The probability of detection (POD) of material defects by non-destructive testing (NDT) is an important input in the risk assessment of aero-engine life-limited parts. Obtaining a POD curve generally requires many assay experiments. To improve the efficiency, based on model-assisted POD (MAPOD) theory, a method to obtain ultrasonic inspection (UT) of POD was developed with the finite element method and reflectance theory as physical models. The method can consider the influence of physical factors in the inspection, which are input as variables in the finite element model. The pressure curve obtained from the finite element model was processed to obtain the signal response distribution, and the POD curve was calculated from the distribution. The method exploits the ability of finite elements to deal with defects of different shapes, while reflectivity theory avoids repeated calculations for defects with different material properties. In this paper, the process of obtaining POD curves by this method is illustrated in the example of hard α material defects in TC4 alloy, where the defect orientation and nitrogen content are used as physical influencing factors. This method assists in reducing the number of experiments and time spent on the acquisition of POD.