ABSTRACT
This paper discusses reactors that experienced numerous cracking problems, over a 8-year period, where cracks are confined to the welded zones. The reactors were in hydrocarbon service. The material of construction is solid TP347 stainless steel and the original construction welds as well as initial repair welds were done with E347-16 consumables. Other factors playing a role are: more than 60 transients resulting in creep-fatigue as well as the poor creep-rupture ductility of welds. Inner surfaces are subject to coke formation and a small degree of low temperature carburization and sulfidation at temperatures in the region of 560° C. Welded joints that have cracked are where the parent material is above 12mm in thickness and where grain sizes are coarser than ASTM no 3.5. Concerns of weld quality, triaxial stress raisers, heat treatment practices, melt practices and sigma phase formation is included. During shut-down conditions some polythionic acid stress corrosion cracking had also been identified. This paper discusses and reveals examples of stress relaxation cracking as well as classical examples of white-phase-fractures. Corrective recommendations, using leaner E16.8.2-15 consumables for repair welding are also covered.
INTRODUCTION
Root cause failure analysis in the petrochemical industry is often hampered by numerous issues. In a meter long crack for example, the knowledge gained at the origin, mid-point and crack tip can be completely different. The information gained from any sample also depends on where the sample was taken and the availability of applicable historical records. To complicate matters further, samples are not always taken where specified and when taken are often removed incorrectly, labeled incorrectly or not supported by clear in-situ photos. These are the realities during a shutdown when there is pressure to get the plant back up and running. As a result one is often forced to work with the samples despite how and where they were taken and try to perform a root cause failure analysis. Moreover, in the case of these reactors there were many factors that masked the fracture face and adding to the complexity of the investigation. The primary root cause for cracking was eventually identified as stress relaxation cracking (SRC). Factors contributing to SRC included: material selection, grain size, restraint, tri-axial stress raisers [undercut, lack of fusion (LOF), lack of penetration (LOP), solidification cracking]. Factors often associated with SRC include white phase fractures (WPhF), HAZ hardening and creep. Secondary cracking mechanisms included creep (cyclic, monotonic & dynamic), interactive creep-fatigue and polythionic acid stress corrosion cracking (PTA-SCC). Cracks that propagated through welds had a very brittle appearance, most probably due to the very poor creep rupture ductility of E347-16 welds. Tertiary degradation mechanisms included: low temperature carburization, metal dusting, sigma phase formation and marginal sufidation at the operating temperatures of 560]>°C. There are numerous similarities to many of the failure mechanisms under discussion and a detailed understanding of all the service conditions and transient realities are required to try and separate them. Needless to say, many of the modes of failure can overlap thus adding to the complexity.
