In this study, the effect of orifice geometries on the detonation propagation is considered systematically in stoichiometric 2H(2)-O-2 mixture. Three various orifice shapes with the same blockage ratio (BR = 0.889) are used firstly, i.e., round, square and triangular. Eight PCB pressure transducers are employed to obtain the average velocity through two adjacent signals while the smoked foil technique is used to record the detonation cellular pattern. The experimental results indicate that three different propagation modes can be observed: (1) when the initial pressure (P-0) is smaller than the critical value (P-c), the steady detonation wave cannot be produced before the orifice plate, afterwards, the mechanism of deflagration to detonation transition (DDT) is seen; (2) near the critical pressure, a steady detonation wave is formed prior to the obstacle, but the failure of detonation is seen after its propagation through the orifice plate due to the diffraction effect and the mass and momentum loss from the wall, and then the phenomenon of detonation re-initiation is observed due to the reflection from the wall; (3) at the initial pressure larger than the critical value, the steady detonation wave can propagate through the orifice plate without decay. Moreover, although the effect of orifice shapes on the critical pressure can be nearly ignored, the re-ignition position is different among three various orifice geometries. For the cases of round and square orifices, the ignition position is produced near the center of the wall. However, the detonation wave is re-ignited from the corner in the case of triangular orifice. Finally, the critical condition of detonation propagation can be quantified as D-H/lambda > 1. But the critical values of D-H/lambda are not uniform among three different orifice geometries. For the cases of round, square and triangular orifices, the critical values of DNA are 8.94, 5.88 and 3.84, respectively. (C) 2019 Hydrogen Energy Publications LLC. Published by Elsevier Ltd. All rights reserved.