Kwabena Ofosu Ph.D., P.E.
Engineer & Researcher in Infrastructure Management Systems
Transportation infrastructure assets include roads, bridges, culverts, tunnels, traffic control systems, and roadside appurtenances such as guardrails, lighting systems, signs and many others.
Transportation agencies employ infrastructure (or asset) management systems to monitor, manage, and maintain these facilities in a cost effective manner.
A deterioration model predicts the future condition or performance of an asset if no maintenance, rehabilitation, or improvements are conducted over a given time horizon. Deterioration models vary in sophistication from expert opinion, to simple linear regression models, to stochastic methods. In recent times, reliability engineering methods have been proposed. Reliability methods have the advantage of being capable of capturing the intrinsic uncertain nature of infrastructure deterioration, and its propagation throughout the service life of an asset.
To develop the reliability-based deterioration model, probability plotting can be used to fit the historic performance or condition data of an asset to a known statistical distribution. The data may include exact observations, as well as censored obervations of sojourn times an asset spends in a given condition/ performance level. Alternately the data may be synthesized from the expected stochastic deterioration model and the data points then fitted to a distribution.
The expected deterioration profile for a class of faclities for example interstate concrete bridges, can be generated from the reliability-based deterioration model by Monte Carlo simulation.
The objective of project-level analysis is to determine the cost-effective treatment from among a set of applicable treatments, for an asset based on its current condition or performance. The process involves modelling known and predicted future events associated with a specific treatment as a "cash flow" diagram. An engineering economic analysis is then performed to select the cost-effective treatment.
Road user cost is incurred by motorists any time they have to make a diversion, detour, or slow down as result of structural or functional failure of a facility such as a pavement, a height restricted bridge, or deficient culvert. User costs are also incurred due to construction activity. Asset management tools must take user cost into consideration when assessing impacts and decisions at the project-level and also at the infrastructre network-level.
The objective of network analysis is to assign resources such as the maintenance budget to a list of candidate projects that were identified at the project-level analysis phase. Traditionally agencies analyze individual classes independently, for example the bridge program may not consider events in the pavement program. The challenge now is to develop network-level analyses tools that consider all sub-systems of the total infrasturucture network as well as the relationships between them, simultaneously, in the network-level analysis. This optimization problem can be formulated in a number ways. An agency may have several goals or objectives for example improving network-wide performance and minimizing the percentage of assets in less than "fair" condition, as well constraints such as overall budget limitations and statutory requirements on fundng levels for a particular asset class. A scenario of this nature can be formulated as a multiobjective integer nonlinear program.
For even a small transportation agency the network-level problem formulation results in a computationally-intensive large-scale multi-dimensional optimization problem. Artificial intelligence methods, particularly the genetic algorithm (GA), have been demonstrated to be flexible and versatile to solve problems of this nature and magnitude.
The concept of the GA is based on Darwinian evolution through adaptation and natural selection. From an initial random population of solutions, successive generations of solutions "reproduce" to generate offspring that are better adapted to achieve the overall objectives of the formulation. By survival-of-the-fittest each generation propagates the genes of its fittest members whiles the less fit individuals die out. An evolutionary fitness curve monitors the progress of a "trait" of interest through the generations untill the overall objective(s) for the network is met or the procedure is terminated after a predefined number of generations.
The output of the network-level optimization is a list of projects to be implementred over a given planning horizon. It is important that projects within a predefined proximity are implemented within the same timeframe. This is called project integration. By so doing agencies can make significant savings and improve cost-effectiveness of their work programs. Also project-level integration has the potential of sgnificantly reducing user cost and other impacts to the motoring public.
As infrastructure ages asset management becomes a dominant view for maintaining a safe, economical, and sustainable transportation system. Neglect, under-funding, and lack of research has potentially catastrophic consequences as witnessed in Minneapolis in September, 2007. The on-going challenge is to advance asset management and to fund and implement the necessary measures to maintain and upgrade our transportation infrastructure.
© June 21, 2010. All Rights Reserved