Rail Cargo Austria Wagon Planning

Background and industry context

Rail Cargo Austria AG is the lead operating company of the ÖBB Rail Cargo Group (RCG). The RCG is one of Europe’s largest rail logistics companies, employing over 5700 people and generating ~$2B in annual sales. To keep rail freight competitive in an increasingly volatile market, the Austrian research project “Backbone PI: Rail” set out to transform wagon planning into a digital, data‑driven process. My involvement spanned 7 months within this 28‑month initiative (Jan 2019 – Apr 2021)

Intro problem

Traditional wagon allocation at Rail Cargo Austria relied on tacit knowledge and spreadsheets passed between experts. Data sources were opaque, often duplicated or missing, and multiple teams spent days assembling a single wagon plan.

Result

Working alongside Fraunhofer Austria, TU Wien and Rail Cargo Austria, our team delivered a PoC application that wraps two predictive models and a plan‑management layer into a single, user‑friendly system. The application consolidates all data sources and business logic, enabling a single operator to generate a wagon plan in minutes rather than days. Centralising the process eliminated the ambiguous spreadsheets and miscommunication that previously plagued wagon planning. Because the system continuously integrates demand forecasts and operational constraints, Rail Cargo Austria can reassign wagons dynamically to maximise utilisation. The project demonstrated multi‑million‑euro potential in savings from more efficient wagon allocation and laid the foundation for broader adoption of anticipatory planning across the rail network.

Description of the solution

My involvement

As lead developer on craftworks’ side, I translated high‑level business rules into executable code for data manipulation. I wrote all the data “glue” for assembling data sources, processing model inputs/outputs and user interaction. Close collaboration with Rail Cargo Austria’s planners and researchers from Fraunhofer and TU Wien helped refine the model and fix logical inconsistencies. I integrated data pipelines from all partners into a cohesive application and exposed clear APIs for model inputs and outputs. This work enabled real‑time data transformations and made it possible to run complex planning algorithms at the push of a button.

Media

• Video overview – Smarte Schiene: Backbone PI Rail system showcased around 1min50.
• FFG Project Description

Technologies used

• Python & Pandas – for data integration and transformation.
• Flask – to expose the planning models as a RESTful service.
• Docker – containerising services for reproducibility.
• Constraint solver algorithms – to assign wagons to demands under multiple constraints.
• Alteryx – for data preprocessing and workflow automation.