A Petri Net Methodology for Modelling the Maintenance of Railway Route Sections

Abstract

Railway utilization is increasing. In 2017-18 there were over 1.7 billion passenger journeys made on the UK railway network, more than double the number made in 1995. Therefore getting maximum performance from the existing network by streamlining maintenance is vital. Previous research has tended to model maintenance interventions as uniform length and assume no spatial dependencies. However, in reality maintenance decisions are generally taken accounting for the condition of the whole of a route section thus prioritizing the utilization of the available resources. In this research, Petri nets will be used to explore a range of techniques for modelling maintenance on a 100 mile section of UK railway. Four methodologies will be implemented and compared. Initially the system will be modelled using a single small (220 yards) section and the values for the route section extrapolated from this. The second method will model the system as 849 individual small sections, with some interactions and dependencies considered between the track sections. The third methodology will consider that maintenance will affect a group of adjacent sections and maintenance will only be performed if all sections are degraded. The final methodology will schedule work using work banks to allow the number of sections maintained during an intervention to vary.

The impact of summer heatwaves on railway track geometry maintenance

“…In a similar fashion, the shorter maintenance schedules 1 and 2 following F routine = N (30,15) and N (60, 15) have improved availability compared to the base case, thanks to sooner completion of the works. These strategy variations are quicker to act on a poor track condition, and so there is a reduced likelihood of reaching the major degradation states.…”

“…These are probabilistic routing transitions, which select one of the two possible output places on firing. In practice, maintenance engineers will select between tamping or stoneblowing as is appropriate to the track section, however trends can be found in maintenance records to simply represent this decision as a ratio 30 . Decision transitions are defined with either routine tamping : routine stoneblowing R.T : R.SB, or urgent tamping : urgent stoneblowing U.T : U.SB.…”

“…In the UK network tamping is strongly preferred for routine maintenance 30 , though stoneblower usage increases for the major degradation conditions as it shows good results when the geometry recovery requirement is larger 31 . For the base case simulation, the routine ratio routine tamping : routine stoneblowing is defined as 9.5 : 1 and the urgent ratio urgent tamping : urgent stoneblowing is 3 : 1.…”

“…27 Petri nets have been used to great success for several track asset management studies. 10,[27][28][29][30] Common features among these models includes the use of the Weibull distribution, which best fits the observed degradation rate data. 7 Also, a modular approach is followed to describe the combination of different system processes.…”

Railway track availability modelling with opportunistic maintenance practice

“…Petri net (PN) modelling has been demonstrated to be a highly flexible method for several successful asset management studies 1,13,[29][30][31][32] . Of these, Prescott and Andrews present the most complete exploration of railway track geometry management with opportunistic maintenance 13 .…”