A delivery plan is easy to admire while every order is already known. Real dispatching has a less tidy rhythm: orders arrive, change, or disappear after vehicles have started moving.

The simulation began with 50 customers. Ten demand events arrived in five batches, leaving 53 customers and 14,133 kg of demand. The comparison held the initial plan and event sequence fixed. Both online strategies completed the work with seven vehicles.

−7.05%Total costRolling reoptimization vs. insertion
−15.81%Total emissionsSame 7-vehicle online comparison
−12.14%Total distanceSame simulated event sequence

Sequential insertion kept the old unserved-customer order and placed each new request into it. Rolling reoptimization allowed new orders and unfinished work to be reconsidered together. It completed three reoptimization batches; two new orders still needed the insertion fallback when no feasible reoptimized plan was found.

The static plan knew every future event in advance and is therefore not a fair real-time rival. The meaningful online comparison is between the two strategies facing the same sequence as it unfolds.