Industries / Construction and property / Commercial and residential construction / Concrete
Repair
Repair finds out why hardened concrete is cracking, spalling or rusting inside, then patches, strengthens or protects it so the structure can stay in service.
7
Milestones achieved, from 1967 to 2013.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1967
Steel plates glued on
L'Hermite and colleagues (1967) and Bresson (1971) proposed gluing steel plates to the concrete with epoxy, and the first uses on a highway bridge and in a building followed in 1966–67, by the account of Empa, the Swiss federal materials laboratory. A beam Empa strengthened this way in 1970 and has kept loaded at 87% of its mean failure load was still in good condition after 47 years.
Credit
L'Hermite and colleagues; Bresson; Empa
1977
Rust mapped by voltage
Steel corroding inside concrete gives off a small voltage that can be read at the surface with a half cell, a reference electrode such as copper in copper sulfate, wired to the steel. ASTM first published a method for this as C876-77. Its appendix says that where readings are more negative than −0.35 V, there is a greater than 90% probability the steel is corroding, and where they are more positive than −0.20 V, a greater than 90% probability it is not; in between the answer is uncertain. FHWA still describes the method for surveying reinforced pavements.
LessonThe standard warns that its numbers should not normally be used on concrete carbonated down to the steel, on indoor concrete that is seldom wet, or after an overlay changes the moisture at the bars, so a map of voltages is a guide to where to look, not a verdict.
Credit
ASTM Committee G01 on corrosion of metals
1991
Carbon strips replace steel
Empa researchers led by Urs Meier proposed in 1987 gluing on strips of carbon fiber reinforced polymer (CFRP) in place of steel plates, and in 1991 the Ibach bridge near Lucerne was strengthened with CFRP strips, which Empa calls a world premiere. The strips are much lighter than steel plates and easier to handle, and Empa says the working hours saved offset the higher cost of the fiber. When the strips were checked in September 2008, the results were "fully satisfactory", and Empa now calls CFRP strengthening a state-of-the-art technique.
Credit
Urs Meier, Empa
1997
A car park deck fails
In March 1997 a 120 t section of the top deck of Pipers Row, a 1965 multistory car park in Wolverhampton, England, collapsed. It began as a punching shear failure, the slab tearing through around the top of one column, and spread to eight more columns. Nobody was hurt. Research for the UK Health and Safety Executive (HSE) found that damage to the waterproof membrane had let water into poor-quality concrete at the top of the slab, where frost weakened it, and that small oversights in design, construction, inspection and maintenance added up. It recommended that deterioration be checked by a structural engineer "rather than considered as a cosmetic repair".
LessonThe HSE warned that routine corrosion surveys may miss loss of strength, so a decaying surface over a column has to be treated as a structural question.
Credit
HSE
2005
Classes for repair mortars
The European Committee for Standardization (CEN) approved EN 1504-3 on 29 April 2005, one part of the EN 1504 series on products and systems for protecting and repairing concrete. It sorts repair mortars into four classes. Structural classes R4 and R3 must reach 45 MPa and 25 MPa in compression and bond to test concrete at 2.0 MPa and 1.5 MPa; non-structural R2 and R1 need 15 MPa and 10 MPa. Member countries had to withdraw conflicting national standards by December 2008.
Credit
CEN Technical Committee 104, sub-committee 8
2007
Half of repairs fail early
CONREPNET, a network of European research and industry bodies funded by the European Commission and coordinated by Stuart Matthews of BRE, collected the records of past repairs on about 230 structures. The survey estimated that almost 50% of repairs showed signs of failure within five years, usually as renewed corrosion, cracking, debonding or spalling. Reviewers traced failures to the design of the repair (38%), poor workmanship (19%), a wrong diagnosis of the original damage (16%), the wrong materials (15%) and other causes (12%). The findings were published by BRE in December 2007 in a report by G. P. Tilly and J. Jacobs.
LessonMost failures traced back to decisions made before anyone mixed a repair mortar.
Credit
CONREPNET; Tilly and Jacobs; Stuart Matthews, BRE
2013
A code for repairs
Codes for new buildings, such as ACI 318, say little about damaged ones, so officials often required repairs to meet new-building rules, which could make repairs costly or lead to demolition. In 2013 the American Concrete Institute published ACI 562-13, its first repair standard and the first written to work with the International Existing Building Code, according to an article by Keith Kesner, Kevin Conroy and Lawrence Kahn in STRUCTURE magazine (September 2014). The same authors put US concrete repair spending at $18 billion to $25 billion a year, much of it "repairs to the repairs".
Credit
ACI Committee 562
Watch for
Cement that reseals its cracks
Researchers say a cement made with less than 0.15% polymer sealed a crack about 2 mm deep within about four hours in laboratory tests, recovered much of its strength over repeated cracking, and has kept healing in a test piece damaged again and again over five years. They described it in Nature Communications on 1 August 2026, and their first target is the cement in geothermal and oil and gas wells.
Achieved whena laboratory outside PNNL reproduces the healing and strength recovery, or a trial on a working structure reports results.
Claimed by
Carlos Fernandez, Chao Zeng and colleagues, Pacific Northwest National Laboratory (PNNL), September 2026
Open
01
An open register of repairs
A public register of at least 300 concrete repairs, each with the structure type and exposure, the diagnosed cause, the repair method and material in EN 1504 terms, the repair date and its condition at each later inspection, drawn from public inspection reports and published case histories. From it, publish how long repairs last by method and exposure.
The briefCredit · unclaimed02
Test R4 mortars independently
An independent test of ten repair mortars sold as EN 1504-3 class R4, bought from ordinary builders' merchants, mixed as their labels say and tested in a university laboratory for compressive strength (EN 12190) and for bond to standard concrete slabs (EN 1542). Publish the results beside each maker's declared values and beside the class limits, with a table of what every product's declaration of performance states.
The briefCredit · unclaimed
Open 01 · the brief
An open register of repairs
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public register of at least 300 concrete repairs, each with the structure type and exposure, the diagnosed cause, the repair method and material in EN 1504 terms, the repair date and its condition at each later inspection, drawn from public inspection reports and published case histories. From it, publish how long repairs last by method and exposure.
Why it is still open
The figures most often quoted on repair life come from CONREPNET's 2007 survey of about 230 structures, whose case records were summarized in a report sold by BRE rather than released as data. A 2020 report to CROSS-UK, the structural safety reporting scheme, says many older car parks received "cosmetic patch repairs with incompatible materials" that may not have restored their strength. Since 1 August 2024 New York City has required the owner of every parking structure in its inspection program to have filed an engineering report or an initial observation report with its Department of Buildings, and the city maps which have filed, so dated condition records are now produced in bulk.
Built with what exists
Public inspection filings and published repair case histories; an AI model that reads PDF inspection reports into the register's fields, with a person checking every record (see the Note "Seven more flagged" for a model reviewing records people had already checked); Python and a public spreadsheet or SQLite file.
First step
Write the record fields from CONREPNET's failure modes and causes, and code 20 repairs from one city's or one state's public reports.
Done looks like
300 repairs, each traced through at least two dated inspections, published under an open license, with the share showing renewed corrosion, cracking, debonding or spalling at 5 and 10 years, by method.
Open 02 · the brief
Test R4 mortars independently
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
An independent test of ten repair mortars sold as EN 1504-3 class R4, bought from ordinary builders' merchants, mixed as their labels say and tested in a university laboratory for compressive strength (EN 12190) and for bond to standard concrete slabs (EN 1542). Publish the results beside each maker's declared values and beside the class limits, with a table of what every product's declaration of performance states.
Why it is still open
EN 1504-3 has the maker carry out the initial performance tests for a product's class, and its identification tests accept a compressive strength above 80% of the maker's declared value. The nearest independent test is Brian Vukindu's 2020 master's thesis at the University of Cape Town, which tested twelve proprietary repair mortars against EN 1504-3. One mortar that its compressive strength put in a structural class bonded at only 0.61 MPa against the 1.5 MPa and 2.0 MPa the structural classes require, and the thesis concludes that makers' data sheets should be verified before a product is specified. Its mortars were supplied by the makers, and it measured bond by ASTM C1583 rather than EN 1542. A 2026 open project on GitHub, Concrete-Repair-RAG, answers questions about Texas DOT repair specifications with Claude, including whether a value meets a requirement, and its author notes that tables in the source PDFs "are not fully captured".
Built with what exists
Ten commercial R4 mortars; a university materials laboratory with a compression machine and a pull-off bond tester; standard concrete slabs as the base; an AI model that reads the makers' PDF declarations into a table, checked by hand; Python.
First step
Collect the declarations of performance for 20 R4 mortars from at least five makers, tabulate them by hand, and ask a university concrete laboratory to host the tests.
Done looks like
For all ten products, at least three compression prisms and five pull-off tests each at 28 days, published with each declared value and the R4 limits of 45 MPa and 2.0 MPa, and a count of products that fall short of their declaration or their class.