One of the key components of a bridge management system (BMS) is the deterioration model, whose accuracy will determine the quality of future maintenance, rehabilitation and replacement (MR&R) decisions. The current state of the art for deterioration models in commercial BMS is the application of Markov chains to determine the probability of transitioning between condition states (CS). However, in recent years research has moved to looking at alternative approaches due to the assumption of a constant bridge population and stationary transition probabilities. This can compromise the efficacy of the deterioration model for predicting deterioration of complex aging bridge structures. This research focuses on the application of survival analysis where the survival time is the time spent in each CS and the ‘failure’ is the transition of the bridge condition to a worse state. Bridge condition states are measured on an ordinal or numerical scale, common examples range from 1-4 or 0-100 for the UK standardised bridge condition index (BCI). This paper presents the application of survival analysis to establish environmental controls and explores the comprehensive assessment of the utilisation of the Cox Proportional Hazards (PH) model. Expanding on the authors previous research which applied survival techniques to identify bridge performance indicators, this research uses live bridge condition data from Northern Ireland, based on over 6000 bridges which form the strategic and regional road network.
|Title of host publication||22nd European Young Statisticians Meeting: Proceedings|
|Number of pages||5|
|Publication status||Published - 10 Sep 2021|
- survival analysis
- Bridge Management Systems (BMS)
- deterioration modelling
- Markov chains