This paper revisits the single-vehicle Pickup and Delivery Problem with Time Windows from a temporal-structural perspective. We introduce a constraint programming formulation that couples a global circuit constraint for routing with conditional difference constraints governing arrival times.This construction naturally induces a dynamic Simple Temporal Network, enabling a unified representation of precedence, waiting, and time-window constraints within a single temporal framework. We study the resulting propagation mechanisms and show how the explicit temporal structure alters constraint interactions relative to classical formulations.The approach is compared to a mixed-integer linear programming baseline. An auditable benchmark instance is developed to allow exhaustive enumeration and controlled sensitivity analysis on time-window parameters. This work was carried out under the supervision of Primous Pomalegni and Henri-Joël Akougbe at the Université de Montréal.