- Email: [email protected]
Corrosion failure analysis of galvanized steel pipes in a closed water cooling system
Engineering Failure Analysis 84 (2018) 46–58
Contents lists available at ScienceDirect
Engineering Failure Analysis journal homepage: www.elsevier.com/locate/engfailanal
Corrosion failure analysis of galvanized steel pipes in a closed water cooling system
T
A. Colombo, L. Oldani, S.P. Trasatti⁎ Department of Chemistry, Università degli Studi di Milano, Via Camillo Golgi 19, 20133 Milan, Italy
AR TI CLE I NF O
AB S T R A CT
Keywords: Galvanized steel Closed cooling system Corrosion failure analysis PET preforms
The present study is focused on the corrosion failure analysis of a water cooling system serving a polyethylene terephtalate (PET) preform molding plant. Piping mainly located both after the water pretreatment and in the injection molding machines was affected by an extensive corrosion phenomenon and several leakages of coolant occurred, involving the risk of loss of functionality of the whole plant. The main problem detected in closed cooling circuits was a marked modification of the water chemistry and a great amount of metal ions leading to recurring fouling of the inner surfaces of molds and cooling channels. Analysis of water flowing in the closed cooling circuit operating at 6 °C revealed a very high amount of metals in the flowing water and in the sediment (zinc and iron). Moreover the water pH was slightly acidic. Data were compared with water from supply well. Preliminary electrochemical characterization was carried out on galvanized steel pipe, in order to evaluate the aggressiveness of water both from the cooling circuits and from the supply wells. The results of laboratory analysis together with the scrutiny of the available technical documentation showed that the damage of the plant was the consequence of an inadequate plant management, mainly for what concerns materials selection and water chemical treatment.
1. Introduction Water cooling systems are an essential part of many industrial processes. They require proper chemical treatment and preventive maintenance in order to ensure continuous plant productivity. The functionality and the service life of the cooling system equipment are provided by selecting suitable materials of construction and supply water of required quality [1,2]. Typical building materials for cooling systems include carbon steel, stainless steel, galvanized steel, copper and copper alloys. The addition of a corrosion inhibitor, sometimes coupled with a biocide, can help to prevent scale formation and corrosion in industrial water cooling systems. Chromate, nitrite and molybdate compounds are the most reliable inorganic corrosion inhibitors [3–6]. Recently, organic/inorganic mixtures have been also developed as corrosion inhibitors [7,8]. Closed cooling systems often require also the addition of antifreeze. Galvanized steel pipes are often used for water cooling systems both in industrial plants and in domestic buildings [9,10]. The extensive use depends on their good performance against corrosion, their mechanical workability and the resistance to biofouling [11,12]. The zinc layer protects steel against corrosion by two effects: a barrier effect and a galvanic protection because Zn acts as a sacrificial anode. However, galvanized steel pipes can be affected by localized and generalized corrosion that sometimes lead to the
⁎
Corresponding author. E-mail address: [email protected] (S.P. Trasatti).
http://dx.doi.org/10.1016/j.engfailanal.2017.10.008 Received 26 May 2016; Received in revised form 9 April 2017; Accepted 11 October 2017 Available online 12 October 2017 1350-6307/ © 2017 Elsevier Ltd. All rights reserved.
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
corrosion failure of water cooling systems before their expected service life [13–16]. The composition of the water heavily affects corrosion phenomena in water distribution pipes. For example, soft waters are referred as aggressive, because they tend to prevent the formation of films of calcium carbonate [17]. Corrosion can also be enhanced either by high levels of chloride and sulphate, low pH or high temperature [18,19]. The rate of corrosion decreases considerably when protective films originate on the surface of the metal. Monitoring and prevention of corrosion phenomena is of outstanding importance in order to avoid main damages like pipe break-down and water leakages, decrease in the cooling efficiency, collapse of the system, shut-down of the whole plant for maintenance services [20]. The present paper reports on the corrosion failure analysis of galvanized steel piping in a water cooling system serving a PET (polyethylene terephtalate) preform molding plant. PET preforms are the basis of very common plastic bottles typically used for drinking waters, fruit juices, soft drinks, oils and milk. PET preforms are usually produced by injection molding. The starting material are PET pellets, melted at 270 °C and injected under high pressure into a mold to make the preform. In each inner mold part there is a water cooling circuit. The temperature of cooling water is generally in the range 6–20 °C. Proper functioning of the cooling circuit is critical for ensuring stability of the end-product. If PET is exposed to temperatures above 270 °C, the polymer may thermally degrade and produce acetaldehyde. Acetaldehyde can migrate into bottled drinks and give the product an acidic taste [21–23]. Therefore the proper control of the process is essential to obtain products in compliance to customer requirements. The aim of the present study was to determine the reasons of unexpected corrosion failures detected at different locations of the whole piping cooling system. The plant under investigation is located in the center of Italy and has 20 lines of production (injection molding machines) that operate 24 h/day. Six closed cooling systems, operating at 6 °C and 20 °C, serve the injection molding machines. A schematic of the plant site is reported in Fig. 1. The plant site includes two different wells: i) Well 1, serving the cogeneration plant to recover part of the energy required for the process; ii) Well 2, providing make-up water to the closed cooling circuit in order to compensate potential leakages. Water is pumped by submerged pumps from both wells. The internal walls of the wells (30 cm diameter) are made of carbon steel. After pretreatment, water from Well 1 is conveyed to water pretreatment and cogeneration system by galvanized steel pipes. Water pretreatment is constituted by sand filter and softener. On the other hand, water from Well 2 is conveyed by polypropylene pipes to a drum, then to sand filter and softener and finally to the chiller. Within the water pretreatment, the interconnections are mainly constituted by galvanized steel, but copper, carbon steel and stainless steel are still present. From the chiller to the inlets of production lines, water flows through pipes made of AISI 304. Then the water enters the closed cooling circuits of the PET preform molds
Fig. 1. Schematic of the plant site. Arrows indicate the original location of analyzed metal samples.
47
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
that are typically made of steel or aluminum. According to the provided information, when the plant was set up, several parts of piping system were made from carbon steel. In the first stages of operation, a mixture of water and ethylene glycol was used as coolant with the addition of an inhibitor. In 2007 carbon steel was completely replaced with galvanized steel but the water/ethylene glycol/inhibitor mixture was partially maintained as coolant, without considering the ratio water/glycol, with the latter that underwent dilution over time. These changes were made to replace damaged parts of the plant. In the following years the piping system suffered with several corrosion failures at different locations where galvanized steel was present. These failures accompanied by frequent leakages of coolant demanded for continuous replacement of damaged parts and involved also the risk of severe loss of functionality of the whole plant. Nevertheless, the molding machine continued to operate without stop up to 2014. An extensive failure analysis was performed on several samples of the galvanized piping system extracted from the plant. 2. Materials and methods 2.1. Materials 2.1.1. Technical documentation Part of the failure analysis activities were devoted to get data and information about the plant management during the period 2007–2014, when corrosion problems occurred. In particular, water analysis reports issued by three water service companies were scrutinized. 2.1.2. Metal samples An unused hot-dip galvanized steel pipe was supplied by the same manufacturer of pipes used to replace damaged parts in the plant. The new pipe was used as reference material. Sections of failed parts were collected at several points of the plant: i) drum; ii) sections of pipes; iii) core element of a mold with internal cooling channel. The sampling sites are indicated by arrows in Fig. 1. 2.1.3. Cooling water samples Three water samples were obtained for chemical analysis and corrosion tests from: i) the well serving the cogeneration plant (Well 1 after softener); ii) the well supplying make-up water to closed cooling circuits (Well 2 after softener); iii) one of the closed cooling circuits operating at 6 °C. 2.2. Method 2.2.1. Visual, metallographic, SEM and XRD examination Macroscopic pictures were taken to provide documentary evidences of the state of collected samples and the distribution of corrosion products. Metallographic examination was carried out to check the microstructure of the base steel. For metallographic purposes, some of the specimens were embedded in thermoplastic resin, polished with abrasive papers down to 2000 grit and diamond pastes down to 0.25 μm. Cross-sections of some of the samples after metallographic preparation by mechanical polishing and microetching were examined by optical microscope to investigate quality and thickness of zinc layer. SEM (Scanning Electron Microscope) analyses were performed to investigate the composition and the morphology of corrosion products and to determine the thickness of zinc layer. The morphology of inner and outer surface of pipes was observed by a SEM mod. LEO 1430 equipped with an EDS microprobe (Energy Dispersive X-ray Spectrometer INCA Oxford Instruments). The elemental composition of corrosion products was determined by EDS. Before SEM analysis, all samples were coated with a very thin layer of gold by sputter Nanotech SEMPREP 2. The corrosion products were characterized by XRD (X-Ray Diffratometry) analysis with a Bruker APEX II powder diffractometer with Cu Kα radiation. 2.2.2. Water analysis A complete physico-chemical analysis was carried out on all collected water samples. pH and redox potential were determined with a Digital pH meter mod. 337 Amel. Conductivity measurements were performed with a mod. 160 Amel conductivity-meter equipped with a mod. 192/K1 Amel cell. Hardness of water was determined by titration with EDTA (following the procedure APAT IRSA CNR [24]) and results are here reported as ppm of CaCO3. Cations and anions were determined by ICP-AES (Inductively Coupled Plasma – Atomic Emission Spectroscopy) and ion-exchange chromatography respectively. The sediment in the water from cooling circuit was analyzed by ICP-AES and EDS, after preliminary filtration, drying and microwave-assisted acid digestion with HNO3. 48
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 2. Actual condition of the plant: a) wellhead; b) tank inlet; c) plant area.
2.2.3. Electrochemical tests Since corrosion is basically an electrochemical process, electrochemical tests could be considered a valuable method to gain insight into the corrosion phenomenon. Accordingly to this approach, polarization curves were acquired for an unused section of pipe using the three collected water samples as test media. Tests were carried out at 25 °C in aerated solutions under atmospheric pressure with a mod. 273A EG & G potentiostat/galvanostat. A classic three electrode configuration was adopted with a platinum wire as a counter electrode and a saturated calomel electrode (SCE) as a reference electrode. A Luggin capillary was used to minimize ohmic drops in solution. The galvanized pipe coupon was degreased with hexane, rinsed with distilled water, dried in air and then placed at the bottom of the cell, with an exposed area of 1 cm2. Before each experiment the system was allowed to stabilize at the open circuit potential (OCP) for 30 min. Cyclic polarizations were performed at a scan rate of 10 mV/min. The polarizations curves were recorded in the anodic direction from the OCP until the current density reached 1 mA/cm2, and then reversed. 3. Results and discussion 3.1. On-site inspection The plant was visited twice at the beginning of 2014 for visual inspection of the condition of the piping system. The inspections were aimed at evaluating the extent of corrosion damages by speaking with operators about the process phases and collecting samples for laboratory analysis. Sample collection concerned both external piping (from the wells to the production area) and internal closed cooling circuits. 3.1.1. External piping Fig. 2 shows the condition of the plant: extensive corrosion phenomena affect all the metallic parts possibly in contact with the circulating coolant from wellhead (Fig. 2a) to the chiller (Fig. 2b), up to the production area (Fig. 2c). Most of the plant equipment is affected by severe corrosion. Leakages of coolant occurred at several points and were detected at different times. Corrosion was mainly in the form of uniform corrosion, though some localized attacks were detected on stainless steel pipes. Persistent rusty deposits were also visible on the facilities and on the ground in all the plant. Following the inspection, some problems about the construction and the management of the plant can be supposed. In particular, different metallic materials were not correctly coupled (galvanized steel coupled to stainless steel), possibly leading to galvanic corrosion phenomena. Talking with plant operators, it emerged that the proper functioning of water pretreatment had never been properly checked up. 3.1.2. Internal closed cooling circuits Fig. 3 shows the connection between the external piping and the coolant inlet (Fig. 3a) into one of the molding injection machine (Fig. 3b). Figs. 4 and 5 show the core part of one mold and the relative cooling channel, respectively. Molds are periodically taken apart for cleaning operation. The internal surfaces of all the molds are always covered by very compact reddish deposits that are usually mechanically removed by ultrasonic cleaning in water. These very adherent rusty deposits are a real concern for the plant manager, since the very high cost of each mold (approximately 500.000 Euros) completely falls on the PET preform company and not on the buyer. This aspect justifies the need that the plant works 24 h a day and the impossibility to schedule stops for maintenance operations. So ordinary maintenance would be highly desirable but it is unfeasible. In spite of the condition of external piping and closed cooling circuits, the quality of produced PET preforms never gets worse. 3.1.3. Analysis of technical documentation The approach to this problem followed a typical corrosion failure procedure, with the preliminary collection of plant service history and background data. The company never adopted proper maintenance actions, such as scheduled inspections, specifically devoted water pretreatment, and adoption of procedure for cleaning piping. Moreover, from meeting with technical staff it was inferred that careful selection of the materials has never been a priority for the management. When severe leakages of coolant happened, damaged pipes were simply 49
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 3. Actual condition of the plant: a) connections between external piping and internal closed cooling circuit; b) injection molding machine inlet.
Fig. 4. Accessory for PET pre-form mold made of high grade steel (surface-hardening heat treatment): a) core part; b) connection to the injection molding machine.
Fig. 5. Accessory for PET pre-form mold made of stainless steel: a) internal cooling channel; b) detail.
Table 1 Water analysis in external piping system. Well 1
pH Conductivity (μS/cm) Hardness (ppm CaCO3) Redox potential vs. Pt (mV) Cl− (mg/l) Fe2+ (mg/l) Zn2+ (mg/l) Cu (mg/l) Microbial count at 20 °C (UFC/1 ml) Microbial count at 37 °C (UFC/1 ml)
Well 2
After softener
Sept 4th 2013
Oct 1st 2013
Sept 4th 2013
Oct 1st 2013
Sept 12th 2013
7.7 480 276 – 30.2 0.01 0 0.43 – –
7.02 649 300 − 216 51.9 7100 – – 400 260
7.8 507 316 23.5 29.8 0.02 0.1 0.9 – –
6.79 753 375 86 – 230 – – 670 270
7.3 590 0 – 16.6 0 – – – –
50
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Table 2 Water analysis in the six internal closed cooling circuits: average values and standard deviations.
pH Conductivity (μS/cm) Hardness (ppm CaCO3) Redox potential vs. Pt (mV) Cl− (mg/l) Fe2+ (mg/l) Zn2+ (mg/l) Cu (mg/l)
Sept 4th 2013
Sept 12th 2013
6.4 ± 0.2 1222 ± 462 428 ± 78 – 205 ± 66 122 ± 63 52 ± 29 5.6 ± 4.3
6.0 ± 0.2 1355 ± 229 180 ± 36 – 204 ± 58 122 ± 63 – –
replaced by new ones. Since the problem has been dragging along for a long time and the plant could not be stopped, the company only turned to independent water service companies. These societies only collected water samples at different locations of the plant. Results are reported in Tables 1 and 2, in term of several physico-chemical parameters. The focus is on the last part of 2013. Careful scrutiny of the water analysis reports revealed lacking and/or conflicting data on pH and chlorides as well as iron content. As data of October 2013 assessed (Table 1), water from wells showed a marked microbial contamination by iron bacteria and a highly negative redox potential that can be related to the corrosion damages of galvanized steel pipes [25]. Microbiologically influenced corrosion (MIC) can potentially contribute to failures in galvanized steel systems containing water. MIC has been suggested to induce failures as leaks through pitting corrosion [26–28]. Management strategies should consider microbial control to prevent corrosion in these systems. However, corrosion by iron bacteria only occurs if the internal zinc layer is damaged [29]. For what concerns the water flowing through the internal closed cooling system (Table 2), analysis revealed very high amount of iron, copper and zinc ions in all the six circuits, probably released from pipes. Chloride levels should also warn, since they could be considered one of the reasons of initial corrosion. Unlike wells water, conductivity is very high and pH is slightly acidic. At the initial stage of the production process, the cooling water was added with ethylene glycol whose concentration was never monitored. The natural degradation of glycol could have led to lowering of pH, thus promoting the beginning and propagation of corrosion.
3.1.4. Preliminary remarks At the end of the on-site inspection and after the examination of technical documentation, some critical factors were identified as follows: - the observed uniform corrosion could be attributed to galvanic coupling of the several alloys used for pipes of closed circuit (copper, galvanized steel, AISI304) and molds (steel, aluminum); - the presence of unexpected levels of chlorides suggested that the water softener did not operate properly over time, thus possibly promoting their release in the coolant; - ethylene glycol, formerly used as coolant fluid, was never monitored even though piping was moved from carbon steel to galvanized steel. Eventual decomposition of glycol to organic acids could have caused the observed increase of acidity. Likely all these factors contributed to the failure of the system. However it was very difficult to draw conclusions only on the basis of technical report scrutiny and visual inspection. To go deeper into the approach to the problem, the failure analysis of metal materials from the plant was carried out in order to evaluate corrosion morphology, corrosion products and zinc layer thickness. A complete investigation required also a detailed water analysis.
3.2. Metallographic analysis and SEM examination 3.2.1. New galvanized steel pipe An unused hot-dip galvanized steel pipe supplied by the manufacturer was taken as reference material (Fig. 6a). The internal diameter is 2.68 cm. From metallographic examination, the substrate shows a typical ferrite-pearlite microstructure of low carbon steel (Fig. 6b). The quality of zinc layer was evaluated by metallographic and SEM examination. The zinc layer is uniform and well-adhered throughout the sections of examined pipes. Usually, a hot-dip galvanized coating consists of several layers. Starting from the steel surface, each layer is an iron–zinc alloy with increasingly lower iron content [30]. SEM images of cross sections in Fig. 7 show the typical morphology of a zinc coated steel, consisting of the coating alloy, an interfacial layer between the coating and the substrate steel (containing a series of intermetallic compounds), and the substrate steel [11]. The European Standard EN 10240 [31] states that a minimum of 55 μm of zinc in the inner side of galvanized steel pipes is required for good protection of steel against uniform corrosion in fresh water. The new pipe exhibits a 60 ± 5 μm zinc layer on the outer side (Fig. 7a) and 53 ± 5 μm zinc layer in the inner side (Fig. 7b). The average thickness of the coating layer is slightly lower than the minimum value specified in the European standard. 51
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 6. Unused galvanized steel pipe: a) section; b) microstructure.
Fig. 7. Cross section of new pipe: a) external surface; b) internal surface.
3.2.2. Damaged metal samples The inner surface of the collected samples had significant material loss from the zinc layer and the base material. The material loss was in the form of general corrosion. In Fig. 8, a coupon of a galvanized steel drum located before the water pretreatment is shown. The typical reddish color of rust is clearly visible on both inner and outer sides. In Fig. 9 the micrograph of the cross section and the EDX analysis performed on the inner side of the drum coupon are shown. The metal underwent severe generalized corrosion with complete depletion of the zinc layer. Sections of failed pipes are shown in Figs. 10 and 11, respectively. Sample in Fig. 10 has been collected from the piping conveying water from Well 1 to cogeneration. Sample in Fig. 11 belonged to piping system within the water pretreatment downstream of Well 2. At a preliminary visual inspection, the outer surface of in-service pipes does not show evident damages, except in the threaded areas. On the other hand, inner surfaces looked severely attacked by corrosion. They have a very compact layer of reddish corrosion products that completely cover the surface. Upon some mechanical removal of corrosion products, corrosion damages of the
Fig. 8. Galvanized steel drum: a) inner side, b) outer side.
52
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 9. Galvanized steel drum: cross section end EDX analysis of the inner side.
Fig. 10. Example of damaged pipe from the piping conveying water from Well 1: a) external view; b) internal view.
Fig. 11. Inner side of failed pipes from water pretreatment piping downstream of Well 2, showing corrosion damages under corrosion products.
underlying steel are also evident. In some cases localized corrosion attack caused the formation of pitting holes in the areas of welding. The morphology of corrosion damages in the welding areas could be related to microbial induced corrosion (Fig. 11b). This aspect is in agreement with the detection of iron bacteria in the water analysis performed on October 2013. The inner surfaces of these failed pipes were characterized by SEM and EDS to check the morphology of corrosion and to determine the elemental composition of corrosion products (Fig. 12). Different layers of corrosion products are clearly visible, pointing to an extensive corrosion phenomenon that started at different times. Chemical analysis showed that deposits consist mainly of iron oxide (Fig. 12b). The degree of oxidation is higher in the upper layer (labeled SP1). Some traces of zinc were still detected, but the amount is about 10%. Other elements were detected in trace levels, namely Si and Al. Higher magnification of the lowest layer SP4 revealed that corrosion products are constituted by particles different in size and morphology (Fig. 12c). The relative elemental composition is also reported and shows that flower-like particles are in higher degree of oxidation than smaller particles. Smaller aggregates displaying acicular features are the prevailing morphology, and they could related to eventual MIC [32]. The presence of manganese, which is an alloying element of steel, was also detected. 53
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 12. a) SEM image of inner surface of damaged galvanized steel pipe (SP indicates the spot site where EDS analysis was performed); b) EDX analysis at different sites; c) higher magnification of SP4 with relative EDS analysis.
XRD analysis of corrosion products near the perforation area is reported in Fig. 13. Lepidocrocite (γ-FeO(OH)) and goethite (αFeO(OH)) are the main constituent. Since no evidences of sulphides were detected, the hypothesis of MIC is not actually supported by experimental data. The micrographs of the cross sections of the failed pipes as well as SEM images (Fig. 14) show that the state of zinc layer is completely different on the two sides of failed pipes. On the outer side (Fig. 14a) there is a fully-dense and continuous layer of zinc, with a thickness of 78 ± 5 μm (Fig. 14c). The inner surface instead is characterized by several directional pits (Fig. 14b). This could be the result of erosion-corrosion of the underlying steel directly exposed to the water flowing into the pipes, as a consequence of high amounts of sediments. SEM image also confirms the complete depletion of zinc layer (Fig. 14d). 3.3. Water analysis A detailed analysis of the water from the wells and from the circuit was performed in order to elucidate the environment that damaged pipes experienced. The water from the wells is completely clear and transparent and it is usually not subjected to any preliminary treatment, except for sand filter and softener. On the other hand, the liquid flowing into closed cooling circuits contains a great amount of slurry reddish sediment.
Fig. 13. XRD analysis of corrosion products from damaged galvanized steel pipe.
54
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 14. Cross sections of failed pipes: micrograph of external surface (a) and internal surface (b); SEM images of external surface (c) and internal surface (d).
Physico-chemical parameters determined in the three water samples collected during the on-site inspection are listed in Table 3. Water from both wells has a pH value near 7, while the water circulating in the closed cooling circuits is acidic (pH 4.8). In the absence of reducing or passivating agents, the corrosion of zinc in aqueous solution is primarily determined by pH values and zinc readily dissolves in acidic or strongly alkaline solutions [10,33]. Conductivity in cooling circuit is nearly the double than in Well 2, indicating that a great amount of salts are therein dissolved. Water hardness is another important variable in zinc corrosion. The content in calcium carbonate for Well 2 is not as low as expected if softener had worked properly. However, hard waters tend to be less corrosive toward zinc because they deposit a protective scale on the surface of the metal. The redox potentials indicate that the water under investigation is normally oxygenated at the time of analysis. Tables 4 and 5 report the amount of anions and cations determined in the three water samples. The amount of chlorides in the cooling circuit is lower than in well water and is very lower compared to September 2013 (Table 2). This is probably due to the introduction of make-up water of different quality at different times. For what concerns the cations, at the time of analysis the quality of sampled well waters is consistent with that of typical subsurface waters from wells. The water from cooling circuit instead exhibits a very high levels of zinc (100 ppm), indicating that a great amount of metal dissolved from inner surface of pipes and passed into solution. A significant amount of iron (43 ppm) was also detected in solution. The reddish sediment is mainly constituted of iron (435 ppm). This finding was confirmed by SEM-EDS analysis of the dried sediment (Fig. 15a) revealing that it is mainly constituted by amorphous iron oxides (Fig. 15b). Since there is any trace of iron in the well water, but there is in the cooling circuit, therefore the source of the iron must be dissolved ions formed during the corrosion process. XRD analysis (Fig. 15c) further confirmed that sediment is mainly constituted by iron oxy-hydroxides. 3.4. Electrochemical tests Electrochemical polarization experiments were performed on a new galvanized pipe to investigate the aggressiveness of the water Table 3 Water analysis: physico-chemical parameters.
pH Conductivity (μS/cm) Hardness (ppm CaCO3) Redox potential vs. C (mV) Redox potential vs. Pt (mV)
Well 1 after softener
Well 2 after softener
Cooling circuit (6 °C)
7.5 762.8 5 0.2274 0.0175
7.3 749.7 103 0.2438 0.0131
4.8 1333 167 0.1765 0.0363
55
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Table 4 Water analysis: anions.
−
Cl (mg/l) SO42 − (mg SO4/l) PO43 − (mg P/l) NO3− (mg N/l)
Well 1 after softener
Well 2 after softener
Cooling circuit (6 °C)
45.93 23.14 0.142 n.d.
45.57 23.11 0.119 n.d.
19.10 18.25 n.d. 0.15
Table 5 Water analysis: cations as determined by ICP-AES (ppm).
Cu Ni Zn Cr Mn Al Fe Na Ca
Well 1 after softener
Well 2 after softener
Cooling circuit (6 °C)
0.005 < 0.005 0.007 < 0.005 < 0.005 0.005 0.037 149 0.62
< 0.005 0.007 0.05 < 0.005 0.03 0.02 0.32 120 30
0.046 0.78 100 0.006 30 0.011 43 146 68
Fig. 15. Reddish sediment in water from cooling circuit: a) SEM image; b) EDS analysis, c) XRD analysis.
samples collected from the plant toward the pipes usually installed in the plant. The results are reported in Fig. 16. The galvanized steel exhibits quite similar trends in all environments. There is a sharp increase of the current and there is no passive region. The increasing potential leads to the active dissolution of zinc that continues until the zinc surface is covered by a layer of corrosion products that do not originate a passive layer. This is the typical behaviour of an active material. However, by the determination of the corrosion current Icorr at the corrosion potential Ecorr, it is possible to draw some general conclusion about the behaviour of zinc layer into the three different solutions under investigation. Icorr was determined by Tafel extrapolation from the anodic branch of the polarization curve, assuming that the cathodic process does not change [34]. The so determined Icorr were 0.112 mA/cm2 for Well 2, 0.117 mA/cm2 for Well 1 and 0.144 mA/cm2 for closed cooling circuit, respectively. The corrosion current is slightly higher for water from Well 1 than water from Well 2, most likely because the water from Well 2 has a higher concentration of carbonates (e.g., is harder). Corrosion occurs more slowly in harder water due to the deposition of protective layers. It is possible also to observe that the OCP of galvanized steel is higher (more noble) in well waters than in water from closed 56
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 16. Cyclic polarization curves of a new galvanized steel pipe in the three different water samples.
cooling circuit. Moreover, the highest corrosion rate was determined for water from the closed cooling circuit, even though its hardness was similar to the water of Well 2 and chloride contents is about half that of water from wells. Since the corrosion rate of zinc is pH dependent, the higher aggressiveness of water from closed circuit can be related to the lower pH of the solution (pH 4.8). On the basis of metallographic analysis and electrochemical test, a pH-related corrosion mechanism can thus be proposed to explain the corrosion failure of galvanized steel piping. 4. Conclusions In the present work an extensive corrosion failure analysis was carried on to elucidate the reason of the premature failure of the galvanized steel piping of a plant for PET preforms. On the basis of the plant history and of the experimental results, it is possible to state that: - an inadequate approach to maintenance problem and material selection was observed; - high levels of hardness after the softener and the variable amount of chlorides suggest that the softener does not work properly; - galvanized steel pipes employed for the closed cooling circuit are in compliance with the European Standard EN 10240 for what concerns quality and thickness of zinc layer: no material anomalies were observed that would have caused the failure; - a relatively low pH (possibly due to degradation reactions of glycol) and erosion-corrosion phenomena by solid deposits are the most probable factors for the failure of the pipes before their expected service lifetime; - fast corrosion of zinc exposed the underlying steel to the aggressive attack of acidic solution, causing extensive corrosion damages. 5. Recommendations Possible solutions to improve plant performance and to prevent future leakages from cooling systems are: -
to to to to
conduct a proper materials selection and design to avoid risk of unwanted galvanic coupling; monitor the quality of the water from the wells and the cooling system and to set-up an eventual proper water pretreatment; implement a scheduled maintenance of the whole plant; design a cleaning procedure to keep under control the amount of suspended solid.
Acknowledgments This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The authors thank Dr. Edoardo Guerrini for SEM analyses, and Mr. Giorgio Termignone for metallographic analysis of metal samples. References [1] [2] [3] [4] [5] [6]
S. Ahmed, J. Bayers, Design basics for cooling systems, Chem. Eng. (New York). 107 (2000) 78–83. J.W. McCoy, The Chemical Treatment of Cooling Water, 2nd ed, Chemical Publishing Company, New York, 1983. A.A.M.D.B. Alexander, Evaluation of corrosion inhibitors for component cooling water systems. Corrosion, Corrosion 49 (1993) 921–928. D.E.A. Marshall, B. Greaves, Low nitrite inhibitor treatment for cooling systems, Mater. Perform. 5 (1986) 45–50. J.P.G.F.M.S. Vukasovich, Molibdate in corrosion inhibition – a review, Mater. Perform. 25 (1986) 9–18. G.O. Ilevbare, G.T. Burstein, The inhibition of pitting corrosion of stainless steels by chromate and molybdate ions, Corros. Sci. 45 (2003) 1545–1569, http://dx.
57
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
doi.org/10.1016/S0010-938X(02)00229-9. [7] A.A. Moccari, Corrosion inhibitor evaluation for materials used in closed cooling water systems, Mater. Perform. 38 (1999) 54–59. [8] N. Ochoa, G. Baril, F. Moran, N. Pébère, Study of the properties of a multi-component inhibitor used for water treatment in cooling circuits, J. Appl. Electrochem. 32 (2002) 497–504 (doi:10.1023/A:1016500722497). [9] Internal corrosion of water distribution systems, AWWRF-DVGW-TZW Cooperative Research Report, Denver, CO, 1996. [10] P.G. Rahrig, Galvanized steel in water and water infrastructure, Mater. Perform. 42 (2003) 58–60. [11] A.R. Marder, The metallurgy of zinc-coated steel, Prog. Mater. Sci. 45 (2000) 191–271. [12] X.G. Zhang, Corrosion and Electrochemistry of Zinc, 1st ed, Plenum Press, New York, 1996. [13] B. Minnoş, E. Ilhan-Sungur, A. Çotuk, N.D. Güngör, N. Cansever, The corrosion behaviour of galvanized steel in cooling tower water containing a biocide and a corrosion inhibitor, Biofouling 29 (2013) 223–235, http://dx.doi.org/10.1080/08927014.2012.763117. [14] D.W.E. Jr, E. William, An investigation of zinc oxide corrosion of galvanized steel in cooling towers, ASHRAE Trans. 91 (2003) 91–100. [15] J.J.R.E. Otero, The assessment of the short term data of pipe corrosion in drinking water. I. Galvanized steel, Corros. Sci. 34 (1993) 1581–1593. [16] C.A. Della Rovere, R. Silva, C. Moretti, S.E. Kuri, Corrosion failure analysis of galvanized steel pipes in a water irrigation system, Eng. Fail. Anal. 33 (2013) 381–386, http://dx.doi.org/10.1016/j.engfailanal.2013.06.024. [17] F. Delaunois, F. Tosar, V. Vitry, Corrosion behaviour and biocorrosion of galvanized steel water distribution systems, Bioelectrochemistry 97 (2014) 110–119, http://dx.doi.org/10.1016/j.bioelechem.2014.01.003. [18] J. Román, R. Vera, M. Bagnara, A.M. Carvajal, W. Aperador, Effect of chloride ions on the corrosion of galvanized steel embedded in concrete prepared with cements of different composition, Int. J. Electrochem. Sci. 9 (2014) 580–592. [19] V. Padilla, A. Alfantazi, Corrosion film breakdown of galvanized steel in sulphate–chloride solutions, Constr. Build. Mater. 66 (2014) 447–457, http://dx.doi. org/10.1016/j.conbuildmat.2014.05.053. [20] B. Hickinbottom, Why does operational management put corrosion prevention and the monitoring of the onset of corrosion at the lowest level of priority and how to overcome this attitude, Proc. - Corros., Australasian Corrosion Association, 2000, pp. 50–57. [21] F. Villain, J. Coudane, M. Vert, Thermal degradation of polyethylene terephthalate: study of polymer stabilization, Polym. Degrad. Stab. 49 (1995) 393–397, http://dx.doi.org/10.1016/0141-3910(95)00121-2. [22] G.W. Halek, The zero-order kinetics of acetaldehyde thermal generation from polyethylene terephthalate, J. Polym. Sci. Polym. Symp. 74 (2007) 83–92, http:// dx.doi.org/10.1002/polc.5070740110. [23] T.Y. Kim, S.A. Jabarin, Solid-state polymerization of poly(ethylene terephthalate). III. Thermal stabilities in terms of the vinyl ester end group and acetaldehyde, J. Appl. Polym. Sci. 89 (2003) 228–237, http://dx.doi.org/10.1002/app.11905. [24] Metodi analitici per le acque, APAT IRSA-CNR, (2003). [25] Roscoe Moss Company, Handbook of Ground Water Development, Wiley, 1990. [26] S.S. Carl, Remediation of corrosion problems inside fire sprinkler system piping, in: Corros. – Annu. Conf. Natl. Assoc, Corros. Eng. (2000) 653. [27] R.J.N. Fernance, P.A. Farinha, SRB-assisted MIC of fire sprinkler piping, Mater. Perform. 43 (2007) 46–49. [28] A.C.E. Ilhan-Sungur, Microbial corrosion of galvanized steel in a simulated recirculating cooling tower system, Corros. Sci. 52 (2010) 161–171, http://dx.doi. org/10.1016/j.corsci.2009.08.049. [29] N. Bolton, M. Critchley, R. Fabien, N. Cromar, H. Fallowfield, Microbially influenced corrosion of galvanized steel pipes in aerobic water systems, J. Appl. Microbiol. 109 (2010) 239–247, http://dx.doi.org/10.1111/j.1365-2672.2009.04650.x. [30] J.D. Culcasi, P.R. Seré, C.I. Elsner, A.R. Di Sarli, Control of the growth of zinc–iron phases in the hot-dip galvanizing process, Surf. Coat. Technol. 122 (1999) 21–23, http://dx.doi.org/10.1016/S0257-8972(99)00404-1. [31] EN 10240, Internal and/or External Protective Coatings for Steel Tubes - Specification for Hot Dip Galvanized Coatings Applied in Automatic Plants, (1995). [32] E. Ilhan-Sungur, A. Çotuk, Microbial corrosion of galvanized steel in a simulated recirculating cooling tower system, Corros. Sci. 52 (2010) 161–171, http://dx. doi.org/10.1016/j.corsci.2009.08.049. [33] F.C. Porter, Corrosion Resistance of Zinc and Zinc Alloys, 2nd ed, Marcel Dekker Inc, New York, 1996. [34] E. McCafferty, Validation of corrosion rates measured by the Tafel extrapolation method, Corros. Sci. 47 (2005) 3202–3215, http://dx.doi.org/10.1016/j.corsci. 2005.05.046.
58
Contents lists available at ScienceDirect
Engineering Failure Analysis journal homepage: www.elsevier.com/locate/engfailanal
Corrosion failure analysis of galvanized steel pipes in a closed water cooling system
T
A. Colombo, L. Oldani, S.P. Trasatti⁎ Department of Chemistry, Università degli Studi di Milano, Via Camillo Golgi 19, 20133 Milan, Italy
AR TI CLE I NF O
AB S T R A CT
Keywords: Galvanized steel Closed cooling system Corrosion failure analysis PET preforms
The present study is focused on the corrosion failure analysis of a water cooling system serving a polyethylene terephtalate (PET) preform molding plant. Piping mainly located both after the water pretreatment and in the injection molding machines was affected by an extensive corrosion phenomenon and several leakages of coolant occurred, involving the risk of loss of functionality of the whole plant. The main problem detected in closed cooling circuits was a marked modification of the water chemistry and a great amount of metal ions leading to recurring fouling of the inner surfaces of molds and cooling channels. Analysis of water flowing in the closed cooling circuit operating at 6 °C revealed a very high amount of metals in the flowing water and in the sediment (zinc and iron). Moreover the water pH was slightly acidic. Data were compared with water from supply well. Preliminary electrochemical characterization was carried out on galvanized steel pipe, in order to evaluate the aggressiveness of water both from the cooling circuits and from the supply wells. The results of laboratory analysis together with the scrutiny of the available technical documentation showed that the damage of the plant was the consequence of an inadequate plant management, mainly for what concerns materials selection and water chemical treatment.
1. Introduction Water cooling systems are an essential part of many industrial processes. They require proper chemical treatment and preventive maintenance in order to ensure continuous plant productivity. The functionality and the service life of the cooling system equipment are provided by selecting suitable materials of construction and supply water of required quality [1,2]. Typical building materials for cooling systems include carbon steel, stainless steel, galvanized steel, copper and copper alloys. The addition of a corrosion inhibitor, sometimes coupled with a biocide, can help to prevent scale formation and corrosion in industrial water cooling systems. Chromate, nitrite and molybdate compounds are the most reliable inorganic corrosion inhibitors [3–6]. Recently, organic/inorganic mixtures have been also developed as corrosion inhibitors [7,8]. Closed cooling systems often require also the addition of antifreeze. Galvanized steel pipes are often used for water cooling systems both in industrial plants and in domestic buildings [9,10]. The extensive use depends on their good performance against corrosion, their mechanical workability and the resistance to biofouling [11,12]. The zinc layer protects steel against corrosion by two effects: a barrier effect and a galvanic protection because Zn acts as a sacrificial anode. However, galvanized steel pipes can be affected by localized and generalized corrosion that sometimes lead to the
⁎
Corresponding author. E-mail address: [email protected] (S.P. Trasatti).
http://dx.doi.org/10.1016/j.engfailanal.2017.10.008 Received 26 May 2016; Received in revised form 9 April 2017; Accepted 11 October 2017 Available online 12 October 2017 1350-6307/ © 2017 Elsevier Ltd. All rights reserved.
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
corrosion failure of water cooling systems before their expected service life [13–16]. The composition of the water heavily affects corrosion phenomena in water distribution pipes. For example, soft waters are referred as aggressive, because they tend to prevent the formation of films of calcium carbonate [17]. Corrosion can also be enhanced either by high levels of chloride and sulphate, low pH or high temperature [18,19]. The rate of corrosion decreases considerably when protective films originate on the surface of the metal. Monitoring and prevention of corrosion phenomena is of outstanding importance in order to avoid main damages like pipe break-down and water leakages, decrease in the cooling efficiency, collapse of the system, shut-down of the whole plant for maintenance services [20]. The present paper reports on the corrosion failure analysis of galvanized steel piping in a water cooling system serving a PET (polyethylene terephtalate) preform molding plant. PET preforms are the basis of very common plastic bottles typically used for drinking waters, fruit juices, soft drinks, oils and milk. PET preforms are usually produced by injection molding. The starting material are PET pellets, melted at 270 °C and injected under high pressure into a mold to make the preform. In each inner mold part there is a water cooling circuit. The temperature of cooling water is generally in the range 6–20 °C. Proper functioning of the cooling circuit is critical for ensuring stability of the end-product. If PET is exposed to temperatures above 270 °C, the polymer may thermally degrade and produce acetaldehyde. Acetaldehyde can migrate into bottled drinks and give the product an acidic taste [21–23]. Therefore the proper control of the process is essential to obtain products in compliance to customer requirements. The aim of the present study was to determine the reasons of unexpected corrosion failures detected at different locations of the whole piping cooling system. The plant under investigation is located in the center of Italy and has 20 lines of production (injection molding machines) that operate 24 h/day. Six closed cooling systems, operating at 6 °C and 20 °C, serve the injection molding machines. A schematic of the plant site is reported in Fig. 1. The plant site includes two different wells: i) Well 1, serving the cogeneration plant to recover part of the energy required for the process; ii) Well 2, providing make-up water to the closed cooling circuit in order to compensate potential leakages. Water is pumped by submerged pumps from both wells. The internal walls of the wells (30 cm diameter) are made of carbon steel. After pretreatment, water from Well 1 is conveyed to water pretreatment and cogeneration system by galvanized steel pipes. Water pretreatment is constituted by sand filter and softener. On the other hand, water from Well 2 is conveyed by polypropylene pipes to a drum, then to sand filter and softener and finally to the chiller. Within the water pretreatment, the interconnections are mainly constituted by galvanized steel, but copper, carbon steel and stainless steel are still present. From the chiller to the inlets of production lines, water flows through pipes made of AISI 304. Then the water enters the closed cooling circuits of the PET preform molds
Fig. 1. Schematic of the plant site. Arrows indicate the original location of analyzed metal samples.
47
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
that are typically made of steel or aluminum. According to the provided information, when the plant was set up, several parts of piping system were made from carbon steel. In the first stages of operation, a mixture of water and ethylene glycol was used as coolant with the addition of an inhibitor. In 2007 carbon steel was completely replaced with galvanized steel but the water/ethylene glycol/inhibitor mixture was partially maintained as coolant, without considering the ratio water/glycol, with the latter that underwent dilution over time. These changes were made to replace damaged parts of the plant. In the following years the piping system suffered with several corrosion failures at different locations where galvanized steel was present. These failures accompanied by frequent leakages of coolant demanded for continuous replacement of damaged parts and involved also the risk of severe loss of functionality of the whole plant. Nevertheless, the molding machine continued to operate without stop up to 2014. An extensive failure analysis was performed on several samples of the galvanized piping system extracted from the plant. 2. Materials and methods 2.1. Materials 2.1.1. Technical documentation Part of the failure analysis activities were devoted to get data and information about the plant management during the period 2007–2014, when corrosion problems occurred. In particular, water analysis reports issued by three water service companies were scrutinized. 2.1.2. Metal samples An unused hot-dip galvanized steel pipe was supplied by the same manufacturer of pipes used to replace damaged parts in the plant. The new pipe was used as reference material. Sections of failed parts were collected at several points of the plant: i) drum; ii) sections of pipes; iii) core element of a mold with internal cooling channel. The sampling sites are indicated by arrows in Fig. 1. 2.1.3. Cooling water samples Three water samples were obtained for chemical analysis and corrosion tests from: i) the well serving the cogeneration plant (Well 1 after softener); ii) the well supplying make-up water to closed cooling circuits (Well 2 after softener); iii) one of the closed cooling circuits operating at 6 °C. 2.2. Method 2.2.1. Visual, metallographic, SEM and XRD examination Macroscopic pictures were taken to provide documentary evidences of the state of collected samples and the distribution of corrosion products. Metallographic examination was carried out to check the microstructure of the base steel. For metallographic purposes, some of the specimens were embedded in thermoplastic resin, polished with abrasive papers down to 2000 grit and diamond pastes down to 0.25 μm. Cross-sections of some of the samples after metallographic preparation by mechanical polishing and microetching were examined by optical microscope to investigate quality and thickness of zinc layer. SEM (Scanning Electron Microscope) analyses were performed to investigate the composition and the morphology of corrosion products and to determine the thickness of zinc layer. The morphology of inner and outer surface of pipes was observed by a SEM mod. LEO 1430 equipped with an EDS microprobe (Energy Dispersive X-ray Spectrometer INCA Oxford Instruments). The elemental composition of corrosion products was determined by EDS. Before SEM analysis, all samples were coated with a very thin layer of gold by sputter Nanotech SEMPREP 2. The corrosion products were characterized by XRD (X-Ray Diffratometry) analysis with a Bruker APEX II powder diffractometer with Cu Kα radiation. 2.2.2. Water analysis A complete physico-chemical analysis was carried out on all collected water samples. pH and redox potential were determined with a Digital pH meter mod. 337 Amel. Conductivity measurements were performed with a mod. 160 Amel conductivity-meter equipped with a mod. 192/K1 Amel cell. Hardness of water was determined by titration with EDTA (following the procedure APAT IRSA CNR [24]) and results are here reported as ppm of CaCO3. Cations and anions were determined by ICP-AES (Inductively Coupled Plasma – Atomic Emission Spectroscopy) and ion-exchange chromatography respectively. The sediment in the water from cooling circuit was analyzed by ICP-AES and EDS, after preliminary filtration, drying and microwave-assisted acid digestion with HNO3. 48
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 2. Actual condition of the plant: a) wellhead; b) tank inlet; c) plant area.
2.2.3. Electrochemical tests Since corrosion is basically an electrochemical process, electrochemical tests could be considered a valuable method to gain insight into the corrosion phenomenon. Accordingly to this approach, polarization curves were acquired for an unused section of pipe using the three collected water samples as test media. Tests were carried out at 25 °C in aerated solutions under atmospheric pressure with a mod. 273A EG & G potentiostat/galvanostat. A classic three electrode configuration was adopted with a platinum wire as a counter electrode and a saturated calomel electrode (SCE) as a reference electrode. A Luggin capillary was used to minimize ohmic drops in solution. The galvanized pipe coupon was degreased with hexane, rinsed with distilled water, dried in air and then placed at the bottom of the cell, with an exposed area of 1 cm2. Before each experiment the system was allowed to stabilize at the open circuit potential (OCP) for 30 min. Cyclic polarizations were performed at a scan rate of 10 mV/min. The polarizations curves were recorded in the anodic direction from the OCP until the current density reached 1 mA/cm2, and then reversed. 3. Results and discussion 3.1. On-site inspection The plant was visited twice at the beginning of 2014 for visual inspection of the condition of the piping system. The inspections were aimed at evaluating the extent of corrosion damages by speaking with operators about the process phases and collecting samples for laboratory analysis. Sample collection concerned both external piping (from the wells to the production area) and internal closed cooling circuits. 3.1.1. External piping Fig. 2 shows the condition of the plant: extensive corrosion phenomena affect all the metallic parts possibly in contact with the circulating coolant from wellhead (Fig. 2a) to the chiller (Fig. 2b), up to the production area (Fig. 2c). Most of the plant equipment is affected by severe corrosion. Leakages of coolant occurred at several points and were detected at different times. Corrosion was mainly in the form of uniform corrosion, though some localized attacks were detected on stainless steel pipes. Persistent rusty deposits were also visible on the facilities and on the ground in all the plant. Following the inspection, some problems about the construction and the management of the plant can be supposed. In particular, different metallic materials were not correctly coupled (galvanized steel coupled to stainless steel), possibly leading to galvanic corrosion phenomena. Talking with plant operators, it emerged that the proper functioning of water pretreatment had never been properly checked up. 3.1.2. Internal closed cooling circuits Fig. 3 shows the connection between the external piping and the coolant inlet (Fig. 3a) into one of the molding injection machine (Fig. 3b). Figs. 4 and 5 show the core part of one mold and the relative cooling channel, respectively. Molds are periodically taken apart for cleaning operation. The internal surfaces of all the molds are always covered by very compact reddish deposits that are usually mechanically removed by ultrasonic cleaning in water. These very adherent rusty deposits are a real concern for the plant manager, since the very high cost of each mold (approximately 500.000 Euros) completely falls on the PET preform company and not on the buyer. This aspect justifies the need that the plant works 24 h a day and the impossibility to schedule stops for maintenance operations. So ordinary maintenance would be highly desirable but it is unfeasible. In spite of the condition of external piping and closed cooling circuits, the quality of produced PET preforms never gets worse. 3.1.3. Analysis of technical documentation The approach to this problem followed a typical corrosion failure procedure, with the preliminary collection of plant service history and background data. The company never adopted proper maintenance actions, such as scheduled inspections, specifically devoted water pretreatment, and adoption of procedure for cleaning piping. Moreover, from meeting with technical staff it was inferred that careful selection of the materials has never been a priority for the management. When severe leakages of coolant happened, damaged pipes were simply 49
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 3. Actual condition of the plant: a) connections between external piping and internal closed cooling circuit; b) injection molding machine inlet.
Fig. 4. Accessory for PET pre-form mold made of high grade steel (surface-hardening heat treatment): a) core part; b) connection to the injection molding machine.
Fig. 5. Accessory for PET pre-form mold made of stainless steel: a) internal cooling channel; b) detail.
Table 1 Water analysis in external piping system. Well 1
pH Conductivity (μS/cm) Hardness (ppm CaCO3) Redox potential vs. Pt (mV) Cl− (mg/l) Fe2+ (mg/l) Zn2+ (mg/l) Cu (mg/l) Microbial count at 20 °C (UFC/1 ml) Microbial count at 37 °C (UFC/1 ml)
Well 2
After softener
Sept 4th 2013
Oct 1st 2013
Sept 4th 2013
Oct 1st 2013
Sept 12th 2013
7.7 480 276 – 30.2 0.01 0 0.43 – –
7.02 649 300 − 216 51.9 7100 – – 400 260
7.8 507 316 23.5 29.8 0.02 0.1 0.9 – –
6.79 753 375 86 – 230 – – 670 270
7.3 590 0 – 16.6 0 – – – –
50
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Table 2 Water analysis in the six internal closed cooling circuits: average values and standard deviations.
pH Conductivity (μS/cm) Hardness (ppm CaCO3) Redox potential vs. Pt (mV) Cl− (mg/l) Fe2+ (mg/l) Zn2+ (mg/l) Cu (mg/l)
Sept 4th 2013
Sept 12th 2013
6.4 ± 0.2 1222 ± 462 428 ± 78 – 205 ± 66 122 ± 63 52 ± 29 5.6 ± 4.3
6.0 ± 0.2 1355 ± 229 180 ± 36 – 204 ± 58 122 ± 63 – –
replaced by new ones. Since the problem has been dragging along for a long time and the plant could not be stopped, the company only turned to independent water service companies. These societies only collected water samples at different locations of the plant. Results are reported in Tables 1 and 2, in term of several physico-chemical parameters. The focus is on the last part of 2013. Careful scrutiny of the water analysis reports revealed lacking and/or conflicting data on pH and chlorides as well as iron content. As data of October 2013 assessed (Table 1), water from wells showed a marked microbial contamination by iron bacteria and a highly negative redox potential that can be related to the corrosion damages of galvanized steel pipes [25]. Microbiologically influenced corrosion (MIC) can potentially contribute to failures in galvanized steel systems containing water. MIC has been suggested to induce failures as leaks through pitting corrosion [26–28]. Management strategies should consider microbial control to prevent corrosion in these systems. However, corrosion by iron bacteria only occurs if the internal zinc layer is damaged [29]. For what concerns the water flowing through the internal closed cooling system (Table 2), analysis revealed very high amount of iron, copper and zinc ions in all the six circuits, probably released from pipes. Chloride levels should also warn, since they could be considered one of the reasons of initial corrosion. Unlike wells water, conductivity is very high and pH is slightly acidic. At the initial stage of the production process, the cooling water was added with ethylene glycol whose concentration was never monitored. The natural degradation of glycol could have led to lowering of pH, thus promoting the beginning and propagation of corrosion.
3.1.4. Preliminary remarks At the end of the on-site inspection and after the examination of technical documentation, some critical factors were identified as follows: - the observed uniform corrosion could be attributed to galvanic coupling of the several alloys used for pipes of closed circuit (copper, galvanized steel, AISI304) and molds (steel, aluminum); - the presence of unexpected levels of chlorides suggested that the water softener did not operate properly over time, thus possibly promoting their release in the coolant; - ethylene glycol, formerly used as coolant fluid, was never monitored even though piping was moved from carbon steel to galvanized steel. Eventual decomposition of glycol to organic acids could have caused the observed increase of acidity. Likely all these factors contributed to the failure of the system. However it was very difficult to draw conclusions only on the basis of technical report scrutiny and visual inspection. To go deeper into the approach to the problem, the failure analysis of metal materials from the plant was carried out in order to evaluate corrosion morphology, corrosion products and zinc layer thickness. A complete investigation required also a detailed water analysis.
3.2. Metallographic analysis and SEM examination 3.2.1. New galvanized steel pipe An unused hot-dip galvanized steel pipe supplied by the manufacturer was taken as reference material (Fig. 6a). The internal diameter is 2.68 cm. From metallographic examination, the substrate shows a typical ferrite-pearlite microstructure of low carbon steel (Fig. 6b). The quality of zinc layer was evaluated by metallographic and SEM examination. The zinc layer is uniform and well-adhered throughout the sections of examined pipes. Usually, a hot-dip galvanized coating consists of several layers. Starting from the steel surface, each layer is an iron–zinc alloy with increasingly lower iron content [30]. SEM images of cross sections in Fig. 7 show the typical morphology of a zinc coated steel, consisting of the coating alloy, an interfacial layer between the coating and the substrate steel (containing a series of intermetallic compounds), and the substrate steel [11]. The European Standard EN 10240 [31] states that a minimum of 55 μm of zinc in the inner side of galvanized steel pipes is required for good protection of steel against uniform corrosion in fresh water. The new pipe exhibits a 60 ± 5 μm zinc layer on the outer side (Fig. 7a) and 53 ± 5 μm zinc layer in the inner side (Fig. 7b). The average thickness of the coating layer is slightly lower than the minimum value specified in the European standard. 51
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 6. Unused galvanized steel pipe: a) section; b) microstructure.
Fig. 7. Cross section of new pipe: a) external surface; b) internal surface.
3.2.2. Damaged metal samples The inner surface of the collected samples had significant material loss from the zinc layer and the base material. The material loss was in the form of general corrosion. In Fig. 8, a coupon of a galvanized steel drum located before the water pretreatment is shown. The typical reddish color of rust is clearly visible on both inner and outer sides. In Fig. 9 the micrograph of the cross section and the EDX analysis performed on the inner side of the drum coupon are shown. The metal underwent severe generalized corrosion with complete depletion of the zinc layer. Sections of failed pipes are shown in Figs. 10 and 11, respectively. Sample in Fig. 10 has been collected from the piping conveying water from Well 1 to cogeneration. Sample in Fig. 11 belonged to piping system within the water pretreatment downstream of Well 2. At a preliminary visual inspection, the outer surface of in-service pipes does not show evident damages, except in the threaded areas. On the other hand, inner surfaces looked severely attacked by corrosion. They have a very compact layer of reddish corrosion products that completely cover the surface. Upon some mechanical removal of corrosion products, corrosion damages of the
Fig. 8. Galvanized steel drum: a) inner side, b) outer side.
52
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 9. Galvanized steel drum: cross section end EDX analysis of the inner side.
Fig. 10. Example of damaged pipe from the piping conveying water from Well 1: a) external view; b) internal view.
Fig. 11. Inner side of failed pipes from water pretreatment piping downstream of Well 2, showing corrosion damages under corrosion products.
underlying steel are also evident. In some cases localized corrosion attack caused the formation of pitting holes in the areas of welding. The morphology of corrosion damages in the welding areas could be related to microbial induced corrosion (Fig. 11b). This aspect is in agreement with the detection of iron bacteria in the water analysis performed on October 2013. The inner surfaces of these failed pipes were characterized by SEM and EDS to check the morphology of corrosion and to determine the elemental composition of corrosion products (Fig. 12). Different layers of corrosion products are clearly visible, pointing to an extensive corrosion phenomenon that started at different times. Chemical analysis showed that deposits consist mainly of iron oxide (Fig. 12b). The degree of oxidation is higher in the upper layer (labeled SP1). Some traces of zinc were still detected, but the amount is about 10%. Other elements were detected in trace levels, namely Si and Al. Higher magnification of the lowest layer SP4 revealed that corrosion products are constituted by particles different in size and morphology (Fig. 12c). The relative elemental composition is also reported and shows that flower-like particles are in higher degree of oxidation than smaller particles. Smaller aggregates displaying acicular features are the prevailing morphology, and they could related to eventual MIC [32]. The presence of manganese, which is an alloying element of steel, was also detected. 53
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 12. a) SEM image of inner surface of damaged galvanized steel pipe (SP indicates the spot site where EDS analysis was performed); b) EDX analysis at different sites; c) higher magnification of SP4 with relative EDS analysis.
XRD analysis of corrosion products near the perforation area is reported in Fig. 13. Lepidocrocite (γ-FeO(OH)) and goethite (αFeO(OH)) are the main constituent. Since no evidences of sulphides were detected, the hypothesis of MIC is not actually supported by experimental data. The micrographs of the cross sections of the failed pipes as well as SEM images (Fig. 14) show that the state of zinc layer is completely different on the two sides of failed pipes. On the outer side (Fig. 14a) there is a fully-dense and continuous layer of zinc, with a thickness of 78 ± 5 μm (Fig. 14c). The inner surface instead is characterized by several directional pits (Fig. 14b). This could be the result of erosion-corrosion of the underlying steel directly exposed to the water flowing into the pipes, as a consequence of high amounts of sediments. SEM image also confirms the complete depletion of zinc layer (Fig. 14d). 3.3. Water analysis A detailed analysis of the water from the wells and from the circuit was performed in order to elucidate the environment that damaged pipes experienced. The water from the wells is completely clear and transparent and it is usually not subjected to any preliminary treatment, except for sand filter and softener. On the other hand, the liquid flowing into closed cooling circuits contains a great amount of slurry reddish sediment.
Fig. 13. XRD analysis of corrosion products from damaged galvanized steel pipe.
54
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 14. Cross sections of failed pipes: micrograph of external surface (a) and internal surface (b); SEM images of external surface (c) and internal surface (d).
Physico-chemical parameters determined in the three water samples collected during the on-site inspection are listed in Table 3. Water from both wells has a pH value near 7, while the water circulating in the closed cooling circuits is acidic (pH 4.8). In the absence of reducing or passivating agents, the corrosion of zinc in aqueous solution is primarily determined by pH values and zinc readily dissolves in acidic or strongly alkaline solutions [10,33]. Conductivity in cooling circuit is nearly the double than in Well 2, indicating that a great amount of salts are therein dissolved. Water hardness is another important variable in zinc corrosion. The content in calcium carbonate for Well 2 is not as low as expected if softener had worked properly. However, hard waters tend to be less corrosive toward zinc because they deposit a protective scale on the surface of the metal. The redox potentials indicate that the water under investigation is normally oxygenated at the time of analysis. Tables 4 and 5 report the amount of anions and cations determined in the three water samples. The amount of chlorides in the cooling circuit is lower than in well water and is very lower compared to September 2013 (Table 2). This is probably due to the introduction of make-up water of different quality at different times. For what concerns the cations, at the time of analysis the quality of sampled well waters is consistent with that of typical subsurface waters from wells. The water from cooling circuit instead exhibits a very high levels of zinc (100 ppm), indicating that a great amount of metal dissolved from inner surface of pipes and passed into solution. A significant amount of iron (43 ppm) was also detected in solution. The reddish sediment is mainly constituted of iron (435 ppm). This finding was confirmed by SEM-EDS analysis of the dried sediment (Fig. 15a) revealing that it is mainly constituted by amorphous iron oxides (Fig. 15b). Since there is any trace of iron in the well water, but there is in the cooling circuit, therefore the source of the iron must be dissolved ions formed during the corrosion process. XRD analysis (Fig. 15c) further confirmed that sediment is mainly constituted by iron oxy-hydroxides. 3.4. Electrochemical tests Electrochemical polarization experiments were performed on a new galvanized pipe to investigate the aggressiveness of the water Table 3 Water analysis: physico-chemical parameters.
pH Conductivity (μS/cm) Hardness (ppm CaCO3) Redox potential vs. C (mV) Redox potential vs. Pt (mV)
Well 1 after softener
Well 2 after softener
Cooling circuit (6 °C)
7.5 762.8 5 0.2274 0.0175
7.3 749.7 103 0.2438 0.0131
4.8 1333 167 0.1765 0.0363
55
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Table 4 Water analysis: anions.
−
Cl (mg/l) SO42 − (mg SO4/l) PO43 − (mg P/l) NO3− (mg N/l)
Well 1 after softener
Well 2 after softener
Cooling circuit (6 °C)
45.93 23.14 0.142 n.d.
45.57 23.11 0.119 n.d.
19.10 18.25 n.d. 0.15
Table 5 Water analysis: cations as determined by ICP-AES (ppm).
Cu Ni Zn Cr Mn Al Fe Na Ca
Well 1 after softener
Well 2 after softener
Cooling circuit (6 °C)
0.005 < 0.005 0.007 < 0.005 < 0.005 0.005 0.037 149 0.62
< 0.005 0.007 0.05 < 0.005 0.03 0.02 0.32 120 30
0.046 0.78 100 0.006 30 0.011 43 146 68
Fig. 15. Reddish sediment in water from cooling circuit: a) SEM image; b) EDS analysis, c) XRD analysis.
samples collected from the plant toward the pipes usually installed in the plant. The results are reported in Fig. 16. The galvanized steel exhibits quite similar trends in all environments. There is a sharp increase of the current and there is no passive region. The increasing potential leads to the active dissolution of zinc that continues until the zinc surface is covered by a layer of corrosion products that do not originate a passive layer. This is the typical behaviour of an active material. However, by the determination of the corrosion current Icorr at the corrosion potential Ecorr, it is possible to draw some general conclusion about the behaviour of zinc layer into the three different solutions under investigation. Icorr was determined by Tafel extrapolation from the anodic branch of the polarization curve, assuming that the cathodic process does not change [34]. The so determined Icorr were 0.112 mA/cm2 for Well 2, 0.117 mA/cm2 for Well 1 and 0.144 mA/cm2 for closed cooling circuit, respectively. The corrosion current is slightly higher for water from Well 1 than water from Well 2, most likely because the water from Well 2 has a higher concentration of carbonates (e.g., is harder). Corrosion occurs more slowly in harder water due to the deposition of protective layers. It is possible also to observe that the OCP of galvanized steel is higher (more noble) in well waters than in water from closed 56
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
Fig. 16. Cyclic polarization curves of a new galvanized steel pipe in the three different water samples.
cooling circuit. Moreover, the highest corrosion rate was determined for water from the closed cooling circuit, even though its hardness was similar to the water of Well 2 and chloride contents is about half that of water from wells. Since the corrosion rate of zinc is pH dependent, the higher aggressiveness of water from closed circuit can be related to the lower pH of the solution (pH 4.8). On the basis of metallographic analysis and electrochemical test, a pH-related corrosion mechanism can thus be proposed to explain the corrosion failure of galvanized steel piping. 4. Conclusions In the present work an extensive corrosion failure analysis was carried on to elucidate the reason of the premature failure of the galvanized steel piping of a plant for PET preforms. On the basis of the plant history and of the experimental results, it is possible to state that: - an inadequate approach to maintenance problem and material selection was observed; - high levels of hardness after the softener and the variable amount of chlorides suggest that the softener does not work properly; - galvanized steel pipes employed for the closed cooling circuit are in compliance with the European Standard EN 10240 for what concerns quality and thickness of zinc layer: no material anomalies were observed that would have caused the failure; - a relatively low pH (possibly due to degradation reactions of glycol) and erosion-corrosion phenomena by solid deposits are the most probable factors for the failure of the pipes before their expected service lifetime; - fast corrosion of zinc exposed the underlying steel to the aggressive attack of acidic solution, causing extensive corrosion damages. 5. Recommendations Possible solutions to improve plant performance and to prevent future leakages from cooling systems are: -
to to to to
conduct a proper materials selection and design to avoid risk of unwanted galvanic coupling; monitor the quality of the water from the wells and the cooling system and to set-up an eventual proper water pretreatment; implement a scheduled maintenance of the whole plant; design a cleaning procedure to keep under control the amount of suspended solid.
Acknowledgments This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. The authors thank Dr. Edoardo Guerrini for SEM analyses, and Mr. Giorgio Termignone for metallographic analysis of metal samples. References [1] [2] [3] [4] [5] [6]
S. Ahmed, J. Bayers, Design basics for cooling systems, Chem. Eng. (New York). 107 (2000) 78–83. J.W. McCoy, The Chemical Treatment of Cooling Water, 2nd ed, Chemical Publishing Company, New York, 1983. A.A.M.D.B. Alexander, Evaluation of corrosion inhibitors for component cooling water systems. Corrosion, Corrosion 49 (1993) 921–928. D.E.A. Marshall, B. Greaves, Low nitrite inhibitor treatment for cooling systems, Mater. Perform. 5 (1986) 45–50. J.P.G.F.M.S. Vukasovich, Molibdate in corrosion inhibition – a review, Mater. Perform. 25 (1986) 9–18. G.O. Ilevbare, G.T. Burstein, The inhibition of pitting corrosion of stainless steels by chromate and molybdate ions, Corros. Sci. 45 (2003) 1545–1569, http://dx.
57
Engineering Failure Analysis 84 (2018) 46–58
A. Colombo et al.
doi.org/10.1016/S0010-938X(02)00229-9. [7] A.A. Moccari, Corrosion inhibitor evaluation for materials used in closed cooling water systems, Mater. Perform. 38 (1999) 54–59. [8] N. Ochoa, G. Baril, F. Moran, N. Pébère, Study of the properties of a multi-component inhibitor used for water treatment in cooling circuits, J. Appl. Electrochem. 32 (2002) 497–504 (doi:10.1023/A:1016500722497). [9] Internal corrosion of water distribution systems, AWWRF-DVGW-TZW Cooperative Research Report, Denver, CO, 1996. [10] P.G. Rahrig, Galvanized steel in water and water infrastructure, Mater. Perform. 42 (2003) 58–60. [11] A.R. Marder, The metallurgy of zinc-coated steel, Prog. Mater. Sci. 45 (2000) 191–271. [12] X.G. Zhang, Corrosion and Electrochemistry of Zinc, 1st ed, Plenum Press, New York, 1996. [13] B. Minnoş, E. Ilhan-Sungur, A. Çotuk, N.D. Güngör, N. Cansever, The corrosion behaviour of galvanized steel in cooling tower water containing a biocide and a corrosion inhibitor, Biofouling 29 (2013) 223–235, http://dx.doi.org/10.1080/08927014.2012.763117. [14] D.W.E. Jr, E. William, An investigation of zinc oxide corrosion of galvanized steel in cooling towers, ASHRAE Trans. 91 (2003) 91–100. [15] J.J.R.E. Otero, The assessment of the short term data of pipe corrosion in drinking water. I. Galvanized steel, Corros. Sci. 34 (1993) 1581–1593. [16] C.A. Della Rovere, R. Silva, C. Moretti, S.E. Kuri, Corrosion failure analysis of galvanized steel pipes in a water irrigation system, Eng. Fail. Anal. 33 (2013) 381–386, http://dx.doi.org/10.1016/j.engfailanal.2013.06.024. [17] F. Delaunois, F. Tosar, V. Vitry, Corrosion behaviour and biocorrosion of galvanized steel water distribution systems, Bioelectrochemistry 97 (2014) 110–119, http://dx.doi.org/10.1016/j.bioelechem.2014.01.003. [18] J. Román, R. Vera, M. Bagnara, A.M. Carvajal, W. Aperador, Effect of chloride ions on the corrosion of galvanized steel embedded in concrete prepared with cements of different composition, Int. J. Electrochem. Sci. 9 (2014) 580–592. [19] V. Padilla, A. Alfantazi, Corrosion film breakdown of galvanized steel in sulphate–chloride solutions, Constr. Build. Mater. 66 (2014) 447–457, http://dx.doi. org/10.1016/j.conbuildmat.2014.05.053. [20] B. Hickinbottom, Why does operational management put corrosion prevention and the monitoring of the onset of corrosion at the lowest level of priority and how to overcome this attitude, Proc. - Corros., Australasian Corrosion Association, 2000, pp. 50–57. [21] F. Villain, J. Coudane, M. Vert, Thermal degradation of polyethylene terephthalate: study of polymer stabilization, Polym. Degrad. Stab. 49 (1995) 393–397, http://dx.doi.org/10.1016/0141-3910(95)00121-2. [22] G.W. Halek, The zero-order kinetics of acetaldehyde thermal generation from polyethylene terephthalate, J. Polym. Sci. Polym. Symp. 74 (2007) 83–92, http:// dx.doi.org/10.1002/polc.5070740110. [23] T.Y. Kim, S.A. Jabarin, Solid-state polymerization of poly(ethylene terephthalate). III. Thermal stabilities in terms of the vinyl ester end group and acetaldehyde, J. Appl. Polym. Sci. 89 (2003) 228–237, http://dx.doi.org/10.1002/app.11905. [24] Metodi analitici per le acque, APAT IRSA-CNR, (2003). [25] Roscoe Moss Company, Handbook of Ground Water Development, Wiley, 1990. [26] S.S. Carl, Remediation of corrosion problems inside fire sprinkler system piping, in: Corros. – Annu. Conf. Natl. Assoc, Corros. Eng. (2000) 653. [27] R.J.N. Fernance, P.A. Farinha, SRB-assisted MIC of fire sprinkler piping, Mater. Perform. 43 (2007) 46–49. [28] A.C.E. Ilhan-Sungur, Microbial corrosion of galvanized steel in a simulated recirculating cooling tower system, Corros. Sci. 52 (2010) 161–171, http://dx.doi. org/10.1016/j.corsci.2009.08.049. [29] N. Bolton, M. Critchley, R. Fabien, N. Cromar, H. Fallowfield, Microbially influenced corrosion of galvanized steel pipes in aerobic water systems, J. Appl. Microbiol. 109 (2010) 239–247, http://dx.doi.org/10.1111/j.1365-2672.2009.04650.x. [30] J.D. Culcasi, P.R. Seré, C.I. Elsner, A.R. Di Sarli, Control of the growth of zinc–iron phases in the hot-dip galvanizing process, Surf. Coat. Technol. 122 (1999) 21–23, http://dx.doi.org/10.1016/S0257-8972(99)00404-1. [31] EN 10240, Internal and/or External Protective Coatings for Steel Tubes - Specification for Hot Dip Galvanized Coatings Applied in Automatic Plants, (1995). [32] E. Ilhan-Sungur, A. Çotuk, Microbial corrosion of galvanized steel in a simulated recirculating cooling tower system, Corros. Sci. 52 (2010) 161–171, http://dx. doi.org/10.1016/j.corsci.2009.08.049. [33] F.C. Porter, Corrosion Resistance of Zinc and Zinc Alloys, 2nd ed, Marcel Dekker Inc, New York, 1996. [34] E. McCafferty, Validation of corrosion rates measured by the Tafel extrapolation method, Corros. Sci. 47 (2005) 3202–3215, http://dx.doi.org/10.1016/j.corsci. 2005.05.046.
58