| Abstract [eng] |
This thesis analyses flow and delays in a dynamic public transport service system using discrete event simulation. A simplified hub-based bus service model was implemented in AnyLogic, where passengers arrive stochastically, wait in a queue, and are served by buses with limited capacity. The initial model was extended with probabilistic disruption, dynamic bus allocation policies, fairness constraints, and cold-weather sensitivity. The main objective is to evaluate how disruptions affect passenger time outcomes and whether dynamic bus allocation can reduce the probability of excessive delays. Several scenarios were compared, including normal operation, disrupted operation, greedy allocation, fairness-protected allocation, and weather-adaptive allocation. Passenger time samples were collected from simulation runs and analysed using Python through summary statistics, probability distributions, cumulative distribution functions, percentiles, and threshold exceedance probabilities. The results show that disruption significantly worsens passenger time outcomes. In the disrupted baseline scenario, the mean passenger time increased to 60.70 minutes, and 52.0% of samples exceeded 60 minutes. When dynamic allocation was accepted, the mean decreased to 25.43 minutes, and the probability of exceeding 60 minutes decreased to 0.0%. The results demonstrate that dynamic service adjustment can reduce extreme delays, but fairness and weather constraints are necessary to avoid negatively affecting passengers on other routes. |
