Tidal Blade Test Analysis

Research-software workflows for full-scale composite tidal blade structural-test data

Python utilities, examples, and documentation for processing full-scale tidal blade structural-test data.

The public repository is designed to show the analysis workflow without exposing private TDMS files, test logs, DIC images, or large generated result folders.

---

Scope

This repository focuses on reusable analysis tasks from full-scale composite tidal blade testing:

• TDMS channel inspection and summarisation;
• static load-displacement processing;
• fatigue-cycle peak/trough summaries;
• natural-frequency and damping helpers;
• actuator-load and root-bending-moment checks;
• single-actuator and multi-actuator test comparison;
• lightweight applied-AI screening for long sensor time series.

The main public scope is the analysis context of the first FastBlade / LoadTide full-scale fatigue test and the later single-vs-multi-actuator comparison study. Later clamping/load-introduction and destructive-testing studies are listed as related FastBlade work, but this repository does not currently claim to reproduce their DIC, clamping, or failure-analysis workflows.

What this is not

This is not a replacement for the private experimental data archive. Raw TDMS files, Excel logs, DIC images, and generated result folders are intentionally ignored by Git. The repository provides a public software layer: code, documentation, synthetic tests, configuration examples, and small public demos.

---

Visual workflow

flowchart LR
    A[Raw TDMS + test log] --> B[Channel inventory]
    B --> C[Join / resample]
    C --> D1[Natural frequency]
    C --> D2[Static load-displacement]
    C --> D3[Fatigue cycle summaries]
    C --> D4[Actuator / RBM checks]
    C --> D5[Applied-AI screening]
    D1 --> E[Results + plots]
    D2 --> E
    D3 --> E
    D4 --> E
    D5 --> E
    E --> F[QA/QC report]

Loading

---

Example outputs

Synthetic examples are included so the workflow can be demonstrated without publishing private experimental files.

Static response	Fatigue-cycle summary	Applied-AI screening

Regenerate the figures with:

python examples/generate_readme_figures.py

---

Repository contents

Area	Contents
src/tidal_blade_test_analysis/	Tested Python utilities and CLI commands
tests/	Pytest tests using synthetic public data
examples/	Configuration templates and synthetic examples
docs/	Publications, test configurations, workflow guide, software card, data dictionary, and applied-AI notes
data/	Local data layout description; raw data ignored by Git

---

Quick start

Conda users

conda create -n tidal-blade-test python=3.10
conda activate tidal-blade-test
python -m pip install --upgrade pip
python -m pip install -e ".[dev]"
python -m pytest

Windows PowerShell

python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install --upgrade pip
python -m pip install -e ".[dev]"
python -m pytest

macOS/Linux

python -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e ".[dev]"
python -m pytest

---

Command-line examples

Create local data folders:

tidal-blade-test init --root .

Summarise channels from a TDMS file or folder:

tidal-blade-test summarise data/raw --output data/results/channel_summary.csv

Find dominant frequencies from a CSV signal column:

tidal-blade-test fft data/processed/free_decay.csv --column load --sample-rate-hz 2500 --n-peaks 5

Fit a static load-displacement relationship:

tidal-blade-test static-fit data/processed/static_case.csv --load-column load --displacement-column displacement

Run the public synthetic examples:

python examples/synthetic_static_demo.py
python examples/synthetic_ml_anomaly_demo.py

---

Applied AI / sensor QA screening

The repository includes a lightweight applied-AI layer for long structural-test time series. The purpose is to help prioritise engineering review, not to replace it.

Included tools:

• fixed-window feature extraction for actuator loads, displacement, strain, root bending moment, pressure, or other processed channels;
• fatigue-cycle response features based on turning points;
• robust drift scoring between a reference section and a candidate section;
• unsupervised anomaly screening using robust scaling and Isolation Forest.

The anomaly labels are screening flags, not confirmed damage labels. A flagged window should be checked against the raw signal, test log, actuator state, sensor health, environmental conditions, and engineering judgement.

Example:

python examples/synthetic_ml_anomaly_demo.py

See docs/applied_ai.md for details.

---

Configuration examples

The repository includes example YAML files for the core test configurations:

• examples/config.single_actuator.yml
• examples/config.single_vs_multi_actuator.yml
• examples/config.example.yml
• examples/config.applied_ai_anomaly.yml

These describe actuator count, actuator positions, target root bending moment, loading direction, analysis outputs, and screening inputs without exposing private data.

---

Publications

See docs/publications.md.

Primary publications supported by the repository scope

1. A Full-Scale Tidal Blade Fatigue Test using the FastBlade Facility.
2. A full-scale composite tidal blade fatigue test using single and multiple actuators.

Related FastBlade studies

1. Clamping parameters in full-scale tidal turbine blade tests: A case study.
2. Destructive testing and failure analysis of a full-scale composite tidal turbine blade.

---

Development

python -m pip install -e ".[dev]"
python -m ruff check src tests
python -m pytest --cov=tidal_blade_test_analysis --cov-report=term-missing

---

Citation

A CITATION.cff file is included so GitHub can generate citation text for the repository. The citation metadata also lists the primary and related FastBlade publications.

---

License

MIT License. See LICENSE.

---

Author

Sergio Lopez-Dubon