Abstract: Critical heat flux (CHF) prediction methods used in subchannel codes are primarily based on tube CHF values. These methods do not take bundle-specific effects into account, such as element gap size, curvature and the presence of an adjacent unheated surface. Furthermore, in order to simulate the bubble behavior in the corner channel and produce better results, it is necessary to use eccentric rod in the annuli to simulate these regions approximately. Fuel cladding swelling at the end of cycle (EOC) causes fuel rods to bow. The fuel rods between the two adjacent sets of grids will bow, which results in a certain decrease in the value of CHF of the bundles. Therefore, it is necessary to study the effect of rod bowing on CHF. The simplest form of a rod bowing section is to place a bowing rod inside a tube. An experiment was carried out to obtain data on CHF in vertical internally heated annuli cooled with Refrigerant (R-134a) at high-pressure subcooled and low-quality conditions (outlet pressure: 1.78-2.72 MPa, mass flux: 587-2 135 kg/(m²·s), critical quality: -0.64-0.12). The test section included three forms: concentric, eccentric and closure. The effect of eccentricity and rod bowing on CHF was studied from the mechanism of two different types of CHF (DNB and Dryout) in vertical internally heated annuli. The experimental results show that the inner rod eccentricity has a punitive effect on CHF at high subcooled condition, and CHF decreases with the increase of eccentricity. The eccentricity effect appears to be diminished at low subcooled conditions. The void drift effect causes the CHF with eccentricity of 0.783 to be larger than the CHF with eccentricity of 0.435 at high pressure and high mass flux conditions (HPHF). Rod bowing experiment results show that bowing with small closure has little effect on CHF. For the type of Dryout CHF with low mass flux, the rod bowing can destroy the stability of liquid film. For the type of DNB CHF with low mass flux, the void drift effect at the bowing annuli is much smaller than the eccentric annuli, and the CHF with rod bowing is smaller than the CHF with rod eccentric with the same minimum gap. For the type of DNB CHF with high mass flux at high closure, the enhancement effect of void drift is not enough to offset the flow reduction effect caused by the increase of closure. As a result, CHF decreases with increasing closure.