Materials Performance

SEP 2018

Materials Performance is the world's most widely circulated magazine dedicated to corrosion prevention and control. MP provides information about the latest corrosion control technologies and practical applications for every industry and environment.

Issue link:

Contents of this Issue


Page 105 of 128

47 MATERIALS PERFORMANCE: VOL. 57, NO. 9 SEPTEMBER 2018 be heated by the seawater on the shell side. Because of liquid oxygen's volatility, highly clean surfaces are required, 2 so the surfaces inside the tubes were to be cleaned after the repair. The absence of any guidelines for removing and reinstalling the spiro- vanes, however, made the cleaning a chal- lenge. Furthermore, seawater led to dam- age mechanisms and failure modes on the shell side, 3 which included cooling water corrosion, graphitic corrosion, microbio- logically inf luenced corrosion (MIC), and dense pitting on the flange faces. The tube bundle assembly was pulled in line with the OEM's guidelines. Hydro-jet cleaning was performed on both the shell and tube bundle, followed by inspection work. Due to the deep pitting on f lange faces, the option of using an epoxy-based repair composite for pitting was not con- sidered. Weld build-up work on the flanges was performed using gas tungsten arc welding, followed by in situ f lange facing that resurfaced the flange mating surfaces by machining to facilitate a tight, leak- proof seal when assembled. The flange material was 100-mm thick Type 304L SS, so thermal distortions in the flanges due to arc welding were expected. However, because the operating pressure of seawater on the shell side was ~0.9 MPa, thermal distortions of the shell f langes could lead to detrimental consequences in terms of leakage from the shell's f lange faces during hydrostatic testing as well as in normal operating conditions. Because of the potential for leakages from tubes at a ma ximum al lowable w orking pressure (MAWP) value of 4.2 MPa, the leak test for the tube bundle was conducted at a pres- sure of ~0.9 MPa, which was in line with the inspection and testing plan. During the test, five to 10 tubes were found to be leaking. The leaky tubes were taken out of service by plugging and seal welding. To access the tube sheet and plug the tubes, the dished head was removed by cutting. Visual inspection of the exposed tube sheet revealed some uniform corro- sion on its face. The presence of this uni- form corrosion could be attributed to either inadequate nitrogen purging during mothballing of the heat exchanger, or nitro- gen leakage after mothballing. Spiro-vanes were removed from th e leaking tubes, which were then plugged and seal welded. After seal welding the tube plugs, the heat exchanger was ready for final testing at 4.2 MPa MAWP without any intermedi- ate pressure testing. Rewelding of th e dished head was carried out following con- ventional Welding Procedure Specifica- tions (WPS) and weld maps using E308L electrodes. While welding the 38-mm thick dished head plates, longitudinal cracks formed immediately after the weld cooled. The E308L electrode was unable to join the plates. Formation of the longitudinal weld cracks, called cold cracking, was attributed to lower ductility of the welding rod and higher stresses in the base metal. 4 A hardness test of the dished head plates was conducted using a Krautkramer MIC 10 † ultrasonic hardness tester, and the hardness values were found to be around 240 HB. This hardness increase was attrib- uted to carburization 3 caused by carbon ingress from the gouging electrodes used when cutting the dished head. The WPS were revised and a new ENiCrFe-3 elec- trode—for higher toughness of weldment than E-308L—was proposed. The new elec- trode was able to withstand the higher stresses without cold cracking during the procedure qualification record phase. The dished head was welded and inspected by ultrasonic testing to check for weld defects. The delays due to review and approval of the new WPS from various stakeholders, including an ASME-authorized inspector and end users, as well as delivery time for the new ENiCrFe-3 electrodes, affected the project schedule. Also, the price of the ENi- CrFe-3 electrode was almost three times higher than the price of the E308L elec- trode, which considerably increased the cost for consumables compared to the budgeted amount. In this case, carburiza- tion, hardness of the dished head, and the risk of cold weld cracking—all unforeseen in the absence of inspection records and mitigation plans—contributed to exceed- ing the time and cost planned for the weld- ing work. All delays, as well as the revision of WPS for welding the dished head, could have been avoided by thoroughly review- ing documentation to ensure the applica- bility of data of as-built equipment and accessing previous inspection records for hardness. After correcting the welding issue, the heat exchangers seemed to be ready for final hydrostatic testing. However, a crucial check for tube bundle inspection by eddy current testing (ECT) was not carried out because it was never addressed during the inspection and testing plan reviews. This was contrary to API standard guidelines for cooling water corrosion diagnosis. 2 Since ECT requires insertion of probes inside the tubes, ECT wasn't even viable because of the presence of spiro-vanes inside the tubes. To check the integrity of the tube bundle at MAWP, the heat exchanger was pressur- ized on the tube side to reach the intended pressure of 4.2 MPa. Upon reaching ~2.0 MPa, however, the shell flanges experienced enormous leakage. The heat exchanger was immediately depressurized and the water was drained. The entire bundle assembly was pulled to check for damage. The major- ity of tubes (~90%) were punctured in a cir- cumferential manner at the tube/tube sheet interface. Although these tubes withstood a pressure of 0.9 MPa, they failed at pressures around 2.0 MPa. Discussion and Remarks Post failure visual analysis revealed that th e tub e s were weak and thin at th e expanded portion due to erosion and cor- rosion. Moreover, inaccessible portions (crevices) of the tube bundle were signifi- cantly affected by cooling water corrosion and somewhat by MIC. Also, some corro- sion found on the tube sheet surface was attributed to inefficient mothballing. Tube damage rendered the complete tube bundle assembly beyond repair. This failure led to huge financial loss and trig- gered further risk of a possible plant outage du e to th e absence of a standby h eat exchanger. † Trade name.

Articles in this issue

Archives of this issue

view archives of Materials Performance - SEP 2018