Answering One Question on the Drawing
Tracing a circuit by eye across a folded A1 sheet, in a switchroom, in poor light, is where mistakes are made. The drawing is rarely the pristine original: it is a photocopy of a copy, marked up twice, with lines that cross each other every few centimetres.
Steel.vision does not ask you to digitise that sheet first. You point the iPad at the drawing as it is, tap the two points you care about, and the software follows the conductor or the pipe run between them and draws it over the frozen frame. The question it answers is the one you actually have in front of the panel: does this line go there, or does it not?
Where Two Lines Cross
On paper, two lines that cross look identical whether they are joined or simply overlapping. A tracer that follows pixels will happily hop from one circuit onto another at that crossing and report a confident, wrong answer. That is the failure mode that matters, because the answer looks the same as a correct one.
Our tracer prices a turn at a crossing differently from a turn along a conductor. Where the only available route turns across an unrelated line, the path is drawn in amber and reported as unconfirmed, not as a connection. Where the drawing carries a junction dot, meaning the lines genuinely are joined, the trace continues through it. A tee off a busbar traces normally, because that is a real branch.
The result is four distinct answers rather than one: the points are on the same line, they are on separate lines, the only route crosses another line and needs checking, or the tap did not land on a line at all. An engineer can act on each of those. A single "no path" cannot be acted on.
Photocopies, Shadows and Folds
Field drawings are faded, and a sheet photographed by hand is bright in the middle and dark at the corners. A fixed brightness threshold fails both ways: it misses grey lines on a faded print, and it turns the shadowed corners of a handheld photograph into imaginary ink. We threshold locally instead, so a clean print, a photocopy and a shadowed snapshot all reduce to the same thin set of lines.
Every change to that pipeline is checked against a fixed set of drawings whose wiring is known in advance, including crossings, junction dots, tees, breaks, heavy busbars, faded lines and handheld photographs. A change that would report a crossing as a connection fails the suite and does not ship.
What the tracer does
- Two-point trace: follows the line between the two points you tap and draws it over the captured frame.
- Crossing detection: distinguishes a crossing from a junction dot and from a tee, and refuses to call a crossing a connection.
- Degraded-drawing handling: local thresholding for faded prints, uneven light and marked-up sheets.
- Symbol and text reading: tapping a recognised symbol snaps the pin to it; tags read off the sheet open the matching manual page.
- On the device: the trace runs on the iPad with no connection. The photograph and the drawing never leave it.
What it does not do
- It does not digitise your drawing set, build a digital twin, or export a netlist. It answers a question on one sheet.
- It does not verify the state of a valve or a switch. It traces the drawing, not the plant.
- It is advisory. It does not replace the drawing, the permit, or the person signing for the isolation.
- The iPad is a standard, unmodified device. It is not certified for use in a hazardous area, and we will say so in writing to your safety team.
Brownfield Sites, As They Actually Are
Older substations and treatment works run on as-builts that were last redrawn decades ago, and the 3D model that would make this easy does not exist. That is the case the tracer is built for: it needs the sheet and nothing else, so it works on the day you arrive rather than after a digitisation project. The same two taps work on a P&ID to ISO 10628 or ISA-5.1 and on a substation single-line diagram.