Carbon/phenolic ablators are successfully used as thermal protection material for spacecraft. Nevertheless, their complex thermal degradation is not yet fully understood, and current pyrolysis models do not reproduce important features of available experimental results. Accurate and robust thermal degradation models are required to optimize design margin policy. We investigate whether the competitive kinetic schemes commonly used to model biomass pyrolysis are appropriate to describe the thermal degradation of carbon/phenolic composites. In this paper, we apply competitive pyrolysis mechanisms for the thermal degradation of the carbon/phenolic ablator PICA. Model parameters are then calibrated using a robust two-step methodology: first deterministic optimization is used to obtain the best estimation of the calibration parameters based on the experimental data, then a stochastic Bayesian inference is performed to explore plausible set of solutions taking into account the experimental uncertainties. The proposed calibrated model provides an accurate description of the pyrolysis process at different heating rates. The model shows great flexibility and robustness at a similar computational cost as the traditional devolatilization models. This opens the possibility for more complex mechanisms when more experimental data becomes available.