ev-flow: The Open-Source Python Tool That Generates Realistic EV Charging Behavior for 8 U.S. Regions—Without a Single Real Charging Record

Electric-vehicle grid-integration studies require large populations of realistic charging profiles, but real telemetry is scarce and privacy-restricted. Existing open generators are either calibrated to non-US mobility surveys or flatten regional heterogeneity. Bertrand Travacca's ev-flow (Python import name pev_synth) solves this with a deterministic nine-stage pipeline (M1–M9) that stitches NHTS survey person-days into 365-day travel calendars, applies temperature-dependent winter energy uplift, samples plug-in start times from the SPEECh Gaussian-mixture parameterization, and propagates a continuous-time state-of-charge ledger with PHEV range-extension terms. Outputs are UTC-stored, timezone-aware, and bit-reproducible from a single master seed.

Ev-flow supports eight US regions and generates both residential and workplace profiles with descriptive EVSE brand and connector enrichment. The bay_area reference residential profile was validated against published bounds: 11 PASS, 0 unexplained FAIL, 6 explained failures, and 4 explained skips across 21 checks. This tool fills a US-focused, NHTS-grounded niche complementary to European generators such as emobpy and VencoPy and simulators like datafev and ACN-Sim. It is MIT-licensed and available on arXiv (arXiv:2606.19520).