About AssetLife
AssetLife is an open source Python library hosted at LF Energy that analyzes historical asset data — failures, repairs, service duration, deterioration measurements, and contextual factors like location or corrosion levels — to build statistical models for lifetime distributions and failure behavior. These models are then used to compare maintenance strategies and determine when to maintain, repair, or replace assets by balancing preventive and corrective costs. It identifies optimal maintenance policies, and projects the expected total cost and replacement count over a planned horizon. Cost calculations include direct replacement expenses and societal costs, such as carbon shadow pricing, which strengthens the economic justification for renewal investments.
Project Lifecycle Stage
Sandbox
Project Special Interest Group
Grid Simulation & Modeling
AssetLife was originally developed and used in production at RTE, the French transmission system operator (TSO), for transmission asset management. It implements survival analysis, renewal theory, and cost calculation from economic engineering.
AssetLife is a statistical toolbox, not a simulation platform.
The Challenge
In the context of aging infrastructures and grid modernization, asset managers face critical investment decisions: when to replace assets, at what cost, and with what consequences for asset stock and budgets? Without a rigorous analytical basis, organizations default to rules of thumb or proprietary tools that cannot be independently validated, reproduced, or adapted to their own failure data.
The consequences are significant. Maintenance scheduled too early wastes capital. Deferred maintenance increases risk exposure and can lead to unplanned outages. Neither outcome is defensible to regulators or investors when a quantitative alternative exists.
Key Features
Lifetime Modeling
AssetLife provides parametric and non-parametric estimators, with and without covariates. Semi-parametric Cox regression is also supported to study the effects of covariates on lifetime distribution.
Recurrent Event Modeling
For assets subject to recurring failures or maintenance events, AssetLife provides non-homogeneous Poisson processes.
Maintenance Policy Modeling
AssetLife implements renewal processes with and without rewards associated with a renewal equation solver. These methods underpin the age-based replacement policies modeling and their cost computions. It allows organizations to compare the full cost and risk trajectory of different maintenance strategies over a defined planning horizon.
Socioeconomic Evaluation
Cost modeling can incorporate the shadow price of carbon alongside direct replacement and failure costs, supporting socioeconomic justification of capital programs. This is particularly relevant for transmission infrastructure, where replacement decisions carry societal as well as financial consequences.
Asset-Type Agnostic Methods
The statistical methods in AssetLife are independent of asset type. While the project originated in electricity transmission, the library applies to any aging infrastructure population for which failure and maintenance records exist.
Key Contributors
Production Users
- RTE
- TenneT
- AusNet
- GRTgaz
R&D and Academic Partners
- Artelys
- CentraleSupélec
- Sorbonne Université
Technical Foundation
AssetLife is built on the standard Python scientific stack and is distributed as an open source Python package.
- NumPy and SciPy provide the numerical computation foundation
- Python is the implementation language, using established conventions for scientific libraries
- Apache 2.0 license governs the codebase
- Source code and documentation are available at https://opensource.rte-france.com/AssetLife/
- PyPI distribution supports standard installation via python -m pip install assetlife
Use Cases
Asset Replacement Policy
Fit lifetime models from assets historical failure records. Then, model and compare run-to-failure policy costs with preventive age replacement policy ones.
Capital Budget Forecasting
Project annual replacement volumes and associated costs for asset populations under alternative maintenance strategies, providing the quantitative basis for multi-year capital expenditure plans and regulatory submissions.
Repairable Asset Management
Model recurrent failure events for assets that are repaired rather than replaced, using non-homogeneous Poisson process counting models to project future failure rates and evaluate repair versus replacement thresholds.
Academic Research
Academic partners at CentraleSupélec and Sorbonne Université use AssetLife in teaching and research programs. Additional research partners are welcome by the community.
Collaboration Opportunities
AssetLife welcomes participation from asset management practitioners, reliability engineers, and academic researchers working on:
- Failure data analysis and statistical modeling for infrastructure asset populations
- Maintenance policy optimization and capital planning
- Renewal theory and stochastic process modeling
Organizations with failure and maintenance data who want a peer-reviewed analytical framework, researchers developing or validating maintenance optimization methods, and engineers building asset investment planning workflows are all encouraged to contribute.
The maintainers have expressed openness to improving the existing Sphinx-based documentation website (https://opensource.rte-france.com/AssetLife/) and to partnerships that can extend the library’s capabilities and real-world application.
Resources: