].

5. Conclusion

The secondary circuit of PWRs composite of a large variety of structural materials, including carbon steels, low-alloy steels, SSs, and nickel-based alloys. Carbon steels suffer from FAC due to the low alloying element contents. FAC is the primary corrosion problem which provides the main non-volatile corrosion products that lead to the formation of fouling on the SG. The non-volatile deposit process will vary depending on the nature of the entrained material, which is influenced by the secondary circuit conditions and the physical and chemical characteristics of the flow. The packed crevices of SGs have the most concentrations of impurities due to boiling and geometry effects. Commonly, the fouling layer is composed of a corrosion layer (or diffusion layer), an oxidation layer, and a deposition layer. The deposits are mainly composite of iron oxides, the rate of which is closely related to the heat flux, iron concentration, and pressure. Ni ion deposition rate has been found not affected by iron oxides concentrations. Other species, sulfate ions, copper, titanium compounds, and silica compounds also have been observed as fouling precipitations. The fouling thickness suppresses the deposition rate. The deposition rate decreases with time as the formerly attractive electrical double layers of the surface become repulsive when particles accumulated.

Based on this study, it can be concluded that the fouling processes including the mechanisms are now fairly well understood. However, significant research activities are still needed which include the kinetics of vapor phase reactions and deposition, the geometry effects, and fouling effects on corrosion especially on the SCC, and predictive model development to predict the secondary circuit corrosion and SG tubes deposition based on modeling of fluid mechanics, analyzing of the coolant environment, and observation of the long-term operation. Specifically, the following opinions for advanced water chemistry are suggested for future studies: (1) the replacement of FAC-resistant structural materials to carbon steels and low-alloy steels; (2) the precise controlling of dissolved oxygen concentration at a certain level to increase ECP and eliminate the occurrence of FAC; (3) the alkaline solution at high pH; (4) high ECP level of structural materials; (5) advanced fabrication technology to eliminate the gaps between heat exchanger tubing and TSPs; and (6) development of the efficiency water chemistry adjustment and impurities removal during the normal operation.

Acknowledgments

This research has been performed using funding received from US Nuclear Regulatory Commission [grant number NRC-HQ-11-G-38-0036].

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. AVT typically involves using ammonia (NH₃) to raise pH and hydrazine (N₂H₄) to minimize oxygen.

2. Alloy 600 wt.%: 76 Ni, 15.5 Cr, 8 Fe, C < 0.15.

3. Alloy 800 wt.%: 32.5 Ni, 21 Cr, 46 Fe.

4. Alloy 690 wt.%: 61 Ni, 29.5 Cr, 9 Fe, C < 0.025.

5. Although the CANDU pressurized heavy water reactor design differs from that of PWRs, many materials degradation issues are similar.

6. pH (25 °C) 9.5–9.6, 260 °C and 5.1 MPa.

7. Wt.%: 20.5 Cr, 32.5 Ni, 0.75 Mn, 0.44 Ti, 0.022 C, 0.01 Co, 0.07 Cu, balance Fe.

8. pH (25 °C) 10.2–10.8, hydrogen content 3–10 cm³/L D₂O.

9. Wt.%: 18.5 Cr, 8.2 Ni, 1.8 Mn, 0.003 Ti, 0.025 C, 0.12 Co, 0.62 Cu, balance Fe.