A crude oil terminal does not have a complicated security problem. It has a very large one. Almost everything worth protecting sits inside a boundary that is long, plain, and a long way from anyone looking at it.
The Site
A crude oil terminal in South Texas. Storage tanks, valve manifolds, pump skids and truck lanes, inside a perimeter that is mostly chain-link running through open scrubland.
The difficulty at a site like this is distance. The perimeter is measured in thousands of feet, and the far end of any given run is nowhere near the nearest camera or the nearest person. A camera covering that much ground is spreading a fixed number of pixels across a very wide scene, so a figure crouched against the fabric at the end of a run, at night, in glare or blowing dust, is a small and low-contrast target. The same geometry applies to a patrol: by the time someone drives the line, whatever happened has had time to finish.
That is the gap. Not an absence of cameras, but the fact that nothing was positioned to notice a breach at the moment and place it occurred.
What We Installed
Fence-mounted detection across the perimeter using IRONCLAD, with camera verification through ヴィダラート.
IRONCLAD mounts directly to the fence fabric and responds to the mechanical signature of the fence being cut or climbed. The perimeter is divided into zones, so an alarm arrives with a location attached rather than as a general alert. Verification is handled by cameras covering the same ground: when a zone reports, an operator is looking at that specific stretch of fence instead of deciding blind or driving out to check.
The change is in what does the noticing. The fence became the detector. The cameras became the confirmation step.
Why It Works Here
The fix matches the failure. The problem was distance, and a fence sensor is indifferent to it. A sensor on the fabric does not need line of sight, lighting, or favorable weather to register that the fence has been cut. It needs contact with the structure it is mounted on. A run three thousand feet away reports exactly the same as one directly in front of the control room.
Site conditions also worked in favor of a fence-mounted approach. Open scrubland gives an approach cover and gives any wide-area system a great deal of moving vegetation to sort through. A fence sensor is not interpreting the whole field. It is monitoring one structure, and the event it cares about is contact with that structure.
None of that makes the cameras redundant. It gives them a job they are good at. Confirming what is happening in a known location is a very different task from being the first thing to notice something anywhere along a perimeter, and the sites that struggle are usually the ones that asked a single layer to do both.
What Similar Sites Should Ask
If your perimeter looks like this one, these questions will tell you whether you have the same gap:
- How far is the farthest point of your fence from the nearest camera, and what does that camera resolve at that distance, at night, in your weather?
- If someone cut your fence right now, at the worst point on the line, what would tell you, and how long would it take?
- Which is the more likely first event here: a cut, a climb, a vehicle at a gate, or an approach under the fence?
- When an alarm comes in, who verifies it, and how quickly can they put eyes on that zone?
- What does the ground do to you: vegetation, wind, blowing dust, wildlife, standing water?
Those answers decide whether fence-mounted detection is sufficient on its own, whether buried detection belongs underneath it, or whether an open-area sensor should cover the ground inside the line. Other deployments are collected in our ケーススタディ.
結論
The problem at this terminal was that the perimeter could be crossed without anything registering it at the time. The solution was to make the fence itself the thing that reports, and to leave the cameras to confirm what it found.
If you have a long perimeter and you are not confident what your current setup would catch at the far end of the run, talk to us.
