Durability Predictions Using Early-Age Durability Index Testing

Durability Predictions Using Early-Age Durability Index Testing JR Mackechnie & MG Alexander Summary: Durability of reinforced concrete structures is often dependent on the corrosion of reinforcement due to ingress of aggressive ions and fluids. Using the premise that the potential durability of concrete is determined by the protection provided by the cover concrete to the embedded steel, the resistance may be defined in terms of transport properties such as absorption, permeation and diffusion. A suite of durability index tests was developed to characterize the earlyage resistance of concrete to transport of fluids and ions that affect corrosion of reinforcement. These tests were found to produce reasonable predictions of durability performance for reinforced concrete structures. Early-age laboratory characterization testing is discussed together with field data regarding carbonation rates and chloride ingress into concrete. Findings suggest that the approach may be usefully applied as performance specifications where durability of reinforced concrete structures must be guaranteed. Keywords. Carbonation, characterization, chlorides, durability, predictions 1 INTRODUCTION The bulk of durability problems concern the corrosion of reinforcing steel rather than deterioration of the concrete fabric itself. The adequacy of the concrete cover layer is therefore critically important in resisting aggressive agents from the surrounding environment. A plethora of durability tests has been developed to measure fluid transport rates by various mechanisms through concrete. Most methods require sophisticated equipment and complex monitoring and therefore have limited practical value for site concrete. The concept of durability index testing was proposed to provide practical means for characterizing the durability potential of concrete (Alexander 1997). Such index tests must be sensitive to important material, processing and environmental factors affecting concrete. The purpose of material indexing is to provide a reproducible engineering measure of microstructure and properties important to concrete durability at a relatively early age (e.g. 28 days). Thus it should be possible to produce concretes of similar durability by a number of different routes: additional curing, lower w/c ratio, different binder types, etc. In order to assess the effectiveness of these laboratory techniques, field studies of concrete in service are essential. These studies allow deterioration mechanisms to be accurately assessed under normal exposure conditions. The disadvantage of field exposure testing is that deterioration may take years to proceed to a measurable extent thereby requiring extrapolations to estimate long-term trends. These extrapolations may be misleading if the results are not independently validated with longterm data. 2 LABORATORY TESTING Transport of aggressive agents into concrete is primarily caused by absorption, permeation and diffusion mechanisms. The influence of each transport type is dependent on environmental conditions and material properties of concrete. Carbonationinduced corrosion of reinforcement is caused by gaseous diffusion of carbon dioxide through relatively dry concrete. Chloride-induced corrosion in contrast is caused by ionic diffusion through near- or fully saturated material. Results reported in this paper have been produced from several different South African studies over the last ten years. Concrete mixes were produced with local materials and are denoted: 100% Portland cement, 30% fly ash, 50% blast-furnace slag and SF 10% condensed silica fume. Grade of concrete refers to the nominal concrete cube strength of the material at 28 days. 2.1 Oxygen permeability test Permeability is defined as the capacity of a material to transfer fluids under the action of an externally applied pressure. The permeability of concrete is dependent on the concrete microstructure, the moisture condition of the material and the characteristics of the permeating fluid. Ballim developed a falling-head permeameter that allowed simple measurement of oven-dried concrete exposed to oxygen under pressure (1993). The test equipment and typical measurements are shown schematically in Fig. 1. 9DBMC-2002 Paper 241 Page 1

PERMEATING GAS OUTLET CONCRETE RUBBER COLLAR GAS INLET PRESSURE 100 kpa PRESSURE VESSEL CONTAINING OXYGEN PRESSURE DECAY CURVE CONNECTION TO PRESSURE TRANSDUCER TIME GAS OUTLET VALVE Figure 1: Oxygen permeability apparatus and typical test data Concrete core samples are initially dried at 50 0 C before being exposed to oxygen at 100 kpa pressure. From the slope of the pressure decay curve (using a logarithmic transformation), the Darcy coefficient of permeability is determined. The coefficient of permeability is an unwieldy exponential number that is simplified by defining the oxygen permeability index (defined as the negative logarithm of the coefficient of permeability). Concrete with oxygen permeability index values above 10.0 may be considered to have excellent impermeability characteristics whilst values below 9.0 indicate poor impermeability. Figure 2 show typical results measured at 28 days after either fully wet or dry curing conditions. The oxygen permeability index values were determined from five different laboratories around South Africa and numerous different concrete types. The high degree of scatter was partly due to the wide range of aggregates types used around the country. 11 10.5 10 9.5 9 Wet Dry Wet Dry 8.5 8 0.3 0.4 0.5 0.6 0.7 0.8 0.9 w/c ratio Figure 2: Oxygen permeability index values at 28 days 9DBMC-2002 Paper 241 Page 2

2.2 Chloride conductivity test Diffusion is the process where fluids or ions move through a porous material under the action of a concentration gradient. Diffusion occurs in partially or fully saturated concrete and is the dominant internal transport mechanism for marine concrete. Diffusion rates are dependent on temperature, saturation level of concrete, type of diffusant, chemical interactions and inherent diffusibility of the material. Natural diffusion tests are extremely time consuming and have lead to the development of various accelerated tests. The chloride conductivity test was developed by Streicher in an attempt to provide a rapid assessment of the chloride resistance of concrete (Streicher & Alexander 1995). Concrete core samples are preconditioned at 28 days to standardize the pore water solution (oven-dried at 50 0 C for seven days followed by vacuum saturation in 5M NaCl solution). The sample is then placed in a two-cell conduction rig containing NaCl solution and a 10V potential difference applied. The apparatus (shown in Fig. 3) allows virtually instantaneous readings under controlled laboratory conditions. DC POWER SUPPLY A V 5M NaCl SOLUTION STEEL CATHODE PLASTIC TUBES CARBON ANODE CONCRETE SAMPLE Figure 3: Chloride conductivity apparatus Chloride conductivity testing has been found to be sensitive to changes in concrete microstructure caused by w/c ratio, initial curing and type of binder. The type of cementitious material has a significant effect on chloride conductivity. Portland cement concrete for instance generally has high conductivity values with only high-grade material achieving values below 1.0 ms/cm, which may be defined as excellent chloride resistance. Slag concrete is contrast has significantly lower chloride conductivity values as shown in Fig. 4. 9DBMC-2002 Paper 241 Page 3

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 w/b ratio Figure 4: Chloride conductivity values measured at 28 days (wet cured concrete) 3 DURABILITY PERFORMANCE Corrosion of steel in concrete is a complex phenomenon, influenced by many internal and external factors. In order to quantify durability performance of concrete, only depassivating effects such as carbonation and chloride ingress are considered. In other words, only the initiation part of the corrosion cycle is considered and factors controlling subsequent corrosion propagation and damage are ignored. 3.1 Carbonation Carbonation of concrete is affected by material, constructional and environmental factors. Effective curing of concrete is known to enhance the near-surface quality of concrete and is particularly important for fly ash and slag concrete. The amount of carbonatable material, in the form of calcium hydroxide, is another important factor. Typical carbonation depths for grade 40 concrete exposed to mild outdoor conditions (average humidity of 80%) are shown in Fig. 5. Concrete samples were large blocks that were cored and tested with phenolphthalein indicator solution. Moist refers to seven days initial moist curing while dry indicates no active moist curing was done. 25 20 15 10 5 - Moist - Dry - Moist - Dry - Moist - Dry 0 0 1 2 3 4 5 6 Time (years) Figure 5: Carbonation depths for grade 40 concrete (outdoor mild exposure) 9DBMC-2002 Paper 241 Page 4

The rate of carbonation is also dependent on environmental conditions during service. A relative humidity of approximately 65% is generally believed to be the optimum condition for carbonation, being dry enough to allow rapid gaseous diffusion of carbon dioxide whilst allowing sufficient moisture for the carbonation reaction to proceed. Table 1 shows carbonation depths for similar concretes exposed to either dry laboratory conditions (temperature 23 0 C and R.H. 60%) or mild outdoor environment (average temperature 17 0 C and R.H. 80%). Results confirm that drier environments allow significantly higher carbonation depths, particularly when concrete is inadequately cured. Grade (MPa) Table 1: Carbonation depths for concrete exposed to different environments 4 years Initial Curing 80% 60% 80% 60% 80% 60% 20 Moist 13.0 18.5 13.0 21.5 15.0 23.5 Dry 14.0 21.0 17.0 37.0 20.0 41.0 40 Moist 5.0 9.5 5.0 10.0 6.0 14.5 Dry 6.0 12.0 6.0 13.0 9.0 24.5 60 Moist 1.0 3.0 1.5 6.0 2.0 4.5 Dry 2.0 6.0 2.0 14.0 3.0 18.5 The rate of carbonation was found to agree with the general power series relationship given in equation 1. x = k c t 0.4 (1) where x is the carbonation depth in millimetres, k c is a material coefficient and t is time in years. The equation allows simple extrapolations of long-term carbonation depths from medium-term data (i.e. one to seven years exposure). 3.2 Chloride ingress Chloride ingress into concrete was assessed by exposing concrete blocks to a range of marine environments. Exposure testing was conducted at two sites on the Cape Peninsula in South Africa with five separate projects being undertaken over periods from one to seven years. Concrete cores were extracted from the concrete after exposure and sliced into fine increments for chloride analysis. Total acid-soluble chloride concentrations were determined in accordance with BS1881 but using a potentiometric titration (British Standards 1988). From the measured chloride profile, the surface concentration and diffusion coefficient were determined using the solution of Fick s second law of diffusion. Typical results for grade 40 concrete exposed to very severe marine exposure (tidal zone location but with limited wave action and abrasion) are compared. Water/binder ratios were in the region of 0.50 for these concretes (Mackechnie & Alexander 1997a). Surface concentrations at the different ages are shown in Fig. 6. concrete was found to have variable levels whereas surface concentrations for and concrete increased significantly with time although appearing to be stabilizing by eight years. The variable surface concentrations of concrete had an impact on measured diffusion coefficients, these two parameters being directly affected by one another in the Fick s law equation. 9DBMC-2002 Paper 241 Page 5

1.0E-10 1.0E-11 1.0E-12 1.0E-13 0 2 4 6 8 Time (years) Figure 6: Surface concentrations with time Diffusion coefficients were found to reduce with time, particularly for and concrete. After eight years exposure, concrete was found to have relatively high diffusion coefficients such that reinforcement was at risk of corrosion to depths greater than 70 mm. Figure 7 illustrates the change in diffusion coefficients with time for grade 40 concrete exposed to very severe marine conditions at Simonstown tidal zone. 1.0E-10 1.0E-11 1.0E-12 1.0E-13 0 2 4 6 8 Time (years) Figure 7: Diffusion coefficients with time Analysis of chloride ingress data revealed what other researchers have observed, namely that Fick s law of diffusion is not able to accurately predict chloride levels in concrete (Bamforth 1994). Some allowance needs to be made for the reduction in diffusion coefficient values with time, particularly for fly ash and slag concrete. A modified solution of Fick s second law of diffusion was therefore formulated based on the empirical relationship given in equation 2 (Mackechnie 1996). D c = D i t m. (2) where D i is the diffusion coefficient at one second, m is the empirical material coefficient relating to the reduction in diffusion coefficient (D c ) with time. The modified solution of Fick s second law of diffusion can therefore be expressed as equation 3. 9DBMC-2002 Paper 241 Page 6

C x = C s (1 - erf[ x ] ).(3) 2 D i t (1-m) where C x is the chloride concentration at depth x and time t, and C s is the surface concentration which is assumed to be constant. 4 DURABILITY PREDICTIONS Service life predictions of reinforced concrete structures are affected by numerous variables that prevent precise estimates of durability performance. Since durability index tests are based on transport mechanisms associated with deterioration, it was thought that these indexes could be used for durability predictions. Durability predictions of an empirical nature were therefore sought using correlations between early-age characterization tests and durability performance results. 4.1 Carbonation predictions Correlations between oxygen permeability index values recorded at 28 days and carbonation depths after exposure were found to be good. Figure 8 shows the relationship for concrete after four years exposure. Oxygen permeability testing was found to be more sensitive than strength in predicting the carbonation resistance of concrete. 40 30 20 10-60% -60% -60% -80% -80% -80% 0 8 8.5 9 9.5 10 10.5 Oxygen permeability index Figure 8: Carbonation versus oxygen permeability index An empirical prediction model for carbonation was therefore formulated using the oxygen permeability test. Early-age characterization testing was used to estimate medium-term carbonation depths, from which long-term results may be extrapolated using equation 1. Using this approach 50 year carbonation depths may be predicted for different environments as shown in Fig. 9 9DBMC-2002 Paper 241 Page 7

1E-10 1E-11 1E-12 SF 1E-13 0 0.5 1 1.5 2 2.5 3 Chloride conductivity (ms/cm) Figure 9: Predicted carbonation depths after 50 years 4.2 Chloride predictions Chloride diffusion through concrete is a complex process involving material, environmental and structural effects. Correlations between early-age properties and durability performance are complicated by material-specific responses. This was observed when correlations between diffusion coefficients and 28 day chloride conductivity values were analysed. Figure 10 shows the relationship found for concrete exposed to very severe marine environment for two years. The two year correlation is shown as results from later ages did not include the full data series but showed similar trends. 1E-10 1E-11 1E-12 SF 1E-13 0 0.5 1 1.5 2 2.5 3 Chloride conductivity (ms/cm) Figure 10: Diffusion coefficient (two years exposure) versus chloride conductivity A more reliable relationship was found when longer-term chloride conductivity values were compared with diffusion coefficients. Some allowance therefore needs to be made for the differing rates of maturity of concrete and chemical interactions between diffusant and concrete. A prediction model based on this premise has been formulated and has been found to have reasonable reliability when independently validated using long-term site data (Mackechnie & Alexander 1997b). 9DBMC-2002 Paper 241 Page 8

The prediction model consists of using 28-day chloride conductivity testing to predict long-term diffusion coefficients but allows for longer-term cementing reactions and chemical interactions between chloride ions and concrete. From the model, time to corrosion activation may be estimated for different concrete types. Figure 11 shows predictions for moist cured concrete, exposed to very severe marine conditions, with cover to reinforcement of 60 mm (note that silica fume data is somewhat conservative due to limited long-term data, i.e. two year exposure only). 100 80 60 40 SF 20 0 0.3 0.4 0.5 0.6 0.7 Water/binder ratio 5 CONCLUSIONS Figure 11: Time to corrosion for reinforced concrete very severe exposure A pragmatic approach is proposed to solve the problem of lack of durability in concrete structures. This approach takes a broad view, incorporating a proper definition of the environment, characterization of the material and field observations of durability performance. By integrating these various aspects, performance specifications can ultimately be produced to achieve specified durability criteria. Durability index tests such as oxygen permeability and chloride conductivity were found to be sensitive to material, environmental and processing factors that affect concrete durability. The tests were also found to produce reasonable predictions of durability performance under a range of environmental conditions. Further, the durability index approach has a sound theoretical basis while still be sufficiently quick and practical for site use. 6 ACKNOWLEDGEMENTS All the reported work was undertaken in South Africa under the Industry/NRF Collaborative Research Programme funded by South African cement and construction companies. The support and guidance of Drs Graham Grieve and Brian Addis over the last ten years is particularly acknowledged. 7 REFERENCES 1. Alexander, M.G. 1997, An indexing approach to achieving durability in concrete structures, FIP 97 Symposium: The Concrete Way to Development, Concrete Society of Southern Africa, Johannesburg, pp. 571-576. 2. Ballim, Y. 1991, A low cost falling head permeameter for measuring concrete gas permeability, Concrete Beton, 61, pp. 13-18. 3. Bamforth, P.B., Admitting that chlorides are admitted, Concrete, 28(6), pp. 18-21. 4. British Standards Institute, Chloride content determination for concrete, BS 1881 Part 124. 5. Mackechnie, J.R. 1996, Predictions of reinforced concrete durability in the marine environment, PhD thesis, University of Cape Town. 6. Mackechnie, J.R. & Alexander, M.G. 1997a, Exposure of concrete under different marine environments, ASCE Journal of Materials in Civil Engineering, 9(1), pp. 41-45. 7. Mackechnie, J.R. & Alexander, M.G. 1997b, Durability findings from case studies of marine concrete structures, Cement, Concrete and Aggregates, 19(1), pp. 22-25. 8. Streicher, P.E. & Alexander, M.G. 1995, A chloride conduction test for concrete, Cement and Concrete Research, 25(6), pp. 1284-1294. 9DBMC-2002 Paper 241 Page 9