ICRI Project: Agent Based Modelling
Agent Based Modelling
Leaders: Wei Huang, NRC Canada
Agent based modeling is a simulation technique that applies user defined relationships to autonomous “decision-making elements” (agents) using stochastic techniques to increment the simulation process from some initial state to a future state. In the case of broken rails, each segment of rail can have a number of characteristics (e.g. rail section, wear state, level of RCF, neutral temperature, support conditions etc.) over which a simulated loading distribution is applied (wheel loads, impact loading, lateral forces). A series of 10,000 rail segments represents a subdivision that is seeded with a number of internal flaws (based on historical defect data) that grow with accumulated loading. Rail grinding can be applied at regular intervals to reduce levels of surface fatigue, larger defects detected and removed (based on typical rail inspection reliabilities) and new defects initiated. The relationships and agent properties are intended to be simple but sensible with the aim of developing understanding of the contributing influences. The current ABM environment being employed is NetLogo (https://ccl.northwestern.edu/netlogo/).
The initial programing of the model has been established and the next steps are to calibrate the model using real-world data. Currently, informations from “Best Practices for Heavy Haul Railway Operations: Wheel and Rail Interface Issues” (Guidelines to best practices for heavy haul railway operations – Wheel and Rail.pdf) will be used to calibrate the model. Factors that will be taken into considerations may include curvature, super elevation, train speed, rail hardness, rail age, rail weight, rail type, car weight, annual traffic, weather condition, bridges, rail inspection and maintenance practice, etc.
If you have experience to offer this effort please contact Wei Huang.