Set a 9-stop loop with OR-Tools and 15-minute buffers, and the driver called at 4:12 a.m. after reaching Stop 3 before the pallet labels left the printer. Anyone else optimize a route so efficiently it creates a time paradox, and how do you explain that to finance when the KPI looks perfect but reality needs a coffee?
But , been there — “4:12 a.m.” beats the label printer every time. In OR-Tools, I fix it by giving each stop a ready-by window and adding a soft lower-bound penalty on the time dimension so early arrival (waiting) hurts the objective: Vehicle Routing Problem with Time Windows | OR-Tools | Google for Developers. Do you have label-ready timestamps you can feed as lower bounds, or should we pad that 15-minute buffer?
I stopped getting 4 a.m… paradoxes by inserting a dummy ‘labels-ready’ stop at the depot with an earliest 04:30 window and a hard precedence before Stops 1–3, so OR-Tools plans the wait instead of “arriving early.” For finance, I export an extra ‘idle_at_stop’ metric from the time dimension slack so the KPI stays perfect but the coffee time shows — want the tiny tweak? Vehicle Routing Problem with Time Windows | OR-Tools | Google for Developers.