Counter-current flame propagation through fixed packed beds serves as a canonical form for understanding biomass thermochemical conversion phenomenon-this is due to the 'universal behavior' of fuel consumption rate per unit area with superficial velocity variation. This behavior is directly related to the particle density scaled ignition time being independent of biomass type, varying only with superficial velocity. But when the devolatilization time exceeds ignition time for a particle, the predicted fuel consumption rates will be higher than the actual values due to overlap of ignition and devolatilization. This situation is relevant for larger particles (> 30 mm equivalent sphere) and thin wood chips, due to sharp edges aiding quick ignition. The current work aims at generalizing the 'universal propagation model' to account for this effect which is critical to designing grate furnaces. Towards this, packed bed experiments were performed in a 500 mm dia, 1 m height cylindrical reactor with groundnut shell briquettes (GSB; dia 100 mm, lengths varying from 40 to 110mm) and pellets (dia 8mm and length varying between 15 to 20mm) at a superficial velocity of 20 and 30 cm/s. The fuel mass flux is calculated by two methods 1) from the slope of time-mass curve 2) from time-Temperature plots measured at fixed thermocouple locations along the reactor. Fuel flux measured using the two methods match closely for pellets, but with GSB, the fuel flux from method 2 is higher than method 1 by about 15 %. This indicates that subsequent fresh layer of biomass ignites before complete devolatilization of the previous layer leading to oxygen starvation. This also indicates a transition from ignition controlled to devolatilization rate controlled flame propagation, consistent with the expectations based on variation of ignition and devolatilization time with particle size measured for single particles in an earlier study at our lab. These results are used to develop a unified ignition-devolatilization model for single particle combustion, which is used to modify the 'universal propagation model' to account for particle size and shape effects. The predictions from this model are shown to closely match with the observed results. © 2017 The Authors. Published by Elsevier Ltd.