Introduction

Preheater units are regularly used in many industries. Several failure studies have been carried out for piping alloy materials used in applications such as (thermal/nuclear) power plants, preheater tubes of heat recovery boilers, petrochemical plants, superheaters [1,2,3,4,5,6,7,8,9,10,11,12].

Corrosion degradation (material loss) in metal alloys results from several factors, depending on the interaction with the environment to which the alloys are exposed. The formation of scales/layers and/or precipitates/products of oxides, hydroxides, sulfates/sulfides as well as thickness reduction in the internal and/or external surface(s) of the materials is, among others, the common indications of corrosion, where electrochemical cells are established.

This paper reports the corrosion failure analysis of A214 carbon steel tubes [13] in a reformed gas boiler feed water preheater, which operates between the high and low temperature shift converters in the ammonia process of a petrochemical plant. Material characterization and morphology analysis, in the form of optical emission spectroscopy (OES), high-resolution optical microscopy, and energy-dispersive x-ray spectroscopy (EDXS), were employed to verify that pitting corrosion was the incipient failure mechanism inside the tubes.

Experimental

The reformed gas boiler feed water preheater unit is a shell and tube heat exchanger in the ammonia process of a plant located at the Point Lisas Industrial Estate, Trinidad and Tobago. It processes on the shell side a (hot) reformed gases stream exiting the high temperature shift conversion unit, with a (cold) boiler feed water stream in the tube side (water side). The (cold) reformed gas stream leaving the preheater is then sent to the low temperature shift converter. Figure 1a and b shows the unit after being taken from the process for inspection and at its location in the plant before being removed for inspection, respectively. The flow directions of the streams and a view of the design arrangement of the carbon steel A214 and SS304L U-tubes inside the preheater are shown in Fig. 2. The design rationale for the use of the SS304L tubes was to minimize the effect of corrosion due to the condensation of selected reformed gas stream components. Moreover, it was indicated that there were no deviations in flow, pressures and temperatures from the standard operating procedures for the boiler. And, that the feed water treatment was adequately and continuously fulfilled during the regular operation of the boiler.

Fig. 1
figure 1

(a) A panoramic view of the reformed gas boiler feed water preheater unit. Indicated are the bundles for the respective carbon steel A214 and SS304L U-tubes. (b) The preheater unit at its location in the process. It operates vertically. The arrows show the respective flow patterns of the streams

Full size image
Fig. 2
figure 2

Schematic side view illustrations of the hot and cold flows and tube arrangement in the unit. The carbon steel tubes are designed to be located at the “hot” front of the shell side stream (after an impingement plate), while SS tubes are placed at the “cold” end of the shell side stream

Full size image

It was reported that after a preheater unit failure in March 2014, which caused a downtime of about 15 to 16 days, the unit was removed from its location for inspection and to determine the source of the problem. A leaking carbon steel tube triggered the failure of the unit. The failure was observed following water fill and 15 psig pneumatic air tests on the U-tubes, from which a carbon steel tube was identified as having a leak. No leaks were noticed on the stainless steel tubes. Figure 3 illustrates the exact location on the (carbon steel) tube that experienced the leak (row 7, column 20 referred to as Tube 1 sample), as well as the location of another carbon steel tube (row 20, column 19 referred to as Tube 2 sample) which presented a reduction in its outer surface (OD) of 25–29%, as determined by remote field electromagnetic testing (RFET).

Fig. 3
figure 3

Schematic front view of the tubes bundles and shell side flow arrangements in the unit. The numbering, row location and characteristics (carbon steel and SS) of the tubes are indicated. Also shown are the leaking tube (7–20) and a tube (20–19) which experienced a 25–29% OD wall loss

Full size image

A section of that leaking carbon steel tube (Tube 1 sample) was cut and then taken for visual examination, spectroscopic and morphologic analyses. The OD of the tube, at an area away from the failure, was measured (micrometer) to be 0.740 in ± 0.001 in, with a wall thickness of about 0.083 in ± 0.001 in. Figure 4a shows the section of the failed tube where a hole (measuring about 3 mm horizontally and 5 mm vertically) was distinctively formed, causing the leak in the tube. That section of the tube was sequentially cut transversally 2 in from both sides of the hole, and then, the new section was longitudinally cut half of the diameter to examine its internals, as shown in Fig. 4b. Some layers [measuring about 0.024 in ± 0.001 in (0.608 m)] of several colors, including light brown/orange/black scales, were observed on the surface of the pit. The light brown/orange scales were soft and were rapidly separated from the surface, while the black scale was brittle, and chips of it were determined to be magnetic, indicating the presence of magnetite (Fe3O4). Samples from that area on top of the pit were taken and properly prepared (cutting, mounting, grinding, polishing, and etching) for spectroscopic and morphology analyses, in accordance with the appropriate ASTM standard [14].

Fig. 4
figure 4

(a) Top view of the external section of the (finned) failed tube where a hole has been formed. Oxides of iron (light brown/orange) scales are spread all over the surface; (b) close-up view of the internal surface of the failed tube, where the hole has been formed inside a pit. The arrow on the left side of the pit indicates not only iron oxides presence but the action of erosion from the flowing boiler feed water (erosion can be also seen on the right side). The white line on top of the pit denotes where samples for spectroscopic and morphology analyses were taken. In addition, some black scales/regions can be seen around the surface of the pit, indicating the presence of magnetite (Fe3O4)

Full size image

Optical emission spectroscopy was performed using a Bruker Q2 ION Spark Spectrometer. The microscopic studies for the morphological and structural analysis were carried out with a Motic BA 310 optical microscope. Additionally, energy-dispersive x-ray spectroscopy (EDXS) analysis was conducted with a microprobe capable of detecting elements greater than boron by emitting an electron beam to excite the atoms. The average values for the relevant techniques are reported.

Results and Discussion

Table 1 contains the elemental composition based on OES analysis (except for the C and S contents, which were determined by the combustion method), as well as the composition of the carbon steel A214 according to the ASTM standard [13]. The samples from the failed tube conformed to the material specifications of the A214 standard.

Table 1 Elemental composition (wt.%) of the failed tube and from the ASTM A214 standard [13]
Full size table

Figure 5 depicts an EDXS spectrum of a sample from the failed tube (pit area), including the results of the composition of the main elements. The presence of light brown/oranges scales of iron oxides is evident throughout the inner and outer surfaces (ID and OD) of the failed tube, as shown in Fig. 4a and b. The findings from the EDXS analysis corroborate the iron oxides presence, from which the 16.7 wt.% of oxygen (not originally present) found in the area of the hole, pit and surroundings, and the reduction in the Fe composition to about 79.30 wt.% (from the original of at least Fe 99.0 wt.%, a significant decrease of about 20.0 wt.%) clearly indicates the formation of FexOy’s (oxides of iron). Moreover, these results also confirm that the black scale seen at the (ID) area on top of the pit (Fig. 4b) is indicative of the Fe3O4 (magnetite) presence, which in turn confirms the visual inspection and assessment of the sample (previously mentioned). Due to the soft and brittle nature of the scales referred in the Experimental section, which most likely were continuously removed by the flowing water (erosion–corrosion), crevice corrosion (also referred in some instances as under deposit corrosion) was minimized. The (high) oxygen content (16.7 wt.%) came from the (boiler feed) water flowing through the tube. Other elements commonly carried in that water were detected, such as Al (0.6 wt.%), Si (0.5 wt.%), P (0.3 wt.%), Cl (0.3 wt.%). On the other hand, C (1.1 wt.%) and Mn (0.3 wt.%) came from the carbon steel tubes internal walls and/or may have been dragged with the water from upstream equipment(s) [as may be the case of Si (0.5 wt.%) found as well]. The absence of metals such as Ca and Mg, commonly removed with the treatment of the boiler feed water, corroborates that the treatment was carried out normally and up to expected standards. Thus, water treatment may not have been a contributing factor for the origin of the failure (pit).

Fig. 5
figure 5

An EDXS spectrum of a sample from the surface area around the pit of the failed A214 carbon steel preheater tube. Main components are: Fe, O, C, Mn, Al, Si, P, Cl, among others. The average values of composition of the main components from the EDXS tests were determined to be (wt.%): Fe 79.3%; O 16.7%; C 1.1%; Mn 0.3%; Al 0.6%; Si 0.5%; P 0.3%; Cl 0.3%

Full size image

From Fig. 6a, it can be seen that the microstructure of the failed carbon steel tube consists of a ferritic matrix and pearlite regions, expected for this material. In addition, the absence of any irregularities (cavities, grain boundary voids, among others) in the microstructure indicates that other failure modes (such as creep) were not part of the mechanism for the failure of the material. This was expected, as the preheater unit operates at moderately low temperatures (around 204 to 286 °C on the shell side and 130 to 234 °C on the tube side), minimizing creep in the materials.

Fig. 6
figure 6

(a) Surface structure (top view, Nital as etchant, × 200 magnification) of a sample from the area around the pit; (b) cross-sectional view (Nital as etchant, × 10 magnification) of the sample from the failed area, clearly showing a significant decrease in ID thickness, the hole formed and the pit’s edges. The arrow indicates an oxide layer on the surface of the pit, exposed to the flowing boiler feed water. The barbed (finned) external surface of the tube can be seen on the top

Full size image

A defect (originated from a discontinuity/irregularity/non-uniform site) on the internal surface of the tube may have been present originally (from fabrication) at the failed area where the pit was observed. In addition, by looking/inspecting (Fig. 4b) the pit area, its geometry/form and its surroundings, the vortex formation on top of the pit may have contributed to the formation and subsequent growth of the pit (width) and then to the penetration inside of the pit (depth) to finally create the hole. The vortex was enhanced by the flowing water through the tube, accelerated by the flow turbulence due to flow direction change that caused the large cavities (formed by erosion–corrosion) to the left and right side of the pit (as indicated in Fig. 4b), and was acting in a radial direction toward the inner surface. This erosion–corrosion, in combination with the scales of iron oxides continuously developed at the surface of the pit, caused the reduction in the inner tube thickness all the way until the hole was created, triggering the rupture of the material and the immediate and necessary shutdown of the equipment and the entire (ammonia) plant. Figure 6b clearly shows the massive internal surface reduction in the material, with the creation of the pit and the hole inside the pit. The presence of an oxide layer is also seen in the image, verifying the presence of iron oxides and confirming the results from the spectroscopic analysis.

The continuous formation of oxides of iron over the internal surface of the material enhanced by the large content of (dissolved) oxygen in the flowing boiler feed water indicates that oxidation (and reduction) reactions were occurring nonstop over time during the operation of the plant; therefore, electrochemical cells were established, causing the degradation in the material. Thus, corrosion was the prevailing mechanism in the failure of the carbon steel tube.

From the ammonia plant operational and maintenance data, it was established that the failure of the carbon steel tubes (A214) in the waterside of the preheater unit occurred on several occasions, while the SS304L tubes that make part of the total bundle of tubes inside the preheater (Figs. 1a and 2) experienced no failures in any of those occasions. From July 2007 to March 2014 (date of the last reported preheater unit failure and unscheduled operation’s shutdown), there were four failures associated with carbon steel tubes, determined after performing the relevant water fill and 15 psig pneumatic air tests. In the first three events, the (quick and practical) solution to take back the unit to operation and then resume plant process was to plug not only the failed carbon steel tubes but, in some instances, some of the surroundings (carbon steel) ones. This a well-known common practice performed in these types of units/situations/plants, as it allows a fast resumption of the plant and minimizes costs and loss in revenues due to (prolonged) downtimes. That plugging of failed tubes and immediate resumption of plant’s operation did not allow for the cause of those three failure events of the preheater unit to be determined. At the 2014 failure (lasted 15 to 16 days, causing significant loss of revenues), the decision was taken to remove the whole unit for a closer inspection and further evaluation of the possible cause of the failure, with the results previously discussed. If stainless steel 304L tubes were used all over the waterside of the preheater, it is ascertained that the unit’s failures due to corrosion will be considerably reduced. The higher costs of the SS304L U-tubes will be overshadowed with the very costly repetitive plant shutdowns due to the analyzed failure. It is well known that the presence of Ni and Cr in stainless steel both forms protective coating layers of oxides over the SS surface with a sufficient and continuous source of oxygen, which is apparent in the flowing boiler feed water and has been corroborated in this study.

Conclusions

The corrosion failure analysis of carbon steel tubes (A214) in a reformed gas boiler feed water preheater unit operating at an ammonia plant was evaluated. It was determined that the initial pitting corrosion occurred at the internal walls of the failed tube, in the waterside, and was the prevailing mechanism for the failure. Iron oxides, including magnetite (Fe3O4), were present over the surface of the area of failure, in a pit, which was formed over time in the inner surface. The pit gradually experienced growth in width and in depth due to subsequent erosion–corrosion. This caused a significant thickness reduction in the internal wall that sequentially led to the formation of a hole and the tube failure. In addition, there were no voids/cracks in the grain boundaries in the microstructure (ferrite matrix, with dark pearlite regions) of the carbon steel tube (A214). Since the failure of the carbon steel tubes in the preheater unit was established to be recurrent, it is recommended to use stainless steel 304L tubes throughout the waterside of the preheater. The SS 304L tubes that make part of the total bundle of tubes inside the preheater experienced no failures during the same 7-year period in which the preheater unit was reported to fail in four occasions at several carbon steel tubes. Therefore, the use of the SS304L U-tubes will be an attractive cost-effective remedy, making as well the overall operation of the plant and the process more efficient.