Dataset of Electric Autonomous Dial-a-Ride Instances with Local Energy Communities and Electricity Tariffs

This dataset provides benchmark instances for the electric autonomous dial-a-ride problem (e-ADARP) integrated with local energy communities (LECs). The instances are derived from real-world taxi trip data from Porto, Portugal, and incorporate energy data from two LECs, each consisting of 10 prosumers with photovoltaic generation and individual load profiles. The dataset contains: 20 instance files across 10 size configurations (2 to 20 vehicles, 20 to 200 requests), with two electricity tariff scenarios per configuration: ToU/ - Time-of-use tariff with mid-market rate pricing (files: r{K}-{N}.txt) Flat/ — Flat tariff with mid-market rate pricing (files: rF{K}-{N}.txt) LEC_results_midPoint.xlsx — Excel workbook with complete prosumer-level energy data for both LECs (19 sheets): member locations and feeder IDs, load consumption profiles, PV generation profiles, battery storage parameters, electricity buy/sell prices under ToU and flat tariffs, and grid connection limits. All time-series data are at 15-minute resolution (96 periods per day). results_LEC.zip — LEC optimization output CSV files used as direct input to the instance generation script. Contains the aggregated LEC balance (overproduction), MMR buy prices, and grid tariff buy prices for both LECs under both tariff scenarios, at 15-minute resolution. generate_instances.jl — Julia script implementing the full instance generation pipeline: reading taxi trajectory data, extracting the Porto road network via OSMnx, computing pairwise shortest-path travel time matrices, loading LEC energy profiles, generating time windows, and writing instance files in the format described in the README. group_prosumers.py — Python script implementing the compatibility-constrained clustering algorithm used to form the two LECs from a pool of 100 residential prosumers, based on geographic distance and electrical feeder assignment constraints. README.txt - A plain-text file describing the instance file format block by block, the repository folder structure, the required dependencies for running the generation scripts, and instructions for reproducing the dataset. This is the recommended starting point for new users of the repository. Each instance file encodes: geographic node coordinates (WGS84), pickup/dropoff time windows, vehicle parameters (battery capacity, energy consumption, passenger capacity, minimum final state-of-charge), charging station locations and rates, LEC energy overproduction profiles, electricity tariffs (grid and mid-market rate) per 15-minute interval, and a full pairwise travel time matrix computed via shortest paths on the OpenStreetMap road network. The instances enable reproducible benchmarking of routing and scheduling algorithms that jointly optimize transportation service quality and charging costs under realistic energy market conditions.