5
Summary
Public transportation systems such as the RandstadRail light rail network, require a high level of system performance in order to provide safe and comfortable transport services for the customers. The goal of this study was to find the best maintenance policy for the RandstadRail rolling stock of HTM, based on the requirements as stated in the service level agreement with regional authority and asset owner MRDH. In order to achieve the goal and find the best maintenance policy, this study sought to answer the following research question:“What maintenance policy should the HTM use for its rolling stock fleet in order to meet the requirements as stated in the service level agreement with MRDH, regarding safety level, availability, reliability, and passenger satisfaction at minimum maintenance cost?”
In order to be able to answer the research question, a model based on the Tsai PM model is constructed to find the best maintenance policy based on reliability, availability and cost. It is assumed that safety and customer satisfaction are directly related to the reliability of the system. The failure behavior of the system is modeled by means of a two‐parameter Weibull distribution. Two improvement factors are defined to model the impact of maintenance actions on the failure behavior of the system. One affects the reliability degradation after maintenance and the other defines the reliability increase after maintenance. The PM model uses a maintenance benefit function to select the most beneficial of three maintenance actions: inspection, low level repair, and overhaul at the end of each maintenance interval. Due the given time constraints, the PM model is applied to the subsystem with the most potential to improve the overall performance of the Citadis fleet. The braking system is selected based on number of failures, impact on availability, and maintenance cost. The PM model is constructed on subsystem level of the braking system. During analysis of the failure data, it is concluded that 35% of the CM cost is caused by false repairs. The potential cost reduction is substantial.
The maintenance policy of the braking system is optimized from three different perspectives: availability driven, cost driven, and reliability driven. The output of the PM model is calibrated based on the current maintenance policy and performance. The performance of the three maintenance policies is given in Table 1. Note that the performance of the current maintenance policy is corrected for false repairs to provide fair benchmarks.
6 Table 1: Summary of proposed maintenance policy performance
Parameter Description Current Optimized for objective
policy Availability Cost Reliability
T Interval 2427 4331 4380 1008 As System availability 0,967 0,971 0,971 0,966 Rs,mean Mean system reliability 0,79 0,80 0,80 0,95 Rs,min Minimum system reliability 0,61 0,59 0,60 0,90 Cpm PM cost [EUR/year] 12833 4065 3863 13638 Ccm CM cost [EUR/year] 5724 5100 5128 5068 Ctot Total maintenance cost [EUR/year] 18557 9165 8991 18706 It is observed that maintenance cost and system availability are highly dependent upon the minimum reliability requirements. Both availability and cost driven maintenance policies show a potential maintenance cost reduction of over 50%, i.e. 689K EUR per year for the fleet. In addition to this cost reduction these policies also maximize the system availability, without sacrificing the current mean and minimum reliability levels. The reliability driven maintenance policy shows a potential increase of 20% in average system reliability and 48% of the minimum system reliability in comparison to the current maintenance policy, without compromising the availability of the system or raising the total maintenance cost significantly.
When applied to all assets in the system, the solution method presented in this thesis has the potential to cover the total cost of ownership and support both strategic and tactical decision making processes to improve the overall safety, availability, reliability, and passenger satisfaction of the RandstadRail service. It is recommended to: improve the failure repair success rate improve the level of detail for failure registration reduce calculation time of PM model increase the level of system detail of both failure and PM modeling quantify and include the total cost of failures during service