PPCO Twist System

FAILURE ANALYSIS amounts of hydrogen sulfi de (H2S) (1 to 2 wt%) and ammonia (NH3) (1 to 2 wt%). T e pH of the condensed water in the overhead drum is 9 to 9.5. T e tubes of this condenser failed after Rapid Failure of only a few months of operation, following the scheduled turnaround in April 2004, during which the condenser had been re- tubed. T e retubing had been carried out to replace a mechanically damaged Al-brass a Copper/Nickel (ASTM B111, grade C687) bundle, which had been in service for more than 20 years without any corrosion problems. T e CuNi alloy was chosen because of availability and Overhead Condenser was expected to behave similarly to the Al- brass under the prevailing conditions. Available information from literature2 in- dicated that CuNi is more resistant than

Al brass in NH3-containing environments Bundle (as is the case here) and is virtually immune to season (NH -induced) cracking. T e ALBERT VELD, Shell Pétrochimie Méditerranée 3 overhead condenser was found to be leak- ing in August 2004, after only three Rapid and unexpected failure of a CuNi bundle of months of operation. a stabilizer overhead condenser occurred where an Al-brass bundle showed chemical resistance for more Failure Analysis than 20 years. An investigation revealed that the inter- VISUAL EXAMINATION granular attack/cracking of the cupronickel tubes was T ree failed tube samples showed nu- merous longitudinal cracks, although some caused by attack of a wet hydrogen sulfi de (H2S)- axial cracks were also observed (Figure 1). containing medium. Literature indicated that cupronickel The longitudinal cracks were relatively is much more susceptible to attack by wet H2S than Al- straight-lined, whereas the transverse brass. Cupronickel is unsuitable as tube material in cracks showed a more irregular path. All cracks were very branched and covered the overhead systems containing signifi cant amounts of H2S and ammonia (NH ). total circumference of the tube samples; 3 some of them were through-wall cracks. Signifi cant wall loss was not observed, and the cracked zones did not show signs of plastic deformation. T e tube samples are totally covered with a black adherent scale. his article concerns a stabi- lizer overhead condenser of IDENTIFICATION AND TESTING a steam cracker unit. T e OF TUBE MATERIAL tube material is a CuNi Metal identifi cation was done with a alloy (ASTM B111,1 grade portable analyzer. The analysis of the C706), with cooling water cracked tube samples confi rmed the cupro- on the tube side (max. tem- nickel alloy to be CuNi10Fe1Mn perature 35°C), and stabilizer overhead (EN12451), or grade C706, according to T(max. inlet temperature 120°C) on the ASTM B111—88.8% Cu, 9.7% Ni, 1.0% shell side. Fe, and 0.6% Mn. T e stabilizer overhead contains mainly T e tensile properties of the tube mate- light components, with significant rial (blanc sample) do correspond with

52 MATERIALS PERFORMANCE April 2006 FIGURE 1 those of a cupronickel alloy C706 for the (0.3 to 3%), and S (8 to hard-drawn condition (H105), according 20%). The microanalyses to EN12451—a tensile strength of 466 also showed a similar and MPa and a yield strength (0.2%) of 433 homogeneous alloy compo- MPa. T ese data were not in line with the sition inside the grains and materials certifi cate for the tube material; at the grain boundaries. the certifi cate specifi ed the H075 condi- No particular element or tion. T e deviation from the materials inter-metallic phase was de- certifi cate as shown by the tensile proper- tected at the grain boundar- Failed cupronickel tube sample. ties was confi rmed by hardness measure- ies that could have been ments. T e hardness of the failed tube responsible for the inter- sample and the blanc sample were mea- granular cracking and at- FIGURE 2 sured to be 140 to 145 hardness, Vickers tack. A specifi c EDXA for (HV) 5, values corresponding to the hard Hg was negative. drawn condition H105 and much higher than the specifi cation for the annealed Discussion H075 condition (75 to 105 HV 5). At fi rst, the rapid failure of the cupronickel bundle of the stabilizer overhead con- Cupronickel is very denser was a complete mys- susceptible to attack tery, especially because an Al-brass bundle exhibited corrosion resistance in these in H2S-containing wet NH3/H2S overhead con- environments, where ditions for more than 20 years. Available information the Al-brass is almost from literature indicated that CuNi is more resistant

totally resistant. than Al-brass in NH3-con- taining environments and is EXAMINATION AND virtually immune to season MICROANALYSES OF (NH3-induced) cracking. FAILED SAMPLES T e only medium the inves- Metallographic examination showed tigators could think of to the presence of numerous inter-granular explain the severe cracking/ and very branched cracks that initiated at attack of CuNi alloys was the outside tube surface (Figure 2). T e mercury salts, but it was cracking was associated with a superfi cial confi rmed that the salts did not play a role in this par- inter-granular attack of the outside tube Inter-granular cracking/attack of the cupronickel tube material. surface to a depth of ~0.1 mm. T e inter- ticular failure case—a spe- granular attack and cracks were com- cifi c EDXA for Hg revealed pletely fi lled with a grey-colored product no trace of this element. A microanalysis of the corrosion prod- (Figure 3). T e microstructure is in ac- tities of H2S and NH3 available. T e cu- cordance with the specifi ed structure for ucts on the CuNi tubes and in the cracks pronickel (90/10) containing small quan- this CuNi alloy. (Figure 4) revealed the presence of massive tities of Fe and Mn (as is the case for the T e corrosion products present at the quantities of S. From literature,2 it became C706) is even more vulnerable to attack. surface and in the cracks were analyzed and clear that cupronickel is very susceptible T e authors therefore concluded that the found to be predominantly composed of to attack in H2S-containing environ- attack of the cupronickel overhead con- CuNi sulfi de (Figure 4). T e semi-quanti- ments, where the Al-brass is almost totally denser bundle was caused by an exposure resistant. tative energy-dispersive x-ray analysis to wet H2S in the temperature range of (EDXA) indicated the presence of Cu (50 T is is especially true for overhead 40 to 110°C. In the cited literature, how- to 80%), Ni (8 to 25%), Fe (1 to 3%), Mn systems where there are signifi cant quan- ever, uniform corrosion is mentioned and

April 2006 MATERIALS PERFORMANCE 53 FIGURE 3

not inter-granular corro- • T e inter-granular attack/cracking of sion/cracking as found in the cupronickel tubes was caused by at- our case. tack of a sulfur-containing medium

T e same issue of poor (wet H2S), condensed in the overhead performance of CuNi-10 in of the stabilizer column, on the shell

an H2S- and NH3-contain- side of the condenser. ing overhead system of re- • From literature, it was found that fi nery fl uid catalytic crack- cupronickel, being more resistant than

ing unit was the subject of Al-brass in NH3-containing media, is an April 2005 MP Forum much more susceptible to attack in wet 3 Q and A. T ere was some H2S-containing media. T e conclusion experience and feedback, seems to confi rm the excellent perfor- but the exact causes still mance of an Al-brass bundle for more remained unexplained. than 20 years under the prevailing con- T e inter-granular cor- ditions, whereas the cupronickel bundle rosion/attack of the cupro- failed after three months of operation. Detail of the inter-granular cracking/attack showing the gray nickel tubes could possibly • T e mechanical properties (tensile corrosion product at the surface and in the cracks. be explained by the hard- strength and hardness) of the cupro- drawn condition of the ma- nickel tube material did not correspond terial. T e mechanical prop- with the annealed condition (H075) erties (tensile strength and stated in the material certifi cate, but was FIGURE 4 hardness) of the cracked found to represent the hard-drawn tubes did not correspond (H105) condition. with those indicated in the • It seems probable that this metallur- material certifi cate and indi- gical condition played a major role in cate a hard-drawn condition the degradation of the tubes, by pro- (H105) rather than an an- moting inter-granular attack/crack- nealed or light-drawn con- ing. dition (H075). • Cupronickel, used especially because To avoid attack and of its excellent resistance to seawater, is cracking of the tubes of the not suitable as tube material in overhead overhead condenser, cupro- systems containing signifi cant amounts nickel should be avoided as of H2S and NH3, which is the case for construction material. An most of the refi nery units’ overhead alternative material could systems. be carbon steel (CS). CS will resist attack from the References 1. ASTM B111, “Standard Specifi cation for Copper and product side, but, in our Copper-Alloy Seamless Condenser Tubes and Ferrule Stock” experience, will be attacked (West Conshohocken, PA: ASTM). quite severely from the 2. Metals Handbook, 9th Edition, Vol. 13, Corrosion, p. 322 and Dechema Werkstoff Tabelle, H S B1 14 and 15. cooling-water side. 2 3. MP Forum, “Corrosion of FCCU Overhead,” MP 44, 4 (2005): p. 56. Conclusions From the diff erent inves- ALBERT VELD is a corrosion specialist at Shell Petrochimie Mediterranee (SPM), B.P. 14, 13131 tigations and analyses car- Berre L’Etang-Cedex, France. He has more than ried out on several failed 35 years of experience in corrosion engineering. tube samples from the cu- He worked for 20 years at the Shell Research pronickel bundle of the Laboratory in Amsterdam, four years at Shell stabilizer overhead con- Canada, and has been with Shell France for 14 years. He specializes in plant integrity and risk- denser, the following con- EDXA of gray corrosion product present on the tube material based inspection. He has a degree in physical and within the cracks. clusions/explanations can chemistry and has been a NACE member since be drawn: 1992.

54 MATERIALS PERFORMANCE April 2006