The decision at stake
The structure showed cracks at two welded nodes. The alternative on the table was replacing the whole module, with the stoppage that implies. The question was whether a local reinforcement would hold for the remaining service life.
Geometry and mesh
Beam model for the global structure and a solid submodel at the two cracked nodes. The welds were represented with enough detail to read the stress concentration; the remaining joints were left rigid.
Physics and assumptions
Modal analysis to rule out resonance with the operating frequency, and fatigue analysis with the load spectrum measured on site over two weeks. Sound material was assumed outside the cracked zones, which was verified by non-destructive testing before the conclusion was delivered.
Convergence
Mesh convergence checked on the peak stress of the submodel: the change between the second and third refinement levels stayed below 2 %.
Validation
Compared against strains recorded by gauges at two points of the structure, with a deviation under 6 %. The computed modes match those measured in the vibration test.
Conclusion and limits
Two reinforced nodes and an inspection plan, instead of the replacement. The conclusion assumes the measured load spectrum represents operation; a change of ore or feed rate calls for a new measurement.