Optimization of maintenance policy based on operational reliability analysis : application to railway switches & crossings

Projekt:

JVTC

Sammanfattning:
The present communication reports on a collaboration between ALSTOM Transport and Luleå Technology University, under the sponsorship of Trafikverket, the Swedish Infrastructure Manager.

For 2020, the European Rail Research Advisory Council (ERRAC) has set among other the following objectives [2] : doubling passenger traffic and tripling freight traffic and reducing the life-cycle cost of infrastructure by 30%.
This need applies to rail infrastructure in general. Clearly, the highest leverage will be obtained by concentrating the efforts on key cost and availability drivers. For instance, it is reported that switches and crossings (S&C) are one of the subsystems that cause the most delays on Swedish Railways while accounting for at least 13% of maintenance costs. (It is the main reason why we chose to base our study on this subsystem).

Intelligent data processing allows to understand the real reliability characteristics of the assets to be maintained. Furthermore simulation and optimisation techniques are applied in order to adapt the maintenance strategy so as to achieve minimum cost while guaranteeing target availability.

The first step has been to determine the S&C reliability characteristics based on field data collection. Because field failure data are typically strongly censored, we have developed our own statistics software package to process field failure data, as commercial packages have not been found satisfactory in that respect. The resulting software, named RDAT® (Reliability Data Analysis Tool) has been relied upon for this study: it is especially adapted to statistical failure data analysis.
The next step will be to customize the maintenance interval by adapting it to individual switches and crossings behaviour characteristics. In order to predict and optimize life-cycle cost (LCC) and availability, Monte Carlo simulations will be performed with stochastic Petri nets. The failure rates estimated with RDAT® will be used as inputs to the Petri net. Such a model lends itself to maintenance optimization. Indeed, designs of experiments can be used in conjunction with simulations in order to express both LCC and availability as functions of various maintenance-related decision variables (such as preventive maintenance frequency, maintenance efficiency, etc, see [7]).
Further improvement could result from applying condition-based maintenance, where preventive maintenance times would no longer be predetermined but rather based on the observed S&C condition, as measured by : number of displacement cycles, current absorbed, vibration intensity during train passage, traffic intensity etc. It is planned to resort to the Watchdog Agent ® software, of Intelligent Maintenance System, to that end [4].

Författare: Benjamin Bonnet ; Marine Parahy ; Pierre Dersin ; Behzad Ghodrati
Utgivare: ESReDA
Utgivningsdatum: 2011-10-05
Diarienummer: TRV 2011/58769
Språk: Engelska
Kontaktperson: Per Olof Larsson Kråik, UHjbs