Electronic leak detection survey for roof and waterproofing membranes

Electronic leak detection, and what it actually finds

Electronic leak detection is a method that finds a breach in a roof or waterproofing membrane by sending an electrical current through the hole and locating where that current goes. The membrane is an electrical insulator, so an intact sheet blocks current the same way it blocks water. Where there is a hole, water or a wetted path carries current through the breach to the grounded conductive deck underneath, and the instrument reads the current flowing to that one point. The hole completes a circuit, and the survey finds the circuit.

That is the whole idea, and it is why ELD locates the actual breach instead of telling you a roof leaks somewhere. A flood test confirms a leak exists. A drip in the building tells you water got in, days and many feet away from where it entered. ELD points at the hole itself, which is the difference between cutting one repair and tearing up a roof to chase a stain.

The method splits into two families that run the same physics in opposite conditions. High-voltage ELD sweeps a charged brush over a dry membrane and sparks to ground at a pinhole. Low-voltage ELD wets the surface, sets up a grounded perimeter, and maps the current vector to the breach, which lets it work under soil, ballast, and pavers. Pick the wrong one for the conditions and you get a clean survey that found nothing because nothing could conduct.

Breach

A hole, puncture, open seam, or other discontinuity in the membrane that lets water and current through

Overburden

Anything covering the membrane, soil and plants, ballast stone, pavers, or insulation in an inverted roof

Integrity test

An ELD survey run to verify a new or repaired membrane is breach-free before it is covered or accepted

Why ELD beats a flood test or chasing the leak

ELD pinpoints the breach, and that is the advantage everything else hangs on. A flood test fills a plaza deck or a roof area with water, holds it overnight, and watches the ceiling below for a drip. It proves the assembly leaks or holds, but it does not tell you where, so a positive flood test leaves you guessing across the whole area. ELD shows the exact location, which shrinks the repair from a section to a patch and shrinks the cost with it.

It also works where a flood test cannot. You cannot reliably flood a sloped roof, a vertical wall, or a roof that already carries soil and plants, and you should not dump tons of water onto a deck whose drainage and structure you are not certain of. ELD uses little water or none, depending on the method, so it carries almost no risk of damaging the building while you test. On a vegetated or ballasted roof, where the membrane is buried and you cannot see a square inch of it, low-voltage ELD is often the only practical way to test the surface at all.

And it works at two points in the roof's life. At install, ELD is a non-destructive integrity check that finds the construction punctures, the dropped-screw holes, and the open details before the overburden goes on. Years later, forensically, it finds the breach that is leaking now. A flood test is a one-time pass or fail. ELD can be run again over the life of the roof, which is why owners of museums, data centers, and hospitals lean on it.

What does ELD need to work?

ELD needs three things in place, and if any one is missing the survey reads false or reads nothing. The membrane has to be electrically nonconductive, so an intact sheet blocks current. The substrate under the membrane has to be electrically conductive and grounded, so it can receive the current that comes through a breach, commonly a structural concrete deck or a steel deck. And there has to be no insulating layer between the membrane and that grounded substrate, or the current that comes through the hole has nowhere to go.

This is physics, not preference, and it sets the whole job up. The membrane is the insulator. The wet path through a breach is the conductor. The grounded deck is the return. With all three present, current flows only at the breach and the instrument finds it. Take away the conductive return and an intact membrane and a holed one look identical to the meter, because neither completes a circuit.

The conventional roof is where this gets interesting, because insulation and a coverboard sit between the membrane and the deck by design, and they are electrical insulators as much as thermal ones. The fix is to install a conductive ground plane in the assembly: a conductive primer wiped onto the substrate, or a thin conductive wire grid laid under the membrane and grounded. Plan that ground plane into the system before the membrane goes down, because adding it afterward means opening the roof. The membrane manufacturer and the ELD specialist should agree on the grounding approach during design, not on survey day.

Low-voltage ELD: vector mapping on a wetted roof

Low-voltage ELD finds the breach by wetting the membrane and mapping the electric field across the surface to the point where current drains through a hole to ground. A conductor loop is run around the perimeter of the test area and energized at a low potential, commonly on the order of tens of volts DC, with the grounded deck under the membrane as the return. The wetted surface above the insulating membrane and the grounded deck below form two zones separated by the membrane. A breach is the one place the two connect.

The technician then walks the area with two probes set a fixed distance apart, reading the voltage difference between them. The field is symmetrical across an intact membrane, so the probes read a predictable pattern. Near a breach, current is rushing toward that point, and the potential gradient bends toward it. The probes swing, then null, then reverse as they straddle the hole, and that crossover walks the technician right onto the breach. This is the vector-mapping technique, sometimes called electric field vector mapping, EFVM, or low-voltage integrity testing.

The strength of the low-voltage method is that it tolerates a wet, conductive surface and works through overburden. Soil, ballast, and pavers can stay in place, because the water and the wetted overburden carry the field down to the membrane, and the breach still pulls current at one location. That makes low-voltage ELD the practical choice on vegetated and ballasted roofs where the membrane is buried. The cost is setup and time: the perimeter has to be established, the surface kept wet, and the grid walked at a slower pace than a dry spark survey.

High-voltage ELD: the spark test on a dry membrane

High-voltage ELD finds a breach by sweeping an electrified brush or broom across a dry, exposed membrane until the charge jumps through a hole to the grounded deck and trips an alarm. The unit puts a high voltage, commonly in the range of several thousand to roughly 12,000 V DC, at very low amperage onto a conductive brush. Over an intact, dry, insulating membrane, nothing happens, because the membrane blocks the charge. Over a pinhole or a puncture, the charge finds the path of least resistance straight through the hole to ground, the circuit completes, and the instrument sounds.

Dry is the whole requirement, and it is the opposite of the low-voltage method. The membrane has to be dry, because any water on the surface gives the high voltage a path to bridge across the top and either spark where there is no hole or, worse, create a shock hazard for the operator. There can be no overburden either. The brush has to touch the actual membrane, so high-voltage ELD is for exposed, bare membranes only, before anything covers them.

Where it shines is finding the tiny stuff. The spark will jump a pinhole far smaller than anything a probe or a visual walk catches, which makes high-voltage testing the strong tool for install-time integrity on an exposed membrane and for testing details, base flashings, and vertical surfaces that a flood test or a wet vector survey cannot reach. It is also faster and cheaper to set up than a vector survey, since there is no perimeter loop to run and no wetting to maintain. The catch is that it only sees the membrane it can physically sweep, dry and uncovered.

High-voltage vs low-voltage: which method when?

The split is simple and it is the first call you make: dry and exposed goes high-voltage, wet or covered goes low-voltage. A bare membrane with nothing on it, before the overburden or at a repair, gets the high-voltage spark test, because it is fast, it finds pinholes, and the surface is dry. A membrane under soil, ballast, pavers, or any wet condition gets the low-voltage vector survey, because the spark method cannot reach a covered membrane and cannot run on a wet one.

Match the method to the breach size too. High-voltage spark is the one that catches a true pinhole, the kind of holiday left by a stray screw or a hot-air weld skip. Low-voltage vector mapping locates breaches well but is generally better at finding the open path than resolving a microscopic hole the way a spark does. If the question is whether a newly welded exposed membrane has any pinholes at all, that is a high-voltage job. If the question is where water is getting through a green roof, that is low-voltage.

Both methods sit under one umbrella standard and have real limits, so neither is a magic wand. Each one assumes the physics is intact: nonconductive membrane, grounded conductive deck, no blocking insulation, and the right surface condition. A specialist who only owns one tool will tell you that tool fits your roof. The honest read is that the roof's condition, dry or wet, exposed or buried, picks the method, and some surveys need both, high-voltage at the exposed details and low-voltage across the covered field.

ASTM D7877 and D8231: the standards that frame ELD

ASTM D7877 is the umbrella. Its title is the Standard Guide for Electronic Methods for Detecting and Locating Leaks in Waterproof Membranes, and it describes the electrical conductance methods used to test exposed or covered membranes, both high-voltage and low-voltage. It applies across roofs, plaza decks, pools, water features, and covered reservoirs, on insulating membranes such as PVC, polyethylene, polypropylene, and bituminous sheet. D7877 is a guide, which means it lays out the methods and their use rather than dictating a single pass-fail procedure. It has gone through several editions, so confirm the current one when you cite it.

ASTM D8231 is the low-voltage practice. It is the Standard Practice for the Use of a Low Voltage Electronic Scanning System for Detecting and Locating Breaches in Roofing and Waterproofing Membranes, and it covers a low-voltage, dual-sweep scanning method run under roughly 50 V. A practice is more prescriptive than a guide, so D8231 sets out how that specific low-voltage scanning survey is performed. Worth knowing: the low-voltage scanning method in D8231 is a specific platform, and not every low-voltage approach, including some vector-mapping setups, claims compliance with it, so do not assume every low-voltage survey is a D8231 survey.

Above both, the membrane manufacturer's requirements and the project specification govern. D7877 itself says ELD is not a replacement for visual or infrared inspection and is meant to work alongside other methods. IIBEC, the roof consultants' body, has published technical guidance on ELD, and the NRCA Roofing Manual covers testing practice by topic. Name the standard that controls the point, confirm the edition the spec calls out, and let the manufacturer's and the contract's requirements override any rule of thumb.

Setup and grounding: the return path the survey depends on

Every ELD survey starts at the ground, because the method is a circuit and a circuit needs a return. The grounded conductive substrate under the membrane is that return, and confirming it before you test is the step that decides whether the survey means anything. On a concrete or steel deck with the membrane laid directly on it, the deck is the ground, and you connect the instrument's ground lead to it at a reliable point, a drain, a structural connection, or an exposed deck location.

When insulation sits between the membrane and the deck, the deck is no longer reachable as a ground, and you need the conductive plane built into the assembly: the conductive primer or the wire grid mentioned earlier, grounded out at a known point. If that plane is not there, low-voltage testing on a conventional insulated roof will not work, full stop, and finding that out on survey day is too late. This is why ELD belongs in the design conversation for any roof meant to be tested under cover.

For low-voltage work you also establish the perimeter and wet the surface. The perimeter conductor isolates the test area and carries the energizing potential, so it has to make good contact around the whole boundary, and the area inside has to be wetted evenly enough to carry the field down to the membrane. Standing puddles and dry patches both distort the reading. For high-voltage work the setup is lighter: confirm the ground, confirm the membrane is genuinely dry, and clear the area, since the only return that should exist is through a breach.

When the membrane itself conducts, ELD does not work

ELD depends on the membrane being an insulator, so a membrane that conducts breaks the method entirely. The classic case is EPDM. Black EPDM is loaded with carbon black, which makes the rubber electrically conductive, so current does not need a hole to reach ground, it travels through the sheet, and the survey cannot tell a breach from sound membrane. Butyl sheet has the same problem. So do membranes with a metallic coating or a foil facing, and so do metal flashings and metal details sitting in the test area.

This is the single fact that decides whether ELD is even an option, and it gets missed. Confirm the membrane chemistry before anyone schedules a survey. PVC, TPO, and most thermoplastic and bituminous sheets are insulating and test fine. Carbon-black EPDM, butyl, and metallic-faced products do not, and no amount of grounding fixes a membrane that conducts on its own. On a roof that mixes materials, the EPDM areas and the metal details are blind spots to ELD even when the rest of the field tests clean.

The companion guide on single-ply seam QA covers how EPDM is seamed and inspected, and the contrast is worth holding in mind. On a thermoplastic roof, ELD is a whole-membrane integrity check that the seam probe cannot match, because the probe only checks the lap while ELD checks every square foot of the sheet for a breach. On a black EPDM roof, you lose that tool and lean harder on the probe, the vacuum box, and a flood test, because the membrane will not carry the current ELD needs.

Why does ELD give a false reading?

ELD gives a false reading when something other than a breach completes the circuit, or when something blocks the circuit a breach should complete. Both happen, and a survey that ignores them reports holes that are not there or misses holes that are. The two failure directions are worth naming, because the fixes are different.

False positives come from unintended grounds. A lightning protection system, a roof drain, a metal conduit, a counterflashing that touches the overburden, any conductive item that reaches ground draws current and reads like a breach drawing current. On a covered roof, metal in or under the overburden is the usual culprit. Standing water bridging across the membrane is another, especially on high-voltage testing, where surface water lets the spark jump where there is no hole, and on a wet high-voltage survey that bridging is also a genuine shock hazard. The defense is to identify and isolate the metal and the water before reading the area, not to write up every alarm as a leak.

False negatives come from blocked paths. On an overburden roof, the current has to travel from the surface down through the overburden to the wetted membrane, and an electrically insulating layer in the way, a drain mat, a root barrier, a rigid insulation board, can break that path so a real breach never shows. Pavers on pedestals leave air gaps that do not conduct, and split-slab assemblies with reinforcing steel scramble the field. The deeper and the more layered the overburden, the more the signal degrades and the more breaches a survey can walk right past. Know the assembly before you trust a clean result.

Integrity testing at install vs forensic leak hunting

ELD does two different jobs and the difference shapes the survey. Integrity testing happens at install or just after, on a new or repaired membrane, to prove it is breach-free before it is covered or accepted. Forensic leak hunting happens later, on a roof that is leaking now, to find the breach that is letting water in. Same physics, different question and different timing.

Integrity testing is the high-value use, because it is cheap insurance bought at the one moment the membrane is exposed and reachable. Run a high-voltage spark survey over a freshly welded exposed membrane and you catch the construction punctures, the dropped fasteners, the weld skips, and the trade damage from other crews before the overburden buries them. On a roof headed for soil or pavers, this is the last time you will see the membrane without excavating, so an integrity survey before cover is the difference between a five-minute patch and a dig. This pairs directly with the seam QA walk: the probe certifies the laps, the spark certifies the whole sheet.

Forensic work is harder and that is the nature of it. The roof is covered, often wet, full of unintended grounds, and the breach you are hunting may be one of several. You confirm the membrane is testable, you establish or find the ground, you isolate the metal, and you map the area methodically. A forensic survey that finds the active breach saves a tear-off, but it lives with every limitation in this guide at once, which is why the conditions and the assembly have to be understood before the first probe goes down.

Permanent leak-detection systems: the embedded grid

A permanent leak-detection system is a sensor grid built into the roof assembly that monitors for a breach continuously, instead of a one-time survey. Conductive sensor cabling or tape is laid out under or within the construction in a grid, commonly on the order of a 15 ft by 15 ft pattern though it can be tighter, dividing the roof into zones. When water crosses the membrane and reaches the sensor layer, it changes the electrical characteristics of that zone, and the monitoring panel flags which zone went wet.

The point of a permanent system is time. A survey tells you the state of the roof on the day you walk it. A permanent grid tells you the moment water gets in, often before any damage shows inside, and it tells you which zone, which narrows the search before anyone climbs up. The panel commonly ties into the building management system and can push an alert by text or email, so the owner learns about the breach from the roof rather than from a stained ceiling.

These systems earn their cost on buildings where water is a catastrophe rather than a callback: museums, archives, hospitals, and data centers, and on green roofs and plaza decks where finding a leak after the fact means excavating overburden. They are not a substitute for building the roof right, and a grid over bad workmanship just tells you faster that the workmanship was bad. But on the right building, continuous monitoring turns a buried leak from a demolition project into a located zone.

Vegetated, ballasted, and inverted (IRMA) roofs

On any roof where the membrane is buried, ELD is usually the only practical way to test the surface, and that is the whole reason low-voltage vector mapping exists. A green roof carries soil and plants. A ballasted roof carries stone. An inverted roof, the IRMA or protected-membrane assembly, puts the insulation and a wear course on top of the membrane. In every case you cannot see the membrane, cannot flood it sensibly, and cannot run a high-voltage spark over it. Low-voltage vector mapping reads through the wetted overburden to the membrane below.

The catch is everything in the false-reading section, amplified. The deeper the overburden, the weaker the signal, and the more layers it passes through, the more chances for an insulating barrier to block it or a buried metal item to fake a breach. Pavers on pedestals are notoriously unreliable because the air gaps do not conduct. So the survey on a covered roof is only as good as the technician's read of what is actually in the assembly, which is why the assembly drawings and the as-built matter as much as the instrument.

Newer roof types stack the same problem higher. A blue roof that detains stormwater, or a solar array mounted over the membrane, or a cooling-oriented vegetated assembly, all add overburden and hardware between you and the sheet, and the cooling and solar trade-offs are their own topic. The roofing lesson holds across all of them: if you intend to test a covered roof electronically, design the ground plane and the access in from the start, because retrofitting either one means moving the overburden you put there to avoid moving.

Marking and verifying the find

Finding the breach is not the end of the survey. Locating it, marking it, confirming it is real, repairing it, and re-testing is the sequence, and skipping the confirmation step is how a survey ends up chasing the wrong spot. When the instrument calls a breach, you mark it physically on the roof, by area and exact location, before you move on, because a flagged point you cannot find again is a point you did not find.

Then you verify before you trust it. On an exposed membrane the cut-down is the proof: open the marked spot and confirm there is an actual hole, a puncture, an open weld, a damaged detail, rather than an unintended ground or a field artifact. Verification is what separates a breach from a false positive, and on a forensic survey full of metal and water it is not optional. A specialist who marks fifteen spots and verifies none has handed you a list, not a diagnosis.

Repair, then re-test the repair. The breach gets fixed by the method the membrane calls for, a welded patch on thermoplastic, a primed and taped patch on the rubber it can be done on, a detailed repair at a flashing, and then that repair gets tested again to confirm it reads clean. The re-test is the close of the loop. A patch nobody re-tested is a repair nobody confirmed, and on a roof headed back under overburden that confirmation is the last look you get.

When is ELD the wrong tool?

ELD is the wrong tool when the question is how much moisture is trapped in the roof, not where the membrane is breached. Those are different problems and they need different instruments. ELD locates a hole in the membrane. A moisture survey finds wet insulation and saturated material inside the assembly, which is what you map before a re-roof to decide how much to tear off.

The moisture-survey tools are their own family. An infrared survey reads the roof after sundown, when wet areas hold heat and show up warmer than the dry field, and it is fast and economical when the surface gives a clean uniform image. A nuclear gauge measures hydrogen content with neutron moderation, so it reads moisture deep in a multi-layer or ballasted roof where infrared cannot see the bottom of the system. Capacitance meters read the change in electrical properties that water causes near the surface. All three find moisture in the assembly. None of them points at the breach the way ELD does, and a good forensic program often uses both, a moisture survey to map the wet zones and ELD to find the breach feeding them.

ELD is also wrong when the membrane conducts, covered in its own section, and it is wrong as a standalone substitute for looking at the roof. D7877 itself says ELD works alongside visual and infrared inspection, not instead of them. A flood test still has its place on a plaza deck or a shower pan where you want a real hydrostatic head and the assembly can take it. Pick the tool for the question: breach location is ELD, trapped moisture is a moisture survey, hydrostatic confirmation is a flood test.

The survey report and the leak map

The deliverable from an ELD survey is the leak map, and a survey that does not produce one has not finished. The map shows the test area, the method used, the breaches found with their locations, and the conditions on the day, on a plan the building owner and the repair crew can actually act on. A verbal it leaks over there is not a leak map. The value of ELD is precision, and precision that lives only in the technician's head is precision lost.

Tie the locations to the building, not to a sketch nobody can register. Each marked breach goes on the plan by area and position, with a photo of the marked spot and the verification cut where one was made, so the repair crew walks straight to it and the next survey can compare against the last. On a covered roof especially, a breach located to a grid square on a clear plan is the difference between one excavation and several exploratory ones.

Capturing all of that as one survey record, the area, the method, the conditions, the breaches, the verification, the repair, and the re-test, photographed and mapped in the field, is exactly the funnel a tool like FieldOS is built to hold. The survey that records its own findings in place, location by location with the photos attached, is the one the owner can hand to a roofer years later and the one that defends the work when a leak comes back and the question is whether the breach was ever found and fixed.

What to document

The ELD record is what makes the survey defensible and the repair findable, and on a roof headed back under overburden it is the only evidence the membrane was tested before it was buried. The breach you located is invisible the moment the soil goes back on, so the map and the record are the proof the survey happened and the repair closed.

Capture it by area as you work: the roof section, the method used, high-voltage or low-voltage, the surface and weather conditions at the time, the grounding confirmed and how, the breaches found and located, the verification result at each, the repair made, and the re-test result. Note the membrane type and whether it was confirmed testable, because a clean survey on a partly conductive roof needs that caveat on the record. Tie the whole package to a plan and photos, since the location is the entire point of the survey.

Common mistakes

• Running the wrong method for the conditions, a high-voltage spark over a covered or wet roof, or a low-voltage survey on a dry exposed membrane that wanted the spark.
• Testing without a confirmed, reliable ground, so the survey reads nothing or reads noise.
• Running ELD on a conductive membrane, black carbon-loaded EPDM, butyl, or a metallic-faced sheet, where current travels through the membrane and no breach can be resolved.
• Ignoring unintended grounds, lightning protection, drains, conduit, counterflashings, so buried metal reads as a breach that is not there.
• Running high-voltage on a wet membrane, which sparks across surface water where there is no hole and creates a shock hazard.
• Trusting a clean low-voltage result through deep or layered overburden, where an insulating layer can block a real breach into a false negative.
• Marking breaches and never verifying them with a cut-down, handing over a list of alarms instead of confirmed holes.
• Repairing a found breach and skipping the re-test, so nobody confirmed the patch reads clean before the overburden went back.
• Treating ELD as a moisture survey, expecting it to map trapped wet insulation instead of locating a breach.
• Burying a roof that was meant to be tested under cover without designing in the conductive ground plane, so it can never be surveyed without excavation.

Field checklist

Standards and references

The document that governs an ELD survey on a specific roof is the membrane manufacturer's requirements and the project specification, because they decide whether ELD is accepted, which method, and what passes. ASTM is the framework around that controlling document, so confirm the figures here against the spec and the current editions before citing them on a submittal.

ASTM D7877 is the Standard Guide for Electronic Methods for Detecting and Locating Leaks in Waterproof Membranes, the umbrella that describes the high-voltage and low-voltage conductance methods on exposed and covered membranes, and it states plainly that ELD supplements rather than replaces visual and infrared inspection. ASTM D8231 is the Standard Practice for the Use of a Low Voltage Electronic Scanning System for Detecting and Locating Breaches in Roofing and Waterproofing Membranes, the prescriptive practice for the low-voltage scanning method run under roughly 50 V. Both standards have been revised across cycles, so confirm the designation and edition the project calls out.

Around the ASTM standards sits the rest of the framework. IIBEC, the roof and waterproofing consultants' body, has published technical guidance on electronic leak detection that is worth reading for its limits as much as its methods. The NRCA Roofing Manual covers testing and inspection practice for low-slope roofing by topic. The moisture-survey methods are their own standards family, separate from ELD, since they answer a different question. Cite the standard that controls the point, confirm the edition, and let the manufacturer's and the project's requirements override any rule of thumb in this guide.

Units, terms, and conversions

ELD uses a small, specific vocabulary, and the same idea reads differently across a spec, a survey report, and a manufacturer's literature, so the terms are worth pinning down.

Voltage is in volts DC: low-voltage scanning runs under roughly 50 V, vector mapping commonly in the tens of volts, and high-voltage spark testing in the thousands, often quoted up to around 12,000 V at very low amperage. A breach is the hole; a holiday is the coatings-and-membrane term for a pinhole or skip. Vector mapping, EFVM, and low-voltage integrity testing all name the wet, perimeter-loop, two-probe method. The high-voltage method goes by spark test or dry test. Overburden is anything over the membrane; an inverted or protected-membrane roof, the IRMA, puts insulation on top of the membrane. ELD locates a breach; a moisture survey, infrared, nuclear, or capacitance, finds trapped moisture in the assembly.

ELD

Electronic leak detection, locating a membrane breach by the electrical current that flows through it to a grounded conductive substrate

High-voltage spark test

A charged brush swept over a dry, exposed membrane that sparks to ground at a pinhole; for bare membranes only

Low-voltage vector mapping (EFVM)

A wetted-surface, perimeter-loop, two-probe method that maps the current vector to a breach; works under overburden

ASTM D7877

The standard guide for electronic methods of detecting and locating leaks in waterproof membranes, covering both voltage methods

ASTM D8231

The standard practice for a low-voltage electronic scanning system for detecting and locating membrane breaches

Overburden

Soil, plants, ballast, pavers, or insulation over the membrane; why low-voltage ELD is often the only practical test

Integrity test

An ELD survey to confirm a new or repaired membrane is breach-free before it is covered or accepted

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• How do you test a TPO seam in the field?
• What is the difference between wet mils and dry mils?
• What is roof flashing?
• Where do most roof leaks come from?
• Why do data centers need liquid-cooling leak detection?