Open-source, extensible, and modular platform for infrastructure asset management at both project- and network-levels. GIAMS is developed in Python to support research and practice in lifecycle analysis, maintenance–repair–rehabilitation (MRR) planning, and optimization for civil infrastructure (e.g. bridges, buildings).

• Overview
• Features
• Installation
• Quick Start
• Project Structure
• Examples
• Concepts
• Testing
• Extending GIAMS
• Citing
• License

GIAMS provides a unified framework to:

• Model assets (bridges, buildings) as collections of elements with condition states, deterioration, agency and user costs, and utility.
• Simulate life cycles via Monte Carlo over a planning horizon, with optional hazard (e.g. seismic) events, response, loss, and recovery.
• Optimize MRR plans at project level (single asset) or network level (multiple assets under budget) using genetic algorithms, particle swarm, hill climbing, brute force, or custom objectives.
• Plug in new models by extending base classes for assets, elements, deterioration, condition rating, hazards, MRR actions, and optimizers.

Example applications include the Indiana bridge network (4,600+ bridges) with HAZUS-based seismic response and Texas DOT–style user costs, dummy building retrofit with the Rank optimizer, and structural health monitoring (SHM) scenarios.

Requirements: Python 3.9+. Dependencies are declared in pyproject.toml (numpy, pandas, matplotlib, scipy, deap, pyswarms).

git clone <your-repo-url>
cd GIAMS
uv venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
uv pip install -e ".[dev]"

cd GIAMS
python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -e ".[dev]"

The [dev] extra adds pytest and pytest-cov for running tests.

Run the default project-level Indiana bridge example (single asset, LCA + NPV output):

python run.py

From Python:

from Examples import Example1
Example1.example1()

Try different optimizers (uncomment in run.py or call directly):

from Examples import Example1

# Life-cycle analysis only
Example1.example1()

# Optimization
Example1.GA_test()        # Genetic algorithm
Example1.hill_climbing()  # Hill climbing
Example1.brute_force()    # Brute force
Example1.pso()            # Particle swarm

GIAMS/
├── Asset/                    # Asset types and element models
│   ├── AssetTypes/          # BaseAsset, Bridge, Building
│   ├── Elements/            # Bridge/Building elements; condition rating, deterioration,
│   │                        # agency cost, utility (e.g. BRIDGIT, NBI, Sinha, Bai, HAZUS)
│   ├── HazardModels/        # Generator, Response, Loss, Recovery (e.g. Poisson, HAZUS)
│   ├── MRRModels/           # MRR plans and effectiveness (e.g. MRRTwoActions, SHM)
│   └── UserCostModels/      # Texas DOT, dummy building, etc.
├── Network/                 # Network loaders and budget setup
│   ├── Networks/            # Data: INDIANA2019.csv, FirstNetwork.csv
│   ├── BaseNetwork.py
│   ├── IndianaNetwork.py
│   ├── FirstNetwork.py, RandomNetwork.py
│   └── DummyBuildingNetwork.py, DummySHMNetwork.py
├── LifeCycleAnalyzer/       # LCA engine and simulators
│   ├── BaseLCA.py, LCA.py
│   └── Simulators/          # MainSimulator, EnvSimulator, SHMSimulator, DummyRiskAnalyzer
├── Optimizer/               # GA, PSO, HillClimbing, BruteForce, IUC, Rank; objectives
├── utils/                   # NPV, accumulators, distributions, predictive models,
│                            # logging, timing, GeneralSettings
├── Examples/                # Example1–3, EstLca, BuildingRetrofitDummy, SHMExample
├── RLEnvs/                  # BaseEnv, IndianaEnv (reinforcement learning)
├── tests/                   # Pytest suite
├── run.py                   # Entry point (Example1.example1 by default)
└── pyproject.toml

Data files (INDIANA2019.csv, FirstNetwork.csv) are in Network/Networks/.

• LCA (Life-Cycle Analysis): Monte Carlo simulation over a planning horizon; each run uses a simulator to generate user costs, element costs, utilities, and condition states; results are accumulated and used for NPV and objectives.
• MRR (Maintenance, Repair, Rehabilitation): Actions applied to elements over time; modeled via MRR plans (e.g. MRRTwoActions, MRRFourActions, SHMActions) and effectiveness models.
• Hazard: Optional events (e.g. earthquakes via Poisson process); each event triggers response (e.g. HAZUS fragility), loss, and recovery.
• Optimization: Optimizers search over MRR (or similar) decision variables; fitness is typically an LCA-derived objective (e.g. utility / user-cost trade-off). IUC and Rank support network-level project selection under budget.

Run the test suite:

uv run pytest
# or
pytest

With coverage:

uv run pytest --cov=Asset --cov=LifeCycleAnalyzer --cov=Optimizer --cov=Network --cov=utils --cov-report=term-missing

The codebase is structured for drop-in extensions:

For optional refactoring ideas (e.g. type hints, docstrings, layout), see REFACTORING_REMAINING.md.

If you use GIAMS in your research, please cite the paper that describes the platform and its modules (the open-source, extensible asset management system and the Indiana bridge applications). See the project repository or publication list for the exact reference.

This project is licensed under the MIT License — see LICENSE.md for details.