Accurate and computationally efficient models of indoor temperature dynamics are essential for building energy management and control. This paper proposes a simplified thermal model of a building floor, formulated through thermal balance equations and identified using only data recorded from the building management system. The physical structure of the system is represented as an RC circuit, with thermal capacity and thermal resistance estimated via two complementary optimization strategies: a metaheuristic method (PSP) and a deterministic approach (SQP). Simplification assumptions are introduced to improve prediction performance while preserving physical interpretability. A stepwise methodology is employed to progressively refine the model and align it with measured temperature dynamics. Results demonstrate that the proposed approach effectively reproduces the thermal behavior of the building, providing a practical trade-off between model simplicity, accuracy, and applicability for real-time control and optimization tasks in building energy systems. With the proposed identification approach, R2 scores of up to 0.84 and 0.86 were reached using respectively SQP and PSO algorithms.