A manhole is a confined space in the sense that the regulations intend: restricted access, poor ventilation, a hazardous atmosphere that can accumulate. A 600 m cable tunnel is the same category of space and a completely different monitoring problem.
Assumption one: the hazard is at the entry point
In a vessel, the atmosphere is roughly the thing you measure at the opening, allowing for stratification. In a long tunnel, it is not. Ventilation falls away with distance, heavier gases collect in low sections, and the section 400 m in may carry a different atmosphere from the one at the access shaft. A monitoring unit standing at the entrance reports on the entrance.
Assumption two: radio propagates out
Steel and reinforced concrete attenuate radio badly, and a tunnel is both. This is where the consequence of a wireless-only design shows up: the equipment reports correctly for the first section of the tunnel and then stops, without failing visibly. From the office, a tunnel with no data looks like a tunnel with nothing happening.
Assumption three: the workers are near the entrance
In a compact space, everyone is within sight and earshot of the attendant. In a long tunnel, crews spread along the run, work in different sections simultaneously, and come and go through the same access point across a shift. An attendant counting people through a door knows how many are inside and has no idea where any of them are.
Assumption four: power is available at the workface
Frequently it is not. Tunnel sections under construction or in remote galleries have no local supply, and running mains cable along the tunnel is often a project in its own right. Equipment that assumes a socket at the workface turns deployment into an electrical installation.
What a cascaded system does differently
The design response is to stop treating the tunnel as a point and treat it as a line. Units cascade along the run, extending data and power over a cabled bus, so each additional node covers its own section rather than relying on wireless penetration through the whole tunnel. The bus reaches 1,000 m in total, with 20 m per node as standard and up to 100 m per node where the geometry allows.
Three things follow from using cable rather than radio as the backbone. Coverage is predictable because the cable is either connected or it is not, rather than degrading with distance and construction. Power reaches the internal units over the same bus, so the only supply needed is at the entry point. And positioning becomes possible along the run, because each node has a known location, which converts the headcount question into a location answer.
The questions to ask of a long-space installation
Before specifying for a tunnel, culvert or long gallery, ask what the atmosphere is at the furthest workface rather than at the entrance; what carries the signal when radio does not; where each worker is rather than how many are inside; and where the power comes from at the far end. On most projects where confined space monitoring underperforms, the answer to at least one of those was never asked.