Flow accelerated corrosion

Flow accelerated corrosion (FAC) also known as erosion-corrosion (EC) is a degradation mechanism that affects carbon steel (CS) piping carrying single phase, subcooled feedwater and steam lines carrying wet steam. The damage caused by erosion-corrosion is higher than damage attributed to erosion or corrosion alone. Carbon steel feedwater piping corrodes during normal operation, forming a thin layer of iron oxide, mostly magnetite (Fe3O4), on the inside surface. This layer protects the underlying piping material from the corrosive environment, and in the absence of erosion, limits the corrosion rate. However, if stressors causing erosion are present, the layer of iron oxide will dissolve and the uncorroded metal surface will again be exposed to the corrosive environment, and piping corrosion will continue. Thus, the continuous process of oxide growth and dissolution leads to thinning of the pipe wall and ultimately to a catastrophic failure, when the pipe is subject to a pressure pulse of large magnitude.

FAC occurs under high flow rates, if the protective oxide layers on the surface of CS components and pipes cannot be built. The protective oxide layers are built on the CS surface by the reaction of iron ions, which are dissolved from the metal, with water at high temperatures. If these protective layers dissolve in an iron unsaturated fluid medium at the metal-fluid interface, or if iron ions released from the CS surfaces are immediately removed by a high flow, the protective layers cannot be built, and this results in the FAC degradation of CSs.

Figure 4.60 presents a simple model describing the phenomena occurring during erosion-corrosion [72]. The FAC mechanism, their stressors and parameters having influence on the development of FAC are well described in [31–33].

FAC will depend on the nature of the protective layer, which is dependent on the base material composition, the surrounding chemistry conditions and temperature. It has to be well understood that FAC is a two phase process: metal and water. In the case of steam flow, it can produce FAC following the above described mechanism only if a water film forms and flows on the metal surface (annular flow). The high steam velocities of typically 30 m/s ‘push’ the water film, increasing the velocity of the streaming water film on the metal surface. This is represented in Figure 4.61. Although the water film has a lower velocity than steam, it results higher than the usual linear velocities at pipes with single phase (liquid) flow. In the single phase-system a characteristic scalloped or orange skin shape surface appear whereas in two phase system a very thin oxide layer and the ‘tiger striped’ surface (Figure 4.62) are observed.

Figure 4.60. Phenomena occurring during erosion-corrosion [72].

Figure 4.61. Flow velocity profiles in single phase and two phase FAC [31].

Figure 4.62. Flow accelerated corrosion: pipe surface appearance.

Local acidic, oxidative conditions

Carbon steel corrosion Denting

The factors affecting the erosion-corrosion rate are then the following:

 Piping and/or component material.

 Bulk water flow velocity.

 Piping configuration, geometry related flow perturbations (local flow velocity, turbulence).

 Feedwater and/or condensate temperature.

 pH value.

 Oxygen content.

 Moisture content (two phase FAC).

Carbon steel components with less than 0.1 weight per cent Cr are susceptible to erosion-corrosion damage being the Cr-Mo alloys with 2–2.5% Cr highly resistant. The characteristics of these alloys require careful welding procedures and extensive radiographic examinations, what makes the use of these alloys more expensive.

By replacing carbon steel pipes and components with low allowed steel the flow accelerated corrosion (FAC) will be reduced significantly (Figure 4.63). See also Figure 4.48 of Section 4.1.7. Stainless steel is not susceptible at all.

The influence of pH and temperature is of crucial importance and has been also discussed in detail in Section 4.1.5. (See Figure 4.64) updated from [30–32].

Figure 4.63. Parameters influencing FAC: Material type (left side) and flow rate (right side) [31,32].

Figure 4.64. FAC rate as a function of temperature (left), Ammonia concentration and pH required to suppress FAC (right).

For a pH_25°C > 9.5 in contact with the metal, using ammonia as alkalizing reagent, a drastic FAC reduction is expected. In fact, the pH at the operating temperature is important. As established in Section 4.1.4, the corresponding pH190°C for steam systems results in > 6.5.

The combined effect of temperature and material is shown in Figure 4.65. This temperature range of 100–200°C with a maximum close to 180°C automatically defines the areas of the steam water cycle with the highest susceptibility, where the material selection must be carefully considered.

Figure 4.65. Parameters influencing FAC: Combined effect of temperature and material [31, 32].

Specific metal loss rate

Temperature

Oxygen contributes to form a very stable hematite (Fe2O3) as oxide layer is generated on the carbon steel surfaces. Being hematite less soluble than hematite, it provides a barrier against FAC, as described in Section 4.1.4 and clearly shown in Figure 4.23.

At the beginning of 2000s, partial oxygen injections were introduced in several PWRs, to counteract local FAC problems occurring under extremely high flow conditions. It is described in detail in Section 5.3.2.1.

Inadequate geometries (e.g. low radius elbows, etc.) associated to high linear velocity of the fluid in contact with unalloyed carbon steel will result in enhanced FAC.

The influence of the design could be well described by Keller [73, 74], defining a dimensional geometry factors for each configuration of piping. The geometry factors acc. to Keller (Figure 4.66) permit to assess the relative influence of the geometry on FAC rate. The low the factor value is the more slightly is the wall thinning. For example, by choice of enough large bend radius or by using y-pipe instead of t-y-pipe or such like for the purposes of better flow quality, the influence of these design factors on corrosion rate could be reduced

Whereas a straight pipe has a factor of kf~0.04, a low radius elbow (kf=0.5) will be subjected to a FAC rate of about 10 times higher, being the absolute values a function of the other relevant parameters.

Figure 4.66. Geometry factors according to Keller for assessment its influence on FAC rate [73, 74].

Since downstream a flow perturbation (valve, elbow, etc.) there is some distance where the flow has not yet recovered the normal flow profile, the areas close to the perturbation are still influenced, and if two components are too close to each other that located downstream will have an enhanced metal loss rate.

These complex interdependences between material, chemistry and hydrodynamics have been successfully modelled in the form of computer codes allowing perform estimations and predictions leading to a safer and more economical operation, like COMSY [75] or CHECKWORKS.

k_c

stagnation points of secondary flowstagnation points of primary flow

flow

The FAC rate can be predicted by calculations using codes like the WHATEC code (today COMSY), as shown in Figure 4.67 and Figure 4.68 [29, 33]. The calculated predictions fit very well with the field experience. Such programmes are integrated in the plant surveillance programmes, enabling the identification of susceptible areas an orienting/rationalizing the UT inspection programmes.

Single phase FAC at the feed water line caused various severe accidents [36] and has been extensively studied over the last thirty years. The first fatal accident in western PWRs due FAC occurred on 9 December 1986 at an 18 inch (approx. 460 mm) elbow immediately downstream of a T-fitting in the condensate system at Surry Unit 2 (see Figure 4.69). The flow accelerated corrosion attack caused a local reduction of the original wall thickness from 12.5 mm down to 1.5 mm. A plant transient caused a pressure increase inside the pipe, initiating the respective pipe rupture. Due to the sudden occurrence of the break, the persons working in the turbine hall had no opportunity to escape. Therefore four workmen were killed and several more were seriously injured [13].

Figure 4.67. Calculated and measured FAC rate at a horizontal pipe downstream a bend.

Figure 4.68. Calculated FAC rate at pipe section having different low perturbations.

Figure 4.69. Ruptured elbow at Unit 2 of NPP Surry [16].

Millstone Unit 3 is a PWR at what was originally a three unit station located on the Long Island Sound in Connecticut, USA. In December 1990, two 6 inch (approx. 170 mm) lines downstream of level control valves in the moisture separator drain system failed catastrophically.

Post-accident investigation revealed that while the lines that failed should have been included in the single phase FAC programme, a miscommunication between the analyst at the corporate office and the engineer at the plant caused these lines to be omitted from the analysis and inspection efforts.

Figure 4.70. Damaged Pipe at Unit 3 of NPP Millstone [16].

On 9 August 2004, a large pipe break in a secondary side line at the NPP Mihama Unit 3 nuclear power station [29] killed five workers. The incident was caused by a degraded pipe in the condensate system, located in the turbine hall. The respective 22” pipe was suffering from flow corrosion attack. The ruptured pipe was located behind an orifice plate. The flow accelerated-corrosion attack caused a local reduction of the original wall thickness from 10 mm down to 1.4 mm.

This local wall thickness reduction caused the respective pipe rupture, which occurred while workers were preparing for a routine outage.

Figure 4.71. Pipe rupture at Unit 3 of NPP Mihama. (The orifice flange is barely visible on the right hand side of the right picture).

Figure 4.72. Sketch of ruptured pipe at Unit 3 of NPP Mihama.

At this time, most parametric influences are reasonably well understood. A code like COMSY could predict the events at Surry and Mihama with due anticipation, as shown in Figure 4.73 [29, 76].

The chart at Figure 4.73 indicates the predicted progress of wall thinning versus the operating time of the plant. Line No. 3-4 marks the minimum required wall thickness of the pipe for the given stress conditions. The red line indicated the wall thinning rate computed by the FAC code whereas the dashed line marks the wall thinning rate which was actually experienced. The lifetime prediction chart indicates the performance of a wall thickness inspection for the year 1991 and predicts the pipe rupture for the year 1999.

Figure 4.74 illustrates a lifetime prediction chart generated by the FAC code for the Surry 2 pipe break location. The red line indicated the wall thinning rate computed by the FAC code. The computed rate correlates well with the wall thinning rate actually experienced. The predictive result indicates the performance of a wall thickness inspection for the year 1981 i.e. five years before the occurrence of the catastrophic pipe break.

Figure 4.73. Lifetime prediction chart Mihama 3 [29, 76].

Figure 4.74. Lifetime prediction chart Surry 2 [29, 76].

Although erosion-corrosion is a greater concern in PWR feedwater piping, steam generator components have also experienced direct damage from this mechanism. Erosion-corrosion of the thermal sleeve at Diablo Canyon Unit 1 was recently reported [77]. The carbon steel J tubes and feedrings within RSGs have also experienced significant erosion-corrosion induced wall thinning. The affected J tubes have been repaired or replaced with Alloy 600 J tubes.

Erosion-corrosion damage has been reported to some of the carbon steel primary side divider plates in the CANDU steam generators, as well as fatigue damage to the carbon steel divider plate bolts. (The primary side divider plate is located below the tube sheet in the lower plenum of the RSGs.) The erosion-corrosion of the plate and fatigue of the bolts caused increased divider plate leakage and excessive bypass flow, which decreased somewhat the performance of the steam generators. Fatigue of the bolts may also lead to loose parts damage to the tube sheet.

Besides direct SG damage, FAC is an important indirect contributor to SG tube corrosion, as explained in Section 4.1.5, increasing the inventory of corrosion products and creating the adverse local conditions leading to damage tube outer diameter.