Optimal load rating-based inspection planning of corroded steel girders using Markov decision process. (October 2021)
- Record Type:
- Journal Article
- Title:
- Optimal load rating-based inspection planning of corroded steel girders using Markov decision process. (October 2021)
- Main Title:
- Optimal load rating-based inspection planning of corroded steel girders using Markov decision process
- Authors:
- Cheng, Minghui
Frangopol, Dan M. - Abstract:
- Abstract: Steel girder bridges are vulnerable to corrosion. To maintain their safety above a predefined target level, the load rating can be computed from the inspection results and guide the following maintenance actions. Optimizing inspection and maintenance based on load ratings has substantial practical and economic relevance. Load rating-based strategies can be categorized based on whether the inspection interval and replacement criteria are fixed or flexible. Existing studies focus on fixed inspection intervals throughout the service life. In general, their results are not optimal for inspection planning. To reduce life-cycle cost, aged steel girders may be inspected and repaired in an adaptive manner. To this end, a method based on Markov decision process (MDP) is proposed to compare the life-cycle cost of four load rating-based policies (i.e. uniform or adaptive non-uniform inspection interval, and fixed or adaptive replacement threshold). Load rating-based inspection planning is formulated as MDP and the optimal plans are obtained using dynamic programming. The conventional approach to discretize states cannot accurately approximate the non-stationary deterioration process, while state augmentation is successful in doing this but will increase computational cost. A comparison of two approaches is made to investigate their effects on life-cycle cost. A bridge girder under corrosion attack is used as an illustrative example. The results show that the load rating-basedAbstract: Steel girder bridges are vulnerable to corrosion. To maintain their safety above a predefined target level, the load rating can be computed from the inspection results and guide the following maintenance actions. Optimizing inspection and maintenance based on load ratings has substantial practical and economic relevance. Load rating-based strategies can be categorized based on whether the inspection interval and replacement criteria are fixed or flexible. Existing studies focus on fixed inspection intervals throughout the service life. In general, their results are not optimal for inspection planning. To reduce life-cycle cost, aged steel girders may be inspected and repaired in an adaptive manner. To this end, a method based on Markov decision process (MDP) is proposed to compare the life-cycle cost of four load rating-based policies (i.e. uniform or adaptive non-uniform inspection interval, and fixed or adaptive replacement threshold). Load rating-based inspection planning is formulated as MDP and the optimal plans are obtained using dynamic programming. The conventional approach to discretize states cannot accurately approximate the non-stationary deterioration process, while state augmentation is successful in doing this but will increase computational cost. A comparison of two approaches is made to investigate their effects on life-cycle cost. A bridge girder under corrosion attack is used as an illustrative example. The results show that the load rating-based plan with an adaptive non-uniform inspection interval and fixed replacement threshold obtained using the state augmentation technique can be near-optimal. … (more)
- Is Part Of:
- Probabilistic engineering mechanics. Volume 66(2021)
- Journal:
- Probabilistic engineering mechanics
- Issue:
- Volume 66(2021)
- Issue Display:
- Volume 66, Issue 2021 (2021)
- Year:
- 2021
- Volume:
- 66
- Issue:
- 2021
- Issue Sort Value:
- 2021-0066-2021-0000
- Page Start:
- Page End:
- Publication Date:
- 2021-10
- Subjects:
- Load rating -- Markov decision process -- Dynamic programming -- Life-cycle -- Inspection -- Maintenance -- Corrosion under uncertainty
Engineering -- Statistical methods -- Periodicals
Mechanics, Applied -- Statistical methods -- Periodicals
Probabilities -- Periodicals
Ingénierie -- Méthodes statistiques -- Périodiques
Mécanique appliquée -- Méthodes statistiques -- Périodiques
Probabilités -- Périodiques
620.100727 - Journal URLs:
- http://www.sciencedirect.com/science/journal/02668920 ↗
http://www.elsevier.com/journals ↗ - DOI:
- 10.1016/j.probengmech.2021.103160 ↗
- Languages:
- English
- ISSNs:
- 0266-8920
- Deposit Type:
- Legaldeposit
- View Content:
- Available online (eLD content is only available in our Reading Rooms) ↗
- Physical Locations:
- British Library DSC - 6617.209600
British Library DSC - BLDSS-3PM
British Library HMNTS - ELD Digital store - Ingest File:
- 22702.xml