Computer-aided tolerancing (CAT) aims to predict and control geometrical and dimensional deviations in the early design stage. Former simulation models based on the translation and rotation of nominal features cannot fulfill engineering demands or cover the product lifecycle. Nonideal feature-based simulation methods are, therefore, drawing a great deal of research attention. Two general problems for non-ideal feature-based methods are how to simulate manufacturing defects and how to integrate these defects into tolerance analysis. In this paper, we focus on the first problem. There are already many manufacturing defect simulation methods. Although they are derived from different fields and have different names, they share common characteristics in application. In this study, we collected different simulation methods and classified them as random noise methods, mesh morphing methods, and mode-based methods. The theoretical backgrounds of these methods are introduced, and the simulation examples are conducted on a consistency model to show their differences. Criteria such as multiscale, surface complexity, measurement data integration, parametric control, and calculation complexity are proposed to compare these methods. Based on these analyses, the advantages and drawbacks of each method are pointed out, which may help researchers and engineers to choose suitable methods for their work.Read less <