This case study presents the fatigue assessment of the 28L oil separator (Atlas part number 1625481501), performed to verify the vessel’s durability under cyclical internal pressure in accordance with the AD2000 S2 standard. The separator is subject to a cyclic pressure loading from atmospheric pressure up to 15 barg, which requires a thorough evaluation of fatigue life, especially in the context of concurrent experimental fatigue testing being conducted by the client. The assessment was conducted using linear elastic static finite element analysis (FEA), with particular focus on weld details and their fatigue classifications as specified by the code.
Methodology
Fatigue Stress Assessment
The fatigue evaluation follows the AD Merkblatt-2000 S2, focusing on welded components (Paragraph 7.2). The assessment checks whether crack initiation or growth is likely within the anticipated service life by comparing the computed stress range to allowable code-defined values. The total stress range is composed of local membrane, general bending, secondary, and peak stresses. Peak stresses, which typically arise at geometric discontinuities such as weld toes and roots, are directly addressed by weld classification as per AD2000 S2 Table 5. The relevant weld class is used to select the corresponding ferritic steel fatigue curve from AD2000 S2 Figure 12.
Stress Linearization
Fatigue-critical welds are assessed by linearizing the stress through the wall thickness at likely crack initiation sites (weld toes and roots). These paths are defined as Stress Classification Lines (SCLs). Along these paths stress linearisation is performed and the stress range for the linearized membrane stress and bending stress is computed. The stress range is computed as the difference between the fully loaded (operating) and unloaded (atmospheric) conditions, and combined using Tresca’s theory.
Material Properties
Material properties were taken from ASME II-D (2010), using the governing temperature of 92.5 °C determined according to the AD 2000 S2 method. The vessel body is constructed of ST37-2 carbon steel, the flanges of SA-105, and the valve of EN42200 aluminium. The mechanical properties (Young’s modulus, Poisson’s ratio, and density) are defined for each materials ensuring accurate representation in the FEA model.
Model Description
Geometry and Mesh
The separator and its attachments were modeled based on provided technical drawings. Welds are modelled as triangular sections, capturing the geometry’s load transfer characteristics. For perpendicular nozzles a right-triangle section was used, while slanted nozzles had tangentially varying weld sections. The connection between vessel and bottom head featured a 3 mm weld, and a corrosion allowance of 0.1 mm is applied to inner vessel and flange walls.
A 3D finite element model was developed using PTC Creo Simulate 3.0 M030, using higher-order P-elements (up to 7th order) to ensure convergence. Local mesh refinement (down to 1 mm element size) was applied at welds and other assessment locations to capture stress gradients accurately. Mesh convergence was confirmed when the change in von Mises stress between successive refinements fell below 5%.
Loads and Boundary Conditions
A cyclic internal pressure of 15 barg was applied. Pressure thrusts on plugged nozzles were applied at the threaded areas or flange faces, with the plugged nozzles assumed to be stiff. The nozzle face surface is made rigid. The vessel’s legs were fixed in all translational degrees of freedom. The aluminum valve on top of the flange was included for its contribution to flange stiffness, though it was not itself assessed for fatigue. Key interfaces (vessel–flange, flange–valve) were modeled as glued connections.
Weld Classification
Welds at nozzles and certain connections (flange to baffle, internal plate) were partial penetration welds, corresponding to weld cartoon 2.7 per AD2000 S2 Table 5, and were thus assigned the most conservative fatigue class, K3. The vessel-to-flange and vessel-to-bottom head connections were similarly classified as K3 due to their partial penetration weld design. Fatigue curves and allowable stress ranges were applied accordingly.
Results and Discussion
Global Stress and Deformation
The global von Mises stress distribution, computed under maximum operating pressure, revealed stress concentrations at several key weld locations, particularly at the vessel-to-flange connection and nozzles 7 and 10. The deformed shape of the vessel confirmed that the highest stresses were localized at these connections.
Stress Linearization and Assessment
Stress linearization at the most critical locations provided the membrane and bending components in all principal directions. For the inlet nozzle 7 (SCL 1.1), the maximum combined membrane plus bending stress (by Tresca criterion) was 128 MPa, corresponding to an allowable number of cycles of approximately 61,000.
A summary of the most critical fatigue locations is as follows:
- Vessel-to-flange connecting weld: Tresca stress 131 MPa, allowable cycles 57,000 (the lowest in the assessment)
- Inlet nozzle weld (nozzle 7): Tresca stress 128 MPa, allowable cycles 61,000
- Nozzle 10 weld: Tresca stress 126 MPa, allowable cycles 64,000
Other welds, such as those at nozzles 9 and 11, leg attachments, and internal components, exhibited lower stresses governing higher allowable cycles. The most critical components featured a partial penetration weld and were subjected to high local stresses.
Conclusions and Recommendations
The fatigue assessment determined that the 28L oil separator, as currently designed, can safely withstand at least 57,000 full pressure cycles, with the limiting feature being the vessel-to-flange weld. Since most of the applicable welds are partial penetration welds they are classified in the most conservative fatigue category (K3, which governs the allowable number of cycles.
To improve fatigue life, two primary recommendations are made:
- Upgrade Weld Penetration: Employing full penetration welds in place of partial penetration welds would elevate the weld classification from K3 to K2, effectively doubling the allowable number of cycles at critical locations.
- Increase Weld Size: Enlarging the welds at critical joints would increase the load-carrying area, reduce the fatigue stress and thereby increase fatigue life.
These improvements would be particularly beneficial should operational requirements demand a higher number of cycles than currently supported by the existing design. Implementation of such recommendations should be balanced against manufacturing constraints and cost considerations.
Key Takeaways
The fatigue life of the 28L oil separator is determined primarily by the details of its welds, especially at the vessel-to-flange and nozzle connections. The use of partial penetration welds, while expedient, significantly limits fatigue performance under cyclic pressure loading. Adopting improved welding practices and joint designs will yield substantial gains in fatigue resistance, ensuring greater reliability and safety throughout the vessel’s operational life.