Concrete scanning with GPR before coring on data center jobs

Concrete scanning, and why you map the slab before you cut it

Concrete scanning is the work of finding out what is inside a slab before a core bit or an anchor drill goes into it. A concrete slab is not a blank block. It is full of reinforcing steel, sometimes post-tension tendons under enormous load, electrical and data conduit, plumbing, and the occasional void where the concrete never consolidated. Scanning maps that hidden layout from the surface so the hole lands in a clear spot instead of in something that hurts you or the building.

On a data center this is everyday work, because the whole back of house gets anchored to the floor. Switchgear lineups, generators, transformers, UPS and battery racks, CRAH and CRAC units, cable tray supports, and busway hangers all need drilled-in anchors, and every one of those holes is a chance to hit rebar, clip a conduit, or sever a tendon. The drawings show where the steel was supposed to go. They do not show where it actually landed after the crew chaired it, tied it, and walked it during the pour.

The reason you scan and do not guess is that the failures are expensive in two different ways. Hit a live conduit and you can electrocute the operator or kill a circuit the building needs. Cut a post-tension tendon and you have wounded the structure and released stored energy that can injure someone standing there. Neither shows on the surface until it is too late. Scanning is the cheap step that keeps the expensive step from happening.

What happens if you cut a post-tension tendon?

Cutting a post-tension tendon releases the stored energy in a high-strength strand that was stressed to tens of thousands of pounds, and it does it all at once. A single seven-wire strand in an unbonded slab is tensioned to roughly 30,000 lbf or more, and a banded group of tendons stacks that into hundreds of tons of force held in the concrete. Nick that strand with a core bit and it can let go, whip, and shoot strand and concrete out of the slab. People have been injured and tendons have come out of the floor. This is the hazard that makes post-tension slabs scan-mandatory before any penetration.

The structural side of the loss is just as real and slower to show up. A post-tension slab carries its load because the tendons squeeze the concrete into compression along a designed drape. Sever one and that band of the slab loses the precompression it was counting on, so the deflection, the cracking, and the load capacity all move in the wrong direction. The engineer of record has to evaluate and the tendon has to be repaired, which on a finished mission-critical floor is a bad day measured in weeks.

Treat any slab that might be post-tensioned as post-tensioned until the drawings and the scan say otherwise. Tenant work and renovations are where tendons get cut, because the crew assumed a conventionally reinforced slab and drilled. On a post-tension floor, coring and drilling happen only in pre-planned, scanned, and verified windows, and a tendon strike is never a field call you recover from by patching the hole.

How does ground penetrating radar work?

Ground penetrating radar sends a short pulse of electromagnetic energy into the concrete from an antenna and listens for the part of that pulse that bounces back. The pulse reflects wherever it crosses a boundary between two materials with different electrical properties, which the trade calls a dielectric contrast. Steel rebar, a metal conduit, the air in a void, the plastic of a PVC pipe, and the far face of the slab all sit at a contrast the radar can hear. The instrument times how long the echo takes to return and turns that travel time into a depth.

The antenna rides on a small cart or hand unit that you roll across the surface in straight lines. As the wheel turns, the unit fires hundreds of pulses and stacks the returns side by side into a cross-section of the slab, a profile that builds in real time on the screen as you push. You can work two ways. A quick real-time pass tells you on the spot whether a spot is clear before a single hole. A grid scan, where you run a tight pattern of lines both directions and let the software build a plan view or a depth slice, gives you the full map of the mat for layout work.

GPR needs access to only one face of the slab, which is what makes it the field workhorse. You roll it on the floor you are standing on, see the rebar grid and the conduit appear under you, and mark them without shutting down the area or moving anyone out. The catch is that the picture is an interpretation, not a photograph, and reading it is the skill the rest of this guide is about.

What GPR finds in a slab

GPR finds anything in the concrete that sits at a strong enough dielectric contrast to the surrounding paste. In practice that is the reinforcing steel, the post-tension ducts and strands, metal and plastic conduit, embedded plates, rebar dowels, voids and honeycombing, and the back of the slab itself, which gives you the slab thickness when you can see the bottom reflection. Steel returns the strongest signal because it reflects almost all the energy that hits it, so rebar and metal conduit light up first and clearest.

Metal conduit reflects like rebar, strong and bright. PVC and other plastic conduit sit at a smaller contrast and return a weaker echo, so they are easier to miss, though a conduit carrying water or wire reads better than an empty one. A void or a delamination shows as a contrast against the solid concrete around it and as a change in the slab reflection. Slab thickness comes from the time it takes the pulse to reach the bottom and return, which is one of the more useful numbers a quick scan gives you before a full-depth core.

What GPR does not give you is a labeled drawing. It gives you reflections, and you assign meaning to them from their depth, their spacing, their continuity, and how they line up with what the structure should have. A bright reflection at consistent depth running in a regular grid reads as rebar. A pair of strong reflections sweeping up and down across the slab reads as a draped post-tension tendon. The instrument shows the echo. You name it.

GPR vs x-ray and the cover meter

GPR is the default concrete scanning method because it needs one-sided access, works in real time, and carries no radiation. X-ray, electromagnetic cover meters, and electromagnetic locators all have a place, but they solve narrower problems. The choice comes down to what you need to see, what access you have, and whether you can clear the area.

X-ray, or radiography, gives a clearer image than GPR and shows internal detail an experienced eye can read like a photograph, including the difference between a bonded and unbonded tendon. The price is steep. X-ray needs access to both faces of the slab, with the source on one side and the film or detector on the other, and the radiation forces a cleared exclusion zone around the shot that empties the work area. Traditional film also goes off site to develop, so you wait for the answer. On an occupied or congested data center floor, clearing the zone for a radiation shot is usually a non-starter, which is why GPR carries the load and x-ray gets reserved for the spot where you need the detail and can control both sides and the area.

A cover meter, also called a pachometer, is a different and simpler tool. It uses an electromagnetic field to find shallow reinforcing steel and measure concrete cover, and it is good at exactly that and nothing deeper. It finds steel, not plastic conduit, not voids, not the back of the slab, and it runs out of depth fast. Use a cover meter to confirm cover and locate a shallow bar quickly. Use an electromagnetic line locator to trace an energized conductor or a traceable utility. For mapping a slab to drill through, GPR is the tool that sees the whole picture from one side.

What antenna frequency do you use?

Antenna frequency is a direct trade between resolution and depth, and concrete scanning lives at the high-frequency end. A higher frequency antenna resolves smaller and more closely spaced targets but attenuates fast, so it does not reach as deep. A lower frequency antenna pushes deeper but smears small targets together. You pick the frequency for the thickness you have to see into and the detail you need to separate.

For concrete cutting and core drilling the common antennas run high, in the range of roughly 1.5 to 2.6 GHz. A 2.6 GHz antenna gives the sharpest picture and separates closely spaced bars near the surface, but its useful depth is shallow, often only the top several inches. A 1.5 to 1.6 GHz antenna trades a little resolution for more reach and is the workhorse for typical structural slabs, with practical depth on the order of a foot to a foot and a half in good concrete. The exact reach depends on the concrete, the moisture, and how congested the steel is, so treat any depth number as a starting point you confirm on the slab.

The practical rule is to match the antenna to the job in front of you. A thin topping slab or a search for shallow conduit wants the highest frequency you have. A thick mat foundation or a deep post-tension duct wants the lower of the two so the pulse reaches it at all. If a target sits deeper than the antenna can see, the screen goes quiet and quiet is not the same as clear. Knowing the difference is the whole game.

Reading the scan: the hyperbola and the depth

A round target buried in concrete does not show up as a dot on a GPR profile. It shows up as a hyperbola, an inverted U or arch, and that shape is the single most important thing to recognize on the screen. The arch forms because the antenna starts hearing the target before it is directly over it and keeps hearing it after it passes, so the travel time is longest at the edges and shortest right above the target. The peak of the hyperbola sits over the true location of the target, and the depth to that peak is the depth to the object. Rebar, conduit, and tendons all read as hyperbolas when you cross them at an angle to their run.

Depth is a calculation, not a direct reading, and it depends on how fast the pulse travels through the concrete. That speed is set by the concrete's dielectric, which for dry, cured concrete commonly runs around 6 to 8 but climbs as the concrete holds moisture. Get the dielectric wrong and every depth on the scan is wrong by the same proportion. Without calibration, depth accuracy is roughly 15 to 20 percent. Calibrate against a target at a known depth, an exposed bar, a core hole, or a verified drawing dimension, and you can tighten that toward 5 percent under good conditions.

Calibrate on the slab you are scanning, not from memory. The fastest method is to migrate or fit the hyperbola in the software, which back-calculates the dielectric from the shape of the arch, or to set the depth from a known point you can physically check. A scan that reads a bar at 2 inches when it is really at 2.5 is the scan that puts your anchor short of the embedment the engineer assumed. Depth is where loose interpretation turns into a real problem.

c

Speed of light in free space, the baseline the pulse slows down from in concrete

εr

Relative dielectric of the concrete, roughly 6 to 8 dry, higher when wet; sets the pulse velocity

t

Two-way travel time, how long the pulse takes to reach the target and return

Telling post-tension from rebar from conduit

Separating the targets is what a scan is for, because you can drill near rebar in a clear gap but you do not drill near a tendon at all. The tells are depth, spacing, continuity, and pattern, and you read them together rather than one at a time. Conventional rebar lays out as a regular grid: evenly spaced hyperbolas at a consistent depth, the top mat near the surface and the bottom mat near the back of the slab, running square to the structure. Once you see the spacing repeat, you trust it.

Post-tension tendons break that pattern. They are placed on a designed drape, so a tendon dives and rises through the slab depth across the span instead of holding one depth, and on a profile run along its length it sweeps up and down rather than sitting flat. Across its width it often runs in banded groups in one direction with distributed tendons the other way, a layout that does not look like the even rebar grid. A reflection that changes depth as you follow it, or a tight band of strong reflections sweeping through the slab, is the flag to stop and treat it as post-tension until proven otherwise.

Conduit reads as an isolated hyperbola that does not belong to the grid. A single strong reflection running on its own line, often deeper than the top mat and crossing the reinforcing at an angle, is usually a conduit or a pipe. Metal conduit is bright like rebar; plastic is dimmer and easier to lose. When you cannot resolve whether a reflection is a deep bar, a tendon, or a conduit, that uncertainty is the answer: you do not drill there, you move the hole or you confirm by another method before committing.

Marking the slab and finding the safe drill window

The deliverable of a scan is marks on the concrete, and the marks have to be clear enough that the drill operator who never saw the screen drills the right hole. Lay the targets out on the surface as you find them: paint the rebar lines, the conduit runs, and any tendon you locate, and mark them in a way that says what each one is. The hole goes in the safe window, the clear gap between the marked targets where the drill misses everything within the depth you are going.

Use a marking convention that distinguishes post-tension from everything else, and do not let it blur. A common practice is one color for reinforcing and conduit and a separate, unmistakable color for post-tension, with the tendon line marked as a keep-off rather than a target to thread past. Marking a tendon the same as a bar is how a tendon gets treated as a bar and drilled near. The whole point of separating them on the screen is lost if they look the same on the floor.

Define the window with the depth in mind, not just the plan position. An anchor that only goes 3 inches deep can sit in a gap that a full-depth core could not, because the core has to clear targets all the way through and the bottom mat as well. Mark the clear spot, note the depth it is clear to, and if the layout the equipment needs does not fit the clear windows, that goes back to coordination before anyone drills, not after the hole is in the wrong place.

One-mat, two-mat, and elevated decks

Know what kind of slab you are scanning before you trust a single pass, because the reinforcing layout changes what you have to find. A thin slab on grade may carry one mat of reinforcing near mid-depth or a single top mat. A structural slab usually carries two mats, a top mat near the surface and a bottom mat near the back face, and an anchor or a core has to clear both. Scanning the top mat and calling the slab mapped misses the bottom mat entirely, which is the steel a deeper core runs into.

On an elevated slab you also care about both faces and about what is under the deck. A core through an elevated floor comes out the bottom into the space below, so the hole has to clear the bottom mat, any conduit run in the deck, and whatever hangs below, and you confirm there is nothing on the underside in the path. Scanning from the top tells you the top mat well, the bottom mat less well, and the deeper you look the more the top steel gets in the way.

The composite slab on a corrugated steel deck is the case that changes the rules, and it gets its own section next, because the steel pan under the concrete reflects the radar and hides what is at and below it. Before you scan an elevated floor, find out whether it is a flat soffit slab, a slab on metal deck, or a post-tension slab, because each one limits what GPR can promise.

Why a metal pan deck blinds GPR

A composite slab poured on a corrugated steel deck is the limitation you have to respect, because the steel pan reflects nearly all the radar energy. The pulse hits the metal deck and bounces, so the deck reads as a strong, continuous reflection across the whole scan and acts like a mirror. You can map the top mat and the concrete cover above the deck, but the deck blinds you to anything at the level of the pan or below it. GPR cannot see through a solid steel sheet.

That matters most for a full-depth core through a composite floor. The hole has to pass through the concrete, through the steel deck, and out the bottom, and GPR will not tell you what is sitting in the flute or hanging just below the deck, because the pan hides it. The flutes of the deck also vary the concrete thickness across the span, so the cover over the top mat changes from the high rib to the low flute, and a scan that does not account for the deck profile reads the depths wrong.

When the slab is on a metal deck and you need to go through it, scanning the top face is not enough by itself. Confirm what is below the deck before a full-depth core, by checking the underside where it is exposed, by the drawings, or by another method, because the radar cannot do it from the top. Trusting GPR through a steel pan is one of the more common ways a scan gives false confidence.

What GPR cannot guarantee

GPR is the right tool and it still has hard limits, and a scanner who oversells it is more dangerous than one who has none. Depth is the first limit. A high-frequency concrete antenna sees a finite distance into the slab, and a target deeper than that range simply does not appear, so a clean-looking scan can hide something below what the antenna reached. Quiet at depth is not the same as clear.

Congested reinforcing is the second limit. A dense top mat reflects so much energy that it shades what is underneath, so a tight grid of bars near the surface can mask a bottom mat, a conduit, or a tendon below it. The radar spends its energy on the first layer and has little left for the second. The third limit is moisture and conductivity. Wet, green, or freshly placed concrete, and salts or conductive contaminants in the slab, attenuate the signal and can throw false reflections, so the picture degrades exactly when the concrete is young or damp.

Add the metal deck from the section above and the honest summary is this. GPR reliably finds the near targets in sound, cured concrete from one side, and it gives you depth you can calibrate to a few percent. It does not guarantee it saw the deep target, the masked target, or anything below a steel deck, and it does not replace the drawings or the engineer's knowledge of the structure. Scan, but know what the scan did and did not prove.

When do you drill, relocate, or stop and call the engineer?

The go decision is simple and strict: you drill only when the scan shows a clear window to the full depth of the hole, you have calibrated the depth, and the marked location matches what the work needs. If the gap is clean and you are confident in the depth, drill it. Everything short of that confidence is a stop, not a maybe.

Relocate the hole when the planned spot is over a target but a clear window exists nearby and the equipment can tolerate the shift. Small moves to thread between bars are routine, but the move has to stay inside what the anchor design and the baseplate allow, because sliding an anchor to dodge rebar can violate the edge distance or spacing the anchorage was designed to, and that is its own failure. When you genuinely cannot tell what a reflection is, hand-chip or drill a small pilot to expose and confirm it before committing the real hole, rather than guessing.

Stop and go to the engineer of record when the work does not fit the structure: when the only clear window is too small for the anchor pattern, when a post-tension tendon sits where the hole has to go, when a full-depth core has to pass a deck or a congestion you cannot see through, or when relocating breaks the anchor geometry. A tendon strike or a structural penetration is never a field call. The structural engineer of record and the drawings control any penetration, and the scanner's job at that point is to say clearly that it is not safe to drill here yet.

Field example: scanning a switchgear pad for the anchor layout

A switchgear lineup is landing on a 10 in structural slab and needs eight drilled anchors at 4 in embedment on the equipment bolt pattern. The scan starts with a grid over the footprint using a 1.6 GHz antenna, run both directions, with the depth calibrated against a core hole nearby that exposed the top mat at a known depth. The top mat reads as a clean grid at about 1.5 in cover on 12 in centers, and the bottom mat shows fainter near the back of the slab.

Mapping the bolt pattern over the grid, six of the eight anchors fall in clear gaps between the top bars, well inside the 4 in depth the anchors need, since the bottom mat sits below that. Two anchors land within an inch of a top bar. One of those shifts a half inch into a clear gap without breaking the anchor spacing, so it is marked and approved. The other cannot move far enough without crowding the next anchor, so it goes back to coordination rather than getting drilled into the bar.

The scan also catches a single strong reflection crossing the footprint at about 4 in deep that does not belong to the grid, almost certainly a conduit. It is marked as a keep-off, the two anchors near it are checked clear, and the find goes in the record. Nobody cored a switchgear pad into a conduit that day, which is the entire return on a one-hour scan.

Who should run the scan, and the standards behind it

Concrete scanning is an interpretation job, and the result is only as good as the person reading the screen. The reflections do not label themselves, the depth depends on a dielectric the operator has to set, and the difference between a deep bar and a tendon is a judgment. An untrained operator pushing a cart produces a scan that looks authoritative and proves nothing. The scanner should be trained and experienced on the equipment and the structure type, and on a critical penetration the data and the marks should be reviewed by someone who can stand behind them.

The method has a standards framework even though the field call stays with a qualified person. ASTM D6432, the standard guide for the surface ground penetrating radar method, lays out the GPR method, the equipment, the field procedure, and the limitations, and it is the general reference for how a survey is run and what the method can and cannot do. ACI 228.2R, the report on nondestructive test methods for evaluating concrete in structures, covers GPR alongside the other nondestructive methods and is the concrete-side reference for choosing and applying them. For GPR on bridge decks specifically there is a separate ASTM test method, which is its own scope rather than building slabs.

Confirm the current editions and any project requirements before you cite a number on a report, because the standards are reviewed and revised on their own cycles. The drawings, the structural engineer of record, and the post-tension supplier's tendon records are the authorities that govern whether and where a penetration is allowed. The scan informs the call. It does not replace the engineer who owns the structure.

The scan record and the as-built

A scan that lives only in the technician's head is a scan nobody can defend later, and on a data center the anchor layout depends on it. The record is what ties the holes that got drilled to the slab that got scanned, so when a question comes up at startup or years out, someone can show what was under the floor and why the anchors went where they did. The marks fade and the floor gets covered. The record is what is left.

Capture the scan as it was: the location and the slab, the method and the antenna, what was found and at what depth, whether the slab is post-tensioned, where the safe drill windows were, and who scanned and reviewed it. Tie it to the equipment and its anchor layout, because the anchor design assumes the holes landed where the scan said they were clear. When an anchor got relocated to dodge a bar, the record should show the move and that the spacing still held, the same way the anchorage acceptance expects.

This record feeds the anchor and grout acceptance and the broader structural turnover. The anchor layout the equipment is bolted to was made possible by the scan, so the scan record belongs with the anchorage documentation that proves the gear is fastened right. For setting and inspecting the anchors the scan cleared the way for, see the anchor bolt and baseplate grout guide. For where this sits in the whole concrete and steel quality program, see the data center concrete and steel QA overview.

What to document

The scan record answers one question later: was this slab checked before it got drilled, and what did the check find? Capture it per scanned area, because a switchgear pad and a generator base are separate scans with separate findings. The point is traceability, so the holes, the equipment, and the slab all tie back to the same record.

Record the location and the slab type, the method and antenna used, the calibration and the dielectric basis, the targets found and their depth, whether post-tension was identified, the slab thickness if you read it, the safe drill windows and the depth they are clear to, any relocated or rejected holes, and who scanned and reviewed it. The table below is the spine of a scan record that holds up.

Common mistakes

• Drilling or coring a post-tension slab without scanning it first, then cutting a tendon and releasing the stored energy.
• Trusting GPR through a steel pan deck, where the metal reflects the pulse and hides everything at and below the deck.
• Scanning only the top mat on a two-mat slab, so a deeper core runs into the bottom mat the scan never mapped.
• Marking post-tension and rebar the same color, so a tendon gets treated as a bar and drilled near.
• Scanning green, wet, or freshly placed concrete, where moisture attenuates the signal and throws false reflections.
• Reporting depth without calibrating the dielectric on the slab, so every depth is off by the same proportion.
• Treating quiet at depth as clear, when the target was simply deeper than the antenna could reach.
• Relocating an anchor to dodge a bar without checking that the edge distance and spacing still meet the anchorage design.

Field checklist

Standards and references

The GPR method itself is described in ASTM D6432, the standard guide for using the surface ground penetrating radar method for subsurface investigation, which sets out the method, the equipment, the field procedure, data interpretation, and the limitations of the technique. It is the general reference for how a survey is conducted and what the method can resolve, including the use of GPR on concrete structures. There is also a separate ASTM test method for GPR on asphalt-covered concrete bridge decks, which is its own application rather than building slabs.

On the concrete side, ACI 228.2R, the report on nondestructive test methods for evaluation of concrete in structures, covers ground penetrating radar alongside the other nondestructive methods, with the principle, the instrumentation, the procedure, and the data analysis for each. It is the reference for selecting and applying a nondestructive method to a concrete structure. Where a penetration interacts with the anchorage and the structure, ACI 318 Chapter 17 governs the anchoring to concrete that the drilled holes serve, and coring for in-place evaluation runs under the ASTM core method when the engineer directs it.

Edition numbers and clause references change between cycles, so confirm the current editions and any project-specific requirements before citing a standard on a report. The structural drawings, the structural engineer of record, and the post-tension supplier's tendon layout and stressing records are the authorities that control whether and where a slab can be cut. The scan supports the decision. The engineer of record and the contract documents control it.

Units and terms

Concrete scanning carries a vocabulary that reads differently across a GPR screen, a structural drawing, and an anchor evaluation report. The terms below are the ones that cause confusion on site.

Antenna frequency is in megahertz or gigahertz; depth, cover, and slab thickness in inches; travel time in nanoseconds; and the dielectric is a unitless ratio. Match the unit to the document you are reading from, and remember that a depth on the screen is only as good as the dielectric it was calculated with.

GPR

Ground penetrating radar, which sends an electromagnetic pulse into the slab and times the reflections to map embedded targets

Hyperbola

The inverted-U reflection a round target makes on a GPR profile; the peak sits over the target and gives its depth

Cover

The depth of concrete from the surface to the reinforcing steel, which protects the bar and sets a shallow anchor's clearance

Post-tension tendon

A high-strength strand stressed to load the slab in compression; cutting one releases stored energy and weakens the structure

Dielectric

The electrical property of the concrete that sets the pulse velocity; roughly 6 to 8 in dry concrete, higher when wet

Pachometer

A cover meter that uses an electromagnetic field to find shallow rebar and measure cover; steel only, not deep

Composite metal deck

A slab poured on corrugated steel decking; the steel pan reflects the radar and hides what is at or below it

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is a special inspection on a construction project?
• Cast-in vs post-installed anchors: which should I use?
• What is the difference between cold and hot commissioning of a BESS?
• What is a CDU in a data center?
• What is adiabatic cooling in a data center?