Abstract. There is a confused and contradictory literature on the role of small concentrations of niobium on the development of the so-called local brittle zones in steels. These zones consist of a mixture of martensite and austenite and hence, their formation should be predictable using modern microstructure calculation methods. Following an assessment of the most relevant literature, a mathematical model is presented which enables three quantities to be calculated, the fraction of martensite, the carbon concentration of the martensite, and its ability to influence toughness. Examples are presented for particular linepipe steels, and then the generic effect of alloying elements other than niobium, on the development of local zones.
The Local Brittle Zone as an EntityLocal brittle zones (LBZs) are small regions of hard, brittle phase that form in the heat affected zones (HAZs) of multipass welds [1]. They usually contain untempered martensite that can lead to scatter in toughness data when the test specimen samples a brittle zone [2]. Such scatter reduces confidence in design because of the existence of a few poor values which fall out of the desired range. Experimental data have been interpreted to conclude that the lower bound toughness values deteriorate as the volume fraction of martensite in the form of LBZs increases [3]. However, the same data can be interpreted to indicate that there is a minimum in the toughness as a function of increasing fraction of martensite [4]. Local brittle zones are often treated without a consideration of their detailed microstructure but to make rational assessments it is necessary to address the following questions in a quantitative manner:1. How hard does a LBZ have to be in order to influence scatter in toughness? Or is it the hardness relative to the surrounding material that influences its ability to act as a brittle zone?2. How big does a LBZ have to be in order to matter, relative to the size of the plastic zone in the test used to determine toughness?3. Does the shape of the LBZ matter?4. How many tests are required in order to establish a probability of failure based design?Given these unknowns, it is not surprising that the critique by Liessem and Erdelen-Peppler [5] fails to resolve the fourth query in this list, other than to suggest a change in specification that would allow repeated testing of rejected pipes until the specification is met. The practical problem is that all pipes contain LBZs when longitudinally welded, and it is luck that determines whether the small-scale tests hit an LBZ. The real issue should include a specification of an acceptable level of risk, and probabilistic testing to see whether the material and process combination manages that risk. The generation of probabilistic data is of course, expensive.
Martensite/Austenite ConstituentThe martensite-austenite (MA) constituent is a rather clumsy name given to what is essentially a mixture of untempered martensite embedded in carbon-enriched retained austenite. When