During the sampling procedure, it was observed that the hydrocarbon obtained exhibited high water content. Water is necessary when a corrosion process is taking place and the extent of the damage depends on its corrosive characteristics.

In all cases, the samples were identified as condensation water, with low conductivity. However, the Langelier index indicated a corrosive tendency for all samples [7] . Values between −3 and −5 were observed, as shown in Figure 1. This situation implies that water may become very aggressive to the metallic structures.

Additionally, it must be indicated that the water corrosiveness for these systems was also related to the content of dissolved gases, including H₂S, CO₂ and O₂. For this case, the H₂S concentration was between 5 and 45 ppm, O₂ between 0 and 2.6 ppm and CO₂ between 28 and 75 ppm (Figures 2-4). The presence of these gases in water and hydrocarbon represents a corrosion risk for the metallic structures, and a prevention system must be considered [8] - [10] .

It is important to point out that the variations observed between the analyzed samples for P-I and P-II are slight; therefore, their corrosive tendency is similar. At the same time, the O₂ and CO₂ contents exhibit some differences that are considered as normal, due to the fluid characteristic variations.

One of the parameters that present bigger differences is the H₂S content. Here, an increase with time for samples taken from P-I was observed, reporting values between 20 and 40 mg/l. For samples taken from P-II, lower values were reported, in the range of 5 and 30 mg/l. It must be noted that the concentration of corrosive gases in water depends upon the general characteristics of the hydrocarbon, specially pressure and temperature. Therefore, the results allow establishing the presence of corrosive gases in the analyzed water samples.

In the same way, it may be indicated that iron content in the water samples is a reference to establish whether the corrosion process is taking place or not, and to determine the necessity of a control system [11] . For this case, both fluids P-I and P-II exhibited corrosive characteristics, although a difference in the iron content could be observed (Figure 5).

The samples taken in P-I had values around 0.5 ppm, while the samples from P-II reported values between 3.3 and 4.2 ppm.

These differences can be related to some specific system conditions and parameters:

• Chemical composition and/or material resistance;

• Temperature;

• Flow;

• System life time;

• Addition of chemicals, such as corrosion inhibitors, scale inhibitors, among others;

• Process efficiency.

On the other hand, the presence of microorganisms was determined in all water samples. Sulfate reducing bacteria populations around 100 bacteria/cm³ were observed. This situation is considered as a potential problem of localized corrosion.

The fluid is mainly composed of light hydrocarbons, such as methane, ethane and propane. This is considered as a normal situation. However, there were some corrosive gases in the hydrocarbon composition. Contents of H₂S, CO₂ and O₂ were determined, as shown in Figure 6 and Figure 7. Here, both points P-I and P-II, show similar gases proportion.

Considering their effect on the corrosion processes, these gases represent a continuous source of corrosive compounds for the aqueous phase. Once the products are consumed in the reaction, the gases dissolve in water to keep the corrosion process going on.

According to the fluid characteristics, the parameters identified as the responsible for the corrosion process in the systems are:

1) Water content;

2) Presence of dissolved corrosive gases;

3) Presence of microorganisms, mainly Sulfate reducing bacteria (SRB);

The presence of sulfate reducing bacteria (SRB) was identified inside the pipeline, next to the failure and in dif-

ferent areas along the inner wall. SRB have been considered as the main agent for the microbiological induced corrosion occurring in hydrocarbon transport and distribution systems [12] - [14] .

SRB presence was determined using culture media, according to API Recommended Practice No. 38 [15] . This is a specific culture for this kind of microorganisms.

When SRB are present in the samples, they reduce the sulfate of the media to sulfide, which reacts with iron to produce a black precipitate of iron sulfide. This is an indicative of SRB presence. When there is no presence of SRB, the culture media remains transparent with no change (Figure 10).

The formation of FeS deposits at the metal surface is one of the main characteristics of this type of corrosion. These results are in good agreement with those obtained in Stage 1, where the presence of SRB in water was indicated.

It is important to point out that the presence of SRB constitutes a risk for the integrity of metallic structures, as these microorganisms induce localized corrosion. Therefore, its identification and control becomes necessary.

The corrosion products obtained from the metal surface were analyzed by different techniques: X-ray diffraction, X-ray fluorescence, and Mössbauer spectroscopy. The following was observed:

Mössbauer Spectroscopy: This analysis is specific to determine iron compounds. The spectrum obtained is shown in Figure 11. The compounds found mainly correspond to iron sulfides: Troilite (FeS), Mackinawite (FeS_0.9) and FeS₂.

X-Ray Fluorescence: This technique is directed to establish the presence of chemical elements in the sample. The results indicate different elements, mainly S, Fe and O.

X-Ray Diffraction: Using this technique, the presence of diverse compounds with crystalline structure can be identified. The obtained diagram exhibited the following compounds: Mackinawite (FeS_0.9), Marcasite (FeS₂), Pyrite (FeS₂), Troilite (FeS) and Pyrrhotite (FeS).

According to theses analyses, it may be possible to establish that the main corrosion products formed at the inner metal surface are iron sulfides, which may result from:

• The presence of H₂S in the hydrocarbon and the associated water;

• The activity of SRB, which was identified in the system. Iron sulfide is a sub-product of the microorganism metabolism.

It is important to mention that although the X-ray fluorescence analysis indicated oxygen content, no oxides were identified by X-ray diffraction nor Mösbauer spectroscopy. In this way, it must be said that during the first inspection of the corrosion products at the metal surface, some “reddish” products characteristics for oxide compounds were observed. These red products were entrusted in the interstices of the inner metal wall, underneath the metal-corrosion products interface. For this reason, the oxides were not detected, even though its presence was visually corroborated.

Several surface analyses, using the Scanning Electron Microscope, were carried out at the site where the failure

occurred, before and after the corrosion products were removed from the metal surface [16] . The results corresponding to the analysis with corrosion products at the surface, exhibited a rectangular failure, with side dimensions of 253 μm and 629 μm, located at the welding joint (Figure 12). An elemental analysis indicated a typical mild steel composition, including Fe, C, Ni, Cr, Si, and Mn.

Moreover, S and O were observed. These elements are associated to the steel composition, but also could be related to the corrosion products formed at the metal surface. Once the corrosion products were removed, a surface analysis was carried out. A uniform corrosion process was observed in the entire metal surface, with a type of localized corrosion in specific sites (Figure 13).

In this case, the characteristics elements for carbon steel are still observed. However, there is lower oxygen content and the presence of sulfur was detected. These results corroborate the fact that the corrosion products are mainly formed by sulfides and oxides. Additionally, as the failure occurred in a welding joint, a galvanic corrosion effect was considered and a metallographic analysis was suggested.

A metallographic analysis and a hardness profile were carried out at the site where the failure occurred, considering the tube, the welding joint and the elbow. During the sample preparation, inclusions at the metal surface were evaluated. Sulfides and oxides were observed in the tube and oxides were noticed at the elbow surface. These results corroborate previous analysis, where the presence of these two compounds was reported. Regarding the hardness profile, the results indicate similar behavior for the three regions analyzed: tube, welding joint and elbow. Values around 74 Rockwell units (HRBW) were measured. However, the heat affected regions (HAR) exhibited higher hardness values, around 86 HRB, as shown in Figure 14.

On the other hand, according to the metallographic analysis, the tube―elbow microstructure arrangement indicated the presence of Pearlite and Ferrite, as shown in Table 1.

It is important to point out that the steel physical properties and its behavior depend mainly upon the carbon content and its distribution into the iron matrix. Most of the steels are a combination of three phases: pearlite,

(a) Grain size 9; (b) Grain size 7 - 8

cementite and ferrite. The strength and hardness of steel with no heat treatment depend on the proportion of these three phases.

For this specific case, a typical carbon steel microstructure formed by ferrite and pearlite was observed for the tube and elbow. The welding joint material had a microstructure formed mainly by proeutectoid pearlite that can be considered as typical for low carbon steel. The presence of dendrites in the welding joint may be related to a material heat effect. The regions identified as HAR presented microstructures constituted mainly by proeutectoid pearlite and fine pearlite. The different microstructures shown by each region could indicate a galvanic effect that contributes to the corrosion process at the failure site, which also corresponds to the welding joint region.

To complete the metallographic analysis and to identify any difference between the chemical composition of the tube, elbow and welding joint, a chemical analysis was carried out, using the Atomic Absorption Technique. This was done to determine the elements present in the metals and the results are summarized in Table 2. According to this table, the tube and elbow had similar chemical composition, indicating the same type of steel. However, the welding joint material presented some significant differences, mainly related to the content of Fe, C, Mn, Si and S.

A galvanic effect could be explained by these differences and the different microstructures observed in the metals. It seems that the welding joint material acts as an anode and the adjacent regions as a cathode.

The results obtained at this moment indicate that there are right conditions in the system for the development of a corrosion process in the inner wall of the pipeline. There is high water content, in addition to the presence of corrosive agents and appropriate material conditions. For the failure considered in this work, different corrosion processes were observed:

1) The first visual inspection of the pipeline inner wall, after the first metal cut and before any metal cleaning, indicated a severe uniform corrosion, identified by a reduction of the pipeline thickness in specific sites, mainly located at the 6 hours position (Figure 15).

This type of corrosion was generally observed at the elbows and “T” joints, where the fluid changes its flow direction. It is expected that during the hydrocarbon transport, heavier fluids are located at the bottom of the pipeline. For this specific case, water phase is in this position, generating aggressive conditions for the metal.

Due to the morphology, location and distribution of this type of uniform corrosion, an erosion corrosion effect is considered. In this type of corrosion, the hydrocarbon flow is enough to remove corrosion products from the metal surface, decreasing their protective effect and increasing the corrosion rate. Corrosion processes due to fluids flow usually induced a localized impact pattern. The failure was located at the welding joint, in a site where an erosion corrosion effect was also observed, as shown in Figure 16.

In this way, it is very important to verify the fluid velocity, as it should not exceed the recommended material limits. From a corrosion point of view, a smooth flow is always preferable to a turbulent flow. At the same time, gases and solid particles must be eliminated from the fluid as possible.

2) A galvanic corrosion process may be considered at the welding joints regions. This type of corrosion occurs because of potential differences between metallic materials, when in contact and in presence of an electrolyte. The material with more negative potential acts as an anode and exhibited a corrosion process. For this specific case, apparently the welding joint acted as anode in the corrosion reaction. However, the galvanic corrosion was less evident in the site where the failure occurred, because of the effect of the erosion corrosion. At the 12 technical hours position, the erosion corrosion effect is less evident and the galvanic corrosion is clearly observed (Figure 17).

3) Microbiological induced corrosion, which can be associated to localized corrosion processes, was observed along the metal sample. It was more evident at longitudinal regions, between 5 and 7 technical hours positions, although this type of corrosion was also detected in most of the pipeline inner metal surface, as shown in Figure 18. Corrosion products were removed from some pits and SRB populations were detected inside the cavities.

Additionally, microorganisms were also detected at the region where the failure occurred. This situation indicated that microbiological induced corrosion had also an effect on the metal failure, although the evidences were not clear, because of the presence of the other types of corrosion.