H.S. Wong, A.R. Karimi, N.R. Buenfeld, Y.X. Zhao (Zhejiang University) & W.L. Jin (Zhejiang University)

Corrosion of reinforcing steel in concrete causes cracking and spalling of the concrete cover, loss of load bearing capacity and ultimately structural collapse. Cracking of the concrete cover is a critical limit state and this is often modelled as a two-stage process that consists of a) an initiation phase, defined as the time taken for corrosion to commence, and b) propagation phase, where the accumulation of corrosion products induces expansive stresses and damage. Until recently, most research has focused on the time up to corrosion initiation, while the propagation phase leading to failure remains poorly understood.

Fig. 1 Applying image analysis to measure degree of cracking and corrosion
Fig. 1 Applying image analysis to measure degree of cracking and corrosion

One important aspect of the propagation phase that lacks understanding is the amount of corrosion products that must form to cause damage. It is unlikely that all corrosion products contribute to cracking because some are soluble species that dissolve in the pore solution and migrate into adjacent cement paste away from the corroding sites. Thus, the ability of the pore structure in cement paste to act as repositories to accommodate the corrosion products may extend the period between onset of corrosion and crack development.

This project aims to enhance understanding of the effect of corrosion on the microstructure of reinforced concrete, in particular the nature, amount and distribution of the corrosion products, how they accumulate and cause damage. This is then used to develop a more accurate model for the propagation phase.

Fig. 2 BSE montage of samples with different degrees of corrosion, showing rust accumulating at the steel-concrete interface and migrating into the cement paste, cracks and air voids (S: Steel, MS: Millscale, RL: Rust layer, RP: Rust-filled paste, P: Unaffected Paste).
Fig. 2 BSE montage of samples with different degrees of corrosion, showing rust accumulating at the steel-concrete interface and migrating into the cement paste, cracks and air voids (S: Steel, MS: Millscale, RL: Rust layer, RP: Rust-filled paste, P: Unaffected Paste).

Reinforced concrete samples were prepared with Portland cement and blended cement (40% slag, 30% fly ash). The panels were cured for 14 days, then placed in an artificial environmental chamber and subjected to cyclic salt spray for 12 months to induce corrosion. A series of 8mm-thick cross-section slices was produced from each panel.

The slices were then epoxy-impregnated, polished and examined with optical microscopy, backscattered electron microscopy and energy dispersive X-ray microanalysis. Image analysis was used to measure the amount of corrosion products, their distribution at the steel-concrete interface and the degree of cracking (Figure 1).

The study found that corrosion products can migrate through the aggregate-paste interface as well as the ‘bulk’ cement paste. Corrosion products can be deposited in cracks, air voids, inner & outer hydration products, and relicts of reacted slag. A distinct boundary between the affected and unaffected paste can be seen, indicating the extent of the rust penetration (Figure 2).

Energy dispersive X-ray microanalyses showed that the cement paste filled with corrosion products has higher analysis totals, and Fe and O contents, but is depleted in Ca. The latter indicates dissolution of CH and/or decalcification of the C-S-H gel, which increases the local porosity and facilitates rust penetration. When pores become filled or blocked, subsequent corrosion products are forced to accumulate at the steel-concrete interface, inducing expansive pressure that leads to bond failure and cracking.

Fig. 3 Effect of corrosion degree on the amount of damage and area of the rust layer (CL) and rust penetrated paste (RP).
Fig. 3 Effect of corrosion degree on the amount of damage and area of the rust layer (CL) and rust penetrated paste (RP).

Image analysis showed that the distribution of corrosion products is non-uniform and that only a small amount of corrosion, approximately 100μm thick covering about 20% of the rebar perimeter, is needed to generate the first visible cover crack (~0.05mm). Once cracking has initiated, the rust preferentially deposits in cracks rather than pores in the cement paste. Hence, the extent of rust penetration into the cement paste does not increase much with corrosion (Figure 3).

Subsequently, the measured distribution of the corrosion products was used as an input into a FE model [2] to compare the damage induced from uniform and non-uniform corrosion (Figure 4). It was found that the distribution of corrosion products has a significant impact on the propagation phase and affects the build-up of expansive pressure around the rebar. More critically it governs the amount of corrosion products required to reach specified stages of damage, in some cases non-uniform corrosion requiring less loss of steel area and hence a shorter propagation period to violate set performance limit states.

Fig. 4 Finite-element modelling of cover cracking using the measured corrosion product distribution as input
Fig. 4 Finite-element modelling of cover cracking using the measured corrosion product distribution as input

Acknowledgements: This project received support from the EPSRC (EP/F002955/1) and the National Natural Science Foundation of China (50538070 & 50808157)

References

  • H.S. Wong, Y.X. Zhao, A.R. Karimi, N.R. Buenfeld, W.L. Jin (2010), On the penetration of corrosion products from reinforcing steel into concrete due to chloride-induced corrosion, Corrosion Science, 52, 2469-2480.
  • Y.X. Zhao, A.R. Karimi, H.S. Wong, B.Y. Hu, N.R. Buenfeld, W.L. Jin (2011), Comparison of uniform and non-uniform corrosion induced damage in reinforced concrete based on image analysis, Corrosion Science, 53 (9) 2803-2814.