Shewhart's Statistical Process Control Chart

CHAPTER ONE: INTRODUCTION

1.1 Introduction
Shewhart’s Statistical Process Control Chart is One of the most commonly used tools in the area of Quality Control. These control charts are used more than past eighty years. These charts are used and added much popularity because of its easiness and efficiency. But, a noteworthy limitation to these traditional control charts is that these charts can be used only for univariate data. Therefore, the processes determined by multivariate data are not be monitored efficiently. The variables may be correlated with the multivariate structure. The correlations exists between the variables, are called inter-correlations. The correlations exists within each variable over time is called autocorrelation.

If the effect of multiple parameters is not independent MVA became very useful. Also, if some parameters are partial or complete measures of some other parameters (correlation) then too this became very useful. There are some cases in which the true source of variation may not be recognized or may not be measurable. (https:// qualityamerica.com/knowledgecenter/statisticalprocesscontrol/when_to_use_a_multivariate_chart.asp 2012). For an illustration, olumetric Flow and Pressure may be the process parameters being controlled. But the Temperature at some point say x, in the process may sway both. So, Mass Flow is the common factor affecting the

Hence, these tactics was more vigorous and lying to fewer false alarms. Plotting a chart (parametric and nonparametric) with univariate data for each variable from multivariate data may not automatically harvest results as exact as monitoring the multivariate distribution as entire. In multivariate processes, the multivariate breakdown into separate p univariate processes produces a loss of power in those tests because; this breakdown does not reflect correlations between the variables. (Fuchs and Kennet, 1998) The industries like food and chemical industries among others, intrinsically deal with humpty number of variables which are highly correlated. There is a great need for development in the area of MSPC to screen processes more professionally in the industrial setting. (Elsayed, 2000) Old-style methods also tend to produce higher false alarms in the incidence of autocorrelation which is quite usual in