CFRP Strengthening

CFRP strengthening covers the addition of load capacity to existing structures by bonding carbon fibre reinforced polymer composites to them - plates and fabrics glued to the surface, and rods set into slots cut just below it. The technique belongs to the wider family of fibre composite strengthening, which also uses glass and aramid fibres, but carbon is the fibre that dominates structural work and the one this page is named for. The composite acts as external reinforcement: it does the same job as bars cast into the concrete, except that it arrives decades later, on the outside, and is held on by an adhesive layer rather than by cast-in bond.

The reference documents in UK practice form a small and stable set. The Concrete Society's Technical Report 55, "Design guidance for strengthening concrete structures using fibre composite materials", is the design guide - the current edition is the third, published in 2012 with an amendment dated October 2013, running to 187 pages. Its published scope covers strengthening with fibre composites bonded on the surface or embedded in the concrete, taking in material types and properties, a review of applications, structural design of strengthened members, flexure, shear, axially loaded members, emerging technologies, workmanship and installation, and long-term inspection and monitoring. Its companion, Technical Report 57, "Strengthening concrete structures with fibre composite materials: acceptance, inspection and monitoring", published in 2003, covers the other half of the job: long-term performance, installation techniques, inspection regimes and the elimination of future damage. The two are published to be read together, and the industry treats them as a pair - the design lives in TR55, the proving lives in TR57.

For highway structures the same ground is covered with statutory force by the Design Manual for Roads and Bridges. CD 371, "Strengthening highway structures using fibre-reinforced polymers and externally bonded steel plates", replaced the earlier BD 84/02 and BD 85/08 and folded in provisions from the withdrawn BA 30/94 on steel plate bonding; the document names TR55 and CIRIA C595 as the third-party methods it draws on, and it repeatedly refers the designer to TR55 for the detailed models. The revision this page was checked against is Revision 0, published March 2020 - the copy downloaded during research carries a superseded stamp, and the open question about the current revision is held in the caveats section. CIRIA's C595, "Strengthening metallic structures using externally bonded fibre-reinforced polymers", published 2004, extends the technique to steel, cast iron and wrought iron, covering design, selection, installation and maintenance with appendices on case histories, fatigue and the analysis of adhesive stress distribution.

The international frame matters because UK designers cite it constantly. The fib - the International Federation for Structural Concrete - published Bulletin 14, "Externally bonded FRP reinforcement for RC structures", in July 2001, with chapters running from materials through flexure, shear and torsion, confinement and detailing to practical execution and quality control; its successor, Bulletin 90, "Externally applied FRP reinforcement for concrete structures", followed in 2019 covering both externally bonded and near-surface-mounted reinforcement, written with the stated aim of feeding composites into the next versions of Eurocode 2 and Eurocode 8. In the United States, ACI's guide PRC-440.2-23, "Design and Construction of Externally Bonded Fiber-Reinforced Polymer (FRP) Systems for Strengthening Concrete Structures", is the 2023 edition of the document much of the world's project documentation cites, and ACI has since added a code-format document for the same subject. The International Institute for FRP in Construction maintains a public index of the design guides, codes and specifications across countries, which is the quickest way to see how wide the published base now runs.

The regulatory picture in England is the ordinary one for structural work, with one recent change worth stating precisely. Strengthening an existing building is commonly building work under the Building Regulations 2010, engaging the structural requirements supported by Approved Document A and the dutyholder and competence regime added to the regulations under the Building Safety Act 2022. The Building Safety Regulator has been a standalone body corporate since 27 January 2026, when the Building Safety Regulator (Establishment of New Body and Transfer of Functions etc.) Regulations 2026 (SI 2026/20) came into force, transferring the building safety functions previously exercised through the Health and Safety Executive; the regulator is sponsored by the Ministry of Housing, Communities and Local Government. It is no longer part of HSE, and descriptions that place it there are out of date. Site work sits under CDM 2015 as construction work does everywhere else. On the highway network, CD 371 itself is the controlling document and the overseeing organisation's acceptance runs through it.

The trade bodies around the documents form a compact set. The Concrete Society publishes and sells the technical reports through its bookshop. The Concrete Repair Association - whose members carry out most building-side strengthening alongside repair - publishes a free advice note on structural strengthening with fibre reinforced polymers, an overview of the material types and the structures they strengthen, and sits inside the Structural Concrete Alliance with the Corrosion Prevention Association and the Sprayed Concrete Association as a single voice for the structural concrete refurbishment industry. System suppliers - Sika is the example used throughout this page because its UK documentation is public and complete - publish the product data sheets, method statements and design software that carry the practical numbers.

Two boundaries keep the subject defined, because the same fibres appear in adjacent technologies this page is not about. Internal FRP reinforcement - composite bars cast into new concrete in place of steel - is a different discipline with its own American code for glass fibre bars. All-FRP construction - bridges and structures made of the composite rather than strengthened by it - has its own DMRB design document, CD 368, and a European technical specification for fibre-polymer structures listed in the international index. Fabric systems carried in a cementitious rather than polymer matrix are sold alongside the epoxy systems and change the bond and fire questions; they are named here only to mark the edge of the page. Masonry strengthening with bonded composites has a dedicated American guide of its own. This page is about one thing: adding capacity to existing structures with externally applied carbon composites and the adhesives that connect them.

Finally, what this page is not. It is not a design guide and it does not answer "how much stronger can my beam get", because the published guidance does not answer that in general terms either: the gain available depends on the member, the existing reinforcement, the substrate, the failure mode that governs and the load case that must survive without the strengthening. Where this page names a document, the document is the authority. Where it quotes a number from a product data sheet, that number belongs to the named product and edition, not to the technology at large.

Structures get strengthened because something changed - the load, the structure, or the knowledge about one of them. The published use lists in system documentation are unusually candid catalogues of what actually goes wrong in buildings, and the one attached to the plate system used as this page's worked example reads as a fair census of the market. Increased load carrying capacity: heavier floor loading, installation of heavier machinery, stabilising vibrating structures, changes of building use. Damage to structural elements: deterioration of the original materials, corrosion of the steel reinforcement, and accidents - vehicle impact, earthquakes, fire. Serviceability: reducing deflection and crack widths, relieving stress in the existing reinforcement, improving fatigue behaviour. Changes to the structural system: removal of walls or columns, and removal of floor and wall sections to create access and openings. Resistance to events: earthquake, impact, explosion. And the quietest category, repair of design or construction defects - insufficient reinforcement or insufficient structural depth, which is the polite description of a drawing error discovered years later.

Each of those categories has a typical story behind it. The change-of-use case is the office floor asked to carry archive storage, the residential slab asked to take a plant room, the mezzanine that acquires a heavier process line; the loading changed and the structure did not. The openings case is the stair or lift shaft cut through an existing slab: the bars that used to run through that rectangle carried load, they have just been cut, and bonded plates around the opening put the lost tension path back. The corrosion case is the car park or balcony where the reinforcement has lost section; strengthening here commonly follows concrete repair rather than replacing it, restoring capacity that the repair mortar alone does not. The defect case needs no story - a member was built with less steel than the design required, and bonding the missing tension capacity to the outside is commonly cheaper and faster than the alternatives. On bridges, the driver is assessment: structures assessed against current loading - the highway assessment regime that CD 371 points to for establishing what a structure can carry - are found wanting, and strengthening closes the gap between assessed and required capacity so the network keeps running.

Two disciplines sit in front of every one of those decisions in the published process. The first is appraisal of the existing structure. Strengthening design starts from what is actually there - geometry, reinforcement, material strengths, condition, defects - not from the original drawings, and the Institution of Structural Engineers publishes the standard reference for that activity, "Appraisal of existing structures", in its third edition since 2010. The appraisal decides three things the strengthening design cannot proceed without: what the structure can carry now, why it is deficient, and whether the deficiency is real - it is a recurring observation in the field that sharper assessment sometimes removes the need to strengthen at all, because the conservatism in the original design or the earlier assessment was the deficiency.

The second discipline is the option study. CD 371 requires an options report and an economic evaluation comparing the proposed strengthening technique with other methods before the technique is adopted - a requirement worth noting because it treats bonded FRP not as the default answer but as one candidate among several - the alternatives the published guides name run from steel plates and section enlargement to external post-tensioning, propping and replacement - to be justified on the specific facts. Building-side projects have no equivalent statutory step, but the same comparison commonly happens in the structural engineer's options appraisal, and the alternatives section later in this page sets out the candidates the comparison runs through.

One structural fact shapes the whole "why" conversation and is worth stating early because it surprises people arriving from new-build design. Bonded FRP is commonly installed on a structure that is already carrying its permanent loads, and unstressed reinforcement added to a loaded member only participates in the loads applied after it is installed. The dead load is already in the concrete and the existing steel; the plate joins for the live load, the new machinery, the extra storey. CD 371 records the two ways round this where the permanent loads themselves are the problem: prestressing the FRP, or jacking the structure during installation so that some of the existing load transfers into the new material when the jacks come down. Both approaches let a greater proportion of the FRP's strength be used; both add cost and complexity, which is why most installed systems are unstressed and most strengthening briefs are written around additional imposed load rather than existing dead load.

The building-side market runs on the coupling between repair and strengthening, and the trade body's own name is the map: the Concrete Repair Association describes its territory as concrete repair, corrosion protection, structural strengthening and crack bridging - four activities, one industry, because the structure that needs its capacity restored is so often the structure whose reinforcement corroded in the first place. The association's free advice note on FRP strengthening describes the materials in usefully plain terms - flat layers of unidirectional fibres, or woven fibre fabrics, embedded within a layer of light polymer matrix material - and states its own purpose as helping readers avoid basic errors when employing, specifying and applying such systems, which is a fair summary of where building-side projects go wrong: not in the mathematics but in the employing, the specifying and the applying.

The scale of what is being corrected varies from the domestic to the infrastructural, and the technique's economics track access more than area. The material is expensive by weight and trivial by project cost; the money is in the preparation, the access, the specialist labour and the proving. That is why the technique wins where its particular advantages bite - no added dead weight, millimetres of added depth, no hot works, short possessions, installation from a scaffold or a mobile platform without cranage - and why it loses where a simple steel beam and two padstones would do. The published literature is consistent on the pattern: bonded composites are the answer to a specific family of problems, not a universal upgrade, and the projects that go wrong are commonly the ones that skipped the appraisal and the option study and went straight to the product.

Three fibre families make up the published composite strengthening palette. Carbon delivers the highest stiffness and strength, does not corrode, and resists fatigue well; it is the default for structural strengthening and the fibre behind every product named on this page. Glass is cheaper, heavier for the same strength and less stiff; it appears in wraps and in ground-anchor and bar products more than in plate strengthening. Aramid - the para-aramid fibre best known under the Kevlar brand, created by Stephanie Kwolek and now presented by Arclin - brings toughness and impact resistance with inherent heat resistance in the fibre itself, and appears where impact or blast is part of the brief. The current edition of the American guide adds basalt to its keyword list, a signal of where the market is heading, though carbon remains the working answer for load-bearing strengthening. What every family shares is the composite construction: fine fibres carrying the load, embedded in a polymer matrix - commonly epoxy - that holds them in line, shares load between them and protects them.

The properties that make carbon composites worth gluing to a structure are easiest to see in a published data sheet. A standard pultruded carbon plate carries a mean tensile strength of 3,100 N/mm2 with a 5% fractile of 2,900 N/mm2, a mean modulus of 170,000 N/mm2, and an elongation at break of 1.80%, at a fibre volume content above 68%. Two of those numbers repay a second look. The strength is roughly six times that of ordinary reinforcing steel, in a strip 1.2mm thick that a fitter carries under one arm. The elongation figure is the caution: the material is linear elastic to failure, with no yield plateau, so a strengthened member's ductility has to come from somewhere else - a point the design section returns to. The plates ship in standard widths of 50, 80, 100 and 150mm at 1.2mm thickness, with a 60mm wide 2.6mm section for heavier duties, cut to length from rolls of up to 250m, which is why a bonded plate has no joints along a beam however long the beam is.

Form follows application, and the market has settled into five forms. Pultruded plates - unidirectional fibres drawn through a die with resin, cured in the factory to a constant cross-section - are the workhorse for flexural strengthening of flat soffits and beam faces, arriving as a finished composite that only needs adhesive. Fabrics - woven or stitched sheets of dry fibre, in the common UK range from around 235 to 900 grammes of fibre per square metre - are impregnated on site by wet or dry lay-up and conform to curved and irregular surfaces, which makes them the material for column wrapping and shear work on shapes a flat plate cannot follow. Preformed shear angles - L-shaped cured plates - go around beam corners where fabric drape or plate stiffness would otherwise be the problem. Rods - pultruded circular sections - serve near-surface-mounted work, bonded into saw-cut slots just below the concrete surface, where they gain cover from the surrounding concrete and change the bond problem entirely. And carbon cords - unidirectional fibre ropes - are used to form connections and anchors, fanned out and bonded to carry a fabric's force past an edge or into a member the fabric cannot reach around. Beyond these bonded forms sits the prestressed family: CFRP tendon systems, tensioned against anchorages and used, in the manufacturer's published description, to replace damaged steel prestressing cables, for seismic retrofitting and for strengthening where deflection and serviceability govern.

Two entries at the edges of the plate table show where the form is going. The referenced range includes a heavier 60mm by 2.6mm section listed with a tensioning force of 220 kN - a plate shipped for prestressed installation, taking the pultruded form into the active-strengthening territory the tendon systems otherwise occupy. And the plate family splits by stiffness as well as strength: alongside the standard-modulus products sit ultra-high-modulus grades, because a bonded plate only attracts load in proportion to its stiffness relative to the substrate - a distinction that matters little on concrete and decides everything on steel, as the applications section explains.

The adhesive is the component the whole system stands on, and its data sheet numbers matter as much as the plate's. The adhesive used with bonded plates is commonly a thixotropic two-component structural epoxy - resin and hardener in a fixed ratio, 3:1 by weight in the referenced product - designed for application between +8 °C and +35 °C. Cured, it reaches a compressive strength around 85 to 90 N/mm2, a tensile strength in the high twenties, and adhesion values that exceed the strength of the materials it joins: pull-off testing on concrete fails in the concrete at more than 4 N/mm2, and on steel the recorded adhesion exceeds 21 N/mm2. The working characteristics are as deliberately engineered as the strengths: the referenced adhesive needs no primer, cures without shrinkage worth the name - 0.04% is the published figure - is unaffected by high humidity while hardening, is impermeable to liquids and water vapour once cured, and pairs a white component with a black one so that an incompletely mixed batch shows itself as streaks instead of failing silently in the joint. Its stiffness - a compressive modulus around 9,600 N/mm2 - is a fraction of the plate's, which is what lets a rigid strip and an irregular concrete surface share load through a forgiving layer. Two published limits belong in any honest description. The glass transition temperature - the softening point the fire section is about - is +52 °C after a 30-day cure at +30 °C. And the manufacturer's own creep note states that, like all polymers under permanent load, the long-term structural design load must generally stay below 20 to 25% of the failure load - the adhesive is strong, but it is strong the way plastics are strong, and the design factors reflect it.

How the composite was made turns out to be a design input, not a manufacturing footnote. CD 371 publishes partial factors on the material that scale with the manufacturing route: 1.05 for filament winding, 1.1 for resin transfer moulding, 1.2 for hand lay-up, 1.5 for hand-held spray application - values the document records as taken from TR55. The logic is variability: a factory-pultruded plate is the same plate every metre, a hand-impregnated fabric depends on the hands, and the design allows for the difference. The same logic runs through the certification layer: structural bonding adhesives fall under BS EN 1504-4, the part of the concrete protection and repair standard series covering the structural bonding of strengthening materials to existing concrete - including plates of steel or fibre composite - and CE-marked adhesives are tested against it. Proprietary systems commonly carry national technical approvals from the markets they sell into, and the data sheets list them; the observed industry habit, as in other bonded disciplines, is that plate, fabric, rod and adhesive are specified as one named system from one supplier, so that the tested combination is the installed combination and responsibility has one address.

Flexural strengthening is the core application and the one the technique was built on: plates or fabric bonded to the tension face of a member so it carries more bending. On a slab that means plates on the soffit, running in the span direction, spaced across the width; on a beam, plates on the bottom face; on a cantilever or a hogging region, plates on the top, buried under the finishes. Bridge decks, car park slabs, office floors and transfer structures all take the same treatment. The design questions are where the plates end - the ends are where bonded plates fail, as the next section explains - and how much of the plate's strength the bond can actually deliver before the concrete lets go. Stacked plates and multi-layer laminates exist for heavier demands; CD 371 permits factory-laminated multi-layer plates only where testing has verified them, and notes the simpler alternative of placing plates side by side where there is room.

Shear strengthening wraps the vertical faces rather than the soffit. Fabric or preformed L-plates are bonded to the sides of a beam - as side plates, as U-wraps carried under the soffit and up both faces, or as closed wraps where the section can be fully encased - acting as external links crossing the diagonal cracks that shear produces. The wrapping scheme matters: a closed wrap anchors itself, a U-wrap anchors better than side plates, and side plates alone depend entirely on bond. Highway guidance limits the spacing of shear strips so a diagonal crack cannot thread between them, and treats separation as a mode to verify even when a beam is fully encased. Columns take the related treatment for a different reason: hoop-wrapped fabric confines the concrete, and confined concrete carries more axial load and behaves less brittly. The geometry rules the gain - confinement works on circular sections, less on rounded rectangles, and CD 371 records that the effectiveness can be zero for rectangular sections - and the published detailing has the hoop fibre placed over any axial fibre, with anchorage into a concrete collar where a strengthened support continues below ground.

Openings in slabs are the everyday building-side case. A new stair, lift, riser or service opening cuts the reinforcement that ran through it; bonded plates around the opening reinstate the tension paths, commonly arranged as trimming bars would be - along the edges, with anchorage lengths past the corners. The attraction is thickness: the strengthening adds millimetres below a soffit that is usually about to be covered by a ceiling anyway, where a steel trimming frame would cost headroom and connections. The same logic serves members that lost section to drilling, chasing or honeycombing, and members whose as-built reinforcement simply proved less than the drawings said.

Near-surface-mounted work moves the composite from the surface into it. Slots are saw-cut into the cover concrete, and pultruded rods - or narrow plate strips on edge - are bonded into the slots with the same family of epoxies. The published trade-offs run in NSM's favour more often than its market share suggests: the bar is bonded on more of its perimeter, the surrounding concrete shields it from knocks, weather and sunlight, and the system tolerates surfaces whose outer skin is too weak for plate bonding - the referenced plate data sheet itself offers NSM slots as the alternative where pull-off strength is below the externally bonded minimum. The costs are the cutting - cover depth has to exist and existing bars must be found and missed - and the labour of the slots. Soffits with thin cover, top surfaces under screeds and members needing discreet strengthening are its common homes, and fib Bulletin 90 covers NSM alongside externally bonded reinforcement as an equal citizen.

The specialist end of the family extends the same materials to harder problems. Seismic strengthening - the largest FRP application worldwide, if not in the UK - wraps columns, beam-column joints and walls to add confinement and ductility, and has its own chapter in the American guide and its own place in the fib work aimed at Eurocode 8. Blast and impact upgrades use fabric wraps and aramid's toughness for the same reason. Prestressed CFRP - tendons stressed against anchorages rather than bonded slack - appears where the existing prestress is the problem: the manufacturer's published applications are the replacement of damaged steel prestressing cables, seismic retrofitting, installation on cracked surfaces through high deflection, and cases where serviceability rather than strength governs. Metallic structures are their own discipline under CIRIA C595 - cast iron beams, riveted girders, fatigue-prone details - commonly using ultra-high-modulus plates, because ordinary-modulus carbon barely participates when bonded to stiff steel; the UK market sells a UHM plate specifically for strengthening metallic structures. Masonry has a dedicated American guide of its own. And timber appears in the plate data sheets' substrate lists, at the quiet end of the market.

Form selection tracks the location more than the load, and the observed pattern is consistent enough to state. Flat soffits with headroom to protect - car park decks, office floors - take pultruded plates, because a plate is the thinnest, flattest, most factory-controlled way to add tension capacity under a ceiling. Columns and anything curved take fabric, because only fabric follows the shape, and a wrapped car park column commonly serves two briefs at once - confinement and vehicle-impact protection - under one render coat. Top surfaces going under screeds, members in reach of the public, and surfaces too weak for plate bonding take near-surface-mounted rods, trading saw-cutting for cover. Beam webs needing shear take fabric or the preformed angles where corners defeat drape. Bridges under possession pressure take whichever form installs fastest without cranage - the no-temporary-support, no-hot-works character of the whole family is worth more on a live carriageway than any property of the laminate.

Sector by sector, the observed pattern in the UK is bridges and highway structures under CD 371, car parks and their ramps, office and residential floors through change of use and openings, industrial floors under new plant, and the repair-driven work that follows corrosion in exposed concrete. The UK supplier's own project navigation groups the recurring destinations - bridges, car parks, the water industry, balconies, and cooling towers and chimneys, the last a reminder that wrapped strengthening scales up to industrial shells as readily as it scales down to a domestic lintel. The technique's footprint is wherever concrete built for one loading now faces another - which is to say, most of the built environment, a few members at a time.

Three ideas organise the design of bonded strengthening, and every published guide circles them. The first is strain compatibility with a head start. The structure was carrying load before the composite arrived, so the concrete and the existing steel hold a locked-in strain that the new material does not share; the plate starts at zero and only accumulates strain from the loads applied after installation. The design therefore tracks two strain histories - the member's and the plate's - and the plate's usable capacity is whatever strain increment the structure can add before something else fails. This is also why the technique naturally strengthens for imposed and future loads rather than existing dead load, and why CD 371's provisions for prestressing the FRP or jacking the structure exist for the cases where the dead load is the problem.

The second idea is that the material never yields. Steel reinforcement announces overload by yielding and redistributing; carbon composite is linear elastic to a strain of about 1.8% and then breaks. A strengthened member's warning behaviour has to come from the existing steel yielding first, from the concrete, or from deliberate limits that keep the composite well inside its strength - which the partial factors and strain caps in the guidance enforce. Related to it is creep rupture: polymers and fibres under high sustained stress can fail over time at loads they carry easily in the short term, so sustained stress in the FRP is limited at the serviceability limit state - CD 371 requires exactly that and points to the criteria in TR55 - and the adhesive's own data sheet states the general polymer rule that long-term design load stays below 20 to 25% of failure load.

The third idea is the collapse safeguard, and it shapes more designs than any equation. CD 371 requires that a structure to be strengthened is itself sound enough that accidental damage to - or removal of - strengthening that is not protected from such damage does not cause collapse. Read plainly: the un-strengthened structure must still stand under a defined load level, because a bonded plate can be lost - to impact, to fire, to vandalism, to a poor bond nobody detected - in a way a cast-in bar cannot. The American guide's chapter structure carries the same concept as strengthening limits. The practical consequence is a ceiling on how much of a member's capacity may ride on the composite, and the honest way to read it is as the technology pricing in its own failure modes.

Those failure modes are what the detailed design actually verifies, and the highway document lists them with useful bluntness. A strengthened section can fail conventionally - concrete crushing, steel yielding then rupture, FRP rupture - or it can fail by losing composite action: the plate stops working with the member because the connection between them lets go. CD 371 names five separation modes to verify: separation induced by surface irregularity; separation induced by shear cracking; separation induced by longitudinal shear stress in the yield zone; separation induced by longitudinal shear stress near the ends of the FRP; and separation due to insufficient anchorage length. The list is a map of where the system is vulnerable. An uneven substrate levers the plate off locally. A shear crack in the concrete steps the two crack faces and prises the plate away - intermediate crack debonding, the mode a substantial research literature and the fib Bulletin 90 provisions concentrate on. High shear flow where the steel yields, and at the plate ends where all the force must transfer in a short length, peels the plate. And a plate that simply ends too soon never develops its force at all. The guidance's blunt instrument against the family is a strain cap - CD 371 has the FRP strain limited to 0.008 to control separation, well below the material's rupture strain - which is the quantitative way of saying the bond, not the fibre, governs.

Anchorage is the discipline that answers the end-zone modes. The force in the plate at its end has to go somewhere, and the design verifies that the anchorage length - the bonded length beyond the point where the plate is needed - can develop it. Two published observations matter here. First, anchorage resistance through bond has a ceiling: beyond a certain length, extra length adds little, and CD 371's recorded remedy for a shortfall is to widen the bonded interface rather than add plate area. Second, where bond alone cannot do it, mechanical anchorages exist - bolted plates, fibre anchors fanned into the member, wraps turned around corners - and the highway document permits them only where testing has demonstrated their effectiveness, reliability and durability, with TR55 discussed as the source of candidate details, and requires bolt holes through FRP to be verified for the stress concentrations they create. Shear and confinement carry their own version of the same logic: hoop strain limits - 0.004 in the highway document's support provisions - and detailing rules such as hoop-over-axial layering and spacing caps on shear strips.

Around these member-level checks sit the system-level ones the documents insist on, and CD 371's general requirements name them as headings: the safeguard against accidental damage, the integrity of the interface with the structure, and over-reinforced sections. Temperature is in the verification set too - the document's reference list carries the Eurocode part for thermal actions, and the adhesive's published coefficient of thermal expansion, 2.5 x 10^-5 per °C, is one of the numbers a bonded joint between dissimilar materials is checked with. Sustained-load applications get a named escalation: where FRP carries permanent load, the document sends the designer to specialist designers and material suppliers for guidance rather than settling the matter itself. Sections that were over-reinforced to begin with gain little from more tension material and are checked for it. Members whose deflection, vibration or cracking governs need stiffness, and a thin plate adds strength far faster than stiffness - the published route to serviceability gains is prestressed CFRP, not more bonded area. Fatigue is treated by deemed-to-satisfy provisions for the FRP and the adhesive in defined circumstances, with the existing structure's own fatigue-sensitive details checked where the strengthened configuration raises their stress ranges - and CD 371 states that it does not provide specific requirements for strengthening structures for fatigue, which routes fatigue-driven metallic work to C595 and specialist literature. Temperature effects, moisture and the difference between laboratory and site cure all fold into the material factors. None of this mathematics is exotic; what is distinctive is where the margin lives. In conventional reinforced concrete the margin is in the steel's ductility. In bonded strengthening it is in the bond checks, the strain caps and the collapse safeguard - which is why the glue line section that follows is not an installation footnote but the other half of the design.

Everything in this trade converges on a layer of epoxy a few millimetres thick and the few millimetres of concrete behind it, because that is the load path. The fibre does not fail on well-run projects; the interface does. The published controls are therefore aimed almost entirely at the interface, and they start before any adhesive is mixed.

The first control is a number: the substrate must be strong enough to be worth bonding to. CD 371 bars externally bonded FRP and steel plates from strengthening concrete structures where the characteristic tensile strength of the surface to be strengthened is below 1.5 MPa, and requires the integrity of the surface concrete - and of its preparation - to be demonstrated by a series of pull-off tests, with the testing requirements set out in the specification. The manufacturer's data sheet for the referenced plate system publishes the same threshold from the product side: a minimum pull-off strength of 1.5 N/mm2 with a mean of 2.0 N/mm2 after surface preparation, verified on the actual surface, on concrete generally older than 28 days. Where the surface cannot meet the number, the published routes are to repair and replace the weak material first, or to change system - near-surface-mounted rods in slots reach sounder concrete below the skin, and fabric systems spread load over larger areas. The pull-off test itself is standardised: BS EN 1542, the test method within the concrete protection and repair series, measures tensile bond by coring through the system, gluing a dolly and pulling it off - the same method later used to prove the installed system.

Preparation is the second control, and the descriptions in system literature are consistent: the concrete surface is cleaned and prepared - commonly by grit blasting or grinding - to remove laitance and contamination and leave an open-textured surface, with the plate side needing minimal preparation because pultrusion leaves a bondable face. Levelness matters because one of the five separation modes is driven by surface irregularity; local high spots are ground and hollows filled with compatible mortars before bonding. Existing defects matter more: cracks, delaminations, honeycombing and corroding reinforcement under a bonded plate are defects sealed in, and observed practice has the concrete repair completed and cured before strengthening starts. The reinforcement scan belongs in this stage too, on both faces of the work - NSM slots must miss existing bars, and drilled anchorages must miss everything.

The third control is the weather, measured, not assumed. The referenced adhesive is applied at product, substrate and ambient temperatures between +8 °C and +35 °C, with the substrate at least 3 °C above the dew point to prevent condensation, and substrate moisture content no higher than 4%. Each number defends a different failure: too cold and the epoxy cures slowly or not fully; too hot and the working time collapses; a surface at the dew point grows an invisible film of water between adhesive and concrete; a damp substrate weakens the bond the same way. These are data sheet requirements, so they are conditions of the system's published performance rather than advice - and on site they translate into thermometers, hygrometers and a log, because the numbers have to be true at the moment of bonding, not at the morning briefing.

Pot life is the clock the whole operation runs on. Once resin and hardener are mixed, the reaction is under way: the referenced adhesive publishes around 120 minutes of pot life at +8 °C, around 90 minutes at +20 °C, and around 20 minutes at +35 °C, with open time a little longer. The quantity mixed shortens it further - a full pail generates its own heat - and the published workarounds are splitting the mix into portions and chilling components before mixing. Mixing itself is specified: the two differently coloured components are combined in their fixed ratio - 3:1 by weight for the referenced product - and mixed for at least three minutes with a slow-speed drill at no more than 300 rpm until the colour is uniform, then transferred to a clean container and mixed again briefly, the colour change existing precisely so that an incomplete mix is visible. Everything after that is placement: adhesive applied to plate and substrate, the plate pressed and rolled until adhesive extrudes from both edges - the visible proof the joint is full - within the layer thickness the data sheet permits, up to 30mm for the referenced adhesive, with its thixotropy holding it without sag on vertical and overhead work at a few millimetres' thickness.

Fabric systems trade the plate's factory quality for site conformability, and the controls shift accordingly. In wet lay-up the fabric is impregnated in the resin on the day, laid onto a primed surface, and worked with rollers to expel air and align fibres; in dry application the resin goes on the substrate and the fabric is pressed into it. The design already prices this difference - the manufacturing-route partial factors in the previous section put hand lay-up at 1.2 against the pultruded plate's 1.05 - and the installation answers with layer counts, overlap lengths and fibre orientations taken from the drawings, because a fabric laid ten degrees off its design direction is a different structural material. Multiple plies and lap splices have their own published detailing, and stacked pultruded plates are permitted in the highway document only with test evidence behind them. Near-surface-mounted work has a parallel discipline: slots cut to specified width and depth, cleaned, part-filled with adhesive, rods pressed in and the slot struck off flush.

The last control is time and protection. Epoxies gain strength over days, not hours, and the referenced data sheet publishes the curve: at +10 °C the adhesive reaches a compressive strength around 55 N/mm2 in a day and around 75 N/mm2 by seven days, while at +35 °C it passes 85 N/mm2 within twelve hours - the same chemistry, run at two speeds. Shear and tensile strengths develop on the same schedule, which is why cold-weather installations keep their protection and access in place longer and why the records tie every bond to the temperature it cured at. The system is protected meanwhile from rain, frost, dust and disturbance. Cured systems are protected permanently from what the data sheet says they cannot live with: the referenced plate system is to be shielded from permanent direct sunlight, moisture and water, commonly by a cementitious or painted overcoat, and its maximum continuous service temperature is about +50 °C. CD 371's list of reasons for coating a cured system adds the human ones - protection against vandalism and accidental removal, and fire protection where fire resistance is required. And because the whole assembly disappears behind coatings and finishes, the record-keeping is part of the installation, not an afterthought: the highway document's handover expectations include installation records, test findings, photographs of critical details, and accurate as-built drawings of every FRP component including those buried by surfacing, precisely because nobody can find a buried plate later without them.

The fire problem in bonded strengthening can be stated in two data sheet numbers. The carbon plate's glass transition temperature is published as above +100 °C. The structural epoxy holding it to the concrete publishes a glass transition temperature of +52 °C - after a warm 30-day cure - with a heat deflection temperature around +53 °C and a maximum service temperature of +45 to +50 °C for the system. The fibre is not the issue; carbon tolerates heat that destroys most building materials around it. The glue is the issue. An ambient-cured epoxy is a polymer that begins to soften at temperatures a dark façade or a hot plant room can approach on its own, and that a fire reaches in its opening minutes. Once the adhesive passes its glass transition, it stops transferring shear, and a plate that carries no shear is decoration.

The published treatment of this follows directly, and it is unusually honest for a construction technology: the strengthening is commonly assumed lost in fire unless it is protected, and the structure is required to survive the fire load case without it. That is the collapse safeguard from the design section doing its second job. The requirement that accidental loss of unprotected strengthening must not cause collapse covers fire loss exactly as it covers impact, and the American guide's strengthening limits express the same idea: the member without its FRP still carries a defined load level, so the fire case is bounded even when the composite is written off. The American Concrete Institute has a dedicated TechNote on the subject - PRC-440.10-21, "Fire Resistance of FRP-Strengthened Concrete Members" - whose published framing records that performance of strengthened members in fire remains a significant concern to design professionals, building officials and owners, that design examples for the fire condition are not widely available, and that the relatively recent emergence of FRP fire protection systems has created challenges for engineers asked to assess structural fire resistance. It is a short document, and its existence is the point: the field considers the question live.

Where the strengthening must itself survive - because the fire load case needs it, or because the client wants the invested capacity protected - the answer is applied fire protection over the cured system. CD 371 lists fire protection among the reasons a strengthening system is coated, alongside protection from solar gain and the temperature rise it causes in the FRP, and protection against vandalism and accidental removal. In buildings the observed forms are the familiar passive fire protection families - board systems and sprayed or trowelled insulating mortars enclosing the strengthened faces - dimensioned by the protection manufacturer's tested data to hold the adhesive below its critical temperature for the required period, which is a far lower temperature target than steel protection works to. The numbers are unforgiving: keeping a bonded joint below roughly 50 to 60 °C through a standard fire exposure takes real thickness, which costs the very headroom and dead weight the composite was chosen to save. That arithmetic is why protected bonded strengthening is commoner in car parks, plant areas and concealed zones than in habitable interiors with demanding fire ratings, and why the alternatives section lists fire-rated interiors among the cases that push projects toward other methods.

The two sides of the industry weight the problem differently, which is worth knowing when reading the documents. Highway structures are open-air; their governing document treats fire as one item on the coating list rather than a chapter, and solar gain - the black plate on a south elevation - is the temperature case its notes dwell on. Buildings invert the emphasis: the fire resistance of floors and frames is a statutory matter under the Building Regulations, so a strengthened building member meets the fire question on every project, and the answer commonly decides between the composite and its alternatives before any strength calculation does. The cured adhesive's full published service envelope - minus 40 to +45 °C for the referenced product when cured warm - covers every UK exposure except the ones people build: boiler rooms, process plant, and fires.

Heat is not only a fire event, and the published exposure limits catch the everyday cases. The referenced plate system's continuous service ceiling of about +50 °C - extendable to about +80 °C only with a specific elevated-temperature adhesive cured by heating - has to be checked against soffits over boiler rooms, roofs under dark waterproofing, external faces in full sun and industrial members near process heat; the solar-gain coating reason in the highway document exists because a black plate in direct sun is its own heating element. Ultraviolet light degrades exposed epoxies and matrices over time, which is the other half of the data sheet's instruction to shield the system from permanent direct sunlight. Moisture and water exposure carry the same instruction. None of these are exotic conditions; they are the ordinary environments of structures, which is why the cured system's protective overcoat - cementitious, painted or board - is described in the documentation as part of the system rather than a finish.

Near-surface-mounted reinforcement changes the fire arithmetic without solving it. Rods set in slots sit inside the concrete cover, gaining the shielding the cover has always given cast-in steel, and the surrounding concrete slows the adhesive's temperature rise substantially compared with a surface plate. The bond is still epoxy and the ceiling is still the glass transition temperature, so the gain is time rather than immunity - but buried reinforcement plus the collapse safeguard is a materially calmer fire story than an exposed soffit plate, and it is one of the reasons NSM appears in fire-conscious briefs. At the other end of the scale, wrapped columns in car parks - a common impact-plus-confinement application - commonly carry their protective render for all three published reasons at once: fire, vehicle strike and sunlight.

What the field does not have is a settled, codified fire design method of the kind steel and concrete enjoy, and this page says so plainly. The ACI TechNote is an overview with an example, not a design code; the UK guidance handles fire through the collapse safeguard and the coating requirement rather than through calculation of protected system performance; and the protection systems themselves are proprietary, with tested performance that lives in manufacturer documentation rather than in the public standards. On most projects the working method is therefore conservative: prove the structure stands in fire without the strengthening, protect the system where its loss is unacceptable or its environment demands it, and treat any claim of fire-resisting bonded strengthening as a claim about a specific tested assembly, to be read from the test report.

A bonded strengthening system cannot be load-tested into confidence and cannot be inspected open like a bolted connection; once installed, it is a black line on a soffit. The industry's answer is a paper trail with testing at both ends, and the UK's dedicated document for it is TR57 - published in 2003 specifically to cover acceptance, inspection and monitoring of fibre composite strengthening, from installation techniques through long-term performance to the elimination of future damage. Its existence as a separate report is a statement in itself: in this trade the proving is a discipline of equal rank with the design.

The testing starts before the system does. The substrate qualification described in the glue line section - pull-off tests demonstrating the prepared surface exceeds the threshold, to the method of BS EN 1542 - is a formal precondition in the highway document, with the testing regime required to be set out in the specification, and it produces the first records in the file: locations, values, failure planes, dates and the preparation method they qualify. On most projects the same campaign fixes the preparation standard - the trial panel or reference area whose result the production surfaces are prepared to match. The materials arrive with their own paper: batch numbers and certificates for plates, fabrics and adhesive components against the specified system, storage conditions on record because epoxies and fibres both carry storage limits, and the system identity checked as a system - the published practice of not mixing one supplier's plate with another's adhesive exists because the tested performance belongs to the combination.

During installation the record is the weather, the clock and the batch. The ambient and substrate temperatures, the dew point margin and the substrate moisture readings at the time of each bond; the mix batches with their times, against pot life at the recorded temperature; which plates went where, tied to batch numbers - the as-built drawings the highway document requires are only writable if this was captured as it happened. Photographs of critical details before they are covered - plate ends, anchorages, laps, slot fills - appear in CD 371's record expectations by name. Adhesive samples taken during the works for cure verification, and companion pull-off dollies bonded to sacrificial areas or spare substrate, are the commonly observed ways of generating test specimens that represent the day's work without wounding the installed system; the American guide's inspection chapters formalise equivalents through witness panels and material tests, and its evaluation and acceptance provisions define what happens when a result fails.

After cure, the installed system is verified two ways. Adhesion testing - pull-off to the same BS EN 1542 method, on the system itself or on companion areas - confirms the bond the design assumed, with coring through the plate into the concrete giving the definitive answer where the specification calls for it. And the search for voids is done by sounding: the installed surface is surveyed by tapping - a bonded area rings solid, a debonded area sounds hollow - with more instrumented methods (thermography, ultrasonics) available where the stakes justify them. The highway document's handover list includes inspection methods for identifying areas of debonding of externally bonded laminates precisely so the operator can repeat the survey in service. Defect dispositions are part of the same paperwork: small voids within published limits are commonly filled by injection through the plate, larger ones mean removal and replacement of the affected length, and the thresholds and remedies belong in the specification rather than in a site-meeting negotiation.

In service, the regime the documents describe is periodic rather than continuous. TR55's own scope runs to long-term inspection and monitoring; TR57 details it; and the highway handover file carries the inspection method forward. The observed practice is visual inspection on the structure's normal cycle - looking for the specific pathologies of the system: edge peeling, blisters, cracked or lost overcoat, impact damage, staining that suggests water behind the plate - with sounding surveys repeated at intervals or after events, and pull-off testing held in reserve for investigation rather than routine. Instrumented monitoring - strain gauges on plates, periodic load-response measurement - appears at the heavy end of the market, on bridges and on structures whose strengthening carries unusual responsibility. The maintenance file's job is continuity: the system's location including every buried component, its design assumptions, its protection, and what was measured when, so that the engineer inspecting in year twenty is checking against year zero rather than against memory.

The people are part of the quality system, and the published documents say so in their structure. The American guide opens its installation chapter with contractor competency before it reaches temperature or tools, and ACI publishes a code-format construction specification for wet lay-up strengthening - a document whose existence means the lay-up process is considered specifiable and enforceable line by line, not a craft to be trusted. In the UK the referenced data sheets carry the same message in plainer words - the products may only be used by experienced professionals, with a qualified structural engineer responsible for the design and trained specialist contractors for the installation - and the observed market markers of that specialism are manufacturer system training and Concrete Repair Association membership. Where the building being strengthened is higher-risk under the Building Safety Act, the paperwork acquires statutory teeth: the work runs through the building control approval machinery the Act built, with the Building Safety Regulator as the building control authority, the dutyholder and competence regime of the amended Building Regulations applying to designers and contractors, and the strengthening records joining the building's information regime rather than a project archive - one more reason the file is written as if someone will read it in thirty years, because on those buildings someone is required to be able to.

Two features of this regime deserve flagging because they differ from adjacent trades. First, the records are the structure, in a stronger sense than usual: a plate under a coating cannot be re-measured, re-identified or re-qualified later, so a missing installation record converts a compliant system into an unverifiable one - which on a regulated structure means treating it as absent. Second, the acceptance tests are destructive where they are most conclusive - a cored pull-off proves the bond by consuming a small piece of it - so the regime leans on statistics and representativeness: enough tests, in the specified places, on the specified method, rather than a single triumphant number. Both features push the same direction as the rest of this page: the discipline is in the proving, and the proving is designed before the first grinder starts.

The published guides introduce bonded FRP by naming what it competes with. The American guide's opening description is the standard list: an alternative to steel plate bonding, section enlargement and external post-tensioning. The UK highway document goes further and puts the competition inside the process - the options report and economic evaluation that compare the proposed technique with other methods before it is adopted. So the honest way to end a reference page on CFRP is the same way the documents begin: with the cases where the answer is something else.

The first family of limits is the substrate. Externally bonded strengthening is barred by the highway document where the surface concrete's characteristic tensile strength falls below 1.5 MPa, and the bar is not bureaucratic - a bond is only as strong as the weaker material at the interface, and low pull-off values mean the concrete itself would be the failure plane. Weak cover concrete, friable surfaces, heavily carbonated or frost-damaged skins and poor-quality repairs all fail this test. The published escape routes - near-surface-mounted rods reaching sounder material, or repair and replacement of the weak zone first - both cost money, and past a point the economics tip to methods that do not depend on the cover concrete at all: enlargement, or new members. Deteriorating substrates fail a subtler version of the test: bonding over concrete with active reinforcement corrosion or ongoing chloride attack seals a progressing defect behind a plate, and the observed sequencing in the repair-led side of the industry is diagnosis and repair first, strengthening after, so the composite is applied to a stabilised structure rather than a deteriorating one.

The second family is structural. Members governed by stiffness rather than strength gain little from thin bonded plates - deflection, vibration and crack-width problems respond to stiffness, and the composite's contribution to a member's stiffness is small even when its contribution to strength is large; the published route to serviceability gains within the technology is prestressed CFRP, and outside it, enlargement or new supports. Members already over-reinforced in tension are checked in the guidance precisely because more tension material moves them toward brittle compression failure. Members whose required gain would breach the collapse safeguard - where the structure minus its strengthening cannot be shown to stand - are outside the technique's permitted envelope no matter how much fibre the soffit could physically carry, and the highway document's sustained-stress and creep-rupture limits cap what the composite may be asked to hold permanently. Fatigue-critical steel details have their own reservation: the UK highway document provides no specific requirements for fatigue strengthening, and the metallic-structures guidance treats the adhesive joint's behaviour, locked-in stress and temperature effects as first-order questions - which routes such work through C595's specialist territory rather than through a standard bonded-plate design.

The third family is environmental, and fire leads it. Where a member needs a fire resistance period and needs its strengthening to survive the fire to achieve it, the protection thickness required to keep an epoxy joint below its glass transition temperature for the full period claws back the depth and weight the composite saved, and the comparison with a steel or concrete scheme - whose protection targets are hundreds of degrees higher - moves against the composite. Hot service environments hit the same ceiling without a fire: continuous temperatures approaching the published service limit of about +50 °C, external faces in full sun without shading coatings, and members near process heat all sit at or beyond the system's comfort. Permanently wet locations and exposed marine environments engage the data sheet's protection requirements and the durability literature rather than a prohibition - but every added protection layer is cost and maintenance the alternative methods may not need.

The fourth family is economic and organisational, and it decides more projects than the technical three. Bonded strengthening is a specialist package - system supplier, trained installer, testing regime, records - and its cost sits in preparation, access and proving rather than in material; where access is cheap and simple, a fabricated steel member with two fixings can undercut the composite before the pull-off testing is even priced. The specialist market is real but thin: the work concentrates in the repair-and-strengthening contractors of the Concrete Repair Association's membership, and a programme that cannot secure one buys risk however good the design is. Warranty follows the same line - the tested system belongs to one supplier, the installation to one specialist, and splitting either splits the responsibility the certification was built on. The option study the highway document mandates is where all of this is weighed, and its published contents read as the discipline's own checklist: the deficiency, the candidate techniques, the interface and safeguard questions, prestress where permanent load governs, and the economics - including the alternative of not strengthening at all.

Set against those limits, the traditional methods keep their places. Steel plate bonding - the technique CFRP largely displaced, and the other half of CD 371's own scope - still appears where stiffness matters more than weight, where a plate must be bolted as well as bonded, or where a client wants a material his maintenance team has known for fifty years; its costs are weight, handling, corrosion protection and joints, which is the list bonded composites were developed to escape. Section enlargement - reinforced concrete jackets and overlays - buys strength and stiffness together, needs no exotic materials and shrugs at fire, at the price of dead weight, lost headroom, formwork, cure time and the wet trades the composite avoids. External post-tensioning adds capacity and closes deflection at once, reaches long spans, and suits prestressed structures whose real problem is lost prestress; it brings anchorages, deviators, protection of the new tendons and a specialist discipline of its own. New structure - additional beams, columns, walls - and the quieter answers, propping, load restriction and change of use, complete the option list, and on some structures the assessment route wins outright: a sharper analysis or a measured material strength proves the capacity was there all along. The comparison section that follows puts the main methods side by side; the point of this one is that the published process expects the comparison to happen, and the strengthening trade's own documents are the ones that say so.

The clearest way to see what bonded carbon composites changed is to line them up against the three methods the published guides name as their predecessors - bonded steel plates, section enlargement, and external post-tensioning - because each comparison isolates a different property.

The dates tell the displacement story on their own. The international index of FRP design documents records the first manufacturer design manuals appearing in a rush across the late 1990s - Tonen's tow sheet manual in 1996, Sika's first engineering guidelines for carbon plate strengthening in 1997, the MBrace system guidelines in 1998 - and the European standards machinery moving in step, with the CEB task group on FRP reinforcement established in December 1996 and its work emerging as fib Bulletin 14 in 2001. The UK guidance matured through the same years: TR55's second edition dates from 2004, the third from 2012 carrying the changes brought by Eurocode 2, and the highway documents ran from BD 84/02 and BD 85/08 - their numbers carrying their years - to consolidation in CD 371 in 2020. Across the Atlantic the guide cited on UK approvals lists moved from its 2008 edition through 2017 to 2023, with a code-format document following in 2024. A technique that had no published guidance in 1995 had, within thirty years, national guidance on three continents and a code - which is roughly the speed at which it took work from the steel plate methods that preceded it.

The cost structures differ more than the price totals, and the comparison is commonly decided there. The composite's material cost is minor and its labour is specialist; the money is in preparation, environmental control, access and the testing regime, so its price moves little with member size and a lot with access difficulty - which is why it wins overhead, over water, over live traffic and inside operating buildings. Steel plate schemes spend on cranage, temporary support, joints and corrosion protection - costs that scale with weight and length. Enlargement spends on formwork, reinforcement, pours, cure time and the knock-on costs of lost space, all of which scale with the member. Post-tensioning spends on anchorage structures and stressing operations largely independent of the span between them. None of these structures is cheaper in general; each is cheaper somewhere, which is what the option study exists to find.

Against steel plate bonding, the comparison is nearly a controlled experiment: same concept, same adhesive family, same substrate rules - the UK highway document covers both in one text and aligns their requirements deliberately, and its 1.5 MPa substrate bar applies to both. What changes is the plate. A carbon strip delivers several times the tensile strength of mild steel plate at a fraction of its weight, arrives in one piece at any length from a roll, and is placed by hand against a soffit - the data sheet's published advantages include installation overhead without temporary support, no joints, and low system thickness with simple crossings where plates intersect. Steel plates are heavy enough to need lifting and propping while the adhesive cures, come in transportable lengths that must be joined, rust from the day the protection fails - the composite does not corrode at all - and add stiffness the carbon strip does not match, which is the one column where steel keeps a clear win. The observed market verdict is recorded in the Concrete Society's own description of the technique: strengthening with fibre composites proved much quicker and more cost-effective than the steel plate methods it replaced.

Against section enlargement - concrete jackets, thickened slabs, encased columns - the comparison is about weight and space. Enlargement adds strength and stiffness together and creates a member that behaves, ages and burns like the structure around it; nothing about it is specialist beyond good concrete work. It also adds permanent load to the very structure being rescued, consumes headroom and clearances by the tens of millimetres, and brings formwork, reinforcement fixing, pours and cure time into an occupied building. The composite inverts every one of those terms - millimetres and kilogrammes instead of centimetres and tonnes, days instead of weeks, dry works instead of wet - and pays for the inversion with the bond dependence, the fire problem and the absence of any stiffness gain worth the name.

Against external post-tensioning, the comparison is about what the structure needs. Post-tensioning is active: it loads the structure in reverse, closing cracks, recovering deflection and relieving the existing steel, which passive bonded plates - which wait for new load before they work - cannot do. It suits long spans and prestressed structures whose problem is lost prestress. It also needs anchorages and deviators fixed into a structure that must be strong enough locally to take them, corrosion protection for the new tendons, and a specialist design and installation discipline. The two technologies meet in the middle in prestressed CFRP - carbon tendons stressed against anchorages, published for exactly the cases where passive bonding is not enough: replacing damaged prestressing cables and satisfying serviceability under deflection.

One comparison runs inside the technology rather than against its rivals, and specifications increasingly turn on it: externally bonded against near-surface-mounted. The same fibre and the same adhesive family deliver a different system when the composite moves into a slot - more bonded perimeter, cover protection against impact, weather, sunlight and casual vandalism, a slower temperature rise in fire, and a tolerance for surfaces whose outer skin fails the externally bonded threshold - against the costs of cutting, the need for cover depth to exist, and the survey discipline of missing every existing bar. Where the published record places both options on the table, the choice is a design decision with the same seriousness as the choice between composite and steel.

The table puts the working comparison in one place, in the terms the published documents use.

The last difference is cultural rather than physical, and it is the one this page has been circling throughout. Steel and concrete strengthening are inspectable trades - a welded plate or a cast jacket can be seen, measured and re-checked for as long as the building stands. A bonded composite is a hidden system whose adequacy was only ever directly observable during the days it was installed. Everything distinctive about the discipline - the substrate thresholds, the weather logs, the pull-off statistics, the strain caps, the collapse safeguard, the as-built drawings of buried plates - exists to make a hidden system trustworthy on paper for fifty years. Where that paper discipline is present, the published record is of a fast, light, quiet technique that solved problems the older methods priced out of reach. Where it is absent, there is no way to tell the difference between a strengthened structure and a decorated one - which is the single sentence this trade would put on its own wall.

BuildPedia would rather tell you where the evidence runs out than round it off. The core design documents of this field are paid publications, so more of this page than usual rests on publisher records, a downloadable highway standard and manufacturer data sheets rather than on the primary design texts. The following were the open points at the time of writing, and each is a place to check the primary source rather than this page.