This study investigates the measurements data attained in Murcia Atrium fire tests (Gutiérrez-Montes, Sanmiguel-Rojas, Viedma, & Rein) using general CFD code, CFX-16.2 ( CFX-16.2, ANSYS, Inc.). Experiments were executed by burning heptane as a fuel for three different fires capacities, (1.32, 2.28 and 2.34 MW). Historically, CFX was validated with experimental data for compartment fires including natural ventilation only. In this paper validation was performed for developing fires during transient period in a large space, including the effect of combined natural and forced ventilation. Two different combustion models i.e. Non-combustion model and Eddy Dissipation Model (EDM) were used to represent the combustion of heptane. Four different approaches were employed to determine the predicted dynamic behavior of the smoke height and to evaluate the accuracy of CFX predictions. The results display CFX predictions have a reasonable concord with the experimental data. Therefore, CFX can be a valid tool for performance-based engineering to evaluate the smoke systems required for complex enclosed structures.