What causes roof blisters, and when should you repair them?

What causes roof blisters?

A blister needs space to start: an existing void, incomplete contact or a weak bond. Heating expands trapped gas and raises vapor pressure, while warm bitumen stretches more readily. CRCA Bulletin 46 (July 1996) describes cyclic growth: the membrane resists full recovery as it cools, and the enlarged void can draw in more air or vapor.

NRC Construction Technology Update 38 (June 2000) identifies installation contributors including mopping skips, debris, uneven substrates and poorly relaxed sheets. An unbonded interface establishes where contact failed; it does not identify which installation error caused it. For hot-applied SBS, NRC emphasizes compatible asphalt, correct mopping temperature and prompt sheet placement, with dry materials and intimate interply contact. NRC CTU 38.

This guide covers low-slope BUR, modified bitumen, adhered single-ply and liquid-applied roofing. Confirm the designed attachment first: normal movement in an intentionally unadhered field is different from separation at a bonded interface. Shingle granule blisters and spray-foam defects require their own investigations. Use the BUR installation and modified-bitumen installation guides to establish the actual assembly.

Is the blister caused by air, moisture or adhesive solvent?

Air, water vapor and adhesive solvent can produce similar raised areas; identify the failed layer before assigning the cause. Water vapor in a cavity, liquid water in insulation and an exterior leak are three separate findings. A dry core narrows the condition at that point; it does not prove an air-only cavity or exclude an earlier moisture episode.

NRC report IRC-IR-759 (March 1998) summarizes blister-gas relative humidity of 30% to 75%, but the temperature basis matters: samples 5, 6 and 7 gave 50% to 75% after correction using mean daily field temperatures; samples G and 8 lacked sampling dates and remained on a laboratory-temperature basis. CTU 38 (June 2000) summarizes 50% to 75%. These are blister-gas observations, not insulation moisture percentages or repair thresholds. NRC IRC-IR-759; NRC CTU 38.

Record installation date, adhesive identity, recent solvent cleaning, rain history, daily swelling, interior humidity and whether the blister preceded a leak. Carlisle identifies trapped-solvent blistering for its named low-VOC bonding adhesives; apply that explanation only after matching the product and installation history. Carlisle TecTopics.

Separate exterior entry, moisture enclosed during construction and interior moisture reaching a cold layer. Match opening results to weather, layer position and changes in building use. Use roof leak diagnosis for entry tracing and vapor-retarder and air-barrier investigation for the interior route. A wet layer identifies material requiring attention without establishing how water reached it.

What does the shape of a blister tell you?

Use Table A to select the next observation; its actions are Anvilfield recommendations, with the primary evidence named in each row.

Record two plan dimensions, maximum rise, surfacing condition, nearby seams, traffic exposure, date and surface temperature. Repeat measurements under comparable conditions so a hotter afternoon does not become a false growth trend. CRCA Bulletin 46 (July 1996) advises avoiding foot pressure on blisters, especially when cold. Plan a compatible closure before any diagnostic opening.

Which roof layer has separated?

Locate the separation on the actual assembly. Add or omit layers in these simplified branches to match the roof.

• BUR/SBS: surfacing → plies/interply asphalt → cover board or substrate → insulation/attachment → vapor control if present → deck.
• Adhered single-ply: sheet → bonding adhesive → board facer/core → lower insulation/attachment → vapor control if present → deck.
• RoofPro: liquid-applied membrane/reinforcement → specified primer → identified existing membrane or structural substrate → underlying assembly.

Photograph both mating surfaces before handling changes the evidence. A clean separation at a bond line is adhesive failure; torn fibers or sheet material on the opposing face indicate cohesive failure within the material. Record the actual plane of delamination, because peeling a sheet from a failed facer will not restore the board.

A membrane patch cannot restore a failed insulation facer or deteriorated substrate beneath it. For dry bond failure, map the lost attachment and specify its restoration using the board attachment and deck/substrate records. A core describes local condition; an uplift-capacity conclusion requires the applicable assembly documents and project testing.

Where should you take roof blister cores?

Select openings to test competing explanations: the failure interior, its transition, an associated moisture anomaly, a same-assembly comparison and any different morphology. Anvilfield estimates six initial locations for the one-zone investigation in Table C, adding locations when boundaries, controls or construction differ. Table C assigns positions for a zone containing a field blister, moisture anomaly and lap blister; enter the observed results during the investigation. Six is a planning allowance, not a standard-mandated density.

A small core provides layer and moisture samples; a controlled cutback exposes more of the failure interface. If cavity-gas analysis is needed, arrange collection before opening vents the cavity. Record which plies or boards remain bonded, the condition of each exposed mating face and water at each level; keep wet upper material from contaminating a lower-layer specimen.

For construction work, OSHA 29 CFR 1926.501(a)(2) requires the employer to determine that walking/working surfaces have adequate strength and structural integrity before allowing employees onto them. OSHA 1926.501(a)(2). Plan safe access and fall protection, identify material hazards and locate the deck and services before cutting. Anvilfield recommends stopping roofing samples above the structural deck and assigning deck sampling separately.

For every location record roof zone, fixed-point offsets, date/time, weather, instrument/model/scale, opening dimensions and depth, layer and container IDs, handling, laboratory method, both-face photographs, closure product/detail, closure date and completion photograph. Seal retained specimens promptly and preserve layer order. Arrange owner authorization, any required warranty coordination, compatible closure materials and closure responsibility before mobilization. Open only what the crew can close watertight before rain or the end of the shift.

Why ASTM D3617 does not set a universal existing-roof core count

ASTM D3617/D3617M-17(2022) covers BUR sampling during application before flood coating and top surfacing; its public scope does not establish a universal sampling density for aged blistered roofs. ASTM D3617/D3617M-17(2022).

Add a cause-disposition line to each sample group: suspected mechanism, supporting observation, conflicting observation, failed layer, retained/rejected/unresolved conclusion and next observation needed. Preserve a dry debonding sample or wet comparison even when it contradicts the first diagnosis. State the expansion triggers in the scope so additional openings answer a defined question.

How do you confirm moisture instead of guessing from a scan?

Confirm scan anomalies with physical evidence from the relevant layer, using a dry surface for an electrical-impedance survey. Tramex’s April 23, 2019 procedure requires the surface to be free of rain, dew and debris; record seams, repairs and changes in construction separately. ASTM C1153-23 addresses nighttime infrared location of wet insulation and invasive verification, while ASTM D7954/D7954M-22a addresses comparative electrical impedance and core verification; neither establishes moisture entry or cause by itself. ASTM C1153-23; ASTM D7954/D7954M-22a.

ASTM C1616-25 expresses insulation moisture as a percentage of dry weight. Subtract dry mass from received mass, divide by dry mass and multiply by 100: moisture (%) = (received mass − dry mass) / dry mass × 100. ASTM C1616-25. Scanner scale units are not this percentage. Record the laboratory method and compare the same material and layer against the selected project criteria.

When should you reject the test result?

Reject an impedance interpretation when the assembly falls within Table D’s exclusions. ASTM D7954/D7954M-22a distinguishes aluminum foil from liquid-applied aluminum-pigmented emulsified asphalt coatings, which normally do not affect readings. These are impedance limitations, not blanket infrared exclusions.

Extend checks when an apparently unaffected comparison is wet or an opening contradicts the scan. Keep tested points, interpreted extent and unresolved cause separate; use infrared survey interpretation for acquisition conditions and mapping.

Worked example: what intact and ruptured blister samples revealed

NRC/NRCA’s March 1998 IRC-IR-759 investigation examined 16 membrane samples and 13 gas canisters from five locations. Table E compares intact, ruptured and comparison samples from the seven-year-old group G roof. NRC IRC-IR-759.

Using NRC’s March 1998 results, Anvilfield estimates that the ruptured, water-containing G-2 blister’s cap sheet retained 27.5% of the same-roof control’s mean maximum tensile load, based on 44 N versus 160 N. Divide the blister mean by the comparison mean: 44 / 160 × 100 = 27.5%, giving a 72.5% reduction. The denominator is the same-roof G-4 comparison mean.

NRC tested 25 mm-wide specimens using ASTM D5147-95, a 75 mm gauge length, 50 mm/min crosshead speed and laboratory conditions of 23 ± 2°C and 50 ± 5% RH. Table E preserves specimen counts and standard deviations from Table 4.1. NRC proposed water-related weakening at the reinforcement/bitumen interface as an explanation for G-2; that interpretation is distinct from the measured loads.

The comparison sample that prevents a false conclusion

NRC also found an unusually weak cap sheet at control 8C, where no blister or water was observed; its cause remained unexplained. Weakness alone therefore cannot identify moisture damage. Anvilfield’s repair interpretation is to prioritize the ruptured wet location, retain contrary observations and establish removal limits from layer and boundary checks before extending one diagnosis to the roof.

When should you monitor, reinforce or cut out a blister?

Repair water-admitting ruptures and vulnerable membrane before relying on a size rule. Fatigue cracking, exposed reinforcement, failing laps, drainage obstruction and traffic exposure change disposition. Provide prompt compatible weatherproofing at an active opening, then complete the affected-layer repair scope.

CRCA Technical Bulletin 46 (July 1996) recommends marking BUR blisters approaching 600 mm, about 2 ft across, for periodic measurement and suggests repair near 1.5 m, about 5 ft, or in high traffic. NRC Construction Technology Update 38 (June 2000) recommends repair of SBS blisters that rupture, reach about 1.5 m across or separate one-third or more of a seam’s width. Use the installed system’s instructions for single-ply and liquid-applied roofing.

A dry blister still needs repair when it compromises drainage, a seam or required attachment. Size and moisture findings answer different parts of the decision.

Protective reinforcement is an option for a firm, intact BUR blister with sound plies and a suitable detail, as described in Professional Roofing (March 1997). IKO’s VER1022 smaller/larger BUR categories specify no numerical diameter.

Why do manufacturer blister instructions differ?

The instructions address different materials, failure planes and phases of work. Match the installed product and defect before choosing between waiting, re-brooming and cutting out. Table G compares the named products and failure planes.

Carlisle’s December 22, 2025 instructions for EPDM/TPO Low-VOC and EPDM x-23 Low-VOC Bonding Adhesives specify at least 30 minutes of sheet relaxation and 300 sq ft of finished roof area per 5-gallon pail. That is finished bonded area, not a separate allowance for each mating surface. Use the tack test and broom or roll immediately after setting the sheet. Carlisle TecTopics. Record the return condition and bond outcome; elapsed days alone do not accept an old, wet, ruptured or persistently unbonded area.

Sika’s RoofPro manual, effective May 1, 2024, directs application during falling ambient and substrate temperatures on substrates liable to outgas. Use the specified primer, resin and reinforcement while correcting the cause. Sika RoofPro manual.

Should you slit or vent a roof blister?

CRCA Bulletin 46 describes an X-cut alternative and historical proprietary vents; the March 1997 Professional Roofing article favors removing unbonded membrane because X-cut flaps can retain unbonded edges. CRCA 46; Professional Roofing, March 1997. Follow the current compatible repair detail for the installed assembly. A pressure-relief opening neither demonstrates dry insulation nor restores failed attachment.

Where should the repair stop, and when is replacement necessary?

Set the unbonded-membrane boundary, wet/damaged-material boundary and required sound tie-in separately. Anvilfield’s boundary method uses four plan marks: V for visible symptom, B for debonding, W for confirmed/interpreted wet or damaged layers, and R for final repair footprint. The method combines IKO’s affected-material repair guidance with core verification and bonded-edge repair practice. IKO Roof Repair; ASTM D7954/D7954M-22a; Professional Roofing, March 1997.

Map the dome, expose the failed interface and bracket its transition and each affected underlying layer. Removal follows the failed membrane and wet/damaged materials at each layer; lay out the system’s tie-in beyond those removal lines on accepted substrate. Accept an edge only when it has sound bond or specified attachment, underlying material meets the selected dry/sound criteria, and clean tie-in space is available. Extend any wet, weak or unbonded edge and verify again; assign a next action to inaccessible edges.

Reject a patch sized only to the crown when separation continues beyond it. Reject membrane-only work where facer, board attachment or deck condition remains unresolved. A restoration coating does not correct a loose or wet underlying layer.

Separate cap-sheet replacement, membrane replacement over retained serviceable insulation, board replacement and full assembly tear-off. Widespread failed attachment or the inability to make reliable tie-ins moves the scope toward larger replacement. Use tear-off versus recover for project eligibility; blister count and roof age do not locate removal lines. The manufacturer determines system-specific repair/warranty acceptance, and the structural engineer determines repairs to a compromised deck.

Repair-area takeoff is not the same as blister area

The March 1997 Professional Roofing article describes a first BUR patching ply extending at least 6 in beyond the cutout and succeeding plies at least 3 in beyond the preceding ply. Its preparation distance of at least 18 in beyond the blister provides working space. Professional Roofing, March 1997. Table H applies that dated sequence to a cutout removing two failed BUR plies. The article generally matches replacement ply count to removed ply count, adding a fill ply where needed to correct a depression; use the current approved detail for installation.

Method: Anvilfield estimates dated November 5, 2026 add twice each edge extension to the rectangular dimensions, multiply length by width and sum ply areas for material. The first ply is 42 sq ft and the second 48.75 sq ft, totaling 90.75 sq ft before waste; displayed quantities round to whole square feet. The selected preparation envelope extends 18 in beyond each cutout edge, while the finished patch extends 9 in. Measure board replacement separately and agree unit prices for concealed extent. Compare larger alternatives using the repair/replace break-even worksheet.

What should the investigation cost and schedule include?

Anvilfield estimates $2,650 for the defined six-location investigation and permanent sampling closures, including one four-hour field visit by two people plus preparation and reporting. A weather return raises the total to $3,250. Table I uses selected November 5, 2026 planning rates and laboratory allowances for one accessible roof zone with closures completed during the visit.

Method: multiply person-hours by $125, add six $50 material/closure allowances and six $75 laboratory specimens, then travel. Closure labor is already in field hours. Six locations do not mean every layer is laboratory-tested; additional specimens require their own count and quote.

Price defect repairs, a full-roof infrared survey, special access, hazardous-material work and deck repair separately. Repair quantities follow removal layers, board thickness/attachment, ply count and tie-ins. Obtain the laboratory’s turnaround before promising a report date; added boundary openings add field and closure time. Coating repairs follow the selected product’s cure and weather window.

Require the numbered map, layer log, sample results, rejected causes, confirmed removal limits, repair detail and closure photos. Contractors offering layer-specific blister investigations build service territories with PorchRocket Ground Control commercial building lists selected by roof material and roof age band.

How often should monitored blisters be checked?

CRCA’s June 2014 Roofing Preventative Maintenance advisory recommends inspections twice yearly, in spring and fall, plus checks after major weather or rooftop equipment work. CRCA maintenance advisory. Use that baseline with the actual warranty’s maintenance requirements.

Return sooner when a blister grows, loses surfacing, cracks, reaches a seam, becomes exposed to traffic or corresponds with new water entry. These changes trigger investigation or repair rather than waiting for the seasonal visit. Store the original outline, dimensions and dated photos, compare similar temperature conditions, and assign the next visit and responsible party in the inspection program and roof asset file.

Field checklist and repair acceptance record

Label the record with owner, roof zone and date. Give every blister, test location and closure a unique ID; assign monitor, reinforce, remove locally, extend investigation or replace affected assembly to each location.

• ☐ Identify roof zone, membrane, attachment, age, prior repairs and applicable manufacturer documents.
• ☐ Map each blister with ID, two plan dimensions, rise, date, temperature and photographs.
• ☐ Record rupture, water entry, surfacing loss, traffic exposure and affected laps.
• ☐ Check drainage obstruction and the specified attachment, including dry facer or bond failure.
• ☐ Establish safe access, deck condition, material identification and permanent-closure responsibility before cutting.
• ☐ Select failure interior, transition, comparison and distinct-morphology openings; record the question each answers.
• ☐ Confirm that the survey method suits the membrane, insulation and conductive components.
• ☐ Photograph both sides of the failure plane and log each layer separately.
• ☐ Seal and identify retained specimens; record handling and the laboratory method and denominator.
• ☐ Record supported, rejected and unresolved mechanisms; add locations where boundaries or comparisons disagree.
• ☐ Separate tested points, interpreted moisture extent, debonded extent and the proposed repair footprint.
• ☐ Identify the moisture source or assign the remaining source investigation.
• ☐ Select the current compatible repair detail, materials, tie-in and attachment requirements.
• ☐ Extend failed boundary checks until each affected layer has an accepted perimeter and a constructible tie-in.
• ☐ Measure removal, each replacement layer, finished patch footprint and sheet quantities separately.
• ☐ Close every opening; record attachment, checked laps, restored surfacing/drainage, photos and warranty acceptance.
• ☐ Assign a disposition and next inspection date to every marked blister.

The acceptance record identifies who installed the repair and when, and links each finished footprint to its layer and attachment record. Check completed thermoplastic welds after cooling under the applicable instructions; Sika’s handbook, print code 1328/5M/020524, requires weld-continuity checks. Sika Applicator Handbook. Use single-ply seam QA for finished patches and warranty documentation for required manufacturer acceptance.

Sources, review and citation

Anvilfield. Reviewed November 5, 2026. Update log: verified source editions and repair triggers; clarified sampling, specimen geometry, adhesive coverage and repair quantities. ASTM citations refer to public publisher scope and significance statements.

How to cite this page: Anvilfield, What causes roof blisters, and when should you repair them?, 2026, https://anvilfield.com/field-guides/roofing/roof-blister-investigation-repair-boundaries/

Related tools

FAQ

People also ask

• Where do most roof leaks come from?
• What is the difference between a roof recover and a tear-off?
• How do you test a TPO seam in the field?
• What does ASTM C1153 say about infrared roof surveys?
• What is a roof cover board?

Sources checked

• S1. CRCA, Technical Bulletin 46, Blistering in Built-Up Roofs, July 1996. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S2. National Research Council Canada, Construction Technology Update 38, Blistering in SBS Polymer Modified Bituminous Roofs, June 2000. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S3. NRC/NRCA, IRC-IR-759, Investigating the Cause of Blistering in SBS Polymer Modified-Bitumen Roofing Membranes, March 1998. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S4. IKO Commercial Roofing Technical Specification Manual, Part 16, Roof Repair, VER1022. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S5. Sika, Roofing Applicator Handbook, Sarnafil and Sikaplan, print code 1328/5M/020524. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S6. Sika, Sikalastic RoofPro Job-Site Applicator Manual, effective May 1, 2024. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S7. ASTM D7954/D7954M-22a, publisher public scope and significance/use; active edition shown. Reviewed November 5, 2026. Public publisher scope and significance/use.
• S8. ASTM C1153-23, publisher public scope and significance/use; active edition shown. Reviewed November 5, 2026. Public publisher scope and significance/use.
• S9. ASTM C1616-25, publisher public scope; active edition shown, updated October 7, 2025. Reviewed November 5, 2026. Public publisher scope and significance/use.
• S10. ASTM D3617/D3617M-17(2022), publisher public abstract/scope; active edition shown. Reviewed November 5, 2026. Public publisher scope and significance/use.
• S11. Carlisle SynTec, Cold Weather Application and Storage Tips for Low-VOC Bonding Adhesives and Primers, displayed December 22, 2025. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S12. NRCA, Professional Roofing, March 1997, hands-on blister evaluation and repairs, pp. 20 onward. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S13. Tramex, How to perform a non-destructive flat roof moisture survey to ASTM D7954, April 23, 2019. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S14. CRCA, Roofing Preventative Maintenance advisory, June 2014. Reviewed November 5, 2026. Primary document; edition and application retained in the article.
• S15. OSHA, 29 CFR 1926.501, Duty to have fall protection, paragraph (a)(2). Reviewed November 5, 2026. Official regulation: walking/working-surface strength and structural integrity; construction-work scope.