Prime Air Drone Logistics

A last-mile feasibility study for suburban drone delivery in Massachusetts: where to place hubs, and how much of the state they reach.

Result: A 3-hub network reaching ~550,000 residents (~7.9% of Massachusetts), an estimated ~1,900 packages/day. Stack: Python, pandas, Folium Live case study: https://pranavkaja.vercel.app/projects/prime-air-drone-logistics

The problem

An MK30-class delivery drone covers roughly a 7.5-mile radius on a 15-minute flight. Given that range, where do you put a small number of hubs to reach the most people across Massachusetts, and is the volume worth it?

Approach

This is a multi-criteria decision analysis, not a single formula. Each candidate city is scored 0-10 on three criteria, then weighted:

Criterion	Weight	What it captures
Demand potential	0.40	Population and order density reachable within 7.5 miles
Tech & income readiness	0.35	Income, education, and likely adoption
Operational feasibility	0.25	Airspace complexity, existing infrastructure, room to site a hub

Two hubs whose service areas overlap add little unique coverage, so rather than stacking the top scorers (which all cluster around Boston) the model picks the highest-scoring city in each macro-region (East, Central, West). That spreads the network statewide, a standard max-coverage facility-location move. Daily volume is then projected from the reachable population and an assumed weekly adoption rate that scales with each region's readiness.

Results

Running drone_hub_siting.py:

=== Selected hubs (highest score per region) ===
       city  region  score  pop_15min_radius  daily_packages
  Cambridge    East   9.10            210000             850
  Worcester Central   7.90            185000             600
Springfield    West   6.90            155000             450

=== Network coverage ===
Hubs:              3 (Cambridge, Worcester, Springfield)
Residents reached: ~550,000 (7.9% of Massachusetts)
Projected volume:  ~1,900 packages/day

The script also writes coverage_map.html, an interactive Folium map with each hub's 7.5-mile service circle.

Run it

git clone https://github.com/PranavKaja/prime-air-coverage.git
cd prime-air-coverage
pip install -r requirements.txt
python drone_hub_siting.py   # prints the ranking + coverage, writes coverage_map.html

Inputs live in candidate_cities.csv, so you can adjust the candidate list, scores, weights, or adoption assumptions and rerun.

Notes

A feasibility study from graduate supply-chain coursework, not an official Amazon project. Population, readiness scores, and adoption rates are documented estimates meant to make the siting logic explicit and reproducible, not measured operational data.

---

Part of my portfolio. Built by Pranav Kaja.