Updating the prediction of chloride-induced corrosion in RC structures by considering cracks detected by a CNN


METADATA ONLY
Loading...

Date

2022

Publication Type

Other Conference Item

ETH Bibliography

yes

Citations

Altmetric
METADATA ONLY

Data

Rights / License

Abstract

Reinforced concrete (RC) structures are generally inspected visually at regular time intervals. Due to the continuous ageing of infrastructure, particularly in industrialised countries, there is an ever-increasing need for efficient and reliable condition assessments, which maximise the information gained during an inspection. This allows for improved planning of maintenance works. Recent developments in the field of machine learning, in particular in deep learning, have the potential to allow the analysisof a vast amount of data in a short time. In the case of chloride-induced corrosion, an a-priori depassivation probability can be estimated(e.g. by the fib model code for service life design in non-cracked concrete) andlater spatially updated following Bayes Theorem to approximate the remaining service life of the structure. The update process is composed of two steps: (i) Detection of cracks through image-based inspection by a pre-trained convolutional neural network (CNN) and (ii) modification of the chloride ingress model based on the inspection and detection of cracks. In such a case, the chloride ingress model is modified to account for cracked concrete locally. This contribution will describe the process of updating corrosion predictions based on image analysis in detail and apply it to an example.

Publication status

published

External links

Book title

ICCRRR 2022 Book of Extended Abstracts

Journal / series

Volume

Pages / Article No.

82 - 83

Publisher

University of Cape Town

Event

6th International Conference on Concrete Repair, Rehabilitation and Retrofitting (ICCRRR 2022)

Edition / version

Methods

Software

Geographic location

Date collected

Date created

Subject

Machine learning (artificial intelligence); Corrosion; inspection; Reinforced concrete; service life; CNNs; CRACKS (BUILDING MATERIALS)

Organisational unit

09593 - Angst, Ueli / Angst, Ueli check_circle

Notes

Extended abstract. Conference lecture held on October 3, 2022.

Funding

Related publications and datasets