A robust process is insensitive to the effect of noise variables. Noise variables are the main source for producing variation. Noise variables are included in the outer array in robust design experiment for enhancing robustness. The approach of robust design is to make the process robust (insensitive) to variation due to noise variables. The effect of noise factors can be modelled in a response surface model which helps to determine the settings of the design factors that neutralize the effects of the noise factors and improve robustness. In experimental design the noise factors are assumed fixed value whereas in real world manufacturing noise factors vary randomly. Again for a large scale manufacturing, it is extremely difficult to study robustness using experimentation as there are chances of stoppage of production. In such a situation a simulation-based model can be developed using industrial data to study robustness of a real manufacturing process. This paper proposed a method (a combination of simulation, regression modelling and robust design technique) to study robustness of a hardening and tempering process producing component worm shaft used in the steam power plant. The process capability indices (both univariate and multivariate) are determined based on the model responses. The variation of process performance (process capability values) due to random noise variation is studied using a general purpose process control chart (R-chart). The results show that noise factors in hardening and tempering process are insensitive to manufacturing variation and process capability indices act as a surrogate measure of process robustness.

This content is only available via PDF.
You do not currently have access to this content.