Research on cascade coal procurement model based on Bi-objective and integer optimization with virtual coal blending calibration

China’s thermal power enterprises generally rely on manual experience while lacking robust modelling and optimisation support for coal procurement, leading to non-optimal economic performance and limited feasibility of procurement schemes. To address these issues, a cascade coal procurement model with two layers is constructed. The outer loop comprehensively considers the effects of pollutant emission on coal cost, as well as seasonal-variation effect from procurement to blending on coal quality, and an improved non-dominated sorting genetic algorithm-II (INSGA-II) is proposed for optimisation. The inner loop further considers vessel capacity and freight charges, and introduces a coal quantity adjustment and substitution mechanism to improve the original inner-loop model against the situation of “low planned coal quantity using high-capacity vessels”. Finally, a virtual coal blending model is constructed to calibrate the inner-loop scheme. Case results show that, compared with the experience-based scheme, the model-based inner-loop scheme reasonably adjusts the procurement quantity in the outer loop while ensuring supply requirements, reducing the cost by 5.75%, and the model-based virtual coal blending scheme not only optimally matches the planned load but also achieves greater economic efficiency. In addition, INSGA-II outperforms NSGA-II in cost–quality trade-off under different weights.