Excavation Support Failure Next to a 60-Year-Old Masonry Building
You are the reviewing engineer for a seven-storey office development in Colombo. The basement requires a 7.5 m deep excavation. Four metres from the crest stands a three-storey load-bearing masonry building about 60 years old, founded on shallow strip footings 1.2 m below ground, with a pre-existing vertical crack 0.5 mm wide that has been stable for years. Ground conditions are 3 m of loose silty sand (SPT N = 6 to 9) over a stiff to very stiff clay, with groundwater at 2.5 m. The approved scheme is 600 mm diameter bored soldier piles at 1.5 m centres with two rows of ground anchors, each row stressed before excavation proceeds below it. Instead, the contractor excavated the full 7.5 m in one lift over a weekend and installed no upper anchor row. On Monday the building shows a new diagonal crack 1.2 mm wide, the ground behind the crest has settled 18 mm, and a plumb survey shows the building has tilted 1 in 900 towards the excavation. Work has stopped. The contractor proposes to dump backfill into the excavation with a excavator, and the client asks whether the building is safe for occupancy tomorrow.
1. State your immediate actions in the first 24 hours, in order of priority, and justify each.
Read the worked answer
1. Protect life first, not the programme. A 1 in 900 tilt plus 18 mm crest settlement plus a new 1.2 mm diagonal crack on a 60-year-old unreinforced masonry building is structural damage, not cosmetic. Recommend immediate partial evacuation of the building and exclusion of the zone within a distance equal to the excavation depth (7.5 m) from the crest. Notify the client, the local authority and, if the building is occupied by others, the occupants in writing. Do not let the contractor's backfilling proposal become the emergency plan without control.
2. Freeze the destabilising action. Do not allow uncontrolled dumping of backfill: a excavator bucket dropping soil against the toe of unsupported piles shock-loads the piles and the retained soil, and does not restore the lost anchor restraint. Controlled backfill is acceptable, but as a measured, compacted, layered operation against the face, placed in a way that supports rather than pushes.
3. Instrument before you touch anything else. Install crack gauges across the new and old cracks, level pins on the building corners and along the crest, and an inclinometer in the pile line or in the soil behind the crest. Take a baseline set immediately and repeat every 4 hours initially. Plumb survey the building on two faces to confirm the 1 in 900 tilt and locate its direction.
4. Shore the building from its own side. Temporary raking or flying shores on the side away from the excavation resist further rotation without adding load on the excavation face.
5. Restore the missing restraint properly. Backfill to about 3 m depth, install and stress the missing upper anchor row through the piles at the designed load, then re-excavate in lifts not exceeding 1.5 m, stressing each anchor row before going below it.
Answer to the client: the building is not safe for normal occupancy tomorrow. Occupancy can resume only when crack widths and tilt have been stable for at least 7 days under full instrumentation and a structural assessment of the masonry has been made.
2. The contractor says the 18 mm settlement is 'minor'. Set out the monitoring regime and trigger values you would impose, explaining how each value was chosen.
Read the worked answer
What to measure and why
Settlement of the building: precise levelling on at least six points (four corners plus mid-walls). This is the direct measure of damage potential because masonry cracks on differential movement, not absolute settlement. For a masonry building of this type, the damage threshold in terms of angular distortion (deflection ratio) is roughly 1/500 for the onset of visible cracking and 1/300 for structural concern. With a 15 m long building, 1/500 corresponds to about 30 mm of differential settlement across the length, so 18 mm uniform is not yet severe, but 18 mm concentrated over 4 m near the excavation face already gives about 1/220 locally.
Lateral movement of the retained soil: inclinometer behind the crest, read to the pile toe. For a 7.5 m supported excavation in loose sand over stiff clay, a well-run job would show crest movement of the order of 0.1 to 0.3 percent of the depth, i.e. 7 to 22 mm total. The 18 mm already sits at the top of that band without any anchor installed.
Crack width: tell-tale gauges on all cracks, existing and new.
Anchor loads: load cells on at least 10 percent of anchors, to catch load loss.
Groundwater: standpipe piezometers behind the crest; a rising water level in the sand would indicate the excavation has created a drainage path and could trigger consolidation settlement of the clay.
Suggested trigger values
| Parameter | Alert | Action (stop work) |
|---|---|---|
| Building settlement rate | 1 mm/day | 3 mm/day, or 10 mm total |
| Lateral crest movement | 10 mm | 20 mm, or rate 2 mm/day |
| Crack width | 1.5 mm | 3 mm, or new crack appearing |
| Angular distortion | 1/1000 | 1/500 |
The action level is not a prediction; it is a commitment to stop and redesign. Readings must be taken daily during active excavation and reduced and reviewed the same day by a named engineer, not filed until the next site meeting.
3. The building turns out to be stable after remedial work. How would you apportion responsibility, and what design and supervision changes would you require before work restarts?
Read the worked answer
Responsibility does not turn on who held the shovel. The sequence of anchors being stressed before excavation proceeds below each level is a design requirement, not a matter of site preference: without the upper row, the 600 mm piles act as 7.5 m cantilevers fixed only at the toe in stiff clay, and the retained sand mobilises far more deflection than a propped or anchored wall. So the contractor departed from the approved method, and the site supervision allowed it over a weekend with no hold point. Both the contractor's quality system and the engineer's supervision failed.
However, the designer is not blameless if the drawings said 'install anchors' without stating the maximum unsupported excavation depth at each stage, or if no hold point was defined that required the engineer's written release before excavating below the first anchor level.
Changes to require before restart
Method statement with hold points. Every lift must be signed off by the engineer before the next lift; no excavation below an unstressed anchor row.
Staged excavation with a maximum unsupported depth stated on the drawing (for example 2.0 m below each stressed anchor row), plus a limit on the length of open trench in plan.
Anchor testing regime: proof testing of a specified percentage of anchors to 1.5 times the design load with load-extension records, plus a load cell on a nominated set of anchors.
Observational method formalised. The design should state the anticipated movement range, the trigger values, and the pre-planned response to each trigger. This converts the monitoring from a reporting exercise into a decision tool.
Independent review of the retained heights and anchor capacities using the measured soil parameters and the revised sequence, including the possibility that the clay is fissured and weaker than the borehole suggests.
Agreement with the adjacent owner on a condition survey (already partly done), monitoring access, and a damage threshold, recorded before restart.