Skip to content

Home

Acropole Logo

Acropole — Predict aircraft fuel flow from trajectory data

CI axm-audit axm-init coverage PyPI Python 3.12+ License: AGPL-3.0

acropole predicts the fuel flow of aircraft (kg/s, kg/h and cumulative kg) from trajectory data — groundspeed, altitude and vertical rate — using a portable ONNX model trained on Quick Access Recorder (QAR) data. It accepts a pandas or polars DataFrame, dispatches per aircraft typecode, and returns the same frame enriched with fuel-flow columns.

Features

  • ⛽ Fuel-flow prediction — fuel_flow (kg/s), fuel_flow_kgh (kg/h), fuel_cumsum (kg)
  • ✈️ Multi-aircraft — frames mixing typecodes are scored per typecode
  • 🐼 pandas and polars — same type in, same type out; polars engine internally
  • 🚀 Fast ONNX runtime — 2–4.8× faster than the original TensorFlow model, no TF dependency
  • 📈 Temporal derivatives — accelerations from a second column, or pre-computed
  • 🎯 Column mapping — point each feature at your own column names
  • 💻 Command-line — the acropole estimate command enriches a CSV/parquet file without writing Python

Quick Start

pip install "acropole[pandas]"

import pandas as pd
from acropole import FuelEstimator

flight = pd.DataFrame({
    "typecode": ["A320", "A320", "A320", "A320"],
    "groundspeed": [400, 410, 420, 430],
    "altitude": [10000, 11000, 12000, 13000],
    "vertical_rate": [2000, 1500, 1000, 500],
})

flight_fuel = FuelEstimator().estimate(flight)
# adds fuel_flow (kg/s), fuel_flow_kgh (kg/h)

Prefer the command line? acropole estimate flight.csv writes an enriched flight_fuel.csv — see Estimate fuel from the command line.

Documentation

This documentation follows the Diátaxis framework:

  • Tutorials — learn Acropole step by step, from install to your first estimate.
  • How-To Guides — task-oriented recipes: multi-aircraft frames, column mapping, derivatives.
  • Reference — the Python API surface, auto-generated from the source.
  • Explanation — how the polars pipeline, the ONNX model and the typecode dispatch fit together.