Full-Scale Soft-Sensor Implementations Enable WRRF Hybrid Digital-Twins

Abstract

Digital twins (DTs) are on the rise in the wastewater treatment industry and, with ever-increasing use of online data in more complex applications, it is bringing us towards a digital revolution within water resource recovery facilities (WRRFs). An integral data requirement is dynamic real-time influent concentrations. However, few WRRFs have the ability to deploy sensors to generate the required dynamic influent concentrations. This paper presents an innovative approach to generate dynamic influent profiles with a soft sensor primarily using bioreactor airflow rates and DT models. It has been implemented on three different full-scale facilities and demonstrate it is feasible to retrieve 15-minute interval diurnal influent concentrations. This study overcomes a key bottleneck in DT applications - the lack of realistic high-resolution influent profiles - fostering the development of DTs. The fact that the data used in this study are commonly available ensures wide applicability for most WRRFs and prepares them for the advanced controls that digital twins can unlock.