Time-dependent chloride penetration in concrete in marine environments

Construction and Building Materials 152 (2017) 406–413

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Time-dependent chloride penetration in concrete in marine environments Lingjie Wu, Wei Li ⇑, Xiaoniu Yu ⇑ College of Civil Engineering and Architecture, Wenzhou University, Wenzhou 325035, China

h i g h l i g h t s
A field investigation is conducted in the Beibu Gulf, Guangxi province, China.
Distribution of age factor is theoretical studied and experimental verified.
There commonly used corrosion models are comparatively analyzed.

a r t i c l e

i n f o

Article history: Received 7 March 2017 Received in revised form 1 June 2017 Accepted 2 July 2017

Keywords: Chloride Concrete Exposure condition Age factor Corrosion initiation time

a b s t r a c t In order to study the effects of exposure conditions (atmospheric, tidal and splash zones) on chloride ingress into concrete and time-dependent chloride diffusivity of concrete, the chloride concentration of existing concrete in the Beibu Gulf was tested by field investigation. In addition, a reasonable calculation model was chosen by comparative analysis to achieve an accurate corrosion initiation time of the tested concrete. The results indicate that splash zone affects the durability of concrete structures more harshly than tidal and atmospheric zones. The age factor, which represents the time-dependent properties of chloride diffusion coefficient, is a normally distributed random parameter. In atmospheric, tidal and splash zones, the mean value of the age factor equated to 0.19, 0.35 and 0.43, respectively. Compared with the Life 365 model and the LNEC E465 model, the DuraCrete 2000 model can better characterize the chloride transport in the tested concrete. Based on this model, given the target reliability index bd = 1.3–1.5, the corresponding corrosion initiation time is 42–47.5a. Ó 2017 Elsevier Ltd. All rights reserved.

1. Introduction Reinforced concrete (RC) is one of the most widely used manmade building materials in the world, and chloride ingress is the most deteriorating factor for RC structures exposed to marine environments [1]. Due to the extensive use of reinforced concrete and high maintenance costs, the prediction of chloride transport in concrete has attracted more and more attention from scholars and engineers [2,3]. Fick’s second law is widely used nowadays to describe chloride penetration in concrete, it is given as follows:

   x Cðx; tÞ ¼ C s 1  erf pffiffiffiffiffiffiffiffiffi 2 Dt

ð1Þ

where Cs is surface chloride concentration, D is apparent chloride diffusion coefficient, x is corrosion depth, C(x,t) is the chloride con⇑ Corresponding authors. E-mail addresses: [email protected] (W. Li), [email protected] (X. Yu). http://dx.doi.org/10.1016/j.conbuildmat.2017.07.016 0950-0618/Ó 2017 Elsevier Ltd. All rights reserved.

centration at the corrosion depth x with exposure time t, and erf(.) is the error function. Numerous studies have reported that D and Cs are the key parameters that influence the transport of chloride in concrete [4–7]. And these two parameters were mainly affected by the concrete mix ratio and exposure condition (i.e. atmospheric, splash and tidal zones). Moreover, early studies usually assumed that D and Cs were constant, while the time-dependent properties of D and Cs have become better understood in recent years. Song et al. [4] systematically described the influence of the water-cement ratio, mineral admixture, and curing conditions on chloride diffusion coefficient and surface chloride concentration, and pointed out that D and Cs are time-varying parameters. Valipour et al. [6] also discussed the influence of the concrete mix ratio, curing conditions and exposure conditions on chloride penetration behavior. However, compared to the water-cement ratio and mineral admixture, the influence of curing conditions on chloride transport in concrete is not so obvious [3,7]. When referring to the time dependency of chloride transport, based on the data collected from 11 coastal concrete bridges, Pack

L. Wu et al. / Construction and Building Materials 152 (2017) 406–413

et al. [8] built up a prediction model to describe the timedependent properties D and Cs; the model was also compared with the well-known model, Life 365. Similarly works have also been presented by Petcherdchoo [9] and Muthulingam [10], in which the object was fly ash concrete instead of ordinary Portland cement (OPC) concrete. Besides, laboratory investigations were carried out by Wang et al. [11], in which the impact of sustained compressive loading on time-dependent chloride diffusivity of concrete was studied. For chloride caused reinforcement corrosion of RC structures, scholars and engineers have done a lot of work. However, most of these studies focus on indoor accelerated corrosion tests or field exposure tests. Relatively few studies have been devoted to existing RC structures. Besides, the effect of time-dependent properties of D on service life prediction and the random properties of the age factor still need to be further studied. Moreover, investigation of existing RC structures usually involves a durability assessment of the structure, so choosing a reasonable prediction model is also very important. Therefore, this paper mainly focuses on existing coastal RC structures located in the Beibu Gulf. Based on the test results, the distribution of chloride concentration under different exposure conditions (atmospheric, tidal and splash zones) is discussed, and the time-dependent properties of D are also studied. Furthermore, a reasonable model was obtained through comparative analysis of three calculation models to predict the initial time of reinforcement corrosion. 2. Experimental investigations 2.1. Structure descriptions A field investigation was conducted in the Beibu Gulf, Guangxi province, China. The test objects including Fangcheng, Qinzhou and Tieshan ports, are shown in Fig. 1. 13# dock of Fangcheng port is a gravity type thin-walled cylindrical caisson structure, built in November 2005 and put into operation in 2007. 10# dock of Qinzhou port is 50,000 t docks and belongs to the category of large cylinder gravity structure. 1# dock of Tieshan port, built in August 2009, is 150,000 t bulk cargo docks, and belongs to the category of gravity caisson structure. The concrete mix ratio of these three docks is all the same, with a water-cement ratio of 0.40 and a concrete strength grade of C40. Additionally, the tested concrete docks are located in a tropical area; the average relative temperature is 22.4 °C and the average relative humidity is 80%. By the time of the field investigation, these three concrete docks had been

Fig. 1. Field condition (Qinzhou port).

407

exposed to the chloride environment for 80, 62 and 35 months, respectively. 2.2. Sampling and testing In the in-situ test, the concrete powders were taken from the external wall of the structures. In Fangcheng and Qinzhou ports, 36 concrete samples were taken from the atmospheric, tidal and splash zones. However, since the Tieshan port had the shortest service time, only 12 concrete samples were collected. Therefore, a total of 48 concrete samples were obtained for in-situ tests. Since the concrete cover depth is 60 mm, the drilling depth was set at 56 mm, divided into eight sections, with each section of 7 mm. Collected concrete powders were sieved with a 0.16 mm square hole sieve to remove the coarse aggregates. Then, the concrete powders were placed into the aluminium specimen boxes, and put into 105 ± 5 °C ovens for drying for two hours. After that, the aluminium specimen boxes went into the dryer to cool to room temperature, as shown in Fig. 2. Eventually, the chloride concentration was determined by RCT (Rapid Chloride Concentration Tester). 3. Results and discussion 3.1. Chloride distribution For coastal RC structures, the exposure conditions can generally be divided into four zones (i.e. submerge, tidal, splash and atmospheric zones) [6,12]. The tidal and splash zones are subjected to long-term chloride corrosion with drying-wetting cycles. Due to the coupling effect of convection and diffusion [13,14], the exposure conditions in these two zones are particularly harsh. Therefore, the tidal and splash zones have been a busy area of research for the durability assessment of coastal RC structures. Fig. 3 shows the mean chloride concentration of the tested concrete. It can be observed in Fig. 3 that the splash zone has the highest chloride concentration, followed by the tidal zone and the lowest concentration is present in the atmospheric zone. Taking Fangcheng dock as an example, the mean chloride concentration at a depth of 3.5 mm in the splash, tidal and atmospheric zones is 0.60%, 0.27% and 0.10% by the weight of concrete, respectively. If the mean chloride concentration in the atmospheric zone was set as a reference, then the tidal and splash zones are 2.7 and 6.0 times higher, respectively. It can clearly be seen that, at a depth of 50 mm, the mean chloride concentrations in the splash and tidal zones are 0.02% and 0.07% by the weight of concrete, respectively.

Fig. 2. Concrete powders.

408

L. Wu et al. / Construction and Building Materials 152 (2017) 406–413

0.4

Chloride concentration /%

Chloride concentration /%

0.7 Splash Tidal Atmospheric

0.6 0.5 0.4 0.3 0.2 0.1 0 0

10

20

30

40

50

Splash Tidal Atmospheric

0.3

0.2

0.1

0 0

60

10

20

Chloride concentration /%

30

40

50

60

Depth /mm (b) Qinzhou port

Depth /mm (a) Fangcheng port 0.25

Splash Tidal Atmospheric

0.20 0.15 0.10 0.05 0 0

10

20

30

40

50

60

Depth /mm (c) Tieshan port Fig. 3. Chloride profiles.

However, in concrete powders collected from the atmospheric zone, it is virtually impossible to detect the presence of chloride. It is clear that chloride ions from salt solution ingress into concrete can only occur through the connected concrete pore network [8,15]. Therefore, the saturation level of concrete pore structures will greatly affect the chloride ions transport. In the atmospheric zone, the surface layer of concrete stays in the unsaturated condition, so the moisture in the concrete voids is insufficient, leading to reduced connectivity in the concrete pore network and impeded chloride transport [6,17]. In addition, the source of chloride ions in the atmospheric zone is the small chlorate crystals carried by the sea breeze [6]. A lower chloride concentration gradient will further limit chloride ion ingress into concrete. As a result, the chloride concentration in the atmospheric zone remains relatively low. Compared to the atmospheric zone, the concrete in the splash and tidal zones is exposed to the seawater continuously, and the seawater is the source of chloride ions. For saturated concrete, diffusion is the main mechanism for chloride ingress, while for unsaturated concrete, it mainly depends on capillary adsorption [16,17]. During the dry state, the surface layer of concrete stays in a nonsaturated condition. During the submerged state, chloride ions migrate rapidly into the concrete through capillary adsorption and form a high chloride concentration. Then, chloride transport in concrete will be accelerated by the interaction of capillary adsorption, diffusion and chloride concentration gradient. When the concrete once again enters the dry state, the moisture in the surface layer of concrete is evaporated, so the residual chloride ions crystallize, forming a high chloride concentration gradient. In the subsequent submerged state, the chloride crystals in the voids dissolve and widen the connected concrete pore networks at the same time, which also accelerates the chloride penetration. Therefore, as shown in Fig. 3, at the same depth, the chlorine con-

centration in concrete in the splash and tidal zones is much higher than in concrete in the atmospheric zone. Additionally, according to the pattern of tides, concrete in the tidal zone experiences two drying-wetting cycles a day, while in the splash zone, the concrete suffers repeated and irregular drying-wetting cycles. It has been reported that the chloride concentration increases with the time of drying-wetting cycles [17]. Hence, the chloride penetration in the splash zone is more effective; correspondingly, the measured chloride concentration in concrete in the splash zone is higher than in concrete in the tidal zone. It is worth pointing out that, in the dry-wet cycling zone, the surface layer of concrete will form a so-called convection zone [15]. However, in this work, owing to the relatively short exposure time and limitations of the test method, the measured chloride profiles in Fig. 3 could not exhibit the convective zone. The value of D and Cs can be determined by using Eq. (1) to fit the measured chloride profiles, the corresponding results are presented in Table 1. For OPC concrete, the value of D is mainly determined by the water-cement ratio, and the impact of exposure condition is not so obvious [4,7]. Therefore, as shown in Table 1, the difference caused by exposure condition on the value of D is not so noticeable as on the value of Cs. Taking Fangcheng dock as an example, in the splash, tidal and atmospheric zone, the mean value of Cs is 0.38%, 0.31% and 0.12% by the weight of concrete, respectively, and the mean value of D is 0.83, 0.75 and 0.70  1012 m2/s, respectively. Moreover, it can be observed in Eq. (1) that a higher value of Cs means a greater risk of reinforcement corrosion. Thus, it is reasonable to conclude that the splash zone is the most unfavorable exposure condition for the durability of coastal RC structures. In order to further understand the influence of exposure condition on the durability of RC structures, the values in Table 1 can be

409

L. Wu et al. / Construction and Building Materials 152 (2017) 406–413 Table 1 Apparent chloride diffusion coefficient and surface chloride concentration.

Splash Tidal Atmospheric

D/(1012 m2/s)

Cs/%

Fangcheng

Qinzhou

Tieshan

Fangcheng

Qinzhou

Tieshan

0.83 0.75 0.70

1.01 0.91 0.74

1.18 1.07 0.84

0.67 0.31 0.12

0.42 0.24 0.07

0.29 0.12 0.04

Table 2 Corrosion initiation time. Exposure condition

Fangcheng

Qinzhou

Tieshan

Splash Tidal Atmospheric

13.16 19.48 34.77

12.80 18.30 48.43

12.70 22.58 106.73

substituted into Eq. (1) to match the corrosion initiation time. For demonstration purposes, the critical chloride concentration Ccr was set at 0.015% by the weight of concrete, and the concrete cover depth d was set at 60 mm. The corresponding results are listed in Table 2. Taking the Tieshan port as an example, the corrosion initiation time in the splash, tidal and atmospheric zones is 12.70, 22.58 and 106.73 a, respectively. This result further proves that the splash zone is the harshest exposure condition in terms of durability of coastal RC structures. 3.2. Time dependency of chloride transport Previous studies have reported that, under the same exposure conditions, chloride transport in OPC concrete was mainly affected by the concrete mix ratio [4,7]. In this work, the investigated concrete samples are all located in the Beibu Gulf, and have the same concrete mix ratio. Therefore, it can be inferred from Table 1 that D is a time-varying parameter, and decreases gradually with the increase in exposure time. As Table 1 illustrates, after exposure to the marine environment for 35, 62 and 80 months, the mean value of D is 1.18, 1.01 and 0.83  1012 m2/s, respectively. The time-dependent properties of D can be explained by the cement hydration that improves the pore structures of the concrete with the increase of exposure time [8]. In addition, it can be noted in Table 1 that Cs is also a timevarying parameter; the value of Cs gradually accumulates with the increase of exposure time. After exposure to the marine environment for 35 and 80 months, the mean value of Cs increased from 0.29% to 0.67% by the weight of concrete. However, Song et al. [4] and Saassouh et al. [7] suggested that the timedependent properties of the Cs will gradually weaken with the increase of exposure time, and reach a certain level, 5–15 a. However, the service lifetime of RC structures is generally far longer than 15 a. Therefore, the time-dependent properties of Cs can be ignored in durability evaluation. The time-dependent properties of D can be described as follows [8,18]:

 m t ref DðtÞ ¼ Dref t

ð2Þ

where m is the age factor, Dref and D(t) is apparent chloride diffusion coefficient at the exposure time of tref and t, respectively. Theoretically, after a log logarithm is implemented on the Eq. (2), we obtain:

m¼

lnðDðtÞÞ   t lnðDref Þ reft

ð3Þ

In the literature, D(t) is usually assumed to be a random variable obeying the lognormal distribution [7]. In this case, ln(D(t)) should be a random variable obeying normal distribution. Additionally, since Dref, t and tref are determined parameters, it can be deduced from Eq. (3) that m should be a normally distributed random variable. In practice, the value of the age factor m is usually obtained by using Eq. (2) to fit the tested value of D. The obtained 144 data samples were statistically analyzed, and the histogram is illustrated in Fig. 4. The age factor m is most suitable for normal distribution, with a mean value of 0.43, and standard deviation of 0.09, and the Kolmogorov-Smirnov test verifies this result. Moreover, this statistical distribution is in agreement with the results of the theoretical derivation discussed above, verifying the credibility of the data in this work further. According to Eqs. (1) and (2), in order to make sure that the value of D decreases with the exposure time, the age factor m should be limited to 0.0–1.0. When m > 1.0, the chloride concentration at a depth of x will slightly decrease with exposure time. In other words, the chloride ions will continue to overflow from the concrete with the exposure time. When m = 1.0, the C(x,t) will remain unchanged and the chloride ions will no longer ingress into the concrete. When m < 0, it means that the value of D will increase with the exposure time, which is totally inconsistent with the actual situation. For OPC, Gjørv [2] suggested that m can be taken as 0.40; According to the empirical formula listed in reference [17], the calculated value of m was also equated to 0.40. For OPC concrete under tidal and splash zones, the value recommended by DuraCrete 2000 was 0.37. The Life 365 model recommends a relatively low value of 0.20, while a relatively high value of 0.55 was recommended by LNEC E465 [17,19]. It is clear that the results of this work are in good agreement with previous research. The influence of exposure condition on age factor m was rarely mentioned in earlier works. The fitting results in this paper are as follows: the mean value of m in tidal and atmospheric zones is 0.36 and 0.19, respectively. It has been reported that the time-

25

20

Frequency

Exposure condition

15

10

5

0

0.3

0.4

0.5

Fig. 4. Histogram of the age factor m.

0.6

410

L. Wu et al. / Construction and Building Materials 152 (2017) 406–413

dependent properties of chloride transport in concrete are mainly affected by the hydration of cement [15]. After casting of about 28 d, the degree of hydration of the cement paste could reach up to 85–90%, and the remainder would be hydrated over the next decades [8]. Owing to the hydration, the volume of cement paste may expand more than two times. Consequently, the concrete pore networks become plugged. Hence, the impact of exposure conditions on age factor m can also be attributed to the different degree of cement hydration. Besides, the random characteristic of m is also rarely discussed in the literature. A normal distribution was chosen by Li et al. [20] in the durability design of the Hong Kong-ZhuhaiMacau sea-link project. Gjørv [2] suggested that if nothing else was known, the random parameters could be assumed to be normally distributed. In order to evaluate the influence of time-dependent properties of D on the durability assessment of RC structures, the deterministic model can once again be applied to compare the corrosion initiation time. According to Tang and Gulikers [21], Eq. (2) can be turned into

" 1m  1m #  m Dref tref t ex t ex D¼   1þ 1m t t t

ð4Þ

   x Cðx; tÞ ¼ C s 1  erf pffiffiffiffiffiffiffiffiffi ; D 2 Dt #1    "  m tref Uc 1 1 ð1  RHÞ4 ¼ Dref  1þ  exp  ð7Þ t R T ref T ð1  RHc Þ4 where Uc is activation energy during the chloride diffusion process (35 kJ/mol), R is gas constant (8.314 J/mol/K), Tref is the reference temperature (293 K), T is temperature (K), RH is relative humidity and RHc = 75%.

   x Cðx; tÞ ¼ C s 1  erf pffiffiffiffiffiffiffiffiffi 2 Dt  D ¼ D28

28 365  t

 m Dref t ref D¼  1m t

ð5Þ

Taken Eqs. (5) into (1), we obtain: 1 "  2 #1m 2  ð1  mÞ  d C cr 1 t¼  erf 1 4  Dref  tm Cs ref

ð6Þ

The increment of corrosion initiation time calculated by Eq. (6) are given in Table 3. As Table 3 shows, in the splash and tidal zones of Fangcheng port, and the splash zone of Qinzhou port, the calculated corrosion initiation times are reduced when the time-varying properties of D are taken into account, which is not in tally with the facts [22], while for the rest, the corrosion initiation time has increased by at least 1.46%. Tang and Gulikers [21] reported that when the value of m > 0.3, the simplified mathematics of diffusion in Eq. (4) may underestimate the corrosion initiation time. In this work, the value of m in splash and tidal zones is 0.43 and 0.36, respectively, both larger than 0.3. Therefore, the conflicting results listed in Table 3 may have been caused by oversimplification in Eq. (4). Nevertheless, in comparing Tables 1 and 3, it is apparent that the timedependent properties of D are indispensable for a more accurate corrosion initiation time prediction of RC structures. In addition, the reliability-based stochastic method may be more reasonable for corrosion initiation time prediction [18–23]. 3.3. Probabilistic corrosion initiation time prediction 3.3.1. Corrosion model Corrosion models, such as DuraCrete 2000 model [2], Life 365 model [24] and LNEC E465 model [19], are common in the literature, and are listed in Eqs. (7)–(9): Table 3 Increment of corrosion initiation time/%. Exposure condition

Fangcheng

Qinzhou

Tieshan

Splash Tidal Atmospheric

37.72 8.96 13.56

26.05 1.46 30.3

13.18 57.5 79.37

  exp

  Uc 1 1  R T ref T

whereD28 ¼ 1:0  10ð12:06þ2:4w=cÞ , Cs = 0.033t and Cs  0.5.

w/c

ð8Þ is

water-cement

ratio,

   x Cðx; tÞ ¼ C s 1  erf pffiffiffiffiffiffiffiffiffi 2 Dt 

where tex is the age of the concrete at the time of chloride exposure. Since t is usually larger than tex, i.e. t  tex, thereby Eq. (4) can be simplified as:

0:2

D ¼ D28

28 365  t

0:55  kD;c  kD;RH  kD;T ; C s

¼ 3:0%  2:5  ðw=cÞ  kvert  khor  ktemp

ð9Þ

where kD,c, kD,RH and kD,T, represent the impact of curing conditions, environmental humidity and environmental temperature on D, respectively; while kvert, khor and ktemp indicate the effects of the exposure condition, distance and environmental temperature on Cs [19]. Based on the measured data from the Beibu Gulf, the above three models are comparatively analyzed in order to achieve a more accurate corrosion initiation time. In addition, combined with the above findings, in the DuraCrete 2000 model, the age factor was set at 0.36 and 0.43 for tidal and splash zones, respectively. Figs. 5 and 6 show the calculated chloride profiles under tidal and splash zones. It can be seen from Fig. 5 that, compared with the Life 365 and the E465 LNEC models, the DuraCrete 2000 model can better reflect the pattern of chloride transport in concrete. Even though Life 365 model has taken into account the time-dependent properties of Cs, it underestimates the early accumulation rate of Cs at the same time. Besides, the age factor m in this model is relatively small, and the result in the value of D is overestimated. Consequently, the calculated chloride profiles failed to reflect the pattern of chloride transport in concrete in the Beibu Gulf. On the contrary, the LNEC E465 model not only overlooked the time-dependent properties of Cs but also overestimated the value of Cs, leading to the calculated chloride concentration being significantly higher than the measured one. However, in the case of the watercement ratio being about 0.4–0.5, then the calculated Cs by the LNEC E465 model would be around 0.40–0.50% by the weight of concrete, consistent with the value recommended by the Life 365 model. Hence, it seems that the LNEC E465 model can also be employed to predict the lifetime of RC structures. As noted above, the experiential formulas used in the Life 365 and LNEC E465 models are inadequate to represent the value of Cs for short-term concrete. Therefore, in Fig. 6, peak chloride concentrations from the measured data were introduced to represent the value of Cs. It followed from Fig. 6 that the DuraCrete 2000 model still exhibits better quality in characterizing the chloride transport in concrete. This may be attributed to the overestimation of the value of D in calculating formulas in the Life 365 and LNEC E465 models. Besides, the Life 365 and LNEC E465 models are most likely to be applied in a deterministic approach [20,24]. Further-

411

L. Wu et al. / Construction and Building Materials 152 (2017) 406–413

0.40 Tested DuraCrete 2000 Life 365 LNEC E465

0.30

Chloride concentration /%

Chloride concentration /%

0.40

0.20

0.10

0 0

10

20

30

40

50

Tested DuraCrete 2000 Life 365 LNEC E465

0.30

0.20

0.10

0 0

60

10

20

Depth /mm

30

40

50

60

Depth /mm

(a) Fangcheng port

(b) Qinzhou port

Chloride concentration /%

0.40 Tested DuraCrete 2000 Life 365 LNEC E465

0.30

0.20

0.10

0 0

10

20

30

40

50

60

Depth /mm

(c) Tieshan port Fig. 5. Chloride profiles (Tidal zone).

0.40

0.5 0.4

Chloride concentration /%

Tested DuraCrete 2000 Life 365 LNEC E465

0.3 0.2 0.1 0 0

10

20

30 40 Depth /mm

50

60

Tested DuraCrete 2000 Life 365 LNEC E465

0.30

0.20

0.10

0 0

10

20

(a) Fangcheng port

30 40 Depth /mm

(b) Qinzhou port

0.30

Chloride concentration /%

Chloride concentration /%

0.6

Tested DuraCrete 2000 Life 365 LNEC E465

0.25 0.20 0.15 0.10 0.05 0 0

10

20

30 40 Depth /mm

50

(c) Tieshan port Fig. 6. Chloride profiles (Splash zone).

60

50

60

412

L. Wu et al. / Construction and Building Materials 152 (2017) 406–413

more, Tang and Gulikers [21] stated that the vital contribution that the DuraCrete 2000 model has made is in its probabilistic approach to the durability design of RC structures. Therefore, in this work, it is reasonable to use the DuraCrete 2000 model to predict the initial time of reinforcement corrosion.

dependent properties of D were taken into account. It is apparent that taking no notice of the time-dependent properties of D will overestimate the risk of chloride corrosion, which had also been noted by Song et al. [22]. 4. Conclusions

3.3.2. Corrosion initiation time prediction According to the reliability theory, the limit state equation Z that refers to the chloride induced corrosion initiation time prediction can be expressed as follows [25]:

Z ¼ C cr  Cðx; tÞ

ð10Þ

Based on the in-situ test results, the concrete cover depth obeys normal distribution, with a mean value of 59.5 mm and a coefficient of variation of 0.07. Dref was assumed to obey lognormal distribution, with a mean value of 0.83  1012 m2/s and variation coefficient of 0.3. Correspondingly, the value of tref should be 80/12 years. As Eq. (1) shows, the higher value of Cs associates with higher risk of reinforcement corrosion. Therefore, compared with 0.36% by the weight of concrete recommended by the Duracrete 2000 model, a more conservative value of 0.67% by the weight of concrete was chosen as the mean value of Cs, and was also assumed to follow lognormal distribution with coefficient of variation of 0.2. As mentioned earlier, age factor m follows the normal distribution, and the mean value is 0.43 and coefficient of variation is 0.1. For existing RC structures, according to the work of Alonso and Sanchez [26], Ccr can be expressed as a lognormal distributed random parameter, with a mean value of 0.64% by the weight of concrete and a coefficient of variation of 0.5. Using the Monte Carlo method with 1,000,000 simulations, Fig. 7 gives the time-dependent reliability index curves of the tested concrete. The corrosion initiation time prediction of reinforcement was usually considered to belong to the category of serviceability limit states, with target reliability index bd of 1.5 [7]. It followed from Fig. 7 that corrosion initiation time is about 42 a. However, fib 2010 reported that corrosion initiation time prediction also belongs to the scope of durability limit state, and the recommended value for target reliability index bd is 1.3 [27]; then the corresponding corrosion initiation time should be 47.5 a. So in this case, the initial time of reinforcement corrosion should be around 42–47.5 a. Considering that repair work would further prolong the service lifetime, it is clear that the tested concrete docks can completely meet the design requirement of 50 a. Additionally, in the case of m = 0, as Fig. 7 vividly illustrates that the corresponding corrosion initiation time is less than half of that when the time-

m=0.43 4

m=0 βd=1.5

Reliability index

(1) At the same corrosion depth, such as 3.5 mm, the chloride concentration in the splash, tidal and atmospheric zones are 0.24%, 0.10% and 0.03% by the weight of concrete, respectively. The splash zone is the most unfavorable exposure condition for coastal RC structures. (2) The apparent chloride diffusion coefficient D decreases with exposure time. After exposure to the marine environment for 35, 62 and 80 months, the mean value of D is 1.18, 1.01 and 0.83  1012 m2/s, respectively. In the durability assessment of RC structures, the time-dependent properties of D should be taken into consideration. (3) The age factor m is a normally distributed random parameter. The mean value of m at atmospheric, tidal and splash zones is 0.19, 0.36 and 0.43, respectively. (4) Compared with the Life 365 model and the LNEC E465 model, the DuraCrete 2000 model can better reflect the pattern of chloride penetration in concrete in the Beibu Gulf. Based on the DuraCrete 2000 model, and given the target reliability index bd = 1.3–1.5, the corresponding corrosion initiation time is about 42–47.5 a.

Acknowledgements The authors gratefully acknowledge the financial support provided by the Natural Science Foundation of China (Project number: 51308419, 51178356 and 51108348). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.conbuildmat. 2017.07.016. References

5

βd=1.3

3

2

1

0 10

In this paper, a field investigation was carried out in the Beibu Gulf to study the time-dependent chloride penetration in existing concrete in a marine environment, and the following conclusions are obtained:

20

30

40

Time /a Fig. 7. Time-dependent reliability index.

50

[1] M.H. Tadayon, M. Shekarchi, M. Tadayon, Long-term field study of chloride ingress in concretes containing pozzolans exposed to severe marine tidal zone, Constr. Build. Mater. 123 (2016) 611–616. [2] O.E. Gjorv, Durability of concrete structures, Arabian J. Sci. Eng. 36 (2) (2011) 151–172. [3] X.M. Shi, N. Xie, K. Fortune, J. Gong, Durability of steel reinforced concrete in chloride environments: an overview, Constr. Build. Mater. 30 (2012) 125–138. [4] H.W. Song, C.H. Lee, K.Y. Ann, Factors influencing chloride transport in concrete structures exposed to marine environments, Cem. Concr. Compos. 30 (2) (2008) 113–121. [5] K.Y. Ann, J.H. Ahn, J.S. Ryou, The importance of chloride content at the concrete surface in assessing the time to corrosion of steel in concrete structures, Constr. Build. Mater. 23 (1) (2009) 239–245. [6] M. Valipour, F. Pargar, M. Shekarchi, S. Khani, M. Moradian, In situ study of chloride ingress in concretes containing natural zeolite, metakaolin and silica fume exposed to various exposure conditions in a harsh marine environment, Constr. Build. Mater. 46 (2013) 63–70. [7] B. Saassouh, Z. Lounis, Probabilistic modeling of chloride-induced corrosion in concrete structures using first- and second-order reliability methods, Cem. Concr. Compos. 34 (9) (2012) 1082–1093. [8] S.W. Pack, M.S. Jung, H.W. Song, S.H. Kim, K.Y. Ann, Prediction of time dependent chloride transport in concrete structures exposed to a marine environment, Cem. Concr. Res. 40 (2) (2010) 302–312.

L. Wu et al. / Construction and Building Materials 152 (2017) 406–413 [9] A. Petcherdchoo, Time dependent models of apparent diffusion coefficient and surface chloride for chloride transport in fly ash concrete, Constr. Build. Mater. 38 (2013) 497–507. [10] S. Muthulingam, B.N. Rao, Chloride binding and time-dependent surface chloride content models for fly ash concrete, Front. Struct. Civil Eng. 10 (1) (2016) 112–120. [11] H.L. Wang, J.G. Dai, X.Y. Sun, X.L. Zhang, Time-dependent and stressdependent chloride diffusivity of concrete subjected to sustained compressive loading, J. Mater. Civil Eng. 28 (8) (2016). [12] H.F. Zhang, W.P. Zhang, X.L. Gu, X.Y. Jin, N.G. Jin, Chloride penetration in concrete under marine atmospheric environment – analysis of the influencing factors, Struct. Infrastruct. Eng. 12 (11) (2016) 1428–1438. [13] Z.W. Yu, Y. Chen, P. Liu, W.L. Wang, Accelerated simulation of chloride ingress into concrete under drying-wetting alternation condition chloride environment, Constr. Build. Mater. 93 (2015) 205–213. [14] S.J.H. Meijers, Computational results of a model for chloride ingress in concrete including convection, drying-wetting cycles and carbonation, Mater. Struct. 38 (276) (2005) 145–154. [15] H.L. Chang, S. Mu, D.Q. Xie, P.G. Wang, Influence of pore structure and moisture distribution on chloride ‘‘maximum phenomenon” in surface layer of specimens exposed to cyclic drying-wetting condition, Constr. Build. Mater. 131 (2017) 16–30. [16] A.J.J. van der Zanden, A. Taher, T. Arends, Modelling of water and chloride transport in concrete during yearly wetting/drying cycles, Constr. Build. Mater. 81 (2015) 120–129. [17] C.H. Lu, Y. Gao, Z.W. Cui, R.G. Liu, Experimental analysis of chloride penetration into concrete subjected to drying-wetting cycles, J. Mater. Civil Eng. 27 (12) (2015).

413

[18] A. Attari, C. McNally, M.G. Richardson, A probabilistic assessment of the influence of age factor on the service life of concretes with limestone cement/ GGBS binders, Constr. Build. Mater. 111 (2016) 488–494. [19] P.F. Marques, A. Costa, F. Lanata, Service life of RC structures: chloride induced corrosion: prescriptive versus performance-based methodologies, Mater. Struct. 45 (1–2) (2012) 277–296. [20] K.F. Li, Q.W. Li, X.G. Zhou, Z.H. Fan, Durability design of the Hong Kong-ZhuhaiMacau Sea-Link project: principle and procedure, J. Bridge Eng. 20 (11) (2015). [21] L.P. Tang, J. Gulikers, On the mathematics of time-dependent apparent chloride diffusion coefficient in concrete, Cem. Concr. Res. 37 (4) (2007) 589–595. [22] H.W. Song, S.W. Pack, K.Y. Ann, Probabilistic assessment to predict the time to corrosion of steel in reinforced concrete tunnel box exposed to sea water, Constr. Build. Mater. 23 (10) (2009) 3270–3278. [23] E. Bastidas-Arteaga, P. Bressolette, A. Chateauneuf, M. Sanchez-Silva, Probabilistic lifetime assessment of RC structures under coupled corrosionfatigue deterioration processes, Struct. Saf. 31 (1) (2009) 84–96. [24] I.S. Oslakovic, D. Bjegovic, D. Mikulic, Evaluation of service life design models on concrete structures exposed to marine environment, Mater. Struct. 43 (10) (2010) 1397–1412. [25] H.P. Hong, Assessment of reliability of aging reinforced concrete structures, J. Struct. Eng.-ASCE 126 (12) (2000) 1458–1465. [26] M.C. Alonso, M. Sanchez, Analysis of the variability of chloride threshold values in the literature, Mater. Corros. 60 (8) (2009) 631–637. [27] M. Akiyama, D.M. Frangopol, M. Suzuki, Integration of the effects of airborne chlorides into reliability-based durability design of reinforced concrete structures in a marine environment, Struct. Infrastruct. Eng. 8 (2) (2012) 125–134.