Visual inspection
What is visual inspection: visual inspection explained, and where its limits honestly sit
Visual inspection is the examination of a component, structure or asset by eye, unaided or with optical aids, to detect surface conditions such as cracks, corrosion, leaks, distortion, wear, contamination and missing or incorrect parts, assessed against defined acceptance criteria. Almost every other inspection method is layered on top of it.
In non-destructive testing the formal name is visual testing, abbreviated VT. It sits alongside penetrant, magnetic particle, ultrasonic, radiographic and eddy current methods in ISO 9712, the international standard for qualifying and certifying NDT personnel (ISO 9712:2021). So it is not an informal walk round with a torch. It is a method with a written procedure, personnel levels and acceptance criteria.
What does visual inspection actually detect?
Surface conditions. That is the whole of it, and it is worth being blunt about the boundary.
- Geometry and fit: distortion, misalignment, weld profile, undercut, incomplete fill, dimensional deviation.
- Surface-breaking damage: cracks large enough to resolve at the viewing distance, gouges, scoring, impact damage.
- Degradation: corrosion, erosion, coating breakdown, blistering, fouling and deposits.
- Leakage and its traces: active leaks, staining, tracking, residue trails.
- Assembly errors: missing fasteners, wrong parts, loose fittings, damaged seals, incorrect marking.
What it cannot do is look inside. A subsurface inclusion, lack of fusion in a weld root, internal wall loss in a pipe: none of that appears on the surface until it is late. That is why visual is the first step and rarely the last, and why the other non destructive testing methods exist.
Direct versus remote visual inspection (RVI)
Direct visual inspection means an uninterrupted optical path from the inspector's eye to the surface, with at most a mirror, lens or light in between. ISO 17637, the standard for visual testing of fusion-welded joints, puts numbers on the conditions: illuminance on the surface of at least 350 lux with 500 lux recommended, the eye within 600 mm of the surface, and a viewing angle of not less than 30 degrees (ISO 17637:2016). Those three numbers are the difference between an inspection and a glance.
Remote visual inspection, usually shortened to RVI, is what you use when a person cannot get an eye there: borescopes, videoscopes, crawlers, pole cameras, drones and fixed cameras. ASME Boiler and Pressure Vessel Code Section V, Article 9 permits remote in place of direct on one condition, that the system is demonstrated to have resolution and lighting at least equivalent to direct viewing, referenced against targets such as ISO 12233 or the USAF 1951 resolution target (ASME BPVC Section V). Equivalence is the test, not convenience.
That single condition is the bridge to the rest of this subject. A borescope inspection reaches inside an engine casing, a drone inspection reaches a chimney flue, and both are still visual inspection, governed by the same acceptance criteria as if a person were standing there.
Why two inspectors grade the same surface differently
Reliability is the uncomfortable part of the method. In a benchmark experiment run for the FAA at Sandia National Laboratories' Aging Aircraft NDI Validation Center, twelve airline inspectors performed set visual tasks on a Boeing 737 airframe. Aviation experience and a timed measure of visual acuity both correlated significantly with performance, and video analysis of the inspections separated two distinct failure modes: search errors, where the inspector never looked at the defect, and decision errors, where they looked at it and called it acceptable (Spencer, Visual Inspection Research Project Report on Benchmark Inspections, DOT/FAA/AR-96/65, 1996).
Lighting, access, eyesight, experience and time on task all move the result. None of that is a reason to distrust the method. It is a reason to control the conditions and record them.
Visual inspection explained: a practical example
A surveyor inspects a rooftop plant deck. Direct visual from the walkway finds surface corrosion on two pipe supports and a split flexible connector. The underside is inaccessible, so a pole camera is used and its footage is checked for resolution against a known reference before it is accepted as evidence. The report records 480 lux measured at the surface, the viewing distance, and the findings graded against the client's condition scale. Wall thickness is not called, because visual cannot measure it, so ultrasonic readings are raised as a follow-on task.
What separates a reliable visual inspection from a look
- A written procedure and acceptance criteria, agreed before anyone attends, so findings are graded rather than described.
- Measured lighting, not assumed lighting. Bring a light meter if the standard specifies a level.
- Real access, at the distance and angle the standard requires, not the nearest place you can stand.
- Current visual acuity records for the inspector, near-vision tested annually under most certification schemes.
- Surface preparation, because paint, dirt and product residue hide exactly the features you are looking for.
- An inspection frequency that reflects risk, rather than whatever interval was set when the asset was commissioned.
What it is commonly confused with
Two confusions cause most of the friction. The first is treating visual inspection as a substitute for volumetric methods. It is not, and no acceptance standard says otherwise. The second is treating remote visual as a lesser thing than direct. It is equal when it is demonstrated to be equal, which is a testable claim.
The practical consequence is that remote visual has quietly become the routine option wherever access is the real cost. Where the images come from someone who is already on site rather than an inspector who has to travel, structured capture tools such as Venta Capture, a product of VentaVid, guide the person holding the phone through the shots an inspector needs and record the timestamps and device context alongside them. They collect evidence for a qualified person to judge, they do not make the call, and where a standard specifies direct viewing, they do not satisfy it.
