High-resolution agent-based modelling of non-exhaust emissions reveals the limits of urban fleet electrification

As exhaust emissions decline, non-exhaust emissions (NEEs) from brake and tyre wear are emerging as the dominant source of traffic-related particulate matter. This transition is complicated by the increasing mass of electric vehicles and the push toward high-capacity public transport. We introduce a high-resolution, moving-observer agent-based model calibrated against hyperlocal Google Air View street-level measurements at the Phibsborough junction in Dublin, Ireland, a documented pollution hotspot. Global sensitivity analysis reveals that the system is fundamentally interaction-dominated: all first-order Sobol indices fall below 0.013, while total-order indices for fleet composition (0.27–0.34), mean speed (0.27), and speed variability (0.23) are substantial, confirming that traffic dynamics are as influential as fleet mix. We evaluate four policy domains informed by this sensitivity structure through passenger-preserving scenario simulation. Fleet electrification benefits are entirely contingent on regenerative braking: at 50% penetration, UFP ranges from −19% with effective regenerative braking to +13% without, while PM₂.₅ increases by 1–5% regardless of braking technology. Bus modal shift delivers the largest dual-pollutant benefits of any single intervention, peaking at 50% shift (−30% UFP, −34% PM₂.₅), though a bus paradox emerges at 75% where additional heavy buses partially offset UFP gains. Replacing ICE cars with 1,000 kg microcars yields initially modest reductions (−4% UFP at 25% adoption, rising to −30% at 75%), with diminishing returns as the unchanged SUV fleet dominates brake wear. Speed management is the most robust technology-independent lever: a 30 km/h limit reduces UFP by 16.5% and PM₂.₅ by 9% independently of fleet composition, while "slow + smooth" traffic can reduce UFP by up to 45%. A combined policy applying all five levers delivers −27% UFP and −12% PM₂.₅, demonstrating compounding benefits from coordinated action. These findings establish that speed management provides a guaranteed floor of improvement, that bus modal shift is the most effective single intervention, and that electrification must be paired with effective regenerative braking and mass-conscious vehicle choices to avoid worsening tyre-derived PM₂.₅.