Geotechnical and Foundation Failure

Bearing capacity failure is relatively uncommon in well-engineered structures, because design practice applies generous factors of safety (typically 2.5 to 3.0 on the ultimate capacity) and modern site investigation usually characterises the soil well enough to avoid gross under-design. When it does occur, it is typically associated with inadequate site investigation, unexpected subsurface conditions encountered during construction, or loading that exceeds the design assumption.

The classical bearing capacity formula, developed by Terzaghi in 1943 and extended by Meyerhof, Hansen, and Vesic, expresses ultimate capacity as a function of soil cohesion and friction angle, foundation geometry, and embedment depth. In a post-failure investigation, the investigator back-calculates these parameters from post-failure borings and laboratory tests and compares the resulting capacity with the load at failure. Actual field failures consistently occur at factors of safety well below 1.0 because the soil was weaker than assumed, the load was higher than stated, or the design model did not account for inclined or eccentric loading.

Differential settlement is far more common than outright bearing capacity failure, and its investigation requires a different approach. Settlement is not an event, it is a process. The investigator must reconstruct the time history of settlement from structural damage patterns: diagonal cracking in brickwork, stepped cracking in concrete frames, distortion of window and door frames, survey records if any were taken. Foundation borings taken after the event can show the variation in soil stiffness across the site, explaining why one part of the building settled more than another.

The 1964 Niigata earthquake in Japan produced the first well-documented case of widespread building damage from liquefaction. The Showa Bridge, a multi-span steel girder bridge, collapsed when its pile foundations in the liquefiable river alluvium lost lateral support and several piers toppled. Apartment buildings in the Kawagishi-cho district tilted to angles as great as 80 degrees from vertical while remaining largely intact structurally. The buildings had not failed in a conventional sense; the soil they stood on had failed around them.

The 2010-2011 Canterbury earthquake sequence in New Zealand, centred near Christchurch, produced the most extensively studied modern liquefaction case. The February 2011 aftershock (moment magnitude 6.2) caused severe liquefaction in the Avon River corridor suburbs, with sand ejecta, settlement, and lateral spreading destroying thousands of homes. The New Zealand Geotechnical Database compiled from post-earthquake investigations contains one of the world's largest public datasets of liquefaction case histories, including CPT profiles both from before and after the event in some locations.

Liquefaction investigation combines seismic hazard estimation (characterising the earthquake demand) with soil characterisation (estimating the susceptibility of the soil to liquefy under that demand). The Seed-Idriss simplified procedure, updated by Youd and Idriss (2001) and subsequently by Boulanger and Idriss (2014), uses SPT or CPT data to estimate the cyclic resistance ratio (CRR) of the soil and compares it with the cyclic stress ratio (CSR) induced by the earthquake. Where CSR exceeds CRR, liquefaction is predicted. Post-failure investigations calibrate these calculations against the observed outcomes.

Slope failures occur when the shear stress along a potential failure surface exceeds the shear strength of the soil or rock at that surface. The factor of safety is the ratio of available strength to applied stress. When it drops to 1.0, the slope fails. In post-failure investigation the slope has already moved, so the investigator must work backward: from the failed geometry, the soil properties, and the pore pressure conditions to the pre-failure state.

Back-analysis is the standard tool. The investigator assumes a circular or non-circular failure surface, divides the slope into slices (Bishop, Morgenstern-Price, or Spencer methods), and calculates the factor of safety using the residual strength parameters for the soil (since the soil has already been sheared to residual). The failure surface geometry is adjusted until the factor of safety equals 1.0. The result gives the mobilised shear strength at failure, which can be compared with laboratory test results from samples taken near the failure surface.

Pore pressure reconstruction is often the most uncertain part of a slope failure investigation. Elevated pore pressures from rainfall or from a rising water table are the most common triggering mechanism for landslides that occur in soil with adequate long-term drained strength. If piezometer records existed before the failure, they are invaluable. If not, the investigator must estimate pore pressures from rainfall records, catchment hydrology, and the back-analysis itself.

The Teton Dam on the Teton River in Idaho was completed by the US Bureau of Reclamation in 1975. It was a zoned earth-fill embankment 93 metres high with a silt and silty clay core designed to prevent seepage through the embankment. The reservoir began filling in 1975 and by June 5, 1976, it was nearly at full capacity. Wet spots appeared on the downstream face, grew rapidly, and at about 11:55 am the face slumped and a collapse tunnel was visible from the air. By late afternoon the entire dam had washed out, releasing approximately 309 million cubic metres of water. Eleven people were killed and an estimated 25,000 were left homeless.

The failure investigation was conducted by two independent panels: the Interior Department's review group and the Independent Panel to Review Cause of Teton Dam Failure. Both concluded that the primary failure mechanism was internal erosion (piping) through the core material. The investigation identified two contributing factors. First, the silt-clay core material was highly erodible when seepage flow occurred, lacking the plastic clay content that would have made it cohesive and resistant to particle removal. Second, the contact between the core and the steeply cut abutment rock contained open joints and cracks in the rock that provided an initial seepage pathway.

The sequence of events, reconstructed from eyewitness accounts and the physical evidence of the breach, was: seepage entered cracked rock in the abutment, penetrated the interface between core and abutment, began eroding core material, formed a tunnel that enlarged progressively as more material was removed, and eventually grew to the point where the roof of the tunnel collapsed and the dam face slumped. The accelerating rate from first observation of seeping wet spots to collapse of the face was approximately 90 minutes. Once visible piping begins in an earth dam, the time to failure can be a matter of hours.

Geotechnical failure investigations rely on intrusive site investigation because the failure is underground and the relevant conditions are not accessible from the surface. The primary methods are borings with SPT and undisturbed sampling, cone penetration tests, and laboratory testing of retrieved samples.

• Rotary borings and SPT: borings at strategic locations around the failure zone characterise the stratigraphy and allow samples to be taken for laboratory testing. SPT N-values give a continuous record of soil density and approximate strength that can be compared directly with N-values assumed in the original design.
• Cone penetration test (CPT): a continuous profile of tip resistance and sleeve friction that is more reproducible than SPT and better suited to soft soils. The CPT soil behaviour type index (I_c) classifies the soil continuously with depth. For liquefaction investigation, CPT data feeds directly into the Robertson and Wride (1998) and subsequent procedures for CRR estimation.
• Laboratory testing: unconsolidated-undrained triaxial tests on clay samples give the undrained shear strength relevant to short-term stability. Consolidated-drained tests give the friction angle relevant to long-term drained conditions. In piping investigations, erosion tests such as the pinhole test and the hole erosion test assess the erodibility of core materials directly.
• Groundwater monitoring: installation of piezometers in the failure zone and adjacent areas reconstructs the pore pressure conditions at the time of failure, either directly (if monitoring was in place) or by inference from current piezometric levels and hydrological modelling.

Retaining wall failures follow a predictable set of mechanisms: overturning, sliding, bearing capacity failure of the wall foundation, rotational failure through the retained soil, and internal failure of the wall structure itself. Investigation begins by identifying which mode governed from the post-failure geometry. A wall that has rotated forward about its toe failed in overturning or rotational modes. A wall that translated horizontally without significant rotation failed in sliding or through bearing capacity of the soil in front of the wall.

Pile foundation problems are diverse. Axial capacity failures occur when the pile cannot sustain the load it is asked to carry. Lateral capacity problems arise from horizontal loading from wind, seismic, or excavation-induced forces. Pile integrity failures involve structural damage to the pile itself: defects in the concrete or steel, corrosion, or joints that open under load. Post-failure pile investigation commonly uses non-destructive integrity testing (sonic echo, cross-hole sonic logging), combined with borings alongside the pile to assess the soil conditions and any changes since the original construction.