Feb. 2010, Volume 4, No.2 (Serial No.

27) Journal of Civil Engineering and Architecture, ISSN 1934-7359, USA

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment
Muhammad Sigit Darmawan
Department of Civil Engineering, Institut Teknologi Sepuluh Nopember (ITS), Surabaya 60111, Indonesia Abstract: Like reinforced concrete (RC) structure, Prestressed concrete (PC) structures cannot escape from corrosion related problems, especially when they are subjected to very aggressive environment, such as chloride environment. The corrosion of PC and RC structures can take place if the concrete quality is not adequate, the concrete cover is less than that specified in the design, poor detailing during design and construction. For RC structures, corrosion in the reinforcing steel generally leads to serviceability problems (staining, cracking and spalling of concrete). By contrast, for PC structures, corrosion of prestressing strands may initiate structural collapse due to higher stress levels in the steel and smaller diameter of the prestressing steel. Research on corrosion effect on concrete structure has mainly considered the effect of corrosion have on reinforced and full prestressed concrete structure. In this study, a structural framework will be developed to predict the flexural strength of partial prestressed concrete structures in a chloride environment. The corrosion model previously developed for reinforced and prestressed concrete structures will be combined to predict the effect of corrosion has on partial prestressed concrete structures. Note that in partial prestressed concrete structures, both non prestressing steel (passive) and prestressing (active) reinforcement are utilized to carry the load. The framework developed will be combined with probability analysis to take into account the variability of parameters influencing the corrosion process. This approach allows more accurate prediction of service life of partial prestressed concrete structures in a chloride environment. Key words: Corrosion, partial prestressed concrete structure, chloride, probability.

1. Introduction
Corrosion of reinforcing and prestressing steel due to chloride contamination can result in considerable reduction in service life of concrete structures. In general, corrosion is of most concern because of the associated reduction in steel cross-sectional area, cracking, spalling and loss of bond, which over time will lead to reductions of strength and serviceability of structures. For prestressed concrete (PC) structures, the corrosion of prestressing strands may initiate structural collapse due to higher stress levels in the steel and smaller diameter of prestressing steel. A post-mortem analysis of the collapsed bridge showed that the 40 year

Muhammad Sigit Darmawan, PhD, research field: stochastic modeling of deteriorating of concrete in aggressive environments. E-mail: msdarmawan@ce.its.ac.id.

old post-tensioned bridge failed as a result of pitting corrosion near the box girder joint [1]. Chloride contamination is considered to be the major causes of corrosion of reinforced concrete (RC) and PC structures. This is caused by either from the application of deicing salts in cold regions or exposure to sea-spray in chloride environments. The deterioration of reinforced concrete structures due to chloride attack comprises of two stages. The first stage involves the movement of chlorides through concrete cover until they reach the threshold chloride concentration at the steel to initiate active corrosion (Fig. 1). The second stage is called corrosion propagation, where reinforcing steel corrodes causing loss of steel area (metal loss) and therefore reduces structural capacity. Different methods have been made to model chloride penetration in concrete (i.e., corrosion initiation). The different methods proposed clearly

26

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment

underline that the actual chloride penetration process is very complicated, and may involve a combination of processes contributing towards the overall ingress of chlorides [2]. Of all the available models, it is generally accepted that the model based on diffusion theory represents the chloride ingress in concrete. Hence, in this paper only corrosion propagation will be discussed. Corrosion propagation is mainly modeled by assuming a relatively uniform loss of material thickness [3], see Fig. 2. However, this approach is not accurate for concrete structures subjected to chloride attack, which usually experiences pitting corrosion, see Fig. 3 [4]. Pitting corrosion model for PC structures subjected to chloride attack can be found in the literature [5]. This model was developed from accelerated corrosion test using four slabs, each of dimensions 1500 mm 1000 mm 250 mm with wires/strands. Using a similar approach, pitting corrosion model for RC structures subjected to chloride attack was developed [6]. From these tests, it was found that the distribution of

Fig. 4 Inverse CDF (CDF-1) plots for maximum pit-depths in prestressing wires[5].

Fig. 5 Inverse CDF (CDF-1) plots for maximum pit-depths in a reinforcing bar [6].

Fig. 1 Deterioration model.

maximum pit-depths for prestressing wires and reinforcing bar is best represented by the Gumbel (EV-Type I), see Figs. 4 and 5.

2. Corrosion Model for Partial PC Structures

Research on corrosion effect on concrete structure has so far considered only the effect of corrosion have on reinforced and full prestressed concrete structure. In this study, a structural framework will be developed to predict the flexural strength of partial prestressed concrete structures in a chloride environment. Note that in partial prestressed concrete structures, both non prestressing steel (passive) and prestressing (active) reinforcement are utilized to carry the load. The corrosion model previously developed for reinforced

Fig. 2 General corrosion.

Fig. 3 Pit-configuration [4].

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment

27

and prestressed concrete structures will be combined to determine the effect of corrosion has on partial prestressed concrete structures. The framework developed will be combined with probability analysis to take into account the variability of parameters influencing the corrosion process. This approach allows more accurate prediction of service life of partial prestressed concrete structures in a chloride environment. The following assumptions are made in developing a more general probabilistic model for pitting corrosion: (1) homogeneous environment along the wire/rebar under consideration (corrosion rate assumed constant along wire/rebar); (2) after an initial period of corrosion, the number of pits formed is assumed constant, length of pit is held constant and pit depth continues to increase; and (3) at any cross-section of the wire/rebar only one pit can form. The predicted Gumbel distribution of maximum pit depth (a in mm) at any time T (years), corrosion rate icorr(1) in A/cm2 at start of corrosion propagation and wire/rebar length L (mm) is thus:

Table 1 Statistical parameters for the maximum pit-depths of prestressing strand. To-exp (years) 0.03836 icorr-exp Length (A/cm2) Lo(mm) o-exp 186 650 0.84 o-exp 8.10 a (mm) mean COV 0.91 0.17

-exp, -exp are Gumbel parameters Table 2 Statistical parameters for the maximum pit-depths of reinforcing bar. To-exp (years) 0.076712 icorr-exp Length (A/cm2) Lo(mm) o-exp o-exp 150 325 1.68 2.99 a (mm) mean 1.87 COV 0.23

then = 1 and = 0. The geometric model is then used to predict the loss of cross-sectional area for a pit size of depth a, see Fig. 3 [4].

3. Statistical Parameters of Maximum Pit-depths Distribution

The maximum pit-depth of corroded prestressing or reinforcing steel is an important parameter as it is the likely place of critical (minimum) section of the steel. Therefore, it is also the likely place where failure of the steel occurs. Statistical parameter of maximum pit-depth distribution used in this study is given in Tables 1 and 2 for prestressing strands [5] and reinforcing bars [6], respectively. These statistical parameters were obtained from accelerated corrosion tests using concrete slabs, each of dimensions 1500 mm 1000 mm 250 mm with strands/rebar. The accelerated corrosion process was introduced to the rebar using an electric current which was induced from a power supply through a current regulator. At the completion of each corrosion test the specimen was broken up and the steel then cleaned, dried and weighed using the method as specified by Standard Practice for Preparing, Cleaning, and Evaluating Corrosion Test Specimens [7]. The pit-depth in the corroded steel was then measured using a micrometer gauge. The pitting corrosion statistical parameters o-exp and o-exp are indicative only and increased confidence

f a (T, i corr , L) =
where

e 0.54

a
0.54

e e

a 0.54

T > Ti (1)

2 2 (T Ti ) +1 1 D o D o 0.0232i corr (1) 1 + +1 = 2 2 [To +1 1] D o D o 0.0232i corr (1) 1 + +1 ( + 1)(i Toexp )+ ( 1)(icorr (1)) 1 correxp To = exp ln icorr (1) ( + 1)

(2) (3) (4)

L 1 = oexp + ln = oexp oexp L oexp

i corr (T Ti ) = i corr (1) (T Ti )

T Ti 1 year (5)

obtained from statistical analysis of maximum pit depths recorded from the accelerated corrosion tests (see Tables 1 and 2), Do is the initial diameter of the wire/rebar (mm), and and are corrosion rate empirical factors. If corrosion rate reduces with time then = 0.85 and = -0.29 [3]. Otherwise, if corrosion rate is constant with time (time-invariant)

28

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment

Table 3 Statistical parameters for PC bridge girder [8]. Parameters fcyl concrete cylinder strength ki in-situ concrete strength factor fpy yield strength fy yield strength Cb bottom cover H beam depth (mm) B beam width
d

Mean Fca + 7.5 MPa

b

COV =6 MPa 0.1 0.025 0.10 = 7.9 mm = 3.6 mm = 3.7 0.046 0.20

Distribution Lognormal Normal Normal Beta Normal Normal Normal Normal Normal

Reference [9] [9] [10] [11] [12] [12] [12] [13] [14]

1.2-0.0082mean(fcyl) 0.88 fpkc 465 MPa Cbnom Hnom+0.8 Bnom+2.5 mm 1.01 1.0

ME flexure

ME - corrosion rate (icorr)

a

F c=specified (characteristic) concrete compressive strength; b= standard deviation; cfpk = characteristic tensile strength of prestressing steel; dME= model error.

in predictions will be obtained if these parameters are based on tests which more closely represent field conditions-that is, longer To-exp and lower icorr-exp.

4. Statistical Parameters of PC Structure

Using probabilistic analysis, the statistical parameters of maximum pit-depths distribution are combined with statistical parameters of prestressed concrete beams (i.e., beam dimension, concrete strength, steel yield strength, cover thickness, in-situ strength factor, model error for flexure and corrosion rate) to determine the effect of corrosion on flexural strength of partial PC beam. The statistical parameters of PC beams used in the probabilistic analysis are given in Table 3. Table 3 shows that the parameters influencing flexural strength of partial prestressed concrete beam have some uncertainty (random variables). The coefficient of variation of these parameters ranges from 0.025 to 0.2. For pitting model parameter, the coefficient of variation is also on the high side (e.g., 0.17 and 0.23), see Tables 1 and 2. Therefore, flexural strength determination of partial prestressed concrete beams based on deterministic analysis is inaccurate. Monte Carlo simulation will be used to determine the distribution of flexural strength of partial PC beams.

The flexural strength of partial prestressed concrete beam can be derived from the principal of equilibrium of section and strain compatibility between steel and concrete, as treated in most prestressed concrete design guides[15], see Fig. 6. It is assumed that ultimate load is reached when the concrete compressive strain in the extreme compressive zone equals 0.003 [16]. From Fig. 6, the ultimate flexural strength Mn at time T (in years) can be determined as M n (T) = Fp(T) dp + Fs(T) ds Cc(T) dc (6) Fp(T) = Ap(T)f p (7)
Fs(T) = As(T)f s (8) where Fp (T) and Fs (T) are the forces in the

prestressing strands and reinforcing steels at time T, Cc (T) is the force in the concrete compressive zone at time T and dp, ds and dc are the distances from the application of each force to the top of the concrete section. Ap (T) and As (T) are the prestressing steel and

5. Flexural Strength of Partial PC Structure

Fig. 6 Forces and strain diagram at ultimate flexural strength.

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment

29

reinforcing steel area at time T, and fp and fs are the stress in the prestressing steel and reinforcing steel, respectively. As the time since corrosion (T) increases, the corrosion will decrease the steel area and therefore reduce the flexural strength of partial prestressed concrete beam.

6. Illustrative Example
For illustrative purpose, the corrosion model developed will be used to determine the effect of corrosion has on partial PC beam shown in Fig. 7[15]. The prestressing steels (Ap) comprise two cables, each consisting of 18 supergrade 7-wire strands of 12.5 mm diameter, whereas passive reinforcement (As) comprise of 10 reinforcing bar of 32 mm diameter. The beam is exposed in a near-coastal area. Four different scenarios of corrosion are considered (1) No corrosion, (2) Only reinforcing bars corrode, (3) Only prestressing steels corrode, (4) Both reinforcing bar and prestressing steel corrode. The corrosion rate used in the analysis is 1 A/cm2 (0.012 mm/year), which can be classified as low to moderate corrosion rate [17]. For no corrosion case, the histogram of flexural strength of PC beam (Mn) obtained from Monte Carlo simulation is shown in Fig. 8. The Monte Carlo simulation is performed by generating randomly all variables influencing the ultimate flexural strength Mn
1500

(equation 6) and then determine the ultimate flexural strength Mn. This process is repeated many times as required (i.e., 1000000 simulations). The generation of these variables is based on statistical parameters a shown in Table 3. The detail of Monte Carlo simulation can be found elsewhere [18]. Fig. 8 shows that for the case without corrosion, the flexural strength has a mean value of 666.8 ton-m, with coefficient of variation of 4.44%. For the second case (only rebars corrode), the histogram of flexural strength of PC beam (Mn) after 40 years since corrosion initiation is shown in Fig. 9. Fig. 9 shows that the mean flexural strength has decreased from 666.8 ton-m to 560.6 ton-m (15% reduction) when compared with the case without corrosion.

Fig. 8 Ultimate flexural strength for no corrosion case.

250

900 Ap As

675

815

Fig. 7 Prestressed concrete T-beam (All units are in mm).

Fig. 9 Ultimate flexural strength for corrosion at reinforcing bars.

30

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment

Fig. 10 Ultimate flexural strength for corrosion at prestressing steel.

Fig. 11 Ultimate flexural strength for corrosion at reinforcing and prestressing steel.

For the third case (only prestressing steel corrode), the histogram of flexural strength of PC beam (Mn) after 40 years since corrosion initiation is shown in Fig. 10. Fig. 10 shows the mean flexural strength has decreased from 666.8 ton-m to 494.5 ton-m (26% reduction) when compared with the case without corrosion. For the last case (both passive and active reinforcement corrode), the histogram of flexural strength of PC beam (Mn) after 40 years since corrosion initiation is shown in Fig. 11. Fig. 11 shows the mean flexural strength has decreased from 666.8 ton-m to 486.5 ton-m (28% reduction) when compared with the case without corrosion. Fig. 12 demonstrates the effect of corrosion has on flexural strength of PC beam, by assuming that both prestressing steel and reinforcing bar is corroding. This figure shows that with time the flexural strength of the beam decreases considerably. For example after 40 years since corrosion initiation, the probability of flexural strength less than 500 ton-m is around 50%. For the case without corrosion, the probability of flexural strength less than 500 ton-m is zero. This figure clearly indicates that corrosion has significant effect on flexural strength of PC beam. For corroded high strength steel, such as prestressing steel, there is high possibility to have different mode of failure than yielding. It is known

Fig. 12 Distribution of ultimate flexural strength with time.

Fig. 13 Effect of different mode of failure of corroded prestressing steel on ultimate flexural strength.

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment

31

from literature [19] that with increasing strength of the steel, the possibility to have fracture or stress corrosion cracking (SCC) as mode of failure of the steel also increases. Furthermore, the presence of pitting with various depths and shapes may further raise the possibility of stress concentration associated with pit geometry, leading to brittle failure (not yielding). Hence, Fig. 13 shows the effect of different modes of failure of corroded prestressing steel on the flexural strength of the beam after 10 years since corrosion initiation. This figure indicates that different mode of failure of prestressing steel effect significantly the flexural strength of corroded partial prestressed concrete beam. For example, after 10 years since corrosion initiation time, assuming Stress Corrosion Cracking (SCC) as mode of failure of prestressing steel reduce the mean flexural strength by 60%, while assuming yielding as mode of failure only reduce the mean flexural strength by 10%. Note that as the corrosion progress, the probability of having SCC as mode of failure of prestressing steel will also increase.

strength of partial prestressed concrete beam). Each parameter is first treated as deterministic variables (equal to its mean). The relative sensitivity of each parameter is determined by considering the effect of (1 standard deviation) change from these mean values on the final results, while all other parameters are treated as random variables. The sensitivity analysis identified pitting model parameters for prestressing steel, corrosion rate, yield strength of prestressing steel, model error for flexure, and cover thickness as the most important parameters influencing the flexural strength of partial prestressed concrete beam, see Table 4. Following these results, it is expected that structural reliabilities are also most influenced by these important parameters. Hence, in order to obtain more accurate predictions of structural reliability, more effort should be directed at improving the accuracy of these parameters, including the pitting model statistical parameters.

8. Conclusions
This paper has described the structural framework to predict the service life of partial PC structure in a chloride environment. The framework combined the pitting corrosion model previously developed for reinforcing bar and prestressing wires to determine the effect of corrosion on flexural strength of PC beam. The framework also considers the variability of concrete properties of partial PC beam in the analysis using probability analysis. This will allows accurate prediction of service life of partial PC structure in a chloride environment. From the analysis, the following conclusions can be drawn: (1) Mode of failure of corroded prestressing steel has significant effect on flexural strength of partial PC beam. Assuming Stress Corrosion Cracking as a mode of failure, reduce the mean flexural strength by 60%, while assuming yielding as mode of failure only reduce the mean flexural strength by 10% after 10 years since corrosion initiation time;

7. Sensitivity Analysis
A sensitivity analysis was performed to study the relative importance of the variability of each random variable on the calculated response (e.g., flexural
Table 4 Variation of flexural strength by varying each parameter by . Parameters fcyl fpy fy ki H (Beam Depth) B (Beam Width) Cover Thickness Model Error for flexure icorr Pitting Model Parameter for Reinforcing Steel Pitting Model Parameter for Prestressing Steel Mn (flexural strength) 0.01% 4.37% 0.54% 0.33% 0.35% 0.67% 2.52% 3.57% 6.34% 0.70% 6.43%

32

Probabilistic Assessment of Pitting Corrosion Effect on Flexural Strength of Partial Prestressed Concrete Structures in a Chloride Environment Standard Testing Materials (ASTM), West Conshohocken, Pennsylvania, USA, 1999. M. S. Darmawan and M. G. Stewart, Spatial time-dependent reliability analysis of corroding pretensioned prestressed concrete bridge girders, Structural Safety 29 (2007) 16-31. M. G. Stewart, Workmanship and its influence on probabilistic models of concrete compressive strength, ACI Materials Journal 92 (1995) 361-372. S. A. Mirza, D. K. Kikuchi and J. G. MacGregor, Flexural strength reduction factor for bonded prestressed concrete beams, ACI Journal 77 (1980) 237-246. S. A. Mirza and J. G. MacGregor, Variability of mechanical properties of reinforcing bars, Journal of the Structural Division 105 (1979 a) 921-937. S. A. Mirza and J. G. MacGregor, Variations in dimensions of reinforced concrete members, Journal of the Structural Division 105 (1979b) 751-766. B. Ellingwood, T. V. Galambos, J. G. MacGregor and C. A. Cornell, Development of a Probability Based Load Criterion for American National Standard A58, National Bureau of Standards Special Publication 577, US Government Printing Office, Washington DC, 1980. K. A. T. Vu and M. G. Stewart, Structural reliability of concrete bridges including improved chloride-induced corrosion models, Structural Safety 22 (2000) 313-333. R. F. Warner, B. V. Rangan, A. S. Hall and K. A. Faulkes, Concrete Structures, Longman Press, Melbourne, 1998. AS 3600, Concrete Structures, Standards Association of Australia, Homebush, New South Wales, Australia, 2001. BRITE/EURAM, The residual service life of reinforced concrete structures, Final Technical Report, Report No. BRUE-CT92-0591, 1995. A. S. Nowak and K. R. Collins, Reliability of Structures, McGraw-Hill Higher Education, New York, 2000. D. A. Jones, Principles and Prevention of Corrosion, Macmillan Publishing Co, New York, 1996.

(2) Sensitivity analysis found that pitting model parameters for prestressing steel, corrosion rate, yield strength of prestressing steel, model error for flexure, and cover thickness as the most important parameters influencing the flexural strength of partial prestressed concrete beam.

[8]

[9]

References
[1] E. Proverbio and G. Ricciardi, Failure of a 40 year old post-tensioned bridge near seaside, Proc. Int. Conf. Eurocorr, London, 10-14 September, 2000. V. G. Papadakis, A. P. Roumeliotis, M. N. Fardis and C. G. Vagenas, Mathematical modeling of chloride effect on concrete durability and protection measures, in: R. K. Dhir and M. R. Jones (Eds.), Concrete Repair, Rehabilitation and Protection, E & FN Spon, London, 1996, pp. 165-174. K. A. T. Vu and M. G. Stewart, Structural reliability of concrete bridges including improved chloride-induced corrosion models, Structural Safety 22 (2000) 313-333. D. V. Val and R. E. Melchers, Reliability of deteriorating reinforced concrete slab bridges, Journal of Structural Engineering 123 (1997) 1638-1644. M. S. Darmawan and M. G. Stewart, Effect of spatially variable pitting corrosion on structural reliability of prestressed concrete bridge girders, Australian Journal of Structural Engineering 6 (2006) 147-158. M. S. Darmawan, Pitting corrosion model for reinforced concrete structures in a chloride environment, The Future Development in Civil Engineering Based on Research and Practical Experience (International Conference), University Pelita Harapan and University Stuttgart Germany, Lippo Karawaci, Jakarta, Indonesia, CD-ROM, 26-27 September, 2007. ASTM G1-03, Standard practice for preparing, cleaning, and evaluating corrosion test specimens, American

[10]

[11]

[2]

[12]

[3]

[13]

[4]

[14]

[5]

[15] [16] [17]

[6]

[18] [19]

[7]