An investigation of methodologies to determine the most sustainable bridge work programs

Bridges are vital objects in public road networks. Since they deteriorate over time, maintenance is required to ensure that they continue to provide desired levels of performance. Over the past few decades an increasing number of computer-aided bridge management systems have been developed and used to manage large numbers of bridges around the world (Mirzaei, Adey, Klatter, & Thompson, 2014). One of the main functions of these systems is to assist bridge managers in the determination of candidate work programs for their bridges. It is beneficial if these candidate work programs are optimal, i.e., work programs that result in the lowest negative impacts. The systems that determine optimal work programs have principally defined the optimal work program as being the one that results in the lowest economic impacts incurred by the owner of the infrastructure, i.e., impacts directly related to the execution of interventions. Owners, however, are not the only stakeholders affected by bridge performance. There are also, for example, the users of the bridge who can be negatively affected by poor bridge performance. Keeping these other stakeholders in mind, optimal work programs for bridges should be the ones that result in the lowest overall negative impacts on all stakeholders. The reduction of all negative impacts to all stakeholders for now and the future, if not minimisation, is a concept that is increasingly being considered in many fields in the past few decades and is often referred to as sustainability. With the current state-of-the-art bridge management systems and the increasing desire for sustainability in the management of infrastructure, bridge managers should be provided with the tools and methodologies that can help them determine the most sustainable work programs for their bridges. The determination of the most sustainable work programs for bridges is challenging. One of the reasons, at least with regards to how optimal work programs are developed in the state-of-the-art systems, is that not all the impacts to all of the stakeholders can be attributed directly to the elements of the bridge taking into consideration their conditions (e.g., the direct impacts related to the amount of concrete that is required to repair a concrete abutment or the amount of paint that is needed to be applied to the steel girders). The impacts incurred by, for example, the user of the bridge, can only be indirectly related to the elements, as they depend on how all of the elements of the bridge work together as a whole (e.g., the additional travel time of the users due to inevitable speed reduction due to the excessive unevenness of the road surface), and on the decisions made related to the expected performance of the whole bridge (e.g., the impacts directly related to the traffic control efforts required to close a lane of a bridge during an intervention). The objectives of this thesis were: - to evaluate existing state-of-practice and state-of-the-art methodologies for the determination of optimal work programs for typical road bridges with respect to their ability to be used to determine the most sustainable work programs, and - to give guidance as to how such methodologies should be modified in the future so that they can determine the most sustainable work programs. The evaluation of the existing methodologies was done taking into consideration impact types that could be associated directly to the elements and impact types that could be related only directly to the bridge as a whole. The methodologies investigated included one that is considered to be representative of the current state-of- practice and two that appear to be the most promising from literature, i.e., the state-of-the-art. The state-of-practice methodology (SOP) is one that uses knowledge-based models and Markov chains to determine optimal intervention strategies for the elements of the bridge and then determines the interventions to be included in the work program based on the probable condition of the element at each time being considered and the interventions to be executed on the elements at that time according to the optimal intervention strategies. One of the state-of-the-art methodologies, the structural performance state methodology (SPM) is one that uses Markov chains to determine optimal intervention strategies for the bridge as a whole and then determine the interventions to be included in the work program based on the probable condition of elements at each time being considered and the interventions to be executed at that time according to the optimal intervention strategies. The other state-of-the-art methodology, the joint replenishment methodology (JRM) is one that uses inventory theory to determine optimal work programs for bridges by first fixing the groups of elements to have interventions at the same time and then determining the optimal regular intervals based on which to execute interventions on the bridge. The methodologies were investigated by using each of them to determine the optimal work program for a steel box girder bridge with a reinforced concrete bridge deck. The impact types considered were the same for all three methodologies, and included some that were attributable to the elements, and others that were attributable to the bridge. The former included the impacts on the owner during the executing interventions, e.g., labor and material costs. The latter included the impacts on the user and public, both during the execution of interventions and between the execution of interventions, including travel time costs, vehicle operation costs, accidents costs, sound emission costs, air pollution costs, and climate costs. The work programs developed using these three methodologies were compared with the work program developed using a proposed methodology which, although perhaps currently difficult to implement in an existing computerised management system, takes into consideration both element and bridge level impacts in appropriate ways. This is done by investigating all possible times between the execution of interventions on all combinations of groups of elements, and taking into consideration the contribution that each group of elements makes towards overall bridge performance. In the proposed methodology, the investigated time period is divided into the possible intervention intervals or intervention interval sets, and the relationship between the elements and the bridge performance is modeled using performance limit state functions for the relevant element groups. Through conducting the investigation, it was seen that each existing methodology has both strengths and weaknesses. The SOP is relatively fast but can only take into consideration bridge level impacts by either attributing them as a percentage increase to the element level impacts when determining the optimal intervention strategies or by attributing them on a bridge by bridge basis in the development of the work program. The latter of which makes it theoretically possible to take into consideration reduced impacts from the execution of interventions on multiple elements simultaneously but practically very difficult. The appropriate consideration of the values of impacts related to bridge performance between interventions is not possible. The SPM can appropriately model both element level and bridge level impacts by having average values of each impact type estimated for each structural performance state and then associated all of the element and bridge level impacts to each of these both during and between interventions in the determination of the optimal bridge intervention strategies, from which the optimal work programs are derived, but it is computationally intensive. The JRM can model both element level and bridge level impacts both during and between interventions by directly associating them to the elements and element groups. The direct link between the bridge level impacts and elements is, however, still necessary when the elements are to be treated individually. Although each model run is relatively quick, a large number of simulations would need to be done to appropriately capture the uncertainty associated with the future and, therefore, to determine the work programs that are truly optimal. The JRM investigated in this work, has the additional weakness that it is linear which makes it necessary to over- and underestimate the impacts. In the comparison of the investigated methodologies with the proposed methodology, it was found that all four produced relatively different work programs, but little variations in the total impacts. This is partially attributable to how the candidate work programs were selected in each methodology. It is, however, also partially attributable to the fact that there are many near-optimal work programs, at least for the example selected, in which there was relatively little traffic and a detour of sufficient capacity and short length so that there were not large impacts related to deviated traffic flow when an intervention was executed. Of the four methodologies compared, the proposed methodology was the one that resulted in the lowest overall impacts. In its entirety, the research showed that it is difficult to use state-of-practice methodologies to determine the most sustainable work programs. This is mainly due to their element-oriented focus. There are, however, some promising methodologies to be found in research of which, one is the investigated SPM. The research also showed that even more improvements are possible, at least from the conceptual point of view, by modelling the behaviour of the bridge as a whole more exactly. Coupled with such modelling efforts, however, are the substantial computing issues that first need to be resolved before methodologies (using this amount of detail) could be integrated into existing bridge management systems. Future research in this area should, therefore, be focused on the determination of ways to allow the required modelling but drastically faster. Such research will most likely include the investigation of simplified models to link the bridge performance states to the value of impacts and also the development of appropriate search algorithms to eliminate the need for exhaustive searches, but that can still find the most sustainable work programs.