Construction Verification & QA

The defining construction-phase use of reality capture is scan-versus-BIM comparison. You capture the building as it has actually been constructed -- a floor of structure, a poured slab, an installed services zone -- and you overlay that as-built point cloud onto the design BIM model that says where everything was supposed to be. Software then computes the difference between the two surfaces, point by point, and typically presents it as a deviation map: a colour-coded picture in which areas that match the design sit in one band, small tolerable differences in another, and anything drifting out of tolerance in a warning colour. In one view you can see not just *that* something is off but *where* and *by how much*.

This is a profound change from traditional quality checking. A tape-and-eye inspection samples a handful of points and relies on the inspector noticing the problem; scan-versus-BIM measures effectively the whole surface and quantifies every deviation, so a bowed wall, a slab that is low in one corner, a misplaced embedment or a column drifting off plumb shows up as measured fact rather than as someone's impression. The result is measured evidence: a record, dated and quantified, of what was actually built against what was designed. That evidence is valuable for the work itself and as a defensible account if questions arise later about quality or responsibility.

The honest framing matters as much here as anywhere. The comparison is only as good as the capture's accuracy and the model's correctness: a scan captured to a few millimetres cannot adjudicate a one-millimetre tolerance, and if the design model is itself wrong the deviation map measures the wrong thing. The colour bands are a visualisation tuned to chosen thresholds, not an oracle -- what counts as 'in tolerance' is set by the project's specification, not by the software's default palette. And deciding whether a measured deviation is acceptable, must be reworked, or can be accommodated is an engineering and contractual judgement. Where a verification must be *binding* -- setting-out that legally governs position, a survey that certifies compliance, a measurement that settles a contractual dispute -- that is a licensed surveyor's and the relevant engineer's responsibility, working to the specification and verified methods. Capture gives you an extraordinarily powerful, comprehensive measured check; it does not dissolve the professional judgement about what the measurements mean.

The reason verification during construction pays so well is the brutal economics of *when* a problem is found. A deviation caught immediately after the work is done, before any following trade has built upon it, is cheap to correct -- the element is still accessible, nothing covers it, and the downstream schedule has not yet absorbed it. The same deviation discovered weeks later, after finishes, services and other trades have proceeded on the assumption that it was right, is expensive: the correction now means opening up completed work, disrupting other trades, and often a cascade of secondary fixes. And the worst case -- discovery at handover or, for prefabricated work, when the factory-made part arrives and will not fit -- turns a small measured error into a programme-wrecking dispute.

Regular capture through construction attacks this directly by shortening the time between a deviation happening and being detected. Scan a floor when its structure is complete and compare to the model, and a misplaced embedment or an out-of-position wall is flagged while it is still the cheapest thing in the world to fix. This is particularly critical where later work depends precisely on earlier work being right -- the classic case being prefabrication, where a facade unit or a module is being manufactured off-site to dimensions that assume the frame is where the design put it. Verifying the frame by capture *before* the part is fabricated, rather than discovering the mismatch on delivery, is exactly the kind of early catch that justifies the whole exercise.

There is real discipline in doing this well rather than theatrically. Verification has to be timed to the decision points -- scanning at the moments when catching a deviation still changes the cheap-versus-expensive outcome, not randomly or only at the end when it is too late to matter. It has to be targeted to what is critical -- the interfaces and elements where deviation genuinely cascades, rather than exhaustively scanning everything and drowning in data. And it has to feed a response: a deviation map that nobody acts on has caught nothing. The value is not in the scanning; it is in the early, measured, acted-upon catch. Used that way, construction verification converts the most expensive category of construction error -- the late discovery -- into the cheapest -- the early fix.

Beyond checking correctness, capture during construction answers a second recurring question: how much has actually been built, and are we on plan? Periodic scans of the site -- weekly, at milestones, at each floor -- create a dated sequence of measured snapshots of the work in place. Compared against the planned model and programme, these reveal progress as measured fact: which elements exist, which do not yet, and whether the build is ahead, on, or behind where it should be at that date. This is progress monitoring, and it replaces the familiar fog of optimistic site reports with something you can see and measure.

The value is partly in accuracy and partly in objectivity. A measured progress record is harder to argue with than a verbal claim, which matters for payment applications tied to work genuinely completed, for spotting slippage early enough to respond, and for keeping distant stakeholders honestly informed about a project they cannot walk daily. On large or fast projects, a regular capture cadence -- increasingly from quick mobile or handheld scanning, and on open sites from drones where regulations permit -- turns progress tracking from a subjective weekly ritual into an evidence-based one. It also dovetails with the verification use: the same capture that tells you *how much* is built can tell you *whether it is built correctly*, so a single scanning routine serves both progress and quality.

The usual honesty applies and then some. Progress capture is a measure of geometry in place, not of everything that matters -- it does not see work hidden behind what is captured, quality of workmanship beyond position, or commissioning status, so it informs judgement rather than replacing site management. Its accuracy and completeness are bounded by the capture method and by occlusion, so a fast mobile scan good enough to confirm a floor exists is not the same as a survey-grade check of where it is. Drone-based progress capture on an open site sits under the national drone regulations and must be flown lawfully. And wherever a measurement crosses into binding territory -- a certified quantity for payment, a legally significant position, a dispute resolution -- it returns to the domain of the quantity surveyor, the licensed surveyor and the verified method, not the quick monitoring scan. Within those limits, measured progress monitoring is one of the most practically useful, and increasingly routine, construction applications of reality capture.

Stand back and the common thread through all of this is quality assurance built on measured evidence. Traditional construction QA leans heavily on inspection, judgement and documentation that is often qualitative -- a photograph, a sign-off, an inspector's note. Reality capture adds a layer of hard, comprehensive, dated measurement: a record of what was actually built, where, and how far from the design, captured densely rather than sampled, and preserved as a defensible account. This shifts quality conversations from assertion to evidence -- from 'it was built correctly' to 'here is the measured as-built against the design, and here is the deviation, quantified'.

That evidentiary quality has several uses at once. It supports acceptance and sign-off with data rather than opinion. It creates a record for the project history -- invaluable if a defect emerges later and the question is what was built and whether it met the design. It can clarify responsibility honestly when something is wrong, by showing measured fact rather than leaving fault to argument. And it feeds forward: the verified as-built captured through construction becomes the foundation of the accurate handover model and, later, the facilities record -- the subject of the next lesson. Capture used for QA is not a one-off check but a measured thread running from construction into operation.

The boundaries that have run through this whole module apply with full force where QA meets liability. Measured evidence is powerful precisely because it can carry weight in decisions about acceptance, payment and responsibility -- which is exactly why, when a measurement must be *binding*, it belongs to the professionals who can stand behind it: licensed surveyors for survey-grade position and compliance, the relevant engineers for whether a deviation is structurally acceptable, quantity surveyors for certified quantities. The capture must itself be of verified accuracy, and its limits -- the method, the date, the occlusion gaps, the tolerance thresholds chosen -- must be stated, or the 'evidence' is contestable. Used with that rigour, reality capture gives construction something it has long lacked: a comprehensive, measured, honest record of what was really built, available early enough to act on, and trustworthy enough to rely on. Replace 'we think it is right' with 'here is the measured evidence', and both quality and accountability improve.