Infrastructure Engineering - Research Publications

Permanent URI for this collection

http://hdl.handle.net/11343/347

Filters

2012 1
Reset filters

Settings
Statistics
Citations
Author: Duffield, Colin × Author: Thompson, Russell × End date: 2016 × Start date: 2010 × Type: Conference Paper ×

Search Results

Now showing 1 - 1 of 1
  • Item
    Thumbnail Image
    Road Networks Management under Uncertainty: A stochastic based model
    Mandiartha, I ; Duffield, C ; Thompson, R ; Mathew, J ; Ma, L ; Tan, A ; Weijnen, M ; Lee, J (SpringerLink, 2012)
    Current pavement management systems (PMS) adopted by the Road Authorities are often very complex and data intensive. Other challenges also faced by Road Authorities in managing road networks include budget constraints and the uncertainty associated in predicting the future performance of pavements. In addition, the emphasis in pavement management has shifted from reconstructing completely new roads towards preservation of existing networks. In many cases, existing PMS do not meet these requirements. Thus, an efficient model that is able to accommodate all of those challenges needs to be developed. This paper outlines the development of a stochastic based PMS that includes a performance prediction model using Markov chains and an optimization model based on Markov Decision Processes (MDP). Combinations of pavement preservation strategies and maintenance budget levels are applied as action criteria in contrast to other stochastic models. Despite the apparent influence of uncertainty in road pavement performance during their service live, stochastic models provide promising results for enhancing current PMS. By analysing historical data, the future behaviour of road pavements under different expenditure levels and combination of routine and periodic maintenance measures can be predicted. From an optimization point of view, the utilization of constrained MDP will potentially result in cost savings. This is due to the optimality principal of the model which is capable of finding a optimal multi-year maintenance policy through the direct inclusion of additional constraints into the optimization problem. Hence, the model considers constraints and incorporates relationships between historical maintenance actions and costs. This paper also presents a methodology for developing rationale for long-term maintenance policies by integrating stochastic based performance prediction and optimization models with the experience of Road Authorities in managing roads networks.