Durability of concrete and cement composites

Examples of serious structural failure

Serious examples of corrosion-related damage in prestressed concrete structures seldom occur. The following are examples of serious failures which are briefly described and evaluated below.

Damage due to errors in planning and execution

Congress Hall in Berlin/Germany4,45

One of the most spectacular prestressed concrete structure failures occurred in 1980. The southern outer roof of the Berlin Congress Hall collapsed 23 years after its construction. The shape of the outer roof was characterised by an external hollow reinforced concrete arch, clamped at two counter bearings on the east and west sides. The arch was connected to the inner auditorium roof using 7 cm thick plates, separated by joints. Each of the outer roof slabs contained two to four tendons, which were housed in ducts. The tendons contained seven to ten pre- stressing wires made of a traditional type of quenched and tempered prestressing steel. These tendons back-anchored the external reinforced concrete arch to the internal circulating ring beam and held the arch in a stable position.

On 21 May 1980, there was a partial collapse of the outer roof (southern part). The failure investigation concluded that, when the design calculation was performed, the actual loading of the prestressing bands for the back-anchorage of the arches had not been estimated with sufficient accuracy. Cracks had formed in the concrete in a narrow boundary zone along the arch and the ring, as well as inside the joint concrete. The accompanying large curvatures of the concrete plate in the cracked zones led to high bending stresses in the tendons, and, due to the movements of the arches, these stresses alternated. To these design-related problems were added serious faults during construction, aggravating the effects of the design errors:

· Water could penetrate the roof (e.g., by leaking through the roof membrane) and could reach the tendons in the severely cracked boundary zones.
· The joint concrete was very porous and became strongly carbonated with a high chloride content. Chloride was transported in the water and penetrated into the injection grout around the tendons.
· In an area of the outer roof, some tendons were not placed centrally inside the reinforced concrete slabs, but were placed almost without concrete covering on the ring beam (Fig. 6.12). Due to their own weight and the external loads of the roof, these tendons were curved sharply downward. At the same time, the corrosion protection of the bottom surfaces was almost completely lost.

Hence, in the case of these tendons, high corrosion exposure and high bending loads were superimposed. Moreover, some tendons were either not grouted or only partially grouted.
· Sealing plugs made from cork had been used as spacers for the prestressing wires in many of the anchorage bodies for the tendons. These created par- ticularly unfavourable corrosion conditions because the cork absorbed water and so, where the steel and cork were in contact, there was no alkaline protection, leading to hydrogen formation.
The tendons corroded particularly around the connections between roof beams and ring beams, as well as in the vicinity of the cork spacers in the anchorages. The tendons then fractured in these two areas due to hydrogen- induced stress corrosion cracking. The use of old-type quenched and tempered prestressing steel (which was particularly sensitive to hydrogen ingress), and the alternating bending loading of the prestressing reinforcement, caused by inevitable movements of the arch due to changes of temperature and wind, greatly increased the rate and the extent of the structural component damage.

Bridge over Muckbachtal on WuÈrzburg±Heilbronn motorway, Germany5
A bridge was built over the Muckbachtal river as part of the motorway connecting WuÈrzburg and Heilbronn in 1970±71. It was a box girder bridge  consisting of two decks, each prestressed with eight longitudinal tendons running in webs. Each tendon comprised 12 wires of 12 mm diameter made from quenched and tempered steel. The tendon profiles were modified to the bending moment axis.

In the longitudinal direction, the bridge was slightly tub-shaped, with the centre situated near the lowest point. Because of the small longitudinal slope, the drainage points were arranged with a short distance between them (approxi- mately 5m in most cases). Across the bridge, surface water was transported by 100mm diameter pipes (Fig. 6.13). These cast iron pipes had several sleeved couplings along their length. Over time, the sealing of the sleeves became faulty (as the sealing compounds became embrittled) and in many cases, water could drip onto the inclined inner surfaces of the outer webs of the box girders. The surface of the motorway was treated with de-icing salt during the winter months, and for approximately 25 years, chloride-contaminated water reached the concrete surface of the webs and the adjacent bottom slab via the leaking drains every winter. The relatively short wet winter periods were then followed by longer dry periods, so very high concentrations of chloride developed in the concrete coverings of both reinforcing and prestressing steels. The chlorides were able to penetrate deep inside the concrete at these places, and corrosion was exacerbated because low-quality concrete containing porous aggregates had been used in construction. Chloride concentrations of several per cent were traced to a depth of approximately 5 cm below the points where water could get in.

During a bridge inspection, hollow-sounding areas were found in the concrete inside the box girders below the leaking drains. The reinforcing bars located underneath the leaks were rusty and their cross-section had been reduced by up to 50%. The prestressing tendons were also found to be severely damaged. This non-uniform chloride-induced corrosion attacked the sheath and then corroded the prestressing steels, and also led to fractures of the wires. In one case, 12 wires were broken in four tendons exposed on the inside of the web. The other wires were so damaged by corrosion that future fractures were inevitable.

The extent and the intensity of the damage to the reinforcing steel and prestressing tendons depended on their positioning and the thickness of the concrete cover. The normal concrete cover was 5 cm. The most severe chloride corrosion damage to tendons situated inside in the web and fractures of wires were found below a large number of defective drains where the reinforcing steel meshes had a low concrete cover (1 to 2.5 cm). From this it was concluded that, from a construction point of view, the damage was due to having too thin a concrete cover for the reinforcements located in chloride-contaminated regions. The first stage was corrosion-induced detachment, where the concrete peeled away from the severely corroded reinforcing steel, leaving a cavity in which water containing chloride could collect and concentrate. In the second stage, the chlorides initiated the destruction of the tendons.

These failures, which are representative for others of this type, are attributed to serious construction errors. The defective drains, which led to intensive and unexpected chloride contamination of the concrete, resulted from poor design and execution, which were combined with insufficiently expert bridge inspection. The thickness of the concrete cover inside the webs of the box girders (which was only a problem in wet conditions) was identified as a common design fault and eliminated in the mid-1970s when `The Additional Technical Regulations for Structures’ (ZTV-K 76) were introduced by the German Federal Ministry of Transport. Today, surface water drains get special attention when constructing new bridges of this type, and are examined in detail during regular routine bridge inspections. The Muckbachtal bridge was repaired by placing an external tendon that ran through all the sections of the bridge, and by adding tendons in the most severely damaged areas. The chloride- contaminated concrete was removed and re-profiled before making repairs below all the drain points on the inside of the inclined webs and on the top side of the bottom plates.

Failures due to unsuitable construction materials
Concrete, injection grout

Typical examples of the effect on stress corrosion resistance of particular mineral construction materials include the use of high alumina cements containing sulfides (susceptible to damage in or above humid spaces) and hardening accelerators containing CaCl2, especially in the case of (prefabricated) pre- tensioned concrete elements. The use of these materials was forbidden after numerous failures.25 The use of high alumina cement in reinforced concrete was banned in Germany in 1962. Since 1958, only cements with a limited chloride content (0.1 mass-%) may be used in prestressed concrete, and the use of chloride-containing hardening accelerators in reinforced concrete structures has been forbidden.37

The introduction of concrete additives as part of modern concrete technology caused corrosion problems because active substances and other materials, which may modify corrosion resistance even in very small concentrations, could get into the concrete. A particularly significant example is the use of thiocyanate as an active substance in a plasticiser when making concrete for reinforced struc- tures. Such plasticisers mistakenly got into fresh concrete used for manufactur- ing prestressed concrete elements and caused damage. Thiocyanate has been forbidden as an active substance in concrete plasticisers, because it promotes hydrogen absorption by the prestressing steel.

Railway overpass in Berghausen (Karlsruhe/Germany)3,25 In 1979, numerous fractures occurred in the prestressing steel at the lowest points in the central area of a railway overpass deck with a beam cross-section. These fractures occurred between a few hours after and several days after prestressing, but before grouting. The prestressing steel consisted of a single 36 mm diameter prestressing steel bar with pearlitic structure (strength 1230 N/mm2). Water (approximately 250 ml to 1 litre) was found in the ducts where the prestressing steel had fractured in each sheath.

Investigations revealed initial rusting of the prestressing steels, resulting in shallow corrosion pits up to 0.5mm deep that induced the fractures. The pro- nounced uneven local corrosion occurred within 2 to 3 weeks after concreting, and was due to penetration of water into the prestressing ducts. Salt-enriched water of neutral pH is extraordinarily corrosive locally because the sulfates and chlorides concentrate due to leaching of the fresh concentrate as it hardens. The main constituents of this corrosive medium are the potassium salts K2SO4 and KCl (Table 6.3). In this case, up to 5 g sulfate (SO4) per litre and 0.25 g chloride (Cl) per litre was measured in the water around the tendons. In addition, up to 0.5 g/l of thiocyanate reached the prestressing steel in the ducts via the water from leaching of the concrete. As has been noted, thiocyanate, acts as a promoter for hydrogen absorption, causing pitting-induced stress corrosion cracking, which leads to premature fracture failures. The thiocyanate had originally entered the concrete as a constituent of the plasticiser, which was itself not certified for use with prestressed concrete. This plasticiser contained 4 mass-% SCN± as the effective substance. In the hardened concrete, however, only traces of thiocyanate were found (0.006 mass-%).

Thiocyanate is known to work as a promoter that favours the absorption of hydrogen by prestressing steel. After considering the damage found and the results of subsequent laboratory tests, the use of thiocyanate as an active substance for concrete plasticisers was forbidden. In order to prevent corrosion- promoting substances being transmitted to the concrete when using concrete additives or bonding agents, it is now mandatory to check using a special electro-chemical method36 before any new materials may be used in Germany.

The localised pitting corrosion attacks were primarily attributed to salt- containing water from the concrete. Current recommendations are to remove settling water a short time after concreting by blowing out and, if possible, by additional washing out. Nowadays the harmful effects of concrete settling water are also reduced naturally by adopting the limited periods between concreting and injecting set out in the German standard DIN 1045.

Prestressing steel

Damage caused by stress-corrosion cracking in prestressing steel presupposes the use of steel with sufficient susceptibility to corrosion-induced failure under particular conditions. Appropriate environmental conditions are also necessary, either to favour active corrosion or to interrupt or terminate passivation. Substantial problems have occurred with certain grades of steel where no serious errors appear to have been committed either in the planning or building con- struction stages. In these cases, the rules concerning corrosion protection were obeyed in all important points during transportation, storage and processing of the steel, and no unsuitable building materials were used. Following detailed investigations, these particular grades of steel were classified as highly sensitive to hydrogen and unsuitable for normal building works, and have been prohibited for use in prestressing applications in Germany. The steel characteristics have been modified to try and develop practical grades of steel resistant to hydrogen attack.

Structures manufactured using bainitic hot-rolled prestressing steel

3 Using ferritic-pearlitic steel grade St 55/85 as an example (with tensile strength measured in kp/mm2 units), the strength of hot-rolled prestressing steels has gradually been increased by alloying and processing measures. From early 1974, a new prestressing steel grade St 110/135, a low-alloy hot-rolled steel with bainitic structure, was used for single rods in prestressed concrete. Within a few years of use, several structures using this grade of prestressing steel failed. This was very surprising because investigation of notched-bar tensile properties and fracture-mechanics suggested that this grade of prestressing steel possessed improved (less brittle) failure characteristics in comparison to the previously used prestressing steel St 85/105.37 However, failure in practice demonstrated that behaviour determined during brittle failure investigations in the laboratory cannot be extrapolated to actual applications in every case. Under construction site conditions, the low alloy silicon-manganese-chromium steel St 110/135 proved to be substantially more susceptible to hydrogen-induced stress corrosion cracking than the hot-rolled prestressing pearlitic steel St 85/105, manufactured out of silicon-manganese-carbon steel.

In the numerous structures made with steel grade St 110/135, corrosion- induced fractures occurred during the construction stages, either before grouting the tendons or a few hours afterwards. In total, 65 fractures were found in 27 separate projects on wires with diameters of 26, 32 and 36 mm in the four years this steel was used (1974 to 1978). These projects ranged from bridges to building components and water tanks. With some fractures, internal steel defects (cavities, core segregations) and other surface defects (hardening cracks, laminations) were also present. Fractures occurred from 0.7 hours to 1200 days after prestressing. In the case of the later fractures, the tendons had not been grouted. Rods which had broken within a few days of prestressing and before grouting were generally free from any corrosion visible to the naked eye. In most cases, there were no unusual quantities of aggressive constituents at the fracture initiation sites or on the steel surfaces. The broken prestressing bars were, however, often wet and in the presence of concrete bleed water.

The actual tensile strength of the broken steel lay between 1350 and 1880 N/mm2. In many cases, the strength was high within the grade and sometimes well above the prescribed tensile strength. On average, the higher the strength, the sooner the steel cracked. Therefore, increasing the rods’ tensile strength also increased the tendency to hydrogen-induced stress corrosion cracking. Corrosion-induced damage in prestressed concrete structures made using the prestressing steel grade St 110/135, which occurred both during the construction phase and on older constructions, was attributed to the high structure-conditioned suscepti- bility of this steel to stress corrosion under usual construction site conditions. It was found to be difficult to create uniform strength levels under alloying and production conditions, and so many over-strength casts, with an unusually high tendency to crack and fracture, came on the market. It was not possible to separate out these over-strength casts using normal, self-regulated controls. The corrosion problems associated with stronger types of prestressing steel could have had an adverse effect on the general reputation of all prestressed concrete structures, and so the further use of this steel grade for prestressed structures was forbidden.

Structures manufactured using quenched and tempered prestressing steel of older type (Laboratory building in Mannheim/Germany

Quenched and tempered prestressing steels were used in the form of round wires (smooth or ribbed) or ribbed oval wires from the early 1950s, especially in Germany. The ribbed oval wires were prestressed in innumerable pretensioned precast concrete members of many different types to provide large numbers of post-tensioned prestressed bridges and high-rise structures. The development of quenched and tempered steels was driven by a combination of economics and the characteristics of the steel, including high strength with good toughness, very good bond characteristics, and low relaxation.

The behaviour of the early types of carbon-alloyed quenched and tempered steel was not always satisfactory with respect to stress corrosion during processing, and fractures occurred during or shortly after prestressing. Signs of inadequate stability under unfavourable site conditions3 and susceptibility to some testing laboratory testing solutions, led to the development of steel with reduced carbon content and about 0.5 mass-% chromium, giving increased resistance to hydrogen-induced stress corrosion cracking.46 However, numerous structures still exist which were constructed using the old type of steel (manufactured until 1965). In addition, old- style quenched and tempered prestressing steel continued to be used until 1990, particularly in pre-tensioned members, in the former East Germany. From 1989 onwards, examples of collapse and near-collapse have occurred with structural members made using this steel type, even where evidently unfavourable environmental conditions were not present:

· After 28 years of service, beams in a laboratory building collapsed, breaking almost in the centre (Fig. 6.14).
· After 30 years of service, an open crack in the concrete of a beam in a factory was found ± again at the mid-span.
· During the demolition of the 25-year-old roof of another factory, the beams broke while being dismantled.
· After 33 years of service, numerous pre-cracks and fractures of wires were found on a bridge from Henningsdorf prestressed with quenched and tempered steel.

These events were primarily caused by accumulated fractures of the wires in the tendons, and the results of the investigations were practically identical in each case. As an example, the first failure mentioned above will be examined in more detail. The two tendons in the beams, made of 16 ribbed oval wires, were covered with a sufficiently thick and dense (impermeable) concrete cover. The ducts were free from any external corrosion, since their concrete cover was neither carbonated nor contained chloride. After opening the tendons, older fractures were found covered with rust and/or grouting mortar. These fractures probably originated from when the beam was manufactured. The fractures were randomly distributed over several metres at the lowest point of the tendons. There were also well-preserved recent fractures in the ruptured cross-section of the beam. The tendons were correctly grouted. The injection mortar was damp, although it contained neither carbonated material nor chloride. In the case of the broken wires, there was no corrosion visible to the naked eye, but small corrosion pits and outgoing cracks could be seen under the microscope. A large number of cracks had occurred but were not related to any heavily pitted eroded areas. The steel ducts showed initial rusting on the inside and lime-like deposits with some concentration of potassium sulfate along a former water level. This old type of quenched and tempered prestressing steel had a tensile strength of 1750 N/mm2; i.e. the actual strength was clearly above the required nominal minimum of 1570 N/mm2, which heightened its sensitivity to hydrogen.

It was concluded that, during the production of the concrete beams, either concrete setting water penetrated into the ungrouted tendons and caused the wire to pre-crack and fracture, or condensation water had already damaged the highly sensitive steel in a similar way. It was shown that the steel grades used were highly sensitive to hydrogen-induced cracking when in contact with condensa- tion water.23 While embedded in a permanently wet injection mortar, the existing pre-cracks in the steel could grow and eventually lead to the final failure. A large number of wire fractures in a cross-section finally caused the fracture of the beam. In lab tests,8 it was shown that wires with pre-cracked sensitive prestressing steel can fracture when placed in an alkaline environment (e.g. Table 6.5). In a stress corrosion test using a saturated solution of Ca(OH)2, retarded fractures of steel occurred after 330 hours, with the stress level set at 40% of the tensile strength. Under crevice corrosion conditions, usually found in the case of bundles, the service-life of pre-cracked samples was shortened to 40 hours. Samples without pre-cracks were still not broken after 5000 hours.

Studies of serious damage that occurred after several decades of service in structures prestressed with the old quenched and tempered steel showed that the unusually high susceptibility to hydrogen was primarily or exclusively respon- sible for the delayed failures. This type of steel has not been in production for 40 years. Damage analysis and the results of research demonstrate that, before grouting, corrosion pitting, pre-cracks and fractures had already occurred. Due to the high stress-corrosion susceptibility of these types of steel, the damage could develop up to complete failure of the members even in correctly grouted sheaths and without any other environmental conditions that favoured cracking. The early damage in the form of pre-cracks and single fractures is explained by the instability of these wires in the moist climate of the ungrouted ducts,23 or on contact with concrete setting water.7 Both these circumstances are part of everyday practice at construction sites. Because of these correlations and because structures post-tensioned with the sensitive steel are still in use, investigations aimed at controlling the potential risks in existing structures were instigated. Non-destructive magnetic leakage flux measurements and destructive checks with single wire sampling are presently in use.


Related Articles