Metallurgical Failure Analysis of the Fractured Ring of a Gland Seal: Hydrogen Embrittlement? Factography can be Ambiguous

Abstract

Abstract One ring of a gland seal from a gas compressor was found fractured. The failure was detected because water leaked into the leaking gas system. The supplier apparently never had issues with this assembly. The field record is unremarkable, according to the manufacturer. However, the incident natural gas compressor is used in intermittent service only, while the assembly was originally designed for more or less permanent operation, according to the supplier of the compressor. It is hypothesized that hydrogen embrittlement initiated surface cracks in the subject ring, at the fastening hole, where wall thickness is minimal, and eventually caused delayed fracture due to the hoop stresses from shrink fitting of the gland seal ring. This was assisted by a hard and relatively brittle (quenched and tempered) microstructure, with manganese sulfide stringers and grain boundary carbide precipitates as contributing factors. All this eventually lead to brittle overload failure of the subject ring by predominantly transgranular lamellar tearing along MnS stringers, where the hoop stress caused decohesion at the interface between MnS inclusions and the matrix. This fracture morphology is typical of lamellar tearing (Terassenbruch). Actually, a mixed mode fracture morphology was observed. Interspersed in the predominantly transgranular overload fracture were areas of intergranular fracture along grain boundaries, embrittled by secondary carbide precipitates, a common condition in quenched and tempered hard martensitic stainless steels. No evidence of cyclic crack propagation, i. e. fatigue fracture, was found. Dimensions were not checked by the investigating laboratory. The hypothesis of dimensional deviations as a contributing factor to the fracture of the ring can therefore not be proven or ruled out within the scope of this investigation, and is mentioned here just for the sake of completeness and as a recommendation to the supplier of the assembly and to the manufacturer of the gland seal to look into this matter more closely. Finally, the strength of the steel, exceeding specified values, will likely have exacerbated the sensitivity of the microstructure to brittle overload failure by predominantly transgranular lamellar tearing, and partly intergranular cleavage. It also increased the propensity of the steel to failure by hydrogen embrittlement. From a design point of view, the combination of a sensitive microstructure due to MnS stringers, combined with too high a strength level, and the shrink-fitting of the gland seal assembly, is not well suited for this application, particularly if and when corrosion can produce hydrogen and embrittle the material further. It is strongly recommended to avoid in future strength levels exceeding the specification and what is usual for such applications, by sufficiently high tempering temperatures and times at temperature, and to avoid dimensional deviations that might increase hoop stresses in this interference fit. Still better would be to eliminate the shrink fit from this design altogether. It is understood by the authors that, in the meantime, the gland seal was re-designed and no shrink fit of the gland seal rings is needed any more.