Bats fly with highly articulated and relatively heavy wings. To understand power requirements, we have developed a three-dimensional reduced-order numerical model, and have analyzed flights of Cynopterus brachyotis, the lesser-nosed dog-faced bat. Using previously-measured wing kinematics, the model computes aerodynamic forces using quasi-steady Blade Element Momentum Theory and incorporates inertial forces of the flapping wing using the measured mass distribution in the wing and body. The two are combined into a Lagrangian equation of motion to predict the free flight of the animal which is compared with the experimentally observed behavior over a range of flight speeds (3.25 – 7.4m/s). To validate the model, the computed lift and thrust are compared with the forces needed to generate the observed motion in the live experiments. In general, we find good agreement, although the average thrust is slightly underpredicted. We use Monte Carlo simulations to quantify uncertainties due to measurement errors and modelling assumptions, and find that the simulations are most sensitive to the empirical coefficients of the aerodynamic model (lift and drag functions). We use the model to analyze flight power requirements, separating the power into aerodynamic and inertial components. We find that, towards the end of downstroke, the inertia of the heavy wing can be directed towards generation of aerodynamic power, thus alleviating the required muscle power; similarly, aerodynamic forces are found to assist the inertial requirements during the upstroke.