A Methodology for Screening Candidate Dissimilar Metal Welds for Repair or Replacement

Dissimilar Metal Welds are required in high temperature sections of boiler reheater and superheater tubing to join low alloy steel tubing to the finishing stainless tubing. The intrinsic difference in the coefficients of expansion cause high stresses to develop at the relatively weak ferritic/weld interface. The power generation industry began to experience failures of type 309 stainless steel welds in the early 1980’s and has recently begun to experience nickel-based weld failures on tubes with 150,000 or greater operating hours. For utilities who have or wish to institute a program of periodic inspection and a DMW repair/replacement maintenance strategy, one of the difficulties has been reliable NDE. Radiographic techniques have shown to be of some benefit, but the cost and lost maintenance time due to safety concerns makes it less than ideal. Replication and electrical resistance techniques also have drawbacks from the standpoint of accuracy. In 1996, SI developed an ultrasonic scanning technique for DMWs. In 1998, the DMWs at the Morgantown Station of PEPCO were examined using this technique and reasonably good correlation was found between UT and destructive metallographic techniques. This paper outlines the application of the focused scanning technique used to quantify the damage in DMWs, the means for selecting damaged DMWs for correlation with the technique and the manner in which this information is incorporated into a reliabilitycentered maintenance strategy for improving boiler availability. Introduction In the early 1980’s, EPRI sponsored workshops to deal with the problem of excessive DMW failures in power boilers (1). The main culprit was the use of a 309 type filler material, which has the greater difference in coefficient of thermal expansion vis-à-vis low alloy steel when compared to nickel-based filler materials. Utilities embarked on programs of inspecting and repairing/replacing damaged welds before they would fail and cause excessive consequential damage to the boiler tubing and significant losses in unit availability. A computer code, PODIS, was developed by EPRI for the purpose of predicting the rate of damage accumulation in DMWs as a function of tube operating temperature, type of weld, degree of bending stress and number of unit cycles. Some double wall radiographic techniques were developed which were useful in detecting DMW damage, but the main drawback was the cost per radiograph and the exclusion of personnel from the boiler in the vicinity of the radiographic tests. Replication and electrical resistance techniques were also employed, but their accuracy was questionable, as they are primarily surface, as opposed to volumetric, techniques. In the early 1990’s, ESEERCO (Empire State Electric Energy Research Corporation) requested that a review of the PODIS technology be conducted. A utility survey was conducted and it was found that the code was very conservative for nickel-based welds. From this program, a new code (DMW LIFE) was developed for ESEERCO. In the mid1990’s, SI used the probabilistic DMW LIFE on a project for PEPCO, wherein failure projections were made for DMWs , based on a few tube samples per boiler. In general, the predictions were reasonable and showed that weld failures could begin and increase over the next 5 – 10 years of operation. None did occur before this inspection. Focused Ultrasonic Techniques SI had developed TestPro, a PC-based digital UT system, in the early 1980’s. One of the first utility applications was for determining the wall and internal oxide thickness of SH/RH tubes. Using the principle of focused sound waves, a similar technique was developed for interrogating the weld metal/ferritic interface using a scanning transducer as shown in Figure 1 (2). Figure 1. DMW Scanning Transducer Assembly. Based on laboratory and early in-boiler tests, SI felt that an accuracy of +/10% on through-wall damage was feasible. Inspection Results During April 1998, the Morgantown Station Unit #1 boiler SH DMWs were examined by SI. The DMWs were located in the penthouse, consisting of 171 platen assemblies with DMWs in tube rows 19 through 24, exclusively. Due to access limitations and the small size of the tubing (1.5-inch diameter), only one scan could be performed on tubes 19, 20, 23 and 24. Two (2) scans were performed on tubes 21 and 22. The inspections were carried out from the space under the header between rows 21 and 22, which limited the available access and speed of the inspection. A total of 870 DMWs were specified for examination, resulting in a total number of scans of 1160 scans, when 2 per tube 21 and 22 are included. Of the specified 1160 scans, 23 (or about 2%) were not performed due to tube surface conditions or access limitations. 336 scans (or about 30%) exhibited no recordable indications. After all data was collected (four twelve hour shifts with two crews of two each), the permanently stored scans of those tubes exhibiting recordable indications were reviewed and sized in approximate 10% increments of tube wall thickness. In addition, a “best effort” characterization was performed on the indications to differentiate fabrication flaws from service-induced damage. Representative B-Scan images from the DMW inspection were saved to disk, and an example is shown in Figure 2. Of the tubes examined, two (2) were found to have damage levels of about 80%, four (4) were found to have damage levels of about 70%, seven (7) were found with 60% and fourteen (14) were found with levels of 50%. It should be noted that in some instances, a good inspection of the ID of the DMW (left side of Figure 2) could not be performed due to the presence of a wide weld crown or accessibility considerations. This meant that the damage level in the weld could be higher, especially since the ID is a primary initiation spot when bending stresses are high. Thus a damage level of 70% (of the UT visible weld could be somewhat larger when the entire cross section of the weld is examined. Based on the analysis, five welds were recommended for removal and rapid turn-around analysis by PEPCO’s Metallurgical Laboratory. Figure 2. Example of a B-Scan Image from a DMW. Metallurgical Analysis The five samples were sent to the laboratory where each weld was cross-sectioned at approximately 90 intervals. A summary of the correlations is found in Table 1. Table 1. Summary of Metallurgical and UT Results Wall Thickness Extent of OD Damage UT Results Metallurgical Comments 0.345” 0.233” 68% 50% Large OD Notch + Midwall Microcracking 0.412” 0.339” 82% 50% “ 0.375” 0.315” 84% 50% “ 0.413” 0.320” 77% 80% OD Creep Microcracking 0.411” 0.118” 29% 70% Isolated Slag Inclusion at 31% Through-Wall 0.411” 0.118” 37% 70% Isolated Slag Inclusion at 71% Through-Wall 0.400” 0.154” 39% 50% OD Creep Microcracking 0.398” 0.320” 80% 60% Fine Mid-Wall Microcracking 0.373” 0.285” 69% 60% Large OD Notch and Fine Mid-Wall Cracking 0.372” .320” 86% 60% Large OD Notch + Midwall Microcracking Typical photomicrographs of the damage are shown in Figure 3.