This work presents a novel detailed analysis of entropy generation mechanisms in a desiccant wheel (DW) by the the direct method, enabling differentiation of entropy sources by type and location. A detailed model of the desiccant-based system, implemented in Modelica, is used to compute entropy generation from various phenomena. Furthermore, a novel idealized dehumidification process is proposed and compared to simulation results. Results show that irreversible heat transfer dominates entropy generation (ca. 40 % in the reference case), followed by mass transfer, dissipation, and mixing. Through parametric variation of regeneration temperature and wheel rotational speed, optimal operating points that minimize entropy generation are identified. It is demonstrated that entropy generation can be minimized by reaching an operating point with minimal temperature spread at the wheel outlet surface. Comparison with the ideal process reveals that a significant portion of the system's exergy requirements stems from irreversibilities within the DW itself, rather than from losses in discharged air streams.