rolland: Time-domain simulation framework for coupled wave propagation and dynamics in railway tracks

Abstract

Rolland v26.08 We are excited to announce the initial major release of Rolland, a highly optimized, Devito-based time- domain simulation framework designed for coupled wave propagation and dynamics in railway tracks. This release introduces the foundational numerical modeling mechanics validated in our publication: Mantel, M., & Sarradj, E. (2026). Time-domain modeling of coupled wave propagation in discretely supported railway tracks. Computers & Structures. doi:10.1016/j.compstruc.2026.108432 Highlights & Core Features Time-Domain Solver: Applying Explicit Finite Difference Method (FDM) to solve 13 differential equations of motion alongside 14 auxiliary differential equations. Full Track Dynamics: Comprehensive structural modeling capturing vertical and lateral bending waves, longitudinal waves, axial torsion, cross-sectional warping, and sleeper dynamics under eccentric excitation and support conditions. Flexible Track Configurations: Support for both ballasted and slab tracks with continuous or discrete supports, including spatial variations (periodic and stochastic) of track properties. Infinite Track Boundaries: Time-domain formulation of Complex Frequency-Shifted Perfectly Matched Layers (CFS-PML) to eliminate artificial boundary reflections. Dynamic Excitations: Seamless configuration for stationary excitations (e.g., Gaussian impulses) and moving sources along the track waveguide. Post-Processing & Validation: Clean, dataclass-based utilities (TrackResponse, TrackDecayRate, and VehicleResponse) to calculate point mobilities, transfer mobilities, and track decay rates (TDR). Five built-in analytical frequency-domain benchmark models for rapid verification (Euler-Bernoulli and Timoshenko beam models on continuous and discrete supports). What's Changed since v26.08a0 Post-Processing Modernization: Streamlined post-processing module with typing protocols and dataclasses (TrackResponse, TrackDecayRate). Replaced legacy PointResponse and removed deprecated PostProcessing classes. Added extensive unit tests and test suites for post-processing accuracy. Domain & Boundaries: Added Dirichlet boundary conditions for rigid boundaries in domain setup. Documentation & Visualization: Added animated visualizations and platform support notes to README and Sphinx documentation. Updated citation metadata (CITATION.cff, README, and documentation) with formal DOI references. Added example scripts for evaluating track response and decay rates. Installation pip install rolland Or install with optional extras: pip install rolland[dev,docs,io] ────── Citation If you use Rolland in your research, please cite: @article{mantel2026timedomain, title = {Time-domain modeling of coupled wave propagation in discretely supported railway tracks}, journal = {Computers \& Structures}, volume = {314}, pages = {108432}, year = {2026}, doi = {10.1016/j.compstruc.2026.108432} } @software{mantel2026rolland_api, author = {Mantel, Maximilian and Wagner, Benjamin and Sarradj, Ennes}, title = {Rolland: A Time-Domain Simulation Framework for Railway Track Dynamics and Rolling Noise}, year = {2026}, version = {v26.08}, doi = {10.5281/zenodo.21920225}, url = {https://github.com/mantelmax/rolland} }