A forecast-driven rolling-horizon-based Wagner–Whitin approach for creating a smart coal-port inventory system

Inventory management at coal-port terminals is challenged by substantial price uncertainty and market fluctuations. This study develops a forecast-driven inventory management framework that adaptively determines horizon length based on predicted price trends and solves each horizon using an enhanced Wagner–Whitin algorithm to optimize replenishment decisions and inventory levels. To better amortize set-up costs while preserving decision flexibility, a deferred-ordering mechanism is incorporated within the horizon strategy. Short-term coal prices are predicted using a long–short-term memory model that captures long-range dependencies. Computational experiments show that the proposed approach consistently outperforms the conventional fixed-horizon strategy in terms of the total cost. Sensitivity analyses further show that volatility magnitude and detection thresholds are first-order determinants of planning granularity, order frequency and inventory cost, thereby explaining the framework’s adaptive behaviour. Overall, the approach is readily implementable and provides a scalable, data-driven policy for bulk-commodity inventory control at coal terminals.