Demand Responsive Transport (DRT) offers flexible alternatives to conventional fixed-route transit in low-demand areas, yet implementations often fail due to inadequate service design. Existing research overlooks how passenger adaptation to service reliability affects system performance over time. This study presents an agent-based model incorporating multi-day user learning, where passengers adjust booking time windows based on experienced delays from previous trips. The model implements hybrid service architecture with mandatory stops in high-density central zones and virtual stops enabling demand-responsive routing in peripheral areas. We simulate a 2×2 km area with six lines serving 1000 requests/day over five days, comparing Fixed-Route Transit, hybrid DRT without learning, and hybrid DRT with user adaptation. Results indicate that DRT can expand service coverage while reducing operator unit costs, even when users impose stricter scheduling requirements. Behavioural adaptation improves matching between passengers and vehicles, contributing to performance stabilisation over time. Efficiency gains strengthen as demand grows, suggesting that hybrid DRT can scale competitively in contexts where demand is uneven but locally concentrated.
Calabrò et al. (2026) studied this question.