It is natural to look inside the trench after a collapse, because that is where the failure surface is. But a substantial share of excavation failures involve soil no weaker than the classification assumed. What changed was the loading: a spoil pile placed at the crest, an excavator tracking along the edge, pile driving on the next block, a dewatering pump that stopped overnight, a second excavation opened alongside the first. These act on the wall from outside it, they are frequently documented by someone, and they are analysed differently from a soil-strength question.
What a protective system is rated against
Lateral earth pressure on a trench wall increases with depth, and the tabulated data and shield ratings under 29 CFR 1926 Subpart P are built around that distribution for a given soil class and depth. A surcharge at the ground surface adds an increment on top of it, spread over the height of the wall.
That is why surcharge is treated separately rather than folded into soil type. Ratings generally assume no surcharge unless they say otherwise, so a system correctly selected for the depth and class can still be loaded past its basis by what sits beside it.
Spoil and stored material at the edge
Subpart P requires that employees be protected from loose rock or soil that could fall or roll in, by keeping excavated material and equipment at least two feet from the edge, by retaining devices, or by both. That two-foot line is a minimum for the falling-material hazard, not a statement that a pile placed there imposes no surcharge.
Spoil placement is among the easier facts to establish afterwards. It shows up in site photographs, in aerial imagery, in the pile still present after the collapse, and in the plain logic of where a machine could have cast material given the working width.
Equipment, deliveries and traffic
Excavators, pipe trailers, concrete trucks and road plates all bear on the crest, and unlike spoil they arrive and leave. The load that mattered may have been present for twenty minutes. Subpart P separately requires a warning system — barricades, signals, or stop logs — where mobile equipment operates adjacent to an excavation and the operator lacks a clear view of the edge.
Reconstructing transient loads leans on timestamps rather than the site as found: delivery tickets, telematics, equipment logs, dashboard and phone video, daily reports.
Vibration as its own mechanism
Vibration appears explicitly in Appendix A, which bars a soil from Type A where it is subject to vibration from heavy traffic, pile driving or similar effects. The exclusion reflects real behaviour: cyclic loading densifies loose granular material and can drive up pore pressure in saturated sands, reducing the effective stress that gives the soil its strength.
Sources are often off the contractor's own site. Where a neighbouring project monitored vibration for its own purposes, those records exist independently and predate any dispute.
Groundwater, seepage and base instability
Water changes a trench in more than one way. Seepage through the face erodes fines and undercuts the wall. Rising pore pressure lowers effective stress and with it shear strength. An upward gradient at the bottom of the cut produces boiling and heave, destabilising the base beneath a shield rather than the face beside it.
The visual signatures differ. A wall that sloughed from a seeping horizon does not look like a base that boiled, and neither looks like desiccated cohesive soil toppled along a tension crack. Photographs taken before recovery work usually preserve that distinction.
Dewatering is a system with a service history
Subpart P prohibits work in excavations with accumulated water unless adequate precautions are taken, requires water removal equipment to be monitored by a competent person, requires surface water to be diverted away, and requires re-inspection after rainstorms. Each creates a record. Pump run-time logs, piezometer readings, well-point installation records and rainfall from a nearby station together bound when water was and was not being controlled.
Adjacent structures and undermining
Excavation removes lateral support from whatever sits beside it. Subpart P requires support, bracing or underpinning where an excavation may endanger the stability of adjoining buildings, walls or other structures, and does not permit sidewalks or pavements to be undermined without equivalent protection.
The failure can run either way. The adjacent foundation may settle while the trench stands, or the loss of support may mobilise a wedge that arrives in the trench. Which came first is usually answered from crack patterns, survey data, and the geometry of the deposited material.
Utilities crossing the excavation
Subpart P requires underground installations to be located before opening an excavation, owners to be contacted, and installations to be protected, supported or removed while the excavation is open. An unsupported line spanning a trench is both a load and a hazard. It is also a potential water source: a pressurised main or leaking sewer saturates material that design treated as drained, and the leak may predate the excavation by years.
Building the loading timeline
External-load questions resolve into a chronology rather than a single measurement: when the trench was opened, where spoil went, what equipment worked where, what rain fell, when pumps ran, and what changed in the hours before the wall moved.
That chronology is assembled largely from outside the contractor's own paperwork — weather records, neighbouring projects, one-call tickets, telematics, imagery. Their independence is the point.
This article is general technical orientation, not a failure analysis, an engineering opinion, or advice on any specific matter. Determining the cause of a particular incident requires hands-on examination by a credentialed expert.