Structures stay standing when safety leads the site.

On paper, a permit-to-work system is simple: no work starts until the hazards are identified, the controls are verified, and the right people have signed off. In practice, the system is tested every time a schedule slips. That’s when a permit gets rushed, a verification step gets treated as a formality, or a “temporary” workaround becomes the way things are done for the rest of the shift.

The sites I’ve worked that maintained a strong safety record didn’t have a more complicated permit system than anyone else — they had supervisors who treated the permit as a live document, not paperwork. That means walking the job before signing, re-verifying isolations after a shift handover, and being willing to stop work and re-permit when the scope changes, even when it’s inconvenient.

The discipline isn’t in the form. It’s in the habit of never letting the form get ahead of the actual condition of the site.

H2S is one of the few hazards on a process site that can disable your ability to detect it at the exact concentration where it becomes lethal — it deadens the sense of smell at high concentrations, which is the opposite of how most people expect a warning sign to work. That single fact is why H2S training has to go further than a toolbox talk and a badge check.

In the H2S courses I’ve delivered, the goal was never just to get people to recognize the rotten-egg smell. It was to build the reflex: gas detector checked and worn correctly, escape route known before work starts, and an assumption that any alarm is real until proven otherwise. Complacency around H2S usually isn’t a knowledge gap — it’s familiarity. People who’ve worked around it for years without incident start treating the risk as smaller than it is.

Good H2S programs fight that familiarity deliberately, with refreshers, drills, and supervisors who model the same caution on day one thousand that they did on day one.

Every HSE engineer wants to point to a zero-incident record. But a number on its own doesn’t tell you whether a site is safe or whether people have simply learned to stay quiet. The sites where I’ve seen the strongest actual safety culture were rarely the ones with the cleanest-looking logs early on — they were the ones where near-miss reporting went up after leadership visibly acted on the first few reports instead of looking for someone to blame.

That’s the real test of a safety culture: what happens the first time a worker reports something that makes a supervisor or a schedule look bad. Handle that well, and reporting becomes a habit. Handle it badly once, and it stops instantly — and the incident log starts measuring silence instead of safety.

Toolbox talks, campaigns, and audits all matter. But culture is built in the moments no one else is watching how leadership responds.