Dimensional analysis of vehicle components in design phase via variation simulation analysis (VSA)

Dimensions of components in computer aided design phase are defined according to ideal conditions. However, it is not possible to produce components in their nominal (ideal) dimensions in reality. Dimensions of components changes due to uncontrollable reasons like variation of ambient temperature, electrical frequency, speed and distortions on casting and forging parts as they are taken out of the mold/stamp. As a result of this, form, roughness variation occurs on part surfaces. Position, angularity and form variations occur on holes. Parts such as pins or shafts are also not exactly rounded and their axes are not perfectly smooth. This situation can easily be observed by measurements. Each element on the part is called a characteristic element, such as a hole, a pin, shaped surfaces (or profile), two parallel surfaces, a slotted hole. In addition to size tolerance, geometry of characteristic elements also vary according to their form, profile, orientation, position and run-out.In order to keep all these deviations under control and to confirm that the production is acceptable, control symbols are defined using the GD&T language issued by organizations such as ASME and ISO for the technical drawings of the parts. Tolerances are assigned to each characteristic elements of components.Size tolerances can be defined as +/- on characteristic elements. Tolerances would also be defined on characteristic elements by using feature control frames that include control symbols. Details are transferred to the technical drawings in accordance with GD&T standards. Thus, both designers, manufacturers, quality and maintenance units would ensure that they all use the same language. This document refers to ASME standards. If manufactured products can fulfill GD&T limits specified in technical drawings, they are acceptable. Otherwise, products should be scrapped. Because, it directly depends on the GD&T, which ensures that parts can be assembled, assemblies are safely sealed, and the noise level controlled in dynamic parts. In this document, tolerance stack problems caused by tolerance deviations in assembly packages of vehicles and methods applied to solve these problems are discussed. Tolerance stack-up effects are analyzed according to the GD&T definitions given in the technical drawings of parts. We can simply predict why we need GD&T by considering the assembly of the two parts containing the hole and the pin pairs. In the ideal design condition, the hole and the pin are concentric and mating faces (surfaces) of the two parts are a hundred percent in contact with each other. But, in real life, the hole and the pin can not be positioned exactly concentrically. Certainly, there is some misalignments. Mating surfaces also do not contact perfectly with each other. Therefore, problems occur due to deviations arise from misalignments. Moreover, tolerance stack accumulation occurs in the case of large group of components are assembled. Tolerance sensitivity varies depending on manufacturing method of components. For example, a component produced with an injection mold has more sensitive tolerances than a component produced with sand molding method. Tolerances of a machined part,produced with turning or milling method,will be more sensitive than cast parts. Similarly,different tolerancing results can be obtained depending on the material and part shape. For this reason, engineers that apply tolerance analysis should be aware of the manufacturing method of components where tolerance analysis applied. It is a necessity for automotive companies to make an optimized design considering the tolerance sensitivity, cost and function of each components depending on the manufacturing method. Quality control is often applied on mass-production parts such as automotive components with the help of gauges and control fixtures. The critical characteristic elements on the component are generally measured by using specific measuring instruments or three-dimensional CMM measurement machines. The measurement methods and criteria are determined according to the tolerances in the technical drawings. Therefore, the measurement methodology for accurate GD&T identification is required to be known by the tolerance analyst. It is possible to apply one, two or three dimensional tolerance calculation depending on the part design. If components of an assembly package are assembled side by side in a queue (linear assembly), it is possible to define only +/- tolerances. Thus, it would be possible to apply one dimensional calculation by reducing tolerance values to a size. If it is not possible to control components with only +/- dimensional tolerances, it is necessary to perform two dimensional tolerance calculation even if the assembly package is linear. As a result, characteristic element control symbols are also needed. In some cases, position of components are angular to each other and there are too many tolerance deviations in different directions in three dimensions. In this case, three-dimensional analysis is needed. Automotive parts often have three-dimensional shapes in a complex structure. For this reason, three-dimensional tolerance calculations are often required. We can say that there are three different methods mostly used in tolerance stack calculations.We can classify these methods as "worst case method", "statistical method" and "6 sigma & Monte Carlo method".In the case of one and two dimensional calculations, "worst case" and "statistical" methods are mostly preferred. In the case of tree dimensional calculations, 6 sigma and Monte Carlo method generally becomes the first choise. With the aid of the Monte Carlo mathematical model, which is one of the probability computation methods, random numbers are assigned to each characteristic element existing on the part. Random numbers vary in the tolerance range defined to the characteristic element.Distribution of random numbers should be in a normal distribution form. Random number assignment is repeated in each simulation and a different stack result is obtained each time. Stack results,obtained via repeated action by defining random numbers, generates a distribution curve. The result obtained with this distribution curve represents uncontrollable dimensional differences that occur during manufacturing. Thus, the total effect occures in an assembly process is simulated. Effect of each characteristic element on the calculation result would be obtained with the help of HLM (high low median) graphs. HLM graphs show the contribution effect of each element on the tolerance stack result in percentage. This simulation is performed systematically to each characteristic element by applying the maximum, minimum and average tolerance values. The effect of the applied maximum, minimum and average values are reported as an output in the form of an HLM grap. HLM calculation is handled as a separate measurement operation in the computer programs in which the variation analysis is performed. The more number of components in the assembly increses, the less probability of worst case conditions occure. That is, assembly probability of components that would all together have maximum or minimum dimension in the same assembly case, dicreases.Therefore, it can be predicted that there will be some improvement in the results obtained depending on the number of components. This situation would be mathematically predicted with the help of statistical methods. When dealing with the statistical methods, it is necessary to have knowledge about terms such as"normal distribution curve" and "sigma distributions", "spec limits", "process capability analysis", "out of specification". Tolerance problems would generally solved by changing design, changing the assembly method or improving the tolerance. Changing the assembly order of components, using the dowel pin, or using the mounting fixture considerably alters the assemblability of components. Changes in the manufacturing process or changes in the material may improve the tolerances of the parts. Tolerance problems can be solved by making design improvements such as a change in the hole dimension to be able to align the part according to tolerance variations. Changing the constraints or using elastic connection elements in areas where huge tolerance varation occures, would also be the design solution. Package of an EGR system was chosen as an example application of this thesis study. In the first step, problems encountered during the installation of prototypes of the EGR package analysed and verified by tolerance stack-up calculations. In the second step, permanent solution of problems found by changing design, improving tolerance, changing characteristic control symbols in the technical drawings of the parts, and changing the mounting method. In the end, improvements were verified via prototype testing and hand calculations. Vis-VSA programme was used for 3D assessment of tolerance stack calculations. Later, a comparison was applied with results obtained by using different calculation methods. Comparison shown that the results obtained via Vis-VSA are consistent. This study showed that the completion of the tolerance stack calculations while in the design phase before the creation of serial production tools is a necessity in terms of high customer satisfaction and low cost requirements. Additionally, we can say that variation simulation analysis applied with tolerance stack calculation softwares, decreases calculation errors, provides time saving and optimum results. This thesis study has shown that tolerance stack calculations of elastic parts (pipes) such as "EGR outlet tube" and "air duct" in the EGR package is possible.