Fundamental Studies of Hydrogen Attack in C-0.5Mo Steel and Weldments
P. Liu, C. Lundin, M. Prager
Number of Pages
High temperature hydrogen attack (HTHA) occurs in steels exposed to high temperature (>200°C [400°F]), high pressure hydrogen and involves surface decarburization, internal decarburization, and/or intergranular cracking. Hydrogen attack is an irreversible process causing permanent damage, which results in degradation of mechanical properties and possible failure including leakage, bursting, fire, and/or explosion.
The continued occurrence of hydrogen attack in C-0.5Mo steel and weldments operating below the C-0.5Mo Nelson Curve has caused significant concern for the integrity and serviceability of C-0.5Mo pressure vessels and piping in the petroleum refining and petrochemical industries. In accordance with the concerns of these industries, fundamental studies of hydrogen attack in C-0.5Mo steel and weldments were accomplished in terms of:
(1) Quantitative methodologies for hydrogen damage evaluation;
(2) Hydrogen damage assessment of service-exposed weldments and laboratory autoclave-exposed materials, including a one-side hydrogen exposure assembly specifically designed to mimic exposure conditions present in service;
(3) Investigation of the effects of carbon content and alloying element content, heat treatment, hot and cold working, welding processes, and postweld heat treatment (PWHT) on hydrogen attack susceptibility;
(4) Development of continuous cooling transformation (CCT) diagrams for C-0.5Mo steels under heat treatment and welding conditions;
(5) Evaluation of carbide composition and morphology for C-0.5Mo steels after service exposure and heat treatment;
(6) Determination of methane evolution by the reaction of hydrogen and carbides;
(7) Calculation of hydrogen diffusion and methane pressure through the wall thickness of the one-side hydrogen exposure assembly;
(8) Hydrogen attack mechanisms and modeling of hydrogen attack limits.
In conjunction with this research, a state-of-the-art literature review was performed to provide a comprehensive overview of the published research efforts on hydrogen attack. The evolution of "Nelson Curves" for carbon steel, C-0.5Mo, and Cr-Mo steels was historically reviewed in regard to design applications and limitations. Methodologies for hydrogen attack assessment were summarized under the categories of hydrogen exposure testing, mechanical testing, and dilatometric swelling testing.
In the current study, service-exposed C-0.5Mo steel components comprising twenty-four separate heats were received for hydrogen attack evaluation using quantitative metallographic analysis, the "Cryo-cracking" fractographic technique, methane determination, carbon concentration measurements, and advanced ultrasonic inspection. Hydrogen attack damage was found in nine out of the twenty-four service-exposed heats. The hydrogen damaged heats were operated at pressure/temperature conditions below the C-0.5Mo Nelson Curve with the exception of MPC-15 (operating condition was located on the Nelson Curve).
Hydrogen damage parameters in terms of the fraction of damaged grain boundary length (FL%) and the fraction of damaged grain boundary area (FA%) were defined for quantitative damage evaluation using optical light microcopy (OLM), scanning electron microscopy (SEM) in combination with image analysis software (NIH Image). Hydrogen attack damage profiles (FL%, FA% as a function of wall thickness) were determined for the service damaged heats and the specially-designed one-side hydrogen exposure assembly. It was found that hydrogen attack damage is most extensive adjacent to the surface exposed to hydrogen and progressively decreases through the wall thickness. The progression of hydrogen attack in terms of methane bubble nucleation, bubble growth and coalescence, and microcracking at different locations through the wall thickness was characterized using OLM and SEM. The "Cryo-cracking" technique, utilizing fractographic examination in the SEM, was demonstrated to be a unique method for evaluating hydrogen damage and determining methane bubble size. A relationship between methane bubble size and the fraction of damaged grain boundaries (FL%, FA%) was established. Recommended methodologies for hydrogen damage examination are proposed based on the morphology of hydrogen attack damage in terms of isolated bubbles and aligned bubbles (SEM) or microcracks (UT, OLM, SEM).
Methane content in service-exposed and autoclave-exposed materials was determined by measuring hydrogen content (dissociated from methane) using gas chromatography and the LECO vacuum fusion method. A relationship between methane concentration and grain boundary damage (FL%, FA%) was established to enable hydrogen damage assessment using methane determination.
One-side hydrogen exposure testing was developed by using a box-like assembly to simulate hydrogen exposure conditions of pressure vessels and piping. The assembly was designed with a sloping wall to enable a range of wall thicknesses to be tested simultaneously. The one-side hydrogen exposure testing successfully simulated hydrogen attack behavior in pressure vessels and piping. A gradient of hydrogen damage was observed within the wall thickness using SEM and methane determination. Based on the determination of the maximum damaged depth for three sections with different wall thicknesses (44 mm, 29 mm, and 13 mm), it was found that the fraction of damaged depth is independent of the wall thickness if the exposure condition is the same and a steady state hydrogen gradient is established.
The effects of carbon content, alloying element content, heat treatment, welding, postweld heat treatment, and hot and cold working on hydrogen attack susceptibility were investigated. It was demonstrated that low carbon content and the addition of carbide-forming elements (Mn, V, Cr) in steel enhances the resistance to hydrogen attack.
In C-0.5Mo weldments, the coarse-grained heat-affected zone (CGHAZ) was observed to be the region which is most susceptible to hydrogen attack. More extensive hydrogen attack cracks were found in a CGHAZ with low heat input (13.8 kJ/cm [35 kJ/in], SMAW) than in a CGHAZ with high heat input (39.4 kJ/cm [100 kJ/in], SMAW). Hydrogen attack susceptibility in the CGHAZ was reduced by postweld heat treatment. Low-carbon weld metal and the fine-grained HAZ regions (FGHAZ) were demonstrated to be minimally susceptible to hydrogen attack.
Hydrogen attack susceptibility was ranked for a group of materials in normalized & tempered (N&T), annealed only, annealed & tempered (A&T), hot worked, and cold worked conditions based on metallographic examination. It was found that the cold worked and annealed materials were the most susceptible to hydrogen attack. With increasing plastic deformation from cold working, hydrogen attack susceptibility increased as well. The extent of hydrogen attack cracking tended to decrease with an increase in hot working deformation. In heat-treated materials, hydrogen attack damage was reduced as tempering temperature increased. Hydrogen attack susceptibility for the A&T, N&T, and hot worked materials was similar and they exhibited lower susceptibility than the cold worked and annealed materials.
For microstructure characterization, CCT diagrams were developed for two heats of C-0.5Mo steel (0.21 wt.% C and 0.15 wt.% C) under heat treatment and welding conditions (slow and fast heating rates, respectively) using Gleeble thermal simulation and high speed dilatometry. Microstructures and carbides were studied for select tempered materials using OLM, SEM, and transmission electron microscopy (TEM) coupled with convergent electron beam diffraction and energy dispersive spectroscopy (EDS). The predominant carbides were identified as M3C for the service-exposed heats and N&T materials. During prolonged tempering or PWHT (621°C [1150°F], 15 h), carbide evolution occurred from M3C carbides to M23C6 and Mo2C type carbides. M23C6 and Mo2C type carbides are more stable in comparison to M3C and can enhance hydrogen attack resistance.
Methane evolution was investigated by using autoclave exposure testing and methane determination. The results indicated that methane formation starts at the beginning of hydrogen exposure at 454°C (850°F), 3.4 MPa (500 psi) H2. An in-situ methane bubble nucleation mechanism was proposed based on the volume change caused by hydrogen-carbide reactions.
The equilibrium methane pressure distribution through the wall of the one-side exposure assembly was calculated using a model developed from the relevant equations pertaining to hydrogen diffusion and thermodynamics of methane formation. Using a criterion relating methane pressure to sintering force (PCH4 > 2s/r), the maximum damage depths at different wall thicknesses in the one-side exposure assembly were calculated and were demonstrated to be consistent with the measured results.
A hydrogen attack mechanism was developed for the condition of one-side hydrogen exposure, as experienced in pressure vessels and piping in service. The reaction of hydrogen with carbon or carbides occurs resulting in the formation of methane bubbles at grain boundaries or carbide/matrix interfaces. Methane bubbles may nucleate in-situ where the reaction of carbides (Fe3C) and hydrogen occurs. The rates of methane formation, bubble nucleation, and growth depend on hydrogen pressure and temperature. Because of the hydrogen concentration gradient within the wall, the methane bubble nucleation rate is more rapid near the surface exposed to the hydrogen than at locations away from the surface. As the reaction of hydrogen with carbides or carbon continues and methane accumulates, internal methane pressure in the bubbles increases. Consequently, a gradient of methane pressure is established within the wall thickness, with the highest methane pressure occurring at the hydrogen exposed surface. Methane bubbles grow when the internal methane pressure is greater than the sintering force (2s/r). Therefore, a gradient of hydrogen attack damage and a maximum damaged depth were observed through the wall as demonstrated in the service-exposed materials and the one-side exposure assembly.
Based on a review of the data used for development of the C-0.5Mo Nelson Curve, it was found that the C-0.5Mo Nelson Curve in API Publication 941 was defined based on an insufficient and unreliable database. According to the database generated in this research, a modified hydrogen attack limit (Nelson Curve) was defined for C‑0.5Mo steel, below which hydrogen attack might not occur for approximately 28 years. The modified hydrogen attack limit (for C-0.5Mo steel) is below the carbon steel Nelson Curve in API Publication 941. Because hydrogen attack is a time-dependent process, it is suggested that the time for safe operation should be stated on Nelson Curves.
A model for hydrogen attack limits for C-0.5Mo steel was developed by applying a critical value of −9.0 to a previously published hydrogen attack parameter (Pw). The hydrogen attack limit calculated from the model satisfies the modified hydrogen attack limit defined in this research. Based on this model, the exposure time for the C-0.5Mo Nelson Curve published in API Publication 941 was determined as 10,500 h (14 months). This further indicates that the C-0.5Mo Nelson Curve defined in API Publication 941 is not valid for long term operating conditions. The developed model emphasizes that hydrogen attack is a time dependent process. In addition to the operating temperature and hydrogen partial pressure, safe operational times should be stated on the Nelson Curves
 Maximum damaged depth is the maximum extent (depth), measured relative to the ID surface, of hydrogen damage through the wall thickness of a component, as observable by SEM.