303 Fig. 2. Crevice corrosion suffered by the 316L stainless steel pump. Small crystals of sodium chloride can be seen on and around the corroded area. corresponding volumes would be 204.1 and 0.23 1, respectively. There is a slight (- 0.3%) increase in volume on mixing. Data for the solubility of sodium chloride in water-butanone mixtures could not be found. However, there are data for the solubility of sodium chloride in water-acetone mixtures at 20 "C [3]. These are given in Table 2. The 'lower' and 'upper' layers in Table 2 refer to the water- and acetone-rich solutions, respec- tively. Comparison of the data in Table 2 and the water-butanone phase diagram shown in Fig. 1 shows that the solubility of acetone in water is roughly comparable to that of butanone in water, and that water is more soluble in acetone than in butanone. Thus, the solubility of sodium chloride in water-acetone mixtures should only be considered as a guide to what might happen in water- butanone mixtures. Although the lower and upper layers in the immiscible water-acetone mixtures are in equilibrium, and the activity of the sodium chloride in each layer is therefore equal, inspection of Table 2 shows that in acetone-water mixtures the partition coefficient for sodium chloride between the water-rich lower layer and acetone-rich upper layer is 40: 1. If in the present case of a two-liquid water-butanone mixture, the partition of sodium chloride between the water-rich phase and the butanone-rich phase is of a similar magnitude, calculations indicate that the concentration of sodium chloride in the water-rich phase will be higher than in both the butanone-rich phase and the original 8 wt% solution of distilled water in butanone. Table 2. Solubility of sodium chloride in aqueous solutions of acetone at 20 "C Weight % acetone 8.0 16.5 25.3 27.1 84.1 85.3 87.7 Lower layer Upper layer E NaCl Der 100 cm3 of solution 27.18 23.10 19.32 18.05 0.45 0.43 0.25 304 'l'he weight of sodium chloride in the 168 kg of the liquor used in the batch process is 0.084 kg. With a partition coefficient of 40:1, this amount of sodium chloride would be partitioned thus: the water-rich layer would contain 0.082 kg, and the butanone-rich layer would contain 0.002 kg. The calculations given previously show that addition of the minimum volume of water required to cause separation of the 8% water-in-butanone solution used in the batch process into two immiscible liquids results in the formation of 0.23 1 of the water-rich phase. Using the data shown in Table 2 as a guide, the greatest amount of sodium chloride which could dissolve in 0.231 of water-rich solution is 0.045 kg. Thus, if only 0.23 1 of the water-rich phase had been formed, it is probable that sodium chloride would have precipitated from the water-rich phase. Since no solids had been observed when the system was dismantled, it appears that at least 0.5 kg of water-rich phase must have been formed. This led to a reappraisal of the amount of water which had been introduced to the system. Calculations based on the Lever rule indicated that this would have required the addition of just under 4 kg of water to the feed liquor. Although the strict applicability of these calculations to the present case may be questioned, since data for the solubility of sodium chloride in water-acetone mixtures have been used, they do suggest that the hypothesis is tenable. The result would have been that the pump which had suffered the crevice corrosion was not exposed to a very dilute solution of chloride but to a brine, and crevice corrosion of 3 16L would inevitably have occurred. 5. A DEMONSTRATION Since there were doubts about the applicability of these calculations to the present case, a demonstration was carried out in which 5 mm nominal diameter 3 16L stainless steel rods were exposed to: (i) an 8 wt%a solution of water in butanone which contained 0.05 wt% sodium chloride, and (ii) the same solution after the addition of just sufficient water to cause the formation of two liquids. Artificial crevices were formed by slipping Viton O-rings up the rod. In the second case, the O-rings were positioned so that there was one in the water-rich phase and one in the butanone-rich phase. The composition of the stainless steel used in this demonstration is given in Table 3. After 12 days of exposure at 25 "C, a pale yellow discolouration of the originally clear water-rich phase was observed, and a light ring of rust began to appear at the edge of the O-ring immersed in this phase (Fig. 3). Neither discolouration of the solutions nor rings of rust at the O-rings were Fig. 3. Light ring of rust which appeared at the O-ring immersed in the water-rich phase after 12 days of exposure at 25 "C ( x 16) 305 Table 3. Chemical analysis of the stainless steel rod used in the demonstration: composition in wt% C Mn Si P S Cr Ni Mo Rods 0.024 1.55 0.59 0.030 0.023 16.3 11.1 2.02 UNS S31603 0.03 2.00 1 .oo 0.045 0.030 16.0-18.0 10.0-14.0 2.00-3.00 maximum maximum maximum maximum maximum observed in either the butanone-rich phase or the original 8 wt% water-in-butanone mixture. These observations show that crevice corrosion had just begun to initiate in the water-rich phase, but had not initiated in either the 8wt% solution of water in butanone, or in the butanone-rich phase formed after the addition of water. 6. CONCLUSION The hypothesis presented in the discussion is tenable, and can explain how crevice corrosion of 316L occurred in what was supposed to be a very dilute solution of sodium chloride in an 8 wt% solution of water in butanone. The unresolved question was how the additional water required to cause separation of this solution into two immiscible liquids was introduced into the system. REFERENCES I. Francis, A. W., LiquicCLiquid Equilibriums. Interscience, New York, 1963. 2. Seidell, A,, Solubilities oforganic Compounds, 3rd edn. Van Nostrand, Princeton, NJ, 1941. 3. Seidell, A., Solubilities oflnorganic Compounds, 3rd edn. Van Nostrand, Princeton, NJ, 1941. Failure Analysis Case Studies II D.R.H. Jones (Editor) 0 2001 Elsevier Science Ltd. All rights reserved 307 TYPE I PITTING OF COPPER TUBES FROM A WATER DISTRIBUTION SYSTEM PAUL0 J. L. FERNANDES Advanced Engineering and Testing Services, MATTEK, CSIR, Private Bag X28. Auckland Park, 2006, South Africa (Received 9 Augusf 1997) Abstract-Samples of copper tubes from a cold water distribution system which had failed due to pitting whilst in service were subjected to a detailed failure investigation. Analysis of the tubes showed that failure was a result of Type I pitting attack. While the exact cause of pitting was unknown, it was hypothesised that it could have been due to changes in the water quality and/or content. The tubes were found to be made from phosphorus de-oxidised copper and no anomalies were evident in either the chemical composition or the microstructure which could have caused the pitting observed. It was recommended that the tubes be replaced and that due attention be given to ensure that the new tubes are free of internal carbonaceous deposits or other foreign matter. 0 1998 Elsevier Science Ltd. All rights reserved. Keywords Corrosion, pitting corrosion. 1. INTRODUCTION Copper tubes are used extensively in water distribution systems due to their corrosion resistance and ease of installation. In Europe and North America they account for more than 80% of all tubes installed in water service [l], amounting to over 100 million metres of tubing. In spite of these large quantities, tube failures are relatively rare. Of the failures that do occur, pitting corrosion accounts for approximately 60%. This study presents an investigation of the failure of copper tubes from a cold water distribution system carrying potable water in a shopping centre. The tubes, which were built into the brick walls, sprang leaks in several premises in the shopping centre after approximately 12 years’ seMce, causing severe staining of the walls. Examination of the tubes revealed the presence of pin holes perforating the tube walls. 2 EXPERIMENTAL PROCEDURE 2.1. Visual examination Several tubes sections were received for analysis. These were sectioned to reveal the internal surfaces, which were found to be covered with a greenish-white scale (Fig. 1). Furthermore, localized deposits of green corrosion product in the form of tubercules were also evident (see arrow in Fig. I). Some tubercules were carefully removed by light scrubbing to reveal the underlying metal. A shiny, black layer of an unidentified compound was found to exist beneath the greenish-white scale. Beneath this black layer, in turn, pits penetrating into the tube wall were found. An example of the various layers and the underlying corrosion pit is shown in Fig. 2. Some of the pits observed were relatively large and deep, as shown in Fig. 3. 2.2. Chemical analysis of internal scale and corrosion products Samples of the tubes were examined in a scanning electron microscope (SEM) equipped with an energy dispersive spectroscopy of X-rays (EDS) facility. The results of the EDS analysis of the greenish-white scale found on the internal surfaces of the tubes are shown in Fig. 4. The large copper Reprinted from Engineering Failure AnaZysis 5 (l), 35-40 (1 998) 80E 309 Fig. 3. The internal surface of a tube showing extensive pitting (x 3) Fig. 4. The EDS results of the greenish-white deposits found on the internal surfaces of the tubes. 3. METALLOGRAPHY Samples from the tubes examined were prepared for metallographic analysis using standard grinding and polishing techniques. Etching was carried out in acidified ferric chloride. The typical microstructure observed in all cases consisted of large equi-axed grains, indicating that the tubes were in the annealed condition. 310 3.1. Chemical anaZysis An analysis of the chemical composition of the tubes was carried out using a wet chemical analysis method. From the high phosphorus content it was evident that the tubes were made from phosphorus de-oxidised copper. 4. TYPE 1 PITTING Pitting corrosion is the most common failure mechanism for copper tubes in water distribution systems. Essentially two different types of pitting attack have been identified, and these are referred to in the literature as Type I and Type I1 pitting*. The former is known as cold water pitting and occurs more frequently than the latter. Type I pitting is usually encountered in cold water systems carrying borehole or well waters free from organic matter [I]. It occurs sporadically and can result in tube wall penetration within a few months. In some cases, however, penetration occurs only after 15 years or more. The internal surfaces of tubes undergoing Type I pitting are usually covered with a greenish scale of a copper compound called malachite. Beneath this scale, the tube surface is covered with a smooth, shiny layer of dark cuprite which is very friable and easily spalled off. Pits are usually associated with the presence of tubercules which form over pin hole defects in the cuprite layer. The characteristics of Type 1 pitting attack are such that many pits at all stages of development can usually be found [I]. Larger pits are generally linearly arranged along the bottom half of horizontal water lines. When pits are very close together, tubercules can extend over a number of pits to form one long tubercule. Although pitting has been observed in annealed, half-hard and hard-drawn tube, susceptibility is generally greatest in the annealed condition. The pits formed are usually saucer-shaped and relatively wide. A number of causes of Type I pitting have been identified [I]. Firstly, the incidence of pitting has been associated with the presence of carbonaceous films on the internal surface of the tube. These films are residues of the lubricant used for the drawing operation and which are carbonized during annealing. The quantity and distribution of these films on the internal surface appears to affect the severity of pitting. The problems arising from the presence of these carbonaceous films can be overcome in practice by scouring the tubes with a water-sand or a water-air blast. Secondly, pitting has been associated with the presence of foreign matter deposits on the bottom half of horizontal tubes [l]. This is in agreement with observations on the preferential location of pits discussed above. The foreign matter deposits can be introduced into the water lines in a number of ways. Metal chips and filings and dirt can be allowed to contaminate the system during installation. If these are not properly removed before service, they may deposit along sections of the water lines where the water velocity is low. Foreign matter deposits may also be introduced into the system in the water or may be due to corrosion products formed during surface corrosion of the tubes during service. The concentration of these deposits, and hence their deleterious effects, can be reduced by the installation of filters in the water line. Thirdly, another factor said to cause pitting attack is the presence of soldering pastes on the insides of the tubes. This generally results from bad workmanship and can be avoided by ensuring that adequate quality standards are maintained during installation. The soldering pastes may act as deposits in the same way as foreign matter. Alternatively, during soldering or brazing these pastes may be converted to oxides which form as a thin film on the copper surface. These oxides are gcncrally cathodic to copper and can therefore give rise to pitting corrosion. The effect of water quality on the incidence of Type I pitting is the subject of some controversy and no consensus has been reached in this regard. Some general observations have been made, however, on the effects of various constituents and characteristics of water on the extent of pitting, *Some researchers have also reported the existence of Type 111 and Type IV pitting, but these appear to be variations of Type I pitting [I]. 311 +Yes -No Is the ratio of aggressive C02 to total C02 above 0.05? 1 + Yes + yes +No Is the pH in the range 6.8 - 7.51 4 NO Is the ratio of sodium to nitrate greater than I? Table 1. The effect of various water constituents and characteristics on Type I pitting Chemical species Effect Sulphate (SO:-) Chloride (Cl-) Nitrate (NO;) Inhibits pitting PH Dissolved oxygen (0,) Carbon dioxide (CO,) Assists pit initiation and growth, but its effect depends on the concentration of other chemical species. Essential for pitting attack. Assists the breakdown of protective surface films and results in the formation of wide, shallow pits Increases in pH generally decreasing the probability of pitting. Increased 0, content increases the probability of pitting. Increased CO, content increases the probability of pitting due to a decrease in pH. and these are summarised in Table 1. An empirical screening process has also been developed to assess the risk of Type I pitting in various waters [2] (Fig. 5). This process has been used extensively with reasonable success. A number of characteristics of Type I pitting discussed above were evident in the failed copper tubes from the shopping centre. The presence of tubercules of corrosion product and the greenish scale on the internal surface of the tubes were clearly evident (Fig. 1). The friable underlying layer of shiny, dark cuprite was also observed (Fig. 2). The wide, saucer-shaped pits and their approxi- mately linear distribution were also evident and are shown in Fig. 3. It is also evident that pits at various stages of development were observed. 5. CONCLUSIONS It was concluded that the failure of the copper tubes was due to Type I pitting attack. It is not clear at this stage what the exact cause of pitting failure was, particularly given the fact that pitting only became evident after 12 years’ service. It is highly unlikely that it may be due to the presence of foreign matter deposits introduced during installation of the system. The introduction of foreign matter in the water is, however, a possibility, particularly if the water is not filtered. A change in water quality or content (e.g. resulting from mixing of the water with borehole or well waters) could also be responsible for pitting. Once initiated, pitting attack can in some cases be halted through the application of appropriate treatments of the water and the metal. The extent of pitting observed in the present case, however, 0 z 312 suggested that such treatment would be both unsuccessful and unfeasible. It was therefore rec- ommended that the copper tubes be replaced. Careful attention should be given to the usual causes of Type I pitting. In particular, it should be ensured that all tubes be thoroughly cleaned and freed of any carbonaceous deposits prior to installation. The tubes should also be cleaned to ensure complete removal of any foreign matter deposits and solder pastes after installation. The use of water filters could also be considered to prevent the introduction of foreign matter in the water. Furthermore, the quality and content of the water should be determined and its potential to cause pitting assessed. The extent of replacement or modifications to the water distribution system would, to some degree, depend on the results of such water analyses. REFERENCES 1. Internnl Corrosion of Water Distribution Sysrem. Report of Cooperation Research, AWWA Research Foundation, USA, 2. Billiau, M., Drapier, C., Muteriaux et Techniques, Nos 1 and 2. 1985. [...]... Park, OH, 1961, pp 805, 1178, I179 Failure Anaiysis Case Studies 1 1 D.R.H Jones (Editor) 0 200 1 Elsevier Science Ltd All rights reserved 317 Failure of automobile seat belts caused by polymer degradation J.M Henshaw".", V Wood", A.C Hallb The University of Tulsa, Department of Mechanical Engineering, 600 South College Avenue, Tulsa, OK 7 4104 , U.S.A The University of IIIinois, Materials Science and Engineering,... 1997;151(6):19 [7] Busch JV Technical cost modeling of plastic fabrication processes Ph.D dissertation, Massachusetts Institute of Technology, 1987 p 17 Failure Analysis Case Studies II D.R.H Jones (Editor) 0 2001 Elsevier Science Ltd All rights reserved 33 1 Oxidation failure of radiant heater tubes K.B Yoona.*D.G Jeongb a Department o Mechanical Engineering, Chung-Ang University, 221 Huksuk, Dongjak, Seoul.. .Failure Analysis Case Studies II D.R.H Jones (Editor) 0 200 1 Elsevier Science Ltd All rights reserved 313 CORROSION OF FLEXIBLE WAVEGUIDES D PAPATHEODOROU, M SMITH and 0 S ES-SAID* Mechanical Engineering Department,... thermal gradients are not the major cause of cracking Hence, for the analysis of the cause of failure of this case, a metallographic investigation of the failed component seems more appropriate than stress analysis Cracked specimen D was sampled from the cracked area around the supporting guide A as shown in Fig 2 for metallographic analysis Specimen D contained a big primary crack and several secondary... crack Considerable wall thinning was observed at the inside of the cracked tube Since it is known that a failure analysis with a secondary crack (which usually shows the initial stage of cracking) is preferred to that with a primary crack (which shows almost the final stage of the failure) , a failure analysis on the secondary crack which is 8-9 mm long was conducted The failed radiant tube was fabricated... July 1998; accepted 9 September 1998 Abstract This paper analyzes the failure of a particular brand of automobile seat belts The failures described were part of what nearly became the most expensive and widespread automobile recall in US history, affecting about 8.8 x IO6 vehicles and with a potential total cost of U.S $lo9 The failures were caused by the degradation and fracture of the seat belts'... above 900°C, they are generally fabricated by centrifugal casting with HK or HP steels having high contents of Ni and Cr In this study, a failure analysis of locally fabricated radiant heater tubes was performed A * Correspondingauthor - Reprinted from Engineering Failure Analysis 6 (2), I O 1 1 12 (1999) 332 Cracked region Burner support A \ a) Top View b) Side View Fig 1 Schematic illustration of the... Further examination of the receptacle mechanism revealed two other potential failure mechanisms A ‘mode 2’ failure, wherein a small piece of plastic from the release button gets wedged in behind the locking slider (Fig 6), preventing it from sliding to its unlatched position when the release button is pressed This is perhaps the most frightening failure mechanism, since it means that the belt wearer is unable... slider In this case, the fragment keeps the locking slider from sliding to the unlocked position when the release button (not shown) is pressed 3 Degradation and failure of the release button 3.1 Characterization of the button material Once the three failure modes described above had been discovered, attention was focused on the degradation and eventual fracture of the release button SEM studies of a... fragments of the release button to interfere with the function of the release mechanism Had any one of the elements in this failure chain been absent, it is likely that the widespread failures of the subject seat belts would never have taken place 2 Each of the three reported modes of failure can be explained in terms of a specific location within the seat belt mechanism in which a fragment of the release . Solubilities oflnorganic Compounds, 3rd edn. Van Nostrand, Princeton, NJ, 1941. Failure Analysis Case Studies II D.R.H. Jones (Editor) 0 2001 Elsevier Science Ltd. All rights reserved. Billiau, M., Drapier, C., Muteriaux et Techniques, Nos 1 and 2. 1985. Failure Analysis Case Studies II D.R.H. Jones (Editor) 0 200 1 Elsevier Science Ltd. All rights reserved. Texas, 1984, pp. 590. 101 . pp. 805, 1178, I179 Failure Anaiysis Case Studies 11 D.R.H. Jones (Editor) 0 200 1 Elsevier Science Ltd. All rights reserved 317 Failure of automobile