Reinforcement corrosion of circular concrete columns under sustained load

. This paper presents the results of an experimental study of the axial capacity of circular spirally-reinforced concrete columns that were subjected to accelerated corrosion while sustaining a constant compressive service load. Two sets of columns of five columns each were designed with different spiral pitch to investigate the effect of concrete confinement on the corrosion propagation. Columns with non-confining spiral reinforcement had only their longitudinal reinforcing steel corroded, whereas in the other set of specimens, the spiral reinforcement was corroded and isolated from the vertical reinforcing steel. The cracking initiation, pattern, and widening of the concrete cover were monitored for a period of 4-10 months, and relations obtained between the crack widening and propagation and the elastic shortening of the column and steel mass loss were established. After the accelerated corrosion period, eight columns were tested to failure under either concentric or eccentric load; the vertical deformation, cracking pattern and failure type were monitored and recorded. The failure of specimens with non-confining spiral reinforcement was characterized by sudden crushing of the concrete and longitudinal reinforcement buckling, with a reduction in load capacity of 30-40%. Columns with confinement reinforcement failed in a ductile manner by progressive spalling of the concrete cover, maintaining their load capacity in spite of the corroded spiral reinforcement.


Introduction
The high pH of concrete that results from the hydration of cement protects reinforcing steel from corrosion.This high pH environment is disrupted when concrete is contaminated with sufficient chlorides or when carbonation of the concrete cover occurs.The passivity that naturally occurs in steel in a high pH environment is then broken, and corrosion of reinforcing steel begins when sufficient oxygen and water are present.The reaction creates corrosion products that have lower density and higher volume than the original steel reinforcement.The resulting pressure against concrete due to corrosion products build-up leads to cracking, spalling, and delamination of the concrete cover.
Damage resulting from chloride-induced reinforcement corrosion is often found in reinforced concrete (RC) infrastructure located in coastal areas or in cold regions that depend on the use of de-icing salts during the winter months.Canada's use of an estimated 4 to 5 million tonnes of de-icing salts per year for winter maintenance has led to widespread reinforcing steel corrosion-induced damage [1], mostly manifested in parking garages and highway bridges.Corrosion is particularly problematic in RC structures with inadequate concrete cover and high water-cement ratio concrete, and in aging infrastructure that has been in service for at least 30 to 40 years.
Limited studies are reported in the literature with respect to the effect of reinforcement corrosion on the serviceability and structural performance of RC columns [2][3][4][5][6].A major observation from these studies is that the load carrying capacity of corroded columns is lower than that of non-corroded columns, the reduction being attributed to: (i) the increase in load eccentricity caused by uneven corrosion of the longitudinal reinforcement; (ii) buckling of the longitudinal reinforcing bars when column ties are corroded; and, (iii) the actual deterioration of the concrete section from loss of confinement and cracking, spalling and/or de-lamination of the concrete cover [5].
The work presented herein results from an experimental study that explored the effect of reinforcement corrosion on cover cracking and the bearing capacity of short circular RC columns.It differs from previous studies in that the corrosion of specimens took place while the specimens were sustaining compressive service load.

Experimental program
The experiment involved 10 columns, 8 of which were corroded and 2 served as control.The RC columns sustained a constant service load throughout the chlorideinduced corrosion process, which was simulated through an accelerated corrosion regime.Crack propagation and crack widths in the concrete cover were monitored.Once the columns' level of damage reached a prescribed level, eight columns were tested to failure, with four under concentric axial load and four under eccentric axial load (i.e., under combined axial load and flexure).

Test specimens and material properties
Test specimens were circular and 260 mm in diameter with 6-15M longitudinal reinforcement, 10M spiral reinforcement, and a 20-mm clear concrete cover, as illustrated in Fig. 1.Two types of column designs were used in the experimental investigation, with five specimens of each type being built.The difference between the two designs was the spacing of the spiral reinforcement, which was specified to 220 mm for type CV columns, and 80 mm for type CS columns, respectively.The spiral pitch for type CV specimens is much larger than that specified by the Canadian concrete design standard, whereas the 80-mm pitch corresponds to detailing for minimum confinement reinforcement [7].The difference in spiral pitch between the two types of columns allowed investigating the effect of confinement on the behaviour of the corroded columns.The elevation views of the two types of test specimens are shown in Fig. 2. The nominal axial load resistance for type CV columns was calculated as 1,534 kN, whereas the second load peak due to confinement for specimens of type CS was estimated as 1,970 kN [7].The concrete mix had a relatively high water-cement ratio of 0.6 to allow corrosion damage to occur in the short period of 10 months.The mixing water contained 3% sodium chloride (NaCl) to trigger and accelerate the corrosion process.The average compressive strength of the concrete at 28 days was 27 MPa, and the yield strength of both the spiral and longitudinal reinforcements was 400 MPa.
The column ends were further confined with 12-mm thick steel collars, in order to limit damage propagation to the middle zone of the column instead of at anchorage regions.Each cap plate had four 300-mm long anchors bent at their ends to allow application of eccentric loading through the cap plates in the failure load test (see Fig. 3(a)).High-strength non-shrink grout was used for a depth of 50 mm at both ends of the column to make sure the columns were fully adhered to their caps at casting.The steel caps as well as other steel accessories were all painted with zinc-based protective coating in order to minimize the propagation of corrosion outside the columns' reinforcement.Curing was achieved by using moist burlap fabrics, which in turn were wrapped with plastic sheets to minimize evaporation, as shown in Fig. 3(b).The specimens were sprayed with water at least once per day in order to ensure sufficient moist conditions for hydration.The curing regime lasted for 14 days after formwork removal.

Sustained loading
Sustained loading began once the specimens cured properly.Two 332×332×38-mm steel plates placed at the top and bottom ends and connected with four 25-mm diameter steel rods were used to keep the columns under a sustained axial compression load, as illustrated in Fig. 4. A 16-mm circular steel plate with rounded edges was bolted to the 38-mm steel plate and sat against the steel caps of the concrete column to represent a pin connection.By subjecting the four steel rods to a tensile force, the RC columns were subjected to a constant axial compressive load.The tension steel rods were also painted with a corrosion resistant primer.A tension force of 55 kN was applied at each rod, with a total axial compression load of 220 kN on each column.This load represents 22% of the design factored axial load resistance of the column, simulating in-service conditions.
Strain gauges on the steel rods were used to monitor the applied sustained axial load.Linear Variable Differential Transformers (LVDTs) measured vertical deformation of the columns.One LVDT at mid-height of each column measured the circumferential expansion.All

Accelerated corrosion regime
In order to induce corrosion on the reinforcement, the columns were electrically connected to a PGSTAT100 potentiostat.A constant anodic current density was impressed on the steel reinforcing bars, while four stainless steel sheets (25-mm wide by 400-mm long) placed directly at opposite sides of the column exterior served as counter-electrodes.The stainless steel sheets were placed during casting on the wet concrete surface to ensure optimum contact.The reinforcing bars were connected to the potentiostat by means of welded screws with nuts covered with electrical tape to limit the effects of corrosion at these points.These connections were also zinc-paint coated for maximum protection.The specimens were further exposed to wetting and drying cycles, in order to ensure a sufficient supply of moisture and oxygen for the electrochemical reactions.The wet-dry cycles were applied at constant time intervals with a pump that sprayed water through a perforated hose placed around the columns, as shown in Fig. 5.

Fig. 5. Accelerated corrosion setup.
Corrosion was induced in the longitudinal reinforcement for 3 of the type CV columns, in the spiral reinforcement for 3 of the type CS columns, and in the entire cage for the remaining 2 columns.The accelerated corrosion regime lasted 10 months for some of the specimens and 4 months for others, as seen in Table 1.The current density applied on the specimens ranged between 250 and 300 μA/cm 2 to achieve mass losses between 15-25% according to Faraday's law An insulating barrier was necessary between the longitudinal and spiral reinforcements in order to corrode one component and not the other.For CV columns, shrink temperature tubes wrapped around the spiral reinforcement served to isolate it from the vertical reinforcement being corroded.For CS columns, electrical tape at the points of contact between the vertical and spiral reinforcements served as the barrier.

Load testing
After the corrosion process had reached the desired reinforcement mass loss levels, the column specimens were tested to failure by applying a compressive load either concentrically or eccentrically, as specified in Table 1.The specimen was subjected to compression with a W760×147 steel beam, which was attached to the strong floor by two 38-mm steel rods at each end (Fig. 6).
For eccentrically-loaded columns, only two rods at one end were subjected to compression.The tension rods on the other end of the beam were placed to stabilize the system and allow the rotational point to happen at the top of the column.These rods were put in tension with pressure jacks applied against the underside of the strong floor.This mechanism allowed the application of a bending moment at the column/steel beam joint that equals the tension load of the rods times the distance between the rods and the column.Tightening the nuts on the stabilizing rods of the other end distributed the tension forces between the two end-point reactions at the two ends of the beam, therefore controlling the amount of eccentricity and bending moment applied.This allowed testing for several load and moment combinations for the purpose of obtaining a representative axial-moment capacity interaction diagram of the corroded column specimens.Fig. 6.Failure load test setup.

Gravimetric testing
Samples of corroded reinforcement were retrieved from the columns after conducting the failure load test.The samples were cut into smaller bars (80-143 mm) and subjected to a chemical bath to obtain the weight of the specimens after the removal of corrosion products [8].The average mass loss was determined according to ASTM G1-03 [8].

Gravimetric mass loss
Figure 7 shows the results of the gravimetric analysis.The analysis of the reinforcement samples showed that the spiral reinforcement (10M) in series CV partially corroded, even though only the longitudinal reinforcement (15M) was intended to corrode.The corrosion of the spiral reinforcement in these columns could have been due to inadequate electrical isolation or its proximity to the exterior of the column, which was subjected to wet-dry cycles.Isolation was better between spiral (10M) and longitudinal reinforcements (15M) in CS samples as only the spiral reinforcement exhibited significant corrosion.
Specimens which were subjected to a longer accelerated corrosion regime exhibited higher mass losses as expected; steel mass losses in specimens CV1 and CS1 (subjected to a corrosion regime for 10 months) were over 20% compared to an average mass loss of 11% in specimens CV2 and CS2 (only exposed for 4 months).

Corrosion-induced cracking
For CV columns in which the longitudinal reinforcement was corroded, the cracking pattern was typical following the direction of the longitudinal reinforcement.Wider cracks were visible in columns corroded for a longer period.For the CS series, the cracking pattern was more random but generally followed the spiral reinforcement.
Figure 8 shows the rate of crack opening as a result of reinforcement corrosion.The figure displays the average crack width measured for each specimen.Figure 8(a period.All CV specimens followed the same trend of crack propagation with respect to time until approximately two months into the testing period, with crack openings between 0.083 mm and 0.14 mm. Figure 8(b) shows the crack width of CS columns versus time.The lateral crack widths in CS columns were significantly less than the widths of the longitudinal cracks in CV columns.The width of the longitudinal cracks along the vertical bars was approximately 6 to 8 times the width of the lateral cracks.The maximum lateral crack width for columns type CS was recorded as 0.083 mm as opposed to 0.712 mm as the maximum longitudinal crack width for columns type CV.Crack widths in the CS columns were kept smaller due to the sustained compressive load acting along the longitudinal axis of the columns.The theoretical mass loss over the testing period was calculated using Faraday's law per Eq. ( 1) with the assumption that the entire current is being used for corrosion without loss, i.e., where ms is the mass loss (g), M is the metal molar mass (55.85 g/mol for iron), z is the valence of the ion formed as a result of iron oxidation (i.e., z = 2), F is Faraday's constant (F = 96,485 C/mol), I is the corrosion current (A), and t is the time elapsed since corrosion started (s).
Figure 9 shows the mass loss against the average crack width to visualize whether a relationship exists between them.This analysis may be helpful because in assessing structures, the direct measurement of corrosion products is generally not possible.The plot in this case, however, does not demonstrate a clear trend; the mass loss between CS and CV specimens was similar despite the difference in crack widths.The average crack width of CS samples may have been kept smaller due to the effect of sustained loading that is perpendicular to the direction of cracking.Therefore, in terms of crack widths, the effect of corrosion on spiral reinforcement has a less severe impact on cover cracking than the effect of corrosion of vertical reinforcement in an RC column.The axial deformation over time increased at different rates was between CV and CS columns, as seen in Fig. 10.For CV specimens with the longitudinal reinforcement being corroded (Fig. 10(a)), the rate of increase of the deformation is greater at the beginning.Column CVG, whose entire reinforcement was subjected to corrosion, experienced most of its vertical deformation in the first two months and stabilized after that.CV1 displayed a continuous increase of its axial deformation during the entire testing period.Although columns CV2 and CV3 were only exposed to accelerated corrosion for four months, their rate of column deformation increase is similar to column CV1.The rate of increase of the circumferential expansion was very low for a period of 10 months.It is noted that these deformations include the effects of creep and shrinkage of the concrete.
Deformations in columns CV as well as the rate of their increase over time were higher than columns CS.For the latter specimens (Fig. 10(b)), vertical deformations appear to plateau slightly after 3 months of accelerated corrosion, before they start increasing again at 8 months.The circumferential expansion was relatively smaller than that of columns CV.In fact, it is concluded from the results that column expansion is further increased as the area of longitudinal reinforcement participating in the resistance of the axial load is decreased due to corrosion.

Capacity of tested specimens
Table 1 shows the results of the failure test.The corroded CV specimens (CVG, CV1, CV2 and CV3) exhibited reduced capacity and increased deformation.In general, the failure type was brittle, with crushing of the concrete and buckling of the longitudinal reinforcement.Figure 11(a) shows the control for CV samples after testing.Figure 11(b) shows rupturing of the spiral reinforcement and buckling of the longitudinal reinforcement of a specimen in which the entire cage was corroded (CVG).CS specimens retained confinement despite loss of the concrete cover approaching the first peak.Their failure was ductile, with the exception of CS3, which does not appear to be representative of the failure characteristics of CS beams due to the significant deformation when attaining the second peak.This deformation could be attributed to the yielding of the W760×147 load-transfer beam.Figure 12(a) shows the control for CS samples with typical concrete spalling breaching the first peak but was able to attain the second peak while maintaining confinement.Figure 12(b) shows specimen CSG in which the spiral reinforcement was corroded.This column and other specimens with spiral reinforcement corrosion were also able to attain the second peak while maintaining confinement.This indicates that the absence of longitudinal reinforcement corrosion allows for improved column confinement.The buckling of the longitudinal reinforcement occurred after attaining the second peak.The relative ratio of the axial capacity of corroded columns to that of uncorroded columns is plotted in Fig. 14 against reinforcement mass loss.Specimens with spiral reinforcement corrosion that remained confined by longitudinal reinforcement retained axial capacity up to the second peak.Specimens with longitudinal reinforcement corrosion (i.e., unconfined) lost between 30-40% of the bearing capacity for a mass loss ranging 10-25%.

Figure 13
Figure13plots the design axial load versus moment interaction diagram.Columns with corrosion along the longitudinal reinforcement (CV) failed in a region in which they were supposed to be safe.CS columns maintained confinement up to the second peak despite spiral reinforcement corrosion, indicating that spiral reinforcement corrosion has minimal impact on the behaviour of reinforced concrete columns provided there is minimum confinement reinforcement.

Fig. 14 . 4 Conclusions
Fig. 14.Ratio of axial resistance of corroded to uncorroded column versus mass loss.

Table 1 .
Details of column specimens and failure load test results.
) plots the crack width of CV columns versus time.The specimen in which the entire cage was corroded (CVG) exhibited the biggest crack width at the end of the test