For a class of industrial processes described by a second-order plus time delay model, this paper investigates the design of active disturbance rejection control (ADRC) incorporating model information. By using the low-frequency approximation, the lumped plant consisting of all internal loops can shape the closed-loop tracking dynamic to the expected response, allowing the feedback gains to be formulated by the desired tracking factor. In this framework, a rigorous stability criterion is first depicted intuitively in terms of dimensionless parameters. An explicit upper bound on the observer bandwidth is then analytically derived through an in-depth analysis of the decaying behavior in the magnitude of the loop function, which guarantees a more satisfactory margin for mismatched delays than existing related studies. Based on this constraint, the regulable tracking factor can further endow the ADRC system with comprehensive robustness to cope with other uncertainties. Comparative simulations and a water tank experiment reveal the merits of the proposed configuration. Furthermore, the field application of this method in the wet flue gas desulfurization process demonstrates its promising practical prospects.