Bridge Annual Maintenance Prioritization under Uncertainty by Multiobjective Combinatorial Optimization

Abstract

Bridge managers are facing ever-increasing tasks of prioritizing limited budgets to cost-effectively maintain normal functionality of a huge inventory of deteriorating civil infrastructures such as highway bridges over the life cycle. A satisfactory maintenance planning scenario should meet managers' specified requirements for the optimum balance between whole-life costing and structural performance. This article presents a general computational procedure to prioritize on an annual basis maintenance efforts for deteriorating reinforced concrete bridge crossheads over a designated time horizon.Within each year, none or one of the available maintenance types with different performance improvement capabilities could be applied and the time of application for any maintenance intervention is considered to be uniformly distributed within a 1-year time interval. Effects of uncertainties associated with bridge crosshead deterioration processes with and without maintenance interventions are considered by means of Monte Carlo simulation to predict probabilistically structural performance and life-cycle maintenance cost. The resulting combinatorial optimization problem is automated by a multiobjective genetic algorithm. It produces a group of different sequences of annualized maintenance interventions that lead to optimized tradeoff among condition, safety, and life-cycle cost objectives. This enables bridge managers to determine a preferred annual maintenance prioritization solution by comparing different alternatives.

Pareto multi-criteria decision making

“…In fact, it happens that these thirteen designs are each almost identical to the Pareto-compromise design for at least one of the three objective criteria (e.g., from Table 3, to single-decimal accuracy, both the PEG-MCDM design and design 23 have Safety Index = 1.5). It is likely the PEG-MCDM design could have been directly integrated into the original Pareto design set, had Liu and Frangopol [17] employed but a slightly different data gradation scheme for the multi-criteria genetic algorithm used to find the entries of the criteria vectors f …”

“…Liu and Frangopol [17] initially completed Step 1 of the PEG-MCDM procedure, described in Section 3.3, by solving Eq. (32) using a multi-criteria genetic algorithm, to find m = 194 alternative Pareto-optimal designs of the maintenance-intervention protocol, represented by three 194 Â 1 criteria vectors f Having the vectors f Table 2.…”

“…That is, to prioritize on an annual basis, maintenance repair efforts for bridge crossheads. As devised by Liu and Frangopol [17], the design of the protocol is governed by n = 3 conflicting objective criteria concerning maintenance life-cycle cost, crosshead condition, and bridge safety. The life-cycle cost criterion involves minimization.…”

Observation-based Decision-making for Infrastructure

“…As another option, multi-objective optimization may be employed to tackle multiple objectives, delivering a set of compromising decision options along the so-called Pareto front. To this end, Liu and Frangopol (2005) employ multi-objective optimization to solve the combinatorial optimization problem of annual prioritization of maintenance efforts for deteriorating components of concrete bridges. Taflanidis & Beck (2008) proposed a robust stochastic design framework in the context of reliability analysis and related decision support, where probabilistic models of excitation uncertainties and system modeling uncertainties can be introduced.…”

Multiobjective Optimization for Risk-Based Maintenance and Life-Cycle Cost of Civil Infrastructure Systems

“…Five maintenance strategies are considered: replacement of expansion joints, silane, cathodic protection, minor concrete repair, and rebuilding [11,22]. Replacement of expansion joints is statistically the least costly.…”