International Symposium (NDT-CE 2003)Non-Destructive Testing in Civil Engineering 2003
|Start > Contributions >Posters > Corrosion:|
EVALUATION OF CORROSION LOSS OF STEEL REINFORCING BARS IN CONCRETE USINGLINEAR POLARISATION RESISTANCE MEASUREMENTSDW Law and JJ Cairns, Department of Civil and Offshore Engineering, Heriot-Watt University, Edinburgh, UK
SG Millard and JH Bungey, Department of Civil Engineering, Liverpool University, Liverpool, UK
The actual and predicted weight loss data for a number of mild steel bars contained in OPC concrete, subjected to three different environmental regimes and monitored using potentiostaticaly controlled linear polarisation resistance measurements are reported. The three sets of reinforced concrete specimens, were subjected to either
Each set of specimens were exposed to a daily regime of wetting and drying in a controlled environment for a duration of 1700 days. Two sets of testing were conducted at around 1200 days and 1700 days. Prior to casting the mild steel bars were individually cleaned and weighed. The weight loss for each bar due to corrosion was recorded at the end of the period of exposure. Instantaneous linear polarisation resistance corrosion measurements were taken on each bar at regular intervals for the duration of the exposure time. These resistances measurements were then integrated to evaluate a predicted total weight loss. The results show that the weight loss evaluated from experimental linear polarisation resistance measurements give a significant over-estimate of the actual weight losses recorded.
The problem of accurately and rapidly assessing the rate of corrosion of steel in reinforced concrete structures has long been a problem for the Civil Engineering industry. A number of electrochemical techniques have been developed to assess insitu the corrosion equilibrium and corrosion rate of the reinforcing steel. This data can then be used to enable an estimate to be made of the service life remaining to a reinforced concrete structure[1-3]. However, while the instantaneous corrosion rate may be used as a guide to estimate the probable loss of steel, concerns remain over the accuracy of these monitoring techniques and the reliability of such estimates.
Linear Polarisation Resistance
Linear Polarisation Resistance (LPR) measurement has become a well-established method of determining the instantaneous corrosion rate measurement of the surface of a reinforcing steel bar in concrete. The corrosion current, Icorr, is given by :
B - Stern-Geary constant
A value of 25 mV has been adopted for active corroding steel bars and 50 mV for passive conditions. In this investigation active corrosion was taken as one where the value of Rct for the surface of the bar was less than 10 kW/cm.
To measure Rct the surface of a reinforcing steel bar is polarised from its equilibrium potential by a small overpotential, D E, usually in the range 10 to 30mV. The resulting current is then monitored at the end of a selected time period, usually between 30 seconds to 5 minutes. Rct is determined by dividing the applied overpotential by the induced current.
In order to evaluate the corrosion current density, icorr, the surface area, A, of steel that has been polarised needs to be accurately known. The corrosion rate may then be correlated to corrosion current density and mean corrosion penetration rate, Table 1
A series of reinforced concrete specimens have been manufactured for anchorage bond/corrosion investigations. Each specimen also contains two 6mm diameter mild steel bars, Figure 1, to correlate the measured weight loss with the predicted weight loss from LPR measurements. The weight of the cleaned 6mm bars pre- and post exposure have been measured and LPR measurements have been taken at regular intervals, between two and eight weeks, over the duration of the exposure. The weight loss predicted from integrating the LPR readings is then related to the actual measured weight loss on the 6mm steel bars.
A total of 27 laboratory specimens were manufactured, comprising three sets of nine specimens. Each set of specimens was subjected to one of three different environmental regimes. The regimes were:
The three sets of specimens were placed in three environmental control cabinets. All the specimens were subjected to a 22-hour dry/2-hour wet daily cycle. This was changed to a 6-day dry/1-day wet cycle for the carbonation exposure specimens after the initial testing. The environmental exposure regime remained the same for the specimens contained in the chloride cabinet. An electric fan was used during the dry cycle to increase the rate of drying in the initial stages of the exposure. Chloride exposure was achieved by sprayed wetting of the specimens using a 1 Molar sodium chloride solution. Full carbonation beyond the depth of the steel bars was achieved by exposing the specimens to a 100% CO2 atmosphere for two weeks. Following this carbonation procedure the specimens were then subjected to the wet/dry cycling regime using tap water. Both chloride and carbonation specimens were exposed to a standard oxygen/nitrogen atmosphere. For the control specimens a nitrogen rich atmosphere, of over 90% nitrogen, was maintained. This ensured that the oxygen available to support corrosion was reduced to a minimal level. However, these specimens were still subjected to the same wet/dry cycle using tap water as the corrosion specimens, to ensure that any differences measured were solely due to corrosion.
The nine specimens in each cabinet were split into three groups. Each group contained a different type of anchorage bar, used in subsequent corrosion/bond testing. Each of the three specimens within a group was to be tested at a different time, to collect data at different levels of corrosion. To date two sets of tests have been conducted, a third set of testing is to be conducted when corrosion levels have reached a more advanced state. Each specimen (Figure 1) contained two 6mm diameter mild steel bars, two 10mm diameter mild steel hanger bars and four main anchorage bars. The 6mm mild steel bars were electrically isolated from all other bars. A low-grade 20 N/mm to 2 concrete mix was used throughout to promote rapid corrosion.
LPR corrosion rate measurements were taken by first measuring the solution resistance, Rs, of the concrete at an AC frequency of 300Hz, which was later subtracted from the polarisation resistance, Rp, of the bar measured at the surface of the concrete to compensate for the resistance of the zone of cover concrete.
The polarisation resistance was evaluated by perturbing the steel potentiostatically from its equilibrium potential by 10mV and measuring the current, DI after a delay of 30 seconds. To ensure a high accuracy, the average LPR measurement was taken from measurements using both a positive and negative overpotential. The measurements were controlled using a laptop PC, giving a real time current vs time decay curve, Figure 2. Readings on each test bar were taken at regular intervals, varying between two and eight weeks.
Initial weight loss data was measured by breaking open one set of specimens after periods of between 1026 and 1168 days. Weight loss testing was carried out on a second set of specimens between 1625 and 1705 days. A total of 28 weight loss bars from fourteen specimens were tested in total.
For these experiments the size of the bars being measured was similar in size in comparison to the auxiliary electrode and hence it was assumed that the total, exposed, surface area of the 6mm weight loss bars was polarised. For reinforcing steel within real concrete structures, where the steel reinforcement is electrically linked, an accurate knowledge of the area of steel being polarised during a LPR corrosion rate measurement is a major potential source of error. Typical corrosion rates are shown in Table 1.
To determine the total corrosion current over the duration of the exposure period the area beneath the corrosion rate against time graph is integrated. Typical plots for pairs of bars from chloride and carbonation exposure bar specimens are given in Figures 3a and 3b. Once the total corrosion current has been integrated over time the total weight of steel lost can be calculated from:
Icorrtotal = Total current passedC= Charge per mole of iron
M = Atomic mass iron
If a uniform rate of corrosion over the surface of the bar is assumed, the loss of cross sectional area of steel can also be calculated from the total weight of steel lost. The loss of cross section for pitting corrosion clearly cannot be calculated due to the localised nature of this type of corrosion. However, it has been estimated that pitting corrosion may potentially give up to five times the value of uniform corrosion.
The initial weight of a 6mm bar was between 65g and 75 g. The measured weight loss and the integrated weight loss evaluated from LPR corrosion measurements for the individual bars are given in Tables 2, 3 and 4. The location and exposure regime for each bar is also noted.
The control specimens, Bars 1-8, Table 2, show that the nitrogen control environment has inhibited almost all corrosion. The weight losses recorded of 0.1g-0.2g are minimal compared to the overall mass of the samples, less than 0.3%.
The variation of the corrosion rate measurements for bars exposed to chloride and carbonation attack are illustrated in Figures 3a and 3b. This shows the that corrosion was rapidly initiated for all specimens in a 2-4 weeks period and then stabilised at a mean value of between 1-2 mA/cm2 for both the chloride and carbonation specimens. The corrosion rate for the carbonation specimens has remained within this range for the duration of the trial period. Some cycling in the rate is observable which is consistent with seasonal variations in temperature and humidity . The specimens in the chloride environment show a significant increase in corrosion rate after 400-500 days; a corrosion rate of up to 10 mA/cm2 being observed. While there is the undercurrent of seasonal fluctuations there is considerably more scatter on these results when compared to the carbonation data. The results show the sensitivity of the measurements to environmental conditions and illustrate, particularly with chloride induced corrosion the need to take a large number of measurements to determine the average annual rate.
Analysis of the data for the all the corroded bars shows that for all corroded bars the predicted value is an overestimate of the measured value, illustrated in Figure 4. Combining all the weight loss data for the corrosion specimens, Tables 3 and 4, gives an overall ratio mean value of 2.1 with a standard deviation of 1.4. This corresponds to an overestimate of weight loss from LPR measurements of 110%.
However, if the data is analysed according to exposure regime, Table 5, the data shows significantly greater scatter for the chloride specimens compared to the carbonation specimens. Combining the chloride data gives a ratio mean of 3.04 and standard deviation of 1.92 and the carbonation specimens give a ratio mean of 1.56 and standard deviation of 0.26. These correspond to overestimates for chloride corrosion of 204% and carbonation attack of 56%. The data is presented both as the mean ratio of the individual specimens and as the combined values for the total measured and predicted weight loss for each of the sets of specimens. This is to account for only two chloride specimens at 1168 days being tested.
The results suggest that the type of corrosion may affect the reliability, and possibly the accuracy, of the technique. Carbonation attack results in general corrosion, while chloride attack can result in pitting corrosion. Both types of corrosion were observed in a visual inspection of the bars.
However, while the LPR technique does show an overestimate of the weight loss of the bars this still represents a reasonable approximation, and an overestimate of corrosion rate corresponds to an early indication of potential damage and the need for a more detailed inspection. An underestimate of corrosion rate would result in significant damage prior to the monitoring indicating the need for inspection.
It must also be noted that the LPR technique does not indicate the level of corrosion present in the bars prior to the commencement of monitoring and this must be ascertained on site by a detailed visual inspection.
The position of the bars, top or bottom cast, did not appear have an effect on the level of corrosion. The top cast bars have a slightly higher level of corrosion compared to bottom cast bars. The combined total weight loss for the 10 top cast bars was 12.2 g and 11.9 g for the 10 bottom cast bars.