Creation of a screening analytical approach for the efficient detection of anomalous performance across large refrigeration pack estates using electrical usage data.

Historical analysis of electrical energy usage data from over 350 High Temperature and Low Temperature retail refrigeration packs has shown a very high level of variability in pack electrical energy usage. This noisy big data environment makes the task of detecting anomalous pack behaviour energy wastage events, a very difficult one. This paper attempts to address this problem, by presenting a modified process characterization approach that through the meaningful subcategorisation and statistical analysis of a pack’s annualized energy usage, can give the practitioner a much better understanding of the relative contributions of baseload, within day variation, and summer seasonal variation sources, present within a pack. It goes on to show that the resultant creation of a screening analytic using only electrical usage data, when applied across a complete estate, can deliver effective pack level nomaly detection, and subsequent cost savings through the timely detection and avoidance of these significant energy wastage events.