Consistency conditions have been successfully utilized for data-driven artifact reductions in cone-beam computed tomography systems equipped with a large area flat-panel detector. Recently, many formulations and applications of pair-wise cone-beam consistency conditions have been published, including Grangeat, Smith, and fan-beam consistency conditions. Previous works demonstrated that the polynomial coefficients for beam hardening correction could be directly computed from cone-beam raw data by enforcing consistency conditions on projection pairs. This paper compares the effectiveness of pair-wise consistency conditions for mono-material beam hardening correction using a second-degree polynomial. The results from our studies show that similar corrections could be achieved for ideal polychromatic projections. We also investigated the effectiveness of corrections after perturbing the projections with an increasing degree of errors other than those caused by beam hardening. The studies indicate the superior robustness of fan-beam consistency conditions towards Poisson noise, axial truncation, detector shift, and scatter, while Grangeat consistency conditions were less vulnerable to projection intensity errors. The optimal choice of consistency conditions depends on the CBCT system geometry, physical phenomena other than beam hardening, and the availability and accuracy of pre-processing and artifact corrections algorithms before beam hardening correction.