A comparison of different optimization techniques for variation propagation control in mechanical assembly

Abstract

Variation propagation control is one of the procedures used to determine product quality in the manufacturing assembly process. The quality of a product assembly is also greatly dependent on the product type and the optimization criteria employed in the assembly. This paper presents three procedures for optimizing the assembly of component stacks by controlling variation propagation. The procedures considered are: (i) straight-build assembly by minimizing the distances from the centres of components to table axis; (ii) parallelism-build assembly by minimizing the angular errors between actual and nominal planes; (iii) target-axis-build assembly by minimizing the distances from the centres of components to a target axis. Simulation results are presented for the assembly of four cylindrical components. The results show that the variation can be reduced significantly by optimization; for example, Procedure 1 can reduce the variation by 48%, Procedure 2 by 63% and Procedure 3 by 35%, compared to assembly without the minimization. The results also show that the 3 proposed assembly methods have reasonable consistency, the maximum of the standard deviation is 0.0726 mm in Procedure 1, 3.5×10 -4 rad in Procedure 2, and 0.029 mm in Procedure 3. These demonstrate that the three assembly techniques are fully functional on their own. Thus, the relevant assembly techniques need to be selected carefully in accordance with the particular industrial application.