PSI - Issue 63

Petr Konečný et al. / Procedia Structural Integrity 63 (2024) 21 – 26

25

For ordinary Portland cement (OPC) concrete, the time to onset of corrosion obtained using Equations (2) and (3) from (Tang & Gulikers, 2007) is 15.7 years with delayed exposure to chlorides while it is 6.3 years with immediate chloride exposure, closely matching the numerical model's result of 14.0 years and 6.4 years respectively. In contrast, the result from Equation (1) from (Maage et al., 1997) yields an initiation time of 17.8 years and 10.3 years respectively, which deviates more significantly from the numerical model. The ratio between the FEM results and Equations (2) & (3) is 1.12 and 0.98 respectively, while the ratio for the model from Equation (1) is higher at 1.27, and 1.61 respectively. For HPC mixture, the situation differs even more. The ratio between the FEM results and Equations (2) & (3) with considered delay in chloride exposure is lower at 1.06 (and 1.00 with immediate chloride exposure), indicating a closer agreement. However, the implementation of the model from Equation (1) shows a significant discrepancy from the reference FEM model, with a ratio of 5.62 with consideration of delayed chloride exposure. If the delay in chloride exposure is not considered, the ratio between results from Equation (1) and FEM model is 7.38. This indicates that the model from Equation (1) is unreliable. It is observed that there is a difference for cases with normal diffusion coefficient and significant error in case of low diffusion coefficients and high aging factors, typical for HPC concretes. Moreover, the model in Eq. (1) does not provide reliable results even when no delay in chloride ingress is considered as given in Eq. (4). This observation aligns with the criticisms mentioned in (Tang & Gulikers, 2007), highlighting the model's limitations in accurately predicting chloride ingress for such materials. Therefore, the analytical models (Eq. (2) & (3)) show better reliability and closer alignment with FEM results for both OPC and HPC concretes. However, studied implementation of the model from Equation (1) is not suitable for predicting chloride ingress not only in case of HPC due to error in predicting chloride ingress even without consideration of delay in chloride exposure and mainly due to its high sensitivity to low diffusion coefficients and prolonged aging factors.

Table 1. Diffusion and aging coefficients of concretes and theoretical initiation time

Exposure

t i (years)

Concrete type

n (-)

D 0 (×10 -12 m 2 /s)

Delay (years)

FEM

Eq. (2) & (3)

Eq. (1)

Eq. (4)

OPC

5.59

0.284

0 5 0 5

6.4

6.3

10.3 17.8

6.3

14.0

15.7

-

HPC

2.11

0.529

399.5 525.2

398.4 557.4

2950.0 2955.0

398.4

-

5. Conclusion The paper presents a comparison of the computation of time to onset of corrosion, considering delayed chloride exposure, utilizing two types of analytical formulas and a reference result evaluated by FEA. Two types of concrete were employed in this study: ordinary Portland cement with normal diffusion properties and high-performance concrete with very low diffusion properties and a high aging factor. This selection was made in order to evaluate the range of application of the tested formulas. The analysis revealed that the model based on Eq. (2) and Eq. (3) from the study (Tang & Gulikers, 2007) demonstrated promising results, closely aligning with the FEA reference results for both OPC and HPC concretes. Conversely, the model based on Eq. (1) from the paper (Maage et al., 1997) exhibited unreliable results for the case without consideration of the delay in chloride ingress for both types of concrete and extremely high value of onset of corrosion time in case of HPC concrete, which can be attributed to the low diffusion coefficient and high aging factor inherent to HPC. Therefore, this model is not of further interest. These findings suggest that while the model from Equations (2) and (3) is more robust, therefore further testing is necessary to fully validate this model including verification with laboratory data. Future studies should include varying the delay times to exposure time ratios and exploring other combinations of diffusion coefficients and aging factors.