Concrete of tomorrow: corrosion performances in marine environment

. Concrete is the most used construction material worldwide, but if often faces durability issues due to rebar corrosion, and cement production can be energy consuming. Attempts to improve its resilience versus corrosion while reducing its carbon footprint are ongoing at an international level. In this objective, in France within the framework of the national project PERFDUB, a series of reinforced concrete walls were cast with innovative concrete designs, two shapes of rebars and two concrete covers. The standard performances of the different concrete mix designs were evaluated at a laboratory scale. Then the walls were exposed in the French La Rochelle harbor, in the tidal zone in order to monitor the evolution of the corrosion of the bars for a period of twenty years. The first results of the electrochemical follow-up of these reinforced walls are presented in this paper.


Introduction
Within the framework of the French national project PERDUB [1], an intensive series of studies mixing lab scale, natural ageing and durability modelling was launched in 2015. The aims of the project are multiple. One of these is to provide innovative concrete mix designs to improve durability performances and lower carbon footprint. Another one is to characterize the performances of these concretes, especially against carbonation and chloride ingress. For the latest, a vast database of performances is being produced at the lab scale in combination with life cycle modelling. A last part of the project is dedicated to the evaluation of the performances of 11 of the concrete mixes designed against corrosion induced by natural carbonation or exposure to a marine environment, for a period of 20 years. To follow up the initiation and propagation phases of corrosion, an approach combining non-destructive testing, embedded sensors and corrosion modelling was adopted. In the present paper, the experimental program and the first nondestructive electrochemical tests performed on 11 walls exposed for a little more than a year to natural ageing in a marine environment, in the harbor of La Rochelle (French Atlantic ocean shore) in the tidal zone are presented.

Experimental program 2.1 Specimen
A series of 11 T-shaped 0.16cm large walls were cast, with two 1m² faces (1mx1m) and a foot (0.4x0.15m) to ensure their stability against the tide (Fig .1a). Two types of steel bars were embedded: smooth round shaped (RL) representative of ancient constructions (and for which surface area can be determined accurately for electrochemical calculations) and high-bond ribbed (HA) bars, representative of modern ones. These rebars are isolated which means not in electrical contact with the others in the specimen. Two concrete covers were considered: 10mm and 20mm, to evaluate the progression of aggressive agents. Additional bars were added on the sides and foot of the walls for stability issues (Fig. 1b). Eleven concrete mix designs with different binders were selected, from standard CEM I to binders containing mineral additions such as fly ashes (V), limestone filler (L), slag (S), silica fume (FS), siliceous addition (AS), or metakaolin (MK). Due to these additions, the content of clinker was quite diverse with values ranging between 101 and 346 kg/m 3 ( Table 2). Different types of aggregates were chosen: two types of alluvial silica-limestones, one soft limestone and one hard limestone.
For each concrete mix design, water to binder ratios varying from 0.35 to 0.6 were considered ( Table 2).
Compressive strengths measured on 3 cylinders of the same concrete after 28 days of wet curing (stored in water) ranged between 29 and 103 MPa, while open porosities measured after 90 days of wet curing varied between 10 and 20% (Table 3).   [4]. Nevertheless, those monitoring techniques have rarely been installed on real structures in natural ageing conditions [5] and still need to be validated. Therefore, the two types of technologies were considered and compared: embedded sensors and non-destructive testing.
Concerning the embedded sensors, for economic reasons, only three concretes B01, B04 and B31 were instrumented, and three additional walls were specifically cast. On those three walls, a complementary rebar network was introduced for the non-destructive measurements, and a series of sensors was embedded to monitor versus time: resistivity (Multi Ring Electrodes ® ); potential (ERE 20 ® ); resistance, potential and corrosion current (Anode Ladder ® ). In a dedicated reservation, additional temperature and relative humidity sensors (HMP110 ® ) were introduced from outside. Finally, all the sensors were connected to a multiplexer, itself connected to an analyzer, with wiring linked to a computer to provide data gathered in an automatic database [6]. Most of the sensors can deliver data at different depths allowing the gradual ingress of aggressive agents into the concrete to be followed. For the non-destructive testing, two types of techniques were considered: electrochemical tools [7] and evanescent field dielectrometry [8] [9]. Resistivity was measured with a four-point Wenner probe (Resipod ® ) [10] and with the disc method (Gecor10 ® ) [11], potential mapping was performed with a copper/copper sulphate electrode (Canin ® ) [12] and corrosion rate was evaluated using a Gecor10 ® [13] Only the first series of non-destructive electrochemical measurements on the eleven noninstrumented walls are presented in this paper.

Experimental protocol
The reinforced concrete walls were cast from December 2019 to January 2020. After almost one year of exposure to natural ageing in a semi-rural environment, 60km south-west from Paris, they were implanted in December 2020 in marine environment in the harbor of La Rochelle.
The specimens are exposed to natural ageing in the tidal zone (Fig.3a), which means around 6 hours immersed and 6 hours in the open air for every tide. Consequently, the measurements can only be performed at low tide (Fig 3b) which means that to achieve the measurements on the eleven walls without embedded sensors (B01, B02, B04, B05, B07, B31, B36, B37, B38, BM40, B41), 10 days are necessary. The first set of experiments took place in October 2021.
The following protocol was adopted for the electrochemical measurements for each rebar (Fig.4) of each side of the walls (Fig. 5): -Potential mapping [12], with 3 vertical lines spaced 5cm apart, with a measurement every 5cm, -Resistivity measurements on point 8 (40cm from the top of the wall) and point 14 (70cm from the top of the wall), first with a 4-point (5cm spacing electrodes) Wenner probe (4 measurements per point according to the RILEM recommendation) [10] and second with the disc method (5 measurements per point 5cm spaced left and right from the rebar) [11], -And one potential and LPR measurement on point 8 [13].

Potential measurements
First, potential mapping clearly evidenced a difference between the two concrete covers, with less electronegative values of potential with the 20mm concrete cover (between -367 and -200mV, Fig. 6-8) than with 10 mm of concrete cover (between -560 and -317mV, Fig. 6-8). For example, for wall B37, the shift is averagely varying from 100 to 200 mV between the two concrete covers (Fig.6). This could be linked to the moisture and salt contents, which are probably higher at lower depth. Distinct results are also noted between the two rebar shapes, for both concrete covers, but no real trend emerges.
Another phenomenon can be observed on the potential maps, which exhibit an electronegative gradient from the top to the bottom of the walls on both faces, but clearer with the lowest concrete cover (RL10 and HA10, Fig. 6). The explanation could come from the duration of the tide's descent on the one hand, as the bottom of the walls stays wet a longer time, which favors more electronegative values (Fig. 8). On the other hand, when the tidal coefficient is low, the walls are not totally immersed (Fig.  9), which means that the top of the walls remain in the open air and can dry, while the bottom is totally under water.
When comparing the results of the 11 walls, B04, B36 and B40 exhibit the most electronegative values, while B31, B38, B41 lead to the less electronegative ones. For B04, the explanation could come from its W/B, and its high carbonation potential as the concrete is composed of a CEM III with low clinker content ( Table 2). So that it is possible that the carbonation has progressed towards the lowest concrete cover of 10mm during the one year of exposure to natural ageing in semi-rural environment, before being implanted in marine environment. This carbonation could have favored the chloride penetration. The carbonation progression will be checked in the next steps of the project. For B40, its quite high porosity and its high sensitivity to chloride migration (Table 3) may be the cause of these quite electronegative values. The same observation can be done for B01 and B07, which exhibit quite electronegative values. Finally, for wall B36, its behavior is probably linked to its porous limestone as aggregates, its very high open porosity, and its quite high sensitivity to chloride penetration (Table 3).

Resistivity measurements
Resistivity evaluated with the Wenner technique (Fig. 11) is consistent with the potential mapping, with the lowest values obtained for walls B40, B36 and B04. In addition, the walls B01 and B07 also exhibited quite low values, all indicative of corrosion favorable conditions, according to the RILEM recommendation. Contrarily, B31, B38, B41 showed the highest resistivity, which is also consistent with the potential mapping, as for the clear variations observed between the two concrete covers. The results obtained with the Disc method (Fig. 12) are in line with that of the Wenner technique, but the measured values are generally lower (Fig. 13). This result has probably to do with the depth of concrete affected by the measurement, which is smaller for the Disc method than for the Wenner technique. Feliu et al. [11] demonstrated that at a depth equivalent to the disc diameter (2cm), the percentage of ohmic drop is already 71%, reaching 81% at a depth of 3cm. As the Wenner probe used in this study had an electrode spacing of 5cm, the depth concerned by the measurement might be higher [14] than that of the Disc method. As it is probable that the concrete is wetter on the surface than at depth, this could explain why the resistivity measured with the 4points method was generally higher than with the disc method.  It has to be noted that, with both techniques, slight differences can be observed between the two rebar shapes whatever the concrete cover. However, no trend emerges, as the variations are more or less in the range of the standard deviation for both measurement techniques.

Corrosion current density
The corrosion current densities are also consistent with the previous resistivity and potential measurements, with the highest values observed for designs B04, B36, B40 and B07 designs, and the lowest for B31, B38 and B41 (Fig. 14). When referring to the RILEM recommendation [13] almost all the recorded values are above the first threshold of negligible corrosion (0.1μA/cm²). Walls B04, B07, B36, B40, and in a lower extent B02, lead to corrosion current densities higher than 0.5μA/cm², which is the threshold of moderate corrosion. The threshold of 1μA/cm² was exceeded only for the smallest concrete cover with designs B04 and B36. At this early stage of exposure to a marine environment, except for the higher values (B04 and B36) the question of real rebar corrosion can be raised. Several hypotheses can be issued: is there enough oxygen at the rebar depth even at low tide? Is slag oxidation interfering with the rebar corrosion measurements? Is the pH of these new concrete designs high enough to protect the rebars? Are the existing RILEM threshold still adapted to the concretes containing mineral additions?
When comparing the values obtained with the two concrete covers, as for the preceding measurements, a tendency towards higher values for the lowest concrete cover is observed, for the same probable reasons.
Finally, it has to be noted that for the highest concrete cover (20mm), the values are in the same range for both rebar shapes, except for walls B04 and B40. Contrarily, for the lowest concrete cover (10mm), generally higher values are observed for the high-bond shaped rebars, except for B36. This might have to do with the calculation of the surface area of the rebars, which is more approximate for the high-bond shaped rebars.

Conclusions and outlooks
An intensive project PN-PERFDUB dedicated to the production and evaluation of several less polluting and more durable reinforced concretes, was initiated in 2015 in France. Among the aims of the project, one is to produce a database on corrosion performances of 11 of the concretes in either conditions favorable for carbonation, or in a marine environment A second aim is to try to improve the accuracy of the corrosion diagnosis, by mixing instant non-destructive testing and permanent monitoring using sensors.
The first results of the non-destructive electrochemical measurements obtained after almost a year of natural ageing in a marine environment showed noticeable differences between the concrete mix designs, but also a clear influence of the concrete cover and of the tide. Surprisingly, due to the early stage of exposure, some of the concretes exhibited quite high corrosion current densities. At this step of the project, several questions can be raised: is the durability of some mix designs questionable? Is oxygen sufficiently available at low tide at the rebar depth? What is the influence of the mineral additions in the binders, especially the slag content? Are the RILEM threshold still adequate for concretes with high mineral additions content (limestone, metakaolin, slag…) which questions the adequacy of their pH value to protect the steel rebars?
The next phases of the project will help to clarify these points, as they include a comparison with embedded sensors and with complementary non-electrochemical non-destructive techniques (evanescent field dielectrometry). In addition, lab tests will be performed to evaluate, among a series of performances (open porosity, transfer coefficients, chlorides diffusion coefficients…), the chloride ingress and the carbonation depth. Therefore, sampling campaigns are scheduled in parallel to the nondestructive measurements and detailed observation of the law-walls, for the next 20 years. Finally, the data will be compared to that of the extensive laboratory tests performed on these concretes in another part of the project, prior to the manufacture of the walls, and with the service life models.