Performance Prediction Modeling of Concrete Bridge Deck Condition Using an Optimized Approach

Chyad, Aqeed Mohsin; Abudayyeh, Osama · 2020 · Crossref

DOI: 10.32732/jcec.2020.9.3.127

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study addresses the need for accurate deterioration prediction models for concrete bridge decks to improve condition assessment and resource allocation in bridge management systems. While Markov chain methods are commonly used, the authors sought to determine if alternative statistical distributions or combined approaches could yield more reliable predictions. The research focuses on Michigan concrete bridge decks, utilizing 25 years of inspection data (1992–2016) from the National Bridge Inventory (NBI) database. The primary objective was to develop an optimized model for predicting condition ratings and to analyze how specific factors—structure type, average daily traffic (ADT), and skew angle—affect deterioration rates. The methodology involved filtering NBI data to remove incomplete records and inspector errors, applying a three-year clipping threshold at the start and end of the dataset. The authors evaluated five statistical distribution functions (exponential, Weibull, lognormal, normal, and gamma) using the Anderson-Darling test to identify the best fit for Time in Condition Rating (TICR) data. They also developed a Markov chain model using transition probability matrices for condition ratings 9 through 4. To optimize predictions, the authors created a nonlinear regression model that combined the best-performing segments of the lognormal and Markov models. Error rates were calculated by comparing predicted condition ratings against original data to assess model accuracy. The results indicated that the lognormal distribution provided the best statistical fit for the data. However, comparative analysis revealed that the lognormal model had lower error rates for high condition ratings (9 and 8), while the Markov chain model performed better for lower ratings (7 through 4). The combined nonlinear regression model achieved the lowest average error rate (7.37%) across all condition ratings, outperforming both individual models. The study also identified significant impacts from external factors: steel superstructures deteriorated faster than concrete or prestressed types between ratings 8 and 4. Additionally, bridges with ADT exceeding 4,000 vehicles/day and skew angles greater than 30° exhibited faster deterioration rates compared to those with lower traffic or skew angles. Most new Michigan concrete bridge decks were found to take at least 40 years to deteriorate from a rating of 9 to 3. The significance of this work lies in the development of a hybrid nonlinear regression model that offers superior predictive accuracy for concrete bridge deck conditions compared to traditional single-method approaches. By integrating the strengths of lognormal and Markov models, the study provides transportation agencies with a more reliable tool for estimating remaining service life and planning maintenance. Furthermore, the identification of specific deterioration drivers, such as high traffic volume and structural type, enables more targeted allocation of limited funds for rehabilitation and reconstruction, ultimately enhancing the safety and longevity of the transportation network.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• incidence prevalence