Site survey and area-classification check
Every deployment starts with a survey against your area-classification drawings and camera estate, before any hardware is proposed.
Industrial safety
A hydrocarbon site is where camera-based safety monitoring is most valuable and hardest to deploy honestly: the hazards are severe, the areas are classified, the work runs under permits, and most of the people on site work for someone else. This guide covers what that means for a deployment, written for HSE and operations owners rather than for technology enthusiasts.
Industry guide Published 6 min read
01The terrain
Three facts separate a refinery, gas plant or petrochemical complex from the general industrial case. The consequences are severe, which changes the value of early detection and the intolerance for false confidence alike. Large parts of the site are classified hazardous areas, which constrains what hardware can exist there and how anything gets installed. And high-risk work runs under a permit-to-work regime, hot work, confined space entry, lifting, which gives detection a role no brochure mentions: verifying that the conditions a permit assumed are actually holding while the work happens.
The vocabulary matters too. This page says permit to work, SIMOPS, turnaround and classified area, because that is the language of the teams who own the problem; the AI is the mechanism, not the subject.
02What Interkey delivers
Every deployment starts with a survey against your area-classification drawings and camera estate, before any hardware is proposed.
Detections built to carry permit context and land in the workflow that can act, integrated per site against the systems you already run, in-house.
Footage governance decided by the operator, with processing kept in-Kingdom and on-premise.
03Where cameras help
An evaluation framework, not a promise sheet: which detections are worth piloting on a hydrocarbon site, what each amounts to as a detection problem, and what drives its false alarms. Scope a pilot around one or two of these, not all of them.
| Scenario | What detection amounts to | What drives false alarms |
|---|---|---|
| PPE compliance at zone entry | Person detected, required equipment absent, in a defined zone | Camera angle and distance; transitional walkways where PPE is legitimately not yet on. The strongest starting case; covered in depth here. |
| Restricted and red-zone entry | Person present where no person should be, absolutely or during an operation | Zone boundaries drawn tighter than the site actually moves; authorised entries not modelled. The zone map is the tuning work. |
| Person–vehicle proximity | Person and moving plant within an unsafe envelope | Depth ambiguity from single cameras at long range; view geometry decides feasibility per location, honestly assessed in the survey. |
| Hot work oversight | Activity signatures at a permitted location; absence of the fire watch the permit requires | Distinguishing permitted from unpermitted activity requires integration with the permit system, which is precisely what makes it valuable. |
| Confined space entry support | Presence of the attendant at the entry point; entries and exits counted | Camera coverage of the entry point itself; interior visibility is usually not available and should not be promised. |
| Suspended load / lifting exclusion | Person beneath or inside a lifting exclusion zone during an operation | Dynamic zones that move with the lift; needs either operational discipline about lift areas or integration with the lifting plan. |
04The hard constraint
Zones where flammable atmospheres can occur are classified, and everything electrical inside them, cameras and compute included, must be certified for the zone it occupies under the hazardous-area regimes the site operates (ATEX and IECEx are the marking schemes engineers will recognise). Three practical consequences for a monitoring project. Certification is a datasheet fact about specific hardware: it is verified against your area classification drawings during the survey, never assumed. Existing cameras inside classified areas were procured under exactly this constraint, which is an argument for reusing them well. And where a needed view does not exist inside a classified area, the deployment plan prices certified hardware and its installation regime honestly, or places the camera outside the zone looking in, which is often the better answer.
Installation itself follows the site’s rules, not the integrator’s preferences: work permits for every task, approved cabling and power routes, and scheduling that respects operations, which on a live plant frequently means aligning physical work with maintenance windows and turnarounds. A deployment plan that has not asked about your turnaround calendar has not been written for a hydrocarbon site.
Severity decides the order, not the org chart
Consequence severity sets the priority order here, not familiarity: permit-to-work verification and hazardous-area constraints get engineered around first, because a false sense of coverage in either is worse than no camera at all. Contractor-crew fairness and a transparent monitoring policy follow close behind, because a programme the workforce does not trust does not survive contact with a shift change.
05The unclaimed ground
A permit asserts conditions: the fire watch is posted, the area is cleared, the attendant is at the confined-space entry, barriers are in place. Between issue and close-out, those conditions are verified by whoever happens to walk past. This is the verification gap, and it is where camera-based monitoring does something genuinely new: continuous observation of whether the permitted conditions are still true, missing fire watch, entry without an attendant present, activity where SIMOPS restrictions say there should be none, raised as events while the work is happening rather than found in the investigation afterwards.
The engineering honesty: this only works integrated with the permit and incident systems the site already runs, so detections carry permit context and land in the workflow that can act. Those integrations are built per site against the systems actually in use, named as targets during scoping, demonstrated before being relied on, which is systems integration work Interkey does in-house, not a connector logo on a slide.
06Multi-employer reality
Most people on a Saudi hydrocarbon site on any given day work for contractors, and every HSE manager knows the asymmetry that creates: the operator owns the consequences while supervision of the work is distributed across employers whose safety cultures vary. Camera-based detection is usefully indifferent to badges, a missing hard hat is a missing hard hat, which makes it one of the few observation mechanisms that covers the whole workforce evenly.
The design work is the workflow: detections involving a contractor crew routed to that contractor’s supervision as well as the operator’s control room, and the per-contractor pattern feeding the contractor-performance reviews the site already holds.
Responsible for HSE on a hydrocarbon site and weighing automated monitoring? Interkey scopes pilots around one unit, your camera estate, your permit regime and your contractor reality, starting with a site survey that respects how a live plant actually works.
07Buyer questions
Direct answers to the questions that come up in real evaluations. Anything missing, ask us at the bottom of the page.
Only hardware certified for the specific zone, installed under the site’s permit and electrical rules, which is why the survey checks your area classification drawings before anything is proposed. Often the better engineering answer avoids new certified hardware altogether: existing in-zone cameras reused, or new cameras placed outside the classified boundary looking in.
No, and be wary of anyone blurring that line. Fire and gas systems are safety-instrumented, certified for their function, and stay exactly as they are. Camera-based monitoring addresses a different layer, human behaviour and work conditions: PPE, zones, proximity, permit conditions, as a complement to the engineered safety systems, never a substitute.
By using existing cameras and out-of-band processing first, nothing in the pilot touches process systems, and by scheduling any physical work under the site’s normal permit regime, aligned with maintenance windows. The pilot guide covers scope and success criteria; the plant-specific part is simply that the pilot plan respects the turnaround calendar and the permit office from day one.
Five questions filter the field: Which of these scenarios have you deployed, and what were the false-alarm drivers? How do you handle classified areas, specifically? How do detections integrate with our permit and incident systems, demonstrated, not asserted? Where is footage processed and stored, and does it stay in-Kingdom? And what happens in the tuning weeks after go-live, staffed by whom? Vague answers to any of the five are your finding.
From the insights library
The strongest starting scenario in depth: detectability, placement, false positives and privacy.
Evaluation guideSuccess criteria and measurement discipline, written to be attached to a pilot agreement.
ProductThe detection product behind the scenarios on this page, and the delivery work around it.
Next step
If the site is hydrocarbon-processing, say so, along with where the classified areas sit — it changes the design from the first conversation.
Or directly
+966-11-2180999 info@interkey.com.saTawuniya Towers, North Tower, 7th Floor, King Fahad Highway, Olaya, P.O. Box 56835, Riyadh 11564, Saudi Arabia
Elsewhere