Located in the car park, this outdoor structure comprises 22 ‘sidewise-seven’-like 2.5x5.55x0.17 m (height x length x thickness) supports spanned by wooden shading slats. Both concrete and slats (the latter not conserved) were initially painted white (Figure 2). The structural review was based on the original drawings and information in the general design specifications on the type of concrete and reinforcement used to build the supports (Table 2).

Surface spalling on the concrete members was patch-repaired in 2007 when the white paint, biological colonies, rust from the exposed reinforcement and poorly bonded grout were sand-blasted off the supports. An epoxy resin passive seal (Legaran by Degussa) was subsequently applied to protect the exposed steel rebar and ensure the bond (with hydraulic binders and resins (Degussa PCC-20)) to the patch mortar used to restore the detached cover. All the surfaces were then painted with Masterseal 325E, a white anti-carbonation coating to detain reinforcement corrosion.

The present visual inspection (Figure 3) revealed concrete cracking at the top and especially at the tip (Figure 3 (b) ) of the supports, material loss, biological colonisation and corrosion in the exposed rebar.

The pergola concrete was characterised to identify the possible origin of the damage and any necessary repairs to choose compatible materials or treatments.

The most prominent physical properties defined in the original design (40, 57) and the ones determined from the core samples in this study are given in Table 2, along with other provisions on concrete set out in the Eurocode (56) for new structures exposed to similar environmental conditions.

Aggregate size distribution was determined by analysing the macroscopic (with no or low enlargement) and (optical and electronic) microscopic images of the core samples (Figure 4), which also revealed their composition, found to be: in the gravel (d/D=2.5/28), carbonate (micritic texture) and polycrystalline quartz grains (with and without mica and iron oxides); in the sand (d/D=0.05/2.5), primarily monocrystalline quartz, and secondarily metamorphic rock fragments (schist), potassium feldspar (microcline) and plagioclase (albite). Both size and composition were largely in agreement with the design specifications (57), which called for 800 L/m³ to 900 L/m³ of natural or crushed gravel with up to 3 % clay and 400 L/m³ to 500 L/m³ of rinsed siliceous sand with a maximum size of 38 mm and no more than 10 % clay of a maximum size of 5 mm (50 % coarse grain from 2 mm to 5 mm and up to 15 % medium grain, from 0.5 mm to 2 mm).

The mixed (siliceous and calcareous) composition of the aggregates ruled out the use of both the standard procedure for determining cement content (46) based on the HCl-solubility of silica at 5 °C and the calcium oxide sub-procedure. The design specified a cement content of 300 kg/m³ to 350 kg/m³.

The XRD pattern for the high cement content paste contained reflections for the two primary hydrated phases, ettringite and portlandite, along with signals for gypsum, perhaps attributable to the deposition of atmospheric SO₂ (59). The most intense lines identified the quartz, calcite, feldspars, albite and microcline present in the aggregates (Figure 5a). The signals for calcite might also have been generated as a result of cement paste carbonation. Further to the DTA/TG study, this hydrated cement-high sample contained 22 wt% calcite and 5 wt% portlandite, denoting the persistence of a store of alkalinity available for reinforcement passivation (60).

The FT-IR spectrum, in turn, exhibited the bands characteristic of the aforementioned mineral phases (Figure 5b): calcite at 1425, 875 and 712 cm^-1; silicate vibrations at around 1000 cm^-1; the double band characteristic of quartz at 796 cm^-1 and 777 cm^-1; the bending vibrations typical of the S-O in sulfates at 668 cm^-1; and the stretching vibrations generated by the OH groups in portlandite at 3640 cm^-1 (61).

The carbonation front in the sample core was not very deep, despite the high levels of air pollution to which the material has been exposed over the years: mean CO₂ penetration was 8.7 mm, with a maximum of 12 mm (Figure 6, Table 2).

The concrete’s water-accessible porosity was a moderate 9.8 % (Table 2) while its density was 2.3 g/cm³ which, like its Young’s modulus, lay within the range defined in Eurocode BS EN 1992-1-1 (56) At 30 MPa its compressive strength, far higher than the design value, was likewise as recommended in (56) for concrete in standard structures.

Core specimen electrical resistivity, found (prior to mechanical characterisation) to assess the risk of steel corrosion, was 8.45 kΩ·cm under saturated and 145.48 kΩ·cm under dry conditions. Further to the criterion proposed by Polder et al. (53) for saturated conditions, here the reinforcement would be at high risk of corrosion, given its low resistivity, <10 kΩ·cm. The severity of that risk is relativised, however, by the structure’s location in an area where neither humidity nor the presence of chlorides is high. For fuller information in that regard, in situ measurements were made of electrical resistivity, corrosion potential and corrosion rate (E_corr, mV; I_corr, µA/cm²) (see item 3.1.3).

By way of summary, according to the laboratory tests conducted the concrete was in good condition further to its age and location.

Laser scans were captured of the 22 units comprising the structure (Figure 7 (a)) to compare their present 3D geometries to the original design drawings. The results for rib No. 8 reproduced in Figure 7 (b) (ribs numbered consecutively beginning with the one closest to the building) showed that the built geometry was nearly identical to the design, with minor deviations possibly due to stake-out errors or to the time lapsing since construction. The 8 cm to 15 cm slump at the edge of the overhang relative to the design and a curved deformation on its upper side together determined a 0.8 cm to 2 cm difference in mid-span height. Although the value of the slump appearing initially is unknown, the curved deformation along the top of the element might hold a clue to the deflection incurred since it was built (bearing in mind that the formwork used was flat). That relative deformation is in keeping with today’s legislative provisions on the construction of cantilevered structural floors in buildings (62). The rest of the slump may be due to rotation in the restrained end, as the overlain geometries (Figure 7 (b)) would appear to indicate, and/or related to flawed workmanship or levelling during construction. Moreover, the pergola has to bear only its own weight and no pathologies were observed that would entail structural risk. The deformation nonetheless generates tensile stress in the top surface of the overhang that favours the appearance of small or the enlargement of existing cracks on that surface, the one most exposed to rainwater. It should consequently be protected to prevent subsequent pathologies.

The ground-penetrating radar (GPR) findings showed that the concrete cover, at <30 mm (Table 2), fails to meet both the design specifications, which call for a cover thickness of >30 mm, and the 35 mm minimum presently laid down in the Eurocode (56) for class 4 structures (buildings and other common structures) with XC4 exposure (outdoor structures).

The electrical resistivity and corrosion potential measurements were below the threshold values defined for steel de-passivation, indicating negligible corrosion and high levels of steel corrosion resistance at this time (Figure 8). All the corrosion rate readings taken in the on-site inspection were likewise very low, corroborating the results observed for the other parameters (Figure 8). To determine representative values of corrosion it will be necessary to monitor the structure, determining corrosion parameters variation with environmental parameters and concrete saturation degree. So these measurements are informing us about the actual corrosion progress but we have no information about what happened before.

In other words, even where the laboratory-measured saturated resistivity values denoted possible corrosion due to low concrete cover resistivity, such problems are unlikely given the absence of chlorides and scant pore water content.

The moisture meter delivered the same readings (30 % to 44 %) for the horizontal and vertical arms of the car park reinforced concrete units. The first and fourth units, the ones closest to the building (Figure 7) that consequently receive less solar radiation, exhibited around 40 % to 43 % moisture content, compared to the 31 % observed in the more distant, less shaded units.

The approximate uniformity of the Schmidt hammer test results at around 52±2 (dimensionless units) for the pergola supports measured was an indication that the surfaces are in good condition. At 60 MPa, the compressive strength extrapolated from the calibration curves furnished by the hammer manufacturer based on the rebound readings was double the value determined on the core samples (Table 2) and 3.5-fold the design specification. According to earlier studies, in most cases extrapolation is unreliable because the rebound measurements vary depending on factors such as surface roughness, moisture content, carbonation, porosity and measuring facility calibration (63, 64).

The reinforced concrete rib-like shell (Figure 9), designed to serve as an outdoor chapel, was built nearly two decades later than the main building, in 1969. Its constructional details are described in Cassinello et al. (37). The 6.5 m high, 10 m long shell is characterised by a complex geometry defined by a Bernoulli lemniscate in which the 40 cm thick base wanes to 6 cm at the outer edge. It is reinforced with steel bars ranging in diameter from 12 mm to 20 mm. When the structure was stripped of the formwork 3 d after casting, the quick set concrete used in its construction exhibited compressive strength of 19 MPa whilst the tip was positioned 12 cm below the design height.

Unlike the pergola, this structure was not initially painted. The form-determined inner surface texture was not modified, whereas the outer surface was treated with a pneumatic bush-hammer (37). No conservation operations have been undertaken since it was built. In an earlier study Echevarría et al. (50) detected local rusting but no surface cracking, crumbling or weathering nor any other symptom that might denote anomalous structural behaviour.

The damage map drawn on the occasion of this analysis shows corrosion in the areas where the reinforcement is exposed, along with biological colonisation, on the top surface of the shell in particular (Figure 10).

The methodology to determine the key characteristics of the rib-like shell concrete was the same as used for the pergola material.

Two core samples were taken from the rib-like shell, one from the slab and the other from the lower part of the rib (Figure 11), for subsequent mineralogical and physical characterisation. Analysis of the optical micrographs revealed that the mineralogical composition of the gravel differs from design specifications (Table 3), for it comprises not only polycrystalline quartz (with and without muscovite and iron oxides) but biomicrite and sparry calcite. In contrast, the primarily quartz-grain based siliceous composition of the sand, with some K-feldspar and plagioclase, is essentially the same as initially designed. Further to their respective particle size distributions, both aggregates are well-graded, with d/D=2/26.7 in the coarse material and d/D=0.02/4 in the fines.

The most intense reflections observed in the XRD pattern for the high cement paste content fraction were generated by the quartz and calcite aggregates (which had not separated from the paste altogether), although the line attributed to calcite might also be indicative of cement paste carbonation. XRD also detected the presence of gypsum and portlandite, with the latter, according to the DTA-TG findings, accounting for 5.3 wt% and calcite for 20.8 wt% of the total. Those results were corroborated by FT-IR analysis, for the bands characteristic of those minerals were visible on the spectrum.

Compressive strength at 33.3 MPa and Young’s modulus at 41.1 GPa were similar to the values expected of healthy concrete and similar as well to the findings observed for the pergola. Corrosion resistance, at dry (447.29 kΩ·cm) and saturated (20.07 kΩ·cm) electrical resistivity values, was higher in the shell than in the pergola and the carbonation front shallower (mean 2.6 mm, maximum 4.1 mm; Figure 12), due largely to its lower porosity (6 %).

In earlier inspection, Echevarría et al. (50) analysed the structural behaviour of the shell with a 3D laser scan, comparing its present condition to the design specifications. Despite the substantial 373 mm drop at the tip of the overhang, their analysis concluded that the shell complied with existing Spanish legislation on structural safety (65).

Although a cover of up to 40 mm was specified for the slab (Table 3), according to the GPR findings (Figure 13) the concrete protecting the reinforcement from potential environmentally-induced corrosion is <30 mm. In some sporadic instances there is no cover whatsoever (Figure 10), due to the difficulties involved in laying the forms and positioning the reinforcement described by Cassinello et al. (37). Those problems were exacerbated by design specifications calling for a cover of just 10 mm at the unrestrained edge of the structure (Table 3).

The electrical resistivity and corrosion potential findings revealed that the risk of active corrosion in the slab concrete is low at this time (Figure 14), for like in the pergola, the threshold value indicative of steel de-passivation has not been reached.

The dimensionless Schmidt hammer test measurements, at 56±4, denote concrete uniformity, whilst the sole differences in shell moisture content are orientation-dependent, with higher values in the north-facing (39 %) than in the west-facing (26 %) areas.

The building’s 400 low-strength, 1.60 m high window frames were cast at an on-site workshop using mortar with a sand:(white) cement ratio of 1:2. Bearing capacity is provided by the likewise 1.60 m long lintels and the vertical supports (concrete columns). The reinforced concrete columns were cast between the windows into forms comprising the vertical members in the two adjacent window frames as the sides, the indoor building façade at the rear and temporary corrugated sheet steel panel across the front. The latter stripped served as ornamentation in this inter-window space (39) (Figure 15).

The frames have been repainted on a number of occasions predating this analysis.

The present damage map shows both corrosion issues and some cracking. The most prominent problem observed was clogged drainage and the resulting ponding of water in the top member. That in turn has often generated further damage in the form of mortar corrosion and crumbling due to cementitious matrix dissolution. Part of one of the corners on the lintel in poorest condition was observed to have detached completely in response to corrosion (Figure 16).

A 29 cm long, 0.87 kg sample was cored from an area in the lintel on one of the windows on the south side of the building where top reinforcement corrosion had induced severe spalling (Table 4).

The petrographic and SEM/EDX (Figure 17) analyses conducted on the sample to characterise the mortar showed that the sand (d/D=0.1/4) comprised well-graded, sub-angular monocrystalline (together with some polycrystalline) quartz grains ranging in size from 100 µm to 4 mm and smaller (250 µm to 3 mm) angular potassium feldspar particles (Table 4). A minor fraction (around 10 %) of 0.5 mm to 1.5 mm plagioclase was also present. Briefly, the design specified the same mineralogy for the mortar sand as for the concrete fines (57).

In addition to the intense quartz and feldspar reflections, the XRD pattern (Figure 18) exhibited signals attributed to mica, hydrated cement phases ettringite and portlandite and cement paste carbonation-induced calcite, all confirmed by FT-IR analysis (Figure 18).

DTA/TG-quantified portlandite content came to 3.7 wt% and calcite content to 10 wt% to 12 wt%. The broad, asymmetrical DTA signal for CaCO₃ decarbonation denoted the low crystallinity of that compound.

A comparison of HCl dissolution-based solubility to the TG findings for the sample (46) yielded a cement:sand ratio of 1:1.7, only slightly lower than the design specification.

Despite the well-graded quality of the sand, the mortar is fairly porous, exhibiting mean porosity of 13.7 % (Figure 19), with pores <1 µm accounting for 9 %. Those data, found with mercury intrusion porosimetric techniques, are not comparable to the water-accessible porosity findings described above, however, for the two parameters measure different size pores, the former in the 3 nm to 350 µm range and the latter 50 nm pores only (66).

Nonetheless, at 41.1 GPa the Young’s modulus value denotes high mortar quality, although compressive strength could not be determined due to the irregular shape of the sample.

The carbonation depth observed is shallow (mean=1.7 mm), with CO₂ penetration reaching up to 16 mm in only one area, affected by cracking (Table 4).

The mean thickness of the concrete covering the reinforcement in the precast window frames was also determined. The minimum design thickness was 6 mm (Figure 15), whilst in situ GPR delivered a mean of 15 mm (in the south-facing window the readings were more scattered than in the north, with a maximum thickness of 40 mm and minimum of 0 mm).

Similar moisture content values (32 % to 40 %) were found for the north and south façade vertical framing, whereas content was greater in the south- than in the north-facing lintels. The explanation may lie in the larger size of the south frame lintels, splashed by the rainwater gushing out of the rooftop gargoyles (Figure 16).

The Schmidt hammer rebound values were essentially identical for the north and south frames, with means of 56 and 57, respectively.