A review of non‐standard applications of statistical process control (SPC) charts

Abstract

The principal application domain for statistical process control (SPC) charts has been for process control and improvement in manufacturing businesses. However, the number of applications reported in domains outside of conventional production systems has been increasing in recent years. Implementing SPC chart approaches in non-standard applications gives rise to many potential complications and poses a number of challenges. This paper reviews non-standard applications of SPC charts reported in the literature from the period 1989 to 2000, inclusive. Non-standard applications are analysed with respect to application domain, data sources used and control chart techniques employed. Applications are classified into five groups according to the types of problem to which control chart techniques have been applied. For each group the nature of the applications is described and analysed. The review does not show a paradigm shift in the types of SPC control chart applications but does show clearly that the application boundaries extend considerably beyond manufacturing and that the range of problems to which SPC control chart techniques can be applied is much wider than commonly assumed. The paper highlights the critical fundamental and technical issues that need to be addressed when applying SPC chart techniques in a range of non-standard applications. Wider managerial issues of importance for successful implementations in non-standard applications of SPC control charts are also discussed.

Detecting disturbances in supply chains: the case of capacity constraints

“…The underlying philosophy is based on Ansoff (1975), who first advocated the use of early warning or weak signals to manage unpredictable, dynamic and hence difficult to plan contexts, and Haeckel's (1999) sense and respond approach, that involves ones involve modeling the behaviour of system variables in statistical terms such as by fitting a probability distribution or a time series model and then monitoring deviations (vis-à-vis usual/normal behaviour), with an alarm being triggered in case of a specified threshold being breached (Basseville and Nikiforov, 1993;Garcia-Teodoro et al, 2009). Though successfully applied in manufacturing contexts for process and quality control (Montgomery, 2005), these techniques have not seen significant application at a supply chain level (MacCarthy and Wasusri, 2002), which could be due to challenges such as: 1) Difficulty in stochastic modelling of variables' dynamics for multistage systems such as supply chains (Batson and McGough, 2007;Tsung et al, 2008), 2) Difficulty in specifying thresholds due to the non-stationary nature of the supply chain variable profiles (from the continuous changes in the internal and external environment), 3) Involvement of a large number of system variables in a typical supply chain assessment, where these techniques are known to be less effective (Woodall and Montgomery, 1999), and 4) Difficulty in effecting optimal mitigative responses post-detection, as a 'disturbed'/'not-disturbed' kind of detection (rather than indication of the specific disturbance impacting a system) is provided by these techniques.…”

Measuring service outcomes for adaptive preventive maintenance

“…Bersimis et al, 2007;MacCarthy and Wasusri, 2002;Stoumbos et al, 2000;Thor et al, 2007;William and Douglas, 1999;Woodall, 2000), we claim this study brings something new to the table (as summarized in table 1). First of all, in this study we measure the (technical) outcome of a conceptual maintenance process, where the data illustrates rates of outcomes -whereas in most SPC charts each data point represents a single output or outcome.…”

“…When considering DP1 (the design needs to accommodate a probabilistic phenomenon) and DP2 (the design needs to distinguish special-cause failures from common-cause failures), we note that there already exists a performance measurement method that satisfies both, namely SPC (MacCarthy and Wasusri, 2002;Oakland, 2008). However, SPC has its origins in manufacturing and despite more recent service implementations (cf.…”

Monitoring and reducing patient dissatisfaction: a case study of an Iranian public hospital

“…Second, since several correlated variables are involved, interpreting the signals from multivariate control charts is more difficult in comparison with univariate control charts. When a p-variable chart alerts an out-of-control situation, it is necessary to construct p univariate charts or apply other statistical techniques to find out which variable manifests out-of-control behaviour (Bersimis, Psarakis, & Panaretos, 2007;MacCarthy & Wasusri, 2002). Third, by increasing the number of variables they become less efficient (Waterhouse et al, 2010).…”