Predictive Maintenance for Railway Tracks
One of the largest controllable expenditures in railway asset management relates to maintenance engineering.
Improvement in data quality for record keeping, data integration, data analysis and safety compliance reporting
Reduction in false positives compared to pre-existing predictive maintenance models
In projected efficiency savings derived from better whole-life cost and asset life-cycle management
Our client is the owner and manager of railway infrastructure assets in Great Britain. They are responsible for the installation, maintenance and renewal of more than 34,500 kilometres of railway track which supports up to 5 million passenger journeys and the delivery of 200,000 tonnes of goods by rail freight every day. The senior leadership team set themselves an ambitious target to achieve over £3 billion in efficiency savings by 2024 by investing in new technology to allow them to collect, monitor and predict the condition of railway infrastructure assets. By predicting failures before they happen they can plan maintenance work during off-peak times and prevent lengthly and costly passenger disruption during peak-times.
We focused on helping the front-line engineers link the way they prioritise fixing faults (Failure Mode Effects and Criticality Analysis) to the wider organisational performance metrics in order to harmonise reporting and increase transparency across the organisation. This helped to increase awareness that maintenance engineering activities could not be looked at from just a costs, safety and reliability perspective but they must also be viewed as an opportunity to increase efficiency, transportation capacity, quality control, environment sustainability, customer satisfaction and the overall health of the national transport system.
Plotly creates and stewards the leading data visualisation and UI tools for machine learning, data science, engineering, and the sciences. Language support for Python, R and Julia.
We aggregated over 454 terabytes of structured and unstructured data from over 71 different enterprise software systems and advanced sensory equipment to identify where we could automate consistent routines in manual data processing between different regional business units and management support functions. We built analytical dashboards and an extension of a generalised proportional intensities model to offer maintenance engineers an enhanced ability to rank their decisions in scheduling preventive maintenance tasks according to failure mode effects and the criticality of an asset. We also integrated scheduled tasks with a spare-parts and inventory management system to ensure the maintenance engineers had the correct spare parts and equipment onsite to perform their scheduled maintenance activities. This will help the client monitor and control capital and operational expenditures.
Our solution uncovered a significant opportunity to eliminate silos in data collection and data management. This resulted in a 93% increase in data quality, 87% reduction in false positive for predictive maintenance models and over £87 million in projected efficiency savings. These projected efficiency savings are going to be achieved by increased transparency between front-line maintenance engineers and senior leadership teams resulting in better whole-life cost and asset life-cycle management. Our analytical dashboards have also helped the client acquire an evidence based approach to measuring cost-benefits associated with major capital expenditures, which typically range between £4-7 billion per year.