Optimising Transportation Routes of Oil Trucking Taking Into Consideration Various Economic Factors

Petroleum product trucking is exposed to persistent cost pressure. Fuel prices move quickly, maintenance of specialised tankers continues to rise and operators must still satisfy strict delivery windows and driver‐hour rules. In this article, we present a hybrid optimisation scheme that feeds live fuel prices from commercial market APIs into a mixed‐integer linear programme whose remaining cost coefficients are supplied by gradient‐boosting models of maintenance and depreciation. The resulting routes respect tanker capacities, time windows and the 8‐hour driving limit. We tested the approach on 6 months of real urban operations (January–June 2025) involving 47 routes, 23 stations and a fleet of 12 tankers. Against the routes the operator actually used, total cost fell by 14% (bootstrap 95% CI 13.2–14.7). Fuel accounted for roughly half of the absolute saving, maintenance for about a third and labour for the rest; average delivery time shortened by 1 hour ( p = 0.023). The same runs beat both a pure Dijkstra policy (10% saving) and a static MILP that employed only average costs (12% saving). When the network was expanded in simulation to regional and international scales, the projected savings were 12% and 10%, respectively. Even after we artificially inflated fuel‐price volatility by 20%, the cost reduction stayed above 10%. The method therefore appears usable for day‐to‐day planning under realistic price uncertainty.