Abstract
While scale exfoliation from the steam side of superheater and reheater tubes in fossil-fired steam boilers has been a problem for over 40 years, recent trends to increased steam temperature (and pressure), longer operating times at near to maximum load, and the introduction of new alloys have emphasized the importance of this phenomenon as a source of unavailability of coal-fired steam plant, and have led to renewed interest in approaches for managing it. Available mechanistic understanding suggests that thermally-grown oxide scales will detach from the metal surfaces on which they were grown when they are strained beyond a critical limit. Strain accumulates in growing oxides due to two main processes: from volume changes that occur when the metal is converted to oxide, and when changes occur in the oxide morphology as the oxide itself is further oxidized; and due to thermal cycling, when the often large differences in coefficients of thermal expansion between the metal and the oxide (and among different oxides) generate large stresses over relatively short times. Such strain accumulation increases with increasing scale thickness and increasing rate of temperature change, and there is a minimum scale thickness above which exfoliation is likely. Also important is the amount of scale that detaches in an exfoliation event, as well as the size and shape of the oxide flakes, since these influence the ultimate destination of the debris in the steam circuit: either deposition in tube bends (resulting ultimately in blocking and/or tube overheating), or entrained transport in the steam to the turbine (leading to erosion damage of the first stages of nozzles and blades). Mechanistic understanding of the oxidation process allows these issues to be addressed analytically, and this approach has been used to map (the “Armitt Diagram”) the regimes where various forms of exfoliation are expected, as a function of oxide thickness (time at temperature) and accumulated strain (oxide type, oxide and alloy properties). Limited use of this approach has provided quite accurate (post-event) predictions, albeit for only two classes of alloys for which sufficient data are available. The program from which this paper was drawn is aimed at extension of this approach to develop a tool for predicting the conditions under which an exfoliation event will occur, and its consequences. Progress depends on the ability to accurately predict oxide thicknesses and to develop mechanistic descriptions of the evolution of scale morphologies (to determine the features that trigger detachment, and mode of detachment) for a wider range of alloys and at temperatures and steam pressures higher that those accommodated in the original Diagram. Of particular concern are the 9-12%Cr ferritic-martensitic steels which do not appear to follow the behavior established for the lower-Cr steels, since the oxidation behavior of these newer steels appears to be significantly affected by small changes in, for instance, alloy composition (within the alloy specification), rendering problematical the description of the evolution of scale morphologies in a way that allows the application of analytical treatments for calculating stresses. This paper summarizes the issues being addressed, and describes the approach taken to build on the earlier methodology for predicting scale exfoliation.
