Variation Analysis Application Experience for Mechanical Assembly Accuracy Assessment
The article presents experience in using the concrete Teamcenter - Variation Analysis system application to assess the accuracy of mechanical assemblies. When creating a geometric model of the product to be designed, modern CAD systems show the nominal geometry of a displayed part or mounting assembly with all the necessary annotations. It is natural that all necessary tolerances for linear and angular dimensions and also shape and location tolerances are assigned. But as a result of the statistical variation of the assigned tolerances, there are situations when individual parts, being classified as fit, are not assembled, or assembly defects are found just in the course of operation or repair of the finished product.Of all tolerances assigned at the design stage, it is necessary to emphasize the tolerances in face flatness, squareness, and parallelism and other tolerances of shape and location. The fact is that these tolerances ultimately determine the accuracy of an assembled product. Indeed, if the mating-in-the-assembly flat faces turn out to be insufficiently flat, this misalignment “brings to nought” all the efforts to achieve high accuracy of the parts to be joined. Repeated tests have shown that when assessing the individual contributions to a total error, the assembly inaccuracies turn out to be an order of magnitude more significant than errors in manufacturing of individual parts. The described Variation Analysis application is used to assess assembling inaccuracies.When analysing the individual parts, the described application can be effectively used to calculate dimension chains (direct and inverse problem). As is customary in many modern CAT systems, the Variation Analysis application takes into account the statistical variation of all the tolerances described. The user has to set the law of probability density distribution for all assigned tolerances in the intermediate items of the dimensional chain, and to perform a numerical experiment to simulate the entire parent population of random events. As a result, the obtained probability density distribution curve has a magnitude we are interested in, taking into account the statistical variation of tolerances in the intermediate items and the errors that occur during the product assembly. Moreover, the user can also have a certain protocol in which the system gives information on contribution that one or another assigned tolerance makes to the total error. This allows making efficacious corrections to the tolerances in order to obtain useful results.