Langford Analytic · Knowledge Base

Support Structure Failure Under Seismic Loading

A case study of a support structure failure under seismic loading, examining the failure of a welded connection and the role of stress concentration.

Seismic7 min read
seismicsupport failurecase studywelded connectionstress concentration

Case classification

Illustrative case study. This scenario is constructed from common patterns in welded support structure failures observed in industrial facilities after seismic events.

Scenario description

A steel support frame for a pressure vessel was constructed from square hollow sections welded to baseplates. The frame was 3 m tall and supported a vessel weighing 2000 kg at the top. During a seismic event, a weld at the junction of a vertical member and the baseplate cracked, causing the frame to lose one support column. The frame did not collapse because the remaining columns carried the load, but the vessel shifted and the attached piping was damaged.

Analysis of the failure

The seismic analysis of the frame had been performed using response spectrum analysis with SRSS modal combination. The peak stress at the weld was reported as 180 MPa, below the allowable of 250 MPa. However, the stress was recovered from the element stress at the weld location, which did not account for the stress concentration at the weld toe. The actual stress at the weld toe, with a stress concentration factor of approximately 2.0 for a fillet-welded connection, was approximately 360 MPa — exceeding the allowable. The weld failed at the toe, initiating a fatigue-like crack that propagated under the cyclic seismic loading.

Root causes

  • The stress concentration at the weld toe was not accounted for in the analysis — the nominal element stress was compared directly with the allowable
  • The weld detail had a high stress concentration factor due to a sharp transition between the tube and the baseplate
  • The seismic loading was cyclic, causing crack initiation at the stress concentration and propagation over the duration of the event
  • The weld was a partial-penetration weld with a smaller throat than assumed in the design
  • The support frame was stiff, attracting high seismic forces through its short, rigid members

Lessons learned

  • Stress concentrations at weld toes must be accounted for — the nominal FE stress must be multiplied by the appropriate stress concentration factor
  • Weld details should be designed to minimise stress concentration — using full-penetration welds, grinding the weld toe, or adding gussets to reduce the stress concentration
  • Cyclic seismic loading can cause crack initiation and propagation at stress concentrations, even if the stress is below the static allowable
  • The weld throat must be confirmed against the design assumption — partial-penetration welds have less capacity than full-penetration welds
  • Stiff support frames attract high seismic forces; the frame stiffness affects the seismic demand

Remediation

The failed weld was replaced with a full-penetration weld with a ground toe. Gusset plates were added at the base connections to reduce the stress concentration and provide a redundant load path. The frame was re-analysed with stress concentration factors applied to the weld locations, confirming that the revised detail had adequate capacity. The weld inspection specification was updated to require ultrasonic testing of critical welds.