Maintenance intervention optimisation for existing railway bridges

Transcription

1 Maintenance intervention optimisation for existing railway bridges D. Nielsen CQUniversity Australia, Rockhampton, Queensland, Australia CRC for Rail Innovation, Australia T. McSweeney CQUniversity Australia, Rockhampton, Queensland, Australia D. Raman CQUniversity Australia, Rockhampton, Queensland, Australia CRC for Rail Innovation, Australia ABSTRACT: Maintenance of aging infrastructure is a burden on many organisational budgets, and Life Cycle Cost Analysis (LCCA) is accepted as an effective method to reduce these costs. However, LCCA becomes extremely complex when applied at the individual bridge component level within a large bridge population while considering multiple maintenance strategies, varying deterioration rates and intervention intervals. To address this concern, a decision support model is presented that proposes maintenance actions for future interventions based on a selection of maintenance strategies and given organisational budget constraints. The remaining service life of surrounding components is considered when assessing the most appropriate maintenance actions. Three popular bridge management strategies are considered in this model, comprising refurbishment, upgrade and renewal strategies. The infrastructure manager may apply one of these three strategies at the bridge level to estimate future network budgets or align with existing budgets. The developed decision support application allows what-if scenarios to be generated to determine life cycle costs by changing life gained from maintenance actions, maintenance costs, maintenance strategy, discount rate, intervention trigger time and minimum intervention intervals. KEYWORDS: Life cycle costing, Bridges, Railway, Maintenance, Optimisation, Maintenance strategies 1 INTRODUCTION Railway bridges are essential structures in most rail networks and they play an important role in sustaining the national economy for many of the world s developed nations. As with other civil structures, bridges deteriorate with age and use, and would eventually become unsafe without maintenance. Maintaining structures within operationally safe parameters becomes increasingly more costly over time, placing an increasing burden on the budgets of infrastructure owners. The strain on maintenance budgets has encouraged many infrastructure owners to seek new ways to reduce bridge upkeep costs through innovation and improved maintenance practices. A popular method to reduce the total cost of bridge maintenance is to conduct Life Cycle Cost Analysis (LCCA) for maintenance activities (Hawk, 3, Frangopol, 11, Kong and Frangopol, 3). LCCA is conducted by comparing the cumulative cost of maintenance actions, including organisational and user costs, over the remaining life of the structure and then selecting the alternative that provides the lowest total cost (Hawk, 3). However, maintenance optimisation is required to balance trade-offs between operational performance and financial performance (Liu and Frangopol, 5). Operational performance forms a part of infrastructure management, including reliability, availability, maintenance, condition, safety, assessment (inspection and structural health monitoring), redundancy and robustness (Frangopol, 11). Future maintenance costs on a large population of aging bridge structures is often difficult to predict over longer planning horizons, particularly when considering individual component deterioration and maintenance alternatives. Furthermore, bridge structures consist of many components and the failure of one of these components may not lead to the failure of the structure (Yang et al., 4). Therefore, having a large number of bridge components with differing interactions on surrounding components, together with costing and deterioration factors, has traditionally made sub-bridge maintenance optimisation complex and unwieldy. One method to reduce bridge life time maintenance costs is to optimise maintenance actions and interventions over a set planning horizon. Kong and Frangopol (3) proposed optimisation of deterio-

2 rating structures using reliability index and maintenance interventions. Yang et al. (6), later expanded this research by including failure costs and maintenance strategies within the constraints of component condition states. However, these models do not consider the updated inspection results from routine bridge assessments and assumes preventative maintenance actions delay deterioration by a set value. In addition to these issues, these models assume the infrastructure owner has the budget to meet the maintenance schedule set by the optimisation output. The model presented in this paper aims to resolve such issues by considering a new life cost approach to maintenance intervention optimisation. Total maintenance costs over the lifespan of a structure may be higher than the structure s initial construction cost. As some bridge structures can function well beyond years life, the initial construction cost may seem small compared to current maintenance costs. Due to this reason, a discount rate is applied to future expenditure to show the changing value of money (Nielsen et al., 13b). The discount factor can have a large effect on delaying expensive upgrade maintenance actions, causing the model to select lower cost maintenance actions. Due to the vast geographical coverage of Australian railway networks, many rail organisations report that ad hoc repairing of bridge components is not financially efficient due to the long distances travelled between maintenance depots and structures (Nielsen et al., 13b). Instead, they prefer to visit structures every few years and conduct adequate maintenance to ensure suitable performance until the next major maintenance intervention. A few organisations with lower budgets accept they may need to revisit bridges multiple times per year to complete essential repairs as required. Bridge geographical location and distances from maintenance depots was required for inclusion into this project by partner railway organisations. Therefore, structure distance to depot costs and the minimum maintenance intervention period is also considered. The concerns of industry and the perceived need to further enhance previous research prompted the development of the Bridge Maintenance Optimisation-Major Works (BMO-MW) application described in this paper. 2 APPLICATION FRAMEWORK The BMO-MW application forms one application within a larger project, and inputs information from other applications within the framework as shown in Figure 1. The framework includes other applications that address assessment, costing, deterioration, minor and routine maintenance works and is designed to accommodate data from organisational asset management systems found in Australian railway organisations. This paper focuses on the maintenance optimisation process for major works on existing railway bridges in Australia. In the context of the BMO-MW application, major works are maintenance actions that increase the service life of components or structures by replacing, upgrading or reconstructing the component or structure. Structural assessment Maintenance cost estimation application Deterioration application Bridge maintenance optimisation: Minor works Bridge maintenance optimisation: Major works Short and long term maintenance plan Figure 1 - Bridge management application framework The BMO-MW application inputs the remaining service life of each component from the deterioration application. Remaining service life was selected as the dependant variable in this analysis due to the linear nature of this parameter and to reduce computations, thus increasing the speed of analysis. As the rate of deterioration of component performance changes due to time and other factors, the complex calculation of future component performance was assessed by the deterioration application. This application outputs the remaining service life for each component as a discrete number of years until the component reaches a set performance threshold. The deterioration application derives the remaining service life from performance measurement data including condition, safety, availability and reliability. The remaining service life is a critical input for the BMO-MW application. Therefore, to improve the accuracy of this parameter, structural assessment updates are sent to the deterioration model by way of visual condition data or structural health monitoring data. After processing, the deterioration model updates the BMO-MW application with the latest forecast of remaining service life. Major works optimisation is conducted on the maintenance strategy selected at the bridge level and the resulting maintenance actions and estimated costs are provided at the bridge component level. LCCA is conducted on maintenance alternatives (for upgrade and renewal strategies) and considering organisational performance criteria, provides a solution based on strategic objective functions. The output of the application is a major works maintenance plan for the network, detailing annual maintenance tasks and associated costs for a longer term planning horizon with interventions planned before components reach set performance thresholds. The BMO- MW application presented allows maintenance strategies to be changed at the bridge level to provide greater cost and timing flexibility for managing bridge networks.

3 Remaining Life (years) 3 MAINTENANCE STRATEGIES In this research, maintenance actions are categorised into three maintenance strategies. Two of these strategies (upgrade and renewal) are assigned an objective function for optimisation analysis. Hence the BMO-MW application calculates the solution of their optimisation objective functions by stepping through and comparing each major maintenance action at the subsequent intervention intervals, while the third strategy (refurbish) will always select the lowest cost option. The outputs of the three maintenance strategies have vastly different maintenance cost requirements over the life of the structure. These strategies may be interchanged at the bridge level to balance the budget of bridge infrastructure owners. The refurbish maintenance strategy usually has the lowest short-term cost of the major maintenance strategies. It generally requires components to be replaced with the same or similar component, design and material. The expected life gain from the new component is similar to the life gained from the replaced component and interventions may be required frequently. Consequently, this strategy may have a relatively higher cumulative maintenance cost over the life of the structure when compared to other strategies. The objective of this strategy is to maintain a safe structure at the lowest immediate cost with a disregard to possible higher future maintenance costs. Accordingly, this strategy does not use LCCA and is commonly employed by rail organisations with low maintenance budgets. The purpose of this strategy is to plan maintenance on short life railway lines or lines that are expected to close within a short time period. Another purpose of this strategy is to allow a comparison of costs between strategies. The second strategy, called the upgrade strategy, typically has a higher upfront cost than the refurbish strategy and aims to replace deteriorated components with longer life components. This may include upgrading a timber bridge deck to a steel or concrete deck. The objective of the upgrade strategy is to maximise the life of existing structures through refurbishment and upgrading maintenance actions to give the longest structural life at the lowest life cycle cost. The objective function for optimisation of the upgrade strategy is given in Equation 1. The application of this strategy initially minimises cumulative maintenance costs over three maintenance interventions and reduces the average condition of components with lower cost maintenance actions before commencing an upgrade of the structure. Every combination of refurbishment and upgrade maintenance activity is evaluated in the application until the greatest cost benefit over three intervention periods can be realised. life cycle maintenance costs Minimise ( ) (1) life gained by maintenance The third and last strategy is the renewal strategy, and its objective is to find the optimal time when the average remaining component value within a bridge structure is minimal. Due to a lack of data on original component construction costs, remaining value is calculated from replacement value multiplied by percentage of remaining component life. The upgrade strategy evaluates every combination of refurbishment and upgrading maintenance activities to find a future point in time where all bridge components will have the lowest average remaining value. The resulting future bridge replacement date is scheduled, allowing adequate time for reconstruction before set performance limits are reached by deteriorated components. The renewal strategy can have the highest upfront cost of the three strategies, yet it is often found that the model may select other strategies to reduce the average component cost before suggesting bridge reconstruction. This strategy aims to replace the entire structure with a new long life structure within three maintenance interventions. Maintenance improves the remaining life of components and more than one maintenance action is usually available to achieve this task. However, different maintenance actions incur different costs with varying improvement on the remaining life of the component. An example in Figure 2 shows the effect of maintenance on the components of a small single span bridge and Figure 3 shows the average remaining life of all the components Age of bridge (years) Comp. 1 Comp. 2 Comp. 3 Figure 2 - Effect of maintenance on component life Another maintenance strategy found within literature is called the Do Nothing strategy. This strategy is redundant within this model due to the nature of how the three strategies are employed to meet their objective functions. Each strategy calculates the next year of intervention based on the remaining life of all bridge components including repaired components from recent maintenance actions. The years between maintenance interventions become the years where no major maintenance is conducted, thus imitating the do nothing strategy.

4 Remaining Life (years) 1 Age of bridge (years) Average Figure 3 - Average remaining life of all bridge components As bridge components approach the end of their functional life, infrastructure owners are required to make a decision regarding which maintenance alternative would be best suited to meet organisational and social requirements placed on that structure. To assist in this decision, infrastructure owners can select the appropriate maintenance strategy suiting their maintenance budget. The three strategies proposed were deemed adequate by industry partners to meet the immediate needs of bridge management decision support. The model of applying maintenance strategies to individual bridges and calculating the remaining service life of each bridge component is computationally intensive and requires the various inputs from other applications within the framework. The advancement in processing speed of modern computers now enables decision makers to quickly analyse their entire sub-bridge inventory to determine the most effective course of action for strategic network maintenance while considering sub-bridge component interaction. A screenshot of the BMO- MW application is provided in Figure 4 showing future interventions of individual components with expected cost. 4 OPTIMISATION PROCESS As discussed earlier, the BMO-MW application has a number of inputs required for optimisation. The inputs are: 1. component life gained from each maintenance action; 2. maintenance action cost estimates; materials, plant and labour; 3. bridge configuration data; 4. remaining life of each bridge component without maintenance; 5. organisational structure performance thresholds; and 6. maintenance budget constraints. To assist infrastructure owners with the maintenance cost estimation of various maintenance strategies, this model augments the estimation method and techniques presented previously by Nielsen et al. (13b). The cost of the refurbish and upgrade maintenance strategies is estimated from previous material, labour, plant and user costs based on the number and types of replaced components. The cost of bridge renewal is estimated from the length of the gap spanned and the height of bridge. Factors are added for additional earthworks and if construction is required over water. The estimation accuracy for the refurbish and upgrade strategies is acceptable for this level of analysis, and the automated structure renewal estimation is adequate for most small, low Australian railway bridges less than m in length. To allow for estimation of complicated works and certain structural types, estimated maintenance values may be manually overwritten by the user with an estimate of greater accuracy. This is particularly relevant for estimates for the renewal strategy on larger Figure 4 Screenshot of BMO-MW application

5 Maintenance Strategy Optimisation Process bridges or bridges with complex construction requirements. Inspection and routine maintenance costs are included in the analysis to provide the total life cycle cost within the analysis period. Inspections are divided into three categories of general, detailed and special (engineering, underwater, NDT) (Nielsen et al., 13a) and estimation calculations and schedules are provided in previous publications (Nielsen et al., 13b, Nielsen et al., 12). Bridge configuration, remaining life and performance assessment data is collected as documented in publications of partner Universities (Aflatooni et al., 12, Aflatooni et al., 13b, Aflatooni et al., 13a). A distinction is made between the remaining component life and defects related to the component. The remaining component life after defect repair is accepted as the remaining component service life for analysis purposes. This may include the possible reduction or increase in remaining service life due to damage or the repair method respectively. Therefore, this model requires the collection of additional component and defect data before repair, plus component condition after repair. Categorisation of maintenance actions between major and minor works is provided in Table 1. Minor works addresses defect repairs and routine maintenance tasks. This includes temporary works to maintain structure functionality as the lead times to conduct major works can be quite long. Paint or surface protection is considered a sacrificial asset under this model, and repainting falls under the major works refurbishment maintenance when it reaches the end of its protective life, while spot painting is classified as a routine maintenance task. The inclusion of both major and minor works within the application framework has the added benefit of identifying components with minor works repair costs above replacement costs. These components are scheduled for replacement or upgrade instead of repair in the BMO-MW application, based on the maintenance strategy selected. Table 1 Maintenance actions included in each strategy Maintenance Actions Minor works Major works Strategy Options Clean/clear/remove Coat/spot paint/poison Strengthen/repair/adjust Replace with same/paint Upgrade components Construct new bridge Default minimum intervention timeframe (years) Refurbish X X X X 1 Upgrade X X X X X 5 Renew X X X X X X Maintenance actions available for each strategy option The final optimisation inputs relate to the organisational requirements and available bridge maintenance budgets. Organisational requirements include minimum intervention intervals, minimum allowable bridge/component performance levels, safety factors (for earlier intervention based on probabilistic deterioration methods) and the discount rate for LCCA. Due to the small number of components and limited maintenance actions per strategy, the optimal solution was found by comparing all possible maintenance actions within the strategy, as provided in Table 1. 5 OPTIMISATION PROCEDURE Major maintenance actions were determined over three intervention periods to remove the need to stipulate a planning horizon. As over 3% of Australian railway bridges are over years old (Nielsen et al., 11), the typical 75 or year planning horizon is no longer acceptable and many Maintenance Intervention 1 Maintenance Intervention 2 Maintenance Intervention 3 Bridge Components (1 to ) Maintenance Action (1-3) Bridge Components (1 to ) Maintenance Action (1-3) Bridge Components (1 to ) Maintenance Action (1-3) Step through every component at each action Find next maintenance intervention Step through every component at each action Find next maintenance intervention Step through every component at each action Record cost and intervention data for each element. No No No Are all actions considered on all defects? Yes Are all actions considered on all defects? Yes Are all actions considered on all defects? Yes Show best solution based on objective function Figure 5 BMO-MW application process flowchart

6 Remaining life (years) railway organisations wish to obtain longer structure life with minimal maintenance. The first step in the optimisation procedure was to calculate the first intervention date from the component with the shortest remaining life. Cost and life gain for each maintenance action is conducted on components reaching the end of their life within the intervention period plus the safety factor. Intervention dates are calculated by: Remaining service life minimum intervention interval safety factor (shown as intervention trigger in Figure 4). These results determine the components requiring maintenance and the first intervention date, at which point the process is reiterated until component s maintenance actions meeting the selected maintenance strategy s objective function is found. This computationally intensive process follows the flowchart provided in Figure 5. Appropriate maintenance actions (based on the selected maintenance strategy) are placed into a set of nested computational loops for components requiring maintenance at each intervention. Each iteration of the nested loops records the maintenance action with future cost (from discount rate) and expected component life after maintenance. The result is compared to the previous result and the result best meeting the objective function is saved for comparison in the next iteration. At the end of analysis, the result meeting the selected maintenance strategy s objective function is presented to the user via a graphical interface. The application provides the intervention cost, and average remaining life of components before and after maintenance interventions. The life cycle cost/year is provided to highlight the change in cost based on the selected maintenance actions. Other analysis outputs include maintenance action for each bridge component, year maintenance conducted, maintenance life gained and cost of maintenance. An estimated bridge life realised after three interventions is also provided. Minor works optimisation and scheduling are conducted before major works to ensure components with uneconomical minor works costs are identified and included into the major works program. Minor works are deemed uneconomical if the repair cost exceeds the refurbishment cost. Minor works scheduling is based on defect ratings applied during inspection. More information on defect management and their ratings related to this model can be sourced from the publication by Nielsen et al. (12). The infrastructure manager can select the minimum number of years between maintenance interventions and set the safety factor to an acceptable level of risk based on organisational requirements. 6 RESULTS AND DISCUSSION One bridge was selected for analysis and the results were compared with traditional maintenance planning methods to determine the practicality and effectiveness of the model. Each bridge component and optimisation inputs were entered into the application including maintenance action costs and subsequent life gained for each action. The analysis processing time for a small two span bridge with major components was under a second for the refurbish strategy and approximately 22 seconds each for the upgrade and renewal strategies. The processing times were taken from analysis conducted on a standard consumer notebook computer. These results are indicative only and can vary greatly based on the number of components found within the structure, the number of components requiring maintenance at each intervention and the computational power of the processing computer. In one rail organisation, the traditional planning method was closely replicated in the model by employing the refurbish strategy with yearly interventions as shown in Figure 6. Component codes provided in the Figures are S=span, G=girder and D=deck, and numbers beside the codes indicate their location within the structure S1G1 S1D1 S2G1 S2G2 S2D1 Figure 6 Remaining component life under refurbish strategy (yearly interventions) The refurbish strategy provided a fairly stable average remaining component life over the maintenance interventions as shown in Figure 7. This was due to the ability to conduct maintenance interventions as soon as individual components reached their performance thresholds. The ability to consider the ramifications of placing new, long life components on structures with older, shorter life components was identified as a shortfall of traditional maintenance planning methods. The BMO-MW application displayed this issue in a visual format by graphing each component and its remaining life, including the effect from future maintenance actions.

7 Remaining value ($) Remaining life (years) Remaining life (years) Remaining life (years) 3 Figure 7 - Average component life under refurbish strategy (yearly interventions) Running the application with the upgrade strategy option selected allowed the inclusion of all maintenance actions as alternatives into the analysis. Under this strategy the application rejected all upgrade maintenance actions and opted for the refurbish actions until the average component performance decreased, before then upgrading both the components together at the same maintenance intervention, as shown in Figure 8. In this case, the timber deck was replaced with another timber deck until the girders deteriorated before upgrading the deck to concrete. This maintenance plan demonstrated the lowest life cycle cost over the 7 year planning horizon Average remaining component life S1G1 S1D1 P1P1 P1P2 S2G2 A2H1 Figure 8 - Upgrade strategy ( year interventions) An analysis of bridge maintenance strategy results has shown that a component upgrade or bridge renewal may not always be the most cost effective option at any given point in time. Analysis results indicate that delaying an upgrade or bridge renewal and conducting lower cost maintenance actions, until the average condition of other bridge components deteriorate to a lesser (yet still safe) condition will provide a greater life cycle cost benefit. This allows the organisation to realise the greatest return from more components in a bridge investment before scheduling major maintenance actions. Running the renewal strategy with the ten year minimum intervention option resulted in the component average remaining life displayed in Figure 9. The application identified opportunities for bridge renewals at years 5, 21, 35 and 52, with the lowest average remaining component life being at year 21. Calculation of the lowest remaining component cost at the years of interventions is provided in Figure. As mentioned earlier, the objective function of the renewal strategy is to construct a new structure when the average remaining value of components in the bridge is at the lowest. Figure shows the most cost effective time for this bridge renewal is at the year 5 intervention. This implies that, without a bridge renewal, planned maintenance works conducted after this date will improve the overall value of the structure Average remaining component life Identified years to conduct bridge renewals Figure 9 - Renewal year options (min. year interventions) Adjusting the minimum intervention intervals produced a greater number of opportunities to intervene with maintenance and bridge renewal. This allows organisations to conduct what if scenarios and balance future works with maintenance budgets. Regardless of the maintenance intervention interval selected for renewals, the application provided results for the lowest average remaining life within a range of 2 years. This suggests that selecting any maintenance intervention interval is acceptable to determine the renewal year based on the objective function of the renewal strategy. $, $1, $, $, $ Average remaining component value Figure - Average remaining component value The optimisation process presented in this paper is sensitive to the accuracy of the remaining service life as any errors will be reflected in the accuracy of the optimised maintenance solution. As this model is based on a life cycle cost analysis of bridge components, knowledge of accurate intervention times and life gained by relevant maintenance alternatives have a large impact on the total life cost. Therefore, remaining service life is recommended to be calculated through time-dependent probabilistic modelling methods, for example Markovian analysis (Lounis and Vanier, 1998).

8 7 CONCLUSION This model optimises the life and maintenance cost performance of bridge components to meet the objective functions assigned to maintenance strategies. This allows infrastructure managers to change maintenance strategies for individual bridges to meet existing budget constraints, estimate future network budgets and develop long term maintenance plans. The BMO-MW application is amalgamated with other bridge applications within a framework to produce a comprehensive bridge management system. Structural upgrades planned by traditional methods did not adequately consider the remaining life of other critical components. This resulted in components with a long life being placed on older components with a far shorter remaining life. This application has addressed this issue by presenting results in a user friendly graphical interface where planning issues are clearly evident. Another benefit of this model over traditional maintenance planning methods is the addition of optimised life cycle cost performance at the component level and the ability to change maintenance strategies, thus allowing improved control and management of individual bridges within the network. An analysis of one bridge found that this model produced sound maintenance plans over three major maintenance interventions as opposed to traditional planning over a set planning horizon. The BMO-MW application includes the ability to conduct what-if scenarios by changing maintenance strategies, maintenance costs, component life gained from maintenance, discount rate, intervention periods, safety factors and remaining service life of components. Future work includes an expanded trial of this model in other railway organisations and to determine the suitability of this model on other civil structures. 8 ACKNOWLEDGEMENTS The authors are grateful to the CRC for Rail Innovation (established and supported under the Australian Government's Cooperative Research Centres Program) for the funding of this research Project No. R Project Title: Life Cycle Management of Railway Bridges. The authors also acknowledge the support of the Centre for Railway Engineering, Central Queensland University and the many industry partners that contributed to this project. REFERENCES AFLATOONI, M., CHAN, T. H. T., THAMBIRATNAM, D. P. & THILAKARATHNA, I. (12) Classification of Railway Bridges Based on Criticality and Vulnerability Factors. Australian Structural Engineering Conference. Perth, Australia, Engineers Australia. AFLATOONI, M., CHAN, T. H. T., THAMBIRATNAM, D. P. & THILAKARATHNA, I. (13a) Susceptibility of the Critical Structural Components of Railway Bridges to the Changes in Train Speed. APVC13 15th Asia Pacific Vibration Conference Jeju, Korea. AFLATOONI, M., CHAN, T. H. T., THAMBIRATNAM, D. P. & THILAKARATHNA, I. (13b) Synthetic Rating System for Railway Bridge Management. Journal of Civil Structural Health Monitoring 3, FRANGOPOL, D. M. (11) Life-Cycle performance, management, and optimisation of structural systems under uncertainty: Accomplishments and challenges. Structure and Infrastructure Engineering, 7, HAWK, H. (3) NCHRP Report 483: Bridge Life Cycle Costs Analysis. National Cooperative Highway Research Program. Washington, DC, Transportation Research Board. KONG, J. S. & FRANGOPOL, D. M. (3) Life-cycle reliability-based maintenance cost optimization of deteriorating structures with emphasis on bridges. Journal of Structural Engineering, 129, LIU, M. & FRANGOPOL, D. M. (5) Multiobjective maintenance planning optimization for deteriorating bridges considering condition, safety, and life-cycle cost. Journal of Structural Engineering, 131, LOUNIS, Z. & VANIER, D. J. (1998) Optimization of bridge maintenance management using Markovian models. International Conference on Short and Medium Span Bridges. Calgary, Alberta, NRC Publications. NIELSEN, D., CHATTOPADHYAY, G. & RAMAN, D. (11) Proposed Australian Rail Bridge Management Framework. IN GALAR, D. (Ed. Maintenance Performance Measurement & Management. Lulea, Sweden, Lulea University of Technology. NIELSEN, D., CHATTOPADHYAY, G. & RAMAN, D. (12) Life cycle management of railway bridges Defect management. IN DHANASEKAR, M., CONSTABLE, T. & SCHONFELD, D. (Eds.) CORE - Conference on Railway Engineering. Brisbane, Australia, Railway Technical Society of Australasia. NIELSEN, D., CHATTOPADHYAY, G. & RAMAN, D. (13a) An Australian railway bridge management framework. Int. J. Strategic Engineering Asset Management,, 1, NIELSEN, D., CHATTOPADHYAY, G. & RAMAN, D. (13b) Life Cycle Cost Estimation for Railway Bridge Maintenance. IN BANERJEE, A. K. & SINGHAL, A. K. (Eds.) International Heavy Haul Association Conference. New Delhi, India, Indian Railways. YANG, S.-I., FRANGOPOL, D. M., KAWAKAMI, Y. & NEVES, L. C. (6) The use of lifetime functions in the optimization of interventions on existing bridges considering maintenance and failure costs. Reliability Engineering and System Safety, 91, YANG, S.-I., FRANGOPOL, D. M. & NEVES, L. C. (4) Service life prediction of structural systems using lifetime functions with emphasis on bridges. Reliability Engineering and System Safety, 86,