Selective corrosion of type 316L austenitic stainless steel welds during the production of organic acids resulted in losses in production due to unscheduled downtimes to perform repairs. Estimated corrosion rates of type 316L filler material welds were an order of magnitude higher than that of the base material. Alternative higher alloyed commercial filler materials were evaluated under actual production conditions. The evaluated filler materials were types 316L, 317L, 309L, 309MoL, 2205, 2507, 625, 825 and 904L. The effect of nitrogen on the corrosion properties of type 309L filler material was evaluated by manipulating the nitrogen concentration of the shielding gas during MIG welding. These changes in nitrogen concentration did not influence the corrosion resistance of the type 309L filler material. No correlation could be established between the corrosion rates, analysed chemical composition of the product and operating temperature during production. In almost all the cases where the chemical composition of the filler material was comparable with that of the base material the corrosion rates of the filler materials were higher than that base material. It might be expected that the ferrite phase with higher molybdenum and chromium should be more corrosion resistant while the austenite should be less resistant. This was, however, not the case with the corrosion of type 309L filler material. It would thus appear that in this case nickel enrichment of the austenite phase had a larger influence on the corrosion resistance of the austenite phase than the chromium and molybdenum had on the corrosion resistance of the ferrite phase. It appears that nickel and molybdenum had the largest contribution to the corrosion resistance of stainless steels welds under these operating conditions. It is, however, believed that a certain minimum concentration of chromium is also required to provide corrosion resistance to these alloys in hot organic acid environments. In contrast with the fact that a substantial alloying content is required to improve corrosion resistance of the filler material, the small difference in composition between ferrite and austenite phases, due to micro segregation, appeared to affect the corrosion resistance on micro scale. This is illustrated by the micrographs, which show corrosion to etch out the dendrite structure. Since the morphology of the austenite and ferrite phases is so similar, it could not always be conclusively established which one of the two phases corroded selectively. Analyses performed on the austenite and ferrite phases did not indicate a concentration difference within the phases itself. However, there were significant differences in the concentration of elements between the phases, with the austenite stabilising elements reporting to the austenite phase and the ferrite stabilizing elements reporting to the ferrite phase, in line with thermodynamic predictions. In the case of the filler materials following the austenite mode of solidification, no significant concentration differences were detected within the matrix. Although all highly alloyed high nickel alloyed filler materials (types 904L, 825 and 625) corroded at a lower rate than the type 316L base material, type 625 filler material was the filler material of choice due to the lack of any pitting of the weld. Pitting was detected in both the 825 and 904L filler materials. Galvanic corrosion was not noted at any of the weld/HAZ interfaces and in no case did the type 316L parent metal adjacent to the weld corrode preferentially to the material further away from the weld. Copyright
Dissertation (MEng)--University of Pretoria, 2010.