Condition Monitoring, Technology and IT

CQUniversity and the digitalisation path to predictive maintenance

How the Centre for Railway Engineering at Central Queensland University is using advanced simulation techniques for railway maintenance applications.

Railway research is one of the main priority research areas at Central Queensland University. As a result, the use of digital technology by the Centre for Rail Engineering (CRE) located at the North Rockhampton campus is helping drive improvements and innovations in the sphere of rollingstock and track monitoring and maintenance.

Research projects and programs performed by the CRE team are focused on achieving a comprehensive understanding and characterisation of railway track component damage processes considering dynamic and impact loading at the wheel-rail interface.

Centre deputy director Professor Maksym Spiryagin said the focus of recent industry research projects was on rail wear and fatigue crack initiation and propagation, known as rolling contact fatigue (RCF).

“The research requires multi-disciplinary knowledge to be applied to cover such complex systems that starts from locomotive and wagon design and finishes at the track design levels,” Spiryagin said.

“The research studies use monitoring and measurement techniques to deliver input parameters (friction characteristic at the wheel-rail interface, wear in terms of changes in wheel and rail profiles, in-train, traction and braking forces, wheel-rail contact forces etc.) for digital twins combining advanced locomotive, wagon and train models with monitored parameters such as dynamic loads, and accelerations.”

The ability of academia and industry to create and execute increasingly complex models with greater accuracy and applicability to practical scenarios is quite important for developing comprehensive and system wide train/vehicle/track simulations. This is an important step forward toward providing more accurate predictions of the maintenance and damage and due to the improved modelling better understanding of the uncertainties and dependencies in the predictions. At the same time, the use of the latest computational methods allows faster and more efficient information delivery than the existing methods.

Simulation technology

The Centre has a long history of not only using simulation but also developing new modelling and simulation codes from scratch. This valuable intellectual property allows totally new simulation modelling of new products and facilitates smarter transducer devices with embedded modelling codes that can achieve on-board processing. Such technology means that solutions can be found to extend big data and internet-of-things technologies into new areas. These capabilities ensure that developments are not held back using “yesterday’s software”. Important developments include work on all aspects of train dynamics in a team led by Professor Colin Cole, a CRE director. Meanwhile Dr Qing Wu, the recipient of an Australian Research Council Discovery Early Career Award, is focussed on parallel computing and track modelling tasks, and their applications on high performance computing platforms.

Professor Spiryagin considers himself a system integrator in advanced simulation developments.

“Digital twin study approach avoids any limitations connected with the costs of experimental programs including interruption of train operational services,” Spiryagin said.

Considering the research problem is focussed on the complexities of the non-linear wheel-rail contact interface characteristics and the numerical characterisation of wear processes considering tribological aspects and train operational scenarios that includes a full-mechatronic model of rail vehicles and a detailed model of track, the virtual simulation platform is implemented on CQU’s HPC cluster. The digital twin model uses the parallel computing and co-simulation technique. It uses one independent processor core on the HPC to simulate each vehicle in the train on the whole railway route.

“Individual cores then communicate with the longitudinal train dynamics simulation through the co-simulation interface to replicate actual train behaviour by means of the application of digital twin technologies,” Spiryagin said.

Modelling and validation

In recent projects, a key emphasis has been on specific application components.

These include creep force and adhesion modelling. Modelling and validation of locomotive dynamic behaviour involves complex multi-disciplinary engineering problems which require coverage of all uncertainties and non-linearities present in the system. Key non-linearities in this complex task are the adhesion and friction processes at the wheel-rail interface and the characteristics of the control and traction system. The current theoretical contact models cannot accurately represent the creep force characteristics at the wheel-rail interface without friction measurements being performed. To avoid high costs and disruption, CRE researchers, Dr Sundar Shrestha and Esteban Bernal Arango, led by Spiryagin studied recent developments in creep force measurement and modelling techniques allowing locomotive/track damage models to use friction-creep curves delivered using laboratory testing of rail and wheel steels on the Centre’s tribology /wear testing   machine and the hand-operated tribometer in the field. The experimental data provided an understanding of adhesion behaviour as well as modelling principles of friction-adhesion behaviour when third body materials (pollutants, lubricants, water) are the present at the wheel-rail interface.

Also examined in detail has been rail heat modelling. The advanced modelling that allowed the estimation of the heat transfer processes at the rail through the temperature measurements in the field and modelling with parallel simulation on the HPC was performed by the Centre’s PhD student, Chris Bosomworth who has developed a low cost self-powered IoT device manufactured for rail temperature measurements. These devices were magnetically attached and built, based on electrically insulated temperature sensors. The work allowed to further improve accuracy of the heat transfer prediction processes at the wheel-rail interface and rail, in addition to detection of its influence on adhesion and rail damage.

All these modelling techniques and approaches bring commercial benefits from virtual testing. “The combination of experimental studies and a virtual testing platform is progressing toward providing a better understanding of the track damage processes between wheels and rails,” Spiryagin said.

“It will allow predicting the service life cycle of Australian rail materials and track components under specific locomotive and wagon operational conditions in several Australian states.

“New analysis tools and techniques have been developed yo allow for analysing train and vehicle/track simulation data with a view to identifying track sections with a high risk of rail damage and possible derailments.

“This is indeed a significant development.”

Final thoughts

The research studies in this area were, and are, funded by the Australasian Centre for Rail Innovation and its industry partners who believe there is a contribution towards better operational safety and performance of the whole train/track system.