Estimating corrosion levels along confined steel bars in concrete using surface crack measurements and mesoscale simulations guided by model predictive control
Introduction
For many reinforced concrete (RC) structures, maintenance is a necessary task. During the serviceable life, periodic monitoring of deterioration is essential for residual performance estimation [[1], [2], [3]]. Among the various potential forms of deterioration, chloride-induced corrosion of the reinforcement is dominant [[4], [5], [6]]. Corrosion produces stresses that may result in cracking of the concrete. Initial surface cracks have been observed at low corrosion levels of micrometer loss of bar radius [7]. If cracking propagates until the cover concrete delaminates, spalling may occur [8]. Penetration of chloride ions to the bar surface depassivates the protective oxide layer on the bar and thus enables corrosion. The corrosion products are more voluminous than the original steel, causing radial stresses in the surrounding concrete [7,8].
Corrosion and related forms of damage are internal phenomena that alter the internal stress flow, possibly changing the overall behavior such as the failure mode [9,10], seismic performance [11] and flexural and shear capacity [12]. Hence, reduction in bar cross-section area and changes in bond performance are important parameters within quantitative analyses that support maintenance decisions. Previous attempts to estimate the influences of corrosion include probability based approaches such as the time-dependent reliability method for risk-cost optimization [13], using the concept of imprecise reliability for safety assessment [14] and durability monitoring [15]. The accuracy of these approaches is limited, and a more direct method is required. Non-destructive techniques to estimate corrosion include the half-cell potential method [16,17], the polarization resistance method [18] and thermography [19]. Others include the measurement of electrochemical potential [20], the Wenner four-electrode set up [21], direct wave radar [22], Fiber Bragg Grating [23], and infrared imaging [24]. In these techniques, degree of deterioration is calculated by a probabilistic approach or using instruments to measure corrosion. Mathematical approaches are approximate, while the use of heavy instruments may be impractical and difficult in the case of large infrastructure [25].
Numerical modeling is another useful tool for understanding concrete behavior and has been used to understand the effect of corrosion in concrete. Many studies have used the finite element method (FEM) to model non-uniform corrosion and study its effects on, for example, bond strength reduction, cover cracking and stress redistribution [[26], [27], [28], [29], [30]]. Other studies have used discrete analysis to accurately model crack propagation and damage in concrete [31,32]. Discrete analysis by the rigid-body-spring model (RBSM) has proven effective in simulating local cracking behavior [33,34], mesoscale behavior and fracturing [35,36] and corrosion [37,38] in reinforced concrete structures. The RBSM proposed by Kawai et al. [39] is one such discrete analysis technique, in which the simulation system developed by Nagai et al. [40,41] is used as a basis of a study to simulate the failure of concrete. The RBSM model can effectively simulate localized cracking in concrete [33,34,36,42]. The discretization of concrete into rigid bodies allows for realistic, discrete-like representation of local crack formation and propagation. Potential bias in cracking direction is reduced by generating random particle geometries. The applicability of RBSM to corrosion has been studied, leading to success in analyzing bond degradation [34,37,38,[43], [44], [45]]. Rebar shape can be accurately modeled in RBSM to account for the interlocking effect of a deformed bar. Therefore, it can be concluded that the damage and reduction in performance could be estimated if non-uniform corrosion data is available.
In practice, however, the corrosion of a rebar is nonuniform and occurs inside concrete, so it cannot be directly measured without removing the cover concrete. The only visible information is surface cracking and any effluence of corrosion products from such cracking. To avoid extensive cover removal, the internal non-uniform corrosion distribution therefore needs to be estimated from the observable surface crack distribution. The model predictive control-rigid body spring model (MPC-RBSM) [46] uses this surface information, which can be readily collected from a corroded structure, to estimate internal corrosion. Model predictive control (MPC) is a process control technique in which inputs are optimized to generate controlled outputs from the model. MPC has been adopted in various fields to produce target outputs [[47], [48], [49], [50], [51]]. In this study, MPC is used to control internal expansion within the RBSM representation of the reinforced concrete so as to produce cracking that is similar to that experimentally observed. The resultant expansion can then be used to estimate degree of corrosion using a linear relationship. For small crack openings and corrosion levels this linear relationship appears to be valid [[52], [53], [54]]. For larger crack openings or higher levels of corrosion, however, the relationship may exhibit nonlinearity due to significant leakage of corrosion products through the crack network, which can lead to discrepancies in the estimation results. In a previous study, the developed MPC-RBSM system successfully simulated the crack widths and damage patterns observed during laboratory testing of steel bar corrosion within concrete. However, due to the large crack width of 1.4 mm, and the presumed associated effect of corrosion product leakage, the estimated corrosion distribution exhibited differences with the experimental results [46]. Furthermore, that earlier study was limited to one specimen type, a single instance of localized corrosion along the bar length, and no lateral confinement (due to, for example, the presence of stirrups).
Since most RC structures contain multidirectional reinforcement, the practical application of MPC-RBSM technique demands the simulation of specimens with transverse confining reinforcement. In the earlier study, only longitudinal reinforcement was considered. Stirrup confinement significantly affects corrosion cracking behavior [55]. Tensile strain resulting from corrosion is borne by the stirrups and delays the propagation of corrosion cracks [56]. However, the threshold degree of corrosion required to initiate corrosion cracking is not affected by the presence of lateral confinement [54]. This delay may result in underestimation of corrosion during visual inspections and hence result in overestimates of residual performance. Mesoscale discrete analysis can directly accommodate these interactions and therefore, in this study, specimens with stirrup confinement are tested.
The research presented herein extends the previous study in several significant ways. Various specimen types are considered, providing a more comprehensive assessment of the MPC-RBSM capabilities for estimating corrosion degrees. In particular, some of the specimens contain multiple instances of localized corrosion, as would occur in practice. Some specimens include stirrup reinforcement to evaluate influences of the additional confinement on the corrosion distribution and its estimation. The specimen designs led to smaller crack openings, which largely avoids the uncertainties regarding the leakage of corrosion products through the crack networks. All simulation results are validated through comparisons with corresponding laboratory tests based on an accelerated corrosion setup. Crack widths for these specimens are recorded and used as targets for the MPC-RBSM system to estimate the distribution of corrosion along the bar lengths. Comparisons between the physically measured and estimated corrosion levels, and damage patterns, reveal that the MPC-RBSM system is accurate for the cases considered. With additional modifications, the MPC-RBSM system is a potentially viable means for the nondestructive evaluation of corrosion levels in structural concrete.
Section snippets
Rigid body spring model (RBSM)
The RBSM was proposed by Kawai et al. [39] and the three-dimensional simulation system for concrete in which aggregate and mortar are modeled as different materials was developed by Nagai et al. [40,41]. This study uses a modification of this system developed by Eddy et al. [33,57], in which the rebars are included in the RBSM. The concrete (mortar and aggregate) is modeled as one material and the rebar as a separate material. In this simulation system, a three-dimensional concrete model is
MPC-RBSM simulation model
Corrosion is simulated by adding expansion to the interface springs. The result is damage and cracking. This process cannot be reversed, so internal expansion cannot be directly estimated from surface cracks. This problem of reverse estimation is solved by integrating model predictive control (MPC) into the CEM-RBSM [46]. MPC is a process control technique used to optimize input(s) so as to produce a specific desired output. MPC has been employed to solve a variety of problems [47] such as
Experimental program
To test the applicability of the MPC-RBSM program to corroded specimens with varying crack profiles as well as different kinds of reinforcement arrangement, an experimental program was set up as explained in this section.
Simulation model
The experimental specimens are modeled using the MPC-RBSM system described in section 2 Numerical model, 3 MPC-RBSM simulation model in order to estimate the corrosion state from the experimentally observed surface cracking pattern. The specimen models, as shown in Fig. 13, have the exact dimensions and details of the experimental specimens, including the 30 mm clear cover over the rebars. Table 4 gives the numbers of rigid bodies in each of the specimens. The models for specimens 1R-1C and
Conclusions
- 1.
The efficiency of the developed MPC-RBSM program has been tested to confirm the applicability to the specimens with smaller crack widths, multiple peaks in the crack width distributions and specimens with confinement by lateral reinforcement. The system simulates the experimental crack widths with three-dimensional mesoscale failure simulation using a discrete analysis model. In all cases, the system provides the estimation of non-uniform corrosion distribution along the bar.
- 2.
For experimental
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This work was supported by JSPS KAKENHI Grant Numbers JP18KK0399, JP19H02210.
References (60)
- et al.
Effects of impressed current density on corrosion induced cracking of concrete cover
Construct. Build. Mater.
(2019) - et al.
Load carrying capacity of concrete structures with corroded reinforcement
Construct. Build. Mater.
(1997) Effect of degree of corrosion on the properties of reinforcing steel bars
Construct. Build. Mater.
(2001)Quantitative estimation of rebar corrosion in reinforced concrete by thermography
- et al.
Using radar direct wave for concrete condition assessment: correlation with electrical resistivity
J. Appl. Geophys.
(2007) - et al.
Finite element modeling of concrete cover cracking due to non-uniform steel corrosion
Eng. Fract. Mech.
(2015) - et al.
Effects of non-uniform corrosion on the cracking and service life of reinforced concrete structures
Cement Concr. Res.
(2010) - et al.
Comparison of uniform and non-uniform corrosion induced damage in reinforced concrete based on a Gaussian description of the corrosion layer
Corrosion Sci.
(2011) - et al.
Modeling pull-out resistance of corroded reinforcement in concrete: coupled three-dimensional finite element model
Cement Concr. Compos.
(2014) - et al.
Lattice discrete particle model (LDPM) for failure behavior of concrete. I: theory
Cement Concr. Compos.
(2011)
Lattice discrete particle model (LDPM) for failure behavior of concrete. II: calibration and validation
Cement Concr. Compos.
Numerical simulation of beam-column knee joints with mechanical anchorages by 3D rigid body spring model
Eng. Struct.
Fracture analyses using spring networks with random geometry
Eng. Fract. Mech.
Crack patterns of concrete with a single rebar subjected to non-uniform and localized corrosion
Construct. Build. Mater.
New discrete models and their application to seismic response analysis of structures
Nucl. Eng. Des.
Meso-scale modeling of non-uniformly corroded reinforced concrete using 3D discrete analysis
Eng. Struct.
Numerical simulation of bond degradation subjected to corrosion-induced crack by simplified rebar and interface model using RBSM
Construct. Build. Mater.
Model predictive control and its application in agriculture: a review
Comput. Electron. Agric.
Model predictive control: theory and practice-A survey
Automatica
Corrosion induced cracking: effect of different corrosion rates on crack width evolution
Construct. Build. Mater.
Bond strength evaluation of corroded steel bars via the surface crack width induced by reinforcement corrosion
Eng. Struct.
Assessment of the structural performance of corrosion-affected RC members based on experimental study and probabilistic modeling
Eng. Struct.
Sustainability of bridge maintenance
Proc. Inst. Civ. Eng. - Bridg. Eng.
Models for the economic evaluation of road maintenance
J. Transport Econ. Pol.
Bridge joint maintenance
Transport. Res. Rec.
Preventive maintenance of concrete bridge decks
Concr. Int.
Condition assessment of the deteriorated reinforced concrete bridge
Civ. Eng. Dimens.
TRRL, reinforcement corrosion and the durability of concrete bridges
Proc. Inst. Civ. Eng.
Cover cracking as a function of bar corrosion: Part I-Experimental test
Mater. Struct.
Bond deterioration of corroded steel bars in concrete
Struct. Infrastruct. Eng.
Cited by (9)
Reproducible estimations of internal corrosion distribution from surface cracks using MPC-RBSM
2023, Engineering Fracture MechanicsCorrosion damage detection in reinforced concrete using Rayleigh wave-based method
2023, Cement and Concrete CompositesProbabilistic similarity of non-uniform corrosion pattern between natural corrosion and accelerated experiment
2023, Construction and Building Materials3D mesoscale simulation of the influence of corrosion on loss of tension stiffening in reinforced concrete
2022, Construction and Building MaterialsCitation Excerpt :Hence, a unique relation has been used throughout the reinforcement-concrete interface. The applicability of the corrosion expansion model in simulating surface cracking patterns has already been verified [10,51,56]. The experimental study conducted by Kato et al. [26] is selected for simulation to investigate the effect of corrosion on the loss of tension stiffening effect with local cracking.
Morphology evolution and corrosion performance evaluation of the amorphous Ni-P coating prepared by picosecond LECD and modified by nanosecond laser remelting
2022, Optics and Laser TechnologyCitation Excerpt :Therefore, crystallization of amorphous Ni-P coating precipitates Ni and forms a stable phase of Ni-P compounds. The corrosion behavior of the Ni-P coating has been carried out, to investigate the influence of the crack density and size on the corrosion resistance [22-24]. As shown in Fig. 9, Tafel curves and EIS of S1-S6 were tested, and the specific values of Ecorr and icorr are shown in Table 4.
Investigating the effect of rebar corrosion order and arrangement on cracking behaviour of RC panels using 3D discrete analysis
2022, Construction and Building MaterialsCitation Excerpt :As per previous [17,42,43], in case of a higher applied current density, diffusion of corrosion products is lower and cracking initiation is earlier as compared to a lower current density for the same degree of corrosion. Hence, using this linking relationship, varying corrosion scenarios can be simulated using the same corrosion expansion model [17,30]. The use of CEM in the RBSM framework has been verified by Jiradilok et al. [29,30,44] in corrosion expansion and pull-out studies.