Pre-ticked for trenches, mobile plant and buried services, with three rows written. The two things that kill on excavation work are what is already in the ground and what is moving above it.
Locate, then prove, then dig
Drawings tell you what somebody recorded. A cable avoidance tool tells you what it can detect. Neither tells you what is actually there, which is why the third step is a trial hole dug by hand or with vacuum excavation to put eyes on the service before a bucket goes near it.
Skipping straight from the scan to the machine is the sequence that strikes cables, and the scan almost always looked fine beforehand. Plastic water main, a fibre duct with no tracer, an unrecorded supply to a shed that was demolished in 1994: none of them read on the instrument.
The trench is also about the air
People tick the trench limb, which is about depth, and miss that a deep excavation can also be a confined space, which is about the atmosphere. They are separate limbs and both can apply to the same hole.
Heavier-than-air gases collect in the bottom of an excavation and stay there. A leaking gas service nearby, exhaust from a pump or a generator sitting at the edge, or ground contamination from a previous use all end up where the person is working, and the air at the top of the trench tells you nothing about the air at the bottom.
Support goes in as you dig, never afterwards
A shoring system installed into an already open trench requires somebody to be in the unsupported trench to install it, which is the exact exposure the system exists to remove. Battering, benching or shoring is designed for the ground and the depth, and it goes in as the excavation progresses.
The other half is the inspection: before each shift, after rain, and after any alteration, by a competent person, recorded. Ground that was fine on Friday is a different material on Monday, and the collapse that follows a wet weekend is the most predictable event on any groundworks job.
Separate the people from the plant
Reversing alarms, high visibility clothing and a banksman are all controls that depend on somebody paying attention at the moment it matters. Physical separation does not, which is why it comes first: a segregated pedestrian route, a one-way traffic plan, barriers rather than paint.
Where somebody genuinely has to work inside the machine envelope, the machine stops and the operator makes eye contact before they approach. Not a wave from thirty metres, and not a radio call. Eye contact, then movement.
Water changes the ground and the plan
A trench that stood open all week can fail on Monday because it rained on Saturday. Water raises the pore pressure in the soil, adds weight, and washes fines out from behind a face that looked solid, and none of that is visible from the top.
So rain is a named trigger for re-inspection rather than a general caution, and the pumping arrangement is part of the design rather than a hire decision. Discharge matters too: pumping trench water onto the ground beside the excavation feeds it straight back in, which is one of the few ways to make an excavation worse while actively working on it.
Groundwater found during the dig is a different problem again, because it means the design assumptions about the soil were made on the wrong material. That is a stop and a conversation with whoever specified the support, rather than something to pump through.
Common questions
Does every excavation need a SWMS?
A shaft or trench deeper than 1.5 metres does, and so does any excavation where powered mobile plant is moving in the area, which is most of them. A shallow hand-dug trench with no plant nearby may sit outside the list, and the general duty still applies to it in full.
Is 1.5 metres the depth at which I need shoring?
No, and it is the most expensive misreading in this trade. It is the depth at which a SWMS is required. Ground support is decided by the ground conditions and the competent person, and trenches well under that figure have collapsed and killed people, particularly in made ground or after rain.
What counts as a competent person for the inspection?
Somebody with the knowledge and experience to recognise the ground conditions, spot the signs of instability and act on them, with the authority to stop the work. It is not a certificate and it is not automatically the most senior person present. The authority half matters as much as the knowledge half.
Do I need a separate SWMS for the plant?
The excavation statement should cover the plant working with it, because separating them produces two documents that each assume the other has dealt with the interface, which is exactly where people get hurt. What does need its own is anything the plant does elsewhere on site.
How close can a spoil heap sit to the edge?
Spoil and plant are kept back from the edge by at least the depth of the excavation as a working rule, because the surcharge from material at the edge is what turns a marginal trench wall into a collapse. It is one of the few figures on a groundworks job that is easy to check by eye.