Optimisation of Bridge Maintenance Programs in Short- and Medium-term Financial Planning
K Nkel, A Stadler, C Walther, PTV AG, DE
In the context of the development of a bridge management system by the German Federal Highway Research Institute (BASt) among others modules have been realized both for the optimisation of maintenance measures in short-term financial planning and for the development of case scenarios in middle- and long-term planning.
The analysis of the task of short-term optimisation of maintenance measures leads to so-called 'knapsack problems'. Yearly budgets are modelled as maximum knapsack weights, measures as objects, costs as weights and benefits as values. The optimisation's objective consists of the maximisation of the sum of the values of all selected objects (benefit of measures) on condition that the sum of the weights (costs) is smaller than the given weight limit (budget). In general, knapsack problems are NP-hard, so for realistic problem sizes they cannot be exactly solved, but for certain knapsack problems there are efficient approximation algorithms, which assure acceptable computation times and an adequate quality of the results.
The paper describes the modelling of the yearly measure planning problem as a multi-choice knapsack problem (MCKP) and its solution by way of an efficient dynamic programming algorithm. For each bridge both combined and mutually exclusive alternative measures are taken into account. The paper analyses different approaches to the computation of the benefit of measures.
The multi-year optimisation problem is solved by iterative solution of the yearly MCKP. Here, besides the simulation of the bridges' states according to the taken measures, additional constraints like minimum quality levels are monitored. Compliance with them is ensured by the a-priori computation of a maintenance program with minimum costs for each bridge. Measures belonging to these programs serve as 'default measures' and may be surpassed in the optimisation process.
The aim of the scenario module is the medium-term prediction/estimation of financial requirements and of the bridge states given different maintenance strategies. For this purpose, each bridge is simulated starting from its actual state taken from the German database 'SIB-Bauwerke'. The simulation considers the empirical aging behaviour of the category the building belongs to as well as foreseen maintenance measures which are planned based on general decision rules.
A graphical user interface (GUI) enables the user to set different threshold values for component states which trigger state-dependent and component-dependent maintenance measures. The computer program generates diagrams describing the development of the yearly costs and of the rating of the bridges' state.
Association for European Transport