Sewer deterioration modeling with condition data lacking historical records.

摘要: Accurate predictions of future conditions sewer systems are needed for efficient rehabilitation planning. For this purpose, a range deterioration models has been proposed which can be improved by calibration with observed condition data. However, if datasets lack historical records, requires combination and models, as the current state network reflects combined effect both processes. Otherwise, physical lifespans overestimated pipes in poor that were rehabilitated no longer represented dataset. We therefore propose model simple calibrated lacking information. use Bayesian inference parameter estimation due to limited information content data identifiability parameters. A sensitivity analysis gives an insight into model's robustness against uncertainty prior. The reveals results principally sensitive means priors specific parameters, should elicited care. importance sampling technique applied permitted implementation regional reasonable computational outlay. Application simulated real shows it effectively compensates bias induced Thus, novel approach makes possible calibrate pipe even when records lacking. Since at least some prior knowledge parameters is available, strength particularly evident case small datasets.