This paper investigates the effect of upper body on balancing by comparing balanced state domain in the phase space. Biped mechanism is simplified to multi-segment model which consists of one stance foot, leg, and upper body in sagittal plane. System parameters such as link mass and link length are chosen appropriately based on human anthropometry data. In addition to the system parameters, the necessary and sufficient conditions for balancing are implemented as constraints. Proposed algorithm iteratively solves nonlinear constrained optimization problem to find velocity extrema for a given set of joint variables and a maximum actuation torque. The balanced state domain of an actively controlled swinging arm demonstrates larger domain compared to the domain without swinging arm. While similar results are shown by other various approaches, the proposed algorithm demonstrates identification of the balanced state domain in a deterministic scheme using numerical optimization.